1987 The typographic contribution to language:

towards a model of typographic genres and their underlying structures

Robert Waller

A note about this edition

After completing my PhD thesis in 1987, I left the research career that had
led up to it, and went into practice as an information designer, although
some ideas from the thesis were published (Waller 1990, 1999). For the
next twenty years, my main focus was commercial survival, and I gave
it little thought until Judy Delin and John Bateman got in touch. Their
AHRC-funded GeM project looked at the role of layout in written text, and
took a genre-based approach. I was delighted when they used my model
as a starting point for their own richer and more robust version (Allen,
Bateman and Delin 1999, Delin, Bateman and Allen 2002).

At their suggestion I created a digital version, by converting the original

WriteNow file into an early version of Word, and upwards through newer
Word versions. The GeM project made this available online, and I have also
placed in on my own website in recent years. However, many illustrations
were missing, and others were scanned from photocopies.

Recently I imported the Word file into InDesign, and I located and
re-scanned most of the images. What you have here is the original text,
although page numbers have inevitably changed, and I have also corrected
a number of typos. One or two exemplar documents are not the originals
but make the same points.

I am mindful that this thesis is over twenty-five years old and was written
before the internet, and in the very early days of electronic documents
and hypertext. And since 1987 much has been published on genre theory
by scholars such as John Swales, Douglas Biber and Vijay Bhatia. I am
also very aware of a huge deficiency in the thesis, because at the time of
writing it I had not encountered Christopher Alexander’s powerful concept
of pattern language (Alexander, Ishikawa & Silverstein 1977). I am certain
Thesis submitted for the degree of that this would have featured significantly in my discussion of how genre
Doctor of Philosophy, Department rules are articulated, and how functional imperatives evolve into genre
of Typography & Graphic
Communication, August 1987. conventions (Waller, Delin & Thomas 2012).
RW January 2014.
Contact information
[email protected] If you are interested in following up key citations of this thesis, I have
www.simplificationcentre.org.uk added a supplementary bibliography at the end. I have added the original
www.robwaller.org page numbers in red in the margin. RW January 2020.
 

Contents

Abstract 4

Acknowledgements 5

Introduction 6

1 Typography and language: a selective literature review 11

Typographers on typography 11
Applied psychologists and typographic research 25
The ‘language element’ in graphic communication 37
Some linguists who have noticed graphic aspects of language 44
Conclusion 55

2 Theoretical knowledge in the world of practice 57

The myth of the two cultures 57
Pure and applied research 60
Paradigms 64
Guidelines and slogans 67
Explicit and tacit knowledge 70
Holistic thinking 72
The specialization of scholarship 73

3 Arbitrariness and linearity: de Saussure’s ‘basic principles’ 76

The criteria for languageness: arbitrary, segmented, systemic and linear 76
Language or paralanguage? 77
Arbitrariness 80
Levels of analysis: letters, words, paragraphs 83
The segmentation problem 86
The problem of linearity 96

4 Writer-syntagm and reader-syntagm 105

Methods of configuration: linear vs non-linear 105
Reading strategy: receptive vs self-organized 108
Eye movements: foveal vs peripheral 112
Reading comprehension: bottom-up vs top-down 113
Oral vs silent reading 115
Punctuation theory: dramatic vs grammatical 117
Resolving the dichotomy 125

Robert Waller • The typographic contribution to language • PhD Thesis 1987 2

 

5 Communication models 128

Coding and decoding 131
Conversational models 134
A genre model of typographic communication 139
Interaction of the structures 143
Genre and textuality 148

6 Topic structure 152

Visual and spatial metaphor 154
Topic diagrams as writing plans 168

7 Artefact structure 173

Levels of graphic segmentation 174
The Procrustean page 182
Grid systems 186
Editorial intervention and artefact structure 191
Medium and message 194

8 Access structure 200

Designing for different purposes 200
Numbering systems 202
Page layout as access structure 205
Co-operative and uncooperative media 208
Grice’s Co-operative Principle 209
Van Dijk’s relevance cues 211
Turn-taking 214
Context 219

9 Genre structure 223

Genres in literature 223
Text types 225
Genres as ordinary-language categories 227
Genre markers and genre rules 229
Genre rules and error detection 234
Genre and design method 236
Conclusion 239

References 242

2020 Supplementary bibliography 261

Robert Waller • The typographic contribution to language • PhD Thesis 1987 3

 

Abstract 3

This thesis presents a model that accounts for variations in typographic

form in terms of four underlying sources of structure. The first three relate
to the three parts of the writer-text-reader relationship: topic structure,
representing the expressive intentions of the writer; artefact structure,
resulting from the physical constraints of the medium; and access structure,
anticipating the needs of the self-organized reader. Few texts exhibit such
structures in pure form. Instead, they are evidenced in typographic genres
– ordinary language categories such as ‘leaflet’, ‘magazine’, ‘manual’, and
so on – which may be defined in terms of their normal (or historical)
combination of topic, access and artefact structure.

The model attempts to articulate the tacit knowledge of expert

practitioners, and to relate it to current multi-disciplinary approaches
to discourse. Aspects of typography are tested against a range of ‘design
features’ of language (eg, arbitrariness, segmentation and linearity). A
dichotomy emerges between a linear model of written language in which
a relatively discreet typography ‘scores’ or notates the reading process
for compliant readers, and a diagrammatic typography in which some
concept relations are mapped more or less directly on the page for access
by self-directed readers. Typographically complex pages are seen as hybrid
forms in which control over the syntagm (used here to mean the temporal
sequence of linguistic events encountered by the reader) switches between
the reader (in the case of more diagrammatic forms) and the writer (in the
case of conventional prose). Typography is thus most easily accounted for
in terms of reader-writer relations, with an added complication imposed by
the physical nature of the text as artefact: line, column and page boundaries
are mostly arbitrary in linear texts but often meaningful in diagrammatic
ones.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 4

 

Acknowledgements 4

Finishing a thesis is a cause for personal celebration, but until one has
achieved a critical distance from one’s own work it is hard to know how
much it will please other people to be associated with it through an
acknowledgement. I am happily not in the position of one writer, who
would have liked to acknowledge his friends but no one helped him, 1 nor,
at the other extreme, of another whose debt to a friend was so great that
‘he alone is to be blamed for any shortcomings this book may have’.2

I owe an enormous debt to my supervisor Michael Twyman, to the

late Ernest Hoch, also my supervisor at the time I started work on this
project, and to colleagues at the Open University, especially Michael
Macdonald-Ross, Peter Whalley, and the late Brian Lewis. I have knowingly
and unknowingly absorbed an untold amount of information and insight
from these colleagues over many years. One never knows whether ideas
that seem original are actually half-remembered from conversations with
others.

I would also like to thank David Hawkridge, Director of the Institute of

Educational Technology, for giving me the time and encouragement to
finish. Others whose encouragement and comments I have greatly valued
include Pat Costigan-Eaves, Paul Stiff, Jane Wolfson, David Wolfson and
Jenny Waller, who had to put up with the traditional eccentricities and
bouts of despair of the thesis-writer.

RW August 1987

1 The preface to Jan V White’s Graphic idea notebook (1980) reads ‘It is customary to thank the people
who have been helpful in the process of book-making, How I wish I could have palmed off some of the
labors onto someone else! Alas…I was stuck with doing it all.’

2 Talbot Taylor, Linguistic theory and structural stylistics, Pergamon Press, Oxford 1981.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 5

Introduction 5

A reasonable common-sense definition of typography with which to start

might be ‘the visual attributes of written, and especially printed, language’.
Like all appeals to commonsense, mine embodies certain assumptions,
preoccupations and interests that will bias the way this enquiry develops.

For one thing, letterforms and layouts are not of interest to this study in
a formal sense but only in so far as they exhibit that quality of difference,
which is at the heart of language. Although at a certain level of analysis
a spoken sentence may be said to be the same as its written equivalent, it
is never exactly the same in substance or effect. It has been diminished in
some respect, but it has also been enhanced: writing has only a crude and
unreliable version of vocal pitch, gesture and tone, but it can contribute
spatial organization and graphic emphasis. Through the technology used to
write, whether a biro or a computer display, written language gives its own
particular clues about its origin. It is typography that has both diminished
and enhanced the subtlety of the message.

There are other visual attributes of written language that have no spoken
equivalent: a table, for example, contains the potential for a large number
of interactions between row and column headings. A skilled reader of tables
can perceive patterns in the data such as would be impossible should the
information be read out aloud – in the case of a large table, a long and
tedious process. In the case of the table, a fairly simple graphic system, the
interface between verbal and visual language has already become blurred,
and it becomes more so when we consider diagrams and diagram-like
typographic layouts.

To those deeply involved with the teaching or practice of typography, such 6

things as creativity, meaning, quality, and style are easier to exemplify than
to explain. But to those outside, in so far as they are aware of them at all,
such aspects of typography are something of a mystery. Engineers designing
text display systems, psychologists studying the reading process, and
even those researching into legibility and other typographic issues, often
seem unaware of the role of typographic subtleties. There is no obvious
conceptual framework within which typography can be discussed in this
public domain, within which logical ties can be clearly seen between its
aims, methods and effects.

‘Conceptual framework’ is a term we tend to use rather loosely to allude

to something we want people to think is rather precise. In concrete terms
it may be realized in a range of scholarly formats. It may be a taxonomy,
which identifies the most relevant dimensions of an issue and arranges the

Robert Waller • The typographic contribution to language • PhD Thesis 1987 6

Introduction

data accordingly; these dimensions may be abstract aspects of typography

or they may be examples of typography in practice. A conceptual framework
may be a set of rules thought to govern the behaviour of such components,
which can be tested empirically – in the natural sciences these may be laws
or theorems; in linguistic terms they may be grammars. It may consist of
theoretical models such as those constructed by cognitive psychologists
to explain the ‘mechanisms of the mind’. Less formal explanations such
as metaphors or analogies can act as organizing principles to direct our
thinking about a field of study. Less formally still, slogans and catchphrases
(‘form is function’, ‘the medium is the message’) can also have a unifying
and directive effect. The imprecision of the term ‘conceptual framework’
may be helpful at this stage: it allows us to refer to something we have not
yet constructed or even specified, without predetermining its status.

Many practical activities get by perfectly well without any articulated

conceptual framework. Is typography any different from, say, plumbing or
car maintenance, that in addition to a range of practical techniques and
strategies there should be an underpinning intellectual system? In the
context of literary theory, Eagleton (1983: 198) remarks that 7
‘Many literary critics dislike the whole idea of method and prefer to work by
glimmers and hunches, intuitions and sudden perceptions. It is perhaps fortunate
that this way of proceeding has not yet infiltrated medicine or aeronautical
engineering.’

No one has died from a poorly constructed novel, we may hope, and we
hear of few accidents involving poems, but bad typography actually can
have quite serious consequences – for example, if instructions or signs are
ambiguous. Typographers have their own response to anti-intellectualism
in this apt, if somewhat condescending, remark from Stanley Morison’s
preface to the second edition (1951) of First principles of typography:
‘The act of organizing a piece of printing so that its correct presentation may be
achieved requires, in the first instance, a sense of method. To be valid this method
must conform to right observation, thinking and reasoning. All men are able to
think, but not everyone is willing to train and exercise that faculty. The process of
thinking is, in fact, often so painful that many prefer to ignore this essential means
to the right solution to the problem.’ (p. 22).

Morison probably overestimates the ability of traditional articulated

reasoning to cope with multi-faceted problems: the apparently unthinking
reliance of craftsmen and women on aesthetic judgements about ‘balance’
and the like may actually represent the only way of expressing a kind
of reasoning that, being so complex, is impossible, or simply tedious, to
express in language.3

3 This issue is taken up in more detail in Chapter 2.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 7

Introduction

Nevertheless, it is one aim of this study, at the outset at least, to try to

suggest a framework within which typography can be discussed and
criticized – but in reasonably everyday terms without the need to dress
one’s thoughts in the full regalia of semiological classification schemes.
Several broad arguments for the development of a conceptual framework 8
may be suggested:4

Typographic education: as it is usually taught at present, typography

primarily involves visual judgement, manual dexterity and holistic problem-
solving ability. These practical skills are taught experientially through
project work, criticism and apprenticeship. But it is also a facet of literacy,
being concerned with the use and interpretation of language. Those
practising it should be literate people skilled in the handling of ideas.
Editors, who also fulfil a mediating role in the publishing process, are
recruited from the graduates of mainstream university disciplines and given
a minimal training. The manual and visual skills required of typographers,
though, are too great for the same system to work: their intellectual
training must therefore grow out of their practical training. It can be
additionally argued that typographic education needs a sounder conceptual
base in order to counteract the strong gravitational pull of the more
glamorous parts of the graphic design world, which many design students
aspire to but few are destined to enter. It should also enable designers to
adapt the specific skills they learn at college to new technologies as they
emerge.

Typographic research: typography has frequently been shown in experiments

to affect the legibility and understanding of texts. But for research to show
exactly how this happens – for hypotheses to be generated and tested
– a coordinating framework is needed. There is also a general cynicism
among practising designers about the worth of such results: partly because
given the lack of a conceptual framework it is hard to generalize from the
‘laboratory’ to the real world, and partly because their own working method
is more instinctive than cerebral.

Design management: although the word ‘design’ describes every aspect

of the planning of a product, the word (and the process it refers to) has 9
eluded easy definition. While typography might be said to be present in all
written texts, whether produced by people termed ‘typographers’ or not, it
frequently forms one element of a production process that takes place in an
institutional environment. Like any other aspect of institutional life, it must
be managed: scheduled, fitted into other processes, explained, or in some
other way articulated. Creativity is regarded as difficult to coordinate within

4 Many of these points were first raised by the Working Party on Typographic Teaching (set up by the
Society of Industrial Artists and Designers and the Society of Typographic Designers) in their 1968
interim report, a version of which was published in the Journal of Typographic Research (since renamed
Visible Language), 1969, volume 3, 91–102.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 8

Introduction

an industrial process. The unfortunate reputation designers have acquired

is highlighted by a recent advertisement for a computer graphics product
which promised ‘a complete studio at your fingertips – with no delays, no
tantrums, no egos’.5 One of the problems may be that designers and their
clients lack a common language.

The de-skilling of printing: the production of typographic displays is no

longer in the hands of a few trained specialists, but available to all.
Technologies that were once complex – typesetting, offset litho, video and
computer displays – are now standard office or even domestic equipment.
Our concept of literacy should be extended to include a wider range of
communication skills, including typography.

The design of communication technologies: the fast growth of the new

communication technologies is involving numerous engineers and
software designers in the design of typographic features and capabilities.
Typographers are often brought in, if at all, only after important decisions
have been taken. For example, Twyman (1982) reports the initial
assumption that only upper case would be necessary on videotex systems. A
conceptual framework might help typographers communicate with systems
engineers, and would make typographic concepts accessible to engineers
who do not have typographers at hand to consult.

Discourse studies: As will emerge from this study, neighbouring disciplines 10

such as linguistics, bibliography and cognitive psychology are starting to
notice more aspects of typography as they converge on the context-sensitive
fields of reader-relations and discourse studies. Typographic scholars can
usefully contribute to this process.

The linguist FW Householder made a half-serious but useful distinction

between ‘hocus-pocus’ and ‘God’s truth’ theories. This study is unashamedly
of the former kind: it does not claim to discover the ‘real’ nature of
typographic phenomena, but suggests structures that may be usefully
applied for the sorts of practical purposes discussed above.

I am painfully aware of the dangers of multi-disciplinary study and my

foolishness in attempting it. I have tried to steer a course between the
naïve positivism of pop psychology and the exaggerated relativism of
what Lakoff & Johnson (1980a) call ‘café phenomenology’. And although
I have sometimes referred to historical examples, I have tried to bear in
mind Eisenstein’s (1979) warning that ‘where historians are prone to be
over-cautious, others are encouraged to be over-bold’ (p. 39).6 I have

5 The quotation is from an advertisement for the ‘Sweet P’ graph plotter which appeared in Byte
magazine for several months during 1984. Computer users will smile wryly at the implication that
computers display none of these characteristics.

6 However, my reading of Eisenstein herself, and others too, leaves me with the impression that many
historians of this subject are not particularly reticent about identifying turning points in civilization – in

Robert Waller • The typographic contribution to language • PhD Thesis 1987 9

Introduction

therefore tried to keep a respectful distance, referring to history only in so

far as the published conclusions of major writers throw light on current
practice through precedent or analogy.

evolution, even: candidates include writing itself (Gelb 1963), alphabetic writing (Havelock 1976),
word separation (Saenger 1982), printing (Eisenstein) and engraving (Ivins 1953).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 10

Chapter 1 • Typography and language: a selective literature review

1
11

Typography and language: a selective literature review

In this chapter I shall review some of the typographic literature to give

an impression of the ‘story so far’. The review is necessarily selective, and
leaves out a great deal of valid and useful information about particular
typographic problems. The purpose is not to list all that is known, but
rather to sketch a background to the theoretical problems addressed in this
study. By ‘typographic literature’ I mean writings that have explicitly set out
to discuss some aspect of typography, rather than writings on other topics
that might conceivably be relevant to typography (although I introduce
quite a number of these in later chapters).

It is appropriate to start by looking at what typographers themselves have

written, before considering the approaches taken by those psychologists and
linguists who have addressed typographic issues.

Typographers on typography
British typography is still heavily influenced by ‘the great and the good’
book typographers of the Anglo-American pre-war tradition, (for example
Rogers 1943, Morison 1951, and Updike 1937). While they contributed a
great deal of intelligent and scholarly writing on typography, it was mostly
of a historical or technical rather than a theoretical character. Above all,
they were concerned with the revival and creation of beautiful and effective
letterforms, and with the design of readable prose, mainly for books but
also newspapers. We should remember that the concept of ‘typography’ as
an abstract entity was relatively new, as was the profession of ‘typographer’.

On the written evidence, it seems that they did not view typography as part 12
of language so much as a channel for its transmission. Beatrice Warde’s
‘crystal goblet’ metaphor (1955) encapsulates the idea of typography as a
transparent vessel adding no colour of its own to the author’s meaning.7
In order that typography could be completely natural and unobtrusive,
letterforms must be perfectly formed and free from mannerisms, lines and
columns of type must be arranged to carry the eye smoothly along free
from distraction or fatigue. This ideal was not original: the best printers
have always been conscious of their responsibility as a channel for clear

7 Models and metaphors for communication are discussed further in Chapter 5.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 11

Chapter 1 • Typography and language: a selective literature review

communication. Perhaps no one has expressed the ideal better than Joseph
Moxon (1683/1962: 211):
‘A good Compositer is ambitious as well to make the meaning of his Author
intelligent to the Reader, as to make his Work shew graceful to the Eye, and
pleasant in Reading.’

While we do not normally associate this school of typography with

‘functionalism’, the crystal goblet tradition established two valuable
characteristics in the evolution of today’s ‘information designers’. Firstly,
the designer’s role is, on the whole, seen as a facilitative one in which the
function of the book as a channel of communication between the author
and reader takes priority over its status as an art-object or as a monument
to its designer.

Secondly, the traditionalists exhibited a finely-tuned sense of appropri-

ateness – alongside legibility, they have a concern for what typographic
researchers were later to call ‘atmosphere value’. Works of literature, in
particular, were best set in a typeface of their period, and conversely ‘If
the subject matter be of a serious or scientific nature the severest style is
the most suitable’ (Rogers 1943). And because he recognized a distinctive
‘eye-catching’ function for publicity material, Stanley Morison was prepared
to break all of the normal rules of elegant typography in his Gollancz book
jackets and other items intended to catch the eye (cf Morison 1928).

Taken literally, pure crystal-goblet typography might lead us to find a 13

‘correct’ way of arranging type and stick to it, whatever the task in hand
– just as some typographers of the Swiss school (discussed below) might
use only a single sans serif throughout their careers. ‘Correct’ might
mean the most ergonomically effective type, or the type most consistent
with the ‘spirit of the age’. Indeed, this was usually the case for the first
four hundred years of printing, when at any one place and time there
was a strictly limited range of ‘normal’ typefaces available to individual
printers. Although taken for granted by present-day typographers, it is
largely a nineteenth and twentieth century opportunity to select the most
appropriate typeface from a range with different historical or expressive
associations. The idea that typography should be in harmony with the genre
of the text was a distinctive concern of these printers and typographers.

Bruce Rogers (1943: 22) expressed it this way:

‘Making an “allusive” format for a book – that is, casting it in the style of the
period of the original text – is in a small way something like planning the stage
setting for a play. An up-to-date style for an ancient text would compare with
staging Hamlet in modern dress. However novel and effective in its own way, you
feel it to be strange, and this sense of strangeness is an annoying distraction; you
are forced to think of the setting and the designer rather than of the text.’

Robert Waller • The typographic contribution to language • PhD Thesis 1987 12

Chapter 1 • Typography and language: a selective literature review

While this might be disputed (it is arguable that modern dress, or modern
(small ‘m’) typography, is more neutral and so less distracting than ‘fancy
dress’) the metaphor is an interesting albeit an old one.8 Rogers does not
elaborate it, but it bears extension: while the script and plot are provided
by the author, who can also control the movement of arguments from
background to foreground and from one episode to another, typography
can contribute the costumes (letterforms, ornaments, symbols, rules, etc)
and the set (the format, layout or grid); typography may also announce or
signal the progression of the plot. It is interesting to see that the ‘staging’
metaphor has recently surfaced in the literature of discourse analysis 14
(Grimes 1975; Clements 1979) to describe the ways in which speakers
and writers use linguistic signalling to ‘place’ parts of a narrative in the
foreground or background.

Figure 1.1 Two ‘allusive’ Bruce Rogers title pages, both reproduced in Paragraphs on printing (Rogers 1943).
Neither the dimensions nor the margins of the originals are indicated, but they are much reduced in size.

To inject a personal note: although this interpretation seems to allow the

typographer considerable scope for making a creative contribution to the
texts for which he or she is responsible, it represents, in part, my own
attempt to be positive about a school of typography that I find somewhat
prefectorial. Morison, in his last work, Politics and script (1972), sees
typographic style in terms of the conferring of authority on a work. Most of
the book is an intricate intertwining of church and typographic history of a
fairly obscure kind, but it concludes with Morison’s view that there is a:

8 The metaphor of language as the dress of thought can be found in Quintilian (Book VIII,introduction,
20).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 13

Chapter 1 • Typography and language: a selective literature review

‘universal consensus upon the principles that should govern the shapes and uses of
the Graeco-Roman alphabet in all works addressed to the intelligence of mankind.’
(p. 339)

The modern guardians of this consensus, which Morison saw as the 15

culmination of ‘a twenty-five-century evolution…under the authority of
emperors, popes, and patriarchs or abbots or masters of guilds’ are the
university presses. The consensus is said to be
‘a reconciliation of authority and freedom…which is the best guarantee against
experiment or innovation or irresponsibility. By irresponsibility is meant any
reduction in the authority proper to the style of permanent literature addressed
to the intelligence of man in favour of the freedom proper to ephemeral matter
addressed to the emotions of man.’ (p. 339)

This exaggerated divide between intelligence and emotion, authority and

freedom may not be representative of anyone’s view other than Morison’s,
but it may nevertheless be symptomatic of what Moran (1978: 14) calls ‘…
an odd world, full of caste and class sympathies’. The context of this remark
is a discussion of the obvious bitterness that marred the relationship of the
Double Crown Club (a select, and male only, dining-club of leading book
typographers) and the largely advertising-oriented British Typographers
Guild.

Quite apart from the risk of excommunication by popes or patriarchs, it

is hard for those trained in this tradition to summon up the courage to be
critical, such is the standard of detailed technical know-how represented by
such as Williamson (1966), and Simon (1945), whose treatment of drama,
poetry and indexes demonstrates immensely good sense and a considerable
virtuosity in the setting of complex matter. However, the problem
remains that standards are mainly passed on in the form of examples to
imitate rather than principles that are adaptable to different technical
circumstances or texts other than novels and poems. And while scholars
and designers of the historical tradition discuss the display of fine literature
for continuous reading in by no means unfunctional terms, they are less
reliable when away from their home ground.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 14

Chapter 1 • Typography and language: a selective literature review

16

Figure 1.2 The telegram form, before and after Morison’s redesign. Reproduced by Moran (1971) 9. Actual size

For example, Stanley Morison’s 1935 redesign of the Post Office telegram
form (Figure 1.2) benefits from fewer words, but otherwise it is not a great
advance on its predecessor: one can imagine his European contemporaries
making a much more adventurous attempt.10

9 Moran attributes the design to Morison. However, Beatrice Warde (1936) also reproduces the form
but, strangely, does not name the designer.

10 A further example is found in Ruari McLean’s Manual of typography (1980), which although
written in the late 1970s draws most of its examples from the 40s and 50s. Having talked sensitively,

Robert Waller • The typographic contribution to language • PhD Thesis 1987 15

Chapter 1 • Typography and language: a selective literature review

At the end of MH Black’s (1961) detailed account of sixteenth- 17

century developments in Bible printing (by Robert Estienne and his
contemporaries), he effectively accuses Bruce Rogers of placing visual form
before function in his ‘allusive’ approach.11 The printers described by Black,
Estienne in particular, did set visual standards that are worth reviving, but
their functional innovations are equally significant. These included chapter
headings, summaries, running heads and versification (see also Black,
1963).12

The age of Morison was one in which typographic opinion-leaders were

considerably more literate, culturally aware and prominent on the
intellectual scene than had previously been the case (or has been since,
probably). But although conscious of the central cultural role of printing
and typography, the fuller implications of that role are often missed. For
example, although Beatrice Warde wrote eloquently of the ‘three great
privileges’ of printing in her foreword to Steinberg’s Five hundred years of
printing (1974) – they are, essentially, the privileges to turn back, to look
forward, and to stop and think – it is not made clear that typography and
layout might be deployed so as to support such activities.13

Modern typography 18
It is usual to contrast this historically inspired, symmetrical style with the
asymmetry of ‘modern typography’ (variously called New, Asymmetric,
Functionalist, International or Modern Typography), whose origins go
back to the Bauhaus and earlier. However, both movements have similar
motivations: to reform the allegedly ‘enfeebled’ typography of the
nineteenth century. Tschichold (1935/1967), who was responsible for the
fullest exposition of the principles of the New Typography, summarized the
different approaches when, after praising William Morris, he said
‘He was right…to go back to the incunabula but wrong in copying their externals
instead of their spirit. They were in their own day a step forward, a bold seizing of

perceptively and at some length about book design (using title-pages rather than text pages as
illustrations), McLean’s discussion of ‘jobbing printing’, a rather archaic and dismissive term, is
extremely short and includes a before-and-after exercise in table design in which the after is more
elegant but rather less effective than the before.

11 Black’s comment reads: ‘The Doves is unreadable, and is plainly a piece of fifteenth-century
revivalism based on a confused analysis. Bruce Rogers is impressive, but this article will have sufficiently
indicated where most of his ideas came from – Estienne. Could it be that a twentieth-century printer
might be moved by such examples to provide a worthy competitor that paid the necessary respect to
tradition without dwindling into eclecticism and pastiche?’ (p. 203)

12 Twyman (1986: 189) comments that ‘oddly enough, specialist historians, whom we might expect
to have been interested in the development of graphic configurations have, almost without exception,
shunned approaches that deal with their specialty in functional terms as a branch of graphic language’.

13 Morison (1936: 1) talks similarly of ‘the inherent courtesy of print: that it can be skipped or
skimmed …’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 16

Chapter 1 • Typography and language: a selective literature review

new opportunities. Morris’s copying of them was a backward step…a shirking of

reality.’ (p. 16)

In the usual ‘fog in channel – continent isolated’ manner, there appears

to have been considerable ignorance in Britain about the graphic arts
revolution happening in Europe. Modern typography harnessed the visual
idioms of new art movements (Futurism, de Stijl, Dada, Constructivism)
to printing, using visual tension, contrast and rhythm to manipulate
the reader’s attention. Laszlo Moholy-Nagy, a major figure in the New
Typography and a teacher at the Bauhaus, explained the method thus:
‘In contrast to the centuries-old static-concentric equilibrium, one seeks today to
produce a dynamic-eccentric equilibrium. In the first case the typographical object
is captured at a glance, with all the centrally focused elements – including the
peripheral ones; in the second case, the eye is led step by step from point to point,
whereby [the awareness of] the mutual relationships of the individual elements
must not suffer.’ (Quoted by Kostelanetz, 1970: 80)

19

Figure 1.3 Herbert Bayer’s 1925 design for a catalogue of Bauhaus products (from Spencer, 1969a).
Original 210mm x 296mm.

While the ‘crystal goblet’ idea relies entirely on the author to make his
or her meaning clear, modern typography was thus prepared to offer the
reader additional support. In Tschichold’s words,
‘modern man must read quickly and exactly. Every effort must be directed to
transferring the words smoothly to the reader. This can be achieved by correct
groupings to express the sense of the words’ (p 46).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 17

Chapter 1 • Typography and language: a selective literature review

Display type was bold and large, text type was clear and simple (often sans
serif), the ‘bullet’ was introduced, thick rules abounded, and ‘white space’
was used to group information. Such features, startlingly new and even 20
shocking at the time, have become absorbed into the general typographic
repertoire.

In their day, such devices made considerable demands on printers and

their equipment: from the technical point of view, functionalist typography
was often rather less functional than the style it replaced. Instead, the
‘functional’ label (originally, at least) indicated a concern for typography
which articulated the meaning or structure of the information, and which
made the printed document function better for the reader.

This functional concern is expressed nowhere better than by Tschichold

(1935/1967). For example, sans serif type was preferred by more formalist
designers because it lacks extraneous or decorative embellishments, but by
Tschichold because ‘its wide range of weights (light, medium, bold, extra
bold, italic) gives every colour in the black-and-white scale’. And while
others are nervous about indented paragraphs because they spoil the neat
blocking of information and strict alignment of everything to the margin,
Tschichold thinks the issue through from the reader’s viewpoint. He points
out that when the last line of a paragraph is a full one, the start of the next
paragraph is hard to see; even if line spaces are used, the reader is confused
when they fall at the page break.14

Like other writings by busy practising typographers, Asymmetric typography

is an uneven mixture of substantial argument and brusquely delivered,
though immensely sound, rules, and of concern for both functional and
technical excellence as well as beauty. But although he has much to say
about the clarity of information as a criterion for good design, elsewhere
Tschichold wrote:
‘We must ask ourselves…whether the result is pleasing, whether we have achieved 21
a balance. Provided the work is all right technically, there is no other criterion for
typographical design.’ (Tschichold 1934/1975: 124)

In this two-fold standard of technical and aesthetic excellence, the

information structuring function seems to have been lost. In the absence
of a critical tradition, it is easy for typographers to lose sight of the needs
and reactions of distant readers and concentrate on the problems that lie
close at hand: the technical problems of the printing process, and their
aesthetic problems as visual artists. This certainly seems to have been the
way that the functionalist ideal developed among the ‘Swiss School’ of

14 Interestingly, a similar point is made by De Vinne (1901:193), who refers to ‘a new school of
typography [which] disapproves of the old-fashioned method of indenting paragraphs’. In the layout of
this thesis I have set the paragraphs full out in order to reserve the use of indention for the considerable
number of quotations. An alternative strategy might have been to reduce the type size of the latter, but
this would have been time-consuming and might have had the effect of diminishing their status.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 18

Chapter 1 • Typography and language: a selective literature review

typographers who emerged in the 1950s as the most thorough exponents

of the modern typography. For example, in their textbook Basic typography,
Ruëgg and Fröhlich (1972) divide their attention between technical matters
(typesetting systems, for example) and aesthetic aspects of visual form
(contrast, proportion, rhythm, colour etc).

Emil Ruder’s Typography (1967; see also the review by Kinross 1984)
provides further examples of this aesthetic revisionism. While warning of
the ‘standing temptation for the typographer to use his type primarily as a
tone of grey and thus to allot it a purely aesthetic and decorative role’, in
another chapter we find him asking:
‘What is the relationship between the colour value and quality and the grey of the
type matter? How do the various tones of grey compare? The proper observation
of these principles is crucial for the beauty of a printed work, and for its formal
and functional qualities.’

Functional Swiss typography of the most tightly-argued and informed

kind is represented by Karl Gerstner (1959, 1974). His concept of ‘integral
typography’ reaffirms the ideals of Tschichold:
‘a marriage of language and type resulting in a new unity, a superior whole. Text
and typography are not so much two consecutive processes on different levels as
interpenetrating elements.’ (Gerstner 1959: 66)

Gerstner’s Compendium for literates (1974) represents an attempt to make 22

the connection, and it is something of a tour de force of systematic analysis.
Indeed, it is so systematic that where most of us would be content to list the
theoretical possibilities of writing as ‘embossed’, ‘punched out’ and so on,
Gerstner’s printer is obliged to actually emboss and punch out examples.
Among the rather tedious but magnificently thorough exemplification (large
writing, small writing, upside-down writing, black writing, white writing…
and so on) are well chosen references to major figures in other disciplines
– including linguistics and psychology – and a systematic view of the
expressive properties of type.

Gerstner’s Compendium deals mostly with the display of words and

sentences, but perhaps the most influential and lasting contribution of Swiss
typography is the grid system of page layout. From the earliest days, most
books have been designed with standard text areas and margins – simple
grids. But Swiss typographers, responding to the needs of technical and
information publishing in a multi-lingual country, developed the grid as a
system for complex multi-column page layout. Their distinctive contribution
was not the multi-column layout, also common in medieval manuscripts
and modern newspapers, but the addition of horizontal grid lines, resulting
in a modular layout system. The immediate practical problem this solved
was the parallel display of different languages. The three main languages
of Switzerland (German, French and Italian) could hang side by side from

Robert Waller • The typographic contribution to language • PhD Thesis 1987 19

Chapter 1 • Typography and language: a selective literature review

a horizontal line; although of different lengths the alignment could be

restored at the next heading down (Figure 1.4). Such pages have some of
the characteristics of tables.

The grid system was also found useful for the design of illustrated books
and magazines, whether multi-lingual or not. Its modularity restricted
the number of possible column widths and simplified the specification of
type. More controversially, though, grids were found to solve aesthetic
‘problems’. By following the modular principle, a visual unity could be
imposed on complex material. Illustrations could be restricted to certain
shapes; tiresome bits and pieces, such as folios, captions, headings and
even paragraph indention, could be aligned on the grid. Eventually the grid 23
achieved a cult status, which it still enjoys in some circles.

Figure 1.4 A typical three-language, three-column page for which the Swiss grid system is ideal. Source: Rüegg &
Fröhlich 1972. Dimensions: 250mm x 250mm.

An example of the degree to which good design was equated with

adherence to the inviolate grid can be identified in a paper by Bonsiepe
(1968). He proposed an index of the orderliness of a page, analogous to
those used to predict the readability of prose. The main measures were
to be the numbers of horizontal and vertical points at which components

Robert Waller • The typographic contribution to language • PhD Thesis 1987 20

Chapter 1 • Typography and language: a selective literature review

were aligned. But whereas readability formulas are routinely subjected to

extremely thorough empirical testing, in which the index scores are related
to the performance or preferences of users,15 the grid principle seems to
have been so self-evidently right that Bonsiepe apparently saw no need to
justify his proposal further.

Bonsiepe’s mistake is to confuse orderliness with tidiness. I discovered

the difference as a child. If my clothes and toys were scattered around
the bedroom they could be located easily but the room was deemed to be
untidy. On the other hand, if they were stuffed into drawers or arranged
in straight lines, the room passed my mother’s inspection but I could never
find anything. In effect, Bonsiepe is insisting that all the information on a
page is fitted into equal size boxes.

This high value placed on visual tidiness is symptomatic of a more general

minimalism, in which simplicity of materials and form is valued for its own
sake. At its best it can result in sober typography of the highest integrity, but
its adherents sometimes make the mistake of assuming that all typographic
contributions to a text are content-free adornments.16 In the wrong hands,
grid typography tempts designers to seek visual alignments wherever
possible, regardless of a real connection existing between the things
aligned. When coupled with a minimalist style that restricts the number of
type variations, grids have sometimes resulted in extremely cryptic layouts
that are hard work to interpret. The problem is that by the time the reader
encounters the page, the designer’s grid has disappeared. The reader must
impute imaginary grid lines to the page and use them to make decisions
about the relationships intended by the designer.

In spite of such problems – and every system can be abused – grid

typography is too important to dismiss. It has become absorbed into
every designer’s working methods and so must be an essential part of 25
any systematic overview of typography. It should also be said that most
typographers qualify their advocacy of grids carefully (eg Hurlburt 1978).

15 Although subject to recent criticism, and consequently somewhat out of fashion now, readability
formulas can still provide a reasonably accurate index of the difficulty of prose, but can neither pinpoint
specific problems nor be used as a prescriptive guide for writers. Klare (1984) has recently published a
comprehensive and intelligent account of their history, use and current status.

16 Minimalism is not the exclusive preserve of the New Typography. The American printer Theodore De
Vinne (1901) appears to have thought the printer’s job would be much simpler if authors could rely on
words alone: ‘The desire to make written language clear to the reader is to be respected, but some of the
methods now in general use are unsatisfactory and will not stand critical examination. A hundred years
ago it was the duty of the printer to begin every noun with a capital letter and to compose in italic every
word that needed or seemed to need emphasis. It was hoped that capitals and italic would help the
reader to a better comprehension of the subject … Experience has proved that readers do not need these
crutches, and that ordinary matter can be made readable and intelligible without them. It is probable
that the next generation will put greater restrictions on the use of quotation-marks…’

This evolutionary view that readers can learn to handle ever simpler arrangements, which anticipates
the attempted lower-case-only alphabet reform of the Bauhaus school (Spencer 1969a,1969b), is also
reflected in De Vinne’s discussion of hyphenation (see Chapter 7).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 21

Chapter 1 • Typography and language: a selective literature review

Moreover, it is a thoroughly two-dimensional and typographic idea which

is impossible to integrate in a theoretical framework derived only from a
(linear) linguistic model. Grids are discussed further in Chapter 7 where
they are seen in the context of other ‘artefact structures’ in text.

In summary, I have briefly introduced three related but distinct strands

among twentieth-century typographic writers: related in their effort to
identify a role for the professional designer and establish high standards
for the design of printed matter; distinct in their systems of visual logic.
Stripped of their various aesthetic dogmas, all three seem to make good
sense as constituents of a typographic criticism. The historical revival
contributes a sensitivity to genre, the New Typography introduces an
enlarged repertoire of graphic techniques with which to display the
structure of a message, and Swiss typography contributes the grid system
and its emphasis on the flexible and ordered page.

It is not a realistic or desirable aim for a typographic theory, in the manner

of spelling reformers, to seek to replace existing standards with some
revolutionary new system that claims to be uniquely rational.17 Instead,
the model that will be presented in Chapter 5, together with the discussion
of typographic genres in Chapter 9, can be seen as an articulation of the
eclectic blend of these three schools of typography that characterizes
current practice.

Gestalt theory
Many of the techniques of the New Typography (and subsequently the Swiss
school) owe something to the contemporary Gestalt psychologists, who
were working in Germany in the 1920s and 1930s (for example, Koffka
1935). Although the connection may not be always explicitly declared (at 26
least in those publications most easily available in the English language)
the influence is undoubtedly there. Gerstner, for example, does not cite
Gestalt psychology in his Compendium for literates, although he frequently
uses Gestaltist terms, like closure, figure and ground. Rivlin (1987),
who has recently published a detailed application of Gestalt principles
to typography, appears to have found few prior publications making the
explicit connection.

Gestalt theory is best summed up by the catch-phrase often associated with

it: ‘the whole is greater than the sum of its parts’. Gestalt psychologists
identified a range of principles which together provide evidence of an
overall Law of Prägnanz, which, to avoid the theoretical commitment
implied by that term, we might call the simplicity principle. It is a natural
instance of Occam’s razor: if several alternative structures are possible, the

17 As Venezky (1970) has shown, English spelling is considerably more rational than is usually thought.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 22

Chapter 1 • Typography and language: a selective literature review

simplest and most stable will be selected. Figures 1.5 to 1.7 illustrate the
three principles most relevant to typography.

a b c

Figure 1.5 Grouping by proximity

Proximity: Things which are close together are seen as groups. While we see
Figure 1.5a as a matrix of equally spaced dots, Figures 1.5b and 1.5c are
visually organised into rows and columns. When typographers use space to
group components, they employ the proximity principle.
27

Figure 1.6 Grouping by similarity Figure 1.7 The closure principle:

it is possible to see this as a square

Similarity: In Figure 1.6, it is the similarity of elements that creates the

grouping, not the use of space. When typographers use a consistent
typeface to signal a particular kind of, say, heading, they are grouping by
similarity.

Closure: Figure 1.7 demonstrates the tendency to ‘close’ gaps between

graphic elements and see stable shapes wherever possible. Typographic
grids are based on the closure principle, together with the similar ‘good
continuation’ principle.

The Gestaltists’ observations and demonstrations of how we perceive

visual structure are too compelling for easy denial but not easily explained
psychologically; their explanation that perceptual principles correspond
to in-built ‘brain field forces’ has given way to more verifiable cognitive
explanations (Marr 1982).18 The basic principles of Gestalt psychology
– or, at least, the observations on which they are based – have become

18 Marr’s explanation is that the Gestalt illusions correspond to characteristics of normal physical
objects and thus they are learned, not innate: many natural objects are symmetrical, have smooth
contours, contrast in various ways with their background and so on. Bruce and Green (1985) provide an
up-to-date review of current perceptual theory.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 23

Chapter 1 • Typography and language: a selective literature review

absorbed into the art school curriculum and, now largely divorced from
their theoretical origins, form a basic part of the designer’s craft knowledge.
They might be seen as relatively inflexible perceptual rules that act as a
fundamental constraint for the typographer alongside such conventional
rules as the left-to-right direction of the writing system.19

Figure 1.8 illustrates some gestalt problems in an illustrated book.20 28

Figure 1.8 This page from The pocket camera handbook by Michael Langford (Ebury Press) shows some of the
perceptual rules that typographers must anticipate. For example, because the small illustration at the bottom
right (‘Batteries’) has been aligned with stages 3 and 4 of ‘Loading the film’, we tend to see it as part of the same
sequence (proximity/good continuation principles). The vertical rule between is too weak to counteract the effect.

19 Rivlin (1987) has argued cogently for a central role for Gestalt principles in typographic theory,
but he may be pushing on an open door – the standard Gestalt demonstrations are too convincing to
deny, and applications to typography are not hard to identify. The real problem, surely, is to relate these
two-dimensional graphic phenomena to language and communication.

20 Hereafter I shall use an anglicized version of the word ‘gestalt’. Thus it is not capitalized and the
plural is ‘gestalts’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 24

Chapter 1 • Typography and language: a selective literature review

Applied psychologists and typographic research

Over the years a great many studies have been published by psychologists
(of various specialisms) who have examined the effect of typography on
readers. Educational theorists (for example, Rowntree 1982) traditionally
divide the outcomes of education into three ‘domains’: the psycho-motor
(physical skills), the affective (aesthetic and moral appreciation) and
the cognitive (intellectual skills). The distinction provides a convenient 29
framework for discussing the applied psychology of typography. Literacy
involves the attainment of skills in all three domains, and all three have
been addressed by typographic researchers using the methodologies of
applied psychology.

Legibility: the psycho-motor domain

Early typographic research (reviewed by Pyke 1926) was closely integrated
with the more general investigation of the reading process. While Pyke
lists instances of eighteenth- and early nineteenth-century work (eg, by
Babbage in 1827), Javal (1878) is generally credited as the first to apply
the scientific method to typography, and a considerable number of studies
of ‘reading hygiene’, as the field was then called, were published in the first
half of this century. The typographic variables listed by Legros (1922) typify
the scope of much of the legibility research that still appears from time to
time today (Table 1.1).

Size of character
Thickness of strokes
White space between strokes
Dissimilarity of characters
Leading
Line length
Frequency of kerns
Similarity of figures
Width of figures
Separation of lines from adjacent matter
Unnecessary marks in or near characters
Vulgar fractions
Variations in type height
Quality of paper
Colour of paper
Light-reflectance of paper
Colour of ink
Illumination
Irradiation

Table 1.1. Typographic variables listed by Legros (1922).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 25

Chapter 1 • Typography and language: a selective literature review

It is interesting that typography was regarded as just one contribution

to reading hygiene, alongside such things as lighting, paper colour,
reflectance, the angle and curvature of the page and even posture. With the
introduction of electronic displays, similar factors have again become the
focus of research attention.

The most prolific legibility researcher was Miles Tinker of the University
of Minnesota, who with his colleague Donald Paterson21 published several
dozen legibility experiments between 1929 and the publication of his books
The legibility of print (1963) and Bases for effective reading (1965), now
standard sources. Reactions to Tinker differ, generally between those with
practical experience in printing or typography and those without. Among
the latter, Tinker’s research is still widely cited.

A number of general criticisms of legibility research, typical of those with

first-hand knowledge of typography, were first voiced by Buckingham
(1931) – although his own experiment was fairly unconvincing.
In particular he criticizes the univariate research model, in which
experimenters try to vary a single factor while holding all others constant.
Buckingham comments:
‘This is good experimental technique. It is an article of faith among investigators.
Yet it won’t work in the way it has been applied to typography unless one is
prepared to go to very unusual lengths with it.’ (p. 104)

He goes on to note that (mostly paraphrased):

• ‘several of those who have given out standards have simply used their
imagination’ (that is, the recommendations do not always relate to the
data).

• typographic variables interact: recommendations about line length,

for example, ‘are valid only for the interlinear spacing employed, and
the investigators do not tell us what that is. Widen the spacing and the
probability is that a longer line may be employed to advantage.’

• investigators often refer to, say, ‘10pt type’ without reporting the 31
typeface used or the interline space.22

• printers perceive the investigators’ ignorance of typographical matters

and ignore the results anyway.23

21 Paterson & Tinker (1940)

22 Although, according to Spencer (1969b), the need to measure the visual rather than the body size of
type was first pointed out by German researchers in 1903, this single fault mars much of Tinker’s work
and that of his contemporaries.

23 Later critics suggested that psychologists should take advice from designers in formulating their
hypotheses (Spencer 1969b; Burnhill & Hartley 1975; Macdonald-Ross & Waller 1975)

Robert Waller • The typographic contribution to language • PhD Thesis 1987 26

Chapter 1 • Typography and language: a selective literature review

• to do a full study of even a modest range of typefaces, sizes, line lengths

and line spacings would require more effort than anyone is prepared to
put in (he outlines a simple study that would have required 1,792,000
returns).

In addition to Buckingham’s criticisms, others have noted that:

• technical research papers are ignored because they are difficult for
printers and designers to understand (Rehe 1974). This may be a rather
more patronizing version of Buckingham’s similar point. Rehe’s own
book is itself clearly written, although somewhat uncritical. Spencer’s
(1969b) review is a model of both clarity and discrimination.

• ‘the classical research literature in this field has concerned themselves

with molecular issues (ie with tiny details) rather than with molar ones
(ie broad scale issues).’ (Hartley & Burnhill 1977a: 223)

• the research tends to be ‘divorced from the questions which are actually
asked by practitioners when a choice of typeface has to be made.’
(Hartley and Burnhill 1977a: 224). Designers would like more details of
the performance characteristics of individual typefaces: for example, can
they be reduced or photocopied?

In spite of the problems, Tinker and others are still frequently cited where 32
scientific evidence is thought necessary to make design recommendations
more convincing. And there are those who regret the passing of this style of
research. Rehe (1974) concludes his review of legibility research by calling
for more of the same:
‘Univariate research, that is, investigation of individual typographic variables,
should be increased and broadened. These individual research findings are the
particles of the mosaic that make for better legibility.’ (p. 61)

Scientific evidence is also attractive to those who apparently regard all

intuitive judgements, whether by novices or experienced practitioners, as
equally unreliable. In this regard we may recall Tinker’s comment that
‘Before scientific research, printers and type designers were concerned mainly with
the esthetic appearance of the printed page. This preoccupation with esthetics,
together with considerations of economy and tradition, dominated all typography
until about 1920. As a result of these obstructive emphases, a scientific typography
has been slow in developing. Indeed, the printing industry continues to resist
procedural changes suggested by experimental findings.’ (Tinker 1965: 115)

Elsewhere Tinker does acknowledge that, on the whole, printers often

make good decisions without the benefit of research. Together with the
fact that many experiments reveal only small differences, if any, this
has led most typographic researchers to the opinion that it is not worth
investing in traditional legibility research. Since the late sixties, research
on simple matters of legibility has tended to be undertaken only in special

Robert Waller • The typographic contribution to language • PhD Thesis 1987 27

Chapter 1 • Typography and language: a selective literature review

circumstances. New display technologies are of obvious interest and

ergonomists continue to publish numerous studies of the ‘human factors’
of CRT displays (eg, Reynolds 1979; Bouma 1980; Shurtleff 1980). In
addition, new type designs and page layouts are sometimes evaluated
by their designers without the results being published. Foster (1980) has
reviewed recent legibility research.

Herbert Spencer, Linda Reynolds and other colleagues formed the 33

Readability of Print Research Unit at the Royal College of Art from which
a number of publications on legibility emerged in the late sixties and
seventies. With the exception of Sir Cyril Burt (Burt, Cooper & Martin 1955;
Burt 1959/1974), who consulted with Stanley Morison, Beatrice Warde and
other leading typographic pundits of the day, the RCA team was perhaps the
first to combine the skills of psychologists and designers, thus overcoming
at least one of the criticisms of the earlier research. Although initially the
emphasis was on legibility, they also looked at aspects of typographic and
spatial signalling – for example, the layout of bibliographies (Spencer,
Reynolds & Coe 1975). These studies had relatively modest and realistic
goals. Essentially they were comparisons of a range of solutions to easily
identified and frequently recurring psycho-motor problems of scanning or
searching. Searching for a name in an index or bibliography, for example,
is an easily-defined and common task. It is therefore valid to apply the
findings directly to practical situations.

Developments at the RCA were paralleled by another prolific psychologist-

typographer team, James Hartley and Peter Burnhill, who similarly moved
from legibility research to structured information, including the design of
academic journals (Hartley, Burnhill & Fraser 1974), textbooks (Burnhill
and Hartley 1975), indexes (Burnhill, Hartley & Davies 1977), and
bibliographic references (Hartley, Trueman and Burnhill 1979). Where
other researchers were usually content to investigate simple issues and
report the data, Hartley and Burnhill proposed a conceptual framework
for their own work which, since it goes beyond psychological issues, I shall
return to shortly.

Atmosphere value: the affective domain

I have noted that typographers are often aware of the expressive properties
of the typefaces they use. Following the lead of Berliner (1920), a number 34
of psychologists have enquired whether this awareness is shared by
readers. Early studies required subjects to choose typefaces appropriate
for particular products (hers were fish, pancake flour, pork and beans,
and marmalade). One of most thorough series of studies of this kind
was reported by Ovink (1938) whose subjects rated the suitability of
typefaces for different text topics (literary styles, ideas, and commodities).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 28

Chapter 1 • Typography and language: a selective literature review

Unfortunately, as with many typographic studies, his results were obtained

with typefaces that are now mostly obsolete. Ovink does, however, provide
a means for the ‘translation’ of his results to other typefaces, since he
analyses their characteristics using a scheme originally designed to describe
personality differences in handwriting. The same scheme could presumably
be applied to modern typefaces to identify equivalents.

More recent studies (reviewed by Rowe 1982) attempt to overcome this

problem by using the semantic differential technique, considered more
respectable by modern psychologists than handwriting analysis. Typefaces
are related to topics indirectly, via general dimensions such as ‘hard/soft’,
‘active/passive’ and so on. The suggestion is that a typeface with particular
qualities could be used to imbue a message with the same qualities. Walker,
Smith and Livingstone (1986) have also published data demonstrating that
typefaces considered by subjects to be suitable for advertising different
professions, turn out to have similar connotations to those professions when
tested separately.

Zachrisson (1965) has noted about his own and other studies of
atmosphere that researchers have failed to take account of the artistic or
literary education of subjects: that is, their ability to discriminate between
typefaces which, in the case of book faces, can look very similar to the lay
person. Moreover, descriptive terms thought up by experimenters may not
be meaningful or relevant to subjects. Bartram (1982) tried to overcome
this last objection by eliciting descriptive dimensions from subjects
themselves. His purpose was also to provide designers with a means to
test their intuitions against the perceptions of their audience (following
Sless 1980).24 He therefore supplied a procedure and a simple statistical
technique for designers to conduct their own research when necessary. This 35
goes some way towards meeting an objection raised by Spencer (1969b:
29):
‘a review of press advertisements, in which typographic allusion is often a
vital ingredient, published over the last half century suggests that findings on
congeniality may have little temporal stability’.

A reasonable assessment of this work is that, while studies of atmosphere

value do not provide direct guidance about typeface choice, as some
authors claim, they do substantiate the common sense view that
typographic style is noticed by readers and that their interpretations

24 Sless devised an exercise to encourage graphic design students to be more objective. Each had to
make a random ink-blot and identify an object that it resembled. Having done so, they asked others
to identify it also. Inevitably there was considerable disagreement. Students then had to make the
minimum modification that they considered necessary to make the ink-blot look unmistakably like the
intended object, before testing it again. The cycle was repeated until the image was reliably identified by
all observers. Some students found it very hard to accept that others could not interpret the image in the
same way as themselves – it usually takes several cycles more than the student believes possible at the
outset. The exercise has become a regular first-year project in some art schools.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 29

Chapter 1 • Typography and language: a selective literature review

are not random. Although there is some disputed evidence that reader
preferences affect reading speed (Burt, Cooper & Martin 1955),25 it is
reasonable to suppose that anything about a text which is discernible to
readers may affect their perception of the status of a document and thus
their expectations, critical stance, reading strategies, goals and outcomes.
It is hard to see applied psychologists going much beyond the present
findings. Laboratory-style experimentation is rather a clumsy instrument to
probe subtle issues – for example, how texts, through their use of stylistic
nuances, may be seen to be ‘quoting’ other texts.

Typographic cuing: the cognitive domain

Some psychologists have looked at the effect of ‘typographic cuing’ on
learning (reviewed by Glynn, Britton & Tillman, 1985). The term generally 36
refers to the use of typography (bold or italic type, or underlining) to signal
the important ideas in a text. In most studies, importance is assessed not
by the author of the prose passage used, but by the experimenter or a
group of independent judges. It is therefore a separate system of signalling
overlaid onto the signalling already implicit in the author’s prose structure.
In this respect typographic cuing is similar to other devices, sometimes
known as ‘adjunct aids’, proposed and tested by educational researchers.
These include advance organizers (Ausubel 1963), behavioural objectives
(cf Davies 1976), and inserted questions (Rothkopf 1970), although these
devices are more genuinely rooted in pedagogical theories.

There is little doubt that cuing does work in drawing attention to the cued
material. The consensus is that people are more likely to remember cued
ideas. Some researchers, though, (for example, Glynn & Di Vesta 1979)
have found that this is achieved at the expense of uncued ideas. It should
also be noted that most studies of typographic cuing improve immediate
recall, but do not improve delayed recall. Quite apart from methodological
objections raised by Hartley, Bartlett & Branthwaite (1980),26 these
conclusions are not altogether unexpected, since the cuing is effectively
giving subjects the answer to the recall questions beforehand. Indeed,
Coles & Foster (1975: 105) suggest that the failure of typographic cuing
to improve test scores in the first part of their own study might have been
because

25 The evidence is disputed, in part, because of the general discrediting of Burt – a leading psychologist
of his day – who is alleged to have ‘cooked’ the data from his experiments on intelligence, and ‘invented’
co-workers. Hartley & Rooum (1983) have re-examined Burt’s typographic work in the light of this
scandal and expressed doubts.

26 Hartley, Bartlett & Branthwaite (1980) criticized the experimental rigour of some of the studies they
reviewed, in particular the failure of some researchers to test the comprehension of uncued as well as
cued items, and the failure to equate the time taken by subjects in the experimental and control groups.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 30

Chapter 1 • Typography and language: a selective literature review

‘not having been informed that cued material would subsequently be tested, the
students may have found cueing confusing or even distracting rather than helpful’.

Some studies have tested more innovative and unusual typographic

formats. A number of these were published in a special issue of Visible
Language by Hartley & Burnhill (1981). Since they followed the admirable 37
practice of that journal in requiring contributors to practice what they
preach, readers may judge the effectiveness of the new formats for
themselves.

Figure 1.9 A page from Jewett (1981) illustrating his proposal for ‘multi-level writing’.
Dimensions: 228mm x 152mm

Jewett (1981) uses different levels of indention to indicate hierarchical

levels of argument (Figure 1.9). His article is printed sideways, presumably
to allow the generous indention that he uses for his three levels of
importance. There seems no reason why it should not have been printed
conventionally, though, since the resulting line length is excessively long:
ironically, this hinders quick scanning, although the hierarchical system is
meant to facilitate it. Furthermore, it is not possible, as might be thought,
to scan the article while ignoring lower levels of the hierarchy. Higher
level paragraphs sometimes make reference to information contained in
the lower level ones they follow. Although Jewett claims that his format
makes writing quicker by absolving the writer from the responsibility of
verbalizing the hierarchical structure, it seems to have been impossible to
shake off the habit.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 31

Chapter 1 • Typography and language: a selective literature review

38

Figure 1.10 A page from Shebilske & Rotondo (1981) showing their use of typographic cues.
Dimensions: 152mm x 228mm.

Shebilske & Rotondo (1981) distinguish between three kinds of ‘content’:

in addition to uncued text, bold type indicates ‘important’ ideas, and square
brackets indicate the ‘gist’ of each idea. For this reader at least, Shebilske
& Rotondo’s article proved almost impossible to read in the sense in which
it was intended (Figure 1.10). While their bold type (the gist of important
ideas) corresponds roughly to its conventional usage, the use of parentheses
was a major obstacle. Their use of parentheses to signal the gist of an
idea is directly counter to their normal meaning, which is to interpolate 39
unimportant (parenthetical, in fact) material. Moreover, we normally think
of the ‘gist’ as the essence of an idea that particular explanations allude
to; we may expect to see it eventually realized in the form of a summary
or catch-phrase, but we do not expect to see each sentence or paragraph
contain a kernel of words which can be set apart typographically as the
‘gist’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 32

Chapter 1 • Typography and language: a selective literature review

Although Shebilske & Rotondo substantiated their proposals with improved

comprehension and favourable reader reactions, they did not directly
compare their rather complex system with simple cuing: readers might
have been only paying attention to the capitalized ideas. In fact, the
favourable reader comments quoted in the report seem to be referring to
the highlighting of important ideas rather than the bracketing of gists. It is
possible that subjects simply ignored the brackets. Indeed, an earlier study
by Hershberger & Terry (1965) did compare various levels of complexity for
cuing (up to five levels of importance were cued), and found no advantage
in distinguishing between more than two levels.

Researchers working within this tradition see themselves as extending a line

of inquiry initiated by Klare, Mabry & Gustafson (1955) and Hershberger
& Terry (1965). Writers and researchers outside educational psychology
circles are not cited and probably not known about. Apart from this
insularity, a view of text as simply containing gist and some unimportant
stuff between, is somewhat unsubtle; as is a view of reading in which the
task of the reader is limited to rote recall – to remember and repeat certain
instructor-designated ideas.

These studies of innovative typographic cuing reflect two wider trends

in the typographic literature. Firstly, researchers sometimes give an
unfortunate impression of naïvety, both typographically and linguistically.
Special functions are assigned to devices such as indention, bold type,
line spaces, and parentheses as if they have no pre-existing function. Also
ignored is the rich and diverse system of linguistic signalling which can
be used by skilled readers to perceive the author’s deployment of ideas.
Secondly, they exemplify a tendency to want to reform a system which 40
is seen as fundamentally irrational. With the exception of historians
describing past practice, comparatively few people have attempted simple
descriptions of typographic systems without prescriptive overtones. The
reformist tendency is seen most clearly in studies of English spelling (cf
Venezky 1970), suggestions to change the direction of writing or to present
words in visual stacks (Huey 1898, Andrews 1949)27 and in attempts to
design phonetic alphabets or simplify the existing one (cf Spencer 1969b).

Hartley and Burnhill

Since Tinker’s retirement, the most active and widely-cited producers of
psychological research on typography have been James Hartley and Peter
Burnhill. Partly in response to criticisms levelled at Tinker’s generation
– that psychologists should collaborate with designers – Hartley, a
psychologist, and Burnhill, a teacher of typography, have pooled their skills

27 There have been a considerable number of experiments. Huey’s is the first I have found; Andrews
prompted a rash of studies in the 1950s, discussed further in Chapter 7.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 33

Chapter 1 • Typography and language: a selective literature review

to produce research that addresses the realities of typographic decision-

making. While their earlier publications addressed ‘traditional’ legibility
issues, later papers have tackled a wider range of typographic issues. I have
already cited some of their work on structured information. In addition,
Hartley has published a number of papers on other aspects of instructional
text and the reading process.

Some of the earlier research on ‘reading hygiene’ is conceptually rather

barren. It has been briefly reviewed here because it forms a considerable
part of the typographic literature and because of the methodological issues
that emerged. Hartley & Burnhill, though, command special attention
because, unlike many of their colleagues and predecessors, they set out a
coherent framework for their empirical research. Hartley and Burnhill claim
justifiably that most earlier researchers appear not to have felt the need for
any theoretical underpinning:
‘…we would maintain [that] most typographical research has no theoretical base; 41
that is, experimental work has been conducted without reference to a coherent
view of the principles entailed in typographical decision-making.’ (Hartley &
Burnhill, 1977a: 224)

They go on to explain the principles of typographic decision-making on

which their own research is based. These principles have frequently been
restated in a number of books and articles, are widely cited and so deserve
critical attention. The three main principles are:

The use of standard page-sizes. Hartley & Burnhill are strong advocates of
international standard paper sizes, and all their experimental materials are
printed on A4 paper. Indeed, they argue that:
‘recognition of standard page-sizes by research workers is a necessary condition
for further development in the design and evaluation of instructional materials.’
(Hartley & Burnhill 1977a: 227)

On the face of it, the recommendation of a standard page size for research
is a curious requirement. The same overall pattern may be created with,
say 8pt type on an A5 page as with 12pt type on an A4 page, yet there
is no suggestion that type size, line spacing or margins should also be
standardized. Moreover, it is not clear whether we are discouraged from
applying results obtained with one page size to another. Nevertheless, it
is perceptive and entirely reasonable to argue that size is a fundamental
constraint on what may be displayed on a page using space to structure
information, and I will return to this issue in Chapter 7.

The use of grids for pre-planning of pages. A second principle is the use of
typographic grids to ensure that space is used consistently and that the
printed page can ‘provide a reliable frame of reference from which the
learner can move away and to which he can return without confusion’

Robert Waller • The typographic contribution to language • PhD Thesis 1987 34

Chapter 1 • Typography and language: a selective literature review

(Hartley & Burnhill 1977a: 233). Grids are intended to improve upon
traditional practice in which it is claimed that:
‘the absence of consistency in the positioning of related parts indicates that layout
decisions have been made during the process of physically assembling the image 42
(type and illustrations) prior to the process of its multiplication by printing’
(Hartley & Burnhill 1977a: 228; their emphasis)

The consistent use of space to convey information structure. This principle

follows on from the last. It states that the hierarchical structure found in
much information printing can be conveyed by the systematic use of space.
Thus a single line might separate paragraphs, two lines a subsection, and
four lines a new section.28 Various corollaries follow from this principle: the
practice of centring headings is strongly condemned; and the excessive use
of indention is discouraged, as is an excessive variety of sizes, styles and
weights of typeface for headings.

The use of typography to convey information structure must be an

unobjectionable, indeed valuable aspect of typographic theory. Consistency,
too, is an uncontroversial and traditional aim. But the insistence on the
use of space rather than typographic signalling is harder to justify. While
it corrects the amateur’s tendency to overcomplexity, elevated to a general
principle it seems to represent an artificial restriction of a writing system
which has developed in response to functional requirements over many
years. Moreover, no psychological reason is given for this restriction: no
explanation in terms of perception, cognitive processes, working memory or
any other psychological model.

28 This is somewhat reminiscent of elocutionary theories of punctuation, in which different

punctuation marks represent pauses of different length. Lowth (1775/1842: 47) put it this way: ‘The
Period is a pause in quantity or duration double of the Colon; the Colon is double of the Semicolon;
and the Semicolon is double of the Comma. So they are in the same proportion to one another as the
Semibref, the Minim, the Crotchet, and the Quaver, in Music.’ We may also recall the obsolete practice
of following different levels of punctuation by varying amounts of space, reported by Walker (1983)
from her survey of early typing manuals (a comma is followed by one word-space, a colon by two, and
a full stop by three). The difference is that the early typists were using space as a redundant signal,
reinforcing the graphically distinctive punctuation mark; Hartley recommends the sole use of space.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 35

Chapter 1 • Typography and language: a selective literature review

43

Figure 1.11 Demonstration of the principle of using space to articulate content structure. The right-hand version
uses space to group the main stages of the task. Source: Hartley & Burnhill (1977a).

The small scale demonstration in Figure 1.11 is convincing enough, but

its application in a large scale complex text, such as Hartley’s own book
Designing instructional text (1985) is more questionable. In a single page
demonstration, all the space that surrounds a text unit is visible – the
space before and after a section thus gives it shape and distinguishes it
from other units in the page or hierarchy. But when interrupted by one or
more page breaks, this effect is diluted and often lost altogether. Rather
than contributing to a gestalt effect working at a relatively subconscious
perceptual level, the space must be consciously interpreted by the reader,
who cannot be expected to locate the last or next space of equivalent status
in order to complete the gestalt.

This aspect of Hartley & Burnhill’s principles reflects a style of minimalist

typography that is firmly in the Bauhaus and Swiss tradition. Although
they do not share the Swiss pursuit of elegant visual form for its own sake,
similar arguments have been reworked to emphasize Hartley and Burnhill’s
concern for communication clarity. However, whereas printing technology 44
was relatively stable when they started their research in the early
1970s, modern page make-up terminals, and even the cheaper desk-top
publishing systems, allow all sorts of facilities which could previously
be viewed as unfunctional or uneconomic because of their technical
difficulty. Presumably in response to anticipated customer demand,

Robert Waller • The typographic contribution to language • PhD Thesis 1987 36

Chapter 1 • Typography and language: a selective literature review

printing machinery manufacturers have devoted great ingenuity to making

‘irrational’ practices easy and economical.

The disappearance of so many technical constraints does not remove the

arguments for a functional approach. Rather it changes its emphasis, since
typographic decision-making has more than one facet. Not only must the
originators of a printed message make decisions about the placement of
components on a page, but so must the users make decisions on the basis
of the resulting layout – decisions about the order in which to read the
page, decisions about the nature, topic and genre of the text. Designers also
need theories about this kind of typographic decision-making. Their own
decision-making structure will be comparatively clear to them, enforced
by very obvious constraints of equipment, budget and so on. The rationale
with which their readers will approach the text is rather less obvious.

Hartley & Burnhill’s achievement is to have moved typographic research,

as practised by applied psychologists, from the mundanities of ‘reading
hygiene’ towards the altogether trickier area of semantics. Here, the
issue is how the appearance of printed material affects not just how much
is understood, or how fast, but what is understood from it. However,
questions like this cannot be answered in a vacuum. Unless we can describe
the characteristics of a typographic display within a fairly standardized
descriptive framework, we cannot generalize from results obtained with
it. To generalize from an applied psychologist’s experiment to a problem
in hand, we need to know what the two situations have in common. But
whereas psychologists can experiment with sentence comprehension secure
in the knowledge that the concepts such as ‘sentence’ and ‘verb’ will be 45
generally understood (if not agreed upon by all linguistic scientists), no
such agreement exists about variations in page layout.

The ‘language element’ in graphic communication

Twyman (1982) has directly addressed this relationship between
typography and ‘content’. Using a similar example to Hartley’s (a structured
list rather than a set of instructions), he goes on to illustrate the effect of
different technologies on the ways that graphic structuring is realized in
practice.

A particularly telling example is reproduced in Figure 1.12, which shows

a printed version of a manuscript original, both dating from around 1473.
Where the manuscript, with its easily adjusted character widths, is able to 46
achieve a justified right-hand edge within a two-column layout, the printer
needs a wider single column. Again (and closer to Hartley’s concerns),
whereas the manuscript signals a major text division using coloured ink, it
was less convenient for the printer, who instead introduced extra space.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 37

Chapter 1 • Typography and language: a selective literature review

Figure 1.12 The same work in manuscript (left) and printed (right). The comment on the new section which is in
colour in the manuscript (just above the large initial) is signalled by space in the printed version. Both reproduced
from W Hellinga, Copy and print in The Netherlands, Amsterdam, 1960.

The conclusion from such examples is that there is a ‘language element’

of graphic communication that underlies such equivalences, and that
should be studied seriously ‘in much the same way as linguistic scientists
have studied spoken language’. The paper was originally addressed to an
audience interested in the technology of printing: its purpose is therefore
to show that whereas, historically, the opportunities for graphic expression
have been constrained by technology, the design of future technologies
might be better informed by a proper specification of the requirements of
graphic language.

Graphetics and graphology: the place of graphics in linguistics

The obvious place to investigate the language element underlying
typography is linguistics, and some textbooks do indeed mention the terms
‘graphetics’ and ‘graphological’ in symmetrical opposition to ‘phonetics’
and ‘phonology’. (While phonetics describes simple characteristics of vocal
sounds in speech, phonology describes systems and patterns of sounds.)
However, this definitional symmetry should not be taken to imply that
graphic and phonic factors enjoy an equal status within linguistics.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 38

Chapter 1 • Typography and language: a selective literature review

In practice, graphetic and graphological factors have not received anything

like the detailed attention that linguists give to phonology. Moreover,
judging by the relevant entries in dictionaries of linguistics, there is
evidence of considerable terminological confusion.

• Pei and Gaynor (1954) list only graphemics as ‘the study of systems of
writing and their relationship to linguistic systems’, a fairly broad and
inclusive definition.

• Hartmann and Stork (1972) list graphology as equivalent to graphemics.

To them, though, graphemics is limited to ‘the study of the graphic
signs used in a particular language’. The separate term, graphetics, is
introduced to describe
‘the study of the graphic substance and the shapes of written signs without regard
to a particular language or writing system’ (my emphasis).

• Crystal (1980) agrees with Hartmann and Stork’s definition of

graphemics and, like them, equates it with graphology. He clarifies
Hartmann and Stork’s conception of graphetics (which sounds rather
atheoretical and pointless) by including properties of the written
medium such as colour, type size and spacing.29

• In addition, graphonomy has been defined by Hartmann and Stork as the

same as graphetics, but by Pei and Gaynor as the same as graphemics.
Hartmann and Stork also list graphics and grammatology as synonyms for
graphetics.

• Grammatology is given further prominence from its use by Gelb (1963)

to denote the study of writing systems in general (eg, ideographs,
pictographs, alphabets, syllabaries etc). The French deconstructionist
philosopher Derrida (1967/1976) also uses the term; to him writing is
not so much a physical act or product as a complex metaphor at the root
of all philosophy and science (following Freud’s metaphor of the psyche
as a ‘mystic writing pad’). After the usual time lag, Derrida’s ideas are
gaining currency in the English-speaking world, and will possibly be the
next major influence on future typographic theory. 48

For our purposes it is probably best to disregard all these terms except
graphetics and graphology. Crystal and Davy (1969) explain the difference
as follows:
‘at [the graphological] level, we are laying stress on the contrasts that can be made
within the linguistic system, rather than on the system itself, which was studied at
the [graphetic] level’.

29 The graphetic/graphemic distinction follows a general use of the emic/etic suffixes within linguistics.
Crystal (1980) explains the difference in this way: ‘An “etic” approach is one where the physical patterns
of language are described with a minimum of reference to their function within the language system. An
“emic” approach [or “ological”], by contrast, takes full account of functional relationships’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 39

Chapter 1 • Typography and language: a selective literature review

This, however, implies that graphetics is the study of systems of marks (that
is, rules for their combination), whereas the dictionary definitions quoted
above give the impression that it is simply the marks themselves that are
studied.

We could clarify the graphetic/graphological distinction in the following

way: taken individually, visual techniques such as the design of letterforms,
symbols, rules, tints and boxes might be seen as graphetic; but when they
are used together to structure a whole text, we see a graphological system
at work. The origin of the serif, the design of more legible type, the choice
between the open and closed bowl ‘g’ are examples of graphetic issues,
interesting in themselves but not contributing to our understanding of how
graphic factors are used in the display of textual arguments. For example,
two editions of Shelley’s poems, the one set in Bodoni and the other set
in Univers, may differ from each other and the original manuscript at the
graphetic level (being set in different types) but both are expected to follow
the author’s original graphology – his use of indention, capitalization,
punctuation and spacing.30

This is not to say that typeface selection is a trivial graphetic matter. As

Twyman (1982) pointed out, the same structure can often be signalled
equally well by either spatial or stylistic variation, and the actual method
used is often a function of the production system. Returning to my poetry
example, typeface selection would thus become a graphological matter if
the context allowed a choice: for example, if the poems appeared in a book 49
where different typefaces were used for each poet.

Given this definition, it is hard to see why graphetics is of any potential

interest to a linguist, rather than a palaeographer, since all meaningful
contrasts are classed as graphological. A similar distinction, but of more
use to the linguist, is made by Twyman (1982, 1986), between extrinsic
and intrinsic features of verbal graphic language. By ‘intrinsic’, Twyman
means ‘the range of characters available on a given [composition] system’,
and stylistic variations of those characters (italic, bold etc; letterform
style; letterform size). ‘Extrinsic’ refers to manipulations of the characters
(configuration, micro- and macro-spacing, colour).

There is a superficial similarity between the graphetic/graphological and

the intrinsic/extrinsic distinctions, since both graphetic and intrinsic factors
involve an inventory of ‘graphemes’ or characters – minimal units of graphic
language.31 The difference is that Twyman applies both his descriptors at
the graphological level of analysis, pointing out that a particular semantic

30 The question of the extent of the author’s responsibility for these matters is the subject of some
controversy among bibliographers. The issue is discussed further in Chapter 7.

31 A recent paper by Henderson (1985) discusses the range of uses of the term ‘grapheme’. An even
smaller unit, the ‘allograph’ is sometimes also referred to; an allograph is a graphic feature which may

Robert Waller • The typographic contribution to language • PhD Thesis 1987 40

Chapter 1 • Typography and language: a selective literature review

distinction might be made with a combination of intrinsic or extrinsic

features: the status of a quotation in a textbook, for example, might be
signalled by the use of a change in typeface (intrinsic to the composition
system)32 or by the use of space (extrinsic to the composition system).
Twyman uses the intrinsic-extrinsic distinction to illustrate differences
between the typical output of four composition systems: manuscript,
hot-metal, photocomposition and videotex. In this respect the model is an
important tool for analysing artefactual influences on graphology.

A broader treatment of graphology is by Mountford (1969). Owing to 50

the context of his paper (an encyclopaedia entry), he confines himself to
matters of definition and classification rather than the analysis of examples.
He proposes and discusses ten ‘groups of features’ with which to describe
‘the graphological structure of traditional orthography (in its printed prose
variety)’ (Table 1.2).

Category Example from typical printed prose

1 Colour contrast black on white

2 Orientation left to right

3 Disposition page layout, paragraphing

4 Graphological layering punctuation hierarchies

5 Graphemic composition sets of letters, figures, punctuation marks

6 Graphomorphemic typology syllables

7 Differentiation resources italics, bold

8 Capitalization proper names

9 Graphetics actual letters etc (as distinct from ‘paradigm’ letters

described in 5)

10 Flexibility open-ended (as distinct from less flexible ideographic

writing systems)

Table 1.2 Ten groups of graphological features (Mountford 1969).

It should be remembered that terms like these are intended to describe

writing systems rather than examples of writing. The relevance of some of
the distinctions made becomes more apparent when applied to non-roman
or innovatory writing systems (eg, Chinese, shorthand, etc). In later papers,

appear in a number of different graphemes – for example, a near-identical allograph is usually used for
the ascender of the lower-case ‘h’, ‘k’, ‘b’, ‘d’ and ‘l’

32 A further linguistic parallel with Twyman’s intrinsic/extrinsic distinction is that between inflected
and positional languages. Inflected languages (such as Latin) rely on word endings and other inflections
to establish grammatical relations, and, unlike positional languages (such as English), are relatively
indifferent to word order.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 41

Chapter 1 • Typography and language: a selective literature review

Mountford (1980, 1982) developed further ideas on the analysis of written

language, which I shall return to shortly.

In practice, the discussion of definitions is rather academic: a search of

Language and Language Behavior Abstracts revealed only a handful of papers
on any of these topics between 1973 and 1985, and few were relevant to 51
this review. Many of the 86 papers using ‘graphemes’ as a key word were
about detailed matters of grapheme-phoneme relations, in the context
of spelling reform, palaeography or reading instruction. Some papers on
‘graphology’ turned out to be on personality assessment from handwriting.
Other papers were about non-Roman scripts. Indeed, Crystal (1980)
remarks in his dictionary entry for ‘graphetics’:
‘So far little analysis of texts in these terms has taken place, and the relationship
between graphetics and graphology remains unclear.’ (p. 169)

Why have graphic factors received so little attention from modern linguists?
Compared with other, weightier, matters that preoccupy the relatively
young discipline of linguistics (such as ‘what is language?’), they are
presumably seen as relatively trivial,33 although necessary to mention
when the existence of writing is to be acknowledged. It is not always
acknowledged as a proper subject for linguists to study. This view stems
directly from de Saussure, usually regarded as the founder of modern
linguistics and semiology, who placed writing outside the linguistic domain:
‘Language and writing are two distinct systems of signs; the second exists for the
sole purpose of representing the first.’ (de Saussure 1916/1974: 23)

So the terminological problem reflects not so much a major debate on

fundamental issues as the largely peripheral status of graphic factors within
linguistics. This is, for the most part, a necessary restriction: illustrations,
to take an extreme example, are part of ‘the text’ as it is viewed by an
ordinary reader, but they are clearly non-linguistic. But while graphic
factors may at present be at the periphery of the linguistic field, at least two
major boundaries of that field have come under pressure in recent years. 52
One is the primacy of speech; the second is the restriction of the sentence
boundary.

The primacy of speech

Vachek (1973) documents much of the debate concerning the status of
writing in linguistics. In addition to de Saussure, Vachek cites the opposition
of many of the most influential twentieth century linguists to the view that
writing is something more than the transcription of speech. Bloomfield
(1935: 21), for example, considered that ‘writing is not language, but

33 As Wilson (1844: 4), speaking of punctuation, puts it: ‘The mental philosopher and the philologist
seem to regard it as too trifling for attention, amid their grander researches into the internal operations
of mind, and its external workings by means of language.’

Robert Waller • The typographic contribution to language • PhD Thesis 1987 42

Chapter 1 • Typography and language: a selective literature review

merely a way of recording language by means of visible marks’. Vachek

quotes similar remarks from influential linguists from both earlier (eg Sapir
1921) and later generations (eg Hockett 1958).

The tone of the primacy of speech advocates is emphatic, even intemperate

at times. Thus de Saussure (1916/1974) speaks of the ‘tyranny of writing’,
of its ‘usurping’ role, of ‘abuses’, of the ‘annoying’ tendency of grammarians
who ‘never fail to draw attention to the written form’. The title of one
section of his Course in general linguistics, though, may explain the tone:
‘Influence of writing; reasons for its ascendance over the spoken form’. At
the time (the Course is based on lectures given between 1906 and 1911),
de Saussure’s purpose was to replace prescriptive grammars based on
literary forms34 with a more fundamental description of natural language.
Bloomfield’s remarks were made in the context of the development of
techniques for the description of unwritten Native American languages.

The influence of de Saussure and Bloomfield was such that Bolinger

(1975: 476) could refer, in his linguistics textbook Aspects of language, to
the ‘old-fashioned’ relationship between writing and speech, in which the 53
only level of equivalence is that of grapheme and phoneme. His alternative
approach is to view writing and speech as ‘more or less independent
systems that tend to run parallel but converge more and more and finally
intertwine’. The two systems, he suggests, are related at various levels
(grapheme to phoneme, morphographeme to morphophoneme etc) but are
not dependent on one another. Instead the listener or reader can usually
interpret speech or writing without reference to the other mode. Indeed,
each mode can use elements that have direct meanings without the need
for ‘arbitrary’ units such as phonemes and graphemes: speakers may use
gestures, writers may use pictures or symbols.

Although linguistic scientists have claimed to study spoken, not written

language, it is ironic that transcriptions of actual speech show that
‘grammatical’ sentences are rare (cf Tannen 1982). When linguists discuss
syntax, they generally use idealized examples which conform to the norms
of written language (in which gesture, feedback and a shared context is
absent). Taylor (1984) and de Beaugrande (1984) have identified some of
the tacit rules by which linguists commonly edit language samples into a
form they can analyse.

34 Cohen (1977: 50) remarks on the relatively detailed attention given to graphic factors by early
linguists: ‘The language texts of the period [1640–1785], reflecting an effort to represent the obvious
sense of the written language, include sections on punctuation, capitalization, and often, handwriting
and type styles. These sections are significantly prominent.’

Robert Waller • The typographic contribution to language • PhD Thesis 1987 43

Chapter 1 • Typography and language: a selective literature review

The sentence level

The preoccupation of linguistics with speech was accompanied for many
years by Bloomfield’s additional restriction of linguistic enquiry to the level
of the sentence. De Saussure had earlier made the important distinction
between langue and parole, sometimes translated as ‘language system’ and
‘language behaviour’. The task of linguistics has generally been to reveal the
language system or grammar that underlies language behaviour. Since the
sentence seems to be the highest level at which concepts of grammaticality
are intuitively agreed by language users, the proper study of linguists is
restricted to sentences. The construction of larger units, such as paragraphs,
is more a matter of rhetorical choice than the application of grammatical 54
rules.

In view of this restriction, it is not surprising that graphic factors have

featured so little in linguistics. Indeed, we may wonder why graphetics
and graphology should ever have been posited by linguists in the first
place. The sentence is a level at which few complex graphological events
occur. Graphology becomes more interesting in non-sentences (such as
bibliographic lists or equations) or in texts with headings, tables, footnotes,
and other components which lie outside the scope of sentence grammar and
which have received relatively little attention from linguists.

Some linguists who have noticed graphic aspects of language

It is noticeable that all of the linguists who have written about or
acknowledged graphic factors (other than for the limited purpose of
comparing writing systems) have moved away from the restriction to
sentence level linguistics. Such linguists are few enough to be able to
list here.

Crystal
David Crystal and Derek Davy’s Investigating English style (1969) describes
their approach to the study of stylistics, a branch of linguistics that tries
to describe and account for variations in the language of, for example,
religion, sports journalism or advertising. Since they divide their examples
equally between spoken and written language, Crystal and Davy are
clearly sensitive to the differences between them. Indeed, they preserve the
typography of their examples of written language, and comment on it in
their discussion.

Crystal and Davy’s descriptive method is hierarchical, using five levels of

description. While the three higher levels – grammar, syntax and semantics
– are common to both speech and writing, parallels are identified at the two
lower levels between phonetics & graphetics and phonology & graphology.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 44

Chapter 1 • Typography and language: a selective literature review

However, although the two graphic terms are apparently to be given equal
weight to their phonic equivalents, in practice most attention is given 55
to speech. As they acknowledge, ‘there is no agreed terminology for the
discussion of graphetic and graphological contrasts’ (p 23).

This is borne out in Crystal and Davy’s commentary on the written

examples. It is of a lay person’s commonsense sort, and is not phrased in
a particularly technical manner or in a specifically linguistic sense. They
refer, for instance, to ‘eye-catching’ features or ‘places for the eye to rest’.
This may not be a problem in itself – there are good reasons why we should
resist the cloaking of typographic study in scientific mystique – but in the
context of linguistics, where such cloaking is the norm, it is symptomatic
of theoretical neglect. Two reasons for the dearth of terminology might be
suggested.

First, graphological samples do not present the same problem of

transcription as phonology: they are already in written form and available
for inspection and analysis. Crystal and Davy’s examples of spoken language
are transcribed using an essentially selective notation which includes such
things as pitch, timing and emphasis but ignores other paralinguistic or
contextual features such as the vocal timbre, sex, age and appearance of
the speaker. So the problem of transcribing speech is bound up with its
analysis – with the selection of its salient features and the identification
of relevant units and boundaries. Since written language does not need
transcription, it does not receive the corresponding analysis. For example,
while their transcriptions of speech show evidence of detailed thinking
about the relative importance and the role of each feature, Crystal and
Davy’s examples of written language are presented in an unmediated form.
Their comments about graphological aspects of the written examples do not
give the impression that they are the result of a careful sorting of linguistic
from non-linguistic features.

Second, our alphabetic writing system enforces a simple segmentation

on language and so seems to exclude ‘non-segmental’ or ‘suprasegmental’
effects. In speech, pauses and ‘tone unit’ boundaries do not always occur at 56
points where a writer would punctuate, and not necessarily at word breaks.
To cope with this, notations for transcribing speech normally embody a
technique for indicating ‘prosodic features’ – changes in speed, tone of
voice, and pitch. Musical notation is an obvious parallel: notes are grouped
on the basis of timing, loudness and expression – three systems that are
fairly independent of each other and may not share the same segment
boundaries. It may be that the relatively simple segmentation of written
language makes graphology appear theoretically less interesting than
phonology to linguists. This certainly appears to be the case for Crystal and
Davy who devote much space to the problem of describing non-segmental

Robert Waller • The typographic contribution to language • PhD Thesis 1987 45

Chapter 1 • Typography and language: a selective literature review

phonology (also termed ‘prosody’); their discussion of grammar, too, is

effectively weighted towards spoken language with its problems of vague
sentence boundaries and frequent ‘ungrammatical’ constructions.

Schematic relationship between graphological features of text and other linguistic levels

Graphology Phonology Grammar Semantics

1 feature feature — —
2 letter phoneme — —
3 letter cluster phoneme cluster — —
4 graphic syllable syllable — —
5 graphic word phonic word word lexeme
6 word cluster some prosodic features sentence analysis information
7 line — — —
8 line cluster — — information
9 paragraph — — information
10 paragraph cluster — — information
11 layout — — information
12 page — — information
13 page cluster — — information
14 text — — information

Figure 1.13 Crystal’s levels of graphic organization (redrawn)

Crystal considers written language in more detail in a later paper (Crystal

1979), identifying fourteen levels at which graphological units could be
distinguished (Figure 1.13). In this exploratory paper, originally read at
conference of the UK Reading Association, Crystal addresses the question:
what levels of response might be expected from readers to different levels
of organization? He applies each of fourteen levels of graphic organization
to three aspects of language, semantics, grammar and mode of transmission 57
(writing and speech).

The main point Crystal wishes to make with this framework has to do with
the status of the line, the only feature which, being entirely an artefact
of the printing process, has ‘no statable correlation with any other level’.
He goes on to review research on semantically-controlled line endings for
beginning readers (who can be observed to have problems coping with
line breaks). This particular issue is discussed in more detail in Chapter 7.
In relation to the present review, though, two other notable points emerge
from Crystal’s analysis. Firstly, above the level of the line, all links with
phonology and grammar break down, while below that level the links are
fairly trivial. From a traditional linguistic point of view this suggests that
there is little interesting that can be gained from written data that cannot
be equally well gained from spoken data.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 46

Chapter 1 • Typography and language: a selective literature review

Secondly, at and above the level of the line, the links are solely semantic,
and consist of rather vaguely-indicated correlations with ‘information
structures’. This suggests, again from a strictly linguistic point view, that
graphology above the level of the line is outside the domain of the most
central and exclusively linguistic concern of linguistics – syntax. As we
review the work of other treatments of typography by linguists we will
find further evidence of the importance of the communication context and
purpose – rhetorical factors that ‘pure’ linguistic science has often been
happier to ignore.

Vachek
Josef Vachek (1948/1967, 1959, 1973) is an old campaigner for the
recognition of written language as autonomous from spoken. A member of
the Prague School of linguists, he follows two practices associated with that
school.

Firstly, Vachek defines speech and writing from a functionalist perspective. 58

Functionalism, in the linguistic context, refers to the idea that language
features stem from the function of language in the community of language
users. For example, since asking questions, making statements and giving
orders are universal uses for language, grammarians can expect to find
interrogative, declarative and imperative forms in most languages. Better
known examples concern vocabulary: one doesn’t expect tropical dwellers
to have a word for snow, since they would never have a function for such
a word.35 The Prague School’s functionalism reflects their rejection of
Saussure’s distinction between langue and parole.

Vachek’s functional analysis of writing and speech is as follows:

‘The spoken norm of language is a system of phonically manifestable language
elements whose function is to react to a given stimulus (which, as a rule, is
an urgent one) in a dynamic way, i.e. in a ready and immediate manner, duly
expressing not only the purely communicative but also the emotional aspect of the
approach of the reacting language user.

The written norm of language is a system of graphically manifestable language

elements whose function is to react to a given stimulus (which, as a rule, is not
an urgent one) in a static way, i.e. in a preservable and easily surveyable manner,
concentrating particularly on the purely communicative aspect of the approach of
the reacting language user.’ (Vachek 1973: 15-16)

This analysis suggests that, since speech and writing are distinct in function
as well as in mode, we should not expect to find exact parallels between
phonology and graphology. For example, the static, surveyable nature of

35 One is reminded of President Reagan’s famous gaffe when he asserted that there is no word for
‘peace’ in Russian.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 47

Chapter 1 • Typography and language: a selective literature review

writing suggests a role for typography quite distinct from the emotional and
immediate role of ‘equivalent’ features in speech, such as tone of voice.

Secondly, Vachek makes frequent use of binary opposites. Applied to lexical 59

structure, for example, pairs of opposites such as ‘lion’ and ‘lioness’ are
said to contain a marked and unmarked member. In this example, ‘lion’ is
unmarked and ‘lioness’ is marked. The marked can be distinguished from
the unmarked not only by the formal addition of, in this case, the suffix
‘-ess’, but also by their asymmetrical functions: thus the two terms can be
defined as ‘male lion’ and ‘female lion’, but not as ‘male lioness’ and ‘female
lioness’ (the one is contradictory, the other tautological).36 Applied to such
examples as actor/actress, or waiter/waitress, this analysis amply illustrates
the feminist case.

Vachek (1973) identifies the ‘written norm as the marked member of an

opposition whose unmarked member is the corresponding spoken form’.
The distinction is made on functionalist grounds:
‘that the situations for which the use of the written norm appears specifically
indicated have always something specialized about them, and very frequently such
use serves higher cultural and/or civilizational purposes and functions (use in
literature, research work, state administration, etc.).’ (p. 16)

In a later paper, Vachek (1979) discusses typographic signalling in some

detail, listing a range of functions for which marked sets of graphic symbols
(for example, italics) might be used to distinguish text features requiring
emphasis or stylization from the unmarked norm (for example, roman
type).

A particularly significant point that emerges from Vachek’s discussion is

that he appears to consider markedness to be a matter of distributional
frequency within a linguistic community rather than just within a particular
document. Referring to the Czech and German practice of printing extended
passages such as prefaces in italics, he points out that in such circumstances
printers have to reverse normal practice by using roman type, an unmarked 60
form, for emphasis instead of italic. Although providing only anecdotal
evidence, Vachek maintains that such signalling fails to convince the reader,
and that such signalling in an italic context can only be achieved with some
other marked set such as bold italic or small capitals. My own observations
as a reader convince me of the probable accuracy of Vachek’s position
(Figure 1.14 shows an example).

36 This example is borrowed from Lyons (1977) who goes on to distinguish between different kinds of
marking, but for present purposes this simple illustration should suffice.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 48

Chapter 1 • Typography and language: a selective literature review

Figure 1.14 This reversal of the normal roman/italic markedness relationship is not very convincing, and is
aggravated by the switching of roles in the eighth line down. This example is from the second edition of Fowler
(1926). Actual size.

This is in contrast to a commonly held view, possibly originating with

experiments on the psychology of perception, that figure-ground contrasts
are largely a matter of proportion, and that therefore one might expect
markedness to be a relative to the proportion of two forms within a
particular text. The well known vase-faces illusion (Figure 1.15) illustrates
how we are able to switch between seeing the white and the black areas
as the figure and the ground. The effect is symmetrical in that if either the
black (faces) or the white (vase) occupies too high a proportion of space,
we are no longer able to switch between images.

Figure 1.15 The faces/vase illusion

What Vachek describes as ‘the inability of italics to figure as the unmarked 61

member of the opposition ‘italic type / roman type’ suggests that, as with
‘lion’ & ‘lioness’, italic type can be defined as ‘not roman type’ but not vice
versa. Further, it could be argued that graphic conventions such as the
italic-roman distinction37 can develop, through frequent usage or reasons
of historical development, something approaching the comparatively
immutable status of natural language (such a status being confined, as with
natural language, to a particular language community at a particular time).

This brings us back to the debate about the linguistic status of written
language. Although the primacy-of-speech advocates argued that spoken

37 The history of this particular development is chronicled by Carter (1969).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 49

Chapter 1 • Typography and language: a selective literature review

language is universal while written language is dispensable since it only

exists in a proportion of language communities, Vachek’s response is that
‘the goal to which language development has been directed in any community
is the highest possible efficiency of lingual communication and the maximum
development of its functional range’,

and furthermore that

‘language “optimals” should not rank lower in importance than language
universals.’ (Vachek 1973: 17)

The choice of an optimal language form is, of course, a pragmatic one,

dependent on the communication context, the means available, and the
purposes and limitations of both speaker/writer and listener/reader. Thus
theoretical advances in written language, and especially typography, are not
to be expected from a view of language which is confined to explaining how
words are combined into sentences.

Werlich
In recent years there has been a significant move away from an exclusive
concern with the sentence towards whole texts. Egon Werlich’s A text
grammar of English (1976) has not been widely cited (perhaps not widely 62
noticed) by Anglo-American linguists. Theoretically (at least it appears so
to this non-linguist author) it is rather sparsely argued, leaving numerous
issues raised but unsettled; however, this brings the accompanying benefit
that Werlich (and the reader) does not lose sight of the broader issues
by concentrating overmuch on precise details of linguistic form. It is a
descriptive exercise, considering an unusually wide range of texts – from
advertisements to committee minutes – and describing their typical
components and characteristics. In the present context, Werlich deserves
mention because he notices typography and, like Crystal and Davy whom
he cites, he is usually meticulous in his preservation of the typographic form
of his examples, even where no special conclusion is drawn from it. For
instance, examples which originated as newspaper articles are printed in
narrow columns with rules between.

Although Werlich is clearly aware of graphic and spatial factors in text,

he presumably regards them as unproblematic or outside the scope of his
grammar. There is no special section on typography, and it does not appear
in the index. Where he does mention typography or layout, it is generally
accorded the role of text type identifier. Thus we recognize a leading article
by its conspicuous position and the newspaper’s emblem at its head. I shall
look in more detail at Werlich’s classification of text types in Chapter 9.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 50

 Chapter 1 • Typography and language: a selective literature review

Bernhardt
One of the broadest and most impressive studies of typography from a
linguistic viewpoint was recently published by Steven Bernhardt (1985).
His paper ‘Text structure and graphic design: the visible design’ was
published in the proceedings of a conference on ‘Systemic perspectives
on discourse’, a field much influenced by the work of MAK Halliday and
his co-worker (and wife) Ruqaiya Hasan, whose taxonomy of cohesive
relationships lists a wide range of techniques used by writers to link text
components (but no graphic ones).38 63

A question apparently not answered by Halliday & Hasan is: what leads
a speaker or writer to choose a particular texture (their term for a set of
cohesive techniques in actual use) over another? They point to social and
contextual influences such as the nature of the audience and the purpose
of the communication. Bernhardt set out to investigate this question by
comparing four texts on the same subject written for different purposes.
They are a research report, a legal statute, a brochure and a ‘fact-sheet’,
each addressing the topic of a wetland area of the Great Lakes. Bernhardt
comments that:
‘In my attempt to explain patterns of rhetorical strategy and the consequent
realizations of cohesion with regard to context of situation, it soon became
apparent that graphic design must figure prominently in the analysis of patterns of
cohesive structuring’ (p. 18).

Visually informative lists forms pamphlets directions legal texts textbooks articles novels Non-visually informative

Figure 1.16 Bernhardt’s continuum of visual organization.

Bernhardt proposes a continuum of visual organization (Figure 1.15) in

which various kinds of text are ranged from the visually informative to
the non-visually informative. His choice of terms is interesting, since it
enables him to confine his analysis to verbal language (that is, to exclude
pictures) while admitting spatial and graphic features. Through an analysis
of examples, he arrives at a more elaborated schema (Figure 1.16) which
characterizes the poles of the continuum at various levels of ‘rhetorical
control’. The characteristics described look like useful ways of describing
particular texts, given Bernhardt’s qualification that not all texts will
evidence all features.

38 Halliday & Hasan’s (1976) theory of linguistic cohesion is described in Chapter 6.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 51

 Chapter 1 • Typography and language: a selective literature review

64
Visual Organization of Written Texts

Visually Informative Rhetorical Control Non-visually Informative

varied surface offers aesthetic possibilities; Visual Gestalt homogenous surface offers little possibility of
can attract or repel reader through the conveying infor­mation; dense, indistinguished
shape of the text; laws of equilibrium, good block of print; every text pre­sents the same face;
continuation, good figure, clo­sure, similarity. formidable appearance assumes willing reader.

localized: each section is its own locale with Development progressive: each section leads smoothly to the
its own pattern of development; arrests next; projects reader forward through dis­course-
reader’s at­tention. level previewing and backwards through
reviewing.

iconic: spacing, headings reveal explicit, highly Partitioning integrated: indentations give some indication
visible divi­sions; reader can jump around, of boundaries, but sections frequently contain
process the text in a non-linear fashion, access several paragraphs and some­times divisions
information easi­ly, read selectively. occur within paragraphs; reader must read or
scan linearly to find divisions.

emphasis controlled by visual stress of layout, Emphasis emphasis controlled seman­tically through
type size, spac­ing, headings. intensifiers, con­junctive ties; some emphasis
achieved by placement of infor­mation in initial
or final slots in sentences and paragraphs.

subordinate relations signaled through type Subordinate controlled semantically within linear sequence
size, headings, in­denting. Relations of paragraphs and sentences.

signalled through listing struc­tures, expanded Coordinate controlled semantically through juxtaposition,
sentences, par­allel structures, enumerated Relations parallel struc­tures, and cohesive ties, espe­cially
or iconically signalled by spacing, bullets, or additive ties.
other graphic de­vices.

linkage controlled visually; lit­tle or no use Linking/ liberal use of cohesive ties, espe­cially
of semantic ties be­tween sentences and Transitional/ conjunctives and deictics; frequent
sections; reliance on enumerative se­quences Intersentential interparagraph ties or transitional phrases.
or topicalization of a se­ries. Relations

variety in mood and syntactic patterning; Sentence Patterns complete sentences with little variation in
much use of Q/A se­quences, imperatives; mood; sentences typically declarative with full
fragments and minor forms; phrases used in syntax.
isolation.

Figure 1.17 Bernhardt’s list of rhetorical techniques related to visual informativeness (redrawn).

For some linguists, Bernhardt’s introduction of ‘visually informative’

texts is problematic, since arbitrariness has traditionally been one of the
distinguishing features of language, as distinct from other sign systems
(Saussure 1916/1974: 67). While an arbitrary sign bears a purely
conventional or denotative relationship with its referent, an iconic one
resembles or connotes it in some way. Being visually informative, a list
(Bernhardt’s example of a visually informative text) provides iconic
information about the number, order and grouping of its constituent parts.
I shall return to the problem of iconicity in Chapter 3.

On the evidence of his citations, Bernhardt appears to be unaware of

the graphic design literature, and design features are described in a

Robert Waller • The typographic contribution to language • PhD Thesis 1987 52

Chapter 1 • Typography and language: a selective literature review

non-technical manner. Much of the paper restates familiar ideas (familiar, 65

that is, to most typographers) within Halliday’s theoretical framework.
This, though, is a major contribution in itself, since Halliday’s framework
is important but rather difficult for non-linguists to understand (and to be
sure that they have understood). In a sense Bernhardt’s analysis benefits
from his lack of typographic baggage. Unencumbered by typographic
dogma, he is able to report things as he finds them, reflecting also the
linguist’s commitment to the synchronic study of systems at a fixed point in
time rather than the diachronic study of systems developing in a historical
context.

It is significant that Bernhardt’s interest in graphic design arose out of

an interest in rhetorical strategy and in the influence of context, rather
than in primary message-making. That is, graphic design is placed in his
scheme at a metalinguistic level, describing or structuring a message within
a social framework rather than contributing to its propositional content.
For Bernhardt, the presence of graphic structuring seems to represent a
prediction by the writer about the need to attract readers and allow them a
choice of pathways through the message. His analysis is an attractive one,
and he has moved apparently effortlessly to an integration of linguistic and
typographic ideas that typographers have been struggling with for some
time. His paper is perhaps the most significant theoretical work on graphic
design in recent years.

Mountford
In a series of papers, John Mountford (1969, 1980, 1982) has addressed
the place of the written medium within linguistics. Central to his position is
his concept of ‘writing-system’ which he contrasts with ‘system of writing’.
The latter term is a broad concept, applied to systems such as the Roman
alphabet, the Cyrillic system or the Chinese system. Writing systems
(unhyphenated) have been extensively documented by Diringer (1962),
Gelb (1963) and Trager (1974).

Mountford uses ‘writing-system’ to describe particular systems which 66

are ‘predicated upon a particular language…the relationship between a
language and its writing-system is (or can be) one-to-many’ (1980: 224).
As an example, he cites four special-purpose writing systems for English
which supplement the general-purpose system: stenographies (shorthands),
cryptographies (private or secret systems), paedographies (eg the Initial
Teaching Alphabet), and technographies (which include special phonetic
alphabets).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 53

Chapter 1 • Typography and language: a selective literature review

3
Spelling
Abbreviatory devices
Punctuation
Serialization devices
Differential resources
Identifying devices
Numeric resources
Referral devices
Symbolic resources
Continuity devices
Diacritic resources
Script features
Distinguishing devices
Layout

Table 1.3 Mountford’s (1980) list of writing-system components

Mountford (somewhat tentatively) breaks down writing-systems into

functional components (Table 1.3), which he uses to classify a great many
common graphic techniques. This is a great deal broader in scope than his
1969 classification of graphological factors (cited earlier), which appear
mainly to be grouped under ‘script features’.

Compared with Crystal’s levels of graphic organization, Mountford’s

scheme accounts for a great many more dimensions of graphic language.
It does not seem to be very satisfactory, though, to conflate features that
Crystal analysed into fourteen levels of graphological organization under
a single heading: ‘punctuation (or some wider term to embrace the whole
hierarchy of units-within-units from, say, ‘book’ down to sentences and
below’. Whereas Crystal relates his levels to semantic and syntactic aspects
of the text, Mountford’s functionalism does not extend very far beyond his 67
distinction between writing-systems on the basis of their broad purposes.
As a result, the table (which, it should be said, is not the main focus
of Mountford’s paper, although the only place where he deals with the
specifics of writing) is somewhat unbalanced, combining highly inclusive
categories (such as ‘punctuation’ and ‘script features’, already mentioned,
or ‘layout’) with categories containing only a few members (diacritical
resources, numeric resources). The problem seems to be that Mountford,
in this particular paper, still seems restricted by the linguist’s traditional
concept of data: that is, by what may be intuitively deduced from language
samples. Any further ordering of the rather arbitrary Table 1.3 would have
to be based on a consideration of the semantics and pragmatics of the
text. Table 1.4 shows one way in which the various components could be
grouped:

Robert Waller • The typographic contribution to language • PhD Thesis 1987 54

Chapter 1 • Typography and language: a selective literature review

Spelling 1
Abbreviatory devices 1
Punctuation 2
Serialization devices 2/3
Differential resources 2
Identifying devices 2/3
Numeric resources 1
Referral devices 3
Symbolic resources 1
Continuity devices 3
Diacritic resources 1
Script features 1
Distinguishing devices 1/2
Layout 2/3

where:
1 = ways of symbolizing verbal language;
2 = ways of displaying the organization of content ;
3 = ways of helping readers to negotiate a course through the text

Table 1.4 Functions of Mountford’s writing-system components.

In spite of these reservations, the concept of ‘writing-system’ is a useful

one, although the purpose for which Mountford’s scheme was designed is
rather specialized (the comparison of shorthand and other orthographies).
This present study is only concerned with one of his systems, the general
purpose ‘Standard English Orthography’. Within that single system, though,
there may be a number of distinctive sub-systems or genres. It may be 68
possible to find a range of common functions for which different techniques
and conventions are used within different genres of text. To do so will
require further analysis of, firstly, the techniques and conventions that are
intrinsic to typographic genres and, secondly, the functions typographic text
is expected to perform within different genres.

Conclusion
Prompted by the failure of applied psychologists to adequately specify
the nature of their stimulus materials in a generalizable way, I turned to
linguistics. However, it emerges that, although linguists sometimes refer
in passing to graphic aspects of language, the study of such things is very
far from the centre of a discipline concerned centrally with words, their
meaning and their rule-bound combination. If we are to find a place for
typography within linguistics it will be within that departure from the
mainstream that is sometimes called ‘text linguistics’ or, particularly where
interdisciplinary links are made, ‘discourse processes’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 55

Chapter 1 • Typography and language: a selective literature review

This last term reflects two significant departures from the traditional
ways in which language has been studied: firstly, the objects of study are
whole, purposeful discourses rather than isolated sentences; secondly,
those discourses are seen in relation to processes of construction and
interpretation. Typical contributions to the journal Discourse Processes or to
the series of volumes Advances in Discourse Processes (published by Ablex)
include contributions from cognitive psychologists, socio- and psycho-
linguists, ethnomethodologists, and rhetoricians. In their effort to reach out
from language to its context, then, links are made with other disciplines,
and some of this work will be reviewed in the course of the following
chapters.

Before moving on to consider the relationship between typography and

language in more detail (in Chapter 3), it is worthwhile to consider further
some of the methodological problems raised by this review of the literature. 69
Typographers, psychologists and linguists clearly have different approaches
to similar issues. In the next chapter, therefore, I will review some problems
associated with the interdisciplinary study of practical problems, in order to
refine the goals of this study.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 56

Chapter 2 • Theoretical knowledge in the world of practice

2
70

Theoretical knowledge in the world of practice

The literature review revealed a certain amount of difficulty in reconciling

different research traditions, and some uncertainty about the form which
we might expect knowledge about typography to take. Before following up
the specific lines of enquiry suggested at the end of the last chapter, then, it
is worth taking some time to focus on some general problems raised there
about the conduct of research and the communication of its outcomes.

The myth of the two cultures

Those involved in teaching or researching typography often feel the need
to justify the business of theorizing about design. Their critics (for example,
Chapman 1978) charge them with irrelevance to real situations, and
sometimes imply an irreverence for the mysteries of creativity. Perhaps
because of a schooling which forced a choice between arts and sciences at
an early age, many of us have inherited what may termed the ‘two cultures’
attitude, following CP Snow’s famous essay (1959). Depending on your
bias, either artists are unsystematic and disgracefully ignorant of the basic
facts of Science, or else the creativity, mystery and uncertainty of Art is seen
as stifled or tainted by the philistine and plodding empiricism of Science.
That dichotomy never did stand close examination and, after thirty years
of particle physics, molecular biology and artificial intelligence research
– areas of science which question the nature of matter, life and thought,
and border on philosophy – it seems somewhat quaint. To some degree
it is the researchers’ own fault that they are caricatured as insensitive to
the complexities of design: as the previous chapter showed, it is easy to
explain why research is rejected by some designers and design teachers as 71
unusable.

On close examination it is easy to get the impression that the distinction

between the sciences and the arts is dissolving. On one hand, philosophers
of science have replaced our confidence in the certainty of Science with
a highly relativist view that has gained wide acceptance: where we
might once naïvely have thought that scientific theories derive from the
observation of facts, we are now told that the facts themselves only exist
in terms of theoretical frameworks. Popper (1959), for example, argued
that scientific theories (or conjectures), far from representing certainties,

Robert Waller • The typographic contribution to language • PhD Thesis 1987 57

Chapter 2 • Theoretical knowledge in the world of practice

are actually characterized by the possibility of refutation. Polanyi (1958)

convincingly demonstrated the role of creative intuition in scientific
discovery. Kuhn (1962) argued that fact-gathering in the absence of a body
of theory or beliefs is a virtually random activity. And Feyerabend (1975)
suggested that scientific rationality has been neither the dominant nor the
best model, but that progressive science is essentially anarchical. As far as
Art is concerned, the twentieth century has seen the Muse replaced by the
quasi-linguistic theories of semioticians and structuralists; and we have
witnessed the quest for a ‘science’ of design (Simon 1969), although this
has met with difficulty and some disillusionment (reviewed by Cross 1980).

In spite of such developments in the philosophy of science, old attitudes

certainly survive in typographic research. Some researchers distance
themselves from designers through an exaggerated respect for their
‘artistic’, ‘intuitive’ or ‘aesthetic’ judgement, which is seen as entirely
closed to the scientific method. The converse of this is a ritual sneer at the
unsystematic, even primitive, intuitive approach. For example, Jonassen
(1982: x) proposes a ‘technology of text’ which he describes as ‘the
application of a scientific approach to text design.’ He continues:
‘It exists as a counterpoint to the artistic and unsystematic approach to text design
and layout that has prevailed since petroglyphs were first inscribed on walls’.

He goes on to emphasize that it would be a mistake to apply the technology 72

to anything but expository text: again, the arts are quarantined from the
rigours of the scientific approach.

Ravetz (1971) has warned of the dangers of this assumption that practical
problems always have technical solutions. Drawing a distinction between
the technical and the practical, he cites the inability of the USA and
USSR, two great technological nations with the (technical) ability to send
people into space, to solve their (practical) social or managerial problems.
According to Ravetz, technical problems have simple, easily identified
goals (for example, a typical civil engineering problem might be to build
a bridge over a certain river to carry a specified volume of traffic), while
practical problems are bound up with competing social and historical
pressures. Technical problems are indeed practical, but practical problems
are not always just technical. Those familiar with the typographic research
literature may recognise this consequence of technical solutions:
‘If the inquiry avoids “theory” and becomes “empirical”, it can encounter the
pitfall of simplifying its objects of inquiry to homogeneous populations defined by
classes of simple data; then the complexity and contrariness of the situation, which
created the problem situation in the first place, is lost from view.’ (Ravetz 1971:
355).

Taken to extremes, this can lead to experimental hypotheses which appear

obvious, trivial and even self-fulfilling. An example of this is a much-cited

Robert Waller • The typographic contribution to language • PhD Thesis 1987 58

Chapter 2 • Theoretical knowledge in the world of practice

paper by Dooling and Lachman (1971) who investigated the effect of titles
on the comprehension and recall of text. They prepared a text which was
deliberately vague and, depending on the title, was interpreted by subjects
as being about Columbus discovering America or men landing on the moon
(it referred to unfamiliar rocky landscapes, the ship, the long journey, a new
discovery). But no psychologist has yet managed to convince me that this
is not a self-fulfilling experiment, since an ambiguous text, by definition, is
one which can be assigned alternative interpretations. If it turned out that 73
the text was always understood in one particular way, whatever the title, it
could simply be rewritten until vague enough for the experiment to work. 39

Although written in a specialized context, Dooling and Lachman’s paper is

frequently cited in general research reviews on text design, suggesting that
reviewers imagine that their readers need proof before they will believe that
vague documents need titles. As the educational psychologist John Carroll
once remarked, ‘it is a poor science that does not improve on common
sense’.

Social scientists have the problem that whereas most physicists deal with
matters (in both senses of the word) far removed from everyday life,
everyday life is what social scientists study. Taylor (1980) has highlighted
the difficulty they have in finding scientific ways to define terms, such as
‘personality’ or ‘style’, which ordinary people use constantly in conversation
without any trouble.
‘An approach attempting to adhere completely to a methodological imitation of
the natural sciences could never tell us even as much as what we do already know,
prescientifically, about behaviour, nor about how we could have arrived at such
knowledge.’ (p, 4)

Reviewing the educational psychology literature on text design, it is quite

common to find what at first sight appears to be a staggeringly naïve view
of what counts as knowledge. In effect, a game is being played where a new
‘fact’ is admitted to the circle of those playing only when an experiment has
appeared in the literature to support it. No other knowledge counts. The
game is played in code: ‘nothing is known about...’ or ‘we do not know…’
means ‘no one has published an experiment about…’. My mind seized up
for a few seconds when I encountered the following conclusion to a recent 74
review of classifications of research questions:
‘On the whole, little is known about the kinds of questions that may be posed for
research’ (Dillon 1984: 327).

39 In fairness to Dooling and Lachman, at the time of publication their paper was not especially
intended to contribute to the practical literature on text design where it is so often cited. Instead it
played an influential role in the general reawakening of cognitive psychology. Their purpose was simply
to demonstrate the degree to which contextual inference contributes to the comprehension of language,
and meaningful comprehension contributes to memory. Even so, it surprises the lay observer that such
things needed demonstration.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 59

Chapter 2 • Theoretical knowledge in the world of practice

Pure and applied research

I drew a distinction between the purpose of Dooling and Lachman’s study
and the circumstances in which it is often cited. This highlights another
dichotomy in addition to one between the arts and science. Within the
scientific tradition there is the issue of how pure and applied research relate
to each other and to the world of practice.

Pure research in psychology is concerned with the investigation of

fundamental and relatively abstract aspects of perception and cognition,
isolated from real-life contexts: psychology’s basic task is to explain how
people perceive, think, remember and so on. Cognitive psychologists build
models which embody hypothesized components and mechanisms of a
cognitive system. These must be consistent with known data and testable
by experiment. Consequently, many studies of, for example, the reading
process address themselves to the theoretical problems of a particular
model rather than to the actual problems encountered by readers. In
contrast, applied psychology uses the same methodologies to tackle
real-life problems more or less directly (‘more or less’, because even applied
psychology can seem extremely abstract to lay people).

Wright (1978) has explored the relationship of pure and applied research
into language comprehension. She argues that both branches are interested
in the connection between ‘factors prior to reading’ (differences in texts,
subjects and reading goals) and performance measures (such as retention
or comprehension). The difference lies in the nature of the connection:
pure psychologists are interested in ‘Theories of HOW’ while applied
psychologists are interested in ‘Theories of WHEN’. For example, a ‘pure’
theory might explain how language is represented in memory, while
an ‘applied’ theory might state that when certain kinds of questions are 75
inserted in a text it is memorized more easily. However, some have doubted
that such applied ‘theories’ are in fact theoretical unless linked by some
general framework: Rickards (1977), for example, concluded that most
research into inserted questions was atheoretical and thus impossible
to integrate and apply. And Anderson and Biddle (1975), reviewing the
same literature, dubbed it ‘mindless empiricism’. Brian Lewis (personal
communication) used the term ‘dustbowl empiricism’ to describe the
endless succession of atheoretical experiments on this and similar
educational issues. If there is no foundation for the process of systematic
enquiry we are dealing not with science but what Hudson (1972) has called
‘the cult of the fact’.

Indeed, Wright goes on to remark that:

‘more typically in practice the contents of the Theories of WHEN box [in a diagram
she provides] seem to be an accumulation of statements about the conditions in
which performance is improved or impaired.’ (p. 263)

Robert Waller • The typographic contribution to language • PhD Thesis 1987 60

Chapter 2 • Theoretical knowledge in the world of practice

She cites the legibility research reviews of Tinker (1963) and Spencer
(1969b) as examples. We have already reviewed some of the difficulties of
applying such research.

A related problem is that it is very easy for research results to become

detached from the conditions and qualifications alongside which they
are originally presented. Most psychologists are well aware of the multi-
dimensional nature of the phenomena they investigate, but there is a
trade-off between different purposes of research. Frase (1973) warned
users of psychological research on text that they should distinguish between
three kinds of problem – theoretical, methodological and practical. There is
a danger of attempting to interpret research which is primarily intended to
solve methodological or theoretical problems as a source of practical advice.

In an influential book Cognition and reality, Neisser (1976) made a plea

for cognitive psychologists to respect what he terms ‘ecological validity’. 76
This term describes the extent to which theories of cognition account
for pragmatic considerations of the subjects’ world view in addition to
‘content-free’ mechanisms of the mind. He cites Newell (1973) who listed
fifty-nine different research paradigms in use at that time, fifty-seven of
which were based on artificial laboratory situations. Neisser appeals to
cognitive psychologists to
‘make a greater effort to understand cognition as it occurs in the ordinary
environment and in the context of natural purposeful activity. This would
not mean an end to laboratory experiments, but a commitment to the study
of variables that are ecologically important rather than those that are easily
manageable.’ (p. 7)

For an example of the easily manageable being preferred over the

ecologically important, we might turn to research on educational texts.
Many experiments, including some of those on ‘typographic cuing’
(reviewed in Chapter 1), continue to use literal recall as a convenient
measure, although outside the laboratory the need to recall the exact words
of a text is rare. Furthermore, Neisser is talking not about applied but about
pure research on cognition:
‘A satisfactory theory of human cognition can hardly be established by experiments
that provide inexperienced subjects with brief opportunities to perform novel and
meaningless tasks.’ (p. 8)

If only one or two variables are measured it is possible to produce

statistically significant data which is easy to interpret. Although
psychometricians have produced methodologies and statistical techniques
for handling multi-dimensional issues, they are complex and technically
demanding and are less likely to give clear answers; they may also be of
doubtful use in the hands of those who only half understand them. And if
such studies were to use a wide variety of subjects (age-groups, cognitive

Robert Waller • The typographic contribution to language • PhD Thesis 1987 61

Chapter 2 • Theoretical knowledge in the world of practice

styles, educational and socio-economic groups etc) the technical problems

of experimentation could be overwhelming. This is not to discount the
contribution of psychologists and experimentation, but to argue that the
most usable research results are those that show an awareness of their 77
limitations and their place within a wider theoretical framework.

This view is echoed by applied linguists (eg Widdowson, 1979; Brumfit,

1980) who have similarly had to address the relationship between
theoretical and applied linguistics. The way forward for theoretical
linguistics was established by de Saussure’s distinctions between langue &
parole and diachronic & synchronic. By studying language as a symbolic
system frozen in time, practical issues to do with translation or second
languages could be ignored. Language could be studied in a ‘pure’ form
undistracted by its social context. However, as soon as we need to
study language as a communication system, it is complicated by human
motivations and conversational roles.

This is not to say that research or scholarship must be unselective. All

scholars obviously have to sort out that which is relevant from that which
is irrelevant to the problem in hand. Popper (1957: 145) remarks in the
course of a discussion of the nature of historical study:
‘If we say that the cause of death of Giordano Bruno was being burnt at the
stake, we do not need to mention the universal law that all living things die
when exposed to intense heat. But such a law was tacitly assumed in our causal
explanation.’

A crudely expressed ‘law’ like that might be disputed by biologists who

might conceivably know different, or who might have a better way of
putting it, but it is good enough for the historian’s purpose. Putnam (1978)
calls this ‘interest relativity’.

At one time it was possible to talk of the ‘unity of science’, whereby

sociology was reducible to psychology, psychology to biology, biology to
chemistry and chemistry to physics. Individual fields of study are often
themselves divided into levels of analysis. Linguistics, for example, is
sometimes divided between various applied fields which are reducible,
through semantics, syntax and phonology, to phonetics (as Figure 2.1
illustrates).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 62

Chapter 2 • Theoretical knowledge in the world of practice

78

Figure 2.1 The linguistic sciences, as diagrammed by Aitchinson (1978).

There are two problems here. The first is that in order to focus their efforts
on a particular level of analysis, scholars have to simplify their assumptions
about peripheral matters. Those simplified assumptions may be made from
a state of ignorance about other levels of analysis. That risk will always be
present and calls for good interpreters or popularizers. The second problem
is more fundamental: we have to distinguish between different levels of
analysis within a unified world view, and different, incompatible, world
views.

The unity of science view has been rejected by most philosophers of science,
who talk instead about the incommensurability (which may be roughly
translated as incompatibility) of scientific paradigms. Feyerabend (1975)
likens this concept to the gestalt switching which we experience when
looking at well-known ‘impossible’ figures (Figure 2.2). In both the figure
and in a scientific theory, apparently immutable things may change their
function or disappear altogether when viewed in a different way.

79

Figure 2.2 An impossible figure used by Feyerabend (1975) to demonstrate the principle of incommensurability.

The clash of incommensurable paradigms may explain the tone of the

primacy of speech debate among linguists (Chapter 1), symptomatic,
perhaps, of the fundamental incompatibility of the structuralist and

Robert Waller • The typographic contribution to language • PhD Thesis 1987 63

Chapter 2 • Theoretical knowledge in the world of practice

functionalist linguistic paradigms. Indeed, similarly intemperate language

can be found in well-known paradigm clashes in other fields: when
evolutionists discuss creationism, for example, or when monetarist
economists discuss green politics. Where modernist graphic designers
condemn symmetrical layouts as irrational, or where applied psychologists
appear to dismiss intuitive knowledge, there, too, is evidence of paradigm
clashes.

Paradigms
The term ‘paradigm’, used in this way to refer to systems of thought, is
associated with Thomas Kuhn (1962). Kuhn defines science primarily
in social terms – for him, a mature scientific discipline is a community
of scholars who share a common paradigm. A paradigm is a model or
pattern. It may be a theoretical statement or law, agreed within a scientific
community, that anchors a field of enquiry and suggests a firm course for
further study. The essence of Kuhn’s theory is that science progresses not
by the steady accumulation of facts but by a succession of revolutions (or
‘paradigm-shifts’), where paradigms (and many of the ‘facts’ attached to
them) are discarded in favour of new ones that are seen as better able to
resolve key problems. Together with other influential post-war philosophers 80
of science, he effectively laid to rest the popular ideal of the ultimate unity
of science, and placed science firmly in the context of human activity rather
than a quest for Universal Truth.

Kuhn emphasizes that paradigms rarely start out as complete systems of

thinking. In fact it seems to be important that paradigms are not complete.
According to Kuhn, most scientific work (what he calls ‘normal science’)
consists of trying to fit previously known observations into the paradigm’s
framework, or trying to make new observations which it predicts.
Paradigms are valuable because they define the agenda for a discipline: they
pose a series of puzzles that fascinate individual scientists and motivate
them to do detailed work of the sort that would be impossible if each saw it
as his or her task to build the field from scratch.

Kuhn’s ideas, concisely and persuasively written, have proved attractive to

members of relatively new specialisms who are struggling for consensus.
I have found him cited not only by other philosophers of science but by
writers on research areas relevant to this study – for example, discourse
analysis (Coulthard, Montgomery & Brazil 1981), linguistic history (Cohen
1977) and reading research (Venezky 1984). Members of such young
disciplines perhaps feel challenged by Kuhn’s remark that
‘…it remains an open question what parts of social science have yet acquired any
paradigms at all. History suggests that the road to a firm research consensus is
extraordinarily arduous.’ (p. 15)

Robert Waller • The typographic contribution to language • PhD Thesis 1987 64

Chapter 2 • Theoretical knowledge in the world of practice

While not explicitly condemning the social sciences, he goes on to point

out that in the absence of the focusing and unifying power of a paradigm,
fact-gathering is restricted to readily available data and commonsense
observation.

It is not altogether accurate, though, to suggest that the social sciences have
no paradigms; cognitive psychology, Chomskian linguistics, and structuralist
criticism, for example, all replaced previous paradigms in a revolutionary
manner – not in the sense of sweeping them off the face of the earth, but by
being incompatible with their predecessors they forced individual scholars 81
to choose between them. Having said that, though, it should be noted
that Kuhn’s remark on the paradigm-less state of the social sciences was
something of an aside. Elsewhere he refers to them only obliquely and as
a rhetorical contrast to the established sciences. For example (of broader
interest to typographers, perhaps), he notes that technical articles are
preferred to books as the communication medium within an established
science, since they can assume agreement within the shared paradigm and
knowledge of its facts. But:
‘only in the earlier, pre-paradigm stages of the development of the various sciences
did the book ordinarily possess the same relation to professional achievement that
it still retains in other creative fields.’ (p. 20)

In a postscript to the 1969 edition of The structure of scientific revolutions,

Kuhn expresses puzzlement at the extension of his ideas to areas outside
science, since he confesses that they were themselves borrowed from those
fields:
‘Historians of literature, of music, of the arts, of political development, and of
many other human activities have long described their subjects in the same way.
Periodization in terms of revolutionary breaks in style, taste, and institutional
structure have been among their standard tools.’ (p. 208)

This is certainly the way that design is taught: while a dominant paradigm
is hard to discern, we do have distinct schools and periods – the Arts and
Crafts, the Bauhaus, Swiss typography, and the historical tradition, for
example. In certain places and at certain times a particular paradigm has
dominated, and the way has been made clear for a flurry of activity within
its boundaries.

Whereas for Kuhn the contrast between science and ‘immature’ social
sciences is relatively incidental, Ravetz (1971) has taken up the issue in
more detail. Like Kuhn, Ravetz examines science as a social activity. He
seems particularly fascinated by the edges of science: pseudo-science,
folk science and immature disciplines. As well as offering a detailed and, 82
to students of typography, recognizable description of immature fields of
inquiry, Ravetz also offers positive advice to those engaged in such fields,

Robert Waller • The typographic contribution to language • PhD Thesis 1987 65

Chapter 2 • Theoretical knowledge in the world of practice

whose ‘model of genuine science is a very specialized one, which may be

quite inappropriate to their own tasks’. He points us to
‘the history of scientific inquiry in the period before the rise to dominance of
“positive science”. Then, there was a clear distinction between two sorts of inquiry,
“history” and “philosophy”; and they were in turn distinguished from “art”.’ (p.
372)

The terms survived well into the nineteenth century: although ‘natural
philosophy’ is now ‘physics’ or possibly ‘science’, ‘natural history’ has been
kept alive as the name of a museum, and the older meaning of ‘arts’ lives on
in the title of the Royal Society of Arts which promotes practical skills, not
poetry or painting.

Ravetz especially warns of the temptation to assume the outward pretence

of the positive sciences before it is fully justified: elaborate mathematical or
symbolic systems may become grotesque parodies of the realities they claim
to describe. This view can be found in many discussions of applied fields of
study: a typical example is Brumfit (1980) who discusses applied linguistics
in these terms:
‘…a great deal of harm has been done by the enthusiasm of practitioners for
inappropriate statistically-based experimental work, when discussion of a synthetic
rather than analytic nature may have much greater value: there are academic
dangers in formalism and practical risks in the adoption of inappropriate ritual.’

Similar views are to be found among sociologists, educational theorists

and psychologists (some of whom I cited earlier in the chapter), so much
so that it is fair to talk of a consensus of opinion running across discipline
boundaries – although still a consensus of the minority.

Those who seek support from Kuhn’s work for the establishment of
a unifying paradigm sometimes appear to assume that although the
‘immature’ social sciences may not yet have recognizable paradigms or are
not yet mature sciences, the attainment of scientific status is nevertheless 83
an ideal to be pursued. An important distinction, though, between the
sciences and the social sciences is embodied in the term itself: the social
sciences recognise the social context in which their subject of study is
found. While scientists are typically removed from their subject, which may
be viewable only through an apparatus or methodology, social scientists
are participants in theirs. A single paradigm may be essential to scientists
trying to make sense of a reality mediated by an oscilloscope, but is unlikely
to satisfy the social scientist who, as a human being with complex social,
spiritual and physical needs, can sense the futility of pretending the world is
one-dimensional.

Ravetz proposes a substitution of terms that might help young areas of

study (such as typography) avoid changing from immature sciences to mere

Robert Waller • The typographic contribution to language • PhD Thesis 1987 66

Chapter 2 • Theoretical knowledge in the world of practice

pseudo-sciences. He suggests that ‘research’ might be replaced by the less

formalized concept of ‘history’ and ‘theory’ by ‘philosophy’.
‘It would then not be a cause of surprise or shame that effective new insights come
only very rarely…and it could be recognized that an essential part of a genuine
education in the discipline is a dialogue with its great masters.’ (p. 374)

This view is of some comfort to teachers of typography who traditionally

engage in just such a ‘dialogue with its great masters’, focusing on key
personalities whose work is catalogued and style analysed. Typographic
philosophies may thus be viewable through typographic history, and
may be embodied in the work of ‘masters’ (whether Aldus Manutius or
Herb Lubalin) whose work is representative of a particular coherent and
distinctive approach accessible through criticism and, like fine artists who
copy paintings in galleries, through exercises in pastiche.40

But while attention to the past is common among typographers and

especially artists, it is rare among scientists, who typically see progress as 84
cumulative. Feyarabend (1975) suggests that scientists, too, would do well
to study the past:
‘No idea is ever examined in all its ramifications and no view is ever given all the
chances it deserves. Theories are abandoned and superseded by more fashionable
accounts long before they have had an opportunity to show their virtues.’ (p 49)

As if to prove his point, the same idea was expressed seventy years earlier:
‘Do you want to get at new ideas? read old books, do you want to find old
ideas? read new books.’ (attributed to Robert, Earl Lytton by the Times Literary
Supplement, October 19th, 1906; quoted by Dobson 1917)

Ravetz’s third component, ‘arts’, is also fully compatible with traditional

typographic study. The term is used to describe ‘the set of principles
defining the methods of any class of tasks’. The linking of history,
philosophy and arts, for Ravetz, gives theorizing a goal and retains its links
with the real world. It also recognizes the value of craft skills, expressed
aphoristically rather in the form of ‘laws’ or theories, as providing a first
generation of facts for the new discipline.

Guidelines and slogans

The communication of aphoristic, practical knowledge presents certain
problems. Practical books and articles often present advice or research
findings as simple guidelines. In their more general form, guidelines
have been termed slogan language (‘form is function’ is an example of a
frequently repeated typographic slogan).

40 Macdonald-Ross (1977) has argued for the study of ‘master performers’ as a means of arriving at
generalizable principles of good design; indeed, the currently fashionable quest for ‘expert systems’ is
based on the study of high-performing individuals.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 67

Chapter 2 • Theoretical knowledge in the world of practice

Wright (1985) has been particularly critical of low-level (ie, detailed)

guidelines which, applied without sensitivity to their inevitably numerous
exceptions, can do more harm than good. She further points to the sheer
number of guidelines needed to cover the range of problems encountered
in text design, citing Hartley whose ‘Fifty guidelines for improving 85
instructional text’ (Hartley and Burnhill, 1977b) were subsequently
expanded to eighty (Hartley 1981). She remarks, ‘Why stop there?’.

In his reply Hartley (1982) suggests that guidelines are helpful to

novices if not treated as inflexible rules. Indeed, his papers and books
are enthusiastically sought by educators wishing to know how they can
write and present good textbooks. My own concern is that guidelines
should not become detached from supporting evidence. A typical guideline
might say ‘Use simple language (Some name, 1979)’, without detailing
those circumstances under which simple language might be misleading,
or what constitutes simple language. More seriously, on following up the
reference given one can find that the cited author has simply remarked,
say, ‘Use simple language’. Research references have sometimes been used
for persuasive purposes to lend authority to the guideline – indeed, when
non-experts seek research references it is frequently for this reason.

Since guidelines are often neither detailed enough for exact application
nor generalizable through a theory, they can appear to offer contradictory
advice. For example, Winn & Holliday (1982) offer research-based
guidelines for diagramming which separately suggest that the components
of diagrams should be arranged in a left-right, top-bottom reading
sequence, should be arranged to reflect the physical arrangement of
the system described, and should be arranged so that graphic proximity
reflects conceptual relatedness. In reality these are all good alternatives to
consider when faced with a diagramming problem. Presented in guideline
form, though, they can too easily become detached from their supporting
argument.

In the field of typography an example of well-presented guidelines is

provided by Felker and his colleagues at the Document Design Center,
Washington DC (1981). Twenty five principles for clear writing and
design are presented together with generous examples, a critical review of
relevant research, a bibliography, and, most important, qualifications which
emphasize the exceptions to the rule. When guidelines are presented as the 86
conclusions to a sound argument or a well-documented case study, then
they can indeed be useful and effective. However, there is still no unifying
framework for the advice. The guidelines do not exemplify or embody a
practical theory, mastery of which would enable the reader to build insight
in order to tackle problems not anticipated by the guideline writers.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 68

Chapter 2 • Theoretical knowledge in the world of practice

In a different context (a discussion of the practical curriculum advice

offered to schoolteachers), the problem of guidelines has been attributed
by Anderson (1981) to a false equation of practicality with simplicity. He
argues that features often associated with good practical manuals, such as
simplicity, readability and the use of familiar examples, do not necessarily
make researchers’ advice usable. Good practical manuals, he suggests,
must be sincere in their practicality: too many curriculum researchers
use a rhetorically contrived appearance of practicality, either because it
is expected of them, or simply to ease standards of academic criticism. In
addition, Anderson argues, researchers usually underestimate both the
complexity and the orderliness that lies behind the apparently chaotic (to
the outside researcher) classroom situation. Practicality is not the same
as simplified theory, but is related to the context of a manual’s use as
it is perceived by its users. The question is, of course: how are writers of
guidelines to get at such perceptions? Anderson (a sociologist) suggests
ethnomethodological techniques to model the intentions and actions
of people in practical contexts. Such techniques are briefly reviewed in
Chapters 5 and 8.

Earlier I discussed the relationship between pure and applied research

(mostly in relation to psychology). But although psychologists might see
the debate about the pure/applied distinction in terms of their real-world
relatedness, to most designers both are in the realm of ‘theory’. Far from
being at opposite ends of a continuum of practicality, both pure and applied
psychology are dimensions of scientific investigation, itself distinct in most
designers’ minds from ‘commonsense’ or intuitive knowledge. 87

When psychologists conclude research reports related to language

with practical recommendations or guidelines, they switch from a
mode of reasoning with whose rules they are familiar to one they often
underestimate. While following methodological conventions with precision,
they risk cutting across the accumulated and, often widely shared, practical
knowledge of their audience. Whereas most scientists seek universal laws
and dislike exceptions, practical knowledge is value-laden and context-
sensitive. Because it is very often tacit rather than articulated, it is easy to
underrate.

Stefan Körner (1970), a philosopher of science, sees the distinction between

science and commonsense as a matter of degree, arguing that it is not the
case
‘that the aims of science – prediction, explanation and mastery of natural
phenomena – are foreign to commonsense, but rather that science pursues them in
a more methodical manner.’ (p. 39)

Robert Waller • The typographic contribution to language • PhD Thesis 1987 69

Chapter 2 • Theoretical knowledge in the world of practice

He goes on to suggest that

‘Commonsense thinking and the logic underlying it do not satisfy the requirement
of the exactness of all attributes. Many, possibly all, of its classifications rely on
the recognition of similarities of objects to standard examples and to standard
counterexamples.’ (p. 43; my italics)

Returning to the context of curriculum advice, Ilene Harris (1981) has

similarly argued for the replacement of ‘slogan language’ with what she
terms a ‘case rhetoric’ in which theoretical precepts are interleaved with
practical examples which build personal insight alongside intellectual
understanding.

Explicit and tacit knowledge

Harris quotes Dreeben’s (1970) analysis of the problems of teacher training,
with which typographic teachers and researchers might identify:
‘...there probably exists enough individual knowledge and experience stored in 88
individual heads to provide the basis for sophisticated technologies – were that
knowledge and experience ever brought together, codified, tested for efficiency
and communicated to teachers both in training programs and on the job.’ (p.212)

This wistful longing to encapsulate knowledge in objective form is at the

heart of all applications of the scientific method to practical areas, and
is exemplified by current proposals to apply computer expert systems to
typography (Hewson & Lefrere 1986; Rivlin 1987). But it could be argued
that all this ‘individual knowledge and experience stored in individual
heads’ is what constitutes a technology. The scientific method is a system
of knowledge which, even if it started as a creative intuition or hunch, is
publicly demonstrable by reasoning and experiment: the method as well
as its outcomes are supposedly open to question and debate (‘supposedly’
since the extreme technicality of many scientific processes places them out
of the lay person’s reach). But although designers can draw on scientific
theories from time to time, the solving of complex problems stems from a
different kind of knowledge that, although possibly beyond analysis, can
nevertheless be effective.

Michael Polanyi, himself a scientist, developed this notion of ‘tacit’

knowledge in a series of books and lectures. He argues that it is not only
as real and valid as explicit knowledge but that it forms the root of all
knowledge, citing numerous examples of the intuitive nature of scientific
discovery:
‘Let us recognise that tacit knowing is the fundamental power of the mind, which
creates explicit knowing, lends meaning to it and controls its uses. Formalization
of tacit knowing immensely expands the powers of the mind, by creating a
machinery of precise thought, but it also opens up new paths to intuition; any
attempt to gain complete control of thought by explicit rules is self-contradictory,

Robert Waller • The typographic contribution to language • PhD Thesis 1987 70

Chapter 2 • Theoretical knowledge in the world of practice

systematically misleading and culturally destructive. The pursuit of formalization

will find its true place in a tacit framework.’ (Polanyi 1969: 156).

Another important discussion of the problem of theory and practice is by 89

Ryle (1949), who addressed the distinction between ‘knowing that’ and
‘knowing how’. Ryle’s ultimate concern is rather more fundamentally
epistemological than our present one, to destroy what he terms ‘the dogma
of the ghost in the machine’: that is, the mind/body distinction of Descartes.
In doing so he argues that practical skills and actions exhibit intelligence in
exactly the same way as mental skills. According to Ryle, Cartesian dualism
implies that thinking (by the mind) and doing (by the body) are separate
operations, that thinking must precede doing and that intelligent practice
exhibits a prior mental process. He argues instead that when we describe
an action as intelligent we are not discussing a mental process of which the
action is just an outcome, but we are discussing the action itself. Indeed,
turning the tables on the ‘intellectualist tradition’, Ryle suggests that
arguing logically is itself a practical skill:
‘Rules for correct reasoning were first extracted by Aristotle, yet men knew how
to avoid and detect fallacies before they learned his lessons…They do not plan
their arguments before constructing them. Indeed if they had to plan what to think
before thinking it they would never think at all.’ (p. 30)

Ryle’s polemic against the intellectualist tradition is attractively democratic

– each of us (‘the boxer, the surgeon, the poet and the salesman’) applies
similar intelligence to our particular tasks. But what does this intelligence
consist of? It does not simply mean performing well – clocks and
performing seals do that. Ryle argues that intelligence is instead to do with
responsibility for effective performance:
‘To be intelligent is not merely to satisfy criteria, but to apply them; to regulate
one’s actions and not merely to be well-regulated. A person’s performance is
described as careful or skilful, if in his operations he is ready to detect and correct
lapses, to repeat and improve upon successes, to profit from the examples of
others and so forth. He applies criteria in performing critically, that is, in trying to
get things right.’ (p. 29)

In effect, Ryle is suggesting here that the critical process is the essential 90
mark of intelligent performance. Indeed, it could be said that criticism (or
feedback) is at the heart of any successful performance. It is at the heart
of behaviourist Stimulus-Response psychology, and it is a basic concept in
systems theory where unintelligent systems are seen to maintain stability
through feedback mechanisms (a thermostat is a typical example). In
fields closer to the present enquiry, Kulhavy (1977) has discussed the role
of feedback in instruction, and Eco (1976) applies a feedback model to
semiological communication. And students of graphic design will confirm
from experience that design is easier to criticize (in the non-pejorative sense
of the word) than to prescribe.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 71

Chapter 2 • Theoretical knowledge in the world of practice

Whether or not there is a dualism of thought and action is not the issue to
us, but there is surely a sense in which the criteria referred to by Ryle can
be asked to stand independently from particular actions. To detect lapses
or to profit from the example of others implies the existence of some sort
of paradigm performance, or set of rules, with which actual performances
are compared, although this may remain in the tacit domain. The critical
loop (or feedback) may not be articulated in everyday practice by the boxer,
surgeon or designer, but that is not to say that it cannot be articulated or
might not benefit from being articulated.

Holistic thinking
Polanyi, from his particular perspective, also recognizes the critical process
as central to the growth of both tacit and explicit knowledge. He uses the
term ‘physiognomy’ to refer to situations which can be recognized but not
described; the metaphor is that of the infinitely varied instances of the
human face which we can identify without being able to articulate. Polanyi
argues that defining a physiognomy (which in our own terms might be a
design problem, a magazine page or a typeface) will involve two stages:
a focal awareness of its particulars and a subsidiary awareness of those 91
particulars in relation to their participation in the whole. Using a series of
examples from ordinary life and science, he argues that most ‘knowing’
involves an alternation of focal and subsidiary awareness, analysis and
integration.41 Interestingly, Polanyi cites the use of the term ‘aesthetic
recognition’ by Pantin (1954) in relation to the recognition of species by
zoologists. Pantin’s fascinating paper describes how biologists in the field
‘…cannot help being struck by the contrast between the way one identifies…
animals in the museum and the way it is done in the field’.

In the case of his own speciality (a species of small worm), it is not possible
in the field to analyse specimens feature-by-feature against a recognition
checklist. Instead,
‘if, when we are collecting Rhynchodemus bilineatus together, I say “Bring me any
worms that sneer at you,” the probability of your collecting the right species is
high.’ (p. 593)

Among the more theoretically inclined of graphic designers the term

‘aesthetic’ has become somewhat discredited, with its implication that
one personal preference is as good as any other. Pantin’s use of the term,
though, suggests that it is more sensibly used to describe the feeling of
recognizing the physiognomy of, in the typographic context, a well-formed
letter or a problem solved. Unaccompanied by an articulated critical process

41 Polanyi carefully distances himself from Ryle’s ‘absurd’ conclusion to ‘his [Ryle’s] powerful
arguments’. Instead of dismissing dualism out of hand, Polanyi prefers to ‘dispose of the Cartesian
dilemma by acknowledging two mutually exclusive ways of being aware of our body’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 72

Chapter 2 • Theoretical knowledge in the world of practice

though, designers can be easily deceived by what we might term aesthetic

fallacies – false symmetries and alignments, for example.

This holistic style of thinking need not be accepted uncritically. Although

the context was somewhat different to the present one, in the course of
an attack on historicist and Utopian social policy Popper (1957) makes an
important distinction between two kinds of ‘wholes’: first, the literal totality
of all the components of a system and their relations; and, second, certain
properties of a system which make it more than the sum of its parts – the
gestalts of that school of psychology. The fact that the second sort of whole 92
can be studied scientifically does not mean that the same can be said of
the first sort. For Popper, science – indeed, all description – is inherently
selective. Gestalts, although holistic properties, are not the only properties,
or even the only holistic properties, of systems. Popper cites melody and
rhythm as examples of co-existing holistic properties in music. The lesson
for typographic theory is that we can expect to find similarly co-existent
holistic properties of typographic displays which, although they may be
hard to reconcile in terms of explicit theory, may be as easy as music for
readers to discern and designers to create, using tacit knowledge.

The specialization of scholarship

We may reduce many of the issues raised in this chapter to two key problem
areas: one is the specialization of scholarship, with its distinctions between
disciplines, between pure and applied research, and between systems and
components; the second is the communication of practical advice, with its
associated problems of balancing theory, guidelines and examples, and
establishing a critical method.

The scholar’s job is to analyse and categorize messy real world situations
into sub-problems that can be handled by detailed methodologies.
Conversely, the user of scholarship has to synthesize the different views
and theories into his or her tacit understanding. So although an analytic
approach may be necessary, frequent reference must be made to the holistic
context of each sub-problem if the research is to be useful.

Holistic overviews are themselves selective, though, in their view of

the world, representing not the totality of all sense-data in a system,
but only those data relevant to a particular theme or paradigm. Thus
the psychologist, the philosopher and the linguist all claim to study the
structure of knowledge at various different levels of analysis, but each takes
a distinctive holistic quality as primary.

A problem for the typographer seeking a rationale is that commitment to a 93

single scholarly paradigm, whether psychological, linguistic or otherwise,
seems unduly restrictive. A design problem (or a design product) derives

Robert Waller • The typographic contribution to language • PhD Thesis 1987 73

Chapter 2 • Theoretical knowledge in the world of practice

constraints from a range of sources and so is likely to have more than one
physiognomy – like the impossible figure in Figure 2.2. While a goal of this
investigation is to point towards a critical method, we must not expect to
find a single paradigm. Instead, I shall be discussing a range of alternative
critical paradigms (which I shall term ‘structures’) each deriving from a
different source of design constraints. It will be suggested that the nature
of designing, and interpreting design, is to attain the skill of multiple (as
distinct from selective) perceptions, and to balance or harmonize their
competing demands.

Here it may be objected that I am drawing an inappropriate parallel

between the way that scientists view communication and the perceptions
of ordinary people trying to communicate. After all, while scientists
construct formal languages to make ideas publicly accessible, in everyday
conversation we use natural language, simultaneously drawing on
linguistic, logical, social and aesthetic resources in an automatic way that
does not call for an explicit critical method. However, the relationship
between writer and reader is asymmetric in a way that the relationship
between speaker and hearer is generally not. In a typical spoken
conversation, both speaker and hearer can use natural language. Written
language, though, differs from spoken language in its provenance as well as
its channel. In its printed form it is typically mediated by editors, designers,
printers and others. Although the result may be interpreted ‘naturally’ by
readers who are unaware of the production process, the assembly of printed
texts is a deliberate, planned formal process in which a range of competing
demands and constraints must be carefully balanced.

But it is not just at the producers of text that this multi-faceted critical
method would be aimed. Classical rhetoric not only taught orators to make
persuasive speeches, but it also enabled listeners to spot logical tricks and
biased arguments. In the same way, a critical method for typography is not 94
just for the refinement of technique, but might also represent an extension
of literacy for readers. The greater the critical awareness of all aspects
of text, the greater chance readers have of exercising control over their
reading in terms of both strategy and outcomes. Ivins (1943), with typical
insight, linked the critical awareness of media (printed illustrations, in his
case) to a more general critical awareness:
‘In view of all this [evidence he has just presented] the importance of being able
to recognise the technique or process by which a printed or otherwise precisely
duplicable image was made becomes obvious – for this knowledge enables us to
discount or make allowances for the limitations, the blind spots, the distortions,
implicitly and unknowingly introduced by techniques and processes into duplicate
images and their testimony about the world. These implicit distortions are a
most important part of the unconscious, unphrased, common assumptions of
any society, which basically determines its ideas and action. Very few people ever

Robert Waller • The typographic contribution to language • PhD Thesis 1987 74

Chapter 2 • Theoretical knowledge in the world of practice

realize the extent to which “objective facts” as known by us are actually no more
than peculiarities of our instruments of observation and record.’ (p. 143)

In any case, there is something to be said for treating scientists and lay
persons, users of formal and natural language respectively, in the same way.
The psychologist George Kelly (1955) has recounted how he developed the
concept of ‘man the scientist’, central to his theory of personal constructs.
His appointments during a typical working day would alternate between his
psychotherapy patients and his post-graduate students. He came to realize
that the patients were asking much the same sort of analytical questions
about their personal relationships as the students were asking about their
projects. By providing them with the Repertory Grid technique with which
to analyse their perception of personal relationships (the same analytical
tool he recommended to his students), Kelly enabled his patients to come to
a more objective understanding of their problems from the scientist’s critical
distance.42

I have suggested that, in this context at least, the same Kuhnian paradigms 95
might serve practical as well as scholarly purposes. But where do paradigms
come from? According to Kuhn:
‘Paradigms gain their status because they are more successful than their
competitors in solving a few problems that the group of practitioners has come to
recognise as acute.’ (p. 23, my italics)

Although Kuhn means ‘practitioners’ to refer to scientists, in the absence of

a science we might adapt his argument to mean those who are most deeply
engaged in the field of enquiry, or indeed the field of practice. In effect,
Kuhn is suggesting that paradigms grow from a critical tradition.

The term ‘critical tradition’ is apt, implying both a historical dimension and
an evaluative approach. The historical dimension informs about possible
connotations of candidate solutions to a design problem, and provides
models for particular genres of text. Evaluation is suggestive both of
minimum standards (catastrophe avoidance) and an ideal or typical model
against which design solutions may be measured, and towards which
guidelines might point. The model of typographic communication proposed
in Chapter 5 is directed towards this end. First, however, I shall return to
consider linguistic aspects of typographic study in more detail, since, as
became apparent in Chapter 1, it appears to be the most likely source of
insight. Whatever else typography is, it is a quality of language.

42 Kelly’s technique has been applied across a very wide range of subject areas, including several
relevant to this study – for example, study methods (Thomas & Harri-Augstein 1980), and the expressive
properties of type design (Bartram 1982).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 75

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

3
96

Arbitrariness and linearity: de Saussure’s

‘basic principles’

I concluded Chapter 1 by identifying the field of ‘discourse studies’ as a

possible home for typographic theory. Although interdisciplinary, discourse
studies is essentially language-based. This chapter therefore explores some
further aspects of the relationship between typography and language. In the
first half of this chapter I shall discuss some aspects of the reconciliation of
the verbal and the visual, before moving on to discuss the linearity of verbal
language, and the effect on the writer-text-reader relationship of freeing
readers from that linearity. A key dichotomy is identified, between writer-
control and reader-control of the order of presentation, and this is discussed
further in Chapter 4.

The criteria for languageness:

arbitrary, segmented, systemic and linear
In addition to the twin doctrines of the primacy of speech and the sentence
boundary, discussed in Chapter 1, de Saussure presents us with two
formidable barriers to the application of linguistic principles to the study of
visual aspects of verbal language:
‘the linguistic sign…has two primordial characteristics. In enunciating them
I am also positing the basic principles of any study of this type.’ (de Saussure
1916/1974: 67)

The first principle is the arbitrariness of the bond between the signifier and
the signified. Arbitrary signs are distinguished from iconic (or motivated)
signs. ‘Cat’, ‘chat’ and ‘gatos’, for example, are arbitrary signs, since they do
not resemble any aspect of real cats. ‘Meow’ and ‘miao’, being motivated by
onomatopoeia, are usually cited as exceptions which prove the rule. They 97
are said not to be strictly linguistic because they can be interpreted by direct
reference to experience rather than through knowledge of the language.

De Saussure’s second principle is the linearity of the signifier (that is,

the language ‘surface’). Most linguists are primarily concerned with
‘syntagmatic’ relations between components: the relationship of each
word to its predecessors and successors in the linear sequence. Those
text linguists who take sentence linguistics as their model are similarly

Robert Waller • The typographic contribution to language • PhD Thesis 1987 76

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

concerned with the relationship of sentences and paragraphs within the

linear series.

These two principles seem to be a necessary precondition for the linguistic

method, which seeks to reveal systematic relations between clearly
segmented components. They are clearly important if language is to be seen
as an abstract or virtual system, existing apart from its context of use. And
they are among the most important of the ‘design features’ of language,
as distinct from other sign-systems, listed in linguistics textbooks (for
example, Hockett 1958). But although that is the preferred view of many
linguistic scientists, few real utterances actually conform to these principles.
Unscripted speech, for example, is usually accompanied by motivated
signs (such as gestures, expressions, and changes of pitch) which signal
the frequent false starts, topic switches and grammatical ‘errors’ that result
from its time-bound linearity (Tannen 1982).

Ironically, in view of the insistence on the primacy of speech, it is only

really possible to find actual utterances which conform to the linguistic
ideal in the form of printed continuous prose which, being mechanically
produced, is formed from a limited set of identical characters. In its usual
printed form, prose is verbal, linear, clearly segmented and typographically
neutral. It is ‘non-visually informative’ in Bernhardt’s terms, ‘unmarked’ in
Vachek’s, or ‘arbitrary’ in de Saussure’s. And the systematic ideal is realized
through the application of spelling rules and the opportunity writers have
to carefully revise their sentence structures to ensure their grammaticality.

A number of important practical issues are at stake when we determine 98

whether typographic features can be handled within a linguistic framework.
In spite of the obvious differences between written and spoken utterances,
verbal language can still be recognized as having an existence apart from its
mode or channel of transmission. Even stripped of the intonations available
in speech and the graphic emphasis available in writing, its segmented,
arbitrary and linear nature makes it not only translatable, but transcribable
in a variety of media. If graphic features transgress on these essential
points, we would have to find some other basis for a systematic analysis
on which to base what Twyman (1982) called ‘graphic translatability’. A
topical and pressing issue is how to store graphically organized information
(for example, timetables, or diagnostic charts) in electronic form in such
as way that it can be accessed in formats as different as printed paper and
electronic screens.

Language or paralanguage?
Linguists traditionally deal with segmental aspects of language – the
segments themselves (phonemes, morphemes, words and sentences)

Robert Waller • The typographic contribution to language • PhD Thesis 1987 77

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

and the rules for their combination. Since most actual utterances contain
features which are not strictly verbal, are iconic in some respect or not
clearly segmented, but which contribute both to meaning and to structure,
linguists have introduced the term ‘paralanguage’. Some have proposed that
paralanguage has a counterpart in written language: Bolinger (1975: 478),
for example, refers in a diagram to ‘paragraphology’.

In spoken language, pointing, winking, waving, shrugging and smiling

are all uncontroversially paralinguistic since they are not phonological in
nature. Features that are phonological, such as variations of stress, rhythm,
tone and pitch are usually deemed ‘prosodic’, or ‘suprasegmental’. This
terminological problem need not concern us too much, though, since the
distinction between prosody and paralanguage – the one being articulated 99
in sound, and the other not – is easier to make in relation to speech than it
is in relation to written language. In written text, everything is to a degree
visible (although ‘prosody’, used in its literary sense of metrical structure in
verse, is only as visible as English orthography is regular – ie, unreliably so).
I shall therefore use only the term ‘paralanguage’.

It could be argued that the term ‘suprasegmental’ is rather misleading,

given that both prosody and paralanguage can be used to emphasize the
segmentation of language units – generally at the discourse level. For
example, in many variants of English, changes in pitch mark the relative
position of words within the sentence; and parenthetical remarks are
normally signalled as such by a change in tone of voice.

Lyons (1977) reflects this point by distinguishing between two kinds of

paralanguage, modulation and punctuation. I will straight away substitute
the term segmentation for ‘punctuation’, since Lyons appears to be using
an everyday term in a special technical sense. Confusion could arise since
punctuation, in its everyday sense, could be said to have a modulating as
well as a segmenting function.

Modulation describes the way in which the meaning of an utterance may be

coloured or emphasized by tone of voice, facial expression or gesture. For
example, a sentence like ‘Don’t be boring’ may be taken as an instruction,
an insult, a mild protest at an idea rejected or a joke, depending on how it
is said. In written language we can achieve a similar, but still ambiguous,
effect by italicizing a word or adding an exclamation mark (‘Don’t be
boring!’). Advertising copywriters have developed this use of punctuation to
a fine art: the period after short headlines, single word sentences, frequent
paragraph breaks with excessive indention. These ‘score’ our reading of the
advertisement (the musical term is suggested by Nash, 1980, and discussed
further in Chapter 4) in imitation of an intimate television voice-over –
‘Kleeno. Because you care.’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 78

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

Our writing system has normally been considered inadequate, though, by 100
linguists wishing to transcribe speech in its full paralinguistic richness.
They have had to invent special notations to give some impression of rises
and falls in pitch and the relative stress given to parts of a sentence. It is
possible to use italics and bold type to add some vocal quality to writing but
only to a strictly limited degree. At the discourse level, though, typographic
modulation is common. Textbook designers, for example, often specify
different typographic ‘voices’ to distinguish between, say, the main text,
quotations, captions and study guidance.

Segmentation describes the marking of boundaries in spoken or written

language. In speech, this may done with pauses, with gesture or with
tone of voice. In writing, boundaries may be represented by space, rules
or punctuation marks. At levels higher than the sentence (for example,
sections of a book, or when the subject of a conversation changes),
boundaries are also typically marked by the use of ‘metalanguage’
– language whose function is to structure or monitor the discourse
as a whole. Words like ‘Well’ (in speech) or ‘Introduction’ (in books)
are metalinguistic. In writing, metalanguage is itself often signalled
typographically: headings, for example, function because of the way they
look as well as through what they say.

Many of those who have directly compared speech and writing comment
that, whereas cohesion and structure is achieved in speech through
paralanguage, in writing it is established through a more elaborate and
formal syntax (Cook-Gumperz & Gumperz 1981; Chafe 1982; Tannen
1982). Cook-Gumperz & Gumperz have studied the implications of this
difference for children’s ‘initiation’ into literacy. They comment that
‘children’s use of intonation is an essential, rather than [a] background or
additional part of the information signalling load for a message’ (p. 101, their
emphasis)43

and that
‘For children, the essential change between written and spoken language is the 101
change from the multi-modality of speech to lexicalized discursive sequences of
written language.’ (p. 99)

Interestingly, Cook-Gumperz & Gumperz go on to report that children

compensate for the lack of paralanguage and prosody in writing by
employing graphic means: heavy and dramatic punctuation,44 the free
mixing of pictures and words, and the unconstrained use of space and

43 This should be clear from the fact that in the adult context we generally regard speech with
exaggerated intonation as childish or patronizing.

44 However, Baldwin & Coady (1978) reported that children up to the fifth grade often ignore
punctuation when reading.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 79

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

writing direction. Typographic and spatial features may, it seems, be more

‘natural’ than we normally think.

Not surprisingly, since the prefixes ‘para-’, and ‘supra-’ imply borderline
status, linguists disagree about how exactly paralanguage should be
handled and which features should be included. Crystal (1974) represents
the liberal view, arguing that
‘any vocal effect which can be shown to have a systematic, shared, contrastive
communicational function is by definition part of the over-all sound system of a
language, and thus linguistic.’ (p. 280)

Once one leaves the securely segmented world of phonemes and

morphemes, though, one encounters extreme difficulty in discriminating
between linguistic and non-linguistic noises (or marks on paper,
presumably). Crystal therefore suggests a scale of linguisticness. At the
‘most linguistic’ end of the scale are features which are ‘most readily
describable in terms of closed systems of contrasts’ and therefore ‘relatively
easily integrated with other aspects of linguistic structure (particularly
syntax)’. At the other end of the scale would be features which may be
‘relatively indiscrete’ or have ‘a relatively isolated function’ and so ‘seem to
have little potential for entering into systemic relationships’.

In the context of semiology, Eco (1976) similarly suggests that

‘The universe of visual communication reminds us that we communicate both on
the basis of strong codes (such as language) and indeed very strong ones (such 102
as Morse code) and on the basis of weak codes which are barely defined and
continuously changing…’ (p. 214)45

To determine whether we are dealing with language, paralanguage, or

something in between, we can perhaps best assess the linguisticness of
typography by considering each of de Saussure’s basic principles in turn.

Arbitrariness
Crystal’s concept of relative linguisticness is reinforced by an examination
of the arbitrariness criterion. This is the criterion by which Bolinger (1975)
seems to exclude graphic devices from the linguistic domain when he refers
to ‘paragraphology’. (His point is made only in the form of a diagram,
so his reasoning is not made very explicit.) Although one might have
expected such examples as italicization or underlining, which seem directly
analogous to intonation in paralanguage, he instead cites punctuation
marks and mathematical signs, on the grounds that they are interpreted
directly rather than by their equivalence to a phonological feature.

45 However, a code that is weakly defined and subject to change seems to stretch the meaning of ‘code’
rather far. A basis for inference that happens to be shared by more than one person need not constitute
anything as formal as a ‘code’. The distinction between coding and inference is discussed further in
Chapter 5.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 80

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

Punctuation indeed seems paralinguistic – not for that reason but rather
because one of its functions is to indicate how sentences should sound.
According to one view of punctuation, the various stops mark pauses of
varying length, or the use of exclamatory or interrogative intonation.46

Mathematical signs are also a debatable example because what

distinguishes them from language is not their phonological status.
Considered separately (rather than in combination, when they can be 103
diagrammatic), they are often simple alternative representations of words
(‘2+2=4’ is an alternative transcription of ‘two plus two equals four’).
While it is true that they do not correspond to phonological features at the
level of the phoneme, they do at the level of the word. Both ‘four’ and ‘4’
are pronounced fʊǝr

Westcott (1971) disputes the arbitrariness criterion altogether, not only in

relation to written language. For example, he cites numerous morphological
examples (‘longer’ is longer than ‘long’, and ‘longest’ is longer than either),
and syntactic examples (the normal subject-verb-object order represents the
actual order of transitive events). He also lists a range of different kinds of
iconism in writing (Table 3.1). Similar examples are cited by other writers
on this theme (for example: Martin 1972; Lotz 1972).

➔ pictogram

? ideogram (but see footnote, previous page)

$ logogram

pp (meaning ‘pages’) morphogram (the second ‘p’ only)

O in ‘IOU’ homophonic phonogram

& syllabic phonogram (when it appears in ‘&c’, meaning ‘etc’)

, (comma) prosodic phonogram (when used to indicate a pause)

Table 3.1 Categories of iconic symbols in the English writing system (adapted to table form from Westcott 1971).

Whereas iconicity and motivation, two terms used as the opposite to

arbitrariness, are usually regarded as synonymous in relation to spoken
language, Westcott’s examples suggest that in written language it might be
useful to distinguish between them. This is because ink offers the possibility
of a much more literal iconicity than air. Written texts can contain not only
traditionally-defined motivated words (like ‘meow’), and motivated graphic 104
effects like emboldening for emphasis, but also iconic displays (that is,

46 According to Husband & Husband (1905: 13), at least two punctuation marks owe their shape to
abbreviations of words. If this is the case, then they can lay claim to linguistic, not paralinguistic, status.
‘It is said’, say the Husbands, ‘that the question mark originated as the first and last letters of “Querio”
placed one above the other. The “o” becoming in time a dot.’ They suggest that the exclamation mark (or
‘note of admiration’ as it was once called) is a similar development from ‘Io’ (joy).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 81

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

pictures or symbols) which are interpreted more or less directly, not via the
(supposedly) phonetic writing system. It is the latter that Bolinger picks out
as paralinguistic.

It is possible to exaggerate this distinction, though, since while it may

be pedagogically convenient to give children a working model of writing
as a phonetic system, it is not wholly phonetic in practice, as Bolinger
(1946) himself demonstrated in an earlier paper on what he termed ‘visual
morphemes’. Since mature readers have little difficulty in distinguishing
between differently-spelled homophones, such as ‘meat’ and ‘meet’,47 it
is obvious that it is not only mathematical symbols that are understood
directly from the written surface without the need for phonological
equivalence. Besides the usual ‘pair’/‘pare’/‘pear’ examples, Bolinger cites
the use of ‘-or’ as a suffix of prestige, citing attempts to upgrade professions
such as ‘advisor’, ‘expeditor’, and even ‘weldor’. In an earlier incarnation of
the same debate, Henry Bradley (1928) cites a number of similar examples,
including the attempt of the compilers of the Oxford English Dictionary to
determine whether ‘grey’ or ‘gray’ is correct:
‘Many of the replies, especially those from artists, were to the effect that the
writers apprehended grey and gray as different words, denoting different varieties
of colour.’ 48

Bradley suggests that one of the consequences of the partly ideographic

nature of writing is the divergence of written and spoken language.
As a lexicographer he was aware that new ‘graphic’ words can be
readily constructed from Greek or Latin roots, with little regard to their
pronunciation: 49
‘For these words the normal relation between alphabetic writing and speech is 105
simply reversed: the group of letters is the real word, and the pronunciation
merely its symbol.’ (Bradley 1928: 178)

That writing is treated as ideographic by readers is confirmed by

psychologists, who have long debated whether written symbols need to be
recoded into a phonological form before they can be understood. Reviewers

47 It is ironic that the distinction between aural and oral can be neither articulated orally nor detected
aurally.

48 The converse of such observations is that many (iconic) pictograms are culturally biased (Mangan
1978) and are thus arbitrary to those from other cultures. For example, when using a guide-book with
numerous pictographic symbols, we often have to look them up in a key in much the same way as we
look up unfamiliar words in a dictionary. Their iconic origins may only become apparent after we are
aware of their intended meaning. And Baron (1981) reports that iconicity is a surprisingly unimportant
factor in the learning of sign-languages for the deaf, autistic or mentally retarded.

49 The best examples are found in the multi-syllabic compound words coined by chemists, which can
be as complex as the chemical compounds they denote. The word ‘syntagm’, used later in this chapter is
another example of a graphic word with no obvious pronunciation in English. Although Wade Baskin’s
translation of de Saussure (1916/1974) uses the word ‘syntagm’, it does not appear in my dictionary.
Some take ‘paradigm’ as a guide and pronounce it ‘syntam’, others say ‘syntagum’, while most, I suspect,
treat it like Polish names in a newspaper report – we note their graphic shape but don’t actually attempt
to pronounce them.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 82

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

(such as Massaro 1979; Baddeley 1979, 1984) have reported that

subvocalization is not a necessary stage in the fluent reading of relatively
easy sentences, although sometimes used for complex comprehension
tasks. This, Baddeley argues, is because subvocalization helps retention
in short-term memory by means of what he terms the ‘articulatory
loop’ (analogous to an audio-tape loop that can be instantly replayed
for checking). However, Baddeley & Lieberman (1980) also propose an
equivalent sub-system for visual information: the ‘visuo-spatial sketch
pad’ and Kleiman (1975) suggested a model that contains both a visual
and a phonological store. Although there is still some disagreement at the
sentence level, there seems to be agreement at the word level that, although
sometimes used by readers, phonological equivalence is not in itself a
criterion for a readable symbol. The current view is fairly represented by
Kolers (1985: 410) who remarked:
‘The linguist’s view of reading as requiring phonological mediation might be said
to imply that vision is dumb but hearing is smart … This claim cannot be taken
seriously any longer, and the wonder is that it was taken seriously for so long
during the 1960s and 1970s. Are faces, scents and music recognized by finding
their surrogates in speech?’50

Levels of analysis: letters, words, paragraphs 106

It is also worth noting that the concept of writing as a completely phonetic

transcription of speech is a mistake that can only be made by users of
alphabetic writing systems such as our own. The inadequacy of that
assumption must be obvious to the Chinese, whose own writing system is
not phonetic, and who, having recently implemented major changes in the
way their language is romanized, must be only too aware that alphabetic
graphemes are but a crude approximation of phonemes.

Moreover, most historians of writing place pictographic or ideographic

systems prior to phonetic ones (for example, Gelb 1963; Diringer 1962).
Although our evolutionary perspective may lead us to conclude that the
earlier systems were therefore proto-stages in the development of the
all-conquering alphabet, they can hardly have been unfunctional – indeed,
they were used for centuries. The Chinese experience is that there are
trade-offs between the simplicity of the phonetic method and the multi-
lingual comprehensibility of the ideograph. Indeed, Harris (1986) has
argued that writing was developed because its independence from speech
gave it certain advantages:

50 An interesting and perceptive variation of subvocalization is mentioned by IA Richards, who

observes that the visual image of words in poems is accompanied not only by an auditory image but
also by ‘the image of articulation – the feel in the lips, mouth, and throat, of what the words would be
like to speak’ (Richards 1926: 119). The related issue of oral and silent reading is discussed further in
Chapter 4.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 83

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

‘Hence it is particularly perverse of modern scholarship to present progress in

human written communication as consisting in working towards devising one
system, namely the alphabet, which was an improvement over its predecessors in
being specifically tied to pronunciation.’ (p. 119)

The alphabetic system obviously does have a phonetic basis, but its real
advantage is its economy of symbols: something modern linguists, with
their spectrograms and computers, might not have achieved. Our own
limited alphabet provides an approximate phonetic system while preserving
etymological clues about word origin and meaning, and enabling the
exploitation of printing with moveable type. The earlier Chinese and Korean
inventions of moveable type (McMurtrie 1937) were not destined to last, 107
given the multiplicity of characters in their writing systems.

It may be that the scope of the arbitrary/iconic distinction is relative

to particular levels of linguistic analysis. Indeed, Westcott (1971: 426)
suggests that
‘iconism is a relative rather than an absolute characteristic of any communication
system, language included. As regards iconism, then, the only realistic question we
can ask about a given form is not “Is it iconic?” but rather “How iconic is it?”.’

If the existence of limited sets of highly iconic signs (such as pictograms)

simply exploits the way we normally read, why should there be any
problem in analysing a sentence which contains a pictogram, say, of a %
instead of the word ‘telephone’?51 That pictograms are out of bounds is
understandable only if we are looking for systematic relations between
language components within the word (that is, phonemes and morphemes).
Above that level it seems irrelevant how particular words are graphically
rendered, so long as they are comprehended in an equivalent way by
readers. This is the view taken by Trager (1974) who, although somewhat
uncompromising with regard to the primacy of speech, is prepared to
accept symbols as writing if they constitute
‘a systematic representation of linguistic elements – specific morphological (words,
phrases) or phonological (phonemes, syllables) items.’ (p. 380)

In practice, it should be added, there are limits to this. Firstly, because

there is a strictly limited vocabulary of symbols or formulaic pictures which
we can rely on others to understand as reliably as if they were words;
and, secondly, because many words contain grammatical as well as lexical
information (that is, ‘inflective’ information about case, tense and so on). 108
In practice, pictograms can most reliably substitute for words in what

51 The rebus – the use of pictograms to indicate the sound of the name of the thing depicted rather
than its meaning – is regarded by historians of writing as an important transitional stage between
ideographic and phonetic writing systems. I suspect, however, that the fact the rebus is now largely
confined to the status of a curiosity indicates just how non-phonological the reading process has become.
The rebus principle can be demonstrated with letter-games where we are meant to say the letters and
listen to the sounds they make: for example, U R N NML, I M A UMN BN (You are an an-im-al, I am a
hu-m-an be-ing).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 84

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

Quirk et al (1985) call ‘block language’ – single word captions, headings

and labels – as distinct from sentenced language. For example, while some
Open University textbooks use the words ‘audio-cassette’ or ‘reading’ to
draw attention to links between the main text and supplementary course
components, others substitute directly-equivalent icons of audio-cassettes or
televisions:

The word level may also be significant to historians of writing, who

discriminate between the different levels of analysis at which writing can
display language. Harris (1986) argues that, since it is more likely that the
progressions from pictographies to syllabaries to alphabets were partial
rather than total revolutions, these writing systems must have something in
common:
‘All three are equivalent at a linguistic level of great practical utility, but for which
we have no current linguistic term: and this is, significantly, because modern
linguistics insists on talking about language in terms of hierarchies of discrete
units. The nearest approach to what we want would be to call it the level of “word
identification”.’ (p. 116, my italics; Harris puts quotation marks around the term to
indicate its disputed status among linguists.)52

If pictograms can only be treated as words, more elaborate iconic displays

such as pictures might perhaps be viewed as linguistic components at
a higher level of analysis: as equivalent to paragraphs or other verbal
segments larger than the sentence. Indeed, Eco (1976) suggests that the
verbal equivalent of an iconic sign
‘(except in rare cases of considerable schematization) is not a word but a phrase or
indeed a whole story.’ (p. 215)

A picture of, say, a horse, is at a much greater level of particularization than

the word ‘horse’: it shows, for example, a black horse galloping, or a white 109
horse standing still.53

In such cases, however, the image alone may be insufficient for its own
interpretation. Indeed, Gombrich (1960) argues that no pictorial image
gains the status of a ‘statement’ unless an explicit reference is made to
what it is supposed to represent. In the case of propaganda photographs
of alleged war atrocities, for example, it is the false captions not the
photographs which lie. Barthes (1977) uses the term ‘anchorage’ to

52 This view is rather spoilt, though, by the fact that the universal use of word separation was
apparently a seventh or eighth century innovation (Saenger 1982). In any case, an alternative phrase
might have been ‘concept identification’, since a clear notion of the ‘word’ may not predate literacy, but
may instead be a consequence of it. According to Goody (1977), some modern societies where literacy is
not fully established do not have a concept of the word.

53 The greater level of particularization of pictures points to an essential difference between pictures
and pictograms which is reflected in their normal graphic treatment. The modern pictograms typically
found in airports and travel guides are intended to convey generalities of the same order of abstractness
as words. Their characteristic graphic neutrality is perhaps the most significant aspect of their invention
by the Isotype Institute (Neurath 1936).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 85

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

describe the relationship of pictures to captions or other accompanying

verbal language: most pictures are capable of several interpretations until
anchored to one by a caption. One way to handle the picture-caption
relationship is to regard them as a single textual unit (or, at least, as tied
units in the manner of a noun phrase, or a compound word). Garland
(1979), who classifies components of diagrams, significantly includes
‘caption’ as an integral characteristic.

We may take it, then, that iconic forms (or even iconic qualities of verbal
forms – display typefaces with special associations, for example) need to
be welded in to the context, or overall cohesive structure, of a particular
text. But this is no less true of verbal components of texts: words, and
even sentences, however well-formed, are meaningless in isolation from a
context. The experiments of Meyer (1975) showed that even paragraphs
can be interpreted in different ways according to the context in which
they are found. However, it is rather more of a challenge to achieve such
cohesion in the case of typographic, pictorial or diagrammatic displays.
Whereas prose is submitted for publication in the order in which it is to
appear, illustrations are generally submitted separately and integrated (if
at all) at a later stage of text production over which the writer traditionally
has little control. Some implications of this will be noted in later chapters – 110
in particular, the need for closer integration of writing and design processes
will become apparent (Chapter 9), and the effect of adopting the page or
double-page spread as a ‘linguistic’ unit will be noted (Chapter 7).

The segmentation problem

It is apparent, then, that the strict arbitrariness criterion can be
circumlocuted to some degree. More central to the linguistic model than
arbitrariness is the ability to identify systemic relations amongst the data
(that is, samples of language or typography). This in turn would seem
to depend on the data being clearly segmented at the appropriate level of
analysis, since mainstream linguistic science relies on a distinction between
an inventory of components (at the sentence level, the lexicon) and rules
for their combination (the grammar).

Seen from this perspective, iconicity and motivatedness only become

problematic to the linguist when they prevent such segmentation. For some
typographers that might be the end of the matter, since their task is simply
to render an author’s linear string of words and pictures in an acceptable
but still linear form, perhaps embellishing the text with a moderate amount
of, for example, italicization, emboldening, or colour. But although all
typographers start with segmented matter (words, lines, pictures and so
on), many take the opportunity to break out of the linear string to use
both dimensions of the page – in Twyman’s (1982) terms, to use extrinsic

Robert Waller • The typographic contribution to language • PhD Thesis 1987 86

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

features of the composition system as well as intrinsic features. In such

cases, meaning may be added to the segmented string through the analogue
shape of the whole layout. Such non-linear, unsegmented and analogue
features present something of a challenge to the linguistic model. To
pursue this issue further it is necessary to make a brief excursion from the
typographic context to that of pictorial imagery, a much more severe test of
the linguistic model.

Writers on graphic design often talk as if there is a language of visual

imagery which mirrors verbal language. Booth-Clibborn and Baroni (1980),
for example, claim that they ‘have analysed the graphic language over
the last few decades and have found that a universal syntax emerges’.
Thompson and Davenport (1980) similarly claim that ‘graphic design is
a language. Like other languages it has a vocabulary, grammar, syntax,
rhetoric’.

These are bold claims and, although made in short introductions to

illustrated books and backed up by no specific evidence, are probably
seriously meant. They are in good company, since others with more serious
theoretical intentions have also maintained that apparently unsegmented
graphic images can be analysed in a linguistic manner.

Ivins (1953) used the term ‘syntax’ to describe the conventions used by
engravers to make reproducible images before the days of the photographic
half-tone. For Ivins, visual syntax referred not to the objects depicted but
to the manner in which black and white lines were deployed in order to
produce the complete grey scale. For example, wood engravers use closely-
spaced cross-hatching to create the illusion of light and shade.54 Although
it is less obvious from a distance, the same technique of illusion is used
when photographs or paintings are ‘screened’ (broken into dots of various
sizes) for printing. According to Ivins, different syntaxes grew from the
various print-making technologies and stylistic inventions of particular
artists and eras. The typographic equivalent might be the different
repertoires of variants offered by different composing systems (for example,
the upper case/lower case/underlining/ second colour repertoire of the
mechanical typewriter as compared with the greatly extended range of the
phototypesetting machine).

Illustrators and engravers are responsible for making every mark contribute 112
to the depiction of a visual scene. In effect, each line, mark or smudge
explains some aspect of the object depicted or the play of light upon it:
hence Ivins’ use of the term ‘syntax’. He also refers to perspective as a
‘logical grammar for the representation of space relationships in pictorial

54 Elsewhere (Ivins 1943), Ivins makes explicit reference to psychological experiments into visual
illusion, but the full extent to which representational art rests on illusions of all kinds has been discussed
at length by Gombrich (1960) and Arnheim (1969).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 87

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

statements’ (p. 23; my emphasis). Twyman (1985), who reviews a number

of implications of the language analogy, also uses the word ‘grammar’,
although within quotation marks, in relation to conventions of drawing,
such as cut-away, isometric and perspective techniques.

Ivins’ use of the term ‘syntax’ is appropriate only in the limited sense that
he is dealing with systems for the combination of separate components.55
In verbal language it is syntax that dictates which word-orders are legal
and which are illegal. Without separate components, then, syntax is an
inappropriate term. Ivins’ components, though, are of a very low semantic
status – visible marks, such as cross hatchings or dots, rather than signs
with an independent meaning comparable to that carried by words in
verbal language.

A ‘linguistic’ approach to iconic displays at a higher semantic level is

proposed by Gombrich (1960), who chooses vocabulary rather than syntax
as the point of comparison.56 Gombrich does not so much look at the
exact mark-making system of the artist as the people and things that he
or she depicts. With some notable exceptions (such as the impressionists,
nineteenth-century wood-engravers working from photographic originals
and the recent school of super-realists), most artists have not behaved 113
like digital scanners, transcribing exact retinal images onto the canvas.
Instead, Gombrich argues, most artists have seen their task as the depiction
of separate (or at least, potentially separate) semantic units, albeit
considerably modified and merged into unitary compositions. Twyman
(1985) makes a similar distinction between synoptic images and images
composed of discrete elements.

55 Since they do not consist of separate marks, Ivins regards photographs as ‘pictorial statement
without syntax’. Ivins’ view of photographs as unmediated samples of reality may not have been a fully
considered one, since he sees them largely in contrast to engravings where the mediation is extreme.
However good the verisimilitude of a picture or photograph, it is still the product of an artist or
photographer who must frame the image and select an appropriate technique for projecting it onto the
two-dimensional surface of the page. Even the most descriptive pictures are selected for a purpose and
so are to some degree explanatory.

56 In an early instance of the linguistic analogy (in De pictura, drawn to my attention by Van Sommers
1984), Alberti draws a parallel between the learning of writing and of painting: ‘I would have those
who begin to learn the art of painting do what I see practised by teachers of writing. They first teach all
the signs of the alphabet separately, and then how to put syllables together, and then whole words. Our
students should follow this method with painting. First they should learn the outlines of surfaces, then
the way in which surfaces are joined together, and after that the forms of all the members individually.’
(Alberti 1435/1972: 97).As Twyman (1985) shows, manuals for teaching artists often take a similar
approach, in which images are built up from schematized elements.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 88

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

Figure 3.1 Objected-oriented computer graphics (a), Isotype diagrams (b), and heraldry (c) are examples of graphic
displays formed from discrete semantic units.

This approach is the basis of object-oriented computer graphics programs.

As Figure 3.1a demonstrates, these programs can treat parts of images as
separate objects which can be copied, rotated, enlarged and so on. It is also
the main principle of the Isotype system of pictorial communication which
uses a vocabulary of standard symbols (Neurath 1936; see Figure 3.1b). An
older example of a visual system with linguistic parallels is heraldry (Figure 114
3.1c), with its vocabulary of symbols, grammar for their combination and
high-priesthood of grammarians (the heralds).

Gombrich, though, does not confine his schema theory to images composed
of discrete elements. He relates a process he calls ‘schema and correction’
to both the creation and the reading of synoptic images.57 According to
Gombrich, artists draw on a vocabulary of visual schemata which are then
corrected to fit the task in hand. His alternative phrase is ‘making and
matching’: a previously made image is matched to the purpose in hand.
This is easy to see in children’s drawings where houses, cars and people
are typically represented in stereotyped ways, with special details added
to identify the particular house or person concerned: a child’s drawing of
‘Mummy’ may be a combination of a woman-schema and an accessory such
as a hat or bag to identify which woman. Eco (1976: 206) calls such details
‘recognition codes’ for pertinent information. Gombrich cites historical

57 In contrast to Ivins, who regards syntax as necessary for the making of images but says that ‘once
they are put together there is no syntax for the reading of their meaning.’ (p. 61).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 89

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

examples as evidence for his claim that the same effect can be seen in
adults’ pictures as visual schemata develop within a culture over many
years. (Figure 3.2 shows an example).

Figure 3.2 Gombrich uses this example to explain his concept of making and matching. A German artist has
drawn a flood scene in Rome from verbal reports. The timber castle with the steep roof represents his schema of
a (German) castle, but it is modified by a number of features he knows the real castle possesses: round towers, for
example.

Although he appears to be using the word ‘language’ as a loose analogy, 115

Gombrich maintains that:
‘Everything points to the conclusion that the phrase “the language of art” is more
than a loose metaphor, that even to describe the visible world in images we need a
developed system of schemata.’ (p. 76)

He goes on to claim that particular cultures have ‘vocabularies’ of schemata

which might cover, say, people, animals, architectural styles, landscapes and
so forth.

Not surprisingly, the parallel between graphic images and verbal language
has proved controversial. Critics of the ‘language of art’ viewpoint, such
as the philosopher Susanne Langer (1942), point out that there are no
pictorial equivalents to the syntagmatic nature of language (its unfolding
in time) and or to words with their relatively fixed equivalences that enable
the construction of dictionaries. Langer draws on Gestalt psychology to
contrast what she terms the logical form of holistic art objects with the
discursive form of verbal language. Verbal language forces us
‘to string out our ideas even though their objects rest one within another; as pieces
of clothing that are actually worn one over the other have to be strung side by side
on the clothesline.’ (p. 81)

She acknowledges that pictures can function as symbols but ‘a work of

art is a single symbol, not a system of significant elements which may be
variously compounded’. The symbolic function of iconic displays ‘depends

Robert Waller • The typographic contribution to language • PhD Thesis 1987 90

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

on the fact that they are involved in a simultaneous, integral presentation’

(p. 97). Langer’s position is rather metaphysical – perceptual forms are
for the ‘conception, expression and apprehension, of impulsive, instinctive
and sentient life’ – and not entirely helpful to those seeking an articulated
critical method.

Nelson Goodman (1976) has also stressed the unsegmented nature of

visual images. According to Goodman, although it is possible to distinguish
between constitutive and contingent properties of a work of literature, this is
not true of paintings. Goodman maintains that the concept of a fake copy of 116
an existing novel is a nonsense, so long as the words are correctly ordered
and spelled.58 The words originally composed by the novelist are what
constitutes the work: other things, including, presumably, the typographic
layout (except those aspects specified by the author), are in Goodman’s
terms contingent:
‘In painting, on the contrary, with no such alphabet of characters, none of
the pictorial properties – none of the properties the picture has as such – is
distinguished as constitutive; no such feature can be dismissed as contingent, and
no deviation can be dismissed.’ (p. 116)

Goodman compares both verbal and visual forms with an ideal concept
of notation which has a number of syntactic and semantic characteristics:
in particular, the symbols or characters in a notational system must be
unambiguous, disjoint and differentiated. Verbal language fulfils only
some of these requirements, paintings none of them. However, verbal
language does at least satisfy the requirement to be differentiated and
disjoint. This makes it an articulate system, in contrast to the dense nature
of undifferentiated (or, to restore the term we have been using up to now,
unsegmented) systems such as paintings.

In effect, Goodman’s view supports the notion, introduced earlier in

this chapter, that segmentation is more important than arbitrariness in
determining whether a linguistic (or some other systematic) approach
might be applied to graphic forms. ‘Descriptions are distinguished from
depictions not through being more arbitrary but through belonging to
articulate rather than to dense schemes; and words are more conventional
than pictures only if conventionality is construed in terms of differentiation
rather than artificiality.’ (p. 230)

58 Unless, presumably, it is claimed to be a copy of the first edition, in which case we are considering
it as qua printed object or investment, not qua novel. More seriously, while we may accept Goodman’s
analogy at face value in the context of his argument, there are important exceptions. Considerable
numbers of writers – mostly poets, but also some novelists – have taken a detailed interest in the
typography of their work, or have used it as an integral part of their expressive repertoire. This issue is
the subject of debate among bibliographers and some further aspects of it will be discussed in Chapters
4 and 7.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 91

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

The unique position of typography, and its special interest, is that it 117
fits awkwardly into the picture/words dichotomy. When typographic
pages add spatial and graphic qualities to segmented verbal language,
they demonstrate the simultaneous use of dense and articulate symbol
systems. But Goodman, in common with other philosophers and historians
of art, seems reluctant to discuss hybrid forms, and his discussion of
diagrams – another hybrid form – is short and somewhat confused.59 His
main argument on this topic centres around a comparison between an
electrocardiogram (a diagram) with a Hokusai drawing of Mount Fujiyama
(a depiction). He suggests that ‘the black wiggly lines on white backgrounds
may be exactly the same in the two cases’. He does not provide a sample of
such a drawing (as will become apparent, Hokusai prints do not actually
look like diagrams), but Figure 3.3, supplied by Twyman (1985), represents
the principle.

Figure 3.3 A sound spectrogram which was interpreted by one viewer as a petrochemical works

Goodman appears to suggest that, although according to his notational

theory a diagram is syntactically dense, yet it is still somehow articulate
because it is still possible to distinguish between its contingent and
constitutive features:
‘The only relevant features of the diagram are the ordinate and abscissa of each of
the points the centre of the line passes through. The thickness of the line, its color
and intensity…do not matter.’ (p. 229)

In effect, Goodman is saying that the essential meaning of the 118

electrocardiogram is preserved even when these contingent factors are
altered, whereas each such characteristic of the picture (including even
the quality of the paper, conveniently forgotten earlier when the artefacts
were said to be exactly the same) contributes to its overall meaning and
ambience.60

59 Or, at least, confusing – the layman hesitates to argue with any philosopher, especially one who
advises that ‘the reader with no background in logic, mathematics, or technical philosophy may well
skim or skip [his explanation of the syntactic requirements of notation] and rely on gathering from the
applications and illustrations in later chapters the principles expounded here.’ (p. 130, footnote)

60 In fact, Goodman’s examples are not quite as clear cut as he suggests. The apparent spontaneity
of Oriental calligraphy and drawing is often rehearsed many times before the production of the final

Robert Waller • The typographic contribution to language • PhD Thesis 1987 92

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

In practice, of course, we interpret images in the light of what we

understand of their context and purpose rather than the philosophical basis
of their notationality. Conceptual art apart, electrocardiograms are not
framed and put up for auction any more than Hokusai prints are found on
clipboards at the feet of hospital beds. Goodman appears to acknowledge
the importance of context elsewhere, when he suggests that whether a
diagram is analogue or digital is determined by how we are to read it.
However, he chooses an example that seems to demonstrate exactly the
opposite. A graph produced by a barogram is analogue because, consisting
as it does of a continuous trace of a moving pen onto a moving roll of
paper, Goodman can claim that it is syntactically dense: every point on
the line represents real data. But if the curve merely joins up separate
data points, representing, say, annual car production over a decade, it is
syntactically disjoint and therefore digital. This seems to ignore the fact
that an important purpose of such graphs is to convey analogue information
about trends. Although produced from a finite number of data points and
connectives, in practice we read the car production graph as if it were a
continuous curve of data. That is the raison d’être of this type of graph: to
reveal the underlying trends among separate data points. Indeed, we would 119
read the (analogue) trend even if the data points were not joined up, as in a
bar chart. In such cases the relationship between components of the image
can be considered constitutive. The whole may similarly be greater than the
parts in the case of typographic layouts: pages are assembled from separate
components but communicate structural information through their overall
shape or order. Like graphs, they may be syntactically articulate in their
construction but not in interpretation.

That analogue relationships may be constitutive can be seen by comparing

Figures 3.4a, 3.4b and 3.4c which all purport to demonstrate the same
principle (of selective perception). One is supposed to be able to see each as
either a rabbit or a duck but not both at the same time. First demonstrated
in a German magazine in a realist style, the duck-rabbit illusion has become
a standard demonstration among, as can be seen from the sources of these
examples, art critics, psychologists and philosophers. Two of them palpably
fail.

version; in other words, it is highly schematized. Moreover, as Schapiro (1969) points out (albeit in
relation to Chinese art) the blank space, or ground, against which the image appears, is not considered
constitutive of the image in the same way as it might be to Europeans: ‘In China where painting was a
noble art the owner did not hesitate to write a comment in verse or prose on the unpainted background
of a sublime landscape and to stamp his seal prominently on the picture surface.’ Conversely, apparently
contingent aspects of graphs, such as thickness of line and colour, are the subject of substantial research
and debate (for example, Tufte 1984). Furthermore, if an electrocardiogram exhibited variations in the
thickness of the line (said to be constitutive in the case of the Hokusai), the machine might well be sent
for repair and its output regarded with suspicion.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 93

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

Figure 3.4 This illustration is frequently used to demonstrate the principle of selective perception: you can read
it as a duck or a rabbit but not both at once. The illusion works quite well in 3.4a (from Gombrich 1960). In 3.4b
(Wittgenstein 1958) I can see the duck, but the rabbit is rather strange. In 3.4c (Bruce & Green 1985) I am quite
unable to see anything except a stylized squid.

They fail because, although to their authors they have become tokens of
a familiar argument, they don’t include enough essential or constitutive
information to the new reader. To those familiar with it, it seems, there
are just three ingredients to the picture: a head, an eye and a bill/ears
feature. Any sketch containing these three features presumably signals
the duck-rabbit illusion schema adequately to those for whom it has long
ceased to be effective as an illusion anyway. 120

To the unfamiliar, though, the exact spatial relationship between the

ingredients is itself a constitutive feature without which, as the second and
especially the third versions show, the illusion is ineffective. Again, if the
purpose of the picture was to signpost the rabbit cage in a zoo, or to depict
a particular rabbit, our criteria would change further.

Ivins includes a similar example from the sixteenth century (Figure 3.5).
When illustrations had to be copied from book to book, copyists transferred
only what they saw as the major constitutive components (leaves, stalks,
flowers), treating the relationship between the components and their exact
shapes as contingent. Although the picture still symbolized ‘plant’, and no
doubt helped to sell the book, the illusion (of verisimilitude, in this case)
was lost. Again, although the constitutive/contingent distinction is a useful
one, it is clearly relative to particular purposes.

Figure 3.5 Both of these sixteenth-century woodcuts depict violets. Ivins (1953) compares the over-rationalized
example on the left (from the Grete Herbal, 1525) with the naturalistic example on the right (from Brunfel’s
Herbarum vivae eicones, 1530). However, while the earlier woodcut is clearly over-schematized, the later one is
possibly too realistic: it is debatable whether the wilting leaves are really constitutive of the species represented.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 94

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

In a well known analysis of an advertisement for packaged food, the literary

critic and semiologist Roland Barthes (1977) resolves the question of
holistic or analogue qualities by treating them as just another sign:
‘Even when the signifier seems to extend over the whole image, it is nonetheless a
sign separated from the others: the “composition” carries an aesthetic signified, in
much the same way as intonation although suprasegmental is a separate signifier 121
in language. Thus we are here dealing with a normal system whose signs are
drawn from a cultural code (even if the linking together of the elements of the
sign appears more or less analogical).’ (p. 46)

A good analogy might be piano music. Since it is produced by separate

key strokes, and conventionally divided into regular bars, it is clearly
segmented and can, to a degree, be analysed in terms of the relationship
between different notes. But although computers might be able to recognize
a melody from such an analysis, most of us can only do so by hearing it
played. We recognize the overall ‘shape’ made by the notes, together with
other non-separate (or ‘suprasegmental’) features, such as crescendos
and rhythms. Barthes’ suggestion is that such patterns, since they are
meaningful to us, are themselves signs, even if they cannot easily be
segmented for analysis.

Barthes’ use of the term ‘cultural code’ indicates that, like Gombrich, he
does not so much analyse the physical marks (notational or otherwise)
that make up a picture as the cultural significance of the objects portrayed
and the manner of their portrayal. Whether a picture is rendered
photographically or through one of Ivins’ syntaxes, whether we are looking
at reality or depictions, we can still distinguish between separate objects. As
Gombrich remarks elsewhere:
‘We could not perceive and recognize our fellow creatures if we could not pick out
the essential and separate it from the accidental.’ (Gombrich 1982: 106)61

Whether or not we can apply the linguistic method to all of culture – for
no aspect of existence escapes the semiologist’s eye – is another matter
entirely.62 Linguistics can be seen as just a prototype of the broader
structuralism that has become a dominant metaphor for twentieth-century 122
thought.63 However, the existence of formal semiotic codes is not the issue
here, although it will be considered further in Chapter 5. For the time being

61 Gombrich use of these terms reminds us that Goodman’s distinction between constitutive and
contingent features echoes that between essential and accidental properties in Aristotelian logic.

62 I freely confess my alignment with the intellectual cowards chided thus by Sturrock (1986: 89):
‘This dramatic extension of the semiotic field, to include the whole of culture, is looked on by those
suspicious of it as a kind of intellectual terrorism, overfilling their lives with meaning.’

63 De Saussure (1916/1974: 68) saw language as the paradigm symbol-system: ‘Signs that are wholly
arbitrary realize better than the others the ideal of the semiological process; that is why language, the
most complex and universal of all systems of expression, is also the most characteristic; in this sense
linguistics can become the master-pattern for all branches of semiology although language is only one
particular semiological system.’ (my emphasis)

Robert Waller • The typographic contribution to language • PhD Thesis 1987 95

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

it is enough to note that there is consistency and pattern in the world, that
many human activities are highly conventionalized and that we have a
remarkable capacity for inferring meaning from all sorts of circumstances.

The problem of linearity

According to de Saussure:
‘While [the linear nature of the signifier] is obvious, apparently linguists
have always neglected to state it, doubtless because they found it too simple;
nevertheless, it is fundamental, and its consequences are incalculable.’ (p. 70)

Linearity is fundamental for de Saussure because it is the basis for one

of his two fundamental categories of linguistic relations: syntagmatic (as
distinct from associative, often referred to as paradigmatic). Syntagmatic
relations are the relations that a word has with others in the linear string,
or syntagm; associative relations are those that a word has with others that
might take its place in the string.64

Barthes (1964/1967) applies de Saussure’s associative/syntagmatic

dichotomy to a challenging range of semiotic systems, including clothing,
meals and furniture. He is able to do so with little difficulty because all 123
of these systems consist of discrete elements (ie, garments, dishes and
chairs etc). Later, though, (Barthes 1981) he talks of iconic syntagms, by
which he means an analogical representation that cannot be subdivided,
but which can be treated as if it were a verbal syntagm (that is, a sentence
or paragraph). A verbal syntagm is indeed a cluster of signs, but it is
essentially a linearly organized cluster, and there seems to be little point
in transferring the term ‘syntagm’ to other contexts if its associations with
linearity are lost. Van Sommers (1984: 1) suggests that the sequence of
production can be considered as a graphic parallel to the syntagmatic
dimension of verbal language: the artist can only do one thing at a
time. However, this may only work in the context of his own fascinating
experiments on the way people draw simple line images. The problem with
this definition is that, since the sequence of production is rarely apparent to
the viewer, the syntagmatic dimension can carry no semantic load. It will be
suggested later in the chapter that the graphic equivalent of the syntagm is
rather the sequence of inspection or reading.

Linearity may be an obvious feature of language, but that is not to say

that cognitively it is ideal. On the whole, it is not a great problem at the

Eagleton describes structuralism as ‘a symptom of the fact that language, with its problems, mysteries,
and implications, has become both paradigm and obsession for twentieth-century intellectual life.’
(Eagleton 1983:97)

64 For example, in the sentence ‘This is a cat’, the word ‘cat’ stands in syntagmatic relationship to
‘This is a…’ and in associative relationship to ‘pet’ or ‘animal’. A helpful ordinary-language version is
sometimes used: choice and chain.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 96

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

sentence level, where comprehension can be handled within working

memory – the beginning of the sentence is still available for processing
when the end is reached. In a lengthy text, though, readers may need to
be explicitly reminded of earlier stages in the argument which must be
retrieved from deeper levels of memory. Much of the work of text linguists
is directed towards an account of the ways in which language users
compensate for this constraint.

The ideal of a one-to-one relationship between language and ideas is part

of what has been termed ‘the language myth’ by Harris (1981). At its
heart is the ‘surrogationalist’ view – the idea that words are substitutes for
things or ideas – which elsewhere (Harris 1980) he traces from Aristotle
to present day linguistics. This is contrasted with ‘instrumentalism’ –
exemplified by speech act theory (Austin 1962) – in which language is
viewed as a multi-purpose tool, only one of whose uses is describing things 124
or making assertions.65 We might illustrate the distinction by contrasting
the Wittgenstein of the Tractatus and the Wittgenstein of Philosophical
investigations. At one point in the Tractatus, he posits an ideal language
form in which
‘the configuration of objects in a situation corresponds to the configuration of
simple signs in the propositional sign.’ (1971: §3.21)66

Wittgenstein’s suggestion was disputed by Ryle (1951: 34), however, who

cited numerous examples to show how difficult it is to see
‘how, save in a small class of specially-chosen cases, a fact or state of affairs can be
deemed like or even unlike in structure a sentence, gesture or diagram.’67

In Philosophical investigations, Wittgenstein (1958: 11) talks of ‘language-

games’, by which term he wishes ‘to bring into prominence the fact that the
speaking of language is part of an activity, or of a form of life’. Language
does not just describe states of affairs, but is used to achieve objectives – for
many of which the transfer of information is incidental. The examples given
indicate that he includes not only speaking in his definition of language but,
among other things, writing, diagramming and drawing.68

Although Westcott (1971) cites a number of examples of ‘iconic’ syntax,

in which word order reflects the order of the events described, such cases

65 Speech act theory, and its relevance to the role of typography, is discussed further in Chapter 8.

66 The development by logicians of notations and diagrams (Gardner 1958) can be seen as part of a
dissatisfaction with the ability of ordinary language to fulfil this objective.

67 It is interesting to note here that Ryle appears not to consider the two-dimensional form of diagrams
any more suited to the direct representation of fact structures than the one-dimensional form of
sentences.

68 The contrast with the Tractatus is made by Wittgenstein himself: ‘It is interesting to compare the
multiplicity of tools in language and of the ways they are used, the multiplicity of kinds of word and
sentence, with what logicians have said about the structure of language. (Including the author of the
Tractatus Logico-Philosophicus.)’ (Wittgenstein 1958: 12).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 97

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

are rare and the ‘fact structure’ (as van Dijk, 1977, calls it) of the topic 125
of discourse rarely corresponds to its linear sentence structure. With
the exception of very simple narratives, with one participant and no
overlapping episodes, most descriptive texts have to cope with information
which is in some way non-linear. Obvious examples are texts which describe
complex structures such as machines, buildings, organizations or political
situations. In such cases, an essentially multi-dimensional ‘reality’ must be
sorted into a linear string in such a way that it can be re-assembled by the
reader. In any case, even where there is a simple linear fact structure, there
may be rhetorical reasons for describing the facts in some other order.

Fact structure – the structure of a process or state (real, analogous or

imaginary) as posited by the writer – might be contrasted with argument
structure, the surface structure of a particular text, written for a particular
audience or range of audiences. We can see this distinction realized in
Halliday & Hasan’s (1976) internal and external uses of conjunctions, which
they illustrate with the following pair of examples:
a: Next, he inserted the key into the lock.

b: Next, he was incapable of inserting the key into the lock.

The same conjunction, ‘next’ refers in (a) to an event in ‘ “internal” or

situation time’, and in (b) to an event in ‘“external” or thesis time’ (Halliday
& Hasan 1976: 240). This use of ‘internal’ and ‘external’ is somewhat
confusing, though, and it is clearer to talk of (a) as an event in a sequence
of facts, and (b) as an item in a sequence of arguments (always bearing in
mind that ‘fact’ and ‘argument’ are here used rather loosely).

Ivins (1953), comparing verbal language unfavourably with pictures

describes the linearity problem in this way:
‘the very linear order in which words have to be used results in a syntactical time
order analysis of qualities that actually are simultaneous and so intermingled and
interrelated that no quality can be removed from one of the bundles of qualities 126
we call objects without changing both it and all the other qualities. […] In a funny
way words and their necessary linear syntactical order forbid us to describe objects
and compel us to use very poor and inadequate lists of theoretical ingredients in
the manner exemplified more concretely by the ordinary cook book recipes.’ (p.
63)69

And Langer (1942), contrasting presentational (pictorial or diagrammatic)

and discursive (verbal) forms says of pictures that

69 This advantage of pictures over words leads Ivins to view the development of reproducible pictures
as the most significant cultural, scientific and philosophical event since the development of writing, the
lack of which was the main ‘road block’ in the way of classical culture and science. Ivins’ preference
for objects over theories, the museum curator’s perspective, perhaps, might have led him to exaggerate
somewhat.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 98

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

‘their complexity…is not limited, as the complexity of discourse is limited, by what 3

the mind can retain from the beginning of an apperceptive act to the end of it. Of
course such a restriction on discourse sets bounds to the complexity of speakable
ideas. An idea that contains too many minute yet closely related parts, too many
relations within relations, cannot be “projected” into discursive form; it is too
subtle for speech.’ (p. 93)

Grimes (1975), a linguist whose work on discourse has been particularly

influential among cognitive psychologists, shows that even time-based
narratives are subject to the constraint of linearity, since they often involve
several participants who must be identified, and whose actions may be
related by overlapping, co-operation, causality and so on. Besides events
and participants, most narratives contain ‘non-events’, listed by Grimes as
settings, background information, evaluations and collateral information.
I shall discuss other approaches to the analysis of ‘fact structures’ in
Chapter 6.

While the linearity problem is at the heart of all text or discourse studies,
few have directly addressed it as an issue. A recent exception is de
Beaugrande (1981) whose theory of linear action is developed further in
Text production: towards a science of composition (1984), an attempt to build 127
a theoretical understanding of the writing process, and hence the teaching
of writing, on a foundation of cognitive and linguistic theory.

Principle de Beaugrande’s explanation Examples from typography

Core-and-adjunct Distinguishes between core and Typographic signalling of notes,

peripheral entities glosses, etc

Pause Allows the on-line sequence to Interpolated boxes, inserts or

be retarded or suspended footnotes

Look-back Subsumes all consultations of Regularity of layout pattern,

the prior discourse tabular structure

Look-ahead Subsumes all anticipations of the Regularity of layout pattern,

subsequent discourse tabular structure, headings

Heaviness Concerns gradations of Typographic emphasis, spatial

importance, emphasis, focus, isolation
length, salience, or novelty, in
the sense that these all draw a
‘heavier’ load on processing.

Disambiguation Deals with excluding alternative Use of layout to direct reading

patterns, both formal and sequence or to group related
conceptual items; access structures

Listing Handles the enumeration of ‘Bullets’, numbering systems,

comparable items in a sequence tabular structure

Table 3.2 Seven linearity principles (adapted to tabular form from de Beaugrande, 1984) with my suggested
application to typography.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 99

Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

De Beaugrande identifies the seven ‘linearity principles’ listed in Table

3.2.70 They comprise a framework within which he is able to relate the
various phases of cognitive processing involved in reading with the different
rhetorical and linguistic forms used by writers (as well as the cognitive
processes through which writers select and produce those forms). An 128
examination of de Beaugrande’s framework may offer some insight into the
linearity of language, how it is overcome, and, perhaps, how typographic
techniques might contribute in this respect.

The seven principles, de Beaugrande argues, govern the ways in which

writers transcribe multi-dimensional ideas into a linear linguistic form. My
‘transcribe’ telescopes de Beaugrande’s fairly elaborate cognitive model of
reading and writing into a single term, but it deserves a brief summary.
De Beaugrande criticizes earlier serial models of writing which involve a
series of discrete ‘black-boxed’ stages. Ideas progress through pragmatic,
semantic, syntactic, and lexical stages until they achieve surface expression
as phonemes or graphemes. These reflect the structure of linguistics and
are convenient for psychological experiments, but they do not stand close
analysis.71 More recent models can be described as ‘parallel interactive’,
since they allow for the different levels to be activated simultaneously. In
this context,
‘linearity reflects the organization of the language modalities of speech and
writing, rather than one-by-one mental processes.’ (p. 104)

Figure 3.6 De Beaugrande’s parallel interactive model of reading (1984)

70 De Beaugrande’s books, although apparently aimed an interdisciplinary audience, are extremely

hard going, heavily larded with citations and technical terms. Cynics may find a possible explanation
on p. 284 of his Text, discourse and process (1980), where he is discussing the use of readability
formulae for schoolbooks: ‘I consider the principle of “least effort” wholly misconceived as a standard
of human activities at large and of the reading of texts in particular. Readers will gladly expend more
effort, provided that the text awakens interest and rewards the effort with informative insights.’ (author’s
emphasis)

71 Models of reading comprehension are discussed further in Chapter 4.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 100
Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

De Beaugrande’s own model (Figure 3.6) suggests six levels of processing.

As he puts it,
‘the zig-zagged arrow suggests a gradual migration of dominance from deep to
shallow during text production, yet with considerable freedom for shifting up or
down.’ (p. 105)

The psychological problem of how parallel processes are managed is 129

not our concern, but at some stage, although originating as non-linear
conceptual networks and processed at the deeper levels in non-linear ways,
ideas must eventually be linearized at the surface level. Hence the seven
linearity principles.

De Beaugrande does not properly explain the source of the seven categories
(why seven? why these seven?), and they have a rather arbitrary feel about
them. Take, for example, the relationship between surface features of the
text and cognitive processes. In the case of the ‘look-back’ principle there
seems to be a direct link: specific backward-looking features of the surface
text are deployed in order to control or facilitate cognitive looks-back by
the reader. In the case of the pause principle, though, the link is tenuous. Its
function in cognitive processes, to cope with processing overloads, does not
appear to relate at all to its function on the text surface, where it articulates
phrase and sentence boundaries.

The diagrammatic version of the principles (Figure 3.7) indicates that they
overlap considerably in practice. The flow-of-control arrows show that, in
most circumstances, a number of different principles must operate together.
For example, following the arrows in the core-and-adjunct diagram leaves
one at the end of a line, needing to return to the core in order to follow
up the various other adjuncts in turn. In practice, a core-and-adjunct text
might be one where the author makes a proposition and then discusses
various problems and corollaries of it. First, the author must look ahead to
the sub-arguments to be presented; they may even be listed. At the end of
each sub-argument, the look-back principle would operate as readers are
reminded of the main proposition. Finally the disambiguation principle
might demand that certain arguments be dismissed and others selected for
further attention.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 101
Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

130

Figure 3.7 De Beaugrande’s diagrams illustrating his seven principles of linearity

Linearity is certainly implicit in speech, bounded by time, but writing, with

two dimensions available, can present large tracts of discourse instantly
accessible to the searching eye. Certain of de Beaugrande’s diagrams might
even serve as models for typographic layouts which would reveal the
structure of the (non-linear) argument or encourage an appropriate reading
strategy. Indeed, he does recognize that linearity in writing is spatial not
temporal. But since his topic is the composition of continuous prose for
fluent reading, it is perhaps not surprising that de Beaugrande restricts
his view to the one-dimensional spatiality of the line, rather than the
two-dimensional spatiality of the page.

The one-dimensional view of language seems to be remarkably persistent 131

among other scholars also. For example, Vachek (1967) similarly
observes the spatial dimension of writing, and similarly fails to develop
the implications of that fact. And even in a conference devoted to the
re-integration of writing into the linguistic and psycholinguistic domains,

Robert Waller • The typographic contribution to language • PhD Thesis 1987 102
Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

the following statement is reported among the discussion after a debate on

ideographic writing systems:
‘Gough asked if all writing systems are linear. Apart from early pictographic
writing, the group agreed that this was the case.’ (Kavanagh & Mattingly, 1972:
128)

De Beaugrande also appears to miss or ignore a further important

implication of the spatiality of writing, whether one- or two-dimensional.
Because it is presented in space, not time, writing offers the reader the
opportunity to physically look back, look forward, scan a list structure and
so on. Without this opportunity, long and complex arguments could neither
be easily written nor critically read. However, de Beaugrande restricts his
view to cognitive versions of those activities:
‘The processor may routinely consult the mental representation of prior text
and re-scan the surface text only on strategic occasions, e.g. for revision.’ (de
Beaugrande 1984: 175)

However parallel the cognitive processes in de Beaugrande’s model, then,

the input is still assumed to be serial. But since one of the most significant
aspects of writing is the release of the reader from the temporal linearity of
speech, there seems no reason why the cognitive psychologist’s perspective
should not be extended. De Beaugrande attributes linearity principles to
both writers and readers, so the implication is that ‘looks-back’ among
readers can be literal; that is, they can actually look back to an earlier point
in the text rather than just their memory of it.

This suggests a crucial distinction. Still taking looks-back as an example, we

might say that text features that are solely verbal ‘look back’ to an earlier
part of the linear text string in a metaphorical sense; the relationship is 132
implicit in the language and must be cognitively apprehended by the reader.
Text features that are graphic, or at least graphically signalled, transfer the
responsibility for the look-back to the reader; the relationship is explicit in
the graphic form of the text and can be perceptually apprehended by the
reader – the look-back is real not metaphorical.

Another way to express this is to say that the responsibility for the syntagm
has shifted from the writer to the reader. Given that readers of written
text can move around it at will, it seems reasonable to propose a concept
of reader-syntagm in contradistinction to the traditional syntagm which is
entirely controlled by the writer. There is a time dimension to reading, just
as there is to speaking, so however non-linear the text, the reader-syntagm
still represents a linear input to the process of cognition. The order of that
input, though, can be controlled by the reader, on the basis, perhaps, of the
visual syntax, schemata or analogical codes, discussed earlier in the chapter.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 103
Chapter 3 • Arbitrariness and linearity: de Saussure’s ‘basic principles’

A word that appears in a prose sentence is supplied to the reader in a

syntagmatic relationship to its co-text – the other words in the sentence. Its
associative relationships – with words the writer might have chosen but did
not – are supplied by the reader’s prior knowledge. Conversely, a word in a
list, a prototypical graphic configuration, stands in associative relationship
to its co-text.72 That relationship with the other words in the list is defined
or reinforced by the list’s title or introduction. Its syntagmatic relations
are supplied by the reader who can legitimately scan the list in any order.
(Readers can, of course, scan a prose sentence in any order too, but cannot
be sure of gaining any form of sense predicted by the author).

This chapter has examined de Saussure’s two principles of arbitrariness and

linearity as a means of identifying some of the main differences between 133
the idealized verbal strings studied by mainstream linguistic science and
the partly non-verbal, partly non-linear texts that we call ‘typographic’. An
important dichotomy has emerged from the discussion, between writer- and
reader-syntagm. The next chapter will review evidence of it in a range of
debates within the disciplines that comprise discourse studies. Chapter 5
will propose a simple conceptual model based on it.

72 Harweg (1987) has recently pointed out that this characteristic of lists (and other sets of words
that are not anaphorically connected) constitutes a counter-example to the normal assumption that
associative relations are an aspect of langue and syntagmatic relations are a dimension of parole.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 104
Chapter 4 • Writer-syntagm and reader-syntagm

4
134

Writer-syntagm and reader-syntagm

I concluded the last chapter by identifying a dichotomy between

conventional linear text, in which the order of presentation is largely
controlled by the writer, and typographically organized text in which the
reader is afforded a greater measure of control. Texts clearly vary in the
opportunities they offer for ‘syntagm control’. Continuous prose, especially
in the form of novels, offers few visible structural cues to readers wishing
to control their own pace. We are normally expected to read a novel from
beginning to end; to do otherwise we need specially annotated study
editions, or our own marginal notes and underlinings. A table, on the other
hand, cannot sensibly be read in a linear order from top left to bottom
right. In between these extremes lie dictionaries, reference manuals,
textbooks – and the various examples proposed by Bernhardt (1985) in his
continuum of visual informativeness.

The distinction between writer-syntagm and reader-syntagm reflects

a number of similar dichotomies emerging from other approaches to
communication. By examining some of them, we may gain some insight
into the nature of the dichotomy and how to resolve it.

Methods of configuration: linear vs non-linear

One of the main axes of Twyman’s (1979) schema for the study of graphic
language represents
‘methods of configuration…by which is meant the graphic organization or
structure of a message which influences and perhaps determines the “searching,”
“reading,” and “looking” strategies adopted by the user.’ (Twyman 1979: 119–121)

Twyman thus proposes a direct relationship between the configuration of a 135

graphic display and the degree of control enjoyed by the reader. Figure 4.1
reproduces Twyman’s schema.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 105
 Chapter 4 • Writer-syntagm and reader-syntagm

Method of configuration

Pure linear Linear List Linear Matrix Non-linear Non-linear

interrupted branching directed most options
viewing open

Verbal/ 1 2 3 4 5 6 7
numerical

Pictorial 8 9 10 11 12 13 14
Mode of symbolisation

& verbal/
numerical

Pictorial 15 16 17 18 19 20 21

Schematic 22 23 24 25 26 27 28

Figure 4.1 Twyman’s schema for the study of graphic language (Twyman 1979). In its original context, the schema
is used to organize a large number of examples of graphic displays, in order to present a broad perspective of the
range of graphic options available.

Purely linear configurations are so rare that for practical purposes they can
be ignored. The presence of the linear category is important, though, to
emphasize the next category along: it is sometimes forgotten that texts we
may think of as purely linear are actually linear interrupted. Although most
interruptions are arbitrary (discounting, for the moment, the fact that we
normally break lines only between words or syllables), there are exceptions
to the rule, and I shall return to such arbitrary or artefactual effects later in
the chapter, and in Chapter 8.

For now, though, we can see linear interrupted text as representing the
writer’s exercise of strong control over the reader’s use of a document.
The non-linear categories represent much weaker control, and in the case
of non-linear most options open, virtually no control either in terms of
the topic-related focusing that I take to be the basis of non-linear directed
viewing, or in terms of the reading-rules implicit in list, linear branching and 136
matrix. With the possible exception of numbered lists, in all categories to
the right of linear-interrupted, the reader exercises most of the control over
the order of presentation. Matrices are a particularly clear demonstration of
the reader-syntagm. As many ‘propositions’ can be generated from a matrix
as there are cells, quite apart from general observations about patterns

Robert Waller • The typographic contribution to language • PhD Thesis 1987 106
Chapter 4 • Writer-syntagm and reader-syntagm

among the data.73 Table 4.1 describes some reading strategies that might
be implicit in these formats.

Configuration Implied reading rules

Pure linear Start at the beginning and carry on until the end.

Linear interrupted Start at the beginning and carry on until the end
disregarding interruptions, which are arbitrary; at each
interruption, carry on reading on the next line, column or
page.

List If the list is numbered, start at the beginning, taking note of

the interruptions, which are meaningful and separate the
parts within a whole. If the list is unnumbered, the items can
be read in any order.

Linear branching Start at some other relevant point, and let your response to
what you are reading determine which (connected) part you
read next.

Matrix Select one heading from each axis of the matrix and look
at the cell formed by their intersection; or vice versa. Or
compare all the cells for a particular row or column. Or
compare the contents of whole rows or columns.

Non-linear directed viewing Start at the focal point(s) in the display and carry on as
instructed or as seems reasonable.

Non-linear most options open Do what you like.

Table 4.1 Twyman’s methods of configuration (Twyman 1979), with my conjectured reading rules.

It is obviously not realistic to suggest that all actual documents will

only employ a single configuration. Applied to the layout in Figure 4.2,
for example, we might detect elements of linear interrupted, list, linear
branching and non-linear directed viewing.

73 Wright (1981) has described some of the sub-skills required by readers of tables.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 107
Chapter 4 • Writer-syntagm and reader-syntagm

137

Figure 4.2 In this page (from Bond 1980), the introductory paragraph is linear interrupted; the main structure is
linear branching (centre mainsheet vs aft mainsheet), while the boxed item on knots has no particular place in any
linear sequence. It is highlighted in blue, however, and we could regard its placement as an example of non-linear
directed viewing. The schematic drawings at the top right certainly come under that category, since the relevant
parts of the rigging are highlighted in blue.

Reading strategy: receptive vs self-organized

The distinction between writer- and reader-syntagm can also be seen in
texts that are a great deal less visually informative than matrices, and even
in continuous prose. Thomas & Harri-Augstein (1980) and Pugh (1979),
who observed the reading strategies of students (that is, the order and pace
of their progress through a text), found that readers who skimmed ahead,
re-read, and changed pace frequently were more effective in achieving
their goals than those who simply read straight through at an even pace.74
The effective readers, we might say, had taken control of the situation.

74 Thomas & Harri-Augstein’s Brunel Reading Recorder requires subjects to wind a handle in order to
move a roll of text past a window through which they can read. The handle also moves a pen across
a sheet of graph paper which travels at a constant rate. Whalley & Fleming (1975) reported a less
intrusive device based on a light-pen. A number of other techniques for observing reading behaviour are
compared by Schumacher & Waller (1985).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 108
Chapter 4 • Writer-syntagm and reader-syntagm

Figure 4.3 shows five reading types described by Thomas & Harri-Augstein 138
from reading protocols obtained with a special apparatus; actual reading
records consist of combinations of these five types. This uneven, purposeful
style of reading is termed self-organized by Thomas & Harri-Augstein,
and self-paced reading by Pugh. Its opposite – reading which follows
the author’s sentences and paragraphs in a linear and even manner – is
generally described as receptive reading.

text

text

text
time time time

a: smooth read b: item read c: search read

text

text
time time

d: theme session e: check read

Figure 4.3 Five types of read, identified by Thomas & Harri-Augstein (1980).

Quite apart from such direct evidence, the importance of self-organized

reading can be seen by most literate people from an introspection of
their own reading habits, and from indirect evidence of other kinds.
An unpublished study at the Open University in 197575 showed that no
matter how much the reading load varied from week to week on the Arts
foundation course, the study times reported by students did not vary by
more than about 10–20%. Average study times were between 10 and 12
hours, although the workload varied between 12,000 and over 40,000
words. Although there are other relevant factors (such as the time taken for
exercises and assignments), it was clear that students faced with impossibly
long reading lists were adopting highly selective strategies – indeed this
is a frequently-stated teaching goal of university teachers accused of
overloading their students.

Hatt (1976) reviewed research on the reading process from the viewpoint 139
of a librarian, criticizing the usual transmitter-message-receiver model of
communication (discussed in more detail in the next chapter). Librarians,

75 The study was carried out by Michael Macdonald-Ross and Alice Crampin of the Open University
Institute of Educational Technology.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 109
Chapter 4 • Writer-syntagm and reader-syntagm

of course, deal with precisely the opposite direction of flow – ‘receivers’

in search of information, for many of whom, Hatt suggests, it may be
irrelevant that messages have ‘transmitters’ or authors at all. As an
alternative he suggests a model of the reading process that ‘makes the
reader the subject’. Like the one it replaces, Hatt’s model boils down to a
beginning, middle and end:
‘1 A reader finds a text
2 He reads the text
3 He uses the message (or not, as the case may be).’ (Hatt 1976: 20)

Hatt points out that virtually all studies of the reading process focus on
stage 2 to the virtual exclusion of stages 1 and 3. His own review, which
is well-written and perceptive but appears to be little noticed, therefore
concentrates on identifying various routes by which readers reach texts and
exit them. His taxonomy of exit patterns is reproduced, without most of his
examples and comments, in Table 4.2.

Pattern 1
1 The reader perceives the text
2 The reader duplicates the text

Pattern 2
1 The reader perceives the text
2 The reader decodes the text
3 The reader accepts the message
4 The reader discards the message

Pattern 3
1 The reader perceives the text
2 The reader decodes the text
3 The reader accepts the message
4 The reader uses the message to confirm an attitude or opinion
5 The reader discards the message

Pattern 4
1 The reader perceives the text
2 The reader decodes the text
3 The reader accepts the message
4 The reader retains, in his store, knowledge taken from the message

Pattern 5 140
1 The reader perceives the text
2 The reader decodes the text
3 The reader accepts the message
4 The reader modifies his cognitive structure to accommodate new
knowledge in the message
5 The reader retains the new knowledge in his modified cognitive
structure

Robert Waller • The typographic contribution to language • PhD Thesis 1987 110
Chapter 4 • Writer-syntagm and reader-syntagm

Pattern 6
1 The reader perceives the text
2 The reader decodes the text
3 The reader accepts the message
4 The reader uses the message to change an attitude or opinion
5 The reader discards the message
NB Hatt lists two variations of Pattern 6 in which he replaces step 5 with
the outcomes of Patterns 4 and 5.

Pattern 7
1 The reader perceives the text
2 The reader decodes the text
3 The reader accepts the message
4 The reader makes a decision, on the basis of the message

Pattern 8
1 The reader perceives the text
2 The reader decodes the text
3 The reader accepts the message
4 The reader makes a decision, on the basis of the message
5 The reader performs an action

Pattern 9
1 The reader perceives the text
2 The reader decodes the text
3 The reader accepts the message
4 The reader originates a new message
NB Hatt notes that Patterns 7, 8 and 9 will only occur in combination with
one of the earlier patterns

Pattern 10
1 The reader perceives the text
2 The reader decodes the text
3 The reader rejects the message

Pattern 11
1 The reader perceives the text
2 The reader decodes the text
3 The reader distorts the message
4 The reader accepts the distorted message

Table 4.2 Hatt’s (1976) list of eleven patterns of exit from the reading act

It is possible that the ability to organize one’s own reading effectively

may be related to a further set of dichotomies identified by educational
psychologists. Cognitive style describes personality differences between
individuals that affect their approach to learning tasks (a recent review is
by Shipman & Shipman 1985).

Educational technologists have traditionally seen curricula in terms of 141

learning objectives: testable skills or knowledge that students are expected
to attain. In a seminal book, The conditions of learning, Gagné (1965)
proposed that objectives are best thought of in terms of hierarchies,
in which each objective is broken down into sub-objectives that must
necessarily or logically be attained before the higher-level objective can be

Robert Waller • The typographic contribution to language • PhD Thesis 1987 111
Chapter 4 • Writer-syntagm and reader-syntagm

reached. An important contemporary, Ausubel (1963), proposed a similarly

hierarchical ‘theory of meaningful verbal learning’ in which he suggested
the use of ‘advance organizers’ which outline the ‘ideational scaffolding’ of
superordinate concepts on to which lower-level concepts can be fitted.76
However, a widely-cited study by Mager (1961) reported that instructional
sequences planned by instructors (of electrical engineering) were often
different from those elicited from learners. While instructors built up an
explanation of, say, radio engineering from a sequence of basic scientific
facts and theories, students were actually motivated by their practical
curiosity about familiar objects (radio valves, for example).

To some degree the provision of objectives or advance organizers is

intended to impose uniform goals on students. But it is clear from Mager’s
study that, even if students are thus provided with the same starting point,
given a choice they also differ in the order in which they study concepts
within a hierarchy – thus no single order prescribed by the instructor, or
writer, is likely to suit all students, no matter how homogeneous their
prior knowledge. Like Pugh and Thomas & Harri-Augstein, Pask & Scott
(1972) used study protocols to replicate Mager’s findings. They relate
their observations to one of a number of different dichotomies identified
by researchers into cognitive style, that between holists and serialists.
According to Pask & Scott, holists prefer to gain an overview of the total
subject, and only to learn the details once their context is established;
serialists prefer to build up their knowledge from the bottom of the 142
hierarchy of objectives. Holists, we could say, prefer to take control over
their study sequence, while serialists may be content to accept concepts
in whatever order they are given, and if necessary delay their full
comprehension and integration.

Eye movements: foveal vs peripheral

Thomas & Harri-Augstein obtained reading records (or ‘protocols’ as
they are sometimes known) at a very broad level of analysis, but reading
protocols at a much more detailed level, obtained with eye-movement
cameras (reviewed by Morrison & Inhoff 1981), also show uneven patterns
of reading. Records of saccades77 typically reveal an uneven pace with
occasional regressions. In fact, Rayner (1978) reported that 10–20% of
saccades are regressions.

76 Ausubel’s theory has inspired a considerable number of experimental studies, but the results are
inconclusive (Barnes & Clawson 1975). They are of some relevance to typographers, firstly,because they
study the effects of a text component (the advance organizer) that is typographically distinguished from
its accompanying text; and, secondly, because Ausubel’s theory is based on an essentially diagrammatic
metaphor – hierarchies of concepts. Literally diagrammatic versions have also been tried (Jonassen &
Hawk 1984).

77 Saccades are quick jumps by which the eye moves across a line of type between fixations; it is during
the fixation that readers get information from the written characters.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 112
Chapter 4 • Writer-syntagm and reader-syntagm

It is notable that if one were to write out the linguistic inputs encountered
by these readers’ cognitive processes, they would look remarkably similar to
transcripts of speech – just as broken and seemingly incoherent. The false
starts, repetitions and incomplete sentences that represent the speaker’s
(and writer’s) fumbling attempts to unravel complex ideas into linear form
seem to have a counterpart in readers’ attempts to reassemble the meaning.
And if this is a valid comparison, it could indicate a functional basis for
language behaviour that is sometimes considered inarticulate and confused:
both speakers and readers are attempting to sort relevant from irrelevant
information, to try different routes before choosing the right one, and to
monitor their own or their hearer’s comprehension.

Eye movement research also suggests a literal application of Polanyi’s

(1969) metaphorical distinction between focal and subsidiary awareness in
the context of scientific problem-solving (see Chapter 2). Although much
of the earlier eye movement research concentrated on foveal vision (the 143
fovea is the part of the retina which has highest acuity) or, in layperson’s
terms, focused vision, recent studies have emphasized the importance of
peripheral and parafoveal vision (the intermediate zone between focused
and peripheral vision). For example, McConkie & Rayner (1975) found
that the length of saccades can be influenced by information about word
length available through parafoveal vision, and Rayner (1975) showed that
the presence of boundary letters (at the beginnings and ends of words)
affects fixation duration. So, far from being passive recipients of visual
information, it seems that we make some strategic choices even at the
relatively automatic eye-movement level of the reading process.

The inter-connectedness of focal and subsidiary (peripheral) awareness

is easier to see through introspection at higher levels of reading. When
we read one item from a list, for example, our focusing on that item only
makes sense within a subsidiary awareness of the whole. In the case of
tables – intersecting lists – our subsidiary awareness is more complex still.
Since language is a system of contrasts, there is, of course, an obvious
relationship between any linguistic unit and its accompanying co-text –
between figure and ground, we might say.

Reading comprehension: bottom-up vs top-down

Models of reading comprehension suggest cognitive reasons for these
observations of the reading process.78 Some models of reading (and

78 It should be noted at the outset of the brief discussion that follows that research into reading
comprehension has produced a vast and extremely complex literature. Gough (1984: 246), in a recent
review of research on word recognition, notes the pessimistic view that ‘we are learning more and
more about less and less’. In addition to other sources cited here, a number of chapters in Barr, Kamil &
Mosenthal (1984) provide accessible and thorough reviews of the field.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 113
Chapter 4 • Writer-syntagm and reader-syntagm

perception, and other cognitive activities) are known, usually by their

detractors, as bottom-up models. They see comprehension as essentially
data-driven and hierarchical. Gough (1972), for example, maintains that
decisions about letters precede decisions about words, and that decisions 144
about words precede decisions about sentences. Although out of context,
the following short quotations give something of the flavour of an
unabashed bottom-up model of reading (produced, it should be said, in the
context of a debate):
• ‘Reading begins with an eye fixation’ (p. 331)

• ‘letters are recovered…as letters [and] the evident effects of higher levels of
organization (like spelling patterns, pronouncability, and meaningfulness) on word
recognition and speed of reading should be assigned to higher, and later, levels of
processing.’ (p. 334, my emphasis)

• ‘I see no reason, then, to reject the assumption that we do read letter by letter.’
(p. 335)

Gough is able to cite empirical evidence in support of his model, although

mostly obtained in rather artificial laboratory experiments – individual
letters and words can, after all, be recognized (and thus studied and
compared by psychologists) out of context. However, once reading is
studied within an ecologically more valid methodology, using meaningful
text, it becomes apparent that real reading probably does not progress in
the orderly linear way assumed by the bottom-up models.

Gough’s model is contested by Brewer (1972) who cites alternative

evidence, much of it from cognitive psychology’s first incarnation in the
1880s (Venezky, 1984, calls it the ‘golden age’ of reading research). Cattell
(1885/1947), for example, demonstrated that words in prose can be read
almost as fast as lists of letters; from this and other evidence he concluded
that letters must be processed in parallel, not serially. Brewer also notes
that Gough’s model, which has graphemes mapped directly onto phonemes
before word recognition can take place, cannot account for evidence of the
direct processing of words without a phonological stage. We have already
considered some of the evidence for this in Chapter 3.

The alternative top-down approach is explained quite well by the title of

Goodman’s influential paper, ‘Reading: a psycholinguistic guessing game’ 145
(1970). Goodman, who analyzed errors made by beginning readers, found
that children faced with unfamiliar words would often guess their meaning
from the context; this indicated to him that reading involves the constant
generation and testing of hypotheses about forthcoming text. Top-level
cognitive activity (the expectation or construction of meaning) therefore
precedes the bottom-level activity of word recognition instead of the other
way around. A similar view is represented in the same publication by

Robert Waller • The typographic contribution to language • PhD Thesis 1987 114
Chapter 4 • Writer-syntagm and reader-syntagm

Hochberg & Brooks (1970), who stress the purposeful, sampling nature of
‘reading as an intentional behavior’.

Most reading theorists now regard these approaches as extreme positions,

and admit elements of both into their models (Samuels & Kamil, 1984,
review recent developments in modelling the reading process). Such models
(an example is that of de Beaugrande, 1984, discussed in Chapter 3) are
known as ‘interactive’. The most influential of the interactive models was
proposed by Rumelhart (1977) who posits a constantly shifting relationship
between bottom-up, or data-driven, processes and top-down conceptually-
driven processes. The domination of one over the other depends on
readers’ familiarity with the topic, their reading purpose and the extent to
which the text matches their level of reading skill and lexical knowledge.
Interactive models, then, suggest a purposeful, top-down process at the
strategic level with bottom-up processes only emerging from the automatic
or subconscious levels when unfamiliar or difficult text is encountered.
An example of this is the ‘articulatory loop’ described in Chapter 3 – an
optional sub-system only used in cases of special difficulty (Baddeley 1984).

Interestingly, Rumelhart blames the unrealistically serial nature of

bottom-up models on the inadequacy of flow-diagrams,79 whose linear
structure, he claims, is unsuited the representation of parallel interactive
processes. In common with others investigating the cognition of natural
language, Rumelhart’s alternative is computer-modelling; research into 146
reading comprehension may be passing into the domain of artificial
intelligence.

Oral vs silent reading

Rumelhart’s approach seems intuitively rather more reasonable than the
extreme positions, and so it is not surprising to find that it is, to a degree,
a restoration of the model that was current before the behaviourist
domination of Anglo-American psychology, to which the more extreme of
the top-down theories were reacting. Huey’s Psychology and pedagogy of
reading (1908), which summarized the research of the previous thirty years,
was republished in 1968 and still seems remarkably modern. Another early
twentieth century psychologist who, like Huey, was reprinted in the early
1970s just as cognitive psychology was regaining the high ground, was
Thorndike (1917/1971), the title of whose paper ‘Reading as reasoning’
could also sum up the modern view. Thorndike argues that
‘we should not consider the reading of a text-book or reference as a mechanical,
passive, undiscriminating text, on a totally different level from the task of
evaluating or using what is read. While the work of judging and applying

79 Calfee (1981: 8) has also commented on the confused diagramming practices of cognitive theorists.
Whalley (1984) discusses problems of knowledge representation in some detail.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 115
Chapter 4 • Writer-syntagm and reader-syntagm

doubtless demands a more elaborate and inventive organization and control of

mental connections, the demands of mere reading are also for the active selection
which is typical of thought.’ (1971: 433)

Interestingly, Huey’s book and Thorndike’s paper, among other publications,

have been considered largely responsible for the move in the inter-war
years away from a view of reading as the simple ability to translate
written symbols into speech (the often mindless activity aptly known as
‘barking at print’) towards the encouragement of silent reading in which
comprehension is the main criterion of success.80 So at exactly the time 147
that linguists were relegating written language to a secondary status,
educational psychologists were promoting it as a source of direct access to
meaning.81

It is sometimes assumed that silent reading is a comparatively recent

development: Pugh (1978) even suggests that it is a nineteenth-century
development although this seems an exaggeration.82 The problem may be
that it is assumed that the way people describe or teach the reading process
corresponds to the way they actually read. An oral or subvocalized model
might have been the best explanation around, rather than an accurate
description.

Some have seen technological causes for the shift from oral to silent
reading. The idea that the introduction of printing led to a radical change
from an oral to a visual culture, from oral reading in groups to silent
reading by individuals, has been suggested by Chaytor (1945), popularized
by McLuhan (1962), and developed more recently by Ong (1967, 1982).
However, recent reviews of medieval book design (Parkes 1976, Evans
1980, Saenger 1982, Camille 1985, Gullick 1986) indicate that the notion
of a purely oral medieval society is an over-simplification, since complex
non-linear page layouts were common.

In classical times, it is possible that the lack of punctuation and word

division (Thompson 1892, 1912) did impose a largely oral process on both

80 Pugh (1978) and Allington (1984) provide historical background to this debate.

81 Sounding a cautionary note about the move away from oral reading, Dearborn, Johnston
&Carmichael (1949) reported evidence that the presence of oral stress contributes to language
comprehension. Readers who are unable to assign stress to the correct words in a sentence (that is,
where the author would have stressed those words when reading aloud) comprehended less. They
appeal for the use of typographic variation to indicate vocal stress in print – a combination of the
‘typographic cuing’ and the ‘atmosphere value’ suggestions reviewed in Chapter 1.

82 Pugh does not cite any direct evidence for his assertion that ‘silent reading was not a common
activity in schools or elsewhere before the middle of the nineteenth century’ (p. 12). However he does
cite Chaytor (1945) as suggesting a possible reason for the development of silent reading – the fact
that the British Museum reading room would be intolerably noisy if everyone read aloud. But this is to
misunderstand the context of Chaytor’s remark: Chaytor is simply contrasting the open layout of modern
libraries with the design of a medieval library which had carrels to protect readers from the noise of
other readers.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 116
Chapter 4 • Writer-syntagm and reader-syntagm

reading and writing (by dictation).83 Although Chaytor is confident that ‘no 148
one is likely to contest the statement that the invention of printing and the
development of that art mark a turning point in the history of civilization’
(p. 1), Saenger (1982) regards the introduction of word separation as at
least as significant. Word separation – described by Saenger as ‘the singular
contribution of the early Middle Ages to the evolution of Western written
communication’ – allowed silent copying of manuscripts, which in turn
led to the growth of silent reading. The real impetus for silent reading,
according to Saenger, came not from printing but from the functional
requirements of the growth of scholasticism in the twelfth and thirteenth
centuries. The intellectual demands on readers of long books which were
heavily glossed and sometimes diagrammed could only be handled by the
relatively faster technique of silent reading. It seems reasonable to agree
with Saenger’s view that
‘The complex structure of the written page of a fourteenth-century scholastic text
presupposed a reader who read only with his eyes, going swiftly from objection
to response, from table of contents to the text, from diagram to text, and from the
text to the gloss and its corrections’ (Saenger 1982: 393).

However, we cannot know just how swiftly readers went from objection
to response – according to the much-quoted Rule of St Benedict (Chaytor
1945: 10), monks were allowed one book at a time and a year in which to
read it. That may not have been representative of all situations, of course,
but however fast medieval readers went, those living in the era of printed
books certainly have more ground to cover. Eisenstein (1979: 72) remarks
that with the availability of printed books
‘successive generations of sedentary scholars were less apt to be engrossed by a
single text and expend their energies in elaborating on it. The era of the glossator
and commentator came to an end and a new “era of intense cross referencing
between one book and another” began.’84

Punctuation theory: dramatic vs grammatical 149

The debate about oral and silent reading is a cousin of the debate
surrounding the phonological equivalence of phoneme and grapheme.
It recurs in relation to another aspect of the printed word – punctuation.
The history of punctuation, it is often suggested (for example, Honan 1960,
Partridge 1953), presents us with yet another dichotomy – between the
dramatic and the grammatical.

83 Thompson makes it clear that word spacing was sometimes used ‘in the course of documents of
ordinary life, written cursively’ (1912: 56). From a modern perspective it is puzzling to find that the
advantages were not immediately recognized and adopted universally. Presumably it was seen as a
compensation for the relative illegibility of cursive script that would be redundant in professionally
produced manuscripts.

84 Eisenstein attributes this quotation in a footnote to Denys Hay, ‘Literature, the printed book’, p. 366,
in The new Cambridge modern history, edited by GR Elton (Cambridge, 1958).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 117
Chapter 4 • Writer-syntagm and reader-syntagm

Chauvier (1849) expresses it thus:

‘Those who wrote of old…doubtlessly marked the proper divisions of meaning,
but they were essentially orators, and their punctuation would be chiefly
oratical. Modern authors, on the contrary, write their works for the press, or
to be considered in the retired and silent closet...their punctuation therefore is
exclusively grammatical.’ (p. 4; the last two emphases are mine)

Dramatic punctuation scores the ‘performance’ of a text for reading aloud.

Question and exclamation marks obviously indicate intonation, while
commas, semi-colons, colons and full stops indicate different lengths of
pause. Dramatic punctuation is therefore intended for language that is to
be heard, whether aloud or by subvocalization. It reached its heights in
Elizabethan and Jacobean drama (where, after all, dramatic punctuation
is entirely appropriate). Simpson (1911) used the dramatic principle
to defend Shakespeare’s printers against the charge that they could not
punctuate – modern editors feel free to adjust the punctuation of the First
Folio as much as they do its spelling.85 Often, grammatically unnecessary
– even disruptive – commas turn out, when interpreted by Simpson, to
have an important rhythmic or semantic role. And, conversely, one of his
most powerful examples is of apparent under-punctuation. In the edition of
Shakespeare on my bookshelf,86 Pistoll (who is being forced to eat a leek by 150
Fluellen) says:
‘By this leek, I will most horribly revenge: I eat and eat, I swear – ’ (Henry V, Act V,
scene II, 49-50)

The First Folio, though, prints it thus:

‘By this Leeke, I will most horribly revenge I eate and eate I sweare.’87

Simpson’s comment includes a nicely literal interpretation of ‘pointing’:

‘It is a pity to clog this disordered utterance with the puny restraint of commas.
The words come wildly from the victim while he writhes and eats and roars, and
Fluellen’s cudgel supplies a very satisfactory punctuation for them.’ (p. 12)

In contrast to this performance-related punctuation, grammatical

punctuation has the aim of clarifying the structure of the sentence.
Here, the various stops indicate the status of text boundaries: the words
‘comma’ and ‘colon’ actually originated as units of sense. 88 Through this
style of punctuation, then, the reader can see the hierarchical structure
of sentences. It is not stretching the argument too far to suggest that the

85 Further aspects of editorial intervention are considered in Chapter 7.

86 WG Clark & W Aldis Wright (eds) The complete works of William Shakespeare, New York: Nelson
Doubleday.

87 I have changed the ‘u’ in ‘reuenge’ and the long ‘s’s.

88 According to Thompson (1912: 70) ‘Suidas explains a colon as a στιχος [stichos] forming a complete
clause; Joannes Siculus lays it down that a clause of less than eight syllables is a comma, and that one of
from eight to seventeen syllables is a colon.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 118
Chapter 4 • Writer-syntagm and reader-syntagm

dramatic/grammatical distinction reflects the writer-control/reader-control

dichotomy that forms the topic of this chapter. The visible structure of
prose, even in the relatively discreet form of punctuation, can be used as a
basis for free, reader-controlled movement around the text89 in a way that
was not possible before punctuation and word spaces were introduced.
Interestingly, dramatic and grammatical styles of punctuation correspond
directly with the two kinds of paralanguage defined by Lyons (1977) –
modulation and punctuation. That both are functions of punctuation (used 151
in the sense of ‘pointing’) explains why I substituted the word segmentation.

Some manuals of punctuation, like Chauvier’s, suggest that punctuation

developed from a mainly oral90 to a highly rule-bound and systematic
grammatical system, and thence to our modern-day practice which is a
blend of the two. But this appears to be something of an oversimplification.
For one thing, such commentators tend to refer mainly to the way in which
just one punctuation mark, the comma, is described. Most of the other
punctuation marks are interpreted in much the same way by all schools of
thought. For example, even the most grammar-bound of theorists admits
that exclamation marks connote tone of voice. Conversely, dramatic
punctuators have always reserved the full stop to indicate the end of
sentences, even though there may be a case for an equally long pause in
mid-sentence.

Further evidence that the evolution from dramatic to grammatical

punctuation may be an oversimplification is suggested by Husband
& Husband (1905), who analyse the punctuation of a ninth-century
document, Alfred the Great’s preface to a translation commissioned by
him of Pope Gregory’s Cura Pastoralis. The Husbands argue that Alfred’s
punctuation was a systematic attempt to clarify the structure of his
thoughts, and the dot, one of the marks used in the document,
‘is most simply explained with the help of grammatical terms: it indicates a
pause…[they explain various categories of pause]…The pause lessens the risk that

89 Kieras (1985) has reported a number of studies in which the position of sentences within paragraphs
and texts influences their perceived importance (earlier sentences are seen as more important).

90 Dramatic punctuation theories may be linked to the commonsense model of reading as a

subvocalized process, which predisposes some to a role for punctuation which parallels features of
spoken language. Even today we can find those who come to the analysis of punctuation already
strongly committed to the primacy of speech. Some remarks by Quirk et al (1985) are instructive:

‘Although in this book we repeatedly emphasize the primacy of speech over writing, and of prosody
over punctuation, we have to recognize that many types of text take shape first on paper and have
their normal realization in graphic form. Punctuation thus has a greater interest for the study of texts
than for linguistics as a whole, where it can be looked upon as a surrogate and a rather inadequate
substitute for the range of phonologically realized prosodic features at our disposal.’ (p. 1445)

In view of the fact that they devote a whole chapter of their authoritative grammar to text linguistics, it
seems curious to single out ‘texts’ (which elsewhere they use to mean both written and spoken samples
of language in use) as, by implication, unworthy of serious linguistic study.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 119
Chapter 4 • Writer-syntagm and reader-syntagm

either confusion or incomplete comprehension will arise out of the complexity of 152
terms.’ (p. 24, authors’ emphasis)

Confidence in the Husbands’ analysis is enhanced by a more recent study of

another translation of Alfred’s.91 Cyrus (1971: 106) found that spacing was
used not to separate words but syntactic units, remarking on
‘the astonishing degree of correspondence between sequences identified by the
spacings and principle constituent boundaries independently inferable from
syntactic features.’

The Husbands attribute the relative lack of punctuation in classical

manuscripts to the highly inflected nature of Latin and Greek, and suggest
that it is when these languages are translated into the vernacular that
some means has to be found to preserve the structural clarity of the
original. They suggest this as an explanation not only for Alfred the Great’s
translation, but also for the detailed grammatical punctuation evident in
some Elizabethan writings. Although scholars saw the merits of using the
vernacular, they were more used to using Latin as the language of scholarly
exposition. Consequently they tended to impose Latin syntax on the English
language, and, the Husbands argue:
‘this artificiality of construction necessitated the employment of means by which to
make the constructions clear to a reader not exercised in the classics.’ (p. 36)

In view of the Husbands’ view that contemporary punctuation was

grammatical to an extreme, Simpson’s dramatic explanation for
Shakespeare’s punctuation begins to look like an exception to the general
rule. His thesis was in fact vigorously disputed by Fries (1925) who
regards it as over-simple and supported by inadequate evidence.92 Fries
produces counter-examples and demonstrates that where Shakespeare 153
himself alludes to punctuation, which occurs as a metaphor in a number
of instances, it is to its syntactic role that he points. This, of course, is not
proof in itself, since there could have been two systems – one for prose,
to which Shakespeare alludes metaphorically, and one for drama, which
he employed as an additional system of stage direction to his actors; or
Shakespeare, being Shakespeare, might simply have treated punctuation as
he treated other aspects of language – as something to explore and extend.

Although Fries claims to have found evidence of grammatical theories in

all the manuals he examined (dating from between 1589 to 1900), Ong
(1944) cites many examples of the breathing principle during the earlier

91 The Tollemache manuscript of Paulus Orosius’s History of the world (British Museum Additional
Manuscript 47967)

92 Ong (1944) suggests that they are both wrong, arguing that the Elizabethans inherited from late
classical and medieval rhetoricians a system of punctuation based on breathing places for oral reading.
He implies that both the dramatic and the grammatical theories are based on a misunderstanding of this
fact: the performative aspect of oral reading suggests a dramatic function, while the need for breathing
not to disrupt the sense suggests a grammatical theory.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 120
Chapter 4 • Writer-syntagm and reader-syntagm

half of the same period. It may be that you find what you are looking
for; Honan (1960) shows that in many eighteenth-century manuals the
two theories co-exist, sometimes in a manner that obviously confused
contemporaries. Their reconciliation need not be confusing, of course, since
it is an aim of good oratory to enunciate the structure of the message as
clearly as possible. This is implied by Monteith (1704):
‘Pointing is the disposal of speech into certain members for more articulate and
distinct reading and circumstantiating of writs and papers. It rests wholly and
solely on concordance, or government of words, and necessitates a knowledge of
grammar. The wrong placing of points perverts the sense from the true scope of
all speech, which is sound reason.’ (quoted by Husband & Husband 1905: 40; my
emphasis)

The opposite has also been argued: that to follow natural speech rhythms
may be a good guide to sentence construction. Treip (1970), who gives
a detailed account of late sixteenth and seventeenth century punctuation
practice, describes the use of rhythmic constructions in prose. And Sopher
(1977) has argued for a return to dramatic punctuation on exactly these
grounds, even to the extent of challenging taboos such as comma splices
and commas between subject and verb.93 But although he accuses the
Fowlers and Partridge of being grammatical punctuators, a close reading of 154
those authors suggest that they would probably agree with Sopher that ‘it
is this need to satisfy both the eye and the ear of the reader that constitutes
the problem of punctuation’. Whatever view they start out with, in the end
it seems that most punctuation pundits arrive at a Goldilocks theory: not
too much, not too little, but just right – the folklore prototype of thesis,
antithesis and synthesis.

In practice, then, most punctuation manuals eventually enjoin writers to

punctuate to clarify the sense of the sentence. Although it is not always
entirely obvious what they mean by this, it implies that decisions should
be taken not from global principles, but according to local needs to avoid
ambiguities at particular points in the linear sentence.94

93 Support for this view can be found in Quirk et al (1985: 1606) who report a strong tendency among
users of English to insert a comma between a long noun-phrase subject and the verb, reflecting a
prosodic convention in speech.

94 To some degree we can see fashions in punctuation mirrored in typography; in particular the
tendency toward minimalism. For example, Partridge (1953) claims that writers at the turn of
the century would typically use ‘a less varied, less discriminatory, less subtle punctuation’ than
modern writers. He attributes the change in practice since that time to the Fowler brothers (1906)
who recommended authors to write clearly enough not to need much punctuation, although other
contemporary writings, including Husband & Husband (1905) and De Vinne(1901), indicate that such
a view was fairly commonplace at a time when over-formal grammar was falling into disrepute. Possibly
through the enormous influence of The King’s English(Fowler & Fowler 1906) and Modern English
Usage (Fowler 1926), a minimalist approach has grown up reminiscent of the typographic minimalism
discussed in Chapter 1, as writers will testify who have had their commas and hyphens struck out by
pedantic editors. Partridge’s comments could well apply to typographers:

‘The Fowlers have said that everyone should avoid depending on his stops. Well, of course! But
it could with still greater validity be said that to eschew the astonishingly ample resources of

Robert Waller • The typographic contribution to language • PhD Thesis 1987 121
Chapter 4 • Writer-syntagm and reader-syntagm

This conclusion finds agreement in de Beaugrande’s (1984) recent account

of punctuation within a psycho-linguistic framework, one of only two
psychological studies of punctuation in English, so far as I am aware.95 As
we might expect, he suggests his seven linearity principles as motives for
punctuation, thus bypassing the dramatic-grammatical debate (Table 4.3). 155

Principle De Beaugrande’s punctuation strategies

Core-and-adjunct Cores are more likely to be separated from each other by commas; and less
likely to have commas inside them.
Adjuncts are likely to be set off from cores unless closely integrated with
the latter.

Pause Mark with punctuation the points where, reading aloud, you would pause.
Use a dash to make a relevant insertion or addition without affecting the
surrounding format.

Look-back Use punctuation to terminate and characterize the preceding stretch.

Use a comma to make a construction (typically an adjunct) look further
back than if no commas were inserted.

Look-ahead Use punctuation to mark and describe the transition to the next stretch of
text Use a comma to mark where the look-ahead of a construction (typically
an adjunct) ends.

Heaviness The heavier a segment, the more likely it is to be bounded by punctuation.

Disambiguation Punctuate so as to reduce or preclude multiple readings

Listing Use comparable punctuation to set off each element within a list of three
or more

Table 4.3 De Beaugrande’s application of his linearity principles to punctuation strategies (adapted to tabular
form from de Beaugrande, 1984: 192–213). It should be noted that in their original context they are surrounded by
numerous examples and comments

De Beaugrande approaches punctuation in a functional way, to avoid the

apparently arbitrary rules that are typically taught:
‘the illusion of uniformity in punctuation arises mainly from coercion by
publishers, not from agreement in manuscripts.’ (p. 192)

As an alternative:
‘English instructors […] can uncover and present the punctuating motives
observed by skilled writers, and leave the students to decide what options are best.

punctuation, to fail to profit by this storehouse of instruments that clarify and simplify, that variegate
and enliven, that refine and subtilize, closely resembles the action of a pig-headed fellow badly
needing spectacles – and refusing to wear them.’ (p 183)

95 Baldwin & Coady (1978) have studied the effect of punctuation on the comprehension of syntax, but
their study, although resulting in some interesting findings that are described elsewhere in this thesis,
does not place punctuation in an overall linguistic framework comparable to that of de Beaugrande
(1984). A paper by Thorndike (1948) is entitled ‘The psychology of punctuation’ and appears in a
psychology journal, but it is actually an attempt to associate the style of famous authors with a count of
their different punctuation marks.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 122
Chapter 4 • Writer-syntagm and reader-syntagm

An “error” is then a failure to respect motives, not a departure from unexplained

personal biases.’ (p. 193)

For de Beaugrande, then, effective punctuation is deployed with regard to

what the reader is likely to need at the particular point in the linear text.
Again, local needs take precedence over global principle.

Rhetorical structure: scoring vs programming

Nash’s (1980) Designs in prose is a practical manual of composition, a
detailed exposition of the craft of constructing prose. Along with words,
phrases, rhetorical devices and argument structures, he deals with some
aspects of typographic layout. Although he deals with this rather hastily,
he does at least give it a prominent position at the beginning of the book,
rather in the way that layout is the first thing encountered by readers when
they inspect a document.

Nash uses ‘layout’ in a restricted sense to refer to graphological features

within a column of type, rather than the design of whole pages. He first
discusses two extreme forms, ‘lines for copy’ (that is, advertising copy),
and ‘sections for documents’ en route to his treatment of paragraphing.
Copywriting, with its short sentences, often one to a paragraph, its
poem-like use of indention and its creative use of punctuation, has a
strongly phonetic quality, which Nash describes as scoring a potential vocal
performance.

Scoring stands in contrast to the programming of texts such as regulations,

catalogues and others of a technical character. They typically consist of
numbered sections or paragraphs, each self-contained with little need for
discursive links with preceding or succeeding sections. Whereas scored
texts are essentially linear, with any pre-planning heavily disguised under
the surface expression, programmed texts bare their structure for readers
to see and use. In Twyman’s (1982) terms, they are also typified by the use
of extrinsic (ie, spatial) typographic features in addition to the prosodically
more equivalent intrinsic features (ie, variations of typestyle). Programmed
texts offer writers the opportunity to treat their task as an elaborate listing
operation, relieved of the burden of making the argument flow and cohere,
and the fact that their structure is declared and labelled offers readers a
measure of syntagm-control.

The burden of making an argument flow is a considerable one. Glynn et 157

al (1982) attempted to measure the cognitive demands of conforming to
stylistic and grammatical norms. Subjects who were permitted to submit
a persuasive document in note form, or without regard to rules of spelling
and punctuation, produced more points in support of their arguments
than those who had to produce polished prose. The tasks set by Glynn et

Robert Waller • The typographic contribution to language • PhD Thesis 1987 123
Chapter 4 • Writer-syntagm and reader-syntagm

al can be seen as ranging from programming to scoring. Flower & Hayes

(1980: 41) include ‘throw a constraint away’ among the strategies they find
used by writers faced with an excessively difficult task. According to their
observations, writers sometimes handle problems by ‘simply choosing to
ignore their audience or the convention that demands coherence between
paragraphs’. Programming the text with numbered paragraphs enables
them to maintain at least the impression of coherence.

Scoring and programming bear some resemblance to two alternative

hypotheses which Wason (1986) detects in the way writers approach (or
are advised to approach) their task. According to his prescriptive hypothesis,
which I suggest corresponds to Nash’s programming, writing is a ‘process
of transcribing previously formulated ideas’ (p. 288). Prescriptive writing is
audience-centred, and ‘the hypothesis entails that writers adopt a particular
register, such as jargon, simply because they believe it is the required mode
of discourse. Hence it can be used, or omitted, at will.’

Wason’s voice hypothesis, which as the name implies can be likened to

Nash’s scoring, is in contrast writer-centred. It ‘implies that many difficulties
in writing are associated with the lack of an appropriate attitude (trust) in
the writer’. Wason sees the development of trust in one’s own writing, of
faith in the power of writing to reveal and extend one’s ideas, as the key to
sincere and, therefore, clear writing. He suggests that jargon is not just a
style that can be abandoned at will; rather it is a sign that writers ‘are not
writing authentically but are alienated from the topic’.96

Wason clearly values the voice hypothesis above the prescriptive one, and 158
this presents something of a challenge to those who would see typography
as a means of revealing the ‘programme’ of a text in an audience-related
manner. He is, though, talking mostly of creative academic writing, and
he does not directly address other kinds. Certainly his remarks on the
contribution of sincerity, trust and authenticity echo and articulate my own
experience, but it is not clear whether he would agree that it is essential to
subject one’s writing to a process of review – perhaps of transformation –
before anyone else can be expected to read it.

Nash is something of a virtuoso of rhetorical style, writing all his own

examples with an impressive mastery of technique, and pastiche. So it is
not surprising to find that he has introduced the scoring-programming
dichotomy only in order to resolve it when he reaches his real topic:

96 The literary critic Northrop Frye (1957: 331) arrives at a similar diagnosis of jargon in a brief
discussion of officialese:

‘a naive intensification of Mill’s desire to speak with the voice, not of personality, but of Reason itself
… what it actually utters, of course, is the voice of the lonely crowd, the anxiety of the outward-
directed conformist. Such jargon may be called, borrowing a term from medicine, benign jargon: it is
unmistakably a disease of language, but not – yet – a cancerous disease like a demagogue’s oratory.’

Robert Waller • The typographic contribution to language • PhD Thesis 1987 124
Chapter 4 • Writer-syntagm and reader-syntagm

paragraphs for discourse. He suggests a third term, expounding, to introduce

the complex set of rhetorical techniques described in detail in the rest of
the book. But by starting with the two extremes, he compels us to recognize
that writers have recourse both to occasional ‘vocal echoes’ and the open
declaration of their intentions or programme. I shall return to consider
some of the patterns identified by Nash, and ways in which they might be
realized in graphic form, in Chapter 6.

Resolving the dichotomy

The various dichotomies reviewed in this chapter (and some others
discussed elsewhere in this study) can be summarized in a table:

Context Writer-control Reader-control 159

Graphic cohesion (Bernhardt non-visually-informative visually informative
1985)

Markedness (Vachek 1973) unmarked (eg, roman) marked (eg italic type)

Perceptual awareness focal subsidiary, aesthetic,

(Polanyi,1969)

Paralanguage (Lyons 1977) modulation punctuation (segmentation)

Presentational form (Langer discursive logical

1942)

Notationality (Goodman digital analog

1969)

Layout (Goodman 1969) contingent, accidental, constitutive, essential

Configuration (Twyman 1979) linear/linear-interrupted matrix/branching/non-linear

Reading strategy (Thomas & responsive self-organized

Harri-Augstein 1980)

Cognitive style (Pask & Scott serialist holist

1972)

Eye movements foveal parafoveal/peripheral

Reading comprehension bottom-up top-down

Reading teaching oral silent

Punctuation dramatic grammatical

Rhetorical design (Nash scored programmed

1980)

Table 4.4 Writer-control and reader-control of the syntagm, reflected in related areas of study. I have only cited
sources where ideas are closely associated with particular scholars.

A consistent thread running through many of these dichotomies is that

they embody theoretical constructs, which, sometimes overstated by their
initial proponents, have provoked equally overstated reactions. They
are practical instances of thesis and antithesis: the tendency to pursue

Robert Waller • The typographic contribution to language • PhD Thesis 1987 125
Chapter 4 • Writer-syntagm and reader-syntagm

theoretical explanations to pure or extreme forms, the inevitable exceptions

to which provoke equally extreme reactions. They also reflect more general
philosophies of their time: we can associate bottom-up models of reading
with behaviourism, and grammatical punctuation theory with the age of
reason. But the exigencies of everyday communication and the good sense
of ordinary language users resist the idealist inclinations of scholars, and, in
almost every case we looked at, there was evidence of synthesis: the third
element of the dialectic triad of thesis-antithesis-synthesis. We might apply
these words of Eric Partridge (1953: 7): 160

‘In punctuation, grammar represents parliament, or whatever the elected body

happens to be called: logic represents King or president: but the greatest power of
all is vested in the people…’97

The synthesis arrived at in most of these areas is of the order of: most
adults read silently and do not need to move their lips or listen to an inner
voice. They are therefore not bound to follow the order of the text in the
way it is presented to them. Moreover, they have their own ideas, purposes
and questions to which they would like answers and their awareness of
context plays an important part in comprehension. They therefore wish
to see an overview of the structure of the text, in order better to both
comprehend it and search within it. However, verbal language is the
medium through which we conduct argument, debate and discourse, and
in which many of us imagine we do our thinking. So a large degree of
compliance with the author’s linearized structure is necessary if we are
to make progress. Moreover, even though our expectations may strongly
drive our interpretation, all interesting writing combines familiarity with
informativity: if our reading consists only of a search for known (or even
hypothesized) information, we may miss the new information which may
confound our expectations, but, without which, reading would be pointless.

Although in many respects distinct from one another, the items in the
middle column of Table 4.4 share some assumptions about text in common.
They assume a model of text as a linear stream of words and sentences
whose syntagmatic order of presentation is controlled by its writer, and
which is apprehended through its phonological equivalence to speech.
‘Content’ or concept relations are thus linearized for later reconstruction
by the reader’s cognitive processes. Here typography does little more than
‘score’ the reader’s performance.

The right-hand column represents a greater measure of control by the 161

reader over the order in which components of the text are inspected. In this
model, concept relations can be to some degree mapped or diagrammed
directly on the page by typographic and spatial features. The debate is

97 As if to demonstrate that he is no linguistic leveller, Partridge goes on to say ‘…or, rather, in the
more intelligent people – in good sense rather than mere commonsense’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 126
Chapter 4 • Writer-syntagm and reader-syntagm

thus about the location of control – whether in all circumstances the writer
delineates the syntagm by providing a linear path (or a defined choice
of paths), or whether the reader is free to choose from a relatively open
display of text components.

In the next chapter I shall explore the relationship of writer, text and reader
in terms of a communication model.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 127
Chapter 5 • Communication models

5
162

Communication models

From an exploration of the relationship between typography and language,

we have reached a wider perspective in which it is seen in terms of the
relations between writers and readers. The obvious way to diagram this
relationship – so obvious as to make it a truism – is to link writer and
reader by a line passing through text (or medium). All theories of language
or communication at some point have to state their view of the writer-
text-reader relationship, even if it is to exclude much of it from their field,
and many illustrate their explanations with variations of the same simple
diagram. Bloomfield, for example, who excludes contextual or pragmatic
matters from the scope of linguistics, includes a simple model at one point
(Figure 5.1). (It leaves little room for doubt about where he stands on the
primacy of speech.)

speaker’s situation speech hearer’s response

Figure 5.1. From Bloomfield (1935: 139). (Redrawn).

Many more recent versions of this model can be traced to Lasswell’s classic
formulation, ‘Who says what, in what channel, to whom, with what effect?’
(Smith, Lasswell & Casey 1946: 121), and Shannon & Weaver’s (1949)
mathematical model of communication. Their information theory diagram
(Figure 5.2) was originally designed to describe the transmission of
electronic information (signals) along wires, but was subsequently adopted
in many other fields.98 Different versions embellish the diagram in order to
model in more detail how their authors believe that messages are encoded,
transmitted and so on. Johnson & Klare (1961) reviewed a range of general 163
communication models based on information theory.

98 The analogy between electronic and human communication was anticipated by the art historian
Roger Fry (1939): ‘If we take an analogy from the wireless – the artist is the transmitter, the work of
art the medium and the spectator the receiver… for the message to come through, the receiver must be
more or less in tune with the transmitter’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 128
 Chapter 5 • Communication models

noise

Source transmitter message signal channel signal receiver message destination

code

Figure 5.2 A simple Shannon-Weaver type model of communication (this one is from Eco 1976: 33). (Redrawn).

As a model it was attractive because it suggested the application of precise

mathematics to the rather intractable problems of human communication.
Hindsight, however, shows that graphic communication, in particular,
proved rather more resistant to statistical research techniques than was
anticipated. Being unsegmented, graphic images present too great a
challenge to what was essentially a statistical theory about the probability
of ‘bits’ of information surviving transmission without interference (Green &
Courtis 1966).

The information theory metaphor undeniably contains the basic ingredients

of a communication system (the source of a communication, its vehicle
and its user), but it can be easily misunderstood. Although as sensitive a
commentator as Gombrich (1963: 60) could remark that ‘what this theory
has taught us unmathematical laymen to see with greater clarity is the
process of interpretation that is bound up with the reception of any signal’
(my emphasis), a cursory look at the famous diagram too often leads to the
assumption of what might be termed a ‘transport’ or ‘container’ metaphor
for communication: that communication is a unidirectional process,
that messages ‘contain’ meaning, and that the outcome of an idealized
communication act is the assimilation by a receiver of a meaning identical
to that transmitted.99 Communication effectiveness so defined (whether 164
explicitly or not) is easy to measure, and the transport metaphor represents
an implicit assumption of many experiments in educational psychology,100
including those on typographic cuing, reviewed in Chapter 1. Meyer (1985:
29), for example, explains that, according to her cognitive model of text
comprehension,

99 Ong (1982: 166) refers to a ‘pipeline’ model in which ‘the naive reader presumes the prior presence
of an extra-mental referent which the word presumably captures and passes on through a kind of
pipeline to the psyche’.

Richards (1926: 175) also comments critically on those ‘who define communication as the actual
transference of experiences in the strictest possible sense of transference – the sense in which a penny
can be transferred from one pocket to another’.

Lakoff & Johnson (1980b) discuss a number of metaphors for cognition and communication and the
need for scholars to be aware of the metaphors that underlie their models and theories.

100 Brian Lewis talked of a similar ‘medical’ metaphor prevalent among educational theorists. A
knowledge deficiency is diagnosed and a course of treatment prescribed until knowledge levels, as
measured by the psychologist’s instrumentation, are normal. In the context of mass media research,
Tunstall (1970: 4) talks similarly of the ‘hypodermic’ model of communication.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 129
Chapter 5 • Communication models

‘readers are assumed to construct memory representations of text propositions

which are similar in terms of both hierarchical relationships and content to the
content structure of the writer.’

This may be valid for some of the documents that such studies aim to
improve (for example, training manuals), but the transport model in its
usual form attributes an unrealistic passivity to readers, and fails to give
adequate recognition to reader-initiated aspects of reading.

The problem is easy to see in relation to graphic displays such as maps.

Although a route map places limits on what may be found out from it, each
traveller receives a different message depending on where he or she is
starting from and wants to go. There is only a ‘transmitter’ of that message
in a very remote sense, and the reader can hardly be described in such
passive terms as ‘receiver’, since he or she is supplying the structural and
logical framework for the particular message received – in other words, the
argument.

There is perhaps no need to overstress this point, since ‘bottom-up’,

data-driven models of communication have generally gone out of fashion.
Even so, many who research communication processes do not explicitly
declare their model of writer-text-reader relations, and the transport
metaphor seems to be the ‘default option’. This is partly owing to the
words and phrases that spring to mind when we want to talk about
communication. Reddy (1979) has identified what he calls the ‘conduit
metaphor’ for language. He lists a large number of everyday expressions
about language to illustrate his thesis that the metaphor is deeply 165
entrenched in our culture – for example: ‘His words carry little meaning’,
‘Try to pack more thought into fewer words’, ‘It’s hard to get that idea
across to him’.101 Because such expressions are so common, it seems we are
predisposed to think of language and media as containing meanings.

Closely related to the conduit or container metaphor is what we might term

the ‘reader as data terminal’ model. The assumption that readers are input
devices for streams of transmitted data is enshrined in certain editorial
practices. For example, the use of ‘op. cit.’ in footnoting assumes that the

101 Ong (1958) claims that the view of books as containers for knowledge has identifiable roots in the
development of humanist thinking in the sixteenth century and can be seen in changes in contemporary
publishing practices. He describes the medieval revival of the Greek technique of topoi (places) in which
knowledge is thought of as stored under easy-to-remember headings. According to Ong, this essentially
mnemonic technique was (misguidedly) transformed into a system of ‘place-logic’. As evidence for the
new assumption that books are places or containers, Ong cites the development of the title. He detects
a progression from the direct dialogue of the manuscript tradition, where ‘books open with a direct
address to the reader without the formality of a title at all: “Here, dear reader, you have a book…” ’,
to descriptive titles (eg, ‘A book called…’) and eventually the simple label-like titles we have today
– ‘books, and their various parts, were becoming objects which should have simple labels and tags’
(Ong 1958: 313). He does not seem to consider the more prosaic explanation that the proliferation of
books following the introduction of printing might have demanded more precise and succinct titles for
cataloguing and reference.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 130
Chapter 5 • Communication models

reader can remember the work referred to even when it was first mentioned
many pages previously.

Coding and decoding

The classic information theory model (Figure 5.2) assumes that the
transmitter and receiver share a common set of conventions or codes.
The identification of these codes has been the task of the relatively new
discipline of semiology.102 Semiologists find codes in a wide range of
phenomena – even those which by their nature seem indivisible and
analogue, such as film, architecture, fashion and so on.

In graphic design, perhaps the most thorough exponent of the semiological 166
approach is Bertin (1967/1984), who is unequivocal in the introduction to
his Semiology of graphics that:
‘in the visual arts, for example, the semiological approach to graphics provides
a rigorous analysis of the visual means used by the artist. It defines the basic
properties and laws governing the arts and suggests objective criteria for art
criticism.’ (p. xi)

Not all media are equally explicit, and Bertin restricts his analysis to
rule-bound diagrammatic and cartographic images on the grounds that
they are monosemic, as distinct from polysemic or pansemic. Monosemic,
polysemic and pansemic images are those which are, respectively, capable
of only one interpretation, capable of several interpretations, and capable
of an infinite number of interpretations. Pansemic images are comparatively
rare (abstract painting might be an example). In practice most but not
all pictures are polysemic, although as we have noted they are usually
anchored to one interpretation by a caption. Diagrams are (in Bertin’s
opinion, at least) said to be monosemic, so long as they make use of
explicitly coded graphic conventions.103

Bertin is absolutely clear – to the extent of printing it in bold type – about

his assumption of a ‘container’ metaphor:
‘Whether we are studying the means, properties and limits of the graphic
system, or planning a design, it is first necessary to strictly separate the
content (the INFORMATION to be transmitted) from the container (the
PROPERTIES of the graphic system).’ (p. 5, Bertin’s emphasis)

102 Generally speaking, ‘semiology’, the term used by de Saussure to describe his proposed ‘science
of signs’, is used by those working in the European tradition. ‘Semiotics’ is used in connection with the
American tradition founded independently by CS Peirce (1839–1914), although the word itself has a
long history.

103 Against this view, however, we might consider comments made by Anderson (1981: 116) on the
rhetoric of diagrams in academic books: ‘one can get away with a little in prose explanation, a lot in
a table, and an infinity in a diagram’ – quite a nice definition of monosemy, polysemy and pansemy
respectively.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 131
Chapter 5 • Communication models

Although welcomed by statisticians, attracted by the certainties offered

by ‘a grammar for graphics’,104 for some other Anglo-Saxon minds Bertin’s
work confirms their worst fears about semiology: replete with technical 167
terms and classification schemes,105 the book ‘contains’ information but fails
to communicate it. This is not to say that is not frequently insightful and
thought-provoking, but that the needs and questions of the reader seem to
be as absent from Bertin the author as from Bertin the theorist.

Bertin’s consideration of the role of reader is mostly limited to the definition

of what he terms ‘retinal variables’ – aspects of the human perceptual
system that might be thought analogous to the technical limitations of
an electronic receiving device. This seems to reinforce those critics (for
example, Sperber & Wilson 1986; Buchanan 1985) who accuse semiologists
of an obsession with codes and fixed meanings, to the detriment of
inference, rhetoric and other reader-centred factors.

Sperber & Wilson (1986) have recently criticized the coding model that
they attribute to many linguists,106 and to semiologists in particular. They
point out that:
‘[although] it is true that a language is a code which pairs phonetic and semantic
representations of sentences…there is a gap between the semantic representation
of sentences and the thoughts actually communicated by utterances. This gap is
filled not by more coding, but by inference.’ (p. 9)107

Whereas the notion of decoding implies the mechanical recovery of a

message put into coded form by a sender, Sperber & Wilson’s alternative,
an inference model, appears to assign a more creative role to the audience,
who must draw inferences from evidence supplied.

Sperber & Wilson’s characterization of semiology – they are particularly 168

scathing about Barthes – is something of an oversimplification. A number
of prominent semiologists have written perceptively about the role of the
reader. In S/Z, regarded as something of a departure from his earlier, more
rigid structuralism (Eagleton 1983), Barthes (1970/1975) distinguishes
between readerly and writerly literature: the former is self-contained,
explicit and allows the reader to look through the language to a portrayed
world; the latter focuses readers on language itself and gives them a role
in creating meaning. Barthes does, however, remain strongly committed

104 This phrase is from Howard Wainer’s introduction to the 1984 English translation, which he was
instrumental in organizing.

105 If grammar is Parliament and logic is King, this style of comprehensive classification of sign systems
is the Common Market, obsessed by the harmonization of standards.

106 Their criticism, although independent, is based on many of the same arguments as that of Harris
(1981), whose ‘language myth’ was discussed in Chapter 3.

107 Sperber & Wilson’s inference theory builds on the work of the philosopher HP Grice (1975), whose
theory of conversational implicature has made a considerable impact on the study of pragmatics and
discourse processes. It is described in more detail in Chapter 8.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 132
Chapter 5 • Communication models

to the concept of codes, although he stretches the ordinary meaning of the

term somewhat. For example, although one might think of connotation (as
distinct from denotation) as personalized and uncoded, Barthes (1977)
explains it by reference to de Saussure’s distinction between syntagmatic
and associative relationships. For Barthes, words and images connote
certain meanings because they are associated along ‘semic axes’ whose
reconstitution ‘will clearly only be possible once a massive inventory of the
systems of connotation has been carried out’ (Barthes 1977: 49). Barthes
does not attempt such an inventory, and it is unclear whether it is a serious
suggestion, but the very breadth (it would need to encompass images,
words, gestures, objects, indeed anything) and depth (the individual’s
psyche does not escape) of such an agenda is what makes semiology
attractive to some, threatening to others, and impracticable to yet others.

Where Barthes is inspired, polemical and somewhat extravagant in his

claims, Eco’s argument is meticulous, although sometimes complex
(Eco 1976, 1981). He is careful to define the code as a ‘mere regulative
hypothesis’ with which to analyse actual instances of signs, and he argues
that the code has too many transient aspects to be defined. His reasoning is
bound up with the role of the reader, or ‘addressee’, in his semiotic theory.

Eco, whose notion of closed and open texts is similar in some respects to
Barthes’ readerly and writerly, has made the role of the reader a central part 169
of his semiotic theory. He amends the traditional communication model
(Figure 5.2) by separating the sender’s code from the addressee’s, and
introducing ‘context’ and an ‘effort to reconstruct the sender’s codes’ into
the process of interpretation (Figure 5.3). In effect, Eco is giving the reader
more work to do, beyond the simple decoding of a signal. Indeed, he talks
elsewhere (Eco 1976: 156) of the ‘labor of inference’.

sender coded text channel text as addressee interpreted

expression text as content

codes
subcodes
context, codes
circumstances subcodes

‘philological’ effort to reconstruct sender’s codes

Figure 5.3 Eco’s revised communication model, redrawn (Eco 1981: 5)108

As Figure 5.3 illustrates, for Eco the code is something potentially, but
not necessarily, shared by participants in the communication process.

108 It is worth noting that, although the diagram also appears in his earlier A theory of semiotics (Eco
1976), there is a difference between the two versions that could be important. In the earlier version, the
arrows between ‘text as expression’, ‘context, circumstances’ and ‘codes, subcodes’ are reversed: they
flow from top-left to bottom-right .

Robert Waller • The typographic contribution to language • PhD Thesis 1987 133
Chapter 5 • Communication models

Instead of a single code, he considers it more reasonable to speak of a

‘complex network of subcodes’ that may be strong or weak, and that are
subject to constant change as each juxtaposition of elements creates a
new, if temporary, connotation. To deal with the problem of shifting and
potentially incompatible sender- and addressee-codes, Eco proposes the 170
twin concepts of undercoding and overcoding, which, although he does
not explicitly say so, appear to relate mainly to addressee and sender
respectively. Undercoding is mainly a problem for the addressee who
must assign provisional meanings to text when faced with uncertainty:
sometimes these meanings are confirmed or denied in the light of
subsequent text. Overcoding describes the use by the sender of ready-made
phrases, intertextual references, clichés and patterns that narrow down the
possibility of misinterpretation by the addressee. The significance of these
concepts in the context of this study is that Eco regards paralinguistic and
other contextual cues (of which typography might be one) as instances of
overcoding that enable the addressee to select appropriate subcodes for the
interpretation of the message.

Conversational models
An overwhelming impression from the semiology of Barthes, Eco and others
is of an assumption that (with the exception of poetic or aesthetic texts)
messages are created by one person in order to communicate something
to another. However much notice they take of the reader, it appears to be
assumed that messages are created, conveyed and attended to as complete
entities – whether they be complete myths, complete poems, complete
advertisements or complete diagrams. In practice, of course, most spoken
conversation is very far from this model. It may be significant that those
who, like Sperber & Wilson (1986), have emphasized the role of inference
have generally taken conversational discourse as their data.

In the light of the dichotomies discussed in Chapter 4, and their various

compromises, typography was seen as a tool for making the content of
documents accessible to readers with different problems and purposes. I
have, in effect, been assuming a model of typographically organized text
that provides the basis for a conversation between writer and reader in
which control switches between the participants.

text as addressee interpreted

expression text as content

context, codes
circumstances subcodes
The accompanying text does not make it clear whether this is intentional or erroneous. Although it is
tempting to see the reversal of the arrows as a shift towards an attribution of greater initiative to the
reader, a second major difference cancels out that impression. Whereas, in the version shown in Figure
5.3, ‘sender’ is connected to ‘addressee’ via the series of left-to-right arrows along the top of the diagram,
the original version uses lines only, with no direction of flow indicated.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 134
Chapter 5 • Communication models

The idea of textual dialogue is obviously more acceptable when we consider 171
the use of reference books designed for easy access, but it has been
argued that even continuous prose is more conversational than it appears.
According to this view, writers ‘converse’ with an imagined reader whose
questions and objections must be anticipated. Apart from those who study
conversation itself (for example, Coulthard & Montgomery 1981, Gumperz
1982, Coulthard 1985), conversational theories of written language have
also been suggested in the literature of linguistics (Gray 1977; Winter 1977;
Widdowson 1979; Hoey 1983), semiotics (Eco 1981), cognitive psychology
(Wright 1978; Nystrand 1986) and among literary critics of the ‘reader-
response’ school (Tompkins 1980; Suleiman & Crosman 1980). Crudely
summarized, one version of the conversational view is that writers address
themselves to an imagined reader (sometimes referred to in the literary
critical context as a ‘mock’, ‘model’, ‘virtual’ or ‘implied’ reader) whose
characteristics and attitudes the real reader is expected, and – if the writer
is skilful enough – able to assume. It is argued that, just like a participant
in a conversation, the imagined reader has questions and expectations to
be dealt with by the writer. Thus we are asked to empathize with radically
different personalities in order to make sense of books by, say, Austen and
Hemingway. As real readers, naturally, we will ask different questions, but
we must suspend judgement and hope that the imagined reader eventually
asks them.

Conversational theories, discussed further in Chapter 8, are the subject of

vigorous debate and represent only one of a range of theories of author-
text-reader relations.109 However, there seems no reason why readers should
not switch between different reader-roles: between a close identification
with the reader anticipated or imagined by the author, a goal-directed
strategy of their own, and the distanced view required for critical scrutiny.
And this would certainly seem to be easier in non-fiction documents where 172
typographically signalled supports are provided: headings, indexes, lists
of contents and other devices that can be grouped under the term access
structures (Waller 1979a).

If we accept a conversational view of the relationship between readers and

texts, then it would seem that our criteria for well-formed texts should
go beyond the language surface and relate to the context in which texts
are used. Indeed, this is reinforced by the review in Chapter 1 of several
linguists who have addressed typographic issues. It was noticeable that all
of them had seen the scope of linguistics as reaching beyond the sentence
to whole texts or discourses. Another tendency was toward functionalism:
typographic features were seen in relation to the use that is made of texts in
addition to the meanings that they might embody.

109 Further aspects of the debate between conversational (Nystrand 1986) and autonomous models of
text (Olson 1977) is discussed further in Chapter 8.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 135
Chapter 5 • Communication models

Writer-text and reader-text relations

Those who employ a conversational model of communication are, in
effect, suggesting that the arrows in the usual information theory diagram
be reversed. A recent example is Nystrand’s (1982) model of ‘textual
space’ (Figure 5.4), an application of a more general model of what he
calls ‘semantic space’ – the sphere of meaning shared by participants in a
medium of communication.

INTERPRETATIVE INTERPRETATIVE SIGN INTERPRETATIVE INTERPRETATIVE

ROLE ACT/expression ACT/comprehension ROLE

WRITER expression TEXT comprehension READER

Figure 5.4 Top: Nystrand’s model of semantic space. Bottom: its application to ‘textual space’.
From Nystrand (1982: 82), redrawn.

The reversing of the arrows between reader and text represents the 173
distinction between bottom-up and top-down models. However, in spite of
an explicit recognition that students of the writing process need no longer
feel intimidated by the advocates of the primacy of speech, Nystrand insists
on applying a general semantic model to all media (including speech,
music and painting). This is theoretically neat, and enables Nystrand to
draw some interesting conclusions about the problems of learning to read,
but it fails to acknowledge a major contextual difference between written
and spoken texts, and a further distinction between printed texts and both
spoken and hand-written ones.

Nystrand argues that when writers and reader share the same textual space,
the material text becomes transparent:
‘fluent writers are no more aware of pen and paper than fluent readers are aware
of the words they see.’ (Nystrand 1982: 83)

This is true and intuitively acceptable up to a point, but sounds remarkably

like the primacy-of-speech argument in its neglect of the contribution
made by the materiality of the text itself. Stretching the suffix ‘-graphic’
somewhat, we can say that spoken texts and hand-written texts are both
autographic. That is, the text heard or read by the ‘receiver’ is identical
in substance (if not always in significance) to that emitted by its author:
the sound-waves (or the ink and paper) are the same. Most printed texts,
though, are prepared by the writer over a period of time and mediated by
complicated bureaucratic and industrial processes of both production and
distribution.110 As a result, writers of texts and their readers are separated

110 Vachek (1967) recognizes the distinction between written and printed language, but takes the view
that printed language is neutral and unmarked. This is superficially true if we restrict our view to simple

Robert Waller • The typographic contribution to language • PhD Thesis 1987 136
Chapter 5 • Communication models

by a change in time, place and material that can be enormous: an Indian

student studying Hamlet or an English student reading the Bhagavadgita
are separated by hundreds of both miles and years from their writers; those
texts are available in technical forms undreamed of at the time they were
written.

The contribution of these intervening processes is highlighted in the model 174

shown in Figure 5.5 (Waller 1979b). The straight-line relationship has
been replaced by an indirect ‘dog-leg’ one, reflecting the fact that printed
documents separate the addresser and addressee to a degree that the
telephone, for example, does not. This simple model stresses the necessity
of considering the writer-text-reader relationship as two separate systems.
Sless (1981, 1986) has also emphasized the separate consideration of what
he terms author-message relations and audience-message relations.

Originator Medium
Change in time
and place
Medium User

Figure 5.5 The indirect nature of printed communication. From Waller (1979b: 216).

Although still connected by lines, in this version of the model their

directionality has been removed. In its original context – an introduction
to a special issue of the journal Instructional Science on diagrams, the
four possible placements for arrows were used to summarize four ways of
thinking about their role (Figure 5.6).

Originator Medium Statement

Originator Medium Tools for enquiry & thought

Medium User Aids to learning

Medium User Aids to problem-solving

Figure 5.6 Four relationships between writers, readers and texts, suggesting four roles of diagrams.
From Waller (1979b: 217).

I have already stressed the active role of the reader, but this diagram
suggests that the normal direction of flow between writer and text can also 175
be reversed. Writers do not simply make statements that are comprehended
by readers. The very act of self-expression distances the writer from its
content and allows objective inspection and evaluation (Macdonald-Ross,

lines of a common typeface, but if layout, binding and display typography are taken into account, texts
originated by different publishers, designers and printing processes can be as distinct from one another
as two samples of handwriting.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 137
Chapter 5 • Communication models

1979, has documented some aspects of the role of graphic notations in

the generation of scientific ideas). The absence of a physical reader does
not mean that the communication is no longer conversational. Just as
readers enquire of texts as well as receive information, so the composition
process objectifies the writer’s thoughts and reflects them back for analysis
and amendment. In the light of feedback from listeners’ expressions and
questions, speakers hesitate, back-track, repeat or retract; writers do the
same things but privately, in response to their own reaction to what they
have written (Hayes & Flower 1980).

The separate relationships between writer & text and reader & text might
be summarized through a pair of metaphors in which writers and readers
are represented by traders and their customers; the medium is represented
by the counter (or its equivalent) over which the relationship is conducted.

Figure 5.7 The transport model

Figure 5.7 represents the transport model of communication as the bar in a

wild west saloon, along which the barman slides the whiskey to the cowboy.
There is only one brand on offer which exactly matches the cowboy’s need.
As long as the bar (the text) is perfectly constructed and the cowboy’s
drinking (reading) skills are adequate, all the whiskey (knowledge) will
reach his stomach (memory). By the use of a stomach pump (or in the
educational context, a written examination) it should be possible to monitor 176
the success of the transaction. The metaphor could be elaborated along
the lines of information theory to suggest that the barman provide extra
(redundant) whiskey in case any imperfections (noise) in the bar surface
cause the drink to spill.

Figure 5.8 The access model

Robert Waller • The typographic contribution to language • PhD Thesis 1987 138
Chapter 5 • Communication models

The alternative open-access model is represented in Figure 5.8 as a

supermarket display – the focus for choice and self-selection. In this
context, as with text, the trader and customer must make a special effort if
they want to communicate directly. Instead, their relationship is normally
mediated by the display shelves and is separated by the time that elapses
between the stocking of shelves and the purchase of goods. The trader must
predict the requirements of the customers by ordering the right products,
and must present them logically and attractively so that they can easily be
found. The customers make their selection according a pre-planned set of
goals (a shopping list) or on impulse; they can opt for pre-packaged food
or assemble ingredients to process themselves. The trader can of course
attempt to influence the customers’ choice by careful juxtaposition of items
(strawberries next to cream) but must present a coherent overall argument
(separate shelves for different classes of goods).

A genre model of typographic communication

The argument so far has been leading up to a simple model that is intended
to shed light on the functional constraints that govern the typographer’s
role in textual communication. Models of communication are typically
constructed with a particular context or type of text in mind, whether 177
literary, instructional or technical. Any communication model addressed
explicitly to typographers, though, must acknowledge that most of them
encounter a wide range of text types from day to day. This model therefore
aims to reveal the structures that underlie distinct genres of text. I shall
describe the model in three stages.

Writer Writer’s text

Readers’ text Readers

Figure 5.9 Stage 1: the genre model is an adaptation of the ‘dog-leg’ model in Figure 5.5

As Figure 5.9 shows, the model emphasizes the separate relationships

between writer & text and readers & text. The central vertical line
represents the publishing process that results in important physical
differences (traditionally, at least) between the writer’s text and the readers’
text. In conventional book-publishing systems the writer deals with a
document that becomes progressively more formal as production processes
develop: rough notes become typescript, typescript becomes galleys,
galleys become pages. Traditional printing methods require the writer to
make most significant decisions in relation to a manuscript and a type

Robert Waller • The typographic contribution to language • PhD Thesis 1987 139
Chapter 5 • Communication models

specification. The reader, on the other hand, sees a finished product which
is expected to betray little of the complexity and difficulties of the writing
process.

Figure 5.9 also indicates the conversational nature of the model – writer
and readers are seen as ‘conversing’ with surrogates in the form of the
text, beyond which they may imagine a writer or readers whose identity is
implied by the content, structure and style of the text. The plural form of
‘reader’ is used in the model as a reminder that the writer must frequently
provide for the needs of a range of different imagined readers. The singular
form of ‘writer’ is used because, even where a number of writers or a team
of writers and designers contribute to a publication, readers are normally 178
presented with the semblance of a unified single source – a single imagined
writer.111

Writer Writer’s text Imagined readers

Imagined writer Readers’ text Readers

Topic Artefact Access

structure structure structure

Figure 5.10 Stage 2: each of the three stages in the communication relationship – writing, production and reading
– determines an underlying functional constraint on the typographer. In this model, they are termed topic, artefact
and access structure.

It is suggested that the three main stages of this model (writing, publishing
and reading) account for three kinds of structure which may be, and
typically are, overlaid in the same document. I shall call these topic
structure, artefact structure, and access structure.

Topic structure includes those typographic effects whose purpose is

to display information about the author’s argument – the topic of the
discourse. If one purpose of language is to describe the world, then it is
easy to see the linearity of language as an imperfection. In its descriptive
role, at least, an ‘ideal’ language would directly map ‘reality’ (whatever
shape that is). To compensate for its linearity, language is rich in spatial
metaphor, and this can sometimes be reflected in a more literal use of space
to add diagrammatic qualities to an otherwise verbal argument. It should

111 The roles of the narrator and implied author in fiction have been discussed most notably by the
critic Wayne Booth (1961). In the non-fiction context, the nearest thing would be the Open University’s
concept of the ‘tutorial in print’ (Rowntree 1982), the full implications of which have never been
properly explored in detail.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 140
Chapter 5 • Communication models

be stressed that the terms topic structure and topicalization are used here in
a typographic sense; topic structures are also signalled verbally, and it is in
that context that they are usually discussed.

Artefact structure represents those features of a typographic display 180

that result from the physical nature of the document or display and its
production technology. Page numbers, for example, usually describe
units of the artefact (that is, pages) rather than units of the topic. Since
everything typographical or spatial uses an aspect of the artefact as a means
of signalling (for example, a heading generally begins on a new line), the
term must be restricted in some way if it is to be of use. In the context
of this argument, it is intended to cover only those features which are
motivated by the artefact alone, or whose ‘ideal’ form is constrained by the
artefact. An example of this distinction is the signalling of a new chapter
in a conventional book. The use of a new page marks the writer’s topic
boundary and an access point for the reader, but is not constrained by the
shape or size of the page. The amount of blank space remaining at the end
of the previous chapter, however, is solely a function of the page size, and is
therefore artefactual. Any attempt by the reader to interpret it as topically
significant is erroneous.

Access structure represents those features that serve to make the document
usable by readers and the status of its components clear. These may include
aids to interacting with the text as artefact: formats convenient for special
purposes; navigational aids for the self-organized reader (for example, a list
of contents); and isolated signposts that offer guidance at strategic points
in the document (for example, ‘continued on p. 60’). They also include aids
to interacting with the text as topic: typography is often used to delineate
the status of different ‘voices’ in the conversation – components such as
quotations, glosses, pedagogical devices (statements of objectives, for
example).

Writer Writer’s text Imagined readers 181

Imagined writer Readers’ text Readers

Topic Artefact Access

structure structure structure

Conventional structures (genres)

Figure 5.11 Stage 3: the complete model proposes that the three underlying structures (topic, artefact and access)
are synthesized in practice by conventional structures that are associated with different genres of text.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 141
Chapter 5 • Communication models

The complete model (Figure 5.11) suggests that a typical printed text
exhibits a combination of topic, artefact and access structures. The
management of their combination, I suggest, may be described through a
fourth category, conventional structure, which amounts to a definition of
typographic genres. The distinction between the three basic structures,
it should be stressed, is largely a theoretical one, since in practice they
frequently coincide. It has already been noted that new chapters, for
example, serve both as access points for the reader’s ‘conversation’ with the
book and as topic boundaries. Topic, artefact and access structures are thus
heuristic concepts whose main purpose is to form the basis for describing
genres of typographically-organized documents. They are ideal types
which are never or rarely found in isolation, but which are recognizable in
combination.

In most genres, all three kinds of structure appear to be inextricably

bound together in conventional ways to the extent that it becomes hard
to imagine any other way of presenting the same topic, or addressing the
same needs. In particular, it will be argued, conventional ways of expressing
and accessing topic structures develop within the artefactual constraints
of contemporary technologies. When those technologies change, it may be 181
necessary to separate out the three categories of functional imperatives in
order to reassemble them to suit the constraints of the new technology.

An example of this is the problem encountered in making tabular

information available through computer displays (McLaren, 1983, and
Norrish, 1984, describe examples of such a task). The topic structure of
a typical railway timetable, for example, includes two main themes: the
network, usually defined in terms of routes (destinations linked by the
network) and trains (times, stopping points, facilities). The organization
of conventional timetables reflects specific conversational structures, and
several parallel versions are sometimes provided.

General purpose timetables for use by people planning journeys are

organized by route, with places ranged down the y-axis and trains along the
x-axis. The times of trains are placed at the stopping points, while facilities
are typically indicated by footnotes and symbols. In-station timetables
are organized by time: the times of trains are listed in bold type, with the
stopping points and times listed next to each. These are for people at the
station who need to know when the next train is, or where the train just
coming in is going: their particular conversational need requires a different
way of accessing the information. Other in-station timetables are organized
by place: travellers look up their destination and find the times of trains
from the station they are in.

To display this information on computer terminals requires a different

approach. Some new artefactual constraints are introduced: the large

Robert Waller • The typographic contribution to language • PhD Thesis 1987 142
Chapter 5 • Communication models

broadsheets used for printed timetables cannot be displayed legibly

on a screen. But other constraints are lost because the computer does
not restrict us to the permanence of print. In the computer version the
distinction between topic structure and access structure is particularly clear
– in computer jargon it corresponds to the distinction between the data
structure and the interface. The timetable information exists in virtual form
and, given an appropriate access structure can be actualized in numerous 182
different ways, depending on the user’s question: how many trains to
London on Sundays? how much does it cost to get to Bristol? how far can I
go for £10? when is the next train to Glasgow? what is the quickest journey
to Brighton? and so on.

Although it aims to convey the same topic structure as the printed

timetable, the different access and artefact structures of computer timetable
places it in a different genre: perhaps in the more general genre of computer
database, which being relatively new, probably contains a number of
sub-genres whose identity will emerge as they become more commonplace.

Interaction of the structures

In order to further illustrate the distinction between the three structures I
will refer in the next few chapters to a series of pages from an illustrated
non-fiction book which makes heavy use of typographic structuring, both
through typographic signalling of text components such as headings, and
through the layout of pages.112 By attempting to ‘parse’ these pages in
terms of the three structures, it is hoped that some conventions and typical
characteristics of the genre to which this book belongs might be revealed.

The first of these pages is illustrated (reduced in scale) in Figure 5.12,

which demonstrates some simple aspects of the interaction of the three
structures. This page, describing ‘Mainsheet systems’, contains a fairly
clear topic structure, as defined by the main headings (indicated in Table 183
5.1 in bold type according to their hierarchy on the page) and captions to
illustrations.

112 The handbook of sailing by Bob Bond, published by Pelham Books, London, in 1980. The book
was produced for the publisher by a firm of ‘book packagers’, Dorling Kindersley. The development of
book packaging over the last few decades has been an important factor in the evolution of graphically
structured texts. The term refers to specialist firms who develop titles from initial concept through to
printed object, but on behalf of other publishers. The concept is usually sold to several publishers, in
different countries, before it is actually developed in detail. Consequently, development costs which
would be too high for a single publisher can be shared. Printing takes place at a single factory, with
all editions sharing the same colour printing. The text is then overprinted in a separate run for each
publisher’s edition.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 143
Chapter 5 • Communication models

Figure 5.12 A page from The Handbook of Sailing, by Bob Bond (Pelham Books 1980)

Mainsheet systems
Centre mainsheet Aft mainsheet
Centre mainsheet Aft mainsheet
side mounting Basic system
traveller System with pulleys
System with transom traveller

Knotting the sheets

Double overhand Figure of eight

Table 5.1 The topic hierarchy represented by the layout of Figure 5.12. Topics represented by illustrations are
indicated by italics.

The topic structure is reinforced in the introductory paragraph, where 184

we read ‘There are two principal types of mainsheet system: aft or
centre…’ The introduction also includes comment on the topic and
subtopics – it defines mainsheet, indicates that one type is mainly used

Robert Waller • The typographic contribution to language • PhD Thesis 1987 144
Chapter 5 • Communication models

on racing dinghies, and tells us that the mainsheet system is normally left
permanently on the boat. The item on knots, although relevant to the main
topic of the page, is both graphically and topically independent. It repeats
information given in a special section on knots elsewhere in the book, and
has probably been used here as a filler.

However, if we expected the typography of the page to follow this simple

hierarchical topic structure, we might be puzzled by the apparent failure to
align the small contextualizing illustrations (top right) with their matching
sub-headings. If there were no considerations other than to display the
topic structure on the page as one might a diagram, then the page could
have looked much like Figure 5.13.

Figure 5.13 The previous figure shown as an exploded text-diagram, unconstrained by the page boundary.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 145
Chapter 5 • Communication models

Obviously this is not possible, and we must accept the constraints of the 185
artefact structure of the publication, the most obvious features of which are
the page size and the layout grid (Figure 5.14). Comparison of a number
of pages from The handbook of sailing indicates that the designers have
allowed themselves a choice of one-, two-, three- or four-column grids, and
that these can be mixed fairly freely within each page.

Figure 5.14 The grid probably used for The Handbook of Sailing.

Figure 5.15 If the two main topics on the page had been equal in length, a symmetrical two-column layout might
have been possible, displaying a clearly diagrammed topic structure.

Ignoring for a moment the boxed item on knots, we can say that if the two
branches of the main topic had contained the same number of elements, it
would have been relatively easy to fit them onto this page in a two column
format (Figure 5.15).

The two sub-topics are not equal, however, and, as we have seen, an extra
item on knots has been recruited to balance the page. By including the
knots item and using a three column grid, we could produce the layouts
shown in Figure 5.16a and 5.16b, each of which associates the small

Robert Waller • The typographic contribution to language • PhD Thesis 1987 146
Chapter 5 • Communication models

sketches with their correct topics, and each of which assigns equal status to
the two main topics.

186

Figure 5.16a Figure 5.16b

Figure 5.17 Two of the pages that accompany the one shown in Figure 5.12.

The key to the layout is found in its access structure. If we are able to view
the ‘Mainsheet systems’ page in the context of other pages with which it
appears (Figure 5.17), we can see that the top third of most of these pages
follows a consistent pattern (Figure 5.18).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 147
Chapter 5 • Communication models

187

Figure 5.18 The consistent access structure of these pages

Each of these pages113 contains a topic heading, an introductory paragraph

and an identical drawing of a dinghy on which the location of the topic
is highlighted in blue. The apparently ‘ideal’ shape for the topic has thus
been traded against the need for consistency in the access structure. Within
what in Figure 5.18 was termed the ‘information block’, there is a further
consistency in the typographic treatment of text components. Roughly
speaking, they are specified as follows:

Page-level heading 22pt roman bold

Intro 10pt roman

Heading 14pt roman bold

Main text 8pt roman

Caption heading 8pt sans serif bold

Caption text 8pt sans serif

Running head & folio 10pt roman

Genre and textuality

The model proposed here implies the replacement of a coding model of
communication with one that recognizes a greater role for inference and
interpretation. Another way of expressing this is suggested by Eco (1976)
who, citing the Soviet linguist Lotman (1969), distinguishes between
grammar-oriented and text-oriented cultures. While grammar-oriented 188
cultures are governed by a system of rules, text-oriented cultures are
governed by a repertoire of texts, imposing models of behaviour – in effect,
genres.

113 It should be stated that I have selected these pages for the purpose of demonstrating a principle. In
reality, the consistent access structure shown in Figure 5.18 only extends across a limited set of pages.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 148
Chapter 5 • Communication models

The analysis of The handbook of sailing demonstrated some aspects of

this distinction between grammaticality and textuality – I tried to show
how ‘grammatical’ expectations about the display of topic structures were
modified by the circumstances in which they were displayed. The following
brief analysis of two examples of another genre, paperback book covers,
demonstrates a wider range of factors that have been associated with
textuality.

Figure 5.19 Figure 5.20

Figures 5.19 and 5.20 represents the jackets of two books in my home. 189
I correctly interpret the first as a book called Tom Jones by an author called
Henry Fielding. I may have deduced that from a typographic rule that, in
the absence of a specific statement that ‘This is a book called such-and-such,
by an author called so-and-so’, the title is printed larger than and below
the author’s name. Or I may have deduced it from my prior knowledge that
there is an author called Henry Fielding. Those who have not heard of this
author would be entirely blameless if they misunderstood this title.114

If I applied the typographic rule to Figure 5.20, though, I should be puzzled

to find a book called Mary Stewart by The Gabriel Hounds. The fact that
I don’t is not due to the fact that I have heard of an author called Mary
Stewart (I had not), but because I know that dogs can’t write. I therefore
rejected the typographic rule and added to my general knowledge the fact
that there is a famous author of that name.

We can apply the same principle to the interpretation of a simple sentence

such as ‘The cat sat ___ the mat’, Our identification of the missing word

114 I have experienced problems of a similar kind when trying to determine which of several parts of a
foreign-language letterhead contains the address, particularly if there are no words I recognize, like Rue
or Straße, and given the different conventional order of street, city and district in some countries.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 149
Chapter 5 • Communication models

as ‘on’ is not just a matter of parsing the grammar of the sentence,

determining that a preposition is needed and selecting one at random. We
also use common sense to reject unlikely options. ‘The cat sat on the mat’
resembles a reading primer cliché. If the subject had been human, we might
have guessed quite differently: ‘mending’ or ‘weaving’, for example.

These two approaches to language interpretation have also been described

as linguistic and ethnomethodological (Widdowson 1979). While a linguist
might look for logical rules linking linguistic signals and patterns to
meanings, to be shared by the creator and interpreter of a document, the
ethnomethodologist is more interested in the practical reasoning that
occurs on an actual occasion of language use (in this case, in a library or
bookshop) – reasoning that usually goes beyond knowledge of language to 190
include all our knowledge of social interaction.

De Beaugrande & Dressler break textuality down into seven standards that
characterize actual texts. Table 5.2 lists them in table form to demonstrate
why, using practical reasoning, we still understand the cover of the Mary
Stewart book in spite of an apparent breakdown of the first (grammatical)
standard, cohesion.

Standard De Beaugrande & Dressler’s My application to Figure 5.20

explanation
Cohesion Grammatical dependencies on Title is usually larger than
the [text] surface. author, but…
Coherence Conceptual dependencies in ‘The Gabriel Hounds’ is less
the textual world. likely to be an author than
‘Mary Stewart’. Her name is
emphasized because…
Intentionality & The attitudes of the We know the publisher wants
Acceptability participants towards the text. to stress the information that
will sell most books. We want
to choose a book.
Informativity The incorporation into the new Here is a new book by a
and unexpected into the old well-known author.
and expected.
Situationality The setting. The cover is on a book which
is for sale.
Intertextuality The mutual relevance of The illustration, title and blurb
separate texts. identify it as of the romantic
fiction genre.

Table 5.2. Seven standards of textuality, based on de Beaugrande & Dressler (1981: 37).

These seven factors bring together things found in the text and things
outside it, and thus form a useful account of what readers expect to find in
actual texts. They represent categories of expectations which we bring to

Robert Waller • The typographic contribution to language • PhD Thesis 1987 150
Chapter 5 • Communication models

real texts in real situations, and consequently they are characterized by the
flexibility of heuristics rather than the rigidity of grammar. In the case of 191
the romantic novel, one apparent surface meaning is overridden by several
other kinds of expectation and prior knowledge about books, literary
genres, authors and publishers.

In the next three chapters I shall look in more detail at the nature and
interaction of topic structure, artefact structure and access structure, before
moving on to consider, in Chapter 9, the concept of genre.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 151
Chapter 6 • Topic structure

6
192

Topic structure

In the last chapter I proposed a model of typographic communication

in which the visible structure of particular texts reflects three kinds of
underlying structural imperative, each stemming from a different part of
the writer-text-reader relationship, and whose integration is managed by
a fourth structure which I called conventional, or genre structure. In this
chapter I shall explore the first of the three, topic structure, in more detail.

Whatever their ultimate motives – to inform, educate or persuade – authors

of non-fiction texts are also trying to order their ideas, and this chapter
explores the extent to which such ordering may be signalled through
typography and layout. That is not to say that page layouts can often
represent ‘knowledge structures’ in a direct sense – we have already noted
the constraints imposed by the linearity of language – but the example
briefly considered in Chapter 5 (Figure 5.12) indicate that there may be
considerable potential for topic structures to be reinforced by their graphic
arrangement.

In the distinction between topic and access structures lurks a danger

that should be acknowledged at the outset: it may be taken to imply
that information can be encoded in a pure form, unadulterated by
considerations of audience. In the present context, though, the distinction
between topic and access structures is largely a theoretical construct,
convenient for the organization of the argument. The progression from
topic structure to access structure via artefact structure may be seen as
a vehicle to demonstrate the replacement of the transport metaphor for
written language with a context-bound and audience-related model in
which typography plays a key role. The notion, implied by topic structure,
that topics can be easily encoded is complicated by the recognition of the 193
role of the artefact, and further broken down when we consider the role of
the reader.

Nevertheless, perhaps because of the prevalence of Reddy’s ‘conduit’

metaphor in everyday language, many writers do appear to see texts as
autonomous expositions of ideas. Sticht (1985), for example, attributes
the failure of technical manuals to communicate effectively to an excessive
degree of topic-orientation. The technical writers surveyed by Kern & Sticht
(1974) saw their task as simply one of assembling and recording all that is

Robert Waller • The typographic contribution to language • PhD Thesis 1987 152
Chapter 6 • Topic structure

known about their topic. A major problem of manuals is identified by Sticht

as a tendency by their writers to see school textbooks as a model of good
writing. He argues that, whereas textbooks traditionally seek to display the
logical connections between ideas, manuals are used by a wide range of
people in association with job-related tasks.

Olson (1980) identifies the textbook form as an example of the archival

form of written language, which he distinguishes from the communicational,
which, presumably, includes notes to the milkman, personal correspondence
and the like. The implication is that the textbook form contains explicit
and context-free meanings and is thus relatively autonomous. Olson does
not claim that archival texts do not communicate, but that they ‘preserve
their meanings across speakers and situations’. His theory of autonomous
text is discussed more fully in subsequent chapters, but for now we may
simply note that, in practice, most uncontroversially archival texts such as
dictionaries and manuals are able to communicate because their context,
source, status, range of possible readers and organizational principles are
made explicit through typographically signalled means.

Olson’s terms are similar in intent to the distinction drawn in Chapter 3

between ‘fact structure’ and ‘argument structure’. There may be cases where
the facts have some inherent order of their own, and others where the
writer may have reason to prefer one arrangement to another. However,
the term ‘topic structure’ enables us to circumnavigate these distinctions 194
altogether for the time being, since it simply refers to whatever the writer
wishes to talk about. Following Grimes (1975: 337), the topic of a text may
be defined as ‘that part of the surface form that represents the speaker’s
thematic choice’115 – whether that form represents a fact structure, an
argument structure, or one of the other distinctions that arise in the
literature of linguistics, psychology and education – topic and comment,
language and metalanguage, for example. To talk of ‘topic structure’, then,
enables us to avoid some of the trickier philosophical questions concerning
the structure of knowledge and to confine our interest to those aspects of
structure that can be made visible through typography, while still, following
Grimes, concentrating on the writer’s thematic choice. Texts seen as topic
structures represent the writer’s communication goals organized in the form
of arguments, which in turn are expressed at the text surface through verbal
language, pictures and typographic layout.

115 The word ‘topic’ is linked to the speaker’s choice of theme and the surface form of language,
through its origins in the Greek word τοπος, a place (see Chapter 4, footnote 101). Topics are
metaphorical places (ie headings) in which arguments can be found.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 153
Chapter 6 • Topic structure

Visual and spatial metaphor

The distinction between fact and argument structures might in any case
be minimized by the abundance of visual and spatial metaphors in the
literature of linguistics and semantics. For example, the literary critic
Northrop Frye (1957: 335) talked of the link between logic and rhetoric –
or, we might say, a topic and the way it is addressed to an audience – as
‘ “doodle” or associative diagram, the expression of the conceptual by the spatial…
If a writer says “But on the other hand there is a further consideration to be
brought forward in support of the opposing argument,” he may be writing normal
(if wordy) English, but he is also doing precisely what an armchair strategist
does when he scrawls plans of battle on a tablecloth. Very often a “structure” or
“system” of thought can be reduced to a diagrammatic pattern – in fact both words 195
are to some extent synonyms of diagram.’116

Rather than advocating a literal expression of the conceptual by the spatial,

Frye is actually addressing the function of metaphor in non-literary prose.
He is concerned that in the effort to ‘purify verbal communication from the
emotional content of rhetoric’, prose becomes, paradoxically, less clear not
more.117

Analogy and metaphor allow us to discuss argument structures as if

they were fact structures. Instances of spatial metaphor in the technical
vocabulary of linguists suggest that it might be possible to identify graphic
techniques that break away from the hierarchical norm but that still
correspond more or less directly to ways in which we are accustomed to
organizing words and ideas.

Nash (1980), for example, suggests four kinds of ‘rhetorical design’ which,
he argues, are fundamental to all composition (although usually found in
combination). Nash’s categories – the Step, the Stack, the Chain and the
Balance118 – may all be interpreted as visual metaphors,.

The Step is the easiest one to identify in graphic form. Indeed, Nash
suggests that his example (a set of instructions) is an instance of

116 We sometimes talk metaphorically of writers ‘mapping their domain’ and this suggests a happy
coincidence in the similarity of the words ‘typography’ and ‘topography’. As a student of the former I
was sometimes assumed by others to be studying the latter. The misunderstanding might have been
reinforced by the fact that the geography and typography departments shared the same building.

117 Although Frye does not develop the idea in depth, Lakoff & Johnson (1980a) have built a cognitive
theory around their wide-ranging exposition of the pervasiveness of metaphor in everyday thinking.
Besides the transport, pipeline, or conduit metaphor for communication, mentioned in Chapter 5,
other everyday metaphors identified by Lakoff & Johnson include Time is money, exemplified by ‘you’re
wasting my time’ or ‘how do you spend your time these days?’; Argument is war: ‘your claims are
indefensible’, ‘he shot down all my arguments’.

118 Nash’s book Designs in prose is written in textbook form – that is, with student exercises and a
general reading list but virtually no citations. It is therefore hard to see how his ideas fit into the general
linguistics scene. Although two of his categories of rhetorical design are similar to those of Grimes
(1975), they are probably independent. Nash’s four varieties of rhetorical design form the basis of Quirk
et als’ treatment of discourse strategies in their authoritative Comprehensive grammar of the English
language (1985: 1435).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 154
Chapter 6 • Topic structure

‘programming’ (see Chapter 4), and ‘could well have been laid out as 196
separate and perhaps numbered sentences’ (p. 9). An example of a stepped
rhetorical design reflected in typographic layout can be seen in the section
headed ‘Sail onto boom’ in Figure 6.1.

Figure 6.1 The three numbered procedures in the section entitled ‘Sail onto boom’ are in a stepped relationship.

Figure 6.2 The stepped relationship between the elements of this page is indicated by the schematic drawing (top
right). However, it is not particularly well reflected in the layout (see comments in text below)

Robert Waller • The typographic contribution to language • PhD Thesis 1987 155
Chapter 6 • Topic structure

However, we might puzzle over the less clear relationship between

rhetorical and graphic design in Figure 6.2. Although the schematic
drawing at the top right-hand corner of the page has ‘Shackling head
to halyard’ as step 4, preceding step 5, ‘Hoisting the jib’, the layout
seems to treat step 5 as a separate topic from steps 1–3, and step 4 as a
comment on step 5. In both Figure 6.2 and Figure 6.1 the clearly stepped
design is diluted by the failure to repeat the enumeration of the steps in
the sub-headings; furthermore, the wording of the sub-headings is not
consistent with the steps as announced in the schematic summary drawings.

A stack design is characterized by the announcement of a topic, followed by

a series of amplifying or explanatory comments. Stacks are, in effect, lists 197
of attributes or comments, and may be graphically treated as such. Figure
6.3 contains a small stack of ideas relating to the topic ‘rudder and tiller’:
‘parts of the rudder’, ‘fitting the rudder’, and ‘tiller extension’ (there seems
no reason why this should not have a more prominent heading). Grimes
(1975: 245-6) discusses a similar rhetorical pattern, the star, whose name
also suggests a graphic form. The star is a pattern of persuasive argument in
which a number of independent points contribute to a central conclusion.

Figure 6.3 With the exception of the stepped sequence under ‘Fitting the rudder’, most of this page consists of a
stack of information about its topic.

Of his four rhetorical designs, Nash’s chains are the least amenable to
graphic treatment since, as the metaphor suggests, they are essentially
linearized, As he puts it,
‘the writer’s procedures are less predictive than exploratory; he works through the
expository maze, seeing no more than a sentence ahead, placing his trust in the
clues afforded by syntactical or lexical connections.’

Robert Waller • The typographic contribution to language • PhD Thesis 1987 156
Chapter 6 • Topic structure

So whereas each sentence in a stacked paragraph takes the same initial 198
topic sentence as its point of departure, chained sentences simply relate to
their immediate predecessors. In view of this apparent lack of pre-planning
it is hard to see why Nash includes chains as ‘designs’ at all. Judging by his
examples, chain structures are more characteristic of literary prose than
expository or technical information.

Balanced rhetorical patterns present contrasting viewpoints – proposition

and counter-proposition. The Balance would appear to be easily reflected
in layout – the point-by-point comparison of two (or more) contrasting
options can be easily made in a table, for example. Indeed, the bilateral
symmetry implied by the term ‘balance’ points to an advantage of graphic
formats over prose– complex comparisons can be made in a considerably
more orderly way.119 In prose, Nash suggests that balance is often more
apparent than real – the writer may simply want to give the appearance
of considering both viewpoints, while moving us gently towards his or her
preferred view.

In ordinary discourse, Nash suggests,

‘there is a programme of assertions, examples, qualifications, but these are not
presented as a series of distinctly labelled positions. Instead, they are related
to each other in a progressively unfolding pattern, the turns and connections of
which are demonstrated in various ways: sometimes by means of syntactic devices,
sometimes through the kinship of elements in vocabulary, sometimes by the
management of punctuation and typography.’ (p. 6–7)

However, although Nash thus includes typography among the structuring

techniques available to writers, bracketed with punctuation, he does not
provide any detailed guidance. 199

In practice, the main provision for the typographic signalling of topic

structures in most publishers’ style guides is for hierarchical structures of
headings and sub-headings. A typical hierarchy might provide for chapter
headings, and three levels of sub-headings, perhaps termed A, B and C
headings. In effect, a single graphic technique must serve for a variety
of rhetorical purposes. Arguments may be represented as hierarchical
structures, even when the ‘ideal’ text-diagram might be rather different.120

119 Support for this view may be found in the outcome of an experiment recently reported by myself
and my colleague Peter Whalley (Waller & Whalley 1987). We tested two prose versions and a tabular
arrangement of a balanced argument comparing aspects of psychoanalysis and behaviour therapy. One
prose version presented each side of the argument separately, while the other interwove both viewpoints
in an integrated fashion. A previous study (by Schnotz 1982) had supported the hypothesis that the
separated prose version would result in a sound comprehension of each therapy, but would inhibit the
coherent integration of the two points of view (and vice versa). We confirmed our own hypothesis that a
tabular arrangement would disadvantage no one, since it would allow readers to choose an appropriate
strategy for their purpose.

120 The idea that texts are sets of hierarchically related propositions underlies a great deal of research
into text comprehension. See the review by Meyer (1985).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 157
Chapter 6 • Topic structure

Since topic structures do not always correspond to the structures implied

by the hierarchical typographic arrangement enforced by the norms of
book publishing (or to any simple, easily diagrammed structure, for that
matter), the exact relationships between major points in an argument must
usually be specified in some other way – as Nash suggests, through syntax
or parallelisms and other ‘kinships’ in vocabulary. Interestingly, there is a
noticeable similarity between Nash’s fourfold classification of rhetorical
designs and a distinction between four kinds of verbal conjunction made
in Halliday & Hasan’s (1976) important account of linguistic cohesion in
English texts (Table 6.1). So although Nash’s categories simply seemed to
be a useful starting point for this discussion because of their metaphorical
names, confidence in them is enhanced by close parallels with other
classifications suggested independently by scholars in related contexts.
In another context still, the psychology of text comprehension, Meyer’s
categories of rhetorical structure are converging in a similar way. She has
recently conflated her original eighteen categories (Meyer 1975) into five
categories that on examination bear a close relationship to Halliday &
Hasan’s: collection, description, causation, problem/solution, and comparison
(Meyer 1985).

Nash’s rhetorical designs Halliday & Hasan’s Examples of conjunctive 200

conjunctive relations adjuncts

Step temporal first, then, next, finally

Stack additive and, furthermore, for

instance

Chain causal so, because, consequently

Balance adversative but, however, on the other

hand, rather

Table 6.1 A comparison of Nash’s rhetorical designs and Halliday & Hasan’s conjunctive relations121

Although conjunction is just one of Halliday & Hasan’s five kinds of

‘cohesive tie’, it is of special relevance to the present study. Whereas the
other four – reference, substitution, ellipsis and lexical cohesion122 – are

121 Halliday & Hasan’s taxonomy of conjunctive relations is considerably more elaborate than is
represented here. Each major type of relation is divided into ‘external’ and ‘internal’, and further
subdivided as appropriate. The examples of conjunctions shown here are from external categories – a
reason for this is discussed below.

122 The distinction between the other four categories is a subtle one. Reference might be exemplified by
‘Three blind mice. See how they run’, where ‘they’ refers to ‘mice’. Substitution is exemplified by ‘My axe
is too blunt. I must get a sharper one’, where ‘one’ substitutes for ‘axe’. Reference is a semantic relation,
while substitution is a grammatical relation between linguistic items – whereas the first example
could be reversed, so that ‘they’ refers ahead (‘See how they run, the three blind mice’), the same
cannot be said of substitution (‘I must get a sharper one, because my axe is too blunt’ is grammatically
unacceptable). Ellipsis is described as ‘substitution by zero’, as in ‘Joan bought some carnations, and
Catherine some sweet peas’. Lexical cohesion is superficially similar to substitution. Where the latter

Robert Waller • The typographic contribution to language • PhD Thesis 1987 158
Chapter 6 • Topic structure

embedded in the internal structure and wording of sentences, conjunction

is normally achieved through separate, identifiable ‘adjuncts’ – words and
phrases. Halliday & Hasan explain that
‘conjunctive relations are encoded not in the form of grammatical structures but
in the looser, more pliable form of linkages between the components of a text’ (p.
321).

So if cohesive relations can be displayed through typography, itself a means

of linking text components, they are most likely to be of the conjunctive
kind. It should be remembered, of course, that Halliday & Hasan are for the
most part interested in relatively short-range relations, typically between 201
pairs of sentences, rather than the structure of extended arguments. Any
extended prose passage will contain a variety of cohesive ties from many
of their different categories and sub-categories. But the sort of relations
or structures found typographically signalled in the Handbook of sailing
examples are usually less subtle than those in a typical page of prose. They
relate to broad structures found (or imposed) within the page’s topic.

Figure 6.4 The identical frame-size of these four methods of carrying a boat, and the absence of a linear sequence
of their arrangement, is suggestive of ‘or’ conjunctions – classed by Halliday & Hasan (1976) as an additive
conjunctive relation (of the sub-category ‘alternative’).

Additive relations can be seen as inclusive of Nash’s stacks (Figure 6.3).

Figure 6.4 gives a further example. Temporal relations can be seen in
terms of steps (Figure 6.1), although the latter may have causal links also.
However the apparent similarity between the Nash and Halliday & Hasan
schemes becomes rather more blurred when one examines the equivalence
of chain & causal. From Nash’s statement that each sentence in a chain

relies on a set of neutral terms (like ‘one’, or in the case of this sentence, ‘the latter’), lexical cohesion
does not so much substitute as reiterate with a lexically related expression. The following example
includes two instances of lexical replacement, ‘children’ and ‘food’: ‘Patrick and Theresa won’t eat their
macaroni. Why are children so fussy about their food?’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 159
Chapter 6 • Topic structure

takes its predecessor as a point of departure, we can see chain relations as

being both causal and additive. Given our present interest in information
rather than literary texts, ‘causal’ is a rather more useful category than
‘chain’, although it is no easier to show graphically.123 The equivalence of
balance & adversative is also not straightforward, since Halliday & Hasan
class balanced constructions as either adversative or additive, according
to whether they refer to external contrasts (that is, contrasts in the fact
structure) or internal contrasts (in the linearized argument structure).

A problem emerges from this brief comparison of two categorial 202

frameworks. Halliday & Hasan’s four categories only correspond to Nash’s
if we select their external (fact structure) examples. But this is the opposite
of what we might expect when we recall that Nash’s purpose is to classify
not fact structures but argument structures. The answer lies in the highly
metaphorical character of Nash’s categories – although he is describing
argument structures, he uses the vocabulary of fact structures to do so.

If we look more closely at this vocabulary of fact structures in the context

of semantics, once again we find a high degree of visual, or at least
visualizable, metaphor. Table 6.2 lists the lexical ‘sense relations’ discussed
by Lyons (1977). Other textbooks (for example, Leech 1981) use similar
terms.

Contrast
Binary opposites
gradable (eg, hot/cold)
non-gradable (eg, male/female)
converse (eg, husband/wife)
directional (eg, North/South, up/down)
Non-binary sets
Serially ordered
gradable scales (eg, poor…fair…excellent)
non-gradable ranks (eg, private, corporal…field marshal)
Cyclical (eg, …spring, summer, autumn, winter, spring…)

Hierarchy
Class inclusion (eg, animal: cow, sheep, etc)
Part-whole relations (eg, body: arms, legs, etc).

Table 6.2 Sense relations in vocabulary (abstracted in table form from Lyons, 1977, Chapter 9).

Many of these sense relations are suggestive of visual metaphor, and it is

quite easy to find a number of them graphically displayed in the Handbook
of sailing. Figures 6.5 to 6.10 show examples of those compatible with the
segmented character of typography.124

123 The link between ‘causal’ and ‘chain’ is reinforced by Grimes (1975: 246), who discusses chain
patterns in rhetoric using causal examples.

124 Examples of the two gradable categories are not shown, since by their nature they are incompatible
with the segmented (ie, non-gradable) character of typography. They can be found in diagrams, though.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 160
Chapter 6 • Topic structure

203

Figure 6.5 Binary contrast, non-gradable: the use of parallel columns is a typical way of showing an either/or
relationship. The use of a different typeface for the main text vs caption relationship could be seen as an example
of a converse binary contrast.

Figure 6.6 Binary contrast, directional: here the order in which topics are presented reflects the directional or
temporal order of topic – when taking a trip in a boat, you leave before you arrive back. In a different topic, it might
have been more appropriate for arriving to precede leaving, the convention being to show temporal progression
in terms of the norms of the writing system; that is, from left to right, top to bottom, in English. Other conceptual
relationships are assigned directionality by metaphor: senior people thus rank above or before junior ones, and so
on.

For example, the illustration at the bottom right of Figure 6.2 shows a sail in the process of being
hoisted – the binary contrast displayed is ‘up vs down’; the infinite number of intermediate grades are
hinted at by the obvious motion of the sail (indicated by the arrow and the person pulling on the rope).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 161
Chapter 6 • Topic structure

204

Figure 6.7 Non-binary sets, serially-ordered non-gradable: The numbered sequence is an obvious example.

Figure 6.8 Non-binary sets, cyclical: In this case the cycle is indicated by using the same illustration for step 4
as for step 1. An alternative might have been to arrange the steps into a circle, but this arrangement is particularly
suited to the subject – the progress of the boat through the water.

Figure 6.9 Hierarchy, class inclusion: The classic hierarchy, indicated by a hierarchy of headings
of varying prominence.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 162
Chapter 6 • Topic structure

Figure 6.10 Hierarchy, part-whole relations: part-whole relations may be shown by a simple typographic
hierarchy, or, as in this example, it may be possible to combine the pictorial and verbal modes to indicate the
position of the parts within the whole.

Since the scope of all classifications is related to their purpose, it is

understandable that some of Lyon’s sense relations do not have a direct
equivalent in graphic displays, and that some graphic conventions do not
find a place in this list. And it is noticeable that some semantic relations
work better than others within the rectilinear conventions of typographic
layout. In particular, non-gradable sets (equivalent to Nash’s steps and
balances) are easily chunked and therefore tabulated or split into columns.
Gradable sets, on the other hand, can be described in linear prose or by
recourse to a separate diagram but with difficulty through layout alone.

The linearity of language is rarely an obstacle to the connection of concepts

at the sentence level. Halliday & Hasan’s cohesive ties, for example,
usually create links between sentences which are both physically close
and available in short-term memory. But when a link is to be made across
many pages rather than just a few sentences, language alone strains to
compensate for its own linearity. Subtleties of sentence construction or
inflection no longer suffice, and authors usually introduce ‘metalanguage’ – 206
whole sentences or paragraphs in which they step back from their argument
and comment, seemingly objectively, on its progress. At this metalinguistic
level some writers prefer to break out of the linear mode altogether and use
graphic techniques. Diagrams are often used, particularly in textbooks, to
help readers overview the author’s argument (Figure 6.11).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 163
Chapter 6 • Topic structure

Figure 6.11 Part of a structural diagram included in a geography textbook (P. Haggett, Geography: a modern
synthesis, 2nd edition, London: Harper & Row, 1975).

Whether readers actually use or benefit from such diagrams is still an open
question among educational psychologists. Jonassen and Hawk (1984) have
tested similar ‘graphic organizers’ and found advantages for immediate but
not delayed recall. It is possible that training is needed to make use of such
devices. Indeed, lack of familiarity with diagrams is suggested by Holliday
(1976) as a possible explanation of his finding that where the ‘information’
in the diagram was accompanied by the same ‘information’ in prose
form, readers preferred the familiar prose version. However, experiments
which oblige readers to study in controlled conditions cannot measure
how effective these devices are for less formal purposes such as browsing
or revision. And in the absence of a basis for comparing the content,
complexity and style of diagrams, it is difficult to generalize from particular
studies.

Table 6.3 was the outcome of an informal survey of diagrams in Open

University and other textbooks (Waller 1981), part of an attempt to 207
encourage authors of continuing education courses to make more use of
graphic design in their work.125 The courses, which cover subjects of general
adult interest such as consumer choice, health, parenthood and retirement,
were intended to be easy to read and were to some extent modelled
on home reference manuals such as The handbook of sailing. The table
was intended to alert authors to opportunities for displaying their ideas
graphically, and the categories bear some relation to Lyons’.

125 Wright (1985: 93) comments, on the basis of a study of writing and editing, that ‘few amateur
writers appear to introduce illustrations spontaneously, even when describing the rules of a board game
such as draughts’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 164
Chapter 6 • Topic structure

Relationships displayed Examples

Categorial
discrete simple lists
hierarchical chains of command, taxonomies, hierarchical lists
overlapping Venn diagrams, matrices
comparative/contrastive continua, parallelisms, reflections and other symmetrical or
axial graphic structures
Dynamic
temporal calendars, time-lines
serial non-temporal series of events or processes, including cycles
causal algorithms, feedback charts, some operating instructions
cumulative recipes, production process/flow charts where a given
feature acquires new characteristics as a result of inputs and
interactions
Spatial
locational town plans, ‘physical’ maps
territorial organizational charts, ‘political’ maps
networks route maps, circuit diagrams

Table 6.3 Semantic structures displayed by various genres of network diagram

(From Waller 1981).126

Table 6.3 classifies topics for diagrams rather than typographic layouts. 208
But since they are almost completely unconstrained by the conventions
of linear-interrupted written language, diagrams provide instances of
graphically-realizable topic structures in a relatively pure form. And there
is a sense in which we can view typographic layouts in terms of ‘text-as-
diagram’ (Waller 1982, 1985).

As Michael Evans (1980) has shown, such diagrams have a long history.
The medieval preoccupation with order and especially geometry made
diagramming a particularly suitable medium for recording scholastic
analysis. Evans describes the use of branching diagrams (‘stemmata’ is
Evans’ term), geometric diagrams, and visual metaphors such as trees,
wheels, towers and ladders. He includes the diagrammatic use of page
layouts in his account:
‘A different size of initial was used to begin book, chapter and verse in the Bible;
different grades of script were used to distinguish between text, commentary and
gloss’ (p. 34)127

126 The classification scheme in Table 6.4 owes much to a similar, unpublished exercise undertaken
by a colleague, Derek Prior (now of the Community Education Development Centre, Coventry); and it
formed part of a joint evaluation project with Mick Jones of the Open University Continuing Education
Division, and Jane Wolfson (now of Learning Materials Design, Newport Pagnell).

127 Ullman (1932: 117) reports that the typographic indication of the status of text was used as early
as the Carolingian period (ninth century): ‘One of the outstanding characteristics of the Carolingian
writing, especially at Tours, was the careful distinction of different styles for different purposes …

Robert Waller • The typographic contribution to language • PhD Thesis 1987 165
Chapter 6 • Topic structure

Yates (1966) views such diagrams largely as mnemonic devices, seeing

them as concrete manifestations of the ‘artificial memory’ systems of
classical rhetoric. The technique was, typically, to imagine a familiar
building, and associate the various facts to be remembered with rooms of
the building, and things in the rooms. Retrieval was a matter of walking
through the building in the mind, and restoring the connection between
place and fact. As we have seen, our word ‘topic’ stems from the Greek
word τοπος, place. This technique was evidently effective and very
necessary when facts could not so readily be looked up in books or notes.

Ong (1958), who attributes the development of topic diagrams to the 209
introduction of printing, laments what he sees as the replacement of
the medieval oral tradition 128with the ‘reduction to spatial form [that]
fixes everything, even sound’ (p. 109). Ong has published a number of
compelling studies comparing oral and literate cultures (1967, 1982), the
general thrust of which is to remind us of the complexity and validity of the
oral tradition.

Figure 6.12

square capitals were used for book headings, rustic capitals for explicits, uncials for chapter headings,
tables of content, and first lines, half-uncials for second lines prefaces and the like. Thus there was
established what has been called the hierarchy of scripts.’ Further aspects of medieval page design are
discussed in Chapter 7.

128 Ong’s identification of the Middle Ages as an age of oral tradition and of spatial forms with printing
has been contested by others; this and other aspects of technological constraints on expression are
discussed in Chapter 7.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 166
Chapter 6 • Topic structure

Ong (1958) gives a detailed and fascinating account of the career and
widespread influence of Peter Ramus, the sixteenth century French scholar
whose teaching method was based on the subdivision of topics into
sub-topics, typically displayed in branching diagrams that encapsulate
knowledge in a seductively complete form (Figure 6.12).129 Ong is fairly
dismissive of his subject, and somewhat hostile to his method. Since his 210
comments on topic diagrams are not confined to the sixteenth century
context, they are of some relevance to the present discussion – although
at times hard to fathom and apparently containing the seeds of their own
refutation.

For Ong, visualization is at the heart of science and education:

‘We are today more than ever witnesses of attempts to reduce everything supplied
by the other senses – sounds, smells, tastes, pressures – to charts and tables which
can be visually assimilated’ (p. 108)

This would seem to be an advantage, but Ong argues that it is deceptive.

The essence of his objection to diagramming is that there is no spatial or
visual analogue for what he calls ‘enunciation’, the making of judgements,
the ‘coupling of subject and predicate – and this last term conceals an
auditory analogy again; praedicatum is the thing cried out or said’ (p. 110)

Most would agree that diagrams tend to present simplified and often
suspiciously symmetrical arguments, and that they are rather harder to
analyse and criticize than verbal language. But it is hard to see why Ong
needs to generalize from Ramus’ diagrams to all literate culture, as he
appears to do. Diagrams and charts rarely appear on their own but are
mostly surrounded by verbal language. Furthermore, the oral culture that
Ong appears to champion is characterized by most scholars, including Ong
himself, as heavily reliant on mnemonic techniques – not only the place-
mnemonics but rhyme, dichotomy, analogy and myth. These techniques fix
knowledge in as permanent a form as the ‘pseudo-eternity of repose’ Ong
attributes to print. And whatever the advantages of dialogue, it would seem 211
to be at least as difficult to question the guardians of oral tradition – whose
job is to preserve not to improve – as it is to question a printed book.130

129 Ramus’ charts are not altogether different from the hierarchical schemes of 1960s educational
theory (Gagné 1965, Ausubel 1963) and the text structure diagrams of recent cognitive psychologists
(Britton & Black 1984). Indeed, a comment of Ong’s might strike a chord with the more sceptical of
educational technologists: ‘…while many of the significant reactions in intellectual history were taking
place because of new scientific or philosophical insights, they were occurring more inevitably because of
the demands of a practical pedagogy – even when the pedagogical necessity was given a veneer of quasi-
scientific explanation’ (Ong 1958: 306)

130 Indeed, Saenger (1982: 399) comments that ‘psychologically speaking, silent reading emboldened
the reader, because it placed the source of his curiosity completely under his personal control.’

Robert Waller • The typographic contribution to language • PhD Thesis 1987 167
Chapter 6 • Topic structure

Topic diagrams as writing plans

The distinctio stage of the scholastic method, which the diagrams discussed
by Evans embody, preceded the detailed discussion of evidence, authorities
and so on. Today, too, diagrams are frequently used for the initial planning
of prose. Indeed, those offering advice on writing (and thinking) frequently
recommend diagramming as an aid to creativity (Buzan 1974, Field 1982).
And diagrammatic techniques for ‘idea-processing’ have been available
on personal computers for some time, and are integrated into some
word-processing programs. Idea-processors allow writers to plan, overview
and reorganize documents as hierarchically arranged diagrams of headings.

Saenger (1982) describes a medieval precedent of this development,

suggesting that the synthesizing task of twelfth- and thirteenth-century
scholasticism led to important changes in ways writers approached their
task. Where writing had previously been undertaken in a relatively linear
fashion, through dictation or the use of wax tablets of limited capacity,
writers found they could no longer organize their ‘exceedingly complex
thoughts’ within these constraints. The introduction of cursive script131
‘meant that authors could revise and rearrange their texts while composing them.
This facility aided thirteenth-century scholastic writers to prepare texts rich in
cross-references which presupposed that the reader, like the author, had the ability
to flip from folio to folio in order to relate arguments to their logical antecedents 212
and to compare comments on related but disparate passages of scripture’ (Saenger
1982: 386).132

This visual planning of arguments is central to the method of production

used for books like The handbook of sailing.133 Figure 6.13, for example,
shows part of an ‘editorial flow-chart’ (sometimes known as a ‘flat plan’)
used to plan a similar manual, The indoor garden book.

131 Clanchy (1979: 89) regards the introduction of cursive script, with its advantages of speed
and legibility, as an important aspect of the ‘shift from memory to written record’, and as a major
contribution of the twelfth and thirteenth centuries to the growth of literacy.

132 Clanchy (1979: 130) also remarks on the changing nature of scholarship in the thirteenth century,
comparing the library regulations of Dominican monks with those of a community of Benedictines two
centuries earlier. Books were no longer issued once a year for ‘mystical contemplation’ but needed to
be available for rapid consultation and comparison: ‘The difference in approach towards writing of
Lanfranc’s Benedictines and Humbert’s Dominicans is so fundamental that to use the same term ‘literate’
to describe them both is misleading.’ Saenger’s suggestion that the reader is expected to apply the same
flexibility of approach as the writer is echoed in the recent development of ‘interactive’, ‘dynamic’ or
‘hyper-’ text (Weyer 1982; Conklin 1986). These offer readers of electronically-delivered texts the same
facilities that the author had on his or her idea-processor: hierarchical nesting of sub-sections, search
facilities, note-making, glossaries and so on.

133 The short account of the production method of this book is based on interviews with staff members
of Dorling Kindersley Ltd, the firm also responsible for ‘packaging’ the Handbook of sailing. The
interviews form the basis of an audio-cassette for an Open University course on communication (Waller
1987). Rogers (1986) has also recently articulated some of the methods by which book packaging
operates.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 168
Chapter 6 • Topic structure

Figure 6.13 ‘Editorial flow chart used to plan an illustrated book (John Brookes, The indoor garden book, London:
Dorling Kindersley, 1986)

213

Figure 6.14 Part of a large ‘design flow chart’ used to plan the display of topic on pages

By planning the sequence and length of topics in advance, space is allocated

more systematically than might otherwise be the case. Furthermore, as
Figure 6.14 (a ‘design flow-chart’) shows, the design of individual pages
is also planned in advance, before any of the words are written or the
illustrations commissioned. The design of such pages acts as a planning
chart for the organization of concepts, the writing of descriptions and the
composition of illustrations.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 169
Chapter 6 • Topic structure

214

Figure 6.15 This layout has been constructed after the photo-session has helped to determine how many steps are
needed to explain the procedure – in this case, a flower arrangement.

In many cases the next step in the preparation of a spread is the photo-
session, where aspects of the topic are photographed. In the case of a
practical task, the number of illustrations required to demonstrate it
properly has a strong influence on the design of the page – the photo-
session is one way of revealing the structure of a topic (Figure 6.15).
Obviously, the final pages will usually undergo numerous modifications
and so look considerably different from the first plan, but these books are
nevertheless powerful demonstrations of the principle of text-as-diagram –
typography, far from being a decorative embellishment, is as fundamental
as any other aspect of the language of these pages.134

Information Mapping 215

Robert Horn (1985) has attempted to systematize the use of text-as-
diagram through his Information Mapping™ system of ‘structured writing’
(Figure 6.17). 135 His original vision was of a system in which
‘the physical arrangements of the maps provide a special136 analogue to the
connections and relationships of the information.’ (p. 182)

134 Even those aspects of graphic design that might be thought to be simple embellishment turn
out to have an important function apart from their marketing value. Although the jacket design is
conventionally left to last, book packagers often start with it. The effort to agree about the cover design
enables a production team to articulate their thoughts about the genre, philosophy and general aim
of the project. Agnew (1986) has recently described a similar production process in a paper entitled
‘writing backwards’.

135 Horn preserves the term as the trade mark of his technical writing firm by insisting that it is
accompanied by the ™ symbol.

136 This is presumably a misprint or dictation error for ‘spatial’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 170
Chapter 6 • Topic structure

Figure 6.17 A double-page spread from an article by Horn describing his Information Mapping system (Horn
1985). Most pages, such as the left-hand page here, contain series of ‘information blocks’ – paragraphs with
marginal headings. Diagrammatic techniques are sometimes used (right-hand page). (Original 290mm x 227mm.)

Horn claims support from a number of evaluations of the system, which

requires writers to identify the status of each text component through
graphic segmentation, shaping and labelling. Because it is largely directed
at technical publications which are not in the public domain, its impact
since it was developed in the early 1970s is hard to assess. Owing to
conservatism, the high cost of the manual and the extra time taken to
prepare texts using Horn’s guidelines, it is probably rather limited. But 216
although it seems to be a very good idea, there are some significant and
instructive flaws in its implementation.

For one thing, the explicit labelling of every turn of the argument leads
to an unnatural and unsubtle fragmentation of the text. Since every
component is labelled with equal typographic emphasis, it is scarcely easier
to pick out the major turning points than if nothing had been labelled. This
is a classic problem of categorization: to classify each item under a different
label is as unhelpful as to classify them all under one heading.

Secondly, Horn labels each kind of block in the same way: definitions,
examples, summaries and facts are displayed in the same typographic voice.
His problem, again, is over-systematization: since he claims ‘a working
typology of over 200 [types of block] (independent of subject matter)
for different kinds of document’, it would not be possible to distinguish
between them all.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 171
Chapter 6 • Topic structure

Thirdly, although it represents the injection of graphic techniques

into verbal language, the system pays insufficient attention to graphic
subtleties. As Figure 6.17 illustrates, the use of space, emphasis, rules,
and diagramming is often clumsy. In this example, we might identify the
excessive capitalization of headings that makes them hard to scan,137
the equal treatment given to new headings and continuation headings
(this gives inadequate emphasis to the change of topic), and the poor
diagramming on the right hand page.

Though contact with typographers is leading to improvements, the

disappointing graphic execution of published examples of Horn’s structured
writing highlights the uncompromising nature of visual imagery.138 It 217
also reinforces a conclusion reached in Chapter 3: that the exact graphic
configuration and rendering of graphic elements is as important – as
constitutive to their meaning – as their mere presence or absence. The lack
of attention to the graphic implementation of Information Mapping may,
of course, be a deliberate compromise. Given that the method is designed
to be applied by technical writers with few graphic skills and a variety of
reprographic techniques, it is probably wise to keep the rules simple. The
constraints imposed by the technology of writing and printing are the
subject of the next chapter.

This chapter has reviewed some aspects of the use of typography for
displaying the structure of a text’s topic. Typography and diagramming
were seen as literal instances of visual metaphors used in the context of
rhetoric (Nash 1980) and semantics (Lyons 1977). The next chapter will
review the next of the three basic structures posited by the genre model –
artefact structure.

137 Other examples of information mapping do not capitalize headings in this way, so this is probably
the result of intervention by the publisher of the book in which this chapter appeared. Nevertheless, the
Information Mapping manual remains silent on this question.

138 As a result of presenting his work at conferences that included typographers, Horn is aware of the
graphic deficiencies of his system. The Department of Typography & Graphic Communication at Reading
University has produced more elegant typographic solutions to the problems of Information Mapping.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 172
Chapter 7 • Artefact structure

7
218

Artefact structure

This chapter examines some influences of the technicalities of text

production on what may be expressed. First I shall discuss problems
associated with the segmentation of language – how artefactual
units such as the line and page influence the display of semantic or
linguistic segments. I shall go on to consider some broader influences of
communication technology on what is or may be expressed.

Figure 7.1 Diagram from Kinneir (1984: 348) explaining the layout system for British road signs.

In certain special circumstances, there is no predetermined limit to the

size of the page or frame. In British traffic signs, for example, the size is
determined by the content. Kinneir (1984: 347), the designer of the system,
describes how ‘with the layouts there was a fundamental difference of
approach from the usual typographic practice’. British road signs are laid
out as a diagram of the road ahead as viewed by the driver. Text is placed 219
at a minimum distance from lines representing roads, and only when the
layout is completed is the outside frame determined – again, by a minimum
distance. Figure 7.1 illustrates the principle. In book illustration, the

Robert Waller • The typographic contribution to language • PhD Thesis 1987 173
Chapter 7 • Artefact structure

opposite is sometimes the case, since illustrations can, within reason, be

reduced or enlarged to fit a given area. But where both the page size and
the image size are inflexible, the artefact makes itself known.

Levels of graphic segmentation

A child learning to read must come to realize that while some breaks in
the string of letters to be deciphered are meaningful, others are almost
completely arbitrary. Spaces between letters indicate a word break, and in
some early reading materials a new line indicates a new sentence, and a
new page announces a new topic. But not always, of course. At some point
we learn that some, and eventually most, line breaks have no meaning – we
have simply come to the edge of the column.

Writing in columns originally developed, not because of the effect of page

size, since papyrus rolls offered an unlimited page width, but for other
functional reasons. According to Thompson (1912), column widths varied
greatly among Greek papyrus rolls. Apart from considerations of legibility,
the maximum column width was presumably dictated by the amount of
the writing surface to view as the papyrus was unrolled with one hand
and rolled up with the other. However, legibility does seem to have been a
factor quite early on, since columns ‘were generally narrow in texts written
for the market by skilled scribes’ (p. 46). Moreover, after the development
of the codex, when the page width might otherwise have dictated the
column width, ‘continuing the practice observed in the papyrus rolls, the
arrangement in [two, three or four] columns was usual’ (p. 55). This is
confirmed by Turner’s (1977) extensive survey of early codices.

According to Thompson, even though word separation was rare in Greek 220
and Roman manuscripts, line breaks were made much as they are today –
between words where possible, and otherwise between syllables (although
the hyphen was not introduced until the eleventh century).139 O’Hara
(1971: 113) supports this view, but suggests that ‘with the implementation
of printing, both the rigorous employment of the hyphen where it was
called for and the “correct” division of words into syllables in turnovers fell
into disuse’, a fact he ascribes to commercial pressures.

In recent years a number of people have suggested reforms to the

convention by which we end lines according to what Twyman (1979)

139 Saenger (1982: 371) notes contemporary evidence that ‘Caesar Augustus, in his autograph letter,
had the peculiar habit of connecting with a long loop the last syllable of one line to the first syllable
of the next line when the two syllables formed part of the same word, a practice illustrating the
idiosyncratic attempt of one author to overcome difficulties facing all Romans when writing in a script
lacking word division’. This use of lines to connect the end of one textual unit to the beginning of the
next is also manifested in some recent government forms where respondents must follow different
routes depending on their response to an earlier question. The alternative paths are indicated by lines,
reminiscent of those painted on hospital floors to help people navigate complex routes.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 174
Chapter 7 • Artefact structure

calls quasi-semantic rules (‘with the lines broken only between words
or within words according to etymology’). Twyman distinguishes ‘quasi-
semantic’ from ‘semantic’, ‘partially semantic’ and ‘mechanical’ word breaks.
Semantic word breaks are common in the case of unjustified type (with a
ragged right-hand edge). Partially semantic breaks are those ‘with the lines
broken between words or within words either phonetically or arbitrarily’.
In mechanical word breaks lines are broken at the most convenient point
regardless of meaning. There might be a case for merging the quasi- and
partially semantic categories, since in practice the choice of etymological or
phonetic grounds for breaking words is not always consistent and is largely
a matter of taste.

In the discussion that follows I shall use the term ‘arbitrary’ to mean any
break in lines, columns or pages that is prompted solely by the edge of
the type area. Since the word-break system employed is not generally
varied within a particular document, it does not form part of the system
of contrasts through which writers can create meanings,140 and so can be
bracketed with other global stylistic choices such as the page size. 221

De Vinne (1901) considered the hyphenation of words at line-endings to

be a waste of time and a needless source of difficulty to the printer, who
cannot reasonably be expected to have the expertise to make etymologically
correct word-breaks. While modern typographers would solve the problem
by abandoning the justification (alignment) of the right hand edge of the
column, this is rejected by De Vinne, who prefers the more radical method
of arbitrary word breaks, inserted wherever the line-ending happens to
fall. He is able to cite a precedent in the work of the eminent eighteenth
century printer John Baskerville, in whose edition of Paradise Lost he found
such unorthodox word-breaks as ‘e–specially’ and ‘o–therwise’. Figure 7.2
illustrates a modern example of truly arbitrary word-breaks.

Figure 7.2 Instructions to a gadget bought by a friend in Singapore

140 A relatively minor exception to this is where unjustified type, which some typographers prefer
without word-breaks, is used as part of a stylistic distinction between text components. However such a
distinction would normally involve additional variations in typeface and size, since justification is not a
prominent enough cue on its own.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 175
Chapter 7 • Artefact structure

However, De Vinne does not employ such a system himself, regarding it as

an ‘ideal’ rather than a recommended practice. He explains that
‘It is not probable that this innovation will find favor with the critical, but it
may be mentioned as an exhibit of increasing restiveness at grammatical and 222
typographical shackles which annoy the reader and do not help and do hinder the
proper rendering of printed words.’ (De Vinne, 1901: 143)

He goes on to argue that, just as readers have learned to do without catch-

words to help them over page breaks, they can equally easily learn to deal
with arbitrary hyphenation.

Most other would-be-reformers have gone the other way and suggested that
line endings should be made more meaningful not less. That is, that they
should mark significant breaks between linguistic or semantic units. While
this is a normal and uncontroversial practice commonly recommended
when breaking display headings and titles (Dowding 1966), a number of
experimental studies have tested the application of this and even more
radical related principles to continuous prose.141

Andrews (1949) proposed what he termed ‘square span’ typography, in

which phrases were grouped in small stacks, but his experimental results
were inconclusive:
Andrews (1949) what he termed in which phrases were
proposed ‘square span’ typography grouped in small stacks.

North & Jenkins (1951) moderated the proposal by suggesting that it was
the spacing of phrases, not the stacking, that was important. They reported
small increases in both speed and comprehension with their ‘spaced
typography’:
In spaced typography, extra space is added between ‘thought units’.

However, although a number of others have investigated these options,

most fail to prove their advantages or admit that the evidence to support
them, if any, is very slim (for example, Klare, Nichols & Shuford 1957;
Coleman & Kim 1961; Carver 1970; Wendt 1979). The principles for
dividing lines into phrases or clauses are mostly intuitive, although Klare et 223
al articulated some rules followed in their study. All of the studies illustrate
just ‘one line’ of the square span format, and it is not clear how multiple
lines would be spaced.

Semantic or syntactic line breaks offer rather more hope of acceptance

by readers, since they do not look startlingly unusual. Coleman & Kim

141 Perhaps I should just say ‘prose’ since the suggestion under consideration is, in effect,
‘discontinuous prose’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 176
Chapter 7 • Artefact structure

(1961), inspired by children’s books which employed this system,142 did

not obtain a significant result from their pilot study, but others seem to
have been sufficiently encouraged to pursue the idea. Frase & Schwartz
(1979) reported an impressively faster (14–18%) response time for a
task which required subjects to verify the answer to a question from the
experimental text; this represents a typical use of a technical manual
but does not resemble the reading of ordinary prose where fluency is
rather more important. In fact Raban (1982), who studied the effect of
such line-endings on children’s reading, found that syntactic breaks were
mistaken for the ends of sentences. It also seems strange to suggest that
a particular punctuation technique (for that is what line-breaks would
become) should be distributed evenly throughout a text, and thus be
determined by line length as well as sense.143 Hartley (1980) criticized
Frase & Schwartz’s methodology and failed to replicate their findings under
different conditions.

Figure 7.3 shows a rare instance of the system in use. Gerstner (1974) uses
it as a component of his ‘integral typography’, some aspects of which he
defines on the pages illustrated (p. 136-137).

Each reader can decide for himself or herself whether their own reaction 224
corresponds to my own: that the poem-like quality of the system draws
attention to Gerstner’s language and, paradoxically, away from his sense.
Poets, of course, have long been aware of the typographic dimensions
to language, which include the shape of stanzas (even to the extent of
Herbert’s shaped poems) and visual rhymes as well as line breaks.

142 The books in question were published in the 1940s by Lillian Lieber. Burt (1959) also recommends
this practice but does not cite any precedents or research evidence.

143 There is an eighteenth-century precedent for this suggestion. Robertson (1785: 75) cites Walker’s
Elements of Elocution thus: ‘An ingenious writer has observed, that not half the pauses are found in
printing, which are heard in the pronunciation of a good reader or speaker; and that, if we would read
or speak well, we must pause, upon an average, at every fifth or sixth word’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 177
Chapter 7 • Artefact structure

Figure 7.3 Gerstner’s ‘integral typography’ includes semantic line breaks

Crystal (1979) also reviews the idea of semantic line breaks and 225
discusses the issue of line endings in the context of his fourteen levels of
graphological organization in text (described in Chapter 1, Table 1.13).
He correlates each level of graphological organization with other linguistic
levels of analysis – phonology, grammar and semantics. Below the level of
the line various correlations with phonology are identified, as well as two
with grammar, and two with semantics. Above the line all correlations are
with ‘a statable information structure in semantics’, while the line itself
is the only level at which Crystal finds no correlations (with the single
phonological exception of metrical lines in poetry). His analysis effectively
concludes that, in continuous prose at least, graphological units below the
line level can be considered invariant aspects of the writing system and

Robert Waller • The typographic contribution to language • PhD Thesis 1987 178
Chapter 7 • Artefact structure

therefore not really questionable for practical purposes. At and above the
line level, though, there is more room for debate, particularly since the line
emerges from this analysis as the only graphological unit with no linguistic
or semantic status.

This is a surprising conclusion for two reasons. Firstly, it seems to ignore

instances where lines do have an independent semantic status – such as
in lists or examples of block language (signs, headings etc). Secondly, if
line breaks can be arbitrarily determined by the printing process, so can
page breaks, which in this analysis are accorded semantic status. But in
conventional printed prose, line breaks are actually determined with a
greater measure of semantic consideration than page breaks. Lines can only
be ended at word or syllable breaks, whereas it is rarer for page breaks to
be manipulated for equivalent reasons (to prevent a widow, for example).

The problem highlighted here is that line-breaks, page-breaks and, in

the case of multi-column layouts, column-breaks can be either arbitrary
or meaningful. Table 7.1 suggests some of the semantic implications of
meaningful breaks.144

Arbitrary Meaningful 226

Single break Successive breaks
Line Prose New paragraph List
Verse
Column Prose New topic Table
Parallel text
Page Prose New topic Topic frame
New chapter

Table 7.1 Some semantic implications of meaningful breaks in the language string.

At the line level, an arbitrary break is clearly just one of the conventions of
the writing system that we take in our stride. Line breaks within paragraphs
are generally not specified by authors, although they may object to
awkward word breaks when they read their proofs. If a new sentence starts
on an unforced new line, though, we regard it as the beginning of a new
paragraph. If a succession of sentences, words or phrases begin on new
lines we are likely to regard them as forming a list.

144 Harris’ terms, ‘structurally superimposed’ and ‘structurally necessary’ (Harris 1986: 137), might be
substituted for ‘arbitrary’ and ‘meaningful’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 179
Chapter 7 • Artefact structure

Figure 7.4 This bibliography is undercoded: meaningful and arbitrary line breaks are hard to distinguish. Source:
Coulthard & Montgomery (1981).

In practice, meaningful line, column or page breaks are often given extra
coding to prevent ambiguity. Ambiguity is particularly acute when arbitrary
line-breaks occur in a list – where line endings would normally be seen as
significant (Figure 7.4). In such cases a second coding – numbers, bullets,
space between items, or indented turnovers – is normally added to clarify
the structure.

Paragraph breaks are almost always given a double coding – new line plus 227
indention, or new line plus blank line – in view of the frequency with which
sentence-breaks within paragraphs happen to coincide with line-breaks.
Figure 7.5 shows an example.

Figure 7.5 As can be seen on the left hand page, a line space is used here to indicate a major break in the text
(after Chapter 22, verse 29). However, small capitals are also used for the first word of the new section, in case the
line break coincides with a column break. This happens on the right-hand page at the start of Chapter 24 (modern
versions of the Bible follow the breaks in original sources, not the later, inappropriate chapter divisions). Source:
New English Bible. Dimensions: 116mm x 176mm.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 180
Chapter 7 • Artefact structure

An arbitrary break at the foot of a column is also a convention of continuous

prose, while a column break not forced by the foot of the type area would
generally imply the beginning of a new topic (usually additionally signalled
by a heading). A succession of column breaks (particularly when reinforced 228
by some connection among the column headings) may imply a parallel
structure. When each parallel column is broken into parallel lists, a table
may result.

At the page level, an unforced break (again usually reinforced by a

heading) signals a new topic – in practice, a new chapter, section or article.
Successive meaningful page breaks, though, form double-page spreads and
so define a complete ‘topic frame’ which bounds the discussion of a single
topic. It is notable that publications using many graphic effects frequently
opt for the treatment of pages as topic frames – not only popular handbooks
such as the Handbook of sailing but also technical manuals. For example,
Smillie (1985) describes the use of the page as topic frame in US Army
documentation.

Unlike columns, which can vary in height and width as their content
dictates, pages are invariable in size. There is therefore a trade-off
between this inflexibility and the ability of page-organized texts to use
two-dimensional diagram-like graphic effects to indicate topic structures.
One point we may make in defence of the practice of writing and designing
by spreads is that continuous prose is virtually the only format for discourse
that does not place limits on its length. Spoken addresses, such as speeches,
lectures and sermons, are ultimately bounded by the conventions of the
occasion or the attention span of the audience.145 The fixed time of the
school lesson is perhaps the most direct parallel to the treatment of a page
or double-page spread as a topic frame.146 In the educational context,
Duchastel (1982) has suggested larger page sizes for textbooks to enable
them to make better use of graphic techniques – fold-out posters that he
terms ‘unbounded text’.

145 According to Hunt (1970), the division of the Bible into chapters was determined largely by
the length required for a church lesson. As Figure 7.4 shows, the sense points identified by modern
translators do not always coincide with Authorized Version chapter divisions.

146 Saenger (1982) suggests that instances of medieval sermon texts organized as double spreads may
have resulted from the practice of transcribing sermons on to wax tablets, which effectively limit the
writer to a double spread. Margaret Smith (personal communication, University of Reading, January
1987) tells me that she has found several fifteenth century texts organized in pages, and mentions
Sebastian Brant’s Ship of fools in this connection.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 181
Chapter 7 • Artefact structure

229

Figure 7.6

Figure 7.7

The Procrustean page

The relative inflexibility of the page as topic frame can be seen in a
sequence of double page spreads from The handbook of sailing (Figure 7.6
and 7.7). It is slightly curious that the section entitled ‘Rigging the mainsail’
should precede the explanation of ‘Mainsheet systems’. Although most
sailors would agree that it is extremely impracticable to fit the mainsheet

Robert Waller • The typographic contribution to language • PhD Thesis 1987 182
Chapter 7 • Artefact structure

after the sail is hoisted, the instructions on rigging and hoisting the mainsail 230
make no mention of the mainsheet, which is mysteriously blanked out in
the ‘Hoisting the mainsail’ diagram (top right of Figure 7.6). However, if
the single page allocated to ‘Mainsheet systems’ were to precede ‘Rigging
the mainsail’, as would seem sensible, the latter two pages would no longer
form a single spread of facing pages. What appears to have happened is
that the (artefactual) need to fit each topic into a single or double-page
display has influenced the order of presentation. The sequence of topics
no longer reflects the ‘fact structure’ of the task but the technicalities of
the medium. While it illustrates the principle quite well, this example is
perhaps somewhat marginal – the reader is presumably expected to read
both sections (and more besides) before attempting to launch a boat. And
the writer could easily have included at least a mention of the mainsheet in
the rigging instructions. However, the overall impression of this book, and
others written in the same style, is that the argument has to be continually
stretched or condensed in order that it should fit into the Procrustean bed
of the page.

Figure 7.8 is from another home reference manual in which the page 231
has been subdivided into four smaller frames for three sub-topics. The
subdivision is shown through the use of prominent headings and also by
establishing a strong visual gestalt for each sub-topic with (in the absence
of horizontal rules) clear channels of white space. Notice, though, that
the white space is only as clearly defined as it is because the prose in each
section divides evenly into three columns. Since this is achieved not just
four times on one page, but throughout the book, it is clearly no accident.
Moreover, the topic fits exactly into the page, and the space between topics
is identical throughout the book. It is obvious, then, that each section,
subsection and even each caption in this book has been ‘written to length’
and that a process of what has become known as ‘cutting and filling’
(Rogers 1986)147 has taken place as the text has been typeset and made up
into pages.

147 According to Moran (1978: 4), this was once known, rather more imaginatively, as ‘soleing and
heeling’

Robert Waller • The typographic contribution to language • PhD Thesis 1987 183
Chapter 7 • Artefact structure

Figure 7.8 From The Reader’s Digest book of do-it-yourself skills and techniques, (1977). 189mm x 265mm.

De Vinne (1901) sees nothing wrong with requiring authors to adapt to the
constraints of the printing process, quoting Benjamin Drew (no date given):
‘Theories are elastic, – are expandable and compressible; but types of metal have
set dimensions of extension and in some circumstances will refuse to budge…
Types are tyrannical, and will sometimes perpetrate solecisms under the plea of
necessity’ (Pens & Types, p. 89)

Against that we may quote Henry Fielding, who is clearly against the
adjustments that must have been necessary in the Reader’s Digest example
we have just considered. He once likened newspapers ‘which consist of just
the same number of words, whether there be any news in them or not’ to
a stage coach ‘which performs constantly the same course empty as well as
full’ (Tom Jones, Book II, chapter 1).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 184
Chapter 7 • Artefact structure

Figure 7.9 Spread from Baby and child by Penny Leach (1977, London: Michael Joseph)

232

Figure 7.10 Version of the above published in 1980 by Shogakukan, Tokyo

We can see the Procrustean effect particularly clearly in Figures 7.9 and 233
7.10, which represent the equivalent double-page spreads from the English
and Japanese editions of the same book. The problem here is that the same
content is to be fitted into approximately the same size of page, but using
not only a different language but a different writing system. Besides the
obvious differences in script and reading direction, the relative economy
of the English writing system means that it effectively has the advantage
of a larger page format. As a result we can see that the Japanese translator
has had to cut a proportion of the material in order to fit the topic into its
frame: five illustrations have had to be omitted. It is worth noting, though,

Robert Waller • The typographic contribution to language • PhD Thesis 1987 185
Chapter 7 • Artefact structure

that the ability of the Japanese to write from top to bottom as well as right
to left gives them an additional technique not available to the designers of
the English version.148 It is very effectively used on the right hand page to
enhance the bracketing effect of the introductory section, and, on the left
hand page, to recover lost space by running around the central illustration.

We may also note that the designer of the Japanese version, by using rules
and boxes, has been more successful in structuring the material, particularly
on the right-hand page. This could be because the variable direction of
the writing system discourages the assumption that readers will always
move reliably from left to right without the cuing offered by the horizontal
rules; or because the Japanese writing system, unlike the English one, has
not been discouraged from using boxes and rules by five centuries of a
printing process in which vertical rules, especially, were difficult to handle;
or perhaps the smaller format places the alternative structuring technique,
white space, at a premium.

Grid systems 234

Most of the examples considered so far in this chapter are instances of

grid typography, introduced in Chapter 1. By subdividing the page, grids
increase the range of possible topic frames. Instead of just one or two
frame sizes (that is, a single or double page spread), a topic may take
any combination of grid squares as its frame while retaining the standard
widths that have traditionally been required for typesetting and picture
processing.

148 Encouraged in part by evidence that Chinese readers read vertically slightly faster than horizontally
(Tu 1930, cited by Tinker 1955), some have suggested a similar ‘vertical typography’ system for
readers of English, in which one word appears on each line. Tinker (1955) reported that the vertical
arrangement slowed readers down, but that they improved with practice. Coleman & Kim (1961)
obtained promising results with vertical arrangements which Coleman &Hahn (1966) were unable
to replicate. A Japanese acquaintance tells me that the vertical arrangement is more normal and is
preferred, but that the horizontal system was introduced in order to be able to incorporate Roman script
for western names, numerals, and certain scientific terms. This also explains the apparent anomaly that,
although the page sequence in Japanese books is the reverse of the European convention (they start at
what to us is the back of the book), the individual columns are read from left to right.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 186
Chapter 7 • Artefact structure

1 2 3 4 1 1 1 1

2 2 2 2

3 3 3 3

Figure 7.11 Columnar grid Figure 7.12 Parallel columnar grid

We can distinguish several different kinds of multi-column arrangement.

Simple multi-column grids, in which text flows from one column to another
as if they were pages, can be termed columnar grids (Figure 7.11). The
telephone directory is an example of a simple columnar grid. In parallel
columnar grids (Figure 7.12) the text still flows vertically down the page,
but the content of the parallel columns is related horizontally. One column
may contain headings or marginal notes related to contents of the other
column.149 Figure 1.4 showed an example in which the parallel columns 235
represent the same text in three languages.

Figure 7.13 Modular grid

149 Duchastel (1985) discusses marginalia in some detail.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 187
Chapter 7 • Artefact structure

The classic Swiss grids, based on regular rows as well as columns, are
modular (Figure 7.13). Topic frames may be constructed from any (usually
rectangular) combination of modules. Truly modular texts are rarely found
– whereas typography textbooks (for example, Rüegg & Fröhlich 1972)
usually illustrate strictly modular grids, their practical examples rarely make
use of the standardized horizontal alignment points. Most implementations
of the grid system use what might be termed blocked grids, which are
similarly composed of rectangular frames, but only their width, not their
height, is determined by the grid. The Handbook of sailing uses a blocked
grid in which the designer can use a two, three or four column arrangement
in any given horizontal strip of the page.

236

Figure 7.14 Irregular grid

Irregular grids are built up from standard column widths but topic frames
are not always rectangular. Most tabloid newspapers use irregular grids150
in which editorial and advertising items are interwoven to create an
impression of variety and compete for the attention of the browsing reader
(Figure 7.14). Their purpose is the very opposite of the cool impression of
order found in Swiss typography – to prevent, not to create, clear visual
gestalts. The disordered but compact pages of tabloids appear to give value
for money – since the reader can never take the page in at one glance, there
might always be something that has been missed.

De Hamel (1984) has provided a detailed and fascinating account of the

page layouts used for the production of twelfth-century glossed Bibles,
in which the parallel texts of scripture and commentary must be laid
out side by side.151 The problem was that the proportion of scripture to
commentary varied throughout the text. In the first half of the century,

150 Hutt (1967) and Evans (1973) discuss newspaper design in detail.

151 A recent review by Gibson (1986) is somewhat sceptical of de Hamel’s claims.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 188
Chapter 7 • Artefact structure

parallel columnar grids were used. The vertical columns were pricked 237
through an entire quire, although sometimes adjusted in width according
to the proportion of gloss to scripture on a particular page. The scripture, in
large script, would occupy the central column, while the glosses would be
placed at appropriate points in the margins, sometimes forming L-shapes by
extending into the head or foot margins (Figure 7.15).152 The introduction
of greatly expanded glosses, though, placed too much strain on this system,
and were eventually produced as continuous texts with the scripture
omitted – an arrangement which suited the more learned scholars, who
would have memorized much of the relevant scriptures, but which did not
suit some of the wealthier purchasers of glossed Bibles (Figure 7.16).

Figure 7.15 Figure 7.16 Figure 7.17

According to de Hamel, the problem was solved by the development of

what was, in effect, a modular grid system, in which all columns shared the
same horizontal alignments. The scripture column, still in larger script, used
only alternate lines of the grid; and since its width could vary according
to the proportion of related gloss, the individual glosses were irregular in
shape (Figure 7.17). The modular system thus allowed scripture and gloss
to be interleaved in a complex but easily executed way. De Hamel argues 238
that the layout’s modularity (although he does not use this term) also made
it easier to understand since each gloss could be aligned correctly and
reliably with the relevant passage of scripture.

The logic of assembly

Modular grid systems exploit what, adapting Gombrich’s ‘geometry of
assembly’ (1979: 9), we might call the logic of assembly. Gombrich’s
argument is rather different to the present one and he does not develop
the term very far. Discussing decorative art, he is simply concerned to point

152 The layout of de Hamel’s own book illustrates the principle quite well. He uses a parallel grid in
which footnotes, smaller illustration and references to plates appear in the margin opposite their point
of reference in the text. However, occasionally a marginal item gets pushed onto the next page through
lack of space. Unfortunately modern production techniques are less flexible than those of the scribes,
and he is unable to vary the layout page by page to suit its content.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 189
Chapter 7 • Artefact structure

out that simple and predictable visual patterns are easily understood but
monotonous. The pattern made by four or five flagstones explains a whole
pavement and we need look no further:
‘When the expected happens in our field of vision we cease to attend and the
arrangement sinks below the threshold of our awareness.’

The term implies that methods of assembly impose their own visual logic on
the things assembled, whether they are corn cobs, sea urchins or pavements
(Gombrich’s examples) or books. As readers, we understand that many
features of typographic pages are artefactual, not linguistic – line breaks
and page breaks are virtually ignored by fluent readers and not mistaken
for linguistic signals. At the end of a line or page we turn automatically to
the next. A consistently applied grid system might be similarly internalized
by readers, and complex texts read with greater fluency and confidence.
Whereas the tabloid newspaper (Figure 7.14) deliberately never allows us
a sense of completion, the Reader’s Digest manual (Figure 7.8) conveys its
own structure at a glance.

Grids are used as an analogy by Harris (1986) to describe the relationship

of the alphabet to phonemes, likening the letters of the alphabet to map
reference grids imposed on speech. His purpose is to explain why alphabetic
writing is not purely phonetic, not an exact point for point transcription of 239
every contour of the landscape of speech. Harris’ analogy adapts well to the
present context.

Although Harris enjoins us not to make the basic mistake of taking the grid
lines on a map to represent streets, that is exactly the assumption readers
may be expected to make about typographic grids and pages when they are
used to delineate topic boundaries. As Figure 7.8 demonstrated, it is not
hard to find pages where the writer has ‘written to length’ – prepared each
section or paragraph to match the visual slot made available by the grid.
(Presumably the problem facing anyone who wishes to translate such texts
into a foreign language is not just the translation of the page’s meaning, but
also its design.) In terms of the relationship between maps and streets, we
can think of similar cases: the boundary between the USA and Canada, for
example, follows the grid line of a map projection rather than the other way
around. And the planning of streets by reference to grid systems is standard
practice in North America. The difference between the organic nature of the
earlier glossed books discussed by de Hamel, and the later modular ones,
is that between old towns and new. The archetypal typographic grid, like
town-planning grids, is used for pre-planning not description.

The logic of assembly involves not only those aspects of layout that are
predetermined by a grid, but the repertoire of characters and contrastive
techniques that are predetermined by available technologies. Mention
has already been made of the general switch from the use of colour in

Robert Waller • The typographic contribution to language • PhD Thesis 1987 190
Chapter 7 • Artefact structure

manuscripts to space in monochrome printed books (Twyman 1982).

Elsewhere, Twyman (1970, 1986) has traced the connections between
printing technology and the expressive repertoire in some detail.

Editorial intervention and artefact structure

Clearly it is not only printing technology but editorial practice that
can act as a constraint on the writer’s means of expression. One of the
improvements over the traditional straight-line communication model 240
claimed for my ‘dog-leg’ model is that it draws attention to the mediation of
industrial processes in the transmission of text from writer to reader. These
processes are institutional as well as technical.153

House styles, policed nowadays by copy-editors, are meant to ensure clarity

of expression, and to make the printer’s task easier and more consistent, but
authors are sometimes ambivalent about their usefulness. Most copy-editors
would agree with Butcher (1975: 1) that their role is ‘to remove any
obstacles between the reader and what the author wishes to convey’, but
it is not always easy to identify the obstacles to an author’s satisfaction.
Certainly most would accept the concept of a spelling mistake, but the
‘rules’ of syntax, punctuation and especially paragraphing are rather more
tenuous. Even if authors can insist on their own punctuation practice,154
it is almost unheard of for them to be able to influence typography once
it has become enshrined in a house style.155 When highly rule-bound and
institutionalized, it seems, such conventional or genre structures can take
on the inflexibility of artefact structures.

Bibliographers are especially mindful of intervention by editors and printers

in the historical context. For most of this century bibliographers have been
sensitive to the materiality of the texts they study. However, paper, type and
ink mostly seem to be of interest as evidence for the dating or attribution
of particular editions, and for routine description (McKerrow 1928) rather 241
than as anything constitutive of the literary work itself. A central task of
bibliography is to establish critical editions of literary texts that reflect the

153 Many media sociologists would maintain that they are political also – but issues such as censorship
and press ownership would seem to be far removed from questions of typographic theory.

154 Pullum (1984) rails against the punctuation (and other) policies of American copy-editors in an
amusing but, in my experience, entirely accurate article entitled Punctuation and human freedom. He
especially objects to their insistence on placing quotation marks outside adjacent punctuation, whatever
the circumstances: anyone who has been published in the USA can confirm that it is impossible in the
USA to distinguish between, say, /He said ‘I’m leaving!’ / and/He said ‘I’m leaving’!/. As Pullum puts it,
‘many advanced cultures show no sign of the superstitious awe with which we regard copy-editors’.

155 My own experience of co-editing a volume for Academic Press in the USA was of an inability to
persuade the publisher even to drop the excessive capitalization of chapter titles. After some negotiation
on the exact form of words, we were able to insert the following neutral sentence into the preface: ‘The
typography of the volume conforms to the standards of The Educational Technology Series.’ (Duffy &
Waller 1985: xv)

Robert Waller • The typographic contribution to language • PhD Thesis 1987 191
Chapter 7 • Artefact structure

author’s original intentions and that minimize the problems of what Gaskell
(1978) refers to as ‘the variation of transmission’. However, although
extremely detailed analyses of typographic factors are routinely used to
trace the origin of particular editions, there is some controversy about the
extent to which they should be considered part of the ‘author’s intent’.
McKenzie (1981, 1986), especially, has taken issue with the bibliographic
neglect of visual aspects of language.

Printed editions of literary works vary from their manuscripts not only in
the obvious respects of letterforms, line-endings and page breaks, but also
in such matters of spelling, spacing and punctuation as may be included in
the publisher’s house style. The bibliographic problem is to determine which
of any changes that can be found between editions reflect the author’s
intention, and which are the result of unwarranted intervention by the
printer or an editor. On the whole the actual words used are usually safe,
but their typography, spelling, capitalization, emphasis and punctuation
are not. It seems that the propensity of some authors to leave most aspects
of punctuation, typography and even spelling to the printer, and of some
printers to intervene even when it was not required, has led some editors
to regard these matters as fair game for alteration (not only for popular but
also critical editions). Sir Walter Greg (1950-1/1960) regarded the author’s
words as ‘substantives’ and the other matters as ‘accidentals’, but the exact
borderline – indeed, the very distinction – is controversial.

Reference has already been made in Chapter 4 to Simpson’s (1911)

defence of Shakespeare’s printer against later editors who assumed that,
because it did not correspond with their own grammatical practice, the
dramatic punctuation of the first folio was incorrect. This is echoed in the
introduction to PH Nidditch’s critical edition of Locke’s An essay concerning 242
human understanding (1975):
‘The idea still persists that English printers in the seventeenth century took little
notice of the formal features of an author’s work as displayed in his manuscript.’
(p. xlviii/xlix)

Nidditch cites evidence, from contemporary works by Milton and Newton

for which manuscripts survive, that printers were more reliable than
is generally supposed.156 In Kuhnian terms, this may be evidence of a
paradigm clash – the editors in question, applying a modern paradigm, are
perhaps unable to see the reasoning behind the older system and put it
down to ignorance or carelessness.

156 McKenzie (1986), who discusses some lines by Congreve with similar conclusions, suggests
that subsequent editings have their own inherent interest for the critic: ‘By reading other forms of
[Congreve’s prologue to The Way of the World], we can chart meanings that later readers made from it
under different historical circumstances.’ (p. 13).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 192
Chapter 7 • Artefact structure

Price (1939) cites evidence to defend sixteenth- and seventeenth-century

printers against the charge that they made free with an author’s grammar,
but in doing so reveals where he draws the line between substantive and
accidental features. The compositor of Sir John Harington’s translation
of Ariosto in 1591 evidently modernized his spelling, a feat for whose
consistency Price is full of admiration:157
‘All these corrections he did of his head without any changes in the copy to guide
him, and besides he was continually altering the punctuation. Yet his fidelity to the
text is marvelous.’ (p. 543, my emphasis)

Price’s apparent acceptance of the continual alteration of punctuation

as consistent with fidelity to the text presumably assumes fidelity to the
sense of some true text that the manuscript only pointed toward but did
not achieve. Harington was evidently happy with the changes – indeed,
expected them – Gaskell (1978), who discusses the same work, tells us
that Harington probably supervised revisions to the punctuation at proof
stage. McKenzie (1981: 105) points out also that, from the evidence of
his preface, Harington considered the needs of his readers carefully, and
that the layout of the text (an extremely lengthy epic), with its marginal
notes, index and illustrations ‘clearly demonstrates the finely planned and 243
purposive nature of the typography’.158

McKenzie cites a wide range of convincing examples of authors who are

clearly aware of typographic form, and deploy it along with other linguistic
resources.159 He appeals for Greg’s substantive/accidental distinction to be
rejected in favour of a broader ‘sociology of the text’. However, the problem
will presumably remain that, unless books are to be reproduced in facsimile
– and some, including works by Blake, Apollinaire and the concrete poets
cannot be reproduced any other way160 – editors are forced to make
choices: no one expects the modern reader to cope with the long ‘s’, for
example.161 To the modern reader the distinction between the long and the
short ‘s’ is meaningless, and spellings that looked normal to a seventeenth-

157 The Harington manuscript was first discussed in detail by Greg (1923/1960).

158 It is interesting to see from the photograph of Harington’s manuscript reproduced by Greg that,
where a modern author would write his or her manuscript in the same hand and add instructions to set
certain parts in italic, Harington actually adopts a different, cursive, hand for those parts (although this
is apparently not consistent throughout the manuscript).

159 Bronson (1968) and Barker (1981) also describes the relationship between literary style,
typographic format and other contemporary arts in the eighteenth-century.

160 There may be a case for a continuum of graphicness, analogous to Crystal’s continuum of
linguisticness discussed in Chapter 3. Massin (1970) reproduces an extensive range of texts – literary
and otherwise – that merge graphic and linguistic features, including letters made from pictures,
pictures made from letters, concrete poems, picture-letter puns and so on. None of them can be ‘quoted’,
only pictured.

161 Bowers (1959: 148) reports that ‘McKerrow, though reluctantly, modernised the Elizabethan long
[s]; and this procedure has now become standard in old-spelling critical texts’. Presumably McKerrow
must have changed his mind at some stage since in An introduction to bibliography (1928) he implicitly
recommends the retention of the long ‘s’ in his advice on transcribing title-pages.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 193
Chapter 7 • Artefact structure

century reader just look archaic and opaque. If the past is a foreign country
then perhaps the skill of editing is akin to that of translating. It would be
interesting to conjecture about a wider range of such equivalences: for
example, modern readers do not need and may simply not comprehend
the use of braces to group the row headings in tables. Any editorial
discrimination is a recognition of some kind of distinction, if not between
the substantive and accidental, then between the relevant and irrelevant
for some stated historical or critical purpose. A sociology of texts would
presumably require editors to at least declare their position on a much 244
wider range of text features than is traditional. Twyman’s (1982) quest for
the underlying ‘language element’ that underlies typographic arrangements
is clearly as relevant to bibliography as it is to the specification of ‘device
independent’ displays.

Moreover, so long as so many authors continue to be indifferent to the

visual form of their work (or to capitulate to the forces of house style), the
distinction between substantive (author’s) and accidental (artefactual or
editor’s) features has some basis in present reality. Given that, as McKenzie
(1981) suggests, ‘modern books…are notorious for smoothing the text and
dull our sensitivity to space as an instrument of order’, it is little wonder
that editors, from their modern perspective, fail to notice the full texture
of literary works of the past. In addition to the historical dimension of a
sociology of texts, which would presumably seek to sensitize bibliographers
and literary editors to a broader range of factors than they now notice, we
perhaps need to extend our modern concept of literacy to include the full
range of expressive tools implied by that concept. If authors are to treat
the graphic arrangement of their words as substantive, they ought to be as
sensitive and fluent in that aspect of writing as with any other – especially
if they use the newer desk-top publishing systems with their extensive
typographic facilities.

Medium and message

This last suggestion begs an important question. Are writers – or what ever
less-restricted term might be substituted for one who prepares graphic
language of whatever kind – indifferent to graphic forms because they have
nothing they wish to say that requires them, or is it the other way around?
‘The other way around’ being: do they find little to express graphically
because they are unaware of (or unable to operate) the means of saying it?

This question has been debated by linguists for many years, in the form of
the Sapir-Whorf hypothesis, named after the American linguists Edward 245
Sapir and Benjamin Lee Whorf , who maintained that languages determine

Robert Waller • The typographic contribution to language • PhD Thesis 1987 194
Chapter 7 • Artefact structure

not only the way we express our thoughts, but the way we think.162
Gombrich (1984: 188) picks up this debate, describing how:
‘in describing the same painting in German or English I had to take the goods
which were on offer163 and thus had to single out different aspects of the same
painting…The grid or network of language we impose on the landscape of our
experience will inevitably result in different maps.’

Gombrich’s remarks bring to mind Harris’s use of the map reference grid
analogy, discussed earlier in the chapter. Indeed, the metaphor of language
as a grid seems to be quite a common one. It is found in Saussurean
linguistics, where words are said to relate to other words both horizontally
(syntagmatically) and vertically (associatively). Ivins (1953: 53) compares
both words and images to fishing nets that only catch such fish as cannot
swim through or escape:
‘in the same way words and visual images catch only the things or qualities they
are adequately meshed for’.164

Although most scholars who discuss the relationship between language

and concepts are careful to distance themselves from Whorf (for example,
Gombrich 1984: 189; Goody 1977: 9), the relationship of artefact and
expression, medium and message, has been discussed widely in recent
years. ‘The medium is the message’ is a catch-phrase associated with
Marshall McLuhan (1962) whose Gutenberg Galaxy was largely responsible
for bringing the study of media effects to the wider public consciousness. 246
So many aspects of modern Western society are bound up with literacy that
it is of obvious interest to contrast non-literate and literate societies, and
to link the introduction of literacy to social, economic and political events
in history. Some very bold claims have been made for the influence of the
new communication technologies on the history of ideas – and even on the
evolution of human cognitive processes.165

In cultures without writing – parallels are drawn between ancient

pre-literate civilizations and modern non-literate tribal cultures – it is

162 Many textbooks on linguistics and semantics contain summaries of the debate surrounding the
hypothesis, which is also known simply as the ‘Whorfian hypothesis’ or as ‘linguistic relativism’ (eg Lyons
1977: 245; Sampson 1980: 81; Harris 1981: 131). Lyons contrasts it with functionalism – whereas
relativists would claim that languages delimit thought processes, functionalists would maintain that the
structure of different languages results from the expressive requirements of particular societies.

163 Gombrich’s use of the term ‘goods’ alludes to a sonnet by IA Richards in which he, in effect,
comments on the Sapir-Whorf hypothesis: ‘Our mother tongue, so far ahead of me, / Displays her goods,
hints at each bond and link, / Provides the means, leaves it to us to think,/ …’

164 Elsewhere, in a variation of the metaphor, he criticizes the excessively systematized techniques of
certain virtuoso engravers as ‘webs spun by these spiders of the exactly repeatable pictorial statement.’
(Ivins 1953: 71)

165 Scribner & Cole (1981) describe this debate and, on the basis of their comparison of literates and
non-literates in an African society, suggest that any improvements in performance of intellectual tasks
due to literacy are confined to the individuals concerned and are task-specific: ‘There is nothing in our
findings that would lead us to speak of cognitive consequences of literacy with the notion in mind that
such consequences affect intellectual performance in all tasks to which the human mind is put.’ (p. 86)

Robert Waller • The typographic contribution to language • PhD Thesis 1987 195
Chapter 7 • Artefact structure

generally agreed that knowledge is heavily coded for memorization and

recitation.166 It is formulaic and rhythmic in structure, and often expressed
in terms of proverbs, legends, riddles and verses. Meaning is thus not
explicit, and interpretation relies instead on prior knowledge, context and
‘wisdom’ – a priesthood, even, qualified to interpret. Several commentators
(for example, Havelock 1976, 1986; Olson 1977) regard the relative
autonomy of written language – its ability to convey reliable and consistent
meanings across differences in audience and context – as central to its
cultural significance. However, considerable confusion seems to surround
the part played by graphic features in these developments. Whereas
Eisenstein (1979) regards the introduction of spatial features as a major
contribution of the introduction of printing towards the development of
modern science, Havelock – whose subject is literary rather than functional
text – appears to see it as retrograde.

Havelock (1976) has argued that, because oral culture is so ritualized, 247
opportunities for creative expression and individual interpretation are
strictly limited. The mnemonic nature of oral texts means that they act as a
force for the retention of existing knowledge within a society rather than a
tool for exploratory thinking and debate. In Havelock’s view, early writing
systems, too – pictographies and syllabaries – were not sufficiently precise
to act as more than aids to memory and should therefore be considered
as features of oral culture. Havelock sees the development of alphabetic
writing by the Greeks as the key to the true literacy that, since it enabled
the fast and accurate transcription of speech, provided the basis for texts
whose meaning was autonomous and therefore potentially original.
Following on from Parry (1971), who identified the formulaic structure
of Homer as characteristically oral, Havelock (1986) has traced the
development of Greek thought and language as the autonomous nature of
written text was discovered. His main interest is in the tracing of linguistic
changes after alphabetization – the gradual replacement of what he terms
the language of doing with the language of being:
‘the linguistic symptoms of this radical shift away from oralism … occurred in a
proliferation of terms, for notions and thoughts and thinking, for knowledge and
knowing, for understanding, investigating, research, inquiry.’ (p. 115)167

Interestingly, in the present context, he also suggests that ‘topicalization

slowly increases its presence in classic Greek’ (p. 103).

166 There is a considerable literature on orality and literacy, some of it somewhat apocalyptic in
character (for example, Innis 1951, McLuhan 1962). Ong (1967, 1982), who appears at times to be
gripped by a nostalgia for tribal culture, celebrates orality at some length in contrast to what he terms
the ‘logocentricity’ of Western civilization; an influential anthropological and sociological view is
argued by Goody (1977, 1986); and current linguistic and psychological interest is represented in the
collections edited by Tannen (1982) and Olson, Torrance & Hildyard (1985).

167 Lloyd (1966) also remarks on the development of new terms, or at least adaptations and better
definitions of existing terms, for use in logic.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 196
Chapter 7 • Artefact structure

Two puzzling aspects of Havelock’s argument may be noted here. Firstly, his
claim for such overwhelming advantages for alphabetic writing rests on the
assumption that other writing systems provide little more than a mnemonic
for the recital of known information, and are therefore incapable of
supporting abstract or creative thought. Indeed, at one point, with breath-
taking chauvinism, he suggests that
‘a Japanese can orally express what the West has taught him. Transferring the
statement to his own script, he will then be able to recognize and to read what
he already knows, as did the scribes of antiquity. But the free production of novel
statements in his own script will remain difficult.’ (Havelock 1976: 84)

Secondly, the phonetic equivalence of alphabetic writing, coupled with

the normal lack of word separation (see Chapter 4), links Greek writing
closely with speech. Indeed, Havelock (1976) appears to regard the Greek
alphabet as a complete inventory of the phonemes in the Greek language
– a view dismissed by Harris (1986: 118), not in specific relation to
Havelock, it should be added, as ‘simply a fourth-form howler of the most
elementary order’. So, while basing much of his theory on the objectifying
effect of writing that separates language from its speaker, Havelock prefers
to ignore, even to exclude, visual aspects of the medium.168 This is surely a
strangely misguided purism, since the ability to manipulate ideas in space
(rather than in linear acoustic form) would seem to give the logician an
advantage – witness the development of symbols in logic, introduced to
a limited degree by Aristotle (Lloyd 1966).169 It is possible, then, that if
progress in Greek philosophy can be linked to writing – and it seems to be
generally accepted that it is170 – it was achieved not as a direct result of
their particular technology of writing but as a more general consequence
of literacy – that more people could engage in dialogue and that progress
could be recorded.

Although they mostly build on the work of the classicists (Parry and
Havelock), other scholars (Chaytor 1945; Ong 1958, 1967, 1982; McLuhan 249
1962), have seen the invention of printing, not of alphabetic writing, as
the pivotal event that turned Western civilization from a mainly oral to a
literate tradition.

168 Elsewhere Havelock (1976: 15) is uncompromisingly opposed to any attention to the visual
appearance of writing: ‘Strictly speaking, writing should behave solely as the servant of the spoken
tongue, reporting its sounds as accurately and swiftly as possible […] it is a sign of the arrival of modern
scientific and socialized man that calligraphy as an art form has largely expired.’

169 Evans (1980: 35) refers to evidence that ‘Aristotle’s works were almost certainly illustrated with
diagrams, and while it is unlikely that the Platonic dialogues were, commentaries on them employed
figures extensively’. Gardner (1958) discusses a number of instances of diagramming and modelling
by logicians from the middle ages to the present day. Even when they don’t actually use diagrams,
philosophers frequently employ diagrammatic or spatial metaphor: for example, Toulmin (1958) entitles
a chapter of The uses of argument, ‘The layout of arguments’, but solely in a metaphorical sense.

170 In A history of western philosophy, Bertrand Russell (1946) seems to take it for granted that the
writing systems available to early civilizations had a direct bearing on their progress in philosophy.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 197
Chapter 7 • Artefact structure

Ong’s account of the influence of the Ramist method has already been
described (Chapter 6). Just why he should attribute spatial thinking to the
technology of printing – a technology which to this day penalizes attempts
to integrate word and image – is not very clear. Yates (1966: 230) takes
issue with him on this point:
‘Rather, it would seem to me, the printed Ramist epitomes are a transfer to the
printed book of the visually ordered and schematized lay-outs of manuscripts.’

In part, Ong’s argument is similar to that of Ivins (1953), to which

reference was made in Chapter 3 – that reliance on graphic forms in the
manuscript age was inhibited by problems of inaccurate copying.171 But,
waxing somewhat metaphysical, he also suggests that the technology of
movable type, and therefore movable letters and words, was suggestive of
ideas as objects in space:
‘Now the printer’s font where types are kept comes into being – a real “place,”
where elements of discourse, reduced to a visually apprehensible and spatially
maneuverable form, are stored.’ (Ong 1958: 310)

Although he does at one point acknowledge that diagrams and spatial

arrangements were also used in the age of manuscripts, Ong’s idea of print
as a newly spatial medium gains authority in his writings from frequent
repetition.

Ong is rebuked for this oversimplification by Eisenstein (1979), who prefers

to see print as a transition not from an oral to a literate culture, but from
one kind of literate culture to another. Eisenstein, though, is questioned in 250
turn by Twyman (1986: 205) who remarks on her ‘repeated reference to the
proliferation of charts and tables…following the invention of printing’:
‘My impression…is that one of the consequences of the invention of printing was
to stifle the range of configurations of graphic language in much the same way as
it manifestly changed the book from a colored artefact to a monochrome one.’

Even today any arrangement that departs from conventional linear-

interrupted prose is difficult, costly and discouraged by publishers
(Biderman 1980). Twyman excludes non-linear images (printed as single
plates) from his comments, but even here it is clear that print enforced a
separation of word from image. Eisenstein herself remarks:
‘That the printed book made possible new forms of interplay between these diverse
elements is perhaps even more significant than the change undergone by picture,
number or letter alone.’ (p. 55)

Elsewhere, though, she appears to contradict this:

171 Although, whereas Ivins refers mainly to medical and botanical studies where verisimilitude is
relatively important, the Ramist charts Ong attributes to the age of print are rather more robust. In fact,
Evans (1980: 35) reports that even in the manuscript age ‘available evidence suggests that, compared
with representational images, diagrammatic designs are transmitted with remarkably little variation’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 198
Chapter 7 • Artefact structure

‘In the field of book illustration, at least, what happened in the late fifteenth
century resembled a divorce rather more than a reunion. When the graceful lines
that linked text to marginal decoration were severed, pictures and words were
disconnected’ (p. 258).172

This last comment is reinforced by Evans (1980) and Camille (1986), who
demonstrate the pervasive use of images and diagrams, highly integrated
with accompanying words, in the manuscript age. From this evidence,
together with other studies of medieval ‘typographic’ layout, it is becoming
clear that the modern genre of typographically organized book, such as
the Handbook of Sailing, has more affinity with medieval books than with
the typical products of the first few hundred years in which printing was 251
dominated by the letterpress system. The ‘standard book’ that emerged
from the first hundred years or so of printing discouraged the multi-column
layouts of medieval books (Parkes 1977, de Hamel 1984), their variety of
script styles (Ullmann 1932; de Hamel 1984), their use of colour as a cue
(Twyman 1982) and their close integration of illustrations and text.

The model outlined in Chapter 5 suggests that any such affinity between
genres (or common membership of a single genre) will be based not only
on a similar set of artefactual constraints, but also on similar demands of
the topic structure and similar patterns of access among readers. The next
chapter discusses some aspects of what the model termed ‘access structure’,
before going on to consider the nature of genre in more detail (Chapter 9).

172 It is actually rather difficult to determine exactly what Eisenstein’s position is. In a recently
published essay (1985), she discusses some of the motivations underlying The printing press as an
agent of change, suggesting that she developed her view that printing led to an increased use of iconic
images in response to an earlier characterization of herself as holding the opposite view. She claims
that imputed to her had been a formula that the advent of printing moved Western Europe ‘from image
culture to word culture’ – since ‘a latent iconoclasm was reinforced and the medieval justification for
allowing graven images in church was weakened by print’ (p. 20). Her revision of this earlier notion
may account for her apparent overstatement of the opposite ‘word to image’ formula.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 199
Chapter 8 • Access structure

8
252

Access structure

In this chapter I shall discuss further some aspects of the relationship of the
reader to the text that were introduced in Chapter 5. Although it has been
argued that technologies of writing and printing constrain what may be
said, it is also arguable that the relationship of medium and message works
both ways. That is, technologies are themselves developed in response to
the needs of users. While the invention of printing might have accelerated
the Renaissance, it was also a response to it – most of the technology had
been around for some time before the vital connection was made and the
market for books warranted the considerable investment required.

Designing for different purposes

In Chapter 7 it was noted that the requirements of medieval scholarship
(which was dominated by biblical scholarship) led to further developments
in book design to accommodate glosses in a more ordered manner.173
Gullick (1986: 207) suggests that
‘The work of assembling the authorities, comments, and of devising ever improved
layouts to make the act of reading easy is one of the great monuments of medieval
scholarship and page design’.

Even today the design of bibles can provide a good demonstration of the
influence of users on formats, since the Bible is an example of a text whose
wording, while it can be retranslated and glossed, cannot be changed 253
in substance. As Table 8.1 demonstrates, a wide range of user needs are
currently catered for. Bernhardt (1985) has also compared a range of texts
that address the same topic but with different purposes (see Chapter 1). In
his sample texts, though, everything about the text differs in response to the
needs of the anticipated audience – scope, argument and language, as well
as format and typography.

173 Modern biblical scholarship is also surprisingly relevant to the present study. The fields of content
analysis, hermeneutics and discourse analysis all have roots in the need to determine the authorship
of scriptures and to suggest procedures for translation into new languages. In particular the Summer
Institute of Linguistics, founded by the tagmemic linguist Kenneth Pike, is a missionary organization.
One of its associates, Robert Longacre, a major figure in discourse studies, is centrally concerned with
Bible translation.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 200
Chapter 8 • Access structure

Edition Purpose

Traditional leather-bound bible Binding (limp leather, rounded corners, etc) protects
against wear; double column is for legibility and
to display verse structure. May have ‘churchy’
connotations.

Pocket-size bible Requirements of legibility are subordinated to

convenience of carriage.

Tiny white bible To be presented or carried on special occasions;

available in presentation boxes for different occasions
(weddings, first communions, etc).

Paperback bible Cheap enough to be given away by evangelists and

the Bible Society.

Lectern bible Convenience of carriage and storage subordinated to

legibility and symbolic prominence.

The Bible designed to be read as To be read in continuous fashion; design discourages

literature (Heinemann 1937) the ‘proof-text’ style of reading.

Loose-leaf and wide margin To encourage cross-referencing and writing of notes;

versions used for sermon preparation.

‘Red-letter’ bible Words of Christ highlighted for devotional reading and

as an aid to rote learning.

Family bible Archival function, with space for a record of births,

deaths and marriages. May reflect a symbolic function
in its large size

Computer disk bible (For example, The Word Processor) include search
facility to remove need for separate concordance.

Chain-reference bible Designed specifically to encourage doctrinal study

through linked proof-texts.

Parallel & polyglot bibles Different languages or translations are printed on

opposite pages, or alternate lines, for easy comparison.

Children’s bible Includes pictures and explanatory notes.

Table 8.1 The different forms in which the Bible is available reflect the range of uses anticipated by publishers.

The interplay between function and genre is clear from some of these 254
examples – The Bible designed to read as literature, especially, proclaims
what it expects from its imagined reader by explicitly ‘quoting’ another
genre (classic literature). Children’s bibles can look like children’s fiction,
partly because they share the same functional constraints, but partly, one
suspects, to exploit the readers’ loyalty to the more popular genre.

Religious objects are, of course, particularly prone to acquire symbolic

connotations, however functional they may also be. Indeed, the
development of the codex form among early Christians is ascribed by
Roberts & Skeat (1983) to the demands of genre. Although codex-like
notebooks were in existence during the first century AD, the roll was the

Robert Waller • The typographic contribution to language • PhD Thesis 1987 201
Chapter 8 • Access structure

main book format and continued to be used for certain (especially legal)
purposes throughout the middle ages and beyond. Yet the vast majority of
early Christian writings are in codex form. Although its advantages seem
obvious to us today, Roberts & Skeat are not convinced that it was adopted
by Christians for exclusively practical reasons. For one thing, they were
not the only group for whom ease of reference and compactness would be
attractive. Moreover, those who were used to rolls appear to have found
little difficulty in finding their way around, and the surprisingly slow
introduction of seemingly obvious reference devices, such as line or page
numbering, indicates that cross-reference was not a priority in the early
church. Roberts & Skeat’s tentative solution to the problem is that the first
gospel, or, alternatively, earlier notes of the sayings of Jesus, might have
been written on codex-like notebooks and that the format might thus have
acquired a symbolic value (aided by its dissimilarity to pagan and Jewish
rolls).

This pattern of development seems to be entirely normal – that is, access

structures, those most functional and directly audience-related of text
components, are adopted, in part, because of their connotations. The
evolution of new methods happens because people copy good ideas – not
always because they have analysed them in depth. Black (1956, 1961)
traces the establishment of Bible printing practices in the first half of the
sixteenth century – the story seems to be of one major innovator (Robert 255
Estienne) responding to the needs of users (the new style of independent
Bible reader of the Reformation), and other printers copying the model
thus established. Black shows that Estienne himself owes much to the
manuscript tradition.

At much the same time, many of the access devices we now take for granted
developed in response to the growing number of readers, and the build-up
of science and literacy (Steinberg 1974; Eisenstein 1979). Indexes and
cross-references were made possible by the multiple reproduction of books,
but they relied, too, on those books being numbered.

Numbering systems
A document without page numbers is almost unthinkable today, even when
some other system, such as paragraph numbers, is also present. However,
page numbers, a system on which a number of other access systems
depend, appear to have taken some time to become fully established.
Although a fair number of early books had page numbers (Turner 1977),
Roberts & Skeat (1983: 51) report that
‘in the whole of ancient literature there is no example of a page reference being
given, and the reason is obvious, namely that no two manuscripts are identical.’

Robert Waller • The typographic contribution to language • PhD Thesis 1987 202
Chapter 8 • Access structure

Instead they suggest that page numbers were useful for binding and for
checking that no pages were missing.

The ubiquity of page numbering today is partly ensured by its inclusion in

printers’ and publishers’ house styles, but this was evidently not the case
with most early printers. Smith (in press) has suggested some reasons why
the introduction of folio or page numbering was relatively slow. Firstly, she
argues, other reference systems work just as well, and, secondly, in some
circumstances – fictional or devotional works, for example – bookmarks 256
suffice and reference systems are often unnecessary.174

Even so, the advantages of page numbers seem so overwhelming that

it is hard not to put down their slow introduction to the unquestioning
conservatism that is inherent to craft traditions. Quite apart from their use
in indexing and cross-referencing, numbering systems are, as Roberts &
Skeat suggest, useful for the making of books and documents. Even in the
printing of folio editions, quite apart from the multi-page sections common
today, pages must be laid out (or ‘imposed’, in printers’ jargon) so that
they can be printed from one forme and folded with the pages in the right
order. Even in the simplest documents, numbers are useful for collation and
checking.

Smith distinguishes between ‘arbitrary’ and ‘non-arbitrary’ numbering

systems, using the terms in much the same way as my own ‘arbitrary’
and ‘meaningful’ artefact structures (Chapter 7). Page numbers indicate
arbitrary divisions of the text – whether or not pages are treated as topic
frames. Many books contain more than one series of page numbers,
but generally for technical rather than semantic reasons. For example,
preliminary pages traditionally employ roman numerals, with the main
arabic series starting on the first page of the ‘main text’. The functional
purpose of this is to allow the preliminary pages and index to be compiled
after the main text has been paginated (Butcher 1975). Technical manuals
often use a separate numbering series for each chapter or section, so that a
single section can be updated without reprinting the whole text.

Non-arbitrary reference systems include: the numbering of lines (termed

‘stichometry’ by palaeographers),175 where line endings are meaningful,

174 On the whole, this is still the case, although the advent of literary criticism and media studies
means that any text is liable to be cited in an academic context. Page numbers are also useful when the
bindings of cheap novels disintegrate and pages must be reassembled. It should also be remembered
that, in the literary or poetic context, language is the artist’s subject as well as his or her medium.
Eighteenth century writers, such as Sterne, Fielding and Swift, were especially prone to comment on
formal aspects of books as printed objects. McKenzie (1986) has discussed specific allusions to page
numbering by Joyce.

175 Thompson (1912) reports that stichometry was mostly used as a means of computing the payment
due to scribes – much as modern printers are paid per thousand ens of set. However, he also reports
instances of manuscripts with every hundredth line or verse marked for reference purposes (or some
other interval).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 203
Chapter 8 • Access structure

as in poetry, computer programs, and some reproductions of text as data 257

for linguistic or bibliographic analysis; the numbering of paragraphs; and
the numbering of headed sections. Line numbers are clearly a specialized
form, but paragraph or section numbers have a general utility which has
been more obvious at some times than at others. Parkes (1976) has shown
that numbered sections became an essential part of the apparatus of late
medieval scholarship, and they are in common use today in certain types of
text (notably, technical reports and textbooks).

A question at least as obvious as ‘why did page numbering take so long to

catch on?’ is ‘why haven’t non-arbitrary numbering systems become more
widely used?’. Their utility was clear in the manuscript age, since references
could be cited even though each copy of a work would have different page
breaks. They have become a standard feature of Open University courses
because they aid the discussion of texts by groups of writers or students.
They also have advantages for printers, who need not delay the setting of
internal cross-references and indexes until the pages are established.

A confident and accurate answer would require a historical survey of

some kind, but an intuitive response is to focus on the rhetorical effect of
numbering systems.176 For some, numbered sections may be symptomatic
of what Nash (1980) termed ‘programmed’ text (Chapter 4) in which the
numbers are something of a cohesive cop-out. The numerical order gives an
element of apparent continuity which enables writers to avoid making the
connections between paragraphs explicit. This is certainly observable in the
drafting of regulations and technical documents, where no argumental flow
between paragraphs is normally considered necessary. Instead, subsumed
under a common heading, such paragraphs relate to each other as items do
in a list – that is, only by virtue of their common membership. Some writers 258
are sensitive to such associations, and may additionally feel that to show a
new paragraph with a line space and a number, rather than a new line and
indention, is to make more of a break than they would prefer; numbering
may also have an inhibiting effect on the occasional instinct to write a very
short paragraph.

The connotations of paragraph numbering may be of either excessive or

inadequate linguistic cohesion. Whereas simple series of numbers (1…n)
may look as if paragraphs are just a series of unconnected pensées, a
structured series (1.1, 1.2…1.n…n.1, n.2 etc) may look excessively
organized. The distinction is aptly illustrated by comparing the numbering
of Wittgenstein’s Tractatus logico-philosophicus (1922/1971) with that of
his later Philosophical investigations (1958). The highly structured six-level
numbering system of the Tractatus reflects its positivist philosophy and

176 I have published a fuller version of this argument, with examples, elsewhere (Waller 1977).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 204
Chapter 8 • Access structure

its apparent goal of completeness and self-sufficiency.177 The later work,

though, reflects a quite different attitude to language and logic and is
presented as a sequence of sometimes unconnected remarks, numbered
in a simple series. Access structures, although strictly functional, may
nevertheless carry connotations of the genres with which they are most
closely associated.

Page layout as access structure

Even in the days when numbering systems were rare, of course, ideas
always had a constant location within the copy each individual reader
happened to have access to; and individuals would sometimes supply their
own referencing systems. This stability of graphic layout, combined with the
fact that books, being scarcer than today, were probably more intensively
studied, might well have obviated the need for the elaborate access systems
required by today’s readers.

The Roman rhetorician Quintilian appears to have regarded the layout 259
of pages (or wax tablets, rather) as a ‘more expeditious and efficacious’
variation of the elaborate place-memory systems recommended by most
rhetoric teachers of his era. He advises the student
‘…to learn by heart from the same tablets on which he has written.; for he will
pursue the remembrance of what he has composed by certain traces, and will look,
as it were, with the eye of his mind, not only on the pages, but on almost every
individual line, resembling, while he speaks, a person reading.’ (Quintilian, Book
XI, Chapter II, 32)178

Saenger (1982: 396) comments that ‘the new readily available university
texts of the later Middle Ages, replete with chapters, subdivisions, and
distinct words, made possible a form of memorization based on the
retention of the visual image of the written page’.

Many people (and I am one) can supply anecdotal evidence that they are
sometimes able to locate ideas in books, even if not memory, simply from
their location within the book – they remember whether the page is near
the beginning or the end of the book, and whether the idea is at the top
or the bottom of the page. The educational psychologist Ernst Rothkopf
(1971) tested this hypothesis in an experiment and reported evidence that
seems to confirm such intuitions.

This informal use of the appearance of a page for information retrieval is

threatened by recent developments in electronic publishing. ‘Dynamic text’

177 Hewson (1983) has analysed the numbering system of the Tractatus from a typographic viewpoint,
and made some critical observations about its effectiveness.

178 Yates (1966: 41) comments: ‘I understand this to mean that this method adopts from the
mnemonic system the habit of visualizing on “places”, but instead of attempting to visualize shorthand
notae on some vast place system it visualizes ordinary writing as actually placed on the tablet or page.’

Robert Waller • The typographic contribution to language • PhD Thesis 1987 205
Chapter 8 • Access structure

or ‘hypertext’ (Weyer 1982; Conklin 1986) offers the reader an interactive

reading environment. Text is presented on a computer screen in a nested
form – the reader points (with a ‘mouse’) to a heading and the relevant
section of text ‘unwraps’ on to the screen. He or she may also point to
a word and obtain a definition or a cross-reference (diagrams may be
similarly unwrapped).
260

Figure 8.1a. A screen ‘page’ from a text prepared on the Guide dynamic text system (Brown 1986).

Figure 8.1b. The ‘same’ page after a reader has unwrapped a heading.

In the implementation of this concept shown in Figure 8.1 it is apparent

that after the reader has unwrapped a heading, the relative positions of
other headings and components has altered. Although a certain amount
of ‘undoing’ is allowed by the system, it may be impossible to retrieve a
‘page’ on a subsequent occasion in exactly the same form. It may also be
impossible for the writer to predict the precise juxtapositions that might

Robert Waller • The typographic contribution to language • PhD Thesis 1987 206
Chapter 8 • Access structure

arise when a text is actually used: many of the usual cohesive techniques
(for example, the use of forward and backward reference) are placed under
considerable strain by dynamic text. 179 The problem is compounded by 261
the fact that, in some applications (including the one illustrated in Figure
8.1), readers can annotate or change the author’s original text. Unless some
way of attributing such changes to individuals is built into electronic text
systems, this could suggest a bibliographer’s nightmare.

Electronically delivered text focuses us on features of books and reading

that we mostly take for granted, especially their physical nature. Garland
(1982: 5) comments:
‘Whenever I rhapsodize about the opportunities presented by the electronic media,
at the back of my mind I find myself thinking, “Yes, but a book is a book is a book.
A reassuring, feel-the-weight, take-your-own-time kind of thing…”.’

And, as Kerr (1986) has pointed out, electronic text does not allow you
to stick a finger between two pages while examining a third. The active
reading strategies encouraged by educators (Chapter 4) assume that the
text remains stable. Readers need to be able to build a mental map of the
text as a physical object, in which headings, illustrations and other graphic
features act as landmarks. It must also be asked whether the amount
of information to view at any one time has an effect on our ability to
understand complex arguments. In the 25 line display typical of current
computers, there is a higher probability that the beginning or end of the
sentence you are reading will be out of sight.180

Benest & Morgan (1985) recently developed a prototype electronic text

system that emulates traditional books. Readers are presented with
realistically-sized ‘double-page spreads’ with shadows imitative of the bulk
of a real book – the shadow is larger on the right-hand side at the beginning
of the book, and larger on the left-hand side at the end. By touching a
mid-point of the shadow, the book opens at that point, and single pages can 262
by touching a dog-ear at the corner of the page. There is a potential in such
a system for book-marks, note-taking, and all the traditional features of
printed text, without losing sight of the extra electronic potential for on-line
dictionaries, cross-referencing and easy up-dating.181

Benest & Morgan’s system may be seen as an instance of the normal

progress of new communication techniques, which often require a
transitional period in which they imitate the old, and in which new

179 Writers conventionally treat texts as if they are static physical objects – to keep track of their linear
arguments even the most ‘codified’ of prose has indexical features – words or phrases which point to
some other part of the (static) text. References which point backwards, forwards or to the immediate
textual environment are known, respectively, as anaphoric, cataphoric and deictic.

180 I have published a slightly fuller version of this argument elsewhere (Waller 1986).

181 Burrill (1986) has also proposed a system that imitates a number of book-like features.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 207
Chapter 8 • Access structure

expressive and interpretative techniques can gradually develop. For

example early printed books imitated manuscripts,182 and early film-makers
used fixed cameras in imitation of the fixed viewpoint of the theatre
audience. In a welcome contrast to some of the ‘literacy revolution’ theories
already encountered, Hirsch (1967) suggests that
‘the transition from script to print was rarely dramatic…[it] was continuous
and broken, and I venture to say that all great discoveries, all so-called new
movements, harbor the same contrasting elements, continuity and radical change.’
(p. 1–2)

Co-operative and uncooperative media

It is arguable that the introduction of greater accessibility has had the effect
of turning text from what Cherry (1966: 16) termed an uncooperative
medium into a co-operative one. A spoken conversation is the archetypal
co-operative medium, since the participants must agree on the topic,
when to interrupt or give way, and when to finish. An unsegmented
written text, on the other hand, gives the reader little option but to start
at the beginning and continue reading until the end is reached – or to
cope with the insecurity of random encounters. The greater the degree of
segmentation of written language, and the greater the degree to which 263
segments are labelled and indexed, the more co-operative the text becomes.
The accessibility afforded by typographic structuring, and typographically-
structured adjuncts such as headings, contents lists and so on, can be seen
as the basis of a conversation between reader and text.

Although conversational models of written text have been proposed

(Chapter 5), the detailed study of co-operation in discourse has, not
surprisingly, focused on spoken conversations. In fact, with a few
exceptions, ‘discourse’ is normally assumed by linguists, sociologists and
others involved in this interdisciplinary field to be spoken (for example,
Gumperz 1982, Coulthard 1985).183 One of those who uses the term in
relation to text, Hoey (1983: 27), refers to the doctrine of the primacy of
speech to justify his view of text as containing implicit dialogue:

182 Smith (in press) suggests that this conservatism might be in part due to the fact that early printer’s
copy often consisted not of an author’s draft, but a ‘published’ manuscript edition.

183 A problem with a number of accounts of ‘discourse processes’ is that, although they usually
acknowledge important differences between written and spoken texts, the distinction is not carried
through to all stages of analysis. De Beaugrande and Dressler (1981), for example, use only examples of
spoken conversations in their chapter on situationality (the handful of written examples use passages of
dialogue), but their chapter on coherence appears to assume the inspection of a written text by a reader.
Brown and Yule (1983), whose textbook on discourse analysis is in most respects a model of clarity, also
veer between spoken and written examples. In the context of ethnomethodology, McHoul (1982) has
challenged the exclusive concern of its leading figures with immediate social contact. Reading, a solitary
activity, is not regarded by some as a social act.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 208
Chapter 8 • Access structure

‘If dialogue has primacy over monologue, it is but a small step to seeing
monologue as a specialized form of dialogue between the writer or speaker and
the reader or listener’.

Clearly we should be careful about applying concepts developed for one

medium to the other.184 Telecommunications apart, spoken conversations
involve the physical presence of both participants who share a common
situation: they share the place in which the conversation occurs, the
physical presence of objects to which they may wish to refer, and the
social setting. However, since discourse analysts ascribe many aspects of
the management of conversations to prosody and paralanguage, and since
typography and punctuation is seen as the graphological equivalent of 264
paralanguage, it is worth reviewing the role of typography in the light of
some recent studies of the pragmatics of discourse.

Grice’s Co-operative Principle

The philosopher HP Grice (1975) has made an influential and widely cited
study of co-operation in discourse. Although he assumes the context of
a spoken conversation, we must clearly take note of his theory if we are
to apply a conversational model to written text; and in any case it has a
more general significance for our concept of language. Grice’s theory of
‘conversational implicature’ has become widely accepted as an explanation
of the fact that the language of conversations is frequently indirect. Take
the following exchange, for example:
A: I can’t find any whisky

B: John was here

The sentence meanings of this exchange do not adequately explain the

sense actually made of these statements by the speakers, although we have
no difficulty in constructing a scenario in which the conversation might
occur. John might have drunk the whisky, or it might have been hidden
because John, a temperance campaigner, was coming. The knowledge
shared by A and B would ensure that A knows which of the alternatives is
more likely.

Grice describes a ‘co-operative principle’ which governs our contributions to

conversations, and which we assume others will also obey:
‘Make your conversational contribution such as is required, at the stage at which it
occurs, by the accepted purpose or direction of the talk exchange in which you are
engaged’ (Grice 1975: 45).]

184 This caveat applies in both directions. We have already noted (Chapter 3) how, in spite of the
doctrine of the primacy of speech, linguists have to edit language samples to conform with grammatical
rules that are only actually adhered to strictly in text.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 209
Chapter 8 • Access structure

This is expanded into four maxims which we are said to normally obey and
expect others to obey: 265

Quantity Make your contribution as informative as required for the current

purposes of exchange. Do not make your contribution more informative
than required.

Quality Try to make your contribution one that is true, specifically: do not say
what you believe to be false; do not say that for which you lack adequate
evidence.

Relation Make your contributions relevant.

Manner Be perspicuous, and specifically: avoid obscurity; avoid ambiguity; be

brief; be orderly.

In our example, then, A can assume that B is co-operating and that his or
her answer is therefore relevant and adequate.

In practice the maxims are not always easy to distinguish. For example, a
violation of Manner, resulting in incomprehension, might be diagnosed by
a reader as a problem of Quantity (more information required) or Relation
(different information required). In fact for practical purposes, Manner is
not a particularly helpful category – it is contradictory, for one thing (to
‘be brief’ might lead to ambiguity)185 – and could be seen as simply an
injunction to obey the other three maxims.

In a conversational setting, people can directly challenge apparent

violations of these principles by requesting more, clearer or better
information. However, the co-operative principle is so strong that rather
than do so, they may make a further inference that the ‘violation’ was
intentional and therefore ironic. Or they may construct an alternative
scenario in which the violation does in fact make sense.

Quality obviously applies equally well to text as to speech, although, since

readers cannot directly challenge writers, their trust in written testimony
cannot be guaranteed – as Clanchy (1979) observes in relation to the
gradual development of trust in written records by medieval readers.
Today, publishers act as gatekeepers to the public domain, and in scientific 266
publishing there are organized systems of validation and refereeing
(Gordon 1980). Trust is partly engendered by the reputation of a journal
or a publisher, which has to be won from the community of readers and
critics.186 The concept of public relations, though, rests on the assumption
that trust can be engineered by correct attention to forms of presentation –

185 As Shuy & Larkin (1978) have pointed out in relation to the language of insurance policies, the
goals of non-ambiguity and brevity may be incompatible. Bhatia (1983) makes a similar point about
legal texts and suggests a graphic ‘easification’ method as an alternative to simplification.

186 It is interesting that Clanchy (1979: 103) notes that in medieval times ‘the commonest sign of an
amateur writer is bad layout’.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 210
Chapter 8 • Access structure

it is not entirely to the credit of typographers that such a large part of the
profession, and its educational system, is geared to this end. The confidence
engendered through presentation seems to be hard to escape from. A
notable attempt is the scientific journal Evolutionary Theory, apparently
respected by the scientific community, which appears in an extremely
amateurish, home-made form. However, the editors still feel the need to
account for the apparent lack of quality, since displayed prominently on the
cover is the motto ‘Dedicated to the primacy of content over display’.

In so far as real readers take on the role of the imagined reader, authors of
novels can, in effect, ensure that all of Grice’s maxims are met. If they do
not – if they are boring, incredible, irrelevant or cryptic – they simply lose
readers. For writers of functional texts (such as directories or manuals)
the imagined reader cannot be regarded as a fiction in quite the same
way, but must be seen as the range of possible actual users. In terms of
Grice’s maxims, they cannot always be responsible for the relevance of
information for each reader, nor for the appropriate quantity. They can, on
the other hand, be held accountable for quality, and bear a large measure
of responsibility for manner. In this context we can view typographically
signalled access systems as the means by which non-fiction writers can cope
with the requirements of relevance187 and quantity while directing their text
at a composite imagined reader.

Van Dijk’s relevance cues 267

Although he makes no overt reference to Grice, van Dijk (1979) has

published a ‘tentative list’ of the cues through which readers may determine
the relevance of a text or text component (Table 8.2) in which graphic
factors are listed along with lexical, syntactic and semantic ones.

It is noticeable that this list includes a wider range of graphical cues than 268
phonetic ones – in contrast to the conventional linguistic view, which is
that graphic cues are a poor substitute for the richness of prosody and
paralanguage. Van Dijk, of course, is concerned with the text rather than
the sentence level, and his analysis suggests that at this level the position
is reversed – that the graphic medium provides a richer repertoire of cues
than the spoken.

187 Sperber & Wilson (1986) have built a broad theory of cognition and communication around the
relevance maxim.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 211
Chapter 8 • Access structure

1. Graphical Type size, boldness

Italics, spaced, underlining, margin lines, boxes/frames, etc;
Make-up, leads, heads, etc.; indentation; text ordering.

2. Phonetic/ Stress, pitch, volume, length, pause.

phonological

3. Paratextual Gestures, facial expression.

4. Syntactical Word order

Cleft sentence structure
Topicalization
Paragraph and discourse ordering.

5. Lexical Direct relevance expressions: important, relevant, crucial, etc.

Theme indicators: the subject/theme/… is:
Summarizers: in brief/short, in other terms/words, etc.
Concluders: the conclusion, result, etc. Is:, we conclude…
Connectives: so, thus, hence
Superstructure signals: our premises are, the conclusion is, it all
happened in, suddenly…
Complex event names: accident, vacation, etc.

6. Semantic Topic-comment function of sentences

Contrastive/differential structures
Thematic words and sentences (topical expressions)
Summarizing or introducing sentences (topical)
Paraphrase
Repetition
Presupposition and semantic ordering
Descriptive level (relative completeness)

7. Pragmatic Global illocutionary force indicating devices: I (hereby) warn (ask,

congratulate) you; particles, etc

8. Schematic/ Global categorical ordering of the text

superstructural

9. Stylistic Specific variations on the other levels

10. Rhetorical Rhetorical operations: parallelisms, repetitions, contrast, etc. (on all other
levels)

Table 8.2 A tentative list of relevance signals in discourse (van Dijk 1979).

Van Dijk’s analysis can be correlated to some degree with the present genre
model, suggesting a possible harmonization of the function of graphic and
other cues. He relates the cues in his list both to what he terms ‘textual
relevance’ (the internal relations of parts of a text) and to ‘contextual
relevance’ – why a particular topic or theme should be relevant to particular
readers with particular purposes. Textual relevance is itself subdivided into
local and global kinds. Since local relevance is mostly concerned with the
sentence level, the graphic contribution would presumably be limited to the
normal repertoire of punctuation marks, italicization and so on. Broader
typographical cues would then relate mostly to global and contextual
relevance, which we may see as roughly parallel to the distinction drawn

Robert Waller • The typographic contribution to language • PhD Thesis 1987 212
Chapter 8 • Access structure

in the present study between topic and access structures. Van Dijk does
not actually assign particular functions to the cues in his list, which could
clearly be extended to include a richer view of typographic resources and
access systems.

Van Dijk hints at a possible conflict between contextual and textual

relevance. Although, on one hand, parts of the text will be deemed more or
less relevant in relation to the reader’s interests and purposes (contextual
relevance),
‘yet language and communication conventions at the same time require that he
will construct a picture of what was intended to be relevant by the speaker. This
means that the reader will have to look for the “objective” [ie, textual] relevance
cues in the text.’ (p. 123, author’s emphasis)

As a case in point, I find myself reading van Dijk’s paper in precisely this
manner. I know from his other writings that he is not very interested in 269
graphic matters, and that, although they are the main reason for my interest
in his paper, they are probably only included for the sake of completeness.
So as well as trying to relate his ideas to my own model, I have to be
satisfied that I understand his intended message and have represented him
fairly.

This reflects something of the tension in the genre model that arises
from the difficulty in distinguishing between topic and access structures
in practice. In a perfect world, it might be thought, the writer’s choice
and sequence of topic would exactly match the reader’s requirements –
such worlds, although far from perfect, do in fact exist in education and
training.188 Given the variety of prior knowledge, skills and purposes among
less controlled audiences, though, we must distinguish between those
access systems which map exactly onto the author’s topic structure, and
those which, listed in some other rational but not text-dependent order,
can be freely accessed by the reader. The first kind might include headings
and the contents list; the second would include alphabetically arranged
indexes and glossaries, and standardized keywords chosen from a list that
is not specific to the text in question (ie, from a list applied uniformly across
a database). Thus my reading of van Dijk’s paper, as an outsider relative
to his discipline, would have been aided by a broader range of cues, some
of which may not be traditional within the genre of ‘scientific paper’ –
more headings, a glossary, perhaps an index, and tutorial explanations or
critiques by others aimed at a multi-disciplinary audience. This difference
in access structures reflects a distinction between two genres of scholarly

188 Examinations are often set to test knowledge of a standard text, rather than of a subject area
for which a range of texts might be available; a well known example is the part of the driving test
where examinees are questioned about the Highway code. Trainees in subjects like nursing, the police,
accountancy and law will be especially familiar with this kind of exam.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 213
Chapter 8 • Access structure

writing – the textbook, geared to students and those new to a discipline,

and the academic paper, topic-oriented and addressed to experts.

Turn-taking 270

In direct dialogue, relevance can obviously be negotiated between the

participants. Another important consequence of the presence of both
participants is that, for dialogue to take place, they need to agree to take
turns. In the written context we can see that even the most self-organized
of reading strategies (for example, of the book-shop browser) still involve
periodic compliance with the writer’s intended sequence. The sociologists’
concept of turn-taking (Sacks, Schegloff & Jefferson 1974) recognizes that
there is a limit to the degree to which conversations can be interactive
and that most social settings and cultures embody rules for ‘floor
apportionment’. These rules may be tacitly observed, as in social gatherings,
or even explicitly specified, where a chairperson allocates time to members
of committees or assemblies.

Turn-taking suggests a possible interpretation of the different levels of

chunking in verbal language: that each chunk represents a unit of the
conversation between writer and reader, the interruption of which risks
misunderstandings at a corresponding level of analysis. Thus, incomplete
apprehension of a single word risks lexical error, of a sentence risks
grammatical error, and of a paragraph risks an error of logic or argument.
In discourse analysts’ terms, graphic segments may represent ‘transition
relevance points’.

Most people would probably regard the chapter as the basic unit for
turn-taking in reading – we expect to read it at one sitting – and authors
may even give explicit instructions to certain categories of readers to skip
chapters. Some textbooks, in fact, include elaborate charts that show
teachers which chapters should be studied for courses of different duration.
Charts such as those shown in Figure 8.2 are now a standard component of
college textbooks in the competitive US market.

Whereas the textbook in which this chart appears expounds its subject
within chapters in a traditional manner, others are expressly designed 271
to be conversational in style and structure, even within chapters. Open
University courses, for example, were originally conceived as ‘tutorials in
print’ (Rowntree 1982) containing ‘self-assessment questions’ for students
to monitor their progress. Textbooks such as these are based to a large
extent on the work of educational psychologists who have exhaustively
investigated the use of inserted questions (reviewed by Anderson & Biddle

Robert Waller • The typographic contribution to language • PhD Thesis 1987 214
Chapter 8 • Access structure

1975) – although question and answer sequences in the form of Socratic

dialogues and catechisms are, of course, an ancient pedagogic technique.189

Figure 8.2 From P. Haggett, Geography: a modern synthesis, 2nd edition, London: Harper & Row, 1975).

Question and answer structures present typographers with what has

become known as a ‘routeing’ problem. Some recent case studies have been
published in relation to question/answer sequences in textbooks (Waller
1984a) and the design of administrative forms (Waller 1984a; Frohlich
1986). Approaches range from radical alternatives to prose (Lewis, Horabin
& Gane 1967, Wright & Reid 1973, Bhatia 1983), interactive computerized
alternatives (Frohlich, 1986) to enhancements of conventional techniques 272
(Cutts & Maher 1986; Department of Health & Social Security 1983; Waller
1984a).

But whereas textbooks and administrative forms employ explicit

questioning techniques, some have suggested that all text can be seen in
terms of an implicit dialogue between writer and reader. Coulthard (1985)
concludes his book on spoken discourse with an invitation to consider

189 The original source of the steady stream of papers on inserted questions that appeared in the
educational psychology literature of the 1970s was Rothkopf’s theory of ‘mathemagenic’ activity
(Rothkopf 1970). The Greek roots of ‘mathema-genic’, a word coined by Rothkopf, suggest ‘giving birth
to learning’ – his central claim is that it is not so much the structure of texts or curricula that determine
effective learning as the activities and attitude of the learner. That is, the use of inserted questions was
designed to encourage readers to engage in the learning task with a questioning mind; Rothkopf was
able to show the use of questions improved the learning of all aspects of the text, not just the topics
focused on by questions. The theory is no longer very fashionable, but could be reinterpreted as an
attempt to encourage a conversational approach to learning. See also the critical review by Carver
(1972) and reply by Rothkopf (1974).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 215
Chapter 8 • Access structure

the extent to which techniques for conversational analysis might apply to

written text:
‘As you close this book you might like to speculate on the function of full stops. Are
they perhaps interaction points, places where the writer thinks the reader needs
to stop and ask questions about the previous sentence, questions whose range
I initially restrict by the structuring of my argument and which I subsequently
answer in the next or later sentences.’ (Coulthard 1985: 192)

This is exactly the approach taken, independently, by Winter (1977)190 and

Gray (1977).191 Gray suggests that whereas dialogue consists of explicit
questions and answers, monologue consists of answers that ‘contain’ (or
imply) their questions:
‘Composition is “composition” by virtue of the fact that it “puts together” in
subject-attribute assertions what in conversation is separated by the speakers – the
raising of the questions and the rendering of the answers to them.’ (Gray 1977: 4)

A simple demonstration text used frequently by Winter and his colleagues

may serve as an example: ‘I was on sentry duty. I saw the enemy
approaching. I opened fire. I beat off the attack’. Hoey (1983) imputes the
following questions to the imagined reader:

I was on sentry duty. Situation 273

What happened?

I saw the enemy approaching. Problem

What was your response?

I opened fire. Solution

What was the result of this?

I beat off the attack. Result/Evaluation

There is a potential difficulty here in identifying the particular questions

posed by the imagined reader. For example the question in response to ‘I
was on sentry duty’ might just as appropriately be ‘why?’ or ‘where?’. Gray’s
answer would be that they are simply analytic devices:
‘each question is determined as much by the succeeding assertion as by the
preceding one. The question…indicates the relationship between two assertions.’
(Gray 1977: 15)

Gray recommends the use of implied questions as part of what he terms

a ‘generative rhetoric’ – a technique for composition in which writers can

190 An accessible account of Winter’s work has been published by Hoey (1983). It is possible that Hoey,
a colleague of Coulthard at the University of Birmingham and warmly acknowledged in his preface,
inspired the remarks just quoted.

191 Widdowson (1979: Chapter 13; 1980; 1984: Section 3) also takes the same view and uses the same
technique of imputing questions to an imagined reader to explain relationships between assertions. His
argument is not pursued as far as those of Winter and Gray, who construct quite elaborate grammars,
but it is more completely integrated into the wider literature of pragmatics and cognitive psychology.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 216
Chapter 8 • Access structure

determine the direction of their argument by articulating (to themselves,

not in their composition) the questions arising from preceding assertions.

Whereas Gray does not have very much to say about overall patterns
of implied questioning, Winter and his colleagues justify their implied
questions by reference to the formulaic sequence (situation-problem-
solution-result-evaluation) indicated in the example quoted above. They
detect this pattern, with numerous variations and embedded sub-sequences,
in samples of real prose. If there is a normal sequence, as they suggest, then
readers presumably know what question to ‘ask’ by reference not only to
the substance of the initial statement but to their tacit knowledge of the
conversational pattern anticipated by the author within a particular type of
document. Winter’s data tends to be drawn from popular science writing,
hence the prominence of the problem-solution pattern in his analysis.
Other document types, presumably, may reveal a fuller variety of dialogue
patterns.

Two implications for typographers may be drawn from the conversational 274
view of text. Firstly, it suggests that textual units may not always be
linked in the systematic way that a focus on topic structures alone might
suggest. Headings, for example, might have no relationship, hierarchical
or otherwise, with each other but only with their immediately preceding
and following text. Such headings give prominence to an implied question
that requires special emphasis or that constitutes a major transitional point
in an argument, but have little meaning to the browsing reader. Editors
and typographers have to take special care to coordinate this local role
of headings with their global role as part of a hierarchy – to ensure that
headings make sense not only in their local context as transitional devices
but also when collected together in a contents list.

Secondly, our attention has been drawn once more to the significance
of genres or text types. Discourse analysts and ethnographers have
drawn attention to the fact that the context of a conversation affects the
relationship between participants and what is said.

In addition to the conversational maxims of Grice, a further influential

strand of linguistic philosophy that sheds light on such relationships is the
speech act theory of JL Austin. Austin (1962) drew an important distinction
between what he termed the constative and the performative uses of
language.192 Whereas the constative function refers to the use of language
to make statements about the world, the performative function describes
the use of language as an instrument for the completion of a task. The key
to the difference lies in their evaluation: constatives such as ‘This thesis is

192 The brief summary that follows cannot do justice to Austin’s theory, nor to the complexity of the
debate that has ensued from it (reviewed by Levinson 1983). The intention is simply to illustrate the
principle of instrumentality.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 217
Chapter 8 • Access structure

written on white paper’ can aptly be called true or false, but a promise, a
warning or a greeting cannot. To use one of Austin’s own examples, the
sentence ‘I name this ship the Queen Elizabeth’, as uttered when smashing
the bottle against the stem, would not normally be called true or false.
Instead, it would be deemed, in Austin’s terms, felicitous (if uttered on the 275
right occasion by the person officially designated to do so) or infelicitous
(for example, if uttered as a joke).

Austin distinguished between three ways in which an utterance may be

viewed. Considered as a locutionary act it is simply an act of speaking, say,
a sentence (or writing one, although spoken examples are mostly used);
considered as an illocutionary act, we must consider the act performed by
the use of the sentence by virtue of the conventional or illocutionary force
normally associated with it; for example, to say ‘I promise that…’ is to
carry out the act of promising. Austin’s third category is the perlocutionary
act, which describes the creation of an effect through an utterance; for
example, embarrassing or annoying someone. The distinction between
illocutionary and perlocutionary is somewhat technical, and is not relevant
in this context. Austin’s theory is most often plundered for the concept of
illocutionary force and this study shall be no exception.

Illocutionary force carries with it the notion of felicity conditions, which are
the rules defining the valid use of utterances like ‘I name this ship…’ or ‘I
pronounce thee man and wife’. Favourite examples of infelicitous speech
acts include ‘baptizing a penguin’ and ‘ordaining a jar of anchovy paste’.
Typography has its own equivalents to official ceremonies: bank notes,
company seals and educational diplomas are only valuable if made and
issued by authorized people, although an extensive rhetoric of value has
been created around such objects – exploited notably by the advertising and
packaging industries.193 However, once we depart from ‘official’ acts such
as the launching of ships or religious ceremonies the definition of ‘felicity
conditions’ is problematic.

Eco (1981: 11) makes an interesting link between the notion of the
imagined reader and speech acts: 276

‘the Model Reader is a textually established set of felicity conditions …to be met in
order to have a macro-speech act (such as a text is) fully actualized’.

Each text, it is suggested, implicitly signals to whom it is addressed – who

is the ‘legitimate’ reader, and who is cast in the role of observer or outsider.

193 The current leaders in this are Reader’s Digest, from whom I have received phoney stock certificates,
pay slips, and bank books. I have also received car registration documents (from Drive Publications)
and computer punched cards (from Which?, who should know better). In a slightly surreal connection
between the felicitous and infelicitous use of the rhetoric of value, a franking machine company sends
real Bulgarian bank-notes to potential customers to symbolize the money they could save through their
products.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 218
Chapter 8 • Access structure

We may complement this with a similar link between conversational

maxims and surface style is made by Gumperz (1982: 131):
‘this channelling of interpretation is effected by conversational implicatures based
on conventionalized co-occurrence expectations between content and surface style.’
(my emphasis)

Large type and childish pictures suggest that children are being addressed:
adults may choose such a book – as a gift for a child perhaps – and they
may read it aloud to a child, or read it for some critical or evaluative
purpose, but they do so as outsiders. This becomes very obvious when new
newspapers are launched: their choice of format (broadsheet or tabloid),
the size of their banner headlines, and the busyness of their pages signals
their desired readership as much as anything they say.194

Context
The role of typography in signalling the genre and illocutionary force of a
text suggests an extension to Gray’s characterization of a written assertion
as an answer containing its question. A written text, we might say, also
contains its own context. (Although, bearing in mind the problems inherent
in the container metaphor, noted earlier, it might be better to substitute
‘embodies’ or ‘implies’.)

This perception may help reconcile the conversational view of reading

with Olson’s (1977) notion of the autonomy of text (Chapter 7). According
to Olson, written text has the capability of preserving explicit meanings 277
in a reliable context-free manner. Because language is freed from its
interpersonal function, reason and logic come to the fore and readers can
extract meaning directly from the self-sufficient text. The modern belief in
the self-sufficiency, or autonomy, of text is attributed by Olson to Luther’s
concept of scripture as its own interpreter. A major issue in the Reformation
concerned the replacement of the Latin Bible, interpreted by the Church,
with direct access to vernacular translations by ordinary people. The
Protestant view of scriptural authority rests largely on the notion that the
Bible is an autonomous text that contains meanings that can be understood
adequately in cultures very different from the ones which produced it.
The access devices discussed earlier can be seen as ways of enhancing this
self-sufficiency by providing the answers to modern readers’ questions,
unanticipated by the original authors. There is perhaps some irony in the
debate within the non-denominational Bible Society195 that surrounds the

194 Numerous articles in the UK Press Gazette indicate that layout is generally agreed to be crucial to
the success of new launches (eg Today, The Independent, News on Sunday). The October 1986 issue of
Designer also contains a number of articles on newspaper design matters occasioned by the transfer to
electronic page make-up.

195 A founding principle of the British & Foreign Bible Society (now called Bible Society – for some
reason they omit the definite article) was to print the scriptures ‘without note or comment’. My remarks

Robert Waller • The typographic contribution to language • PhD Thesis 1987 219
Chapter 8 • Access structure

provision of headings and summaries – although designed to aid self-study

by ordinary readers, they inevitably reflect the priorities of their compiler.

Olson regards seventeenth-century British essayists, Locke in particular,

as responsible for the archetypal autonomous text, citing also the Royal
Society’s perception of the link between plain language and clear thinking.
Interestingly, Locke’s own view of Bible layout is brought to our attention by
McKenzie (1986). In his own commentary on the Epistles, Locke protests at
their division into chapters and verses:
‘that not only the Common People take the Verses usually for distinct Aphorisms,
but even Men of more advanc’d Knowledge in reading them, lose very much of the
strength and force of the Coherence, and the Light that depends upon it’.196

Locke’s wish was realized in 1937 with the publication by Heinemann of 278
The Bible design to be read as literature (described in Table 8.1), and almost
all modern translations197 follow suit, although marginal or superscript
verse numbers are still provided for reference purposes. Since Locke’s
own Essay concerning human understanding is divided into chapters and
numbered sections, he is presumably objecting not to all reference systems
but only to those imposed on authors by others.

Olson’s view has recently been disputed by Nystrand, Doyle & Himley
(1986), who point out that formal speeches and lectures are as explicit as
any written text, and that ‘public signs, kit instructions and notes left on
refrigerator doors’198 are examples of context-dependent writing.199 They
use Olson’s own paper to demonstrate how almost any writing is context-
bound – it is contextualized by its publication in the Harvard Educational
Review, its date of publication, its introductory literature review and
its accompaniment by an abstract, footnotes and references. We might
add to their list Olson’s affiliation to the Ontario Institute for Studies in
Education.200 These editorial and typographic features allow us, a decade
later, to place Olson’s ideas into a context which, for us, includes material
published since that time (for example, Nystrand’s criticism).

here are based on personal conversations with Bible Society staff during a discussion of their plans for
various special editions of the Good News Bible in 1977. A brief defence of the commentary in that
version can be found in Nida (1977).

196 An Essay for the Understanding of St. Paul’s Epistles. By Consulting St. Paul himself, 1707, quoted
by McKenzie (1986: 46).

197 For example, The Good News Bible, and The New International Version. Hunt (1970) discusses
similar features in the design of The New English Bible.

198 ‘Place-bound’ writing is discussed further by Harweg (1987).

199 Tannen (1982: 3) also disputes Olson’s hypothesis. She suggests that it ‘indeed taps features often
found in spoken and written discourse respectively, but these result not from the spoken or written
nature of the discourse as such, but rather from the genres that have been selected for analysis – casual
conversation, on one hand, and expository prose, on the other.’

200 It may or may not be significant that McLuhan, Havelock and Innis also worked in Toronto.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 220
Chapter 8 • Access structure

Nystrand et al claim that Olson’s article ‘functions not because it is

independent of its context of use but because it is so carefully attuned to
this context’ (p 101). However, they seem to over-egg their pudding when
they go on to claim that:
‘it is difficult to think of many actual situations where writers do not know at least 279
something substantial about their reader’s expectations even if they cannot always
know them personally.’ (p 106)

The readership of the Harvard Educational Review is self-selecting to a large

degree, but many texts cannot be aimed at specific audiences. The problem
of determining the skills and needs of readers is widely recognized by those
responsible for government information, technical literature and other
widely-circulating non-fiction texts (as a number of the papers in Duffy &
Waller, 1985, show). Moreover, the permanence of the written medium
means that the author’s assumptions about the original readers of a text
might be mostly irrelevant at a later date, or in another place. Although,
since the Epistles were one of the last parts of the Bible to be written, their
co-text (the Old Testament and the Gospels) is probably as familiar to
modern readers as those Paul was originally addressing, there is very little
in common between the original audience of, say, the minor prophets and
ourselves – we have little option but to risk Locke’s scorn and use them out
of context, if at all. One of the attractions of the typographically-distinct
access systems described earlier is that they can be added at a later date
without directly affecting the original author’s composition.

Nystrand et al criticize Olson for making an unfair comparison between

informal conversation and formal written exposition, but in fact he does
recognize that, since texts lack a shared situational context, they must
assign ‘the information carried implicitly by nonlinguistic means [ie, in a
conversation] into an enlarged set of explicit linguistic conventions’ (Olson
1977: 272).201 In other words, written exposition attempts to predict
the implications202 of what is said in order to deal with them explicitly.
If we include in that ‘enlarged set of explicit linguistic conventions’ 280
the typographic and editorial adjuncts that enable the text to answer a
wider range of readers’ questions than the author was able to anticipate,
then perhaps there is no real quarrel between the autonomists and the
conversationalists.

Widdowson (1984: 86) defines the achievement of accessibility as ‘an

alignment of different states of knowledge so that a common frame of

201 The linguist Wallace Chafe, who has written extensively on oral and written language, has also
commented that much of the paralanguage that accompanies speech is replaced by grammatical
structures in writing (Chafe 1982).

202 It is not altogether clear whether by ‘implications’, Olson means logical entailments or the sort of
conversational implications discussed by Grice. If he means the former, then his views cannot easily be
reconciled with Nystrand’s.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 221
Chapter 8 • Access structure

reference is created’. He does not develop the concept in much detail, but
it sounds similar to Nystrand’s (1982) concept of shared semantic space
(introduced in Chapter 5). In conversation this is negotiated – terms can be
defined, language simplified, theories exemplified, and objections met on
request. In text this can be achieved partly by the special adjuncts that have
been developed to help readers navigate around complex texts. But it seems
we must define access structure in broader terms also. By establishing and
signalling the context – the genre – of written communication, typography
indicates its relevance and scope and the social relations of its participants.

It is clear that there is considerable overlap between my three structures

in well-designed texts. Topic structures are not just fact structures but
argument structures in which information is focussed, backgrounded,
overlaid – staged, in fact, to use Grimes’ (1975) term – according to the
writer’s conversation with the imagined reader. The argument and the
conversation must be achieved, of course, within the confines of the stage
– the artefact. In the concluding chapter I shall endeavour to pull together
the three different strands of my argument and suggest that typographic
genres, containing implicit (and occasionally explicit) genre rules, are an
important key to an integrated and natural textual communication.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 222
Chapter 9 • Genre structure

9
281

Genre structure

The model of typographic communication proposed in Chapter 5 suggested

that actual texts reflect three underlying structural imperatives, each
corresponding to one part of the writer-text-reader relationship. The
analysis of pages from The handbook of sailing demonstrated some of its
assumptions about the needs of reader, the communicational intention
of the writer and the way that the technology of printing is used. That
analysis might reasonably be generalized to other instances of the genre
‘home reference manual’ (the term used by its publisher). Other genres
differ because of their different combination of topic, artefact and access
structures: genres as different as posters, brochures, textbooks, journals,
conference programmes or bus tickets could be analysed in a similar way.

In this concluding chapter I shall explore the concept of genre as a

contribution to typographic criticism and training, although, for reasons
that should become clear, I do not attempt a comprehensive classification
of genres. Indeed, de Beaugrande (1980: 196) introduces his discussion of
the related concept of ‘text types’ with the warning that they are best left as
‘fuzzy classifications’:
‘Unduly stringent criteria, like the rigorous borderline between sentences and
non-sentences, can either (1) open up endless disputes over the admissibility of
unusual or creative texts to a type, or (2) lead to so many detailed types that any
gains in heuristic usefulness are lost.’203

Genres in literature
Genre is an ancient literary concept that, like so much else, goes back to
Aristotle (the Poetics). In that context, Conley (1979) has sounded a similar 282
warning:
‘If the history of ancient rhetoric teaches us anything, it is that the degree to
which a discipline or method atrophies or declines is directly proportional to the
complexity of the taxonomies it generates.’ (p. 52–3)204

203 A similar comment is made by Graesser & Goodman (1985: 114) about the analysis of text
structures: ‘Many representational systems are so complex that they have alienated virtually all
researchers except for the one who invented the theory’

204 In view of current interest in the ‘generic coding’ of documents for electronic publishing, these
are warnings that should be taken seriously. While I have not managed to find evidence that the concept
of genre has been seriously addressed by those developing such systems, any classification scheme that

Robert Waller • The typographic contribution to language • PhD Thesis 1987 223
Chapter 9 • Genre structure

Although in the present context we are interested primarily in genres, or

varieties, of typography, it seems that there are substantial connections
between typographic and literary forms. The most thorough theoretical
treatment of literary genres is by the Canadian critic Northrop Frye (1957),
whose account of the origin of genres makes an interesting reference to
what I have called artefact structure – the means by which text is delivered
to its addressees:
‘The origin of the words drama, epic and lyric suggests that the central principle
of genre is simple enough. The basis for generic distinctions in literature appears
to be the radical of presentation. Words may be acted in front of a spectator; they
may be spoken in front of a listener; they may be sung or chanted…The basis of
generic criticism in any case is rhetorical, in the sense that the genre is determined
by the conditions established between the poet and his public.’ (Frye 1957: 246–7)

Such technical aspects of the artefact are still an important determinant

of the normal literary use of the term ‘genre’ at the broadest level. In
particular the writing-system rules employed by a writer – rules concerning
the status of the line, in particular – determine whether a work will be
received as a poem or a novel. Hawkes (1977: 137) comments:
‘the two distinctive genres of language in its written form, poetry and prose, emit
iconic messages about their nature through the visual means of typography over
and above (or under and beneath) the symbolic messages of their content.’

However, Hawkes moves immediately to an example that complicates this 283

simple distinction between poetry and prose. He ‘quotes’ a passage from
Ulysses205 where Joyce switches from what we might term novel-prose to
newspaper-prose, complete with a bold centred headline. (It is a moot point
to what extent one can ‘quote’ typography without actually reproducing it
in facsimile206 – his version differs typographically from my own copy of
Ulysses, which itself almost certainly differs from earlier editions.) Even in
the literary context, then, it seems that a wider range of typographically-
distinct genres – wider than just poetry and prose – must be recognized.

Brewer (1985: 189), comparing oral and written story-telling traditions,

regards the printing press as largely responsible for the extensive
divergence of genres in literate societies:
‘Many oral narratives appear to be carrying out a wide range of functions at the
same time. Thus, a single oral narrative may be doing what Western literature
would do through a novel, a dirty joke, a history text, a scientific journal article,
a religious text and a philosophical essay…Literacy, the printing press, and

claimed to identify the characteristics of all types of document would undoubtedly be received with
great interest.

205 Hawkes gives the reference as pp.107–108, but does not say which edition. The relevant passage,
which is headed ‘IN THE HEART OF THE HIBERNIAN METROPOLIS’ appears on page 118 of the 1971 Penguin
edition. The headlines continue for a number of pages.

206 In my own experience, even when one submits facsimile illustrations of typography, publishers are
sometimes tempted to treat them as quotations and reset them.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 224
Chapter 9 • Genre structure

specialization of function in Western society have allowed the development of

highly specialized genres. Along with the specialization of discourse force (e.g., to
inform, or to entertain, or to persuade) has gone specialization of discourse form.’

Whereas these authors (Frye, Hawkes and Brewer) use the concept of
genre to link the way texts are produced with their rhetorical intention,
Miller (1984) has moved the emphasis further toward the writer-reader
relationship. She suggests that although Frye links genre with situation,
his actual criticism is still based on formal characteristics of language: ‘For
[him], situation serves primarily to locate a genre; it does not contribute 284
to its character as rhetorical action’ (p. 153). Her definition of genre as
‘social action’ implies that ‘a rhetorically sound definition of genre must be
centered not on the substance or the form of discourse but on the action it
is used to accomplish’ (p. 151).

Werlich’s theory of text types

Werlich (1976), whose text grammar places the analysis of genre in a
central position, also regards typography as a significant marker of what he
terms ‘text type’. He establishes a hierarchy of what are effectively genres,
although he does not use that term. Werlich’s work does not appear to be
widely cited (in the English language literature, at least), but since, firstly,
it is an unusually thorough treatment of text types, and, secondly, he is
particularly sensitive to the role of typography, it deserves attention in the
present context.

At the highest level of Werlich’s scheme is text type, of which there are five
kinds: description, narration, exposition, argumentation and instruction. The
first three of these stem from the topic of the text – spatial, temporal and
systematic relations between concepts; the last two seem to be related more
to the need to effect a change in the addressee (to persuade or instruct).
De Beaugrande (1980) uses almost identical categories:207 descriptive,
narrative, argumentative, scientific (ie, exposition), didactic (ie, instruction);
he adds literary and poetic. However, having recognized these categories, de
Beaugrande goes on to refer to such things as advertisements, newspapers
and recipes, without making it clear which text types they correspond to.

Werlich overcomes this problem by making a further distinction between

text groups (fictional and non-fictional) and text forms, variants of which
correspond rather more closely to ordinary language categories such 285
as those mentioned by de Beaugrande. I have already noted Werlich’s
relatively faithful reproduction of the typography of his examples (Chapter
1), and he evidently regards typography as a particularly important

207 Both Werlich and de Beaugrande probably draw their categories from German text linguistics,
where, to judge from their citations, there seems to be a long tradition of text classification. In particular,
both cite work by Gülich & Raible (eg, 1972).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 225
Chapter 9 • Genre structure

determinant of the addressee’s critical stance. Depending whether the text

is fictional or non-fictional, the reader must connect textual references to
people, places or events – either to real phenomena or only to the text’s
internal frame of reference. According to Werlich:
‘Non-fictional texts (e.g. news stories, reports, comments, regulations, etc.) are
marked by signals (e.g. types of title and headline, including references to the text
form, and dates, place-names, kind of publication, typography, layout)…’ (p. 42)

He uses an almost identical formulation when he comes to discuss ‘fictional

texts (e.g. short stories, novels, sonnets, plays)’ (p. 43).

Text forms, in Werlich’s scheme, are rather broad categories that

correspond to his five text types, but which are broken down into further
sub-categories: argumentative text forms, for example, include comment
and scientific argumentation. Text forms are often realized in practice as
text form variants which are ‘composed in accordance with a conventionally
fixed compositional plan (e.g. the leading article and the review are variants
of the comment)’ (p. 46). His examples of fictional and non-fictional texts
would thus appear to be text form variants.

Werlich’s scheme appears to offer an attractive theoretical framework which

assigns a linguistic role to typography (although not a role which is defined
in any detail) and copes with typographically distinct text form variants
such as advertisements, newspapers and so on. His detailed descriptions of
examples are perceptive and draw on a wide range of linguistic concepts.
However, the rather dogmatic classification into five text types is not very
satisfactory when actual texts are examined. It is not clear, for example,
whether a particular text is expected to fall into only one category – it
is not hard to find a single prose passage containing a combination of,
say, narration, description and exposition. Werlich does at one point 286
recognize that his text types are ideal structures which might be combined
in practice, but his selection of examples gives the opposite impression. A
further problem is that many texts employ one apparent structure in order
to achieve another covert goal. Advertisements, in particular, are placed
awkwardly into the category of instruction; most do indeed try to influence
behaviour, but often indirectly. And many advertisements require no direct
action of the reader but are published for general public relations purposes;
they might better be classed as exposition or argumentation. Werlich is
therefore vulnerable when he suggests a single compositional plan for
advertisements:
‘(1) headline (with an optional subhead and an optional illustration)

(2) body copy

(3) signature line’ (p. 126)

Robert Waller • The typographic contribution to language • PhD Thesis 1987 226
Chapter 9 • Genre structure

The five text types stem from Werlich’s notion of the ‘cognitive matrix
of the communicant’s mind’ in which he identifies five corresponding
kinds of cognitive perception (of time, space, etc). The communicant (or
writer) is thus assigned the dominant role in Werlich’s communication
model; indeed, he presents a diagram of the encoding-decoding variety
discussed in Chapter 5. While the ‘cognitive matrix’ adds an apparent
theoretical depth to his scheme, it results in the practical difficulties just
noted. De Beaugrande, by way of contrast, makes very little of his text type
definitions, which he tosses in as something that ‘might prove useful for
further research’ (p. 197). He repeats the warning, quoted earlier, that they
are ‘fuzzy sets of texts among which there will be mutual overlap’.208

Genres as ordinary-language categories

By treating genres (text form variants, in Werlich’s terms) as basic
categories, we may avoid the intellectual gymnastics that result from 287
Werlich’s attempt to apply cognitive categories, which apply to individuals,
to texts, which are social phenomena. Ordinary-language genre labels are
generated in response to real needs felt by communities of text producers
and users; they thus have an empirical, perhaps an evolutionary, basis as
social realities.

This is precisely the view of the ‘ordinary language philosophers’, whose

founding figure, JL Austin, justified it thus:
‘Our common stock of words embodies all the distinctions men have found worth
making, in the lifetimes of many generations: these surely are likely to be more
numerous, more sound, since they have stood up to the long test of the survival
of the fittest, and more subtle, at least in all ordinary and reasonably practical
matters, than any you or I are likely to think up in our arm-chairs of an afternoon
– the most favoured alternative method.’ (Austin 1961: 182)

Among professional philosophers there seems to be considerable scepticism

about Austin’s faith in the survival of the fittest (Graham 1977), but his
ideas are widely cited by linguists and others interested in language and
communication.

Journalism can supply a simple example of the evolution of new descriptive

terms to fit everyday linguistic categories. While book editors are usually
content to see headings in terms of simple hierarchies (chapter heading,
sub-heading, sub-sub-heading; or A heading, B heading, C heading),
journalists have coined words that reflect the way headings are used in
newspapers: some terms are based on the location of headings (skyliner,

208 It is also interesting to note that de Beaugrande defines text type in a similar manner to the model
of genres presented here, although referring to a different set of underlying structures: ‘A text type is
a distinctive configuration of relational dominances obtaining between or among elements of: (1) the
surface text; (2) the textual world; (3) stored knowledge patterns; and (4) a situation of occurrence’ (de
Beaugrande 1980: 197).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 227
Chapter 9 • Genre structure

double-decker), others on their purpose (kicker, screamer, teaser).

Journalism textbooks such as Evans (1974) usually supply their own
variations on such terminology.

The ordinary-language status of typographic genres may also contribute

to the quest for the extension of literacy to include typographical factors.
Within their own cultures, readers can develop a tacit knowledge of genres,
even if they do not initially have the explicit technical knowledge needed
to produce accurate examples themselves. In contrast, specialist linguistic
or psychological terminology fits awkwardly into the context of typographic
training – theoretical concepts like ‘schema’ and ‘macrostructure’ are hard
to understand, and especially hard to apply to practical tasks, even by
experts.

On the other hand, their ordinary-language status means that descriptions

of genres reflect the full complexity of human interaction rather than
the symmetry of a theoretical model. It also means that new genres are
constantly being developed as topic, artefact and access structures change,
or new combinations are required. Genres are therefore easier to instantiate
than classify – easier to recognize in retrospect than to specify in advance.
The study of ordinary-language or ‘de facto’ genres, as they are termed by
Miller (1984), is essentially ethnomethodological; in her words, ‘it seeks
to explicate the knowledge that practice creates’ (p. 155). New genres
are probably recognised, and therefore named, by specialists before they
percolate through to ordinary language use.

A further note of relativism and fuzziness is added when we recognize,

firstly, that actual texts may belong to more than one genre, and,
secondly, may contain components that belong to different genres. These
problems were recognized by the sociologist Dell Hymes (1972, 1974)
whose application of the concept of genre to spoken discourse has been
influential among discourse analysts (cf Brown & Yule 1983, Coulthard
1985). He tackles the first problem by distinguishing between a genre and
its performance, suggesting the use of the term ‘speech act’ to denote the
latter. Actual speech acts need not necessarily fall neatly into a single genre
category. He deals with the second problem by recognizing different levels
of genres: elementary or minimal genres which in practice may be typically
found grouped together into complex genres. Thus a religious service might
constitute a complex genre, consisting of elementary genres such as hymns, 289
prayers, sermon and so on. Speech acts are instances of elementary genres,
and Hymes uses the term speech event to describe instances of complex
genres.

Although Basso (1974) suggests the application of Hymes’ approach to

written communication, he is vague about the details. Certain problems
might be anticipated (indeed, Coulthard, 1985, has noted problems with

Robert Waller • The typographic contribution to language • PhD Thesis 1987 228
Chapter 9 • Genre structure

Hymes’ system in relation to its intended context of spoken discourse).

Although it is reasonable enough to see speech acts and events as
hierarchical, it is hard to know where to stop. Hymes only suggests two
levels, but we have to decide whether to add further ones. A church service
might itself be a sub-component of a larger event – a feast-day, for example,
which might, further, be part of Lent or some other sub-section of the
church year.

Rather than propose a detailed hierarchy of genres in parallel to a hierarchy

of speech acts (or text acts, documents or whatever equivalent term one
might choose), it would seem more realistic to recognize that any class of
objects – not only linguistic ones – can be seen in terms of genres, kinds,
types or varieties, and that judgement about genre membership cannot
be restricted to a single dimension. That is, we need not expect to find
an exactly parallel relationship between categories of abstract entities
(genres) and categories of real objects (texts). Campbell & Jamieson (1979)
distinguish between a generic perspective and ‘a crusading search to find
genres’. They remark that
‘The generic perspective recognizes that while there may be few clearly
distinguishable genres, all rhetoric is influenced by prior rhetoric, all rhetorical
acts resemble other rhetorical acts.’ (p. 26)

Genre markers and genre rules

Genres are proposed as a basis for typographic conventions because,
as ordinary language categories, they are intuitively and holistically
understood. But although I have proposed a model in which they are 290
accounted for by three underlying sources of structure, each corresponding
to an aspect of the writer-text-reader relationship, these are abstract
categories that are not usually immediately apparent from looking at a
typographic display.

In practice, it is more likely that genres are recognized by their more

obvious and typical physical characteristics. These might be described or
grouped in a number of ways. A full list might be as comprehensive as van
Dijk’s list of relevance cues (Chapter 8), but concentrating on the most
readily apparent graphic features we might organize the typical features of
typographic genres into four simple categories:

1. Typical context of use: situations (industrial, domestic, educational,

bureaucratic etc); products (books, periodicals, objects, packs and
containers etc); in the case of historical examples, date of origination.

2. Typical format and configuration: page (or field) size and shape, binding
(where appropriate), paper or other surface material, frequency and use
of colour, grid, boundary (line, box, column, page, book, container etc);

Robert Waller • The typographic contribution to language • PhD Thesis 1987 229
 Chapter 9 • Genre structure

extrinsic information structures (Twyman 1982) might be included under

this heading

3. Typical treatment of verbal language: composition system (letter fit,

image quality, etc), typographic style (atmosphere, associations), range of
signalling (underlining, bold, italic, etc), additional features (rules, tints,
borders, etc); intrinsic information structures (Twyman 1982) would be
placed here.

4. Typical treatment of visual elements: pictorial syntax or style,209 proportion

of visual to verbal language, how visual and verbal language are integrated.

291
Instructions for domestic Holiday brochure Traffic sign
appliance

1. Typical context of use Delivered with product In travel agencies or sent by On posts or scaffolds near
post roads.

2. Typical format and Size may be restricted by Mostly A4 to fit standard racks Standard shapes; metal or
configuration container size; usually one or and envelopes. Some slimmer backlit plastic; standard
two colours; major division for timetable racks. Bright colours; Multiple signs
by language; minor division colour; cheap shiny paper; stacked vertically.
by operational task. Short Mostly saddle- stitched; short
examples may be on single ones may be concertina-
sheets or concertina- folded, folded; long ones may have
longer ones stitched. square backs

3. Typical treatment of Sans-serif type, multi-column Display type may have special Standard Dept of Transport
verbal language grid; blocked paragraphs; atmosphere; tables, boxes etc; bold sans-serif type (upper
tables for technical info.; small print at back; booking & lower case)
boxed or bold sections for form on back page.
warnings etc.

4. Typical treatment of Schematized diagram of Colour photographs; some Heavy use of arrows &
visual language product with parts identified of hotels, some symbolic of standard symbols, often
on diagram; in multi-lingual destination (Eiffel tower, etc); used unaccompanied by
examples, diagram folds May include drawings and words. Symbols &maps refer
out with parts identified by decorative or atmospheric to immediate environment.
numbers: separate keys for illustrations. Hotel illustrations
each language. are closely integrated with
relevant prose and tabular info.

Table 9.1 Typical surface features of some typographic genres

Table 9.1 illustrates how three particular genres might embody typical
combinations of such graphic features. Figures 9.1–9.3 illustrate some
examples. In attempting even this limited exercise it becomes clear that
one must discriminate between those features which a text must have to
qualify for membership of a particular genre, and those features which are
completely irrelevant to genre. But as well as such essential and accidental

209 Ashwin (1979) has addressed the problem of characterizing illustrative style.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 230
 Chapter 9 • Genre structure

features, we may identify an intermediate category of typical features. For

example, an essential feature of a match-box label is that it is small enough
to fit on a match-box. It is also essential that it is actually on a match-box
(or, properly captioned, in a collection or illustrated book). It will display
typical formulations such as ‘Safety matches’, ‘Made in Sweden’ or ‘approx.
contents 50’, but other visual and verbal elements will be mostly accidental
(as far as genre assignment is concerned).

2014 editorial note:

new examples have been used
for Figures 9.1-9.3

Figure 9.1 Typical instruction leaflet for a domestic product.

292
Warning signs
Mostly triangular

Dual Road narrows on Road

carriageway right (left if narrows on
ends symbol reversed) both sides
Distance to Distance to
‘STOP’ line ‘Give Way’
ahead line ahead

Crossroads Junction on T-junction with Staggered Traffic merging

bend ahead priority over junction from left ahead
vehicles from
the right

The priority through route is indicated by the broader line.

Double bend first Bend to right Roundabout Uneven road Plate below
to left (symbol (or left if symbol some signs
may be reversed) reversed)

Two-way Two-way traffic Opening or Low-flying aircraft Falling or

traffic crosses straight ahead swing bridge or sudden fallen rocks
one-way road ahead aircraft noise

Traffic signals Slippery road Steep hill Steep hill

downwards upwards
Traffic signals
not in use Gradients may be shown as a ratio i.e. 20% = 1:5

Tunnel ahead Trams Level crossing Level crossing Level crossing

crossing with barrier or without barrier without barrier
ahead gate ahead or gate ahead

108

Figure 9.2 Typical travel brochure Figure 9.3 Road traffic signs from the
UK Highway Code.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 231
Chapter 9 • Genre structure

Those genres with a high proportion of essential features are obviously 293
more coherent, more easily recognized and more strictly rule-bound than
those with few such features. The traffic signs considered in Table 9.1 must
obviously be unambiguous for reasons of safety and law enforcement –
not only ambiguous in their meaning but also in their genre, since drivers
need to distinguish them from a host of competing visual information.
Bank-notes are another kind of ‘text’ in which every graphic detail,
including the paper, is rule-bound, while various other kinds of documents
contain just some parts which are circumscribed by law – credit agreements,
legal documents and company registration documents, for example. These
may be contrasted with other genres which, although identifiable, are only
similar by virtue of typical features. The other two examples in Table 9.1
exhibit no purely graphic features which are essential (although the A4
format of the travel brochures is consistent, it is shared by too many other
genres to be a primary recognition feature). It is because most genres are
similarly identified by clusters of typical features – which may be constantly
changing as new fashions and innovations are absorbed – that any general
classification scheme must remain fuzzy and tentative.

It is, of course, possible – in fact, common – for otherwise fuzzy genres

to be regulated in particular contexts, if not with the force of law then
through institutional rules or authoritative recommendation – house styles,
for example. Other genres may be similarly articulated, but retrospectively
rather than prescriptively – although newspaper design has been described
in some detail, few journalists would treat such descriptions as instructions.
Other genres still – the more avant-garde magazines, perhaps – may be
similarly stereotyped but as yet unarticulated. Table 9.2 suggests a tentative
structure for these varying degrees of rule-boundedness.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 232
Chapter 9 • Genre structure

Basis of genre Examples Where the ‘rules’ are 294

membership articulated

Similar because Legally enforced Certain aspects of Legislation

highly rule-bound credit agreements,
traffic signs, product
labelling, etc

Institutionally Scientific journals APA style (and similar)

enforced
Technical manuals Military contracts 210

Recommended by Book publishing Butcher (1975),

authorities Williamson (1966)

Ritualized but Letter-writing Etiquette manuals

loosely articulated (Walker 1983) 211

Newspapers/ Evans (1973; 1974),

magazines Sellers (1968)

Stereotyped Advertising, Criticism (Thomson &

through style and fashion Davenport 1980),
intertextual magazines
Semiological analysis
reference and
(Barthes 1977)
imitation

Similar only Amateur Similar examples

Similar but through advertisements on analysed by Walker
not explicitly coincidence of office notice-board (1983)
rule-bound constraints

Table 9.2 Some typographic genres on a ‘scale of rule-boundedness’. Note that instances of the genres at all
points on the scale exhibit similarity to one another – there is no implication that genres at the top of the scale
are closer-knit than those lower down. The table simply aims to describe the extent to which similarities among
members of a genre can be attributed to explicit rules.210211

Table 9.2 is meant to focus attention on the relative rigidity of various

kinds of genre conventions, rather than the classification of particular
genres. On a closer look it becomes clear that different aspects of the
same genre may be rule-bound to different degrees and in different
ways. Certain advertisements have to conform to legal requirements (in
particular, cigarette and political advertising), they voluntarily conform to 295
recommended standards (although it is not clear whether ‘legal, decent,
honest and truthful’ applies to graphic aspects of advertisements), they

210 It is usual for military contracts to include a specification for the technical manual that is to
accompany the equipment to be supplied. The specification may include details of format, illustration
and typography. Kern (1985) discusses technical manuals for the US armed forces and provides
numerous references to sources of procedures and standards.

211 Letter-writing formats are institutionally enforced by school examinations. My wife, who is a
teacher of English, tells me that students must remember to use different conventions for each exam –
many students take both ‘O’ level and CSE English.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 233
Chapter 9 • Genre structure

are likely conform to stereotypes,212 and they may share the same practical
constraints.

Genre rules and error detection

In cases of less well articulated genres, the identification of underlying
‘rules’ or slots (or, at least, the basis for similarity) can be something of a
problem. The Danish linguist Hjelmslev (1959) suggested a commutation
test. If the substitution of an element of a text significantly changes the
meaning of the whole then it is clearly essential not accidental. This notion
is taken up by Barthes (1964/67: 65) as a means of detecting semiological
codes in imagery, and Eco (1976) similarly refuses to accept that an
unsegmented work of art is ‘a magic spell that is radically impermeable to
all semiotic approach.’ He explains the commutation test thus:
‘If one changes one contextual element, all the others lose their primitive function
and are usually unable to acquire another; they remain unbalanced, as on a
chessboard where a bishop has been replaced by a third castle. If there is such
contextual solidarity, then there must be a systematic rule.’ (p. 271)

The commutation test deliberately tries to introduce ‘error’ into a text in

order to tease out the underlying relationship of its elements. Some artistic
works use what are effectively commutation tests when they challenge the
conventions of their medium – painters may extend the image beyond the
frame, composers may employ noises made by objects other than musical
instruments and writers may invent new words.

A published example of a typographic commutation test may be found

in Jones (1976), who demonstrates Stanley Morison’s appropriate use of 296
typographic ornament by printing samples of Garamond and Baskerville
type with contemporary and non-contemporary decoration (Figure 9.4).

212 Variations of the concept of stereotype – schemata (Gombrich 1960) and ‘cultural codes’ (Barthes
1967, 1977) – were discussed in Chapter 3 as ways of attributing systematicity to analogue images.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 234
Chapter 9 • Genre structure

Figure 9.4 The top row is correct. From Jones (1976: 40).

Although it is more usual to speak of error ‘analysis’ (cf Lewis 1981), the
resulting imbalance envisaged by Eco provokes a judgement that is more
synthetic than analytic. It is a holistic or aesthetic apprehension of error
that may only be expressed as puzzlement or dislike; its converse – a
problem well solved – may be represented by feelings of pleasure, triumph
and balance. For all their inability to articulate the problem (until it is 297
solved and order is restored, at least), the aesthetic judgement relied upon
by most designers, although termed ‘irrational’ by some, can have real
validity in the hands of skilled typographers (although the questions still
remain: how can such skilled performers be identified? how can this skill be
passed on to others?).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 235
Chapter 9 • Genre structure

Whereas semiologists thus appeal to error detection as a means of revealing

underlying codes, in terms of the model presented in this study, such
contextual (or genre) imbalance may be seen as a failure to reconcile the
requirements of topic, artefact and access structures.

Since topic, artefact and access structures represent participants in the

communication process (and recalling Miller’s [1984] definition of genre
as social action), it is interesting to note that the Shorter Oxford English
Dictionary’s definition of ‘solecism’ links the grammatical and social
notions of error. While the major definition is ‘using incorrect syntax’,
a further definition refers to ‘a breach or violation of good manners or
etiquette’. The converse of this idiomatic recognition that manners are a
‘grammar’ of social behaviour is that grammar is a matter of good manners
– considerateness – towards the reader. ‘Considerateness’ has in fact
been suggested as the main criterion for good textbook design by some
educational researchers (Jones et al 1984; Anderson & Armbruster 1985).
This criterion also calls to mind Otto Neurath’s vision of the transformer213
as the ‘trustee of the public’.

Genre and design method

Although the model presented in Chapter 5 proposes that conventional
or genre structures represent the holistic configuration of the three basic
structures (topic, artefact and access), the relationship is not purely 298
hierarchical. Because genres are stereotyped and conventional, they may
take on a life of their own and provide a rival source of design constraints –
a fourth structure to be considered in parallel to the others.
Genre

Task Design
solution

Functional
constraints
(Topic, artefact &
access structures)

Figure 9.5 An ideal relationship between genre conventions and functional constraints, in which they are
considered in parallel before fixing on a design solution.

213 ‘Transformer’ describes a role developed by the Isotype Institute as an intermediary between
subject-matter specialists (who may lack graphic skills) and graphic artists (who may lack statistical
expertise). The transformer is a communication expert acting on behalf of the audience. Twyman
(1975) describes the original concept, and a modern application of the idea to instructional design was
proposed by Macdonald-Ross & Waller (1976).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 236
Chapter 9 • Genre structure

The practical working of this relationship may be better expressed in

terms of a simple diagram of the design process. In Figure 9.5, the design
process is represented as the linking of a design task and a design solution
through the four structures of the present model. Topic, artefact and access
structures are here conflated under the term ‘functional constraints’, for
economy of expression.

Although an ideal situation might be represented by a careful coordination

of genre and functional constraints, by identifying four distinct routes from
problem to solution we may tease out some aspects of the role of genres in
design method. These are shown in Figures 9.6–9.9.

Genre 299

Task Design
solution

Functional
constraints

Figure 9.6

Figure 9.6 represents a design method totally reliant on genre, in which a

conventional solution is proposed instinctively, and possibly correctly, but
never tested against an analysis of the relevant functional constraints. In
effect, the functional constraints analysis can only be achieved in such cases
through feedback from users – something that may be inconvenient and
expensive to obtain. In practice, it is unlikely to be a priority in situations
where so little attention has been paid to the design process in the first
instance. In any case, unless related to a functional analysis, it may not
be possible to use feedback to make detailed modifications, although
cumulative and overwhelming negative feedback may lead to a radical
redesign – which, if still uninformed by a functional approach, may simply
result in a new set of problems.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 237
Chapter 9 • Genre structure

Genre

Task Design
solution

Functional
constraints

Figure 9.7

The converse of design solely by genre is diagrammed in Figure 9.7. Here 300
an apparently sound and carefully researched technical solution is found,
but, since it lacks the characteristics of a recognized convention, readers
may not know what sort of style, strategy or critical stance is appropriate.
They may take some time to deduce the ‘rules of the game’. In certain
unusual or innovatory situations, this may actually be the only option – but
readers are likely to need special help. Genre-free technical solutions may in
time lead to the development of new conventional structures – for example,
the extremely ‘unfriendly’ user interface to the CP/M operating system for
personal computers became well enough known by users for the writers of
the rival MSDOS system to imitate it.

Genre

Task Design
solution

Functional
constraints
Figure 9.8

The next two alternative routes represent rather healthier design methods
in which functional and genre considerations are considered in parallel.
Figure 9.8 represents the reaching of a solution intuitively by identifying
the appropriate genre; the solution is then validated against the functional
constraints. This cycle of design and criticism was discussed in relation to
typographic research by Macdonald-Ross & Waller (1976).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 238
Chapter 9 • Genre structure

Genre

Task Design
solution

Functional
constraints
Figure 9.9

In Figure 9.9 the critical cycle goes the other way: a solution achieved
through an analysis of functional constraints is then checked against
genre-related expectations that might be anticipated among readers. In
practice, the critical cycle may often be informal and evolutionary – small
modifications being introduced in subsequent editions of a particular text,
and minor functional improvements becoming incorporated into other
instance of the genre.

Conclusion
Reading the products of typography’s neighbouring disciplines, it is possible
to gain an impression of typography as a distant cousin whose existence
is recognized on certain special occasions, but who is not really part of
the interdisciplinary family of discourse studies whose core members are
linguistics, sociology, psychology and literary studies. Yet there is a sense,
too, in which typography may be something of a missing link whose
recognition may be long overdue. Although there seems to be a consensus
that situational context is a vital consideration in any theoretical treatment
of discourse, with very few exceptions (Bernhardt, 1985, is a notable one)
the typographic contribution to that context is ignored, poorly understood
or simply pointed to but not investigated further.

A recurring theme in studies of communication and discourse is the contrast 302

between spoken and written language, conversation and prose, or oral
and literate culture, characterized by Olson’s (1980) distinction between
archival (topic-oriented, autonomous) and communicational (audience-
oriented, context-bound) forms of language. But although extended prose
argument is claimed to replace the mnemonic archival techniques of oral
cultures, the requirements of quick access via headings, classification
schemes, and regular spatial arrangements suggest close parallels between
the requirements of memory in oral cultures and typographic access in
literate ones that suffer from information overload. Typographic systems,
like oral ones, emphasize rhythms, parallelisms, schematizations and
similarly unsubtle but visible (and therefore usable) structures. For

Robert Waller • The typographic contribution to language • PhD Thesis 1987 239
Chapter 9 • Genre structure

example, pages (such as the one shown in Figure 7.6) consisting of evenly
shaped paragraphs reflect the need for all lines to contain the correct
number of syllables in verse forms. The practical need served is not memory
but clear topicalization (Chapter 6) and accessibility (Chapter 8).

In recent years there has been something of a revival in the fortunes of the
word ‘rhetoric’ which for many years had, and for some still has, pejorative
overtones of flowery or demagogic language.214 Rhetoric has many affinities
with typographic design – it can be superficial, merely decorative and
insincere, or it can represent the marshalling of practical techniques of
clear communication. Indeed, a number of teachers of graphic design
have applied rhetorical ideas to their subject (Bonsiepe 1965; Ehses 1984.
Kinross 1986 also discusses the application of rhetoric to graphic design,
but from a less committed standpoint).

Ehses concentrates largely on the use of rhetorical figures in the design of

posters. However, his selection of that part of rhetoric that has to do with
persuasion, and its application to posters – a persuasive medium – may just 303
serve to reinforce old prejudices about the superficiality of graphic design
and does not suggest how the rehabilitated rhetoric might be applied to
typography. Ehses’ rhetoric, as he explains, is an application of just one
part of what was originally a five-part system in classical rhetoric – elocutio,
or style. Leaving aside the two minor stages, memoria (memory) and
pronunciato (delivery), the three main stages were inventio (invention),
dispositio (arrangement) and elocutio (style) (the translated terms are from
Dixon, 1971).

The story of twentieth-century typographic theory might be seen as

some sort of a progression through these stages. Starting with a limited
conception of typography as the simple recording and delivery to the reader
of words (memoria and pronunciato), through a growing awareness of the
stylistic and associative properties of type (elocutio, exemplified by Bruce
Rogers’ theatrical metaphor), and the Modern Typographers’ use of type to
display the structure of text (dispositio), we reach the opportunity afforded
by electronic publishing to incorporate typography into the process of
writing (inventio).

Whatever basis is used to incorporate typography into discourse studies, it

is important that it does not become as technical as linguistics, semiology,
and even rhetoric, with their multifarious abstract categories. A highly
specialized theoretical system may satisfy scholarly criteria but it is unlikely
to succeed in the practical context, where, as was noted in Chapter 2,
simple guidelines and slogan language are generally better welcomed than

214 Dixon (1971) gives a good account of the history of rhetoric, and attitudes to it; accounts of recent
thinking on rhetoric may be found in EE White (1980) and, with particular reference to genre criticism,
in Campbell & Jamieson (1979).

Robert Waller • The typographic contribution to language • PhD Thesis 1987 240
Chapter 9 • Genre structure

theoretical constructs which few will make the effort to understand. It is

hoped that, by treating ordinary language genres as real categories, some
of these problems can be met halfway. Tacit familiarity with genres can be
taught in the traditional manner of design education – through imitation,
pastiche and criticism – but a more controlled and explicit understanding
may also be reached by analysing genres into their three underlying
structures – or some alternative system that others might propose.

Eco (1976) cites an illuminating analogy used by Lotman (1969): 304

‘Adults are usually introduced to an unknown language by means of rules;

they receive a set of units along with their combinatorial laws and they learn
to combine these units in order to speak; a child, on the other hand, is trained
through constant exposure to a continual textual performance of pre-fabricated
strings of that language, and he is expected to absorb his competence even though
not completely conscious of the underlying rules.’ (Eco 1976: 138)

The trend in modern foreign-language teaching is to merge these

two approaches. Even for adults, conversational practice is displacing
grammatical rules. Textual genres might be taught similarly – by
exposure to a continuous textual performance. This calls to mind Körner’s
(1970) description of common-sense classifications – for that is what
everyday terms such as ‘brochure’, ‘manual’, or ‘magazine’ are – as
resting on ‘similarities of objects to standard examples and to standard
counterexamples’ (see Chapter 2).

Typographers and graphic designers do indeed learn their trade by such a

method – by apprenticeship rather than formal teaching. But, applying the
analogy of language teaching, many could do with some basic rules and
strategies to accompany their conversational practice. Recalling Partridge’s
analogy of logic as king and grammar as Parliament, it might be suggested
that genre rules relate to rhetorical and linguistic rules in the way that
Anglo-Saxon common law relates to the Napoleonic code: they rely largely
on precedent, rather than prescribing the range of legal possibilities.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 241
References

*indicates works that were not in the original edition, cited in the 2014 preface.
Agnew GR (1986) ‘Writing backwards: the use of visual models in writing’, Visible
Language, 20, 233–243
Aitchison J (1978) Linguistics, 2nd edition, London: Teach Yourself Books/Hodder &
Stoughton
Alberti LB (1435/1972) On painting and On sculpture: the Latin texts of De Pictura and
De Statua, translated and edited by C Grayson, London: Phaedon
*Alexander C, Ishikawa S & Silverstein M (1977) A pattern language: towns, buildings,
construction, New York: Oxford University Press.
*Allen P, Bateman J & Delin J (1999) ‘Genre and layout design in multimodal
documents: towards an empirical account’, in R Power & D Scott (eds) Using layout
for the generation, analysis, or retrieval of documents, Proceedings of the American
Association for Artificial Intelligence Fall Symposium 1999. AAAI Press
Allington RL (1984) ‘Oral reading’, in R Barr, ML Kamil & P Mosenthal (eds), Handbook
of reading research, New York: Longman, 829–864
Anderson DC (1981) Evaluating curriculum proposals, London: Croom Helm
Anderson RC & Biddle WB (1975) ‘On asking people questions about what they are
reading’, in G Bower (ed), Psychology of learning and motivation, 9, New York:
Academic Press
Andrews RB (1949) ‘Reading power unlimited’, Texas Outlook, 33, 20–21
Anderson TH & Armbruster BB (1985) ‘Studying strategies and their implications for
textbook design’, in TM Duffy & RHW Waller (eds), Designing usable texts, Orlando,
Florida: Academic Press, 159–177
Arnheim R (1969) Visual thinking, Berkeley: University of California Press
Ashwin C (1979) ‘The ingredients of style in contemporary illustration’, Information
Design Journal, 1, 51–67
Ausubel DP (1963) The psychology of meaningful verbal learning, New York: Grune &
Stratton
Austin JL (1961) Philosophical papers, Oxford University Press
Austin JL (1962) How to do things with words, Oxford: Clarendon Press
Baddeley AD (1979) ‘Working memory and reading’, in P Kolers, M Wrolstad, H Bouma
(eds) Processing of visible language, vol 1, New York: Plenum, 355–370
Baddeley AD (1984) ‘Reading and working memory’ Visible Language, 4, 311–322
Baddeley AD, Eldridge M & Lewis VJ (1981) ‘The role of subvocalization in reading’
Quarterly Journal of Experimental Psychology, 33, 439–454
Baddeley AD & Lieberman K (1980) ‘Spatial working memory’, in R Nickerson (ed),
Attention and Performance VIII, Hillsdale, NJ: Lawrence Erlbaum Associates, 521–539
Baldwin RS & Coady JM (1978) ‘Psycholinguistic approaches to a theory of punctuation’.
Journal of Reading Behavior, 10, 363–376
Barker N (1981) ‘Typography and the meaning of words: the revolution in the layout of
books in the eighteenth century’, in G Barber & B Fabian (eds), The book and book
trade in eighteenth-century Europe, Hamburg: Hauswedell
Barnes BR & Clawson EU (1975) ‘Do advance organisers facilitate learning?’, Review of
Educational Research, 45, 637–660
Baron NS (1981) Speech, writing and sign, Bloomington: Indiana University Press

Robert Waller • The typographic contribution to language • PhD Thesis 1987 242
References

Barr R, Kamil ML & Mosenthal P (eds) (1984) Handbook of reading research, New York:
Longman
Barthes R (1964/1967) Elements of semiology, translated by A Lavers & C Smith,
London: Jonathan Cape. First published as Éléments de Sémiologie, Paris: Éditions du
Seuil
Barthes R (1970/1975) S/Z, translated by R Miller, London: Jonathan Cape. First
published in Paris: Éditions du Seuil
Barthes R (1977) ‘The rhetoric of the image’, in Image, music, text, essays translated by S
Heath, London: Fontana
Bartram D (1982) ‘The perception of semantic quality in type: differences between
designers and non-designers’, Information Design Journal, 3, 38–50
Basso KH (1974) ‘The ethnography of writing’, in R Bauman & J Sherzer (eds),
Explorations in the ethnography pf speaking, Cambridge University Press, 425–432
*Bateman J (2008) Multimodality and Genre: A foundation for the systematic analysis of
multimodal documents. London: Palgrave Macmillan
Beaugrande R de (1980) Text, discourse and process, Norwood, NJ: Ablex
Beaugrande R de (1981) ‘The linearity of reading: fact, fiction, frontier?’, in J Flood
(ed), Issues in reading comprehension, Newark, NJ: International Reading Association
Beaugrande R de (1984) Text production: toward a science of composition, Norwood, NJ:
Ablex
Beaugrande R de & Dressler W (1981) Introduction to text linguistics, London: Longman
Benest ID & Morgan G (1985) ‘A humanised model for the electronic library’, in Paper
versus screen: the human factors issues, Digest no 1985/80, London: Institution of
Electrical Engineers
Berliner A (1920) ‘Atmospharenwert von Drucktypen’, Ztschf. angwandte Psychol., 17,
165–172 (as cited by Ovink 1938)
Bernhardt SA (1985) ‘Text structure and graphic design’, in JD Benson & WS Greaves
(Eds), Systemic perspectives on discourse, volume 2, Norwood, NJ: Ablex
Bertin J (1967/1984) Semiology of graphics: diagrams, networks, maps, translated by WJ
Berg, Madison: University of Wisconsin Press. First published in 1967 as Sémiologie
graphique, Paris: Mouton
Bhatia VK (1983) ‘Simplification v. easification – the case of legal texts’, Applied
Linguistics, 4, 42–54
Biderman AD (1980) ‘The graph as a victim of adverse discrimination and segregation’,
Information Design Journal, 1, 232–241
Black MH (1961) ‘The evolution of a book form: The Octavo Bible from manuscript to
the Geneva Version’, The Library, 5th series, xvi, 15–28
Black MH (1963) ‘The evolution of a book form: II The Folio Bible to 1560’, The Library,
5th series, xviii, 191–203
Bloomfield L (1935) Language, revised edition, London: George Allen & Unwin
Bolinger DL (1946) ‘Visual morphemes’, Language, 22, 333–340
Bolinger DL (1975) Aspects of language, 2nd edition, New York: Harcourt Brace
Jovanovich
Bond B (1980) The handbook of sailing, London: Pelham Books
Bonsiepe G (1965) ‘Visual/verbal rhetoric’, Ulm, 15/15/16, 23–40 (cited by Ehses,
1984)
Bonsiepe G (1968) ‘A method of quantifying order in typographic design’, Journal of
Typographic Research, 2, 203–220
Booth WC (1961) The rhetoric of fiction, University of Chicago Press

Robert Waller • The typographic contribution to language • PhD Thesis 1987 243
References

Booth-Clibborn E & Baroni D (1980) The language of graphics, London: Thames &
Hudson
Bouma H (1980) ‘Visual reading processes and the quality of text displays’, in E
Grandjean & E Vigliani (eds) Ergonomic aspects of visual display terminals, London:
Taylor & Francis
Bowers F (1959) Textual and literary criticism, Cambridge University Press
Bradley H (1919) Spoken and written English, Oxford: Clarendon Press
Brewer WF (1972) ‘Is reading a letter-by-letter process?’ in JF Kavanagh & IG Mattingly
(eds), Language by ear and by eye, Cambridge, Mass: MIT Press, 359–365
Brewer WF (1985) ‘The story schema: universal and culture-specific properties’, in
DR Olson, N Torrance & A Hildyard (eds) (1985) Literacy, language and learning,
Cambridge University Press, 167–194
Britton BK & Black JB (eds) (1985) Understanding expository text, Hillsdale, NJ:
Lawrence Erlbaum Associates
Bronson BH (1968) ‘Printing as an index of taste’, in Facets of the Enlightenment,
Berkeley: University of California Press
Brown G & Yule G (1983) Discourse analysis, Cambridge University Press
Brown PJ (1986) ‘A simple mechanism for authorship of dynamic documents’, in JC van
Vliet (ed), Text processing and document manipulation, Cambridge University Press,
34–42
Bruce V & Green P (1985) Visual perception: physiology, psychology and ecology, London:
Lawrence Erlbaum Associates
Brumfit CJ (1980) ‘Being interdisciplinary – some problems facing applied linguistics’,
Applied Linguistics, 1, 158–165
Buchanan R (1985) ‘Declaration by design: rhetoric, argument and demonstration in
design practice’, Design Issues, 2, 4–22
Buckingham BR (1931) ‘New data on the typography of textbooks’ , Yearbook of the
National Society for the Study of Education, 30, 93–125
Burnhill P & Hartley J (1975) ‘The psychology of textbook design: a research critique’,
Aspects of Educational technology, VIII
Burnhill P, Hartley J & Davies L (1977) ‘Typographic decision making: the layout of
indexes’, Applied Ergonomics, 8, 35–39
Burrill VA (1986) ‘VictORia’s TEXT reading and authoring system’, in JC van Vliet (ed),
Text processing and document manipulation, Cambridge University Press, 43–57
Burt C (1959) A psychological study of typography, Cambridge University Press (reprinted
1974 by Bowker)
Burt C, Cooper WF & Martin JL (1955) ‘A psychological study of typography’, British
Journal of Statistical Psychology, 8, 29–57
Butcher J (1975) Copy-editing: the Cambridge handbook, Cambridge University Press
Buzan T (1974) Use your head, London: British Broadcasting Corporation
Calfee R (1981) ‘Cognitive psychology and educational practice’ in DC Berliner (ed)
Review of research in education, Vol 9, Washington DC: American Educational
Research Association, 3–73
Camille M (1985) ‘The Book of Signs: writing and visual difference in Gothic
manuscript illumination’, Word & Image, 1, 133–148
Campbell KK & Jamieson KH (eds) (1979) Form and genre: shaping rhetorical action,
Falls Church Virginia: The Speech Communication Association
Carter H (1969) A view of early typography, Oxford: Clarendon Press

Robert Waller • The typographic contribution to language • PhD Thesis 1987 244
References

Carver RP (1970) ‘Effect of a “chunked” typography on reading rate and

comprehension’, Journal of Applied Psychology, 54, 288–296
Carver RP (1972) ‘A critical review of mathemagenic behaviors and the effect of
questions upon the retention of prose materials’, J of Reading Behavior, 4, 95–119
Cattel J McK (1885/1947) ‘On the time required for recognising and naming letters and
words, pictures and colours’, in AT Poffenberger (ed), James McKeen Cattell: man of
science, vol 1, Psychological Research, Lancaster, Penn: Science Press, 1947, 13–23.
First published as ‘Über die Zeit der Erkennung und Benennung von Schriftzeichen,
Bildern und Farben’, Philosophische Studien, 2, 535–650
Chafe W (1982) ‘Integration and involvement in speaking, writing, and oral literature’,
in D Tannen (ed), Spoken and written language, Norwood, NJ: Ablex
Chapman J (1978) ‘re-thinking research into visual communication’,Icographic, 13,
28–30
Chauvier JH (1849) A treatise on punctuation, (Trans. J Huntington) London: Simpkin
Marshall
Chaytor HJ (1945) From script to print: an introduction to medieval literature,
Cambridge University Press
Cherry C (1966) On human communication: a review, a survey and a criticism, 2nd
edition, Cambridge, Mass.: MIT Press
Clanchy MT (1979) From memory to written record, London: Edward Arnold
Clements P (1979) ‘The effects of staging on recall from prose’, in RO Freedle (ed), New
directions in discourse processing, Norwood, NJ: Ablex
Cohen M (1977) Sensible words: linguistic practice in England, 1640–1785, Baltimore:
The John Hopkins University Press
Coleman EB & Hahn SC (1966) ‘Failure to improve readability with a vertical
typography’, Journal of Applied Psychology, 50, 434–436
Coleman EB & Kim I (1961) ‘Comparison of several styles of typography’, Journal of
Applied Psychology, 45, 262–267
Coles P & Foster J (1975) ‘Typographic cueing as an aid to learning from typewritten
text’, Programmed Learning & Educational Technology, 12, 102–108
Conklin J (1986) A survey of hypertext, Technical report no STP-356-86, Austin, Texas:
Microelectronics & Computer Technology Corporation
Conley TM (1979) ‘Ancient rhetoric and modern genre criticism’, Communication
Quarterly, 27, 47–53
Cook-Gumperz J & Gumperz JJ (1981) ‘From oral to written culture: the transition
to literacy’, in MF Whiteman (ed), Writing: The nature, development and teaching
of written communication. Volume 1: Variation in writing, Hillsdale, NJ: Lawrence
Erlbaum Associates, 89–109
Coulthard M (1985) An introduction to discourse analysis, 2nd edition, London: Longman
CoulthardM & Montgomery M (eds) (1981) Studies in discourse analysis, London:
Routledge & Kegan Paul
Coulthard M, Montgomery M & Brazil D (1981) ‘Developing a description of spoken
discourse’, in M Coulthard & M Montgomery (eds), Studies in discourse analysis,
London: Routledge & Kegan Paul
Cross N (1980) ‘An introduction to design methods’, Information Design Journal, 1,
242–253
Crystal D (1971) Linguistics, Harmondsworth: Penguin
Crystal D (1975 ) The English tone of voice, London: Edward Arnold

Robert Waller • The typographic contribution to language • PhD Thesis 1987 245
References

Crystal D (1974) ‘Paralanguage’, in TA Sebeok (ed) Current Trends in Linguistics, volume

12: linguistics and adjacent arts and sciences, The Hague: Mouton
Crystal D (1979) ‘Reading, grammar and the line’, in D Thackray (ed), Growth in
reading, London: Ward Lock
Crystal D (1980) A first dictionary of linguistics and phonetics, London: Deutsch
Crystal D (1985) A dictionary of linguistics and phonetics, 2nd edition, Oxford: Blackwell
Crystal D & Davy D (1969) Investigating English style, London: Longman
Cutts M & Maher C (1986) The Plain English story, The Plain English Campaign, Vernon
House, Whaley Bridge, Stockport SK12 7HP
Cyrus VJ (1971) ‘Linguistic features of scribal spacing’, Visible Language, 5, 101–110
Davies IK (1976) Objectives in curriculum design, London: McGraw-Hill
Dearborn WF, Johnston PW & Carmichael L (1949) ‘Oral stress and meaning in printed
material’, Science, 110, 404
*Delin J, Bateman J & Allen P (2002) ‘A model of genre in document layout’,
Information Design Journal 11, 54–66
Department of Health & Social Security (1983) The good forms guide, Document Design
Unit, DHSS, Room 510, Ray House, St Andrew St, London EC4A 3AD
Derrida J (1967/1976) Of grammatology, translated by GC Spivak, Baltimore: The John
Hopkins University Press. First published in Paris by Les Editions de Minuit.
De Vinne TL (1901) The practice of typography: correct composition, 1921 edition, New
York: Oswald
Dijk TA van (1977) Text and context: explorations in the semantics and pragmatics of
discourse, London: Longman
Dijk TA van (1979) ‘Relevance assignment in discourse comprehension’, Discourse
Processes , 2, 113–126
Dillon DT (1984) ‘The classification of research questions’, Review of Educational
Research, 54, 327–361
Diringer D (1962) Writing, London: Thames & Hudson
Dixon P (1971) Rhetoric, London: Methuen
Dobson A (1917) A bookman’s budget, London: Oxford University Press
Dooling DJ & Lachman R (1971) ‘Effects of comprehension on retention of prose’,
Journal of Experimental Psychology, 88, 216– 222
Dowding G (1966) Finer points in the spacing & arrangement of type, 3rd edition,
London: Wace & Co
Dreeben R (1970) ‘The school as a workplace’, in R Travers (ed), Second handbook of
research on teaching, Chicago: Rand McNally
Duchastel P (1982) ‘Unbounded text’, Educational Technology, 22, 19–21
Duchastel (1985) ‘Marginalia’, in D Jonassen (ed), The technology of text, volume 2,
Englewood Cliffs, NJ: Educational Technology Publications, 210–236
Duffy TM & Waller RHW (eds) (1985) Designing usable texts, Orlando, Florida:
Academic Press
Eagleton T (1983) Literary theory, Oxford: Blackwell
Eco U (1976) A theory of semiotics, Bloomington: Indiana University Press
Eco U (1981) The role of the reader, London: Hutchinson
Ehses HH (1984) ‘Representing Macbeth: a case study in visual rhetoric’, Design Issues,
1, 53–63
Eisenstein EL (1979) The printing press as an agent of change, Cambridge University
Press

Robert Waller • The typographic contribution to language • PhD Thesis 1987 246
References

Eisenstein EL (1985) ‘On the printing press as an agent of change’, in DR Olson, N

Torrance & A Hildyard (eds) (1985) Literacy, language and learning, Cambridge
University Press, 19–33
Evans H (1973) Newspaper design, London: Heinemann
Evans H (1974) News headlines, London: Heinemann
Evans M (1980) The geometry of the mind, Architectural Association Quarterly, 12 (4),
32-55.
Felker DB, Pickering F, Charrow VR, Holland M, & Redish JC (1981) Guidelines for
document designers, Washington, DC: American Institutes for Research
Feyerabend P (1975) Against method, London: New Left Books
Field A (1982) ‘Getting started: pattern notes and perspectives’, in DH Jonassen (ed),
The technology of text, Englewood Cliffs, NJ: Educational Technology Publications
Flower LS & Hayes JR (1980) ‘The dynamics of composing’ in LW Gregg & ER Steinberg
(eds) Cognitive processes in writing, Hillsdale, NJ: Lawrence Erlbaum Associates,
31–50
Foster JJ (1980) ‘Legibility research 1972–1978: a summary’, Technical report, London:
Royal College of Art
Fowler HW (1926) A dictionary of modern English usage, London: Oxford University
Press
Fowler HW & Fowler FG (1906) The King’s English, London: Oxford University Press
Francillon F (1842) An essay on punctuation, London: Whittaker & Co
Frase LT (1973) ‘On learning from prose’, Educational Psychologist, 10,
Frase LT & Schwartz BJ (1979) ‘Typographical cues that facilitate comprehension’,
Journal of Educational Psychology, 71, 197–206
Fries CC (1925) ‘Shakespearian punctuation’, Studies in Shakespeare, Milton & Donne,
University of Michigan Publications, Language & Literature, volume I, New York:
Macmillan, 65–86
Frohlich DM (1986) ‘On the organisation of form-filling behaviour’, Information Design
Journal, 5, 43–59
Frohlich DM (1986) ‘On the re-organisation of form-filling behaviour in an electronic
medium’, Information Design Journal, 5, 111-128
Fry R (1939) Last Lectures, Cambridge University Press
Frye N (1957) Anatomy of criticism, Princeton University Press
Gagné RM (1965) The conditions of learning, New York: Holt, Rinehart & Winston
Garland K (1979) ‘Some general characteristics present in diagrams denoting activity,
event and relationship’, Information Design Journal, 1, 15–22
Gardner M (1958) Logic machines and diagrams, New York: McGraw-Hill
Lloyd GER (1966) Polarity and analogy: two types of argumentation in early Greek
thought, Cambridge University Press
Macdonald-Ross M (1979) ‘Scientific diagrams and the generation of plausible
hypotheses’, Instructional Science, 8, 223–234
McLuhan M (1962) The Gutenberg galaxy, London: Routledge & Kegan Paul
Garland K & Associates (1982) Ken Garland and Associates: Designers – 20 years work
and play 1962–82, 71 Albert St, London NW1: Ken Garland & Associates
Gaskell P (1978) From writer to reader, Oxford: Clarendon
Gelb IJ (1963) A study of writing, 2nd edition, Chicago: University of Chicago Press
Gerstner K (1959) Designing programmes, Zurich: Arthur Niggli

Robert Waller • The typographic contribution to language • PhD Thesis 1987 247
References

Gerstner K (1974) Compendium for literates: a system of writing, Cambridge, Mass: MIT
Press
Gibson M (1986) Review of de Hamel (1984), The Library, sixth series, 8, 166–169
Glynn SM, Britton BK, Muth D & Dogan N (1982) ‘Writing and revising persuasive
documents: cognitive demands’, Journal of Educational Psychology, 74, 557-567
Glynn SM, Britton BK & Tillman MH (1985) ‘Typographic cues in text: management
of the reader’s attention’, in D Jonassen (ed), The technology of text, volume 2,
Englewood Cliffs, NJ: Educational Technology Publications, 192–209
Glynn SM & Di Vesta FJ (1979) ‘Control of prose processing via instructional and
typographical cues’, Journal of Educational Psychology, 71, 595–603
Gombrich EH (1960) Art and illusion: a study in the psychology of pictorial representation,
London: Phaidon
Gombrich EH (1963) Meditations on a hobby horse, London: Phaidon
Gombrich EH (1979) The sense of order: a study in the psychology of decorative art,
Oxford: Phaidon
Gombrich EH (1982) The image and the eye, Oxford: Phaidon
Gombrich EH (1984) Tributes: interpreters of our cultural tradition, Oxford: Phaidon
Goodman KS (1970) ‘Reading: a psycholinguistic guessing game’, in HS Singer & RR
Ruddell (eds), Theoretical models and processes of reading, Newark, NJ: International
Reading Association, 259–272
Goodman N (1976) Languages of art: an approach to a theory of symbols, 2nd edition,
Indianopolis, Indiana: Hackett
Goody J (1977) The domestication of the savage mind, Cambridge University Press
Goody J (1986) The logic of writing and the organization of society, Cambridge University
Press
Gordon MD (1980) ‘The role of referees in scientific communication’, in J Hartley (ed),
The psychology of written communication, ≠London: Kogan Page, 263–275
Gough PB (1972) ‘One second of reading’, in JF Kavanagh & IG Mattingly (eds),
Language by ear and by eye, Cambridge, Mass: MIT Press, 331-358
Gough PB (1984) ‘Word recognition’, in R Barr, ML Kamil &
P Mosenthal (eds), Handbook of reading research, New York: Longman, 225–253
Graesser AC & Goodman SM (1985) ‘Implicit knowledge, question answering, and the
representation of expository text’, in BK Britton & JB Black (eds), Understanding
expository text, Hillsdale, NJ: Lawrence Erlbaum Associates, 109–171
Graham K (1977) JL Austin: a critique of ordinary language philosophy, Brighton: The
Harvester Press
Gray B (1977) The grammatical foundations of rhetoric, The Hague: Mouton
Green RT & Courtis MC (1966) ‘Information theory and figure perception: the metaphor
that failed’, Acta Psychologica, 25, 12–36
Greg WW (1923/1960) ‘An Elizabethan printer and his copy’, The Library, 4, 102–118;
reprinted in Collected papers, Oxford: Clarendon Press, 95–109
Greg WW (1950–1/1960) ‘The rationale of copy-text’, Studies in Bibliography, 3, 19–36;
reprinted in Collected papers, Oxford: Clarendon Press, 374–391
Grice HP (1975) ‘Logic and conversation’ in P Cole & JL Morgan (eds), Syntax and
semantics, vol 3: Speech acts, New York: Academic press
Grimes JE (1975) The thread of discourse, The Hague: Mouton
Gülich E & Raible W (1972) Textsorten: Differenzierungskriterien aus linguistischer Sicht,
Frankfurt: Athenäum

Robert Waller • The typographic contribution to language • PhD Thesis 1987 248
References

Gullick M (1986) ‘Old books for new’, Fine Print, 12, 204–208
Gumperz JJ (1982) Discourse strategies, Cambridge University Press
Haggett P (1975) Geography: a modern synthesis, New York: Harper & Row
Halliday MAK & Hasan R (1976) Cohesion in English, London: Longman
Hamel CFR de (1984) Glossed books of the Bible and the origins of the Paris booktrade,
Woodbridge, Suffolk
Harris I (1981) ‘Effective communication for guiding practitioners: theoretical
and practical perspectives’, paper presented to American Educational Research
Association Annual Meeting, Los Angeles
Harris R (1980) The language-makers, London: Duckworth
Harris R (1981) The language myth, London: Duckworth
Harris R (1986) The origin of writing, London: Duckworth
Hartley J (1980) ‘Spatial cues in text’, Visible Language, 14, 62–79
Hartley J (1981) ‘Eighty ways of improving instructional text’, IEEE Transactions on
Professional Communication, 24, 17–27
Hartley J (1982) ‘Information Mapping: a critique’, Information Design Journal, 3, 51–58
Hartley J (1985) Designing instructional text, 2nd edition, London: Kogan Page
Hartley J, Bartlett S & Branthwaite A (1980) ‘Underlining can make a difference –
sometimes’, Journal of Educational Research, 73, 218–222
Hartley J & Burnhill P (1977a) ‘Understanding instructional text: typography, layout and
design’, in MJA Howe (ed) Adult Learning, Wiley, London, 223–247
Hartley J & Burnhill P (1977b) ‘Fifty guide-lines for improving instructional text’,
Programmed Learning and Educational Technology, 14, 65–73
Hartley J & Burnhill P (eds) (1981) ‘The spatial arrangement of text’, special issue of
Visible Language, vol 15, no 1
Hartley J, Burnhill P & Fraser S (1974) ‘Typographical problems of journal design’,
Applied Ergonomics, 5, 15–20
Hartley J & Rooum D (1983) ‘Sir Cyril Burt and typography: a re-evaluation’, British
Journal of Psychology, 74, 203–212
Hartley J, Trueman M & Burnhill P (1979) ‘The role of spatial and typographic cues in
the layout of journal references’, Applied Ergonomics, 10, 165–169
Hartmann RRK & Stork FC (1972) A dictionary of language and linguistics, London:
Applied Science Publishers
Harweg R (1987) ‘Remarks on the topology and kinematics of speech and writing’,
Semiotica, 63, 253–267
Hatt F (1976) The reading process, London: Clive Bingley
Havelock EA (1976) Origins of Western literacy, Toronto: Ontario Institute for Studies in
Education
Havelock EA (1986) The muse learns to write, New Haven: Yale University Press
Hawkes T (1977) Structuralism and semiotics, London: Methuen
Hayes JR & Flower LS (1980) ‘Identifying the organization of writing processes’ in LW
Gregg & ER Steinberg (eds) Cognitive processes in writing, Hillsdale, NJ: Lawrence
Erlbaum Associates, 3–30
Hellinga WG (1960) Copy and print in The Netherlands, vols 1 and 2, Amsterdam:
North Holland Publishing Company
Henderson L (1985) ‘On the use of the term “grapheme” ’, Language & Cognitive
Processes, 1, 135–148

Robert Waller • The typographic contribution to language • PhD Thesis 1987 249
References

Hershberger WA & Terry DF (1965) ‘Typographical cuing in conventional and

programed texts’, Journal of Applied Psychology, 40, 55–60
Hewson R (1983) ‘The visual presentation of a philosophical text’, BA dissertation,
Department of Typography & Graphic Communication, University of Reading
Hewson R & Lefrere P (1986) ‘Developing a knowldege-based system for tabular work’,
in JJH Miller (ed), Protext III, Dublin: Boole
Hirsch R (1967) Printing, selling and reading 1450–1550, Wiesbaden: Otto Harrassowitz
Hjelmslev L (1959) Essais linguistiques, Travaux du Cercle Linguistique de Copenhague,
Volume XIII, Copenhagen: Nordisk Sprog- og Kulturforlag
Hochberg J & Brooks V (1970) ‘Reading as intentional behavior’, in HS Singer & RR
Ruddell (eds), Theoretical models and processes of reading, Newark, NJ: International
Reading Association, 304–314
Hockett CF (1958) A course in modern linguistics, New York: Macmillan
Hoey M (1983) On the surface of discourse, London: George Allen & Unwin
Holliday WG (1976) ‘Teaching verbal chains using flow diagrams and text’, AV
Communication Review, 18, 129–159
Horn R (1985) ‘Results with structured writing using the Information Mapping™ Writing
Service standards’, in TM Duffy & RHW Waller (eds), Designing usable texts, Orlando,
Florida: Academic Press, 179–212
Honan P (1960) ‘Eighteenth and nineteenth century punctuation theory’, English
Studies, 40, 92–102
Hudson L (1972) The cult of the fact, London: Jonathan Cape
Huey EB (1898) ‘Preliminary experiments in the physiology and psychology of reading’,
American Journal of Psychology, 9, 575–586
Huey EB (1908) The psychology and pedagogy of reading, New York: Macmillan
Hunt G (1970) About the New English Bible, Oxford University Press & Cambrige
University Press
Hurlburt A (1978) The grid, London: Barrie & Jenkins
Husband TF & Husband MFA (1905) Punctuation: its principles and practice, London:
Routledge
Hutt A (1967) Newspaper design, London: Oxford University Press
Hymes D (1972) ‘Models of the interaction of language and social life’, in JJ Gumperz
& D Hymes (eds), Directions in sociolinguistics, New York: Holt, Rinehart & Winston,
35–71
Hymes D (1974) ‘Ways of speaking’, in R Bauman & J Sherzer (eds), Explorations in the
ethnography pf speaking, Cambridge University Press, 433–451
Innis HA (1951) The bias of communication, University of Toronto Press
Ivins WM Jr (1943) How prints look, New York: Metropolitan Museum of Art
(republished by Beacon Press, Boston, 1958)
Ivins WM Jr (1953) Prints and visual communication, Cambridge, Mass: Harvard
University Press
Javal E (1878) ‘Hygiène de la lecture’, Bulletin de la Société de Médecine Publique, p. 569
Jewett DL (1981) ‘Multi-level writing in theory and practice’, Visible Language, 15,
32–40
Johnson FC & Klare GR (1961) ‘General models of communication research: a survey of
the developments of a decade’, Journal of Communication, 11, 13–26
Jonassen D (ed) (1982) The technology of text, Englewood Cliffs, NJ: Educational
Technology Publications

Robert Waller • The typographic contribution to language • PhD Thesis 1987 250
References

Jonassen D & Hawk P (1984) ‘Using graphic organizers in instruction’, Information

Design Journal, 4, 58-68
Jones BF, Friedman LB, Tinzmann M & Cox BE (1984) Content-driven comprehension
instruction and assessment: a model for Army training literature, (Technical report),
Alexandria, Virginia: Army Research Institute
Kavanagh JF & Mattingly IG (eds) (1972) Language by ear and by eye, Cambridge, Mass:
MIT Press
Kelly GA (1955) The psychology of personal constructs, 2 volumes, New York: Norton
Kern RP (1985) ‘Modeling users and their uses of technical manuals’ in TM Duffy
& RHW Waller (eds), Designing usable texts, Orlando, Florida: Academic Press,
341–375
Kern RP & Sticht T (1974) Guidebook for the development of Army training literature:
rationale and policy implications from interviews with Army writers, Alexandria,
Virginia: Human Resources Research Organization
Kerr ST (1986) ‘Learning to use electronic text: an agenda for research in typography,
graphics and inter-panel navigation’, Information Design Journal, 4, 206–211
Kieras DE (1985) ‘Thematic processes in the comprehension of technical prose’, in BK
Britton & JB Black (eds), Understanding expository text, Hillsdale, NJ: Lawrence
Erlbaum Associates, 89–107
Kinneir J (1984) ‘The practical and graphic problems of road sign design’, in R Easterby
& H Zwaga (eds) Information design, Chichester: Wiley, 341–350
Kinross R (1984) ‘Emil Ruder’s Typography and “Swiss typography”’, Information Design
Journal, 4, 147–153
Kinross R (1986) ‘Semiotics and designing’, Information Design Journal, 4, 190–198
Klare GR (1984) ‘Readability’, in R Barr, ML Kamil & P Mosenthal (eds), Handbook of
reading research, New York: Longman, 681–744
Klare GR, Mabry JE & Gustafson LM (1955) ‘The relationship of patterning
(underlining) to immediate retention and to acceptability of technical material’,
Journal of Applied Psychology, 39, 40–42
Klare GR, Nichols WH & Shuford EH (1957) ‘The relationship of typographic
arrangement to the learning of technical training material’, Journal of Applied
psychology, 41, 41–45
Kleiman GM (1975) ‘Speech recoding in reading’ Journal of Verbal Learning and Verbal
Behavior, 14, 323–339
Koffka K (1935) Principles of Gestalt psychology, New York, Harcourt Brace
Kolers PA (1985) ‘Phonology in reading’ in DR Olson, N Torrance & A Hildyard (eds)
(1985) Literacy, language and learning, Cambridge University Press, 404–411
Körner S (1970) Categorial frameworks, Oxford: Blackwell
Kostelanetz R (ed) (1970) Moholy-Nagy, London: Allen Lane/The Penguin Press
Kuhn TS (1962) The structure of scientific revolutions, University of Chicago Press (2nd
edition, 1970).
Kulhavy RW (1977) ‘Feedback in written instruction’, Review of Educational Research, 47,
211–232
Lakoff G & Johnson M (1980a) Metaphors we live by, University of Chicago Press
Lakoff G & Johnson M (1980b) ‘The metaphorical structure of the human conceptual
system’, Cognitive Science, 4, 195–208
Langer SK (1942) Philosophy in a new key, Cambridge, Mass.: Harvard University Press
Leech G (1981) Semantics: the study of meaning, 2nd edition, Harmondsworth: Pelican
Legros LA (1922) A note on the legibility of printed matter, London: HMSO

Robert Waller • The typographic contribution to language • PhD Thesis 1987 251
References

Levinson SC (1983) Pragmatics, Cambridge University Press

Lewis BN (1981) ‘An essay on error’, Instructional Science, 10, 237–257
Lewis BN, Horabin IS & Gane CP (1967) Flow charts, logical trees and algorithms for rules
and regulations, CAS Occasional Papers No 2, London: HMSO
Lloyd GER (1966) Polarity and analogy: two types of argumentation in early Greek
thought, Cambridge University Press
Macdonald-Ross M (1979) ‘Scientific diagrams and the generation of plausible
hypotheses’, Instructional Science, 8, 223–234
McLuhan M (1962) The Gutenberg galaxy, London: Routledge & Kegan Paul
Langford M (1980) The Pocket camera handbook, London: Ebury Press
Lotman JM (1969) ‘O metajazyke tipologiceskick opisanij kul’tury’, Trudy po znakovym
sistemam, IV (cited by Eco 1976)
Lotz J (1972) ‘How language is conveyed by script’, in JF Kavanagh & IG Mattingly
(eds), Language by ear and by eye, Cambridge, Mass: MIT Press, 117–124
Lowth R (1775/1842) A short introduction to English grammar, with critical notes, new
edition with corrections, London: J Dodsley & T Cadell; reproduced as an appendix
to Francillon (1842)
Lyons J (1977) Semantics, 2 volumes, Cambridge University Press
McConkie GW & Rayner K (1975) ‘The span of effective stimulus during a fixation in
reading’, Perception & Psychophysics, 17, 578–586
Macdonald-Ross M (1977) ‘Graphics in text’, in LS Shulman (ed), Review of research in
education, Itasca, Illinois: Peacock
Macdonald-Ross M (1979) ‘Scientific diagrams and the generation of plausible
hypotheses’, Instructional Science, 8, 223–234
McLuhan M (1962) The Gutenberg galaxy, London: Routledge & Kegan Paul
Macdonald-Ross M & Smith EB (1974) Bibliography of textual communication, Milton
Keynes: The Open University
Macdonald-Ross M & Waller RHW (1975) Criticisms, alternatives and tests: a framework
for improving typography, Programmed Learning & Educational Technology, 12,
75–83
McHoul A (1982) Telling how texts talk, London: Routledge & Kegan Paul
McKenzie DF (1981) ‘Typography and meaning: the case of William Congreve’, in G
Barber & B Fabian (eds), The book and book trade in eighteenth-century Europe,
Hamburg: Hauswedell, 81–123
McKenzie DF (1986) Bibliography and the sociology of texts, (The Panizzi Lectures,
1985), London: The British Library
McKerrow RB (1928) An introduction to bibliography, 2nd impression with corrections,
Oxford: Clarendon Press
McLaren I (1983) ‘Videotex – glimpses of some facets’, Information Design Journal, 3,
231–238
McLean R (1980) The Thames & Hudson manual of typography, London: Thames &
Hudson
McLuhan M (1962) The Gutenberg galaxy, London: Routledge & Kegan Paul
McMurtrie DC (1937) The book, New York: Covici Friede
Mager RF (1961) ‘On the sequencing of instructional content’, Psychological Reports, 9,
405–413
Mangan J (1978) ‘Cultural conventions of pictorial representation: iconic literacy and
education’, Educational Communication and Technology, 26, 245–267

Robert Waller • The typographic contribution to language • PhD Thesis 1987 252
References

Marr D (1982) Vision, San Fransisco: WH Freeman & Co

Martin SE (1972) ‘Nonalphabetic writing systems: some observations’, in JF Kavanagh &
IG Mattingly (eds), Language by ear and by eye, Cambridge, Mass: MIT Press, 81–102
Massaro DW (1979) ‘Reading and listening’, in P Kolers, M Wrolstad, H Bouma (eds)
Processing of visible language, vol 1, New York: Plenum, 331–354
Massin (1970) Letter and image, London: Studio Vista
Meyer BJF (1975) ‘Identification of the structure of prose and its implications for the
study of reading and memory’ Journal of Reading Behavior, 7, 7–47
Meyer BJF (1985) ‘Prose analysis: purposes, procedures, and problems’, in BK Britton
& JB Black (eds), Understanding expository text, Hillsdale, NJ: Lawrence Erlbaum
Associates, 11–64
Miller CR (1984) ‘Genre as social action’, Quarterly Journal of Speech, 70, 151–167
Monteith R (1704) The true and genuine art of pointing, Edinburgh
Moran J (1971) Stanley Morison: his typographic achievement, London: Lund Humphries
Moran J (1978) ‘Fit to be styled a typographer’: a history of the Society of Typographic
Designers, London: Westerham Press
Morison S (1928) The 2 kinds of effectiveness, Manifesto distributed at the Annual
Conference of the Master Printers’ Federation, Blackpool, June 1928 (reproduced by
Moran 1971)
Morison S (1936) ‘The changing newspaper’, The Monotype Recorder, 35, 1–2
Morison S (1951) First principles of typography, 2nd edition, Cambridge University Press
Morison S (1972) Politics and script, (The Lyell lectures, 1957, edited by N Barker),
Oxford: Clarendon Press
Morrison RE & Inhoff A-W (1981) ‘Visual factors and eye movements in reading’, Visible
Language, 15, 129–146
Mountford J (1969) ‘Writing’, Encyclopaedia of linguistics, information and control,
Oxford: Pergamon Press
Mountford J (1980) ‘Writing-system as a concept in linguistics’, Information Design
Journal, 1, 223–231
Mountford J (1982) ‘“Text”, “book”, “writing”, and the medial aspect of language’,
Information Design Journal, 3,110–118
Moxon J (1683/1962) Mechanick exercises on the whole art of printing, edited by H Davis
& H Carter, 2nd edition, London: Oxford University Press (originally published in
1683–4)
Nash W (1980) Designs in prose, London: Longman
Neisser U (1976) Cognition and reality, San Francisco: Freeman
Neurath O (1936) International picture language, London: Kegan Paul
Newell A (1973) ‘You can’t play 20 questions with nature and win’, in WG Chase (ed)
Visual information processing, New York: Academic Press
Nida EA (1977) Good news for everyone, London: Fount
Nidditch PH (1975) (ed), John Locke’s An essay concerning human understanding,
critical-text edition, Oxford: Clarendon
Norrish P (1984) ‘Moving tables from paper to screen’, Visible Language, 18, 154–170
North AJ & Jenkins LB (1951) ‘Reading speed and comprehension as a function of
typography’, Journal of Applied Psychology, 35, 225–228
Nystrand M (1982) ‘The structure of textual space’, in M Nystrand (ed), What writers
know: the language, process and structure of written discourse, New York: Academic
Press, 75–86

Robert Waller • The typographic contribution to language • PhD Thesis 1987 253
References

Nystrand M, with Himley M & Doyle A (1986) The structure of written communication,
Orlando, Florida: Academic Press
Nystrand M, Doyle A & Himley M (1986) ‘A critical examination of the doctrine of
autonomous texts’, in M Nystrand, The structure of written communication, Orlando,
Florida: Academic Press, 81–107
O’Hara FM Jr (1971) ‘The use of the hyphen in printing to indicate divided words’,
Visible Language, 5, 111–124
Olson DR (1977) ‘From utterance to text: the bias of language in speech and writing’,
Harvard Educational Review, 47, 257–281
Olson DR (1980) ‘On the language and authority of textbooks’, Journal of
Communication, 30, 186–196
Olson DR, Torrance N & Hildyard A (eds) (1985) Literacy, language and learning,
Cambridge University Press
Ong WJ (1944) ‘Historical backgrounds of Elizabethan and Jacobean punctuation
theory’, PMLA, 59, 349–360
Ong WJ (1958) Ramus, method, and the decay of dialogue, Cambridge, Mass.: Harvard
University Press
Ong WJ (1967) The presence of the word, New Haven: Yale University Press
Ong WJ (1982) Orality and literacy, London: Methuen
Ovink GW (1938) Legibility, atmosphere-value, and forms of printing types, Leyden: AW
Sijthoff
Palmer RE (1969) Hermeneutics, Evanston, Illinois: Northwestern University Press
Pantin CFA (1954) ‘The recognition of species’, Science Progress, 42, 587–598
Parkes MB (1976) ‘The influence of the concepts of ordinatio and compilatio on the
development of the book’, in JJG Alexander & MT Gibson (eds), Medieval learning
and literature: essays presented to Richard William Hunt, Oxford, Clarendon Press,
115–141 (and plates IX–XVI, at the end of the volume)
Parry M (1971) The making of Homeric verse: the collected papers of Milman Parry (edited
by A Parry), Oxford: Clarendon Press
Partridge E (1953) You have a point there, London: Hamish Hamilton
Pask G & Scott BCE (1972) ‘Learning strategies and individual competence’,
International Journal of Man-Machine Studies, 4, 217
Paterson DG & Tinker MA (1940) How to make type readable, Hew York: Harper
Pei MA & Gaynor F(1954) A dictionary of linguistics, New York: Philosophical Library
Polanyi M (1958) Personal knowledge, London: Routledge & Kegan Paul
Polanyi M (1969) Knowing and being, University of Chicago Press
Popper KR (1957) The poverty of historicism, London: Routledge & Kegan Paul
Popper KR (1959) The logic of scientific discovery, London: Hutchinson
Price HT (1939) ‘Grammar and the compositor in the sixteenth and seventeenth
centuries’, Journal of English & Germanic Philology, 38, 540–548
Pugh AK (1978) Silent reading, London: Heinemann
Pugh AK (1979) ‘Styles and strategies in adult reading’, in P Kolers, M Wrolstad, H
Bouma (eds) Processing of visible language, vol 1, New York: Plenum
Pullum GK (1984) ‘Punctuation and human freedom’, Natural Language and Linguistic
Theory, 2, 419–425
Putnam H (1978) Meaning and the moral sciences, London: Routledge & Kegan Paul
Pyke RL (1926) The legibility of print, London: HMSO

Robert Waller • The typographic contribution to language • PhD Thesis 1987 254
References

Quintilian (1856) Institutes of oratory, Translated by JS Watson, Vol I (Books I–VII),

London: Bohn
Quintilian (1870) Institutes of oratory, Translated by JS Watson, Vol II (Books VIII–XII),
London: Bell & Daldy
Quirk R, Greenbaum S, Leech G & Svartvik J (1985) A comprehensive grammar of the
English language, London: Longman
Raban B (1982) ‘Text display effects on the fluency of young readers’, Journal of
Research in Reading, 5, 7–28
Ravetz JR (1971) Scientific knowledge and its social problems, London: Oxford University
Press
Rayner K (1975) ‘The perceptual span and peripheral cues in reading’, Cognitive
Psychology, 7, 65–81
Rayner K (1978) ‘Eye movements in reading and information processing’, Psychological
Bulletin, 85, 618–666
Reddy MJ (1979) ‘The conduit metaphor – a case of frame conflict in our language
about language’, in A Ortony (ed), Metaphor and thought, Cambridge University
Press
Rehe RH (1974) Typography: how to make it legible, Bloomington: Indiana University
Press
Reynolds L (1979) ‘Teletext and viewdata – a new challenge for the designer’,
Information Design Journal, 1, 2–14
Richards IA (1926) Principles of literary criticism, 2nd edition, London: Routledge &
Kegan Paul
Rickards JP (1977) ‘On inserting questions before or after segments of text’,
Contemporary Educational Psychology, 2, 200–206
Rivlin C (1987) ‘A study of the relationship between visual perception and typographic
organisation’, PhD thesis, Manchester Polytechnic
Roberts CH & Skeat TC (1983) The birth of the codex, Oxford University Press
Robertson J (1785) An essay on punctuation, Scolar Press edition, 1969
Rogers B (1943) Paragraphs on printing, New York: William E Rudge’s Sons (facsimile
edition by Dover Publications, 1979)
Rogers G (1986) Editing for print, London: Macdonald
Rothkopf EZ (1970) ‘The concept of mathemagenic activities’, Review of Educational
Research,40, 325–336
Rothkopf EZ (1971) ‘Incidental memory for the location of information in text’, Journal
of Verbal Learning & Verbal Behavior, 10, 608–613
Rothkopf EZ (1974) ‘Barbarism and mathemagenic actitivities: comments on criticism by
Carver’, Journal of Reading Behavior, 6, 3–8
Rowe CL (1982) ‘The connotative dimensions of selected display typefaces’, Information
Design Journal, 3, 30–37
Rowntree D (1982) Educational technology in curriculum development, 2nd edition,
London: Harper & Row
Ruëgg R & Fröhlich G (1972) Basic typography, Zurich: ABC
Ruder E (1967) Typography, Zurich: Arthur Niggli
Rumelhart DE (1977) ‘Toward an interactive model of reading’, in S Dornic (ed),
Attention and performance, vol VI, Hillsdale, NJ: Lawrence Erlbaum Associates
Russell B (1946) A history of western philosophy, London: George Allen & Unwin
Ryle G (1949) The concept of mind, London: Hutchinson

Robert Waller • The typographic contribution to language • PhD Thesis 1987 255
References

Ryle G (1951) ‘Systematically misleading expressions’ in AGN Flew (ed) Essays on logic
and language, Oxford
Sacks H, Schegloff EA & Jefferson G (1974) ‘A simplest systematics for the organisation
of turn-taking in conversations’, Language, 50, 696–735
Saenger P (1982) ‘Silent reading: its impact on late medieval script and society’, Viator,
13, 367–414
Sampson G (1980) Schools of linguistics, London: Hutchinson
Sampson G (1985) Writing systems, London: Hutchinson
Samuels SJ & Kamil ML (1984) ‘Models of the reading process’, in R Barr, ML Kamil & P
Mosenthal (eds), Handbook of reading research, New York: Longman, 185–224
Sapir E (1921) Language, New York
Saussure F de (1916/1974) Course in general linguistics, edited by C Bally & A
Sechehaye with A Reidlinger, translated by W Baskin, Revised edition, London:
Fontana; originally published in 1916 as Cours de linguistique générale.
Schapiro M (1969) ‘On some problems in the semiotics of visual art: field and vehicle in
image-signs’, Semiotica, 1, 223–242
Schnotz W (1982) ‘How do different readers learn with different text organisations?’ in
A Flammer & W Kintsch (eds) Discourse processing, Amsterdam: North Holland
Schumacher GM & Waller RHW (1985) ‘Testing design alternatives: a comparison of
procedures’, in TM Duffy & RHW Waller (eds), Designing usable texts, Orlando,
Florida: Academic Press, 377–403
Scribner S & Cole M (1981) ‘Unpackaging literacy’, in MF Whiteman (ed), Writing: The
nature, development and teaching of written communication. Volume 1: Variation in
writing, Hillsdale, NJ: Lawrence Erlbaum Associates, 71–87
Sellers L (1968) The simple subs book, Oxford: Pergamon Press
Shannon CE & Weaver W (1949) The mathematical theory of communication, Urbana,
Ill.: University of Illinois Press
Shebilske WL & Rotondo JA (1981) ‘Typographical and spatial cues that facilitate
learning from textbooks’, Visible Language, 15, 41–54
Shipman S & Shipman VC (1985) ‘Cognitive styles: some conceptual, methodological,
and applied issues’, in EW Gordon (ed), Review of Research in Education, volume 12,
Washington DC: American Educational Research Association, 229–291
Shurtleff DA (1980) How to make displays legible, La Mirada, Calif.: Human Interface
Design
Shuy RW & Larkin D ‘Linguistic considerations in the simplification /clarification of
insurance policy language’, Discourse Processes,1, 305–321
Simon (1969) The sciences of the artificial, Cambridge, Mass: MIT Press
Simpson P (1911) Shakespearian punctuation, Oxford: Clarendon Press
Sless D (1980) ‘Image design and modification: an experimental project in
transforming’, Information Design Journal, 1, 17–80
Sless D (1981) Learning and visual communication, London: Croom Helm
Sless D (1986) In search of semiotics, London: Croom Helm
Smillie RJ (1985) ‘Design strategies for job performance aids’, in TM Duffy & RHW
Waller (eds), Designing usable texts, Orlando, Florida: Academic Press, 213–243
Smith BL, Lasswell HD & Casey RD (1946) Propaganda, communication and public
opinion, Princeton University Press
Smith MM (1988) ‘Printed foliation: forerunner to printed page-numbers?’ Gutenberg-
Jahrbuch, 63, 54–70.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 256
References

Snow CP (1959) The two cultures and the scientific revolution, (Rede Lecture, 1959),
Cambridge University Press.
Sopher H (1977) ‘The problem of punctuation’, English Language Teaching, 31, 304–313
Spencer H (1969a) Pioneers of modern typography,London: Lund Humphries
Spencer H (1969b) The visible word, 2nd edition, London: Lund Humphries
Spencer, Reynolds & Coe (1975) ‘Spatial and typographic coding with bibliographic
entries’, Programmed Learning & Educational Technology, 12, 95–101
Sperber D & Wilson D (1986) Relevance: communication and cognition, Oxford:
Blackwell
Steinberg SH (1974) Five hundred years of printing, 3rd edition, Harmondsworth:
Penguin
Sticht T (1985) ‘Understanding readers and their uses of text’, in TM Duffy & RHW
Waller (eds), Designing usable texts, Orlando, Florida: Academic Press, 315–340
Sturrock J (1986) Structuralism, London: Paladin/Grafton Books
Suleiman SR & Crosman I (eds) (1980) The reader and the text, Princeton University
Press
Tannen D (1982) ‘Oral and literate strategies in spoken and written narratives’,
Language, 58, 1–21
Tannen D (ed) (1982) Spoken and written language: exploring orality and literacy,
Norwood, NJ: Ablex
Taylor TJ (1980) Linguistic theory and structural stylistics, Oxford: Pergamon Press
Taylor TJ (1984) ‘Editing rules and understanding: the case against sentence-based
grammar’, Language and Communication, 4, 105–127
Thomas LF & Harri-Augstein ES (1980) ‘Learning-to-learn by reading: towards a
conversational technology’, in R Winterburn (ed) Aspects of Educational Technology,
London: Kogan Page
Thompson EM (1892) Handbook of Greek and Latin palæography, London: Kegan Paul,
Trench, Trübner
Thompson EM (1912) Introduction to Greek and Latin palæography, Oxford: Clarendon
Press
Thompson P & Davenport P (1980) The dictionary of visual language, London: Bergstrom
& Boyle
Thorndike EL (1917/1971) ‘Reading as reasoning: a study of mistakes in paragraph
reading’, Journal of Educational Psychology, 8, 323–332 (reprinted in Reading
Research Quarterly, 1971, 425–433)
Thorndike EL (1948) ‘The psychology of punctuation’, American Journal of Psychology,
61, 222–229
Tinker MA (1955) ‘Perceptual and oculomotor efficiency in reading materials in vertical
and horizontal arrangements’, American Journal of Psychology, 68, 444–449
Tinker MA (1963) Legibility of Print, Ames: Iowa State University Press
Tinker MA (1965) Bases for effective reading, University of Minneapolis Press
Tompkins JP (ed) (1980) Reader-response criticism, Baltimore: The John Hopkins
University Press
Toulmin SE (1958) The uses of argument, Cambridge University Press
Trager GL (1974) ‘Writing and writing systems’, in TA Sebeok (ed) Current Trends in
Linguistics, volume 12: linguistics and adjacent arts and sciences, The Hague: Mouton,
373–496
Treip M (1970) Milton’s punctuation, London: Methuen

Robert Waller • The typographic contribution to language • PhD Thesis 1987 257
References

Tschichold J (1935/1967) Asymmetric typography, translated by R McLean, London:

Faber & Faber (originally published as Typographische Gestaltung, Basle, 1935)
Tschichold J (1934/1975) ‘The placing of type in a given space’, translated by R McLean
and published as an appendix to his Jan Tschichold: typographer, London: Lund
Humphries; first published as ‘Die Anordnung des Schriftsatzes in Flächenraum’,
Typographische Monatsblätter, 7
Tu HTC (1930) ‘The effect of different arrangements of the Chinese language on the
speed and comprehension of silent reading’, J of Genetic Psychology, 38, 321–337
Tunstall J (1970) Media sociology: a reader, London: Constable
Turner EG (1977) The typology of the early codex, Philadelphia: University of
Pennsylvania
Twyman ML (1970) Printing 1770–1970, London: Eyre & Spottiswoode
Twyman ML (1975) ‘The significance of Isotype’, in Graphic communication through
Isotype, University of Reading
Twyman ML (1979) ‘A schema for the study of graphic language’, in P Kolers, M
Wrolstad, H Bouma (eds) Processing of visible language, vol 1, New York: Plenum,
117–150
Twyman ML (1982) ‘The graphic presentation of language’, Information Design Journal,
3, 2–22
Twyman ML (1985) ‘Using pictorial language: a discussion of the dimensions of the
problem’ in TM Duffy & RHW Waller (eds), Designing usable texts, Orlando, Florida:
Academic Press, 245–312
Twyman ML (1986) ‘Articulating graphic language: a historical perspective’, in ME
Wrolstad & DF Fisher (eds), Towards a new understanding of literacy, New York:
Praeger, 188–251
Ullman BL (1932) Ancient writing and its influence, New York: Longmans, Green (1980
facsimile edition published by University of Toronto Press)
Updike DB (1937) Printing types: their history, forms, and use, London: Oxford University
Press
Vachek J (1948/1967) ‘Written language and printed language’, in A Prague School
reader in linguistics, Bloomington: Indiana University Press, 453–460 (originally
published in Recueil Linguistique de Bratislavia, 1948, I, 67–75)
Vachek J (1959) ‘Two chapters on written English’, Brno Studies in English, 1, 7–38
Vachek J (1973) Written language: general problems and problems of English, The Hague,
Mouton
Vachek J (1979) ‘Some remarks on the stylistics of written language’. in DJ Allerton, E
Carney & D Holdcroft (eds), Function and context in linguistic analysis, Cambrifge
University Press, 206–221
Van Sommers P (1984) Drawing and cognition, Cambridge University Press
Venezky RL (1970) The structure of English orthography, The Hague: Mouton
Venezky RL (1984) ‘The history of reading research’, in R Barr, ML Kamil & P Mosenthal
(eds), Handbook of reading research, New York: Longman, 3–38
Walker P, Smith S & Livingston A (1986) ‘Predicting the appropriateness of a typeface on
the basis of its multi-modal features’, Information Design Journal, 5, 29–42
Walker SF (1983) ‘Descriptive techniques for studying verbal graphic language’, PhD
thesis, Department of Typography & Graphic Communication, University of Reading
Waller RHW (1977) ‘Numbering systems in text’, Notes on transforming 4, Milton
Keynes: The Open University Institute of Educational Technology (reprinted, with

Robert Waller • The typographic contribution to language • PhD Thesis 1987 258
References

illustrations omitted, in J Hartley (ed), The psychology of written communication,

1980, London: Kogan Page)
Waller RHW (1979a) ‘Typographic access structures for educational texts’, in P Kolers, M
Wrolstad, H Bouma (eds) Processing of visible language, vol 1, New York: Plenum
Waller RHW (1979b) ‘Four aspects of graphic communication’, Instructional Science, 8,
213–222
Waller RHW (1981) ‘Understanding network diagrams’, paper presented to American
Educational Research Association Annual Meeting, Los Angeles
Waller RHW (1982) ‘Text as diagram: using typography to improve access and
understanding’, in D Jonassen (ed), The technology of text, Englewood Cliffs, NJ:
Educational Technology Publications, 137–166
Waller RHW (1984a) ‘Designing government forms: a case study’, Information Design
Journal, 4, 36–57
Waller RHW (1984b) ‘Darwin’s finches – a case study in criticism and alternatives’ in M
Nathenson & E Henderson (eds) Independent learning in higher education, Englewood
Cliffs NJ: Educational Technology Publications
Waller RHW (1985) ‘Using typography to structure arguments: a critical analysis of
some examples’, in D Jonassen (ed), The technology of text, volume 2, Englewood
Cliffs, NJ: Educational Technology Publications, 105–125
Waller RHW (1986) ‘What electronic books will have to be better than’, Information
Design Journal, 5, 72–75
Waller RHW (1987) ‘The design of print media’, Unit 23 of EH207 Education and
communication, Milton Keynes: The Open University
*Waller RHW (1990) ‘Typography and discourse’, in R Barr et al (eds), Handbook of
Reading Research, vol II, New York: Longman, 341-380.
*Waller RHW (1999) ‘Making connections: typography layout and language’, in R Power
& D Scott (eds) Using layout for the generation, analysis, or retrieval of documents,
Proceedings of the American Association for Artificial Intelligence Fall Symposium
1999. AAAI Press
*Waller R, Delin J & Thomas M (2012), Towards a pattern language approach to
document description. Discours 10 (online journal, accessible at http://discours.
revues.org/8673
Waller RHW & Whalley PC (1987) ‘Graphically organised prose’, in E De Corte, H
Lodewijks, R Parmentier and P Span (eds), Learning and instruction, Oxford:
Pergamon Press, 369–381
Warde BL (1936) ‘What does “modern” mean in typography?’, The Penrose Annual, 38,
44–47
Warde BL (1955) The crystal goblet: sixteen essays on typography, London: Sylvan Press
Wason P (1986) ‘Trust in writing’, in ME Wrolstad & DF Fisher (eds), Toward a new
understanding of literacy, New York: Praeger, 285–302
Wendt D (1979) ‘An experimental approach to the improvement of the typographic
design of textbooks’, Visible Language, 13, 108–133
Werlich E (1976) A text grammar of English, Heidelberg: Quelle & Meyer
Westcott R (1971) ‘Linguistic iconism’, Language, 47, 416–428
Weyer SA (1982) ‘The design of a dynamic book for information search’, International
Journal of Man-Machine Studies, 17, 87–107
Whalley PC & Fleming RW (1975) ‘An experiment with a simple recorder of reading
behaviour’, Programmed Learning & Educational Technology, 12, 120–124

Robert Waller • The typographic contribution to language • PhD Thesis 1987 259
References

Whalley PC (1984) ‘A functional model of similarity’, PhD thesis, Department of

Psychology, The Open University
White EE (ed) (1980) Rhetoric in transition, Pennsylvania State University Press
White JV (1980) Graphic idea notebook, New York: Watson-Gupthill
Widdowson HG (1979) Explorations in applied linguistics, Oxford University Press
Widdowson HG (1980) ‘Conceptual and communicative functions in written discourse’,
Applied Linguistics, 1, 234–243
Widdowson HG (1984) Explorations in applied linguistics 2, Oxford University Press
Williamson H (1966) Methods of book design, 2nd edition, London: Oxford U Press
Wilson JA (1844) A treatise of grammatical punctuation, Manchester
Winn W & Holliday W (1982) ‘Design principles for diagrams and charts’, in D
Jonassen (ed), The technology of text, Englewood Cliffs, NJ: Educational Technology
Publications
Winter EO (1977) ‘A clause relational approach to English texts’, Instructional Science, 6,
1–92
Wittgenstein L (1958) Philosophical investigations, translated by GEM Anscombe, 2nd
edition, Oxford: Blackwell
Wittgenstein L (1922/1971) Tractatus Logico-Philosophicus, translated by DF Pears & BF
McGuiness, 2nd edition, London:
Wright P (1978) ‘Feeding the information eaters: suggestions for integrating pure and
applied research in language comprehension’, Instructional Science, 7, 249–312
Wright P (1981) ‘Tables in text: the subskills needed for reading formatted information’,
in LJ Chapman (ed), The reader and the text, London: Heinemann
Wright P (1985) ‘Editing: policies and processes’, in TM Duffy & RHW Waller (eds),
Designing usable texts, Orlando, Florida: Academic Press, 63–96
Wright P & Reid F (1973) ‘Written information: some alternatives to prose for expressing
the outcomes of complex contingencies’, Journal of Applied Psychology, 57, 160–166
Yates FA (1966) The art of memory, London: Routledge & Kegan Paul
Zachrisson B (1965) Studies in the legibility of text, Stockholm: Almqvist & Wiksell.

Robert Waller • The typographic contribution to language • PhD Thesis 1987 260
2020 Supplementary bibliography

The issues addressed in this thesis have been picked up by scholars in the
field of multimodality. The following sources are a good starting point, and
some of them cite this thesis.

Bateman J (2008) Multimodality and Genre: A foundation for the systematic analysis of
multimodal documents. London: Palgrave Macmillan
Bateman, J, Wildfeuer, J and Hiippala, T (2017) Multimodality: foundations, research
and analysis – a problem-oriented introduction, De Gruyter
Delin J, Bateman J & Allen P (2002) ‘A model of genre in document layout’, Information
Design Journal 11, 54–66
Hiippala, T (2017) An overview of research within the Genre and Multimodality
framework. Discourse Context Media 20: 276-284
Hiippala, T (2015) The structure of multimodal documents: an empirical approach,
London: Routledge
Lickiss, MD (2019) Design perspectives on multimodal documents: system, medium, and
genre relations, London: Routledge
Spitzmüller, J (2017) ‘Schematizing information: The macrotypographic framing of text’
in Colleen Cotter, and Daniel Perrin (eds) The Routledge Handbook of Language and
Media, London: Routledge

Robert Waller • The typographic contribution to language • PhD Thesis 1987 261