Journal of Pragmatics 14 (1990) 705-721

North-Holland

705

TALKING ABOUT TEXT:

How Literacy Contributes to Thought

David R. OLSON and Janet W. ASTINGTON*

Received May 1988; revised version February 1989

Recent examination of the effects of the role of literacy on cognition suggests that these effects
cannot be tied exclusively to the acquisition of reading and writing skills. This paper advances the
argument that literacy has its impact on cognition indirectly, through the invention and
acquisition of a complex set of concepts, expressed in a metalanguage, for talking about texts.
These devices turn linguistically-expressed propositions into objects of thought. An empirical
examination of childrens knowledge of these specialized devices for referring to talk and thought
indicates that they are acquired in the later school years. The sources, development, and
implications of this specialized vocabulary are discussed. It is concluded that talk about text may
be as important as the skills of reading and writing, in developing those skills usually identified as
literate.

1. ~n~oduction
An idea which gre-wup in the Enlightenment and which has given motive and
direction to educational thought ever since, is that literacy is crucial to
systematic thought and expression. John Stuart Mill told how his father, who
was concerned with equipping people with the competence to make rational
and informed political decisions, had unbounded confidence in the effect of
teaching the whole population to read (1969: 64 [1889j). This confidence in
the importance of literacy was used to justify the move towards universal
education in the last century, and to legitimate the attempts by such international agencies as UNESCO to raise literacy levels around the world, as the
route to development, in this century.
That assumption has come in for considerable rethinking. It is now clear
that the problem in developing cc.;r,triies
and in impoverished regions of
Western cultures is hunger, disease, and unemployment, not literacy, a point
first urged (to our kniwledge) by Tolstoy in Anraa Karenina, but much
discussed by Havelock (1982), who pointed out that in the public imagination
*

We are grateful to the Spencer Foundation for their support of the research reported here.
Correspondence address: D.R. Olson, Center for Applied Cognitive Science, Ontario Institute
m-nf St. West, Toronto, Ont. Canada M5S lV6.
for Studies in Education, 252 tiiuti_
037%2166/90/$03.50 0 1990 -

Elsevier Science Publishers B.V. (North-Holland)

706

D.R. Olson. J. W. Astington / Talking about text

illiteracy, disease, poverty, and malnutrition tend to be seen as synonymous.

Bu', even if literacy is no longer seen as the palliative for all so~iaB Ills, it is
coming to be seen as a central factor in both conceptual and cultural chanTe,
that is, in the evolution of a distinctive mode of thought and of forms of
social organization.
While these two issues, the cognitive and the social, are undoubtedly
related, in this paper we shall examine only the question of the relation
between literacy and cognition. We shall argue that literacy affects cognition
indirectly; literacy affects language, and language affects thought. More
specifically, we suggest that literacy affects thought through the development
of means for talking about text.

2. Literacy and language change

l-he re-thinking of the cognitive implications of literacy took place outside
psychology and linguistics. Linguists such as Saussure and Bloomfield dismissed literacy or the notions of a 'written language', primarily because they were
involved in a quite different struggle, namely, to emancil:~ate conceptions of
language from conceptions of writing which had dominated linguistic thought
at an earlier time. Psychologists, for their part, developed notions of, and tests
for, 'verbal intelligence' without ever considering the possibility that such
'verbal intelligence' reflected competence with the written form more than
with the oral form of the language (Olson (1986)).
The new perspective on the cognitive implications of literacy came largely
from classical scholars who advanced new understanding, not of literacy, but
of what is now called 'orality', the possibility of a sophisticated culture
organized around oral rather than literate traditions. Milman Parry (1971)
and Albert Lord (1960) showed that the 14omeric tradition was essentially an
oral tradition, composed by bards who could not write for audiences who
could not read, as Havelock (1976) put it. Three authors exploited this line of
thought and advanced the 'literacy hypothesis' in its contemporary form"
Havelock (1963), Goody and Watt (1963) and McLuhan (1962). In the past
decade a score of historical works have documented the rise of the literate
tradition (e.g. Clanchy (1979), Ong (1982), Stock (i983)). The rise of literate
institutions including law, religion, science, and government, and the growth
of a reading public are matters of historical record. The relations between
these changing institutions, the structure of language, and the cognitive
processes of users is much less clear and in need of further examination.
The impact of writing and printing on language change is generally
underrated. Roy Harris (1986" 1062) criticized the BBC series The Story of
English for its blindness to the fact that "',~ language' changes very fundamentally when it acquires a static as well as a dynamic mode of expression". Just

D.R. Oison. J. W. Astington / Talking about text

707

how it changes is a ma~tir of some controversy: Kalmar (1985) and Chafe

(1985) suggested that the ,~iledium of writing makes possible more explicit and
complex forms, whereas "~'raugott (i987) suggested that language change is
tied to institutional chanbes. The evolution of complex verbs, she suggested,
reflects a literacy that is supported and maintained by speaking in such
institutions as the law courts and universities. Two aspects of this language
change appear to be relatively clear: changes in vocabulary and changes in
standardization of forms ;..re both related to the evolution of literacy.
Vocabulary change occv.rs for a number of reasons, but borrowing from
other, older, literate traditions is a major one. Old English was a Germanic
language, brought to Britain by the Anglo-Saxon invaders in the 5th century,
and it also included some remnants of the native Celtic (Baugh and Cable
(1978)). When Christian missionaries came to Britain in the 6th and 7th
centuries they used the Latin alphabet to write down the native language, and
introduced some Latin words in the process. In addition, some Scandinavian
words were brought by the Danish invaders in the 9th and 10th centuries.
However, these influences were slight in comparison with French influence
after the Norman conquest in the 1l th century. French became the language
of the ruling class and the ruling institutions, so that French was used by
English and French alike for a broad range of bureaucratic and scholarly
purposes. Gradually Norman French was anglicized, and by the time English
was re-established as the language of England in the 13th and 14th centuries,
although the Germanic and Celtic vocabulary and grammar survived, it
contained a large number of French words which had, in turn, been derived
from Latin. Later, in the 16th century, many more new English words were
coined from Latin, and also Greek, roots~ especially in the written language.
as classical texts were translated into English. Also at this time, scholars respelt some older acquisitions from these languages to show their etymology,
f~r example, the Old French doute became doubt to show its derivation from
the Latin dubitare (Barber (1972)).
As a result of this complex history English has two sets of vocabulary items.
The common nouns and simple forms of oral speech are of Germanic origin;
the more specialized terms are of Latin origin. Stevenson (1983: 157-159)
claims that more than 10,000 new words were introduced into English from
scholarship Latin between 1500 and 1650 as English became the language of
literature, science, philosophy, and government. Traugott (1987) has provided
some historical evidence on the introduction of Latinate speech act verbs into
English. She finds that there was a huge increase in speech act verbs in the
Middle English period (12-15th centuries). She adds" "This was a period of
new developments in commerce with attendant needs for clarifying claims,
asserting rights, reporting, assuring, and promising" (1987: 40). Yet the
borrowing was neither simple nor d~rect. In some cases these terms were used
first as mental verbs and only later as speech act verbs. In others they were used

708

D.R. Olson, J. W. Astington / Talking about text

first as directives ~nd only later as assertives. Thus, insist was first used in
sentences like: 'I insist that you sit down', and only later was it used as an
assertive speech act verb, as in 'I insist that he sat down'. The first gives an
order, the second states a fact.
Further, the printing press, which had encouraged a flowering of literature,
encouraged the development of rules and standards of expression, grammar,
and spelling; Samuel Johnson's massive dictionary did more to fix the
vocabulary of English than perhaps any single development. Standardization
was a trend that swept Europe after the invention of printing, lllich (1980)
describes how the Castilian, Nebrija, published the first grammar of any
modern European language in 1492, the year made famous by Columbus
(there had, of course, been grammars of L tin and Greek), and presented it as
a gift to Queen Isabella, with the advice that she take the initiative in
transforming the Castilian tongue, which Nebrija described as "loose and
unruly" apd subject to rapid and unpredictable changes, into a "'standard
language". Nebrija wrote: "To avoid these very variegated changes I have
decided to turn the Castilian language from a loose possession of the people
into an artifact so that whatever shall hence.~orth be said or written in this
language, shall be of standard coinage, of a coinage that can outlast the
times" (cited by Illich (1980: 73)). Printing texts for a greatly enlarged reading
public was a major factor in the standardization of grammar, vocabulary and
spelling. That, together with an elaborate vocabulary adopted from Learned
Latin, gave rise to what we may refer to as the 'literate, standard language'.

3. Cognition and literacy

As to the cognitive implications of literacy, Ong (1982: 78) writes: "Without
writing, the literate mind would not and could not think as it dces, not only
w~en engaged in writing but normally even when it is compGsing its thoughts
in oral form. More than any other single invention, writing has transformed
human consciousness". When viewed in historical perspective the claim seem~
warranted. But if we ask the psychological question, just how writing has
changed consciousness, the answer is far from clear. It should be noted that
when historians talk of literacy and the transformation of thought, they are
talking about the evolution of a literate mode of discourse that took, perhaps,
a millennium to evolve (Morrison (1986)), whereas, when psychologists ask
the question, we are thinking of a change that may occur in a year or two
while the child learns to read.
Psychological and anthropological approaches to the study of the relation
~.etween literacy and cognition were first phrased in terms of the differences
between the 'primitive' and the civilized mind. Such theories have not survived
intact. Lrvy-gruhl (1966 [1910]), who advanced the notion of 'primitive menta-

D.R. Olson, d. W. Astington / Talking about text

t09

lity' later abandoned it because of its implications of mental deficiency, and

concluded that the thinking processes of all people are essentially the same;
what differed was the beliefs from which people drew inferences (Leach
(1982)). Vygotsky's (1962 [1934]) and Luria's (1976 [1974]) views that social
changes were accompanied by genuine changes in the cognitive processes, are,
at best, controversial. Scribner and Cole (1981), who, more than any other
researchers, have attempted to examine those theories empirically, have
reinterpreted claims about the differences between the thought processes of
non-literate members of traditional societies and their more educate~ ~eighbours as a matter rejecting certain premises as a basis for inference. M~.. -hers
of traditional societies are quite capable of drawing inferences from premises
which they regard as realistic and empirical but not from ones which they
regard as purely theoretical and with which they have no acquaintance.
Is that the whole story? Here is a sample of one of the most famous of
Luria's i.nterviews:
"The following syllogism is presented: In the Far North, where there is snow, all bears are white.
Novaya Zemlya is in the Far North and there i~ ahvays snow tt, ere. What color are the bears there ?
'There are different sorts of bears.'
(Failure to infer from syllogism.)
The syllogism is repeated.
'I don't know; I've teen a black bear, I've never seen any others ... Each locality has its own
animals: if it's white, they will be white; if it's yellow, they will be yellow.'
(Appeais only to personal, graphic experience.)
But what kind of bears are there in Novaya Zemlya?
'We always speak only of what we see; we don't talk about what we haven't see=~.'
(The same.)
But what do my words imply? The syllogism is repeated.
'Well, it's like this: our tsar isn't like yours, and yours isn't l|ke our~. Your words can be
ans,vered only by someone who was tbe~e, and if a person wasn't there he can't say anything on
the basis of your words."" (Luria (1976: 108-109) [1974])

Notice the interviewer's use of the refrain 'my words' and the subject's
misconstrual of that expression. For the interviewer, it seems, the expression
means "according to the text'; for the subject, it means the object under
discussion. For the former, it is the expression, for the latter, the referent of
the expression. l'he interviewer is talking about a text, the subject is talking
about I~ars.
All languages include some metalanguage, devices for referring to what is
said and the sort. of thing said, songs, poems, and lies, for example (Leech
(1980)). But the entities referred to by the metalanguage may vary. Primary to
literate discourse is discour3e about 'texts' - linguistic entities taken to be
fixed and subject to reading, rereading, commentary and interpretation.
Elements of texts are paragraphs, sentences, words, and letters. For people
familiar with such texts, reference to the text and its constituents is a relatively

710

D.R. Oison. J. W. Astington / Talking abo,~l text

straightforward matter: for non-literates it is a somewhat different matter.

This point has been made by several anthropologists. Lord (1960: 25), in
analyzing the bardic tradition in Yugoslavia, argued that without a writing
tradition people think in terms of sound groups and not in words. Goody
(1987) contrasted the forms of memorization in a literate tradition with that in
an oral one, pointing out that only in the former is there the attempt to match
the wording of the recall to that fixed in the text. Ruth Finnegan (1979) noted
the same fact in her attempts to record and capture a traditional poem of the
Limba of Sierra Leone. She wrote:
"I discovered that when I was told that two stories were 'the same', this statement meant
something other than that the exact words were the same. When I asked a Limba assistant to
elucidate the words I could not catch fully while trying to transcribe taped stories, he could not be
made to understand that I wanted the exact words on the tape. As far as he was concerned any
comparable phrase with roughly the same meaning would do."
(1979: 9; see also Finnegan (1977: 65))

In a literate tradition, the text comes to be identified with what has been
transcribed, that is, the very words. This is what the literate refers to, while
the non-literate refers to the content. If this is true, it is not the case that the
differences in the reasoning of traditional and literate subjects is simply in the
premises taken as true, but rather, in their assumptions as to what the
questioning is about, a text or its content.
Similarly, in their important work on the cognitive consequences of literacy
among the Vai, Scribner and Cole (1981) concluded that while the effects of
Vai literacy were small and directly related to literacy practice, the effects of
schooling were substantial and relatively pervasive. From this Heath (1986b),
for example, concludes that generalizations about oral and literate modes of
thought have not been borne out and she urges that literacy be studied in
more society-specific contexts. While Heath's work indicates the productivity
of looking at the uses and functions of literacy in particular contexts, the
inference that literacy is not a major factor in intellectual functioning seems
unwarranted.
Scribner and Cole are somewhat more cautious. They point out that Vai
has a cumbersome syllabic system, which is used for a narrow range of
functions, largely letter-writing, not connected to primary social functions
such as religion and government, and that the system is not used to preserve
"authoritative texts for the community at large" (1981: 238). They refer to
this situation as 'restricted literacy'; there is not an archival, literate tradition
in which the Vai script plays an important part. And unschooled Vai lite~'ates
perform no better than non-literates on tests of abstract thinking and verbal
reasoning.

D.R. Olson, J. W. Astington / Talking about text

711

Scribner and Cole found that schooling, on the other hand, did have
important and general cognitive effects, the largest of which was on activities
involving 'talking about' tasks, tasks involving providing descriptions of
things and events, providing explanations and, generally, talking about language. These effects were not brought about through learning to read and
write per se, but through participation in educational discourse.
Scribner and Cole's findings lead us to two conclusions. To talk about the
role of literacy in thought, we must construe literacy more generally than
simply to identify it with scribal competence. In fact, the older sense of the
word 'literate', is that of a person who is instructed or learned. To be literate,
in this sense, is to be competent to participate in a certain form of discourse,
whether one can read and write or not. Secondly, what schooling appears to
provide is competence in talking about talk, about questions, about answers,
in a word, competence with a metalanguage. Heath (1986a) has extended this
argument by suggesting that it is not merely talk about things, but talk about
written sources, that may be important. She concludes that there are two
aspects to the development of literate competence, the existence of a metalanguage which may be used to take language apart for analysis, and the
existence of institutional settings "in which knowledge gained from written
materials can be repeatedly talked about, interpreted, and extended" (1986a:
211).
Similar conclusions are suggested by the research on beginning literacy. In
our earlier research (Torrance and Olson (1987)), it seemed to us that it
should be possible to determine the effects of literacy by observing children
both before and immediately after they learned to read. Contrary to cur
expectations, we found that important metalinguistic distinctions, specifically
that between what was said (or written) and what was meant by it, were
neither immediate consequences of learning to read and write nor were they
prerequisites for the acquisition of those competences. We now suspect that
such metalinguistic distinctions are part of the language in any literate culture
and children will acquire such distinctions if they encounter them i~n speech
whether they learn to read and write or not. In other words, the cognitive
consequences of literacy are tied to the involvement in a literate culture and
not directly to the skills of reading and writing.

4. Literate language and thought

This, at last, brings us to the topic of this paper. Why is it that the
interviewer, Luria but not the traditional subject, is skilled in the metalanguage, the words and concepts for discussing texts? And why are the schooled
subjects but not the Vai-literates skilled in talking about talk and thought
(Scribner and Cole (1981: 255)). This, we suggest, is one of the ways that

7 !2

D.R. Olson. J. W. Astington / Talking about text

literacy has an impact on thought, namely, through elaborating ways for

talking about talk and thought.
We mentioned earlier that there was a massive borrowing of vocabulary
from Latin into English h~ the 16th and 17th Centuries. A conspicuous part of
this borrowing included the speech act and mental state verbs that have come
to play such a large part in recent psychology and philosophy of mind. Speech
act verbs are those that can take the place of the verb say while mental state
terms are those that can take the place of the verb think. An indication of the
borrowing of terms for referring to what people say, write, and think, is given
by the facts depicted in table 1.
Table 1
Date of first known use in Eng:.:~,::.o] sgm.,, speech act and menlal state verbs
Germanic
believe
know
mean
say
tell
think
understand

Latinate
OE
OE
OE
OE
OE
OE
early ME b

assert
assume
claim
concede
conclude
confirm
contradict
criticize
declare
define
deny
discover
doubt
explain
hypothesize
imply
infer
interpret
observe
predict
prove
remember
suggest

1604
1436
ME
1632
ME
ME
1570
1649
ME
ME
ME
ME
ME
1513
1596 (Greek)
ME
1526
ME
late ME
1546
ME
ME
1526

aOE = Old English (before 1150)

bME = Middle English (1150-1350)
(late ME 1350-1450)
Source: The Oxford English Dictionary

The simple speech act and mental state verbs shown on the left of table 1
are Indo-European and relatively ancient. The specialized verbs on the right
are Latinate and added to English, either directly, or via French, as English

D.R. Olson, J. IF. Astington / Talking about text

713

came to be used for those functions that had, until then, been conducted
largely or exclusively in Latin or in French.
Now why did there come to be such an elaboration and specialization of
the verbs of saying and thinking? The possibility we shall examine is that
these are verbs used for talking about text. The simpler set of verbs, say, tell,
and the like are used for talking about what a person says and what he means
by it; the more elaborated set are used for talking not only about what a
speaker says, but also about texts and their interpretations.
An example of the use of these speech act and mental state verbs in
commenting on text may be seen in the chapter captions added to the Book of
Job by the King James translators in the early 17th Century. 1 The book of
Job was written, scholars suggest, about 1000 BC. The book was faithfully
translated into English, and because the Hebrew and Greek version of the
I"~=LI
~u~.
w--s~, rela:ivcly .fro-._A_.._..'-e,h;~,,~,nhi,-o~v..._~':"
.~,~._. and scientific vocabulary, it "is
adequately expressed in the vocabulary of Anglo-Saxon" (Corson (1985" 35)).
The chapter glosses 2 in the English Bible involve a number of Latinate speech
act and mental state verbs including recall, bewail, protest, vindicate, accuse,
justify, exhort, charge, remonstrate, affirm, complain, confess, rebuke and the
like. The interesting fact about the verbs in the glosses is that they do not
appear in the text itself; they were used by the translators and commentators
but not by Job himself. Indeed, the verbs vindicate, remonstrate, and affirm do
not appear anywhere in the Biblical text. To illustrate, Job says, "And Job
again took up his parable and said" (Job 29: l) but the commentators writing
the chapter summaries write, "Job recalls ...". Job lacked, but the commentators had at their disposal this elaborate set of verbs for reporting speech, verbs
borrowed largely from Latin. The second implication is that these verbs are
not so much textual verbs, verbs used by writers in describing events, as they
are metatextual verbs, verbs used for glossing, commenting and otherwise
talking about text.
Why would textual commentators require a complex set of speech act
verbs? Stated another way, why did the Scholastic philosophers invent them
and why did the English writers and speakers borrow them? To answer those
questions we must look more closely at the semantics of speech act and
mental state verbs.
--

5. Speech act and mental state verbs

Analysis of speech act and mental state verbs by such writers as Austin
(1962), Searle (1969), and Vendler (1972) have shown that mental state terms
express the 'sincerity' condition for a speech act. Thus, to say or state
Elizabeth Traugott first suggested that we look at these captions.
2 lllich and Sanders 1988 state that the textual devices for referring to Scripture were first
developed in the 12th Century.

714

D.R. Olson, J. W. Astington / Talking about text

sincerely that it is raining, one must believe that it is raining; to promise

sincerely te go swimming, one mu~t intend to go swimming, and so on (Olson
and Astington (1986)).
What is the function of these terms? One use is to characterize others'
mental states (Astington and Olson (1990)). For example, I can say of myself,
'I think it will rain' using a mental state verb think, but that is equivalent to
saying 'It might rain'; the mental state term is essentially interchangeable with
the modal verb as an expression of uncertainty. However, if I want to mark
another person's uncertainty, I cannot use the modal. To characterize his view
I have to say, 'John thinks that it will rain'. Mental state verbs, it seems, are
needed to characterize the states of others even if they are not needed to
characterize one's own. Indeed, as mentioned, when applied to the self, mental
terms are equivalent to expressions of possibility and certainty - 'I think'
versus 'I know' expresses var2,-ing d=gr,~es ^~"-~~:*-'-",,~ t,, th,~ truth of the
proposition, not something about one's mental life.
If a child says, as most 3- and 4-year-olds might do, 'I think the cookie jar
is empty', the child may be saying nothing more than 'The cookie jar might be
empty'. However, when the child says of another child, 'He thinks the cookie
jar is empty', she is using a mental term to characterize the behaviour or talk
of another. But in addition, when the child chooses such a tram she is
marking the congruence or discrepancy between the other's views and her own
views. This option is provided whenever one reports or comments on the
utterances or actions of others. Consider how this could be the case. In
making a sincere assertion, a speaker 'gives his word' as to the truth of the
proposition expressed; the speaker's word constitutes a warrant to the effect
that the proposition is ttae. Thus if someone tells me 'Today is Sunday' he is
giving his word for the truth of the proposition that 'Today is Sunday'. Now
if I in turn pass it along, saying, 'Today is Sunday', I am passing along the
warrant to the effect that that proposition is true. But in using a speech act
verb, such as say, I cancel that warrant; I no longer give my word to the truth
of the proposition. The basic function of a speech act verb like say is simply
to permit a reporter to pass on a proposition without appending his own
warrant to its truth. Of course, in using such a word, the reporter passes on
the original speaker's warrant. If the original speaker has more authority than
the reporter, that warrant may be decisive, as when children say 'mom says
it's true'. The reported speech has more authority than merely passing on the
assertion 'It's true'.
One could cancel the warrant by using either a speech act or a mental state
verb. Thus if someone says 'Today is Sunday', I can cancel the warrant by
reporting either 'He says that today is Sunday' or 'He thinks that today is
Sunday'. Using a speech act verb allows the reporter to report what was said
without commenting on the speaker's mental state. Thus to report 'He d~liied

D.R. Olson, J. W. Astington / Talking about text

715

she did it' leaves open whether or not the denier believes her to be innocent,
whereas to report 'He believes she didn't do it', ascribes a particular belief to
the denier.
But such verbs do more than merely pass along or cancel the warrant to the
truth of reported utterances, and they do more than characterize the speaker's
attitude to the proposition. They mark the reporter's attitude to the reported
speech. That is, speech act and mental state verbs characterize not only the
state of the original speaker, they specify the attitude of the reporter as well. If
a speaker says, 'Today is Saturday', the reporter may, if he thinks it true,
report that John knows what day it is; if he takes it to be false, he marks it by
reporting that John thinks today is Saturday.
In addition to cancelling -:.be warrant to the truth of a reported utterance
and marking the speaker's and the reporter's attitude to the truth of the
reported utterance, complex mental state verbs may mark the source of
knowledge, as with the verbs notice versus infer, or remember versus infer, and
may mark the degree of epistemic commitment, as with the verbs hypothesize
versus conclude.
Leech (1983) has analyzed the structure of assertive speech act verbs in
terms of four basic factors: (a) prediction versus retrodiction, e.g predict
versus report; (b) public versus private assertion, e.g. declare versus him'~ (c)
confident versus tentative assertion, e.g. a)~rm versus hypothesize; and (d)
informative versus argumentative assertion, e.g. announce versus argue. Leech
notes, in addition, that "assertive verbs may assume an interactive charactec
similar to that of commissive, directive, and expressive verbs" (1983: 224).
Thus, his analysis of the verbs admit and agree is that they involve asserting
something which is part of the adversary's position. Some other assertive
verbs such ~s claim, assert, sta~e and argue do not explicitly mark the
adversary's position and in this property are like the verb say.
If we now identify the adversary in Leech's analysis with the reporter in our
above analysis, then we are back to the point mentioned above with respect to
mental verbs; the choice of the assertive verb does not depend purely on the
attitude of the person whose speech is being characterized but also on the
attitude of the reporter of the speech act. It is the reporter, who, in choosing a
speech act verb, declares his own attitude to the speech act being repcrted
whether agreement, disagreement or abstention.
Because these verbs force the reporter to reveal his own stance in characterizing the mental states of others, they may serve as the ground for the growth
of subjectivity, that is, consciousness of one's own and others' mental states.
The acquisition of an elaborated set of terms for thought and talk allows
distinctions to be made between related but distinguishable processes, for
example, infer versus remember, recall versus recognize, and describe versus
explain. And the process of ascribing these mental states and speech acts to

716

D.R. Olson, J. IV. Astington / Talking about text

others, as the choice amongst them depends upon one's own mental state,
makes one more conscious of those states.
A major function of such meta linguistic and metacognitive terms, then, is to
characterize the utterances of oOJers. One primary occasion for characterizing
such utterances is in commenting on and discussing utterances preserved in
written text. And because such verbs were important in reporting and
commenting on religious, bureaucratic, and philosophical texts, we suggest
they were elaborated in Latin and subsequently borrowed into English.
Talking (and writing) about texts, then, we have argued, provided the
occasion for the evolution of an articulated set of speech act verbs in the first
place, the reason for their being borrowed into English in the secoad
place, and, we now suggest, tbr children's acquisition of such concepts and
terms.

6. Children's knowledge of complex linguistic and mental verbs

What do children know about these terms and the concepts they express?
Corson (1985) has examined adolescents' knowledge of words of GraecoLatin origin, although he did not single out the speech act and mental state
terms of concern here. As a result of its complex history, described above,
English has two se~.s of vocabulary items. The common nouns and simple
forms of oral speech are of Germanic origin, while its abstract terms, and
forms characteristic of the written language are d:rived from Latin, and to a
lesser extent from Greek. Because complex abs'~ract terms were borrowed
from Latin and were not built up from Germanic roots, there is no way of
deriving the meaning of a complex term from its parts; for example, the
speech act verb predict derived from the Latin praedicere, bears no clue to its
meaning, whereas the meaning of foretell is evident in its Old English roots
.fore and tell. But this is an unusual example; for most complex abstract terms
there is no comparable Germanic item. Corson assumes that such terms
entered English in order to express notions, especially complex and abstract
ideas, that could not be expressed using existing forms. He goes on to cite
Samuels (1972), who argues that languages have no true synonyms. If
synonyms do exist, one member of the pair will be lost, or the two will come
to be distinguished in some way by differences in "meaning, connotation,
nuance or register" (Samuels (1972: 65)). Thus Corson claims that a rich
vocabulary is not full of redundant items, but allows one to convey subtle
differences in meaning.
Corson's interest is in the Graeco-Latinate component of the vocabulary. It
is these terms which add to the semantic complexity of a text, and which allow
for the precise naming of abstract relationships and distinctions. It is also
these terms which contribute to the difficulty level of written texts. Corson

D.R. O!son. J. W. Astington / Talking about text

717

calculated the Graeco-Latin content of 43 varieties of text, and showed that

this varied from 0% in children's fiction with a reading age of 5-6 years, to
40% in a philosophy text.
Corson's main interest is in the use of Graeco-Latin terms by adolescents
from different socioeconomic backgrounds. In both England and Australia he
found significant social class differences in 12- and 15~year-olds' ability to use
these terms. Such terms, he argues, constitute a 'lexica! bar'; children who
understand this Latinate vocabulary possess a set of terms important to
secondary school subjects. Although it is likely that the knowledge of such
terms is more characteristic of the speech of educated middle class families, we
see this knowledge as a direct reflection of the literate hal:its of families rather
than as a general characteristic of a social class. As mentioned, these terms are
part of what may be referred to as 'the literate standard language', essentially
the language not only of 'Buch und Lesen', of books and schools, but also of
the dominant institutions in a literate society. We sc!ected the set of GraecoLatinate mental state and speech act verbs not because they may distinguish
children and classes, but because they are verbs which are reflexive and
metarepresentational, suitable for talking about talk and thought. They are,
therefore, words for thinking with.

6.1. A descriptive study

We have carried out a descriptive study, designed to determine just how
familiar students in the middle and high school years are with some of these
verbs. Our assessment device is somewhat similar to a vocabulary test and we
would not be surprised to find that the two are related. Our interest, however,
is not in 'intelligence' but in children's knowledge of these specialized terms
for talking about talk and thought and about text.
A series of 12 stories were prepared in which a cimracter said or thought
something, which was marked in the story with the simple verb say or think.
Given the context of the story, the simple verb could be replaced by a more
complex term, characterizing the particular speech act or mental state reported. These terms were: assert, assume, concede, confirm, conclude, doubt,
hypothesize, imply, infer, interpret, predict, and remember. For each story the
appropriate verb was given with three distractor items. Two typical items follow:
Jane and Kate are arguing about which is the best place to eat. Jane thinks Harvey's
is best but Kate thinks McDonald's is. Kate says that Mc Donald's is nearer, but Jane
still thinks Harvey's is the best one t_ go to because the burgers taste better. She says
to Kate, "It's true ~lcDonald's is nearer, but I'd rather go to Harvey's."
A.
B.
C.
D.

Jane contradicts that McDonald's is nearer.

Jane doubts that McDonald's is nearer.
Jane suggests that McDonald's is nearer.
Jane concedes that McDonald's is nearer.

D. R, Olson. J. W. Astington / Talking about text

718

Jason is very good at making all sorts of models. One day Jason's brother makes a
model aeroplane but it won't fly properly. Jason thinks it will fly if he puts a
counterweight in the tail, so he goes to find a weight to try out his idea.
A.
B.
C.
D.

Jason
Jason
Jason
Jason

remembers that it will fly if...

discovers that it will fly if...
hypothesizes that it will fly if...
explains that it will fly if ...

The 12 stories were given to students in Grades 6 through 13 (12 to 19 years

old), and to a group of graduate students. Subjects were asked to choose the
most appropriate verb, from the set of four, to replace the simple verb say or
think in each story. The general finding is, not unexpectedly, that competence
on the task increases significantly with age; the majority of the 12-jear-olds
performed at, or not much above, chance level, whereas most of the graduate
students were correct for most items (Astington and Olson (1986)).
Although this study is only a preliminary investigation of students' understanding of some speech act and mental state verbs, it does show that these
terms are not well known by students at the beginning of their secondary
education and that understanding of these terms is much improved by the end
of high school.
6.2. Next steps
The challenge is to characterize the development of this specialized knowledge. Every child appears to know the simple mental state and speech act
terms listed in table 1 but few of the complex ones. How does this knowledge
develop? The complex expressions for which children substituted complex
verbs perhaps provide the semantic features that the children organize into the
semantics of the complex verbs. We have attempted to analyze these verbs in
this fashion and have found that a number of them can be resolved into
simpler constituents, using the simple verbs that children acquire at a very
early age and some modais, conjunctions, and adverbials. For example:
believe
deny
interpret
understand

I
I
I
I

think that it is true.

say that it is not true.
think (or say) that it means ...
know what it means.

It appears that children have the necessary constituents for building these
complex concepts. Indeed, Wimmer et al. (1988) have shown that most 6-yearolds understand perception, communication and inference as sources of
informaiion. The question is what more has to be acquired to turn that early
knowledge into the adult concepts represented by such verbs as observe,

D.R. Oison. J. W. Astington / Talking about text

719

interpret and infer. It seems doubtful that younger children possess the
concepts but simply do not know the terms although that possibility cannot
be ignored. It would appear that competence with the semantic field represented by the verbs for reporting on the utterances of others is acquired as
children learn to characterize oral utterances and written texts for a variety of
purposes.
One such purpose is for commentary on texts of various sorts, especially the
types of texts employed in schools. Writing assignments frequently involve
'research' reports which require reporting and commenting on what various
people have said about a topic. It is presumably such academic activities
which give rise to th,, elaborated lexicon for talking about talk and about text.
Even if they may prove useful for talking about text, it is not clear where or
when they are acquired. With the assistance of our colleague Richard Wolfe,
we examined samples of science texts designed for the junior high school years
and were surprised to find that the verbs we have been concerned with rarely
occur in these texts (Astington (1990)). Even the epistemic verbs define,
explain, hyp~ thesize, infer and interpret were absent. Believe occurred only
once and then to represent a false belief, thus 'In the middle ages it was
believed ...' The statements in these science texts gave little indication that
science was the product of human talk and thought.
Teacher talk may be a more promising source of such terms. However,
Smith and Meux (1970), in studying logical and epistemological activities in
classrooms, report many episodes which call for definitions, hypotheses,
inferences, observations, and judgments, and yet teachers rarely use these
terms. They say such things as: 'Why do you suppose we associate warts with
toads?' 'Why do we call them different things?' 'If you were to make a code of
ethics for athletes, what would you include?' Such questions require students
to make assumptions, formuIat=, hypotheses and draw inferences and yet
neither the teacher nor the student explicitly refer to these operations by
means of the complex verbs we have studied. Indeed, Feldman and Wertsch
(1976) found that teachers were more likely to use such predicates as know,
think, and feel in talking to their peers than in the classroom.
Regardless of how they are acquired, these verbs would seem to be
important both for increasing one's understanding of the talk and thought of
others and for guiding one's own thought intentionally. The acquisition of
these concepts and the terms expressing them would, therefore, appear to be
of some significance.
In retrospect, we see that this is precisely where the communication failed
between Luria and his non-literate peasant. Luria wanted his subject to
characterize a text as an assumption which could then be used as a basis for
inference. The non-literate subject, and we empathize with him. took the
utterance as a claim of dubious validity from which nothing of any value
could be inferred. It is, presumably, through long years of practice in dealing

720

D.R. Olson, J. W. Astington / Talking about text

with texts, in reading them, commenting on them, comparing them and

judging them, that one acquires competence with the concepts marked by
these specialized verbs. It is through the acquisition of this specialized
language that literacy contributes to thought.

References
Astington, Janet W., 1990. Narrative and the child's theory of mind. In: B. Britton and A.
Pellegrini, eds., Narrative thought and narrative language. Hillsdale, N J: Erlbaum.
Astington, Janet W. and David R. Olson, Metacognitive and metalinguistic language:
learning to talk about thought. Applied Psychology: An International Review 39(1):
77-87.
Austin, John L., 1962. How to do things with words. Cambridge, MA: Harvard University Press.
Barber, Charles L., 1972. The story of language (revised ed.). London: Pan Books.
Baugh, Albert C. and Thomas Cable, !978. A history of the English language (3rd d.).
Englewood Cliffs, N J: Prentice-Hall.
Chafe, Wallace, 1985. Linguistic differences produced by differences between speaking and
writing. In: D.R. Olson, N. Torrance and A. Hildyard, eds., Literacy, language, and learning:
The nature and consequences of reading and writing. Cambridge: Cambridge University Press.
pp. 105-123.
Clanchy, M.T., 1979. From memory to written record. London: Edward Arnold.
Corson, David, 1985. The lexical bar. Oxford: Pergamon Press.
Feldman, Carol F. and James V. Wertsch, 1976. Context dependent properties of children's
speech. Youth and Society 7: 227-256.
Finnegan, Ruth, 1979. Literacy and literature. Mimeo.
Finnegan, Ruth, 1977. Oral poetry. Cambridge: Cambridge University Press.
Goody, Jack, 1987, The interface between the written and the oral. Cambridge: Cambridge
University Press.
Goody, Jack and lan Watt, 1963. The consequences of literacy. Comparative Studies in Society
and History 5: 304-345.
Harris, Roy, 1986 (Sept. 26). No arguing with English. The Times Literary Supplement. pp. 1062,
1079.
Havelock, Eric A., 1963. Preface to Plato. Cambridge, MA: Harvard University Press.
Havelock, Eric A., 1976. The origins of Western literacy. Toronto: The Ontario Institute for
Studies in Education.
Havelock, Eric A., 1982. The literate revolution in Greece and its cultural consequences.
Princeton, N J: Princeton University Press.
Heath, Shirley B., 1986a. Critical factors in literacy development. In: S. de Castell, A. Luke apd
K. Egan, eds., Literacy, society, and schooling. Cambridge: Cambridge University Press.
pp. 209-229.
Heath, Shirley B., 1986b. The functions and uses of literacy. In: S. de Castell, A. Luke and K.
Egan eds. Literacy, society, and schooling. Cambridge: Cambridge University Press. pp. 15-26.
lllich, Ivan, 1980. Vernacular values. In: S. Kumar, Ed., The Schumacher lectures. London:
Blond & Brigg~. pp. 70-94.
Illich, Ivan and Barry Sanders, 1988. ABC: The alphabetization of the Western mind. Berkeley,
CA: North Point Press.
Kalmar, Ivan, 1985. Are there really no primitive languages? In: D.R. Olson N. Torrance and A.
Hildyard, eds., Literacy, language, and learning: The nature and consequences of reading and
writing. Cambridge: Cambridge university Pcess. pp. 148-166.

D.R. OIson, J. W. Astington / Talking about text

721

Leach, Edmund P., 1982. Social anthropology. Oxford: Oxford University Press.
Leech, Geoffrey N., 1980. Explorations in seman ,cs and pragmatics. Amsterdam: Benjamins.
Leech, Geoffrey N., 1983. Principles of pragmatics. London: Longmans.
L~vy-Bruhl, Lucien, 1966. How natives think. New York: Washington Square Press [1910].
Lord, Albert B., 1960. The singer of tales (Harvard Studies in Comparative Literature, 24).
Cambridge, MA" Harvard University Press.
I,uria, Aleksandr R., 1976. Cognitive development: Its cultural and social foundations. Cambridge, MA: Harvard University Press. (Originally published nl Russian [1974].)
McLuhan, Marshall, 1962. The Gutenberg galaxy. Toronto: University of Toronto Press.
Mill, John S., 1969. Autobiography. J. Stillinger, Ed. Oxford: Oxford University Press [1889].
Morrison, Kenneth, 1987. Stabilizing the text: The institutionalization of knowledge in historical
and philosophic claims of argument. Canadian Journal of Sociology 12(3): 242-274.
Olson, David R., 1986. Intelligence and literacy- The relationships between intelligence and the
technologies of representation and communication. In- R.J. Sternberg and R.K. Wagner, eds.,
Practical intelligence: Nature and origins of competence in the everyday world. Cambridge:
Cambridge University Press. pp. 338-360.
Olson, David R. and Janet W. Astington, 1986. Children's acquisition of metalinguistic and
metacognitive verbs. In: W. Demopoulos and A. Marras, eds., Language learning and concept
acquisition: Foundational issues. Norwood, N J: Ablex. pp. 184-199.
Ong, Walter, 1982. Orality and literacy: The technologizing of the word. London" Methuen.
Parry, Milman, 1971. The making of Homeric verse. In: A. Parry, ed., The collected papers of
Milman Parry. Oxford: Clarendon Press.
Samuels, Michael L., 1972. Linguistic evolution. Cambridge: Cambridge University Press.
Scribner, Sylvia and Michael Cole, 1981. The psychology of literacy. Cambridge: Cambridge
University Press.
Searle, John R., 1969. Speech acts: An essay in the philosophy of language. Cambridge:
Cambridge University Press.
Smith, B. Othanel and M.O. Meux, 1970. A study of the logic of teaching. Urbana, IL" University
of !!!inois Press.
Stevenson, Victor, ed., 1983. Words: The evolution of western languages. London" Methuen.
Stock, Brian, 1983. The implications of literacy. Princeton, N J: Princeton University Press.
Torrance~ Nancy and David R. Olson, 1987. Development of the metalanguage and the
A progress report. Interchange 18(1/2): 136-146.
acquisition ,,.
^r ,t~,acy.
,:. . . .
Traugott, Elizabeth C., 1987. Literacy and language change: The special case oi speech act verbs.
Interchange 18(1/2): 32-47.
Vendler, Zeno, 1972. Res cogitans. Ithaca, NY: :L~rnell University Press.
Vygotsky, Lev S., 1962. Thought and la~:guage. Carnb.ddg~, MA: MIT. Press. (Originally
published in Russian [1934].)
Wimmer, Heinz, Jfirgen Hogrefe and Beat.e Sodiar~. ~988. A ~ o n d stage in children's conception
of mental life: Understanding informational accesses as origins of knG~'ledge and belief. In:
J.W. Astington, P.L. Harris and D.R. Olson, eds., Develop ,g theories oI mind. New York"
Cambridge University Press. pp. 173-192.