ewar

mem e

ESCRIPTORS Affective Behavior; - Cogniti=ve

Ski inguistics o e
Readin _ nstructio
*Rea. esearc

ection of
 U.S:DEPARTMENT OF HEALTH::
EDUCATION A WELFARE-
= - OFFICE OF EDUCATION7
THISsDOCUMENT,=HAS=HEEN _REPRO._
DUCED,EXACTLY=AS RECEIVED FROM=
THE-PERSON OR ORGANIZATION_ORIG =
== INATING ITzPOINTS OF VIEW OR OPIN
-=- =IONS -STATED=DO,NOT-sNECESSARILYn: ==
REPRESENPOFFICIAGOFFICE_IOPEDU-
CATION POSITION OR POLICY,==--,

W/7/11..
47G 7:7 V, n
--\,w
 mversit

ssistan xecutive ecre a ONALD

ewar aVa
lications oor
atioriz ewir

end ing sio iatiori =

,10=ERICfAND=ORGANIZATIONS OPERATING
UNDERAGREEMENTS=WITH THE 1.1WOFFICE
OF=" EDUCATION,i= RRTH ER =iREPROCIOCTION
=OUTSICIE,THEERICISYSTEM-REGUIRES_PE
MISSION OPTHE COPYRIGHT OWNER" yy
eactibrit
Or

attors4n

eactio eoreuca a
a esear srz

u strata-
____- ______ _-
an Um,- ro jret_

iforn a-
 F
I

F
F

F
F
IF

I IF

h
F

1I

F
F}I F F
F

F 1
L
Fik r
L, F' F F

hF

FF I
F
F F

F FF
F

F
F
F

FF1
F F

F
F
hi
,==='-'a_3
 c z__---- i --_____- -___ --_ --:_T__ =
_
-=-_=if_:_--_i_r=i---_---- ____
___

== ___- --_-_ _ -_=_-_= -_-i-_ :.- _

_
rrerace

--=--
=
i=----rAtid-14&=tocomplete what we do when we read
ol 'st's achievements, for it
would be most-intricate workings
of the human iind,= as we o-uthel the tangled -story
ei,thost-irernarkablerispeci
=
ormance that civiliation
ame-
_
-===-
 _

= s

:_=--

_ _ = _ _ =__ _=___
T
=-=

-TT:
RETICAL MODELS 1::AND-= PROCESSES;
- __ = =_ s= -

_ _ _ _

__

r-E
-

-
 trictec
e=weou

=early istu
 -_-__ =-

_
 _

-- ----------_:= - __,,_
_
_
 ereri
vague-
, _-_-
-

THEORETICAL MODELS=AND = PROCESSES -0 MARMOT- ---

individu s
formed come

-reinitiative=4
ro em solvin
leving= readers
su
rsiiteiY

-- -----
 THEORETICALi'MODELS-_AND PRoCESSES-Dr-_READING

tructure twationa

------f-c---7--, ---

- -_. _- --- =
-,_--__,_,__,______
=
 z

---- ,------
habits, of course, has partly been a consequence of the view that
linguistic mastery has been regarded as functionally complete by
school entrance age. His summary of language development stud_ ies
during the elementary school years provides a useful overview of the
kinds of investigations which contemporary language theory= has
=

fostered. It should be pointed out, however, that interpretations of

these studies ought to take into account the stimulus mode employed
in eliciting language samples, since the stimulus situation can have a
fundamental effect on= the kind of language elicited. Nonetheless,
emerging from contemporary language studies is the obiervation that
language, seemingly, is acquired in= a rather fixed order, although
the rate =of acquisition may vary considerably due to individual and
environmental factors.
In conjunction with this point, also,, it is well io keep in mind
that many findings reported from child = language research are ex-
pressed in terms of normative =data with the consequence thatrnean
scores on tests or mean chrinological ages in linguistic development
obscure ranges in-individual = scores or ages that comprise the mean.
Language norms provide= important measures for use by the re-
searcher and the practitioner; but language variation ultimately must
be taken into account. Normative data ought_ not b_ e used to obscure
the fact that ranges around a mean = often are far more informative,
particularly in regard to a developmental- view of language acquisi-
tion. As Ruddell has implied, each individual's = biological endow-
ments, in combination with the- consequences of his particular
linguistic environment, can= lead =to a= substantial variation among
elementary school children's language performances.
Distinguishing among causes of- language variation is another
implication that can be derived from Ruddell's presentation. Causal
factors of language variation might be distinguished-in at least three
ways: developmentally, dialectally, and pathologically. Each factor
raises the possibility that a model' of= reading instruction should take
into consideration how modes of instruction may differ- because of
__

possibly different language inputs= into the reading process.

35

done using college subjects: With= one exception, the study using
college subjects was identical to the one using kindergarten children.
Some =of the college subjects getting look-say training were able to
read words on the transfer list on first presentation whereas virtually
none of the look-say kindergarten subjects could do this An-analysis
of how the college subjects did this revealed that those subjects
who had used their knowledge of reading and who had learned
letter sound correspondencei =on their own, were able to transfer
this knowledge to reading the= transfer list. Taken together, the two
studiei give strong support to the= notion that knowledge of letter
sound correspondence is an important basis for transfer to reading
new words.
In the kindergarten study of the effect of phonic versus look-say
training, a task analysis of the skills necessary for transfer to reading
new words indicated that knowledge of letter-sounds as well as the
ability to blend these sounds was required. As the data from the
experiment revealed, the group getting botli kinds of training was
superior to the groups which did not get the necessary training. The
implications of this experiment for the teaching of reading are that
task analyses can be most helpful in planning instructional sequences.
In doing_task_analyses,_specific objectives are_identified.Then,-pre-
 on the other hand, may pose another kind of possible mismatch,
that which results from differences between the formal features of
the language of instruction and the formal features of language
which the child has learned in his particular linguistic community.
The instructional resolution of such a mismatch is currently a topic
of Considerable* debate- specifically regarding whether or not focus
should be placed =on altering the language of instruction to match
the language of the child or vice versa. Pathological variation, in
turn, poses the possibility that, in addition to whatever develop-
mental and dialectal differences might exist, special instructional
techniques may be needed that-are not ordinarily encompassed in
a reading program; as, for example, the techniques that may be used
in working with the visually or auditorily= handicapped child.
These three factorsdevelopmental, dialectal, and pathological
are not intended-as inclusive fcirmk of linguistic variation. None-
theless, they do call attention to the need_for a model of reading
instruction to outline instructional-interVentions which are cued by
the kinds of linguistic =variations which can and do occur in the
language of the elementary school child.
A concluding remark: despite the growing data about language
acquisition, Ruddell's admonition that subitantial work in other
dimensions of language in relation to reading is needed is war-
ranted. His presentation 'clearly indicates that, despite some signif-
icant theoretical forinulations and equally significant research, there
is still "a good distance to go in = developing =a theory of reading and,
from such a theory, a model of reading instruction. Ruddell's own
contributions, and hie' comprehensive summary of the contributions
of other researchers, are welcome additions to the accumulating data
about the reading process and its language base.

36 THEORETICAL MODELS AND PROCESSES OF READING

4. Bormuth, JR. "Readability: A New Approach," Reading Research= Quar-

terly, 1 (Spring 1966), 79-132.
5. Cattell, J. "The Inertia of the Eye and -Brain," James McKeen
Man of Science, Psychological Research, 1. Lancaster, Penna.: Science Press,
1947.
6. Cattell, J. "On the Time Retluired for Recognizing and Naming Letters and
Words, Picture', and= Colors," James McKeen CattellMan of Science,
Psychological Research, 1. Lancaster, Penna.: Science Press, 1947.
7. Durrell, D. D., et al. "Success in First Grade Reading, =Boston University
Journal of Education, 3 (1958), 2.47.
8. Erdmann, B., and R. Dodge. Psychologische UntersUchungen fib_ er das Lesen
auf Experimenteller Grundlage.-Halle: Neimeyer, 1968.
9. Gibson, E. J., et al. "The Role of Grapheme-Phoneme Correspondence in
the Perception bf Words," American Journal of Psychology, 75 (1962),
554.570.
10. Gibson, E., J. Farber, and S. Shepela. "Test of a Learning Set Procedure for
the Abstraction of Spelling Patterns, Project Literacy Reports #8. Ithaca,
N. Y.: Cornell University, 1967, 21-31.
11. Jeffrey, -W. E., and S. J. Samuels. 'The Effect of Method of Reading Train-
ing on Initial Reading and Transfer," Journal of Verbal Learning and
Verbal Behavior, 6 (1967), 354.358.
12. Marchbanks, G., and H. Levin. "Cues By Which Children Recognize Words,"
Journal of Educational Psychology, 56 (1965), 57.61.
13. Martin, E. "Relation Between Stimulus Recognition and PairedAssociate
 1 Preparation of = this paper was-supported in part by grants to the Univer-
sit), of Minnesota, Center for Research in Human Learning, from the National
Science Foundation (Gs 1761), the National Institute of Child Health and_ Human
Development (nn-01136), and= the Graduate School of the University of Minne-
et er,a
 THEORETICAL =MODELS 4ND- TRocEssts_or - READING_

4. onia. - Individ_uaI letters and letter_ usters

as cud or sounds. These sounds may combined uent.
recognize t e word. Critics o phonic me of - teaching word
recognition claim thatnglish >s not a highly alphabetic language,,
at is =in English ow corms ence _between letters an
sounds. This is only individual letters. When
individual letters =as well as ett-er c usters an eir itions in
word are ta en into-consi eration, recent work in lin istid indicates
that English has higher letter-sound = correspondence an ever before
realized.
5. Context. Word associations an information ovidid in the
context-of a sentence may-provide e_responsenecessary
response for recogniz-
 - -
-
, SAMUZIS 27

I ing the word. Red, white, and

-. Few English speaking people
require the printed stimulus to recognize the missing word. While
context provides an important cue for recognition and for learning
to reada word, it is important to determine if the reader can recog-
nize a word when it is presented in isolation Tf the student does not
visually attend to the stimulus whenhe says the word, he may not
t learn to read it.
When the beginning reader uses strategies such as recognizing
iwords as sight words, using unusual characteristics of words and word
shape as cues, he is learning strategies which not only are not useful
j
for transfer-but will have to be abandoned-if he is to progress to the
point where he can- decode words on his own. Teachers who en.!
I courage their beginning students to use word shape and the whole
word as cues have the mistaken belief that children ordinarily note
a whole configuration. One reading textbook (26) states, "To start
with a whoie word is sound psychologically, for young children are
not prone to be very analytical in their perceptions. Their natural
- tendency is to perceive total patterns."
If there is anything which discrimination studies indicate, it is
that children select the easiest cue for recognition, and the easiest
cue is frequently just a single letter of a word or some incidental
-
detail. Children do not ordinarily attend to total patterns nor to all
the letters in a word. It is only when single letter cues fail to distin-
guish one word from another that children attend to all the letters.
To determine which cues nonreaders and begiining readers use
j in word recognition, Marchbanks and Levin (12) had kindergarten-
ers and first graders select the one word from a set of alternatives
which was similar to a standard. The selection could be on the basis
of word shape or letter cues. The results indicated that children pre.
i ferred to use first letters, final letters, middle letters, and word shape
(in that order of preference) as cues to word identification. This
f study is important bec a it demonstrates that the theorir are un-
tenable which propose tizat beginning readers recognize words as
wholes primarily by shape. It also demonstrates the importance of
letter cues in the word recognition of children.
Under which conditions will children us single letter cues or
all the letters in the word as cues to recognition? To find the answer,
Samuel: and Jeffrey (21) gave kindergarteners a list of four words
 28 THEORETICAL MODELS- AND PROCESSES OF READING

-to learn to-read. One group learned-a list-where the words were easy
to -_visually-discriminate -from each, other. This group ___was called- -the
high diScriminab-, - -group- and theit-Avords were-spelled BE,-
MI, so.- Another groty -was- called-- the low-discriMinability group and
their words-were difficult to Visually_-_discriminate from each other.
The words were spelled_ SE, -SA,- -ME, .MA. coMparison- bet*een -the
two groups on-speed:of-learning :to read :the -words indicated that the
-group gettingthe-list with-highlydiScritninablei-vordS'excelled: Then
a-,test--wasiivento-deterininetthetrettersluiedifas_tties-,forrecOgnition:
The test_reyealed_=_t which had-
learned =more-quickly,- =hatF,Ustd1--single tertues-- as tbt-basis-- for
retognitiOn._Tht ictAkf:distriminabiligEgrou ;--although= earning-less
-rapidlyhad=use&bothzletteislfss-Ahef -_-xteognitis*=14Then
children -own-a!wordpelled,__mo,--a_mord=they
-in

_had=not seen before _those-from- the = high- discriminabilitygroup- were

apt -to say th tword;wa-s:_f -mvor -so =depending_ on whether-they used
a-ifirst or last letter asan ate;in-recognitiOn.-_-Those_ in Ahe-16*--diterim-
inability-grOu0-tended-:taiSay;theEVOrd*as-ibrie had-- not= -seen
before. -Thus, !in itial=itrain oLwOrds-__With loW-rdit-critn-
inability, which forted iatten tionr:on the letters,- -encouraged the
-child_ to- adopt a- strategy --which_- provided-:A better= -basis --for transfer
to learning-new-Words; _

In _rteaching reading _to begiriningreadert,:a deCision must be

made between Speed Of-initiaFlearning:anditrantfer.- The _decision
to foster_ speed- Of, initiat_leariting:it-fthe: expenSt- of transfer-May- be a
false economy _Initial= speed -61Elearning-can be facilitated-- by using
the look-say method -with words Which _are-highly_ditcrirtiinable frOM
each other. For-example, =when- given- =the- sentence, "Yesterday- our
class went to _a_.firestation,",--_the beginning -reader would probably
learn to read the- sentence- using- first or last letter cues. Although
this strategy _would _lead to =rapid -learning, it alsci results in __post'
transfer to learning how--to_iread. new _words-. In learning to read, the
principle of least effort_ operateS. This means that-the- strategy- is to
select from ithe stimulus complex that tut which most easily elicits
the correct response. Cues may be word length, shape, or single let-
ters. These cues are irrelevant -in that they provide no-basis for learn-
ing new words and what is learned will before long inhibit future
learning.
SAMUELS 29

Teachers who begin the teaching of reading by having the

learner recognize a basic group of words as sight words have noted
that at first the learning is rapid, but soon the rate of learning new
words slows down drastically. The initial rate of learning is rapid
because numerous simple strategies- provide cues- for word recogni-
tion. Only so many words can be recognized by length, shape, and
single letters before the strategies prove_ ineffective. When- this
occurs, the rate of learning new_ words decreases, and the learner
reinains on a learning plateat until he learns a= rational system -for
decoding words 'from symbols to-sounds.- -
One strategy for= facilitating_word recognition is to use color cues
with each word as in the words-in-color system. With _this= system
certain sounds are represented by_particular colors. When a word is
printed, it is spelled according to standard-English orthography, but
certain letters are in a particular-col. which represents the pro-
nunciation.- =While this system may increase rate of initial learning,
the critical question- is one of transfer. If the learner focuses his
attention on color and not letter shape, what happens when the color
cues are removed? To answer_ this = question, Samuels (19) had first
graders and college students- learn to =read words printed in color or
words printed in regular type. Samuels == found =that= rate of learning
the words in color was significantly faster than the words in regular
type. But on the transfer testswhen the color cues were removed
the subjects had great difficulty in recognizing the words formerly
in color. Thus, on the transfer tests the tables were turned. In com-
paring recognition between the words which were always in regular
type to the words which had formerly been in color, recognition was
superior for the words which had always been in regular type. What
makes these results so surprising is that the college students knew the
color cues were to be removed. Apparently, the color cue was so
potent they were unable to focus attention on the relevant cue of
letter shape. Again this-study illustrates the principle of least effort
in learning and the dangers of a false economy in which there is rapid
learning at the expense of transfer.
Presently, many teachers are of -the opinion that letter name
knowledge facilitates learning to read. There is mounting evidence,
however, that learning to decode words is not aided by letter name
kndwledge. The basis for the belief regarding the facilitating effect
 of letter name knowledge on reading-may originate from the fact
that causation: is often- mistakenly :imputed= to-correlational findings.
Bond and Dykstra-- (3) :found- in _the first grade -studies -that-- reading
achievement was highly correlatect:With_ letter name- kriOwledge;
in fact, it was _the single_test_predictot of first grade:reading_ suCcess.
Some- -ten -years_ earlier;- Nicholson (7) repOrted that__thei- correlation
betWeen ability to :identiklowercate--:-letters: upon- entrance to first
gradeTand the-rate of_learning-:to:_read=words-ivair--==.51; Which:was
highet--than-__the;correlationilietweenirq'_(r arid;-__the.=rate-=:of
learning_ these=- words: In_-=-Alie:_=-saidekottotto- Linehin--:_stated::that
letter natteanct letter sound-seeined:rd-fatilitatefirst=grade
reading-_-_acluevementi_:Since, :-whick got :lettet:itame: and
Sound=training_reCeive45anditO isCrimination- trainintasi-mell, it
iMpOtsible,toidelerMirie= _this Study_ if the-facilitating effect was
produced-_-_by e=inaine-;:fsonn _itory_-or
DUrrell---(7)_ concludecli-loiveverthtleading difficulties could_
_s_ -prevented if,among- -other ingsi training-in-letter names and=lounds
was given.-
SeveraLcritical=questions-must be _answered-- re cling- the firid-
ing=thati training in:letter EnaMe-s-_ "and - _sounds cilitatet =learning to
read-words:- 1), Is it lettetnamesietter sdUri_ ottheit Coinbination
whidt.-facilitatesi teading-aCquisitidn?:12- the _correlational find--
ing between lettetriaines-iand 'iouri and be--an-_artifact- or
product of some otherifactotS_Ohnmacht (14)-Med'aciaitroomtetting
to study -the effect_Of lettetname andisMind4raining-on reading. One
-grr,up_ wai-giiren_e2rly-,:training=iti letter names A-lecOnd-gro-up was
gi.,-en-training in -names= and=sonndi;=and 4-third- group- terVed as a
cc ntrol.- She found= that the group- getting training on names- and
sounds was s-uperior to the othet-groups--in- word knowledge and
word discrimination. The group training in letter names was
no better than-the control-on these reading- measures. It appeid, then,
that letter name_trainingin an experimental study does not facilitate
reading acquisition._ _=_

Samnels was interested in the Same-question. He did a laboratory

study (18) to deterthine what component_ of letter name knowledge,
if any, facilitates reading acquisition. One Of the-explanations Offered
by educators as to why they belieye letter name knowledge facilitates
learning to read is that many letter names =are similar to the letter
SAMUELS -

sounds: 4t -is possible, however, that reading acquisition may be

influenced by the ability to visually discriMinate one letter _from
another and not by-knowledge of:the letter names To-answer these
questions, three groups of thildren-midway-through first grade were
used. The visual discrimination group was given= a_paired-assoCiate-
task in which_ the subjects had:to visually- distriniinateloUr artificial
letters from -- each --other. The lettername_group-Was _giVenTa-'paited-
associate task with -the Smile four-- letters-=bdt subjects -had -to: learn
letter--nantes--for each of the letters(s,--,'E, A), The control group
got- an-=irrelevaiw-Talred=assottatetask-._, -hen;Jhe_taine===trani rtask-
Ivas,giverirwalLthegroups,_This:tatkEtonsisted-At&learriiri
appropriate-EngliShz-wordifor=0Ordszoilstrutted r_o_nt= of -t_ e=artifitial_
letters- (sE_ SA tsSAI;,ME=_-!ME,-7_11A-:=-_s--MAY 1SuTprisinglyiiio
Significant -differentes--were-sEfound:=AmonaitrOV e
fixiding--ran counter= to_the-torrelational--=fitidin 0 :Add
replicatedt twite;---With-tliffetent_labbratorr Fassistants Arid-5-_-different
&it grade -subjectt,r]b-ut=alivays-,with=thelialil
ferente among =

Ites-uks: from-- the -Ohntnatht --anti-Samuels stndies-sitnggeSt --that

letter name -knoWledge=has,tfo-lositiVe effect-on---readingiacqUitition-
int:- that the= correlational firitlings--betweenletter-naine-iknowledge
and reading-may -beckproductolsomeotherfattor, There- ieviderite--
(25) that rpaired-associate learning7ability- is significantly correlated
wi th_ Intelligente.- Letter naming iisa_paired-asSociate task- and'inay
be taken _as-an index- of intelligence. Sifice we -already- know= that in
the -elementary school I04- is -highly correlated-_-with -readingachieVe-
ment, it-is not surprising that letter name-knowledge is also_ torrelated
with reading achievement, Another- explanation- is that -the kind of
home background_ which enables-a thildto enter first grade-already
knowing many of- the letters of the-alphabet -would- be-the-kind of
home in which academic achievement would be emphasized._Again,
it is well known that socioeconomic status and home--environment
are highly correlated -with school-achievement. --
Although- letter __mine knowledge does not seem to have any
beneficial effect on reading, there is evidence that_letter sound -irain-
ing does have a positive effect. The=Linehan and Ohnmacht studies
both su ested this, and- a study by Jeffrey and Samuels (11), which
will be discussed later, gives further evidence of this._
 THEORETICAL MODELS AND PROCESSES_ OF READING

_
Classical Research -Which Has Influenced
Current Thinking
The research since the -1960's indicates that children tend to
select a detail such as a letter as_ a cue for word recognition. This
finding: is at-variance with the more commonly= held= belief that
children-_userhe whole word or word the cue. How-did this
latter view ongrnate?
-Prior to: 1900;- Cattell,- And -1
e-zettrrenCere_
whole words livor --=_Aticressitri*Or -reco
*art, partly_:_tonsU o
ooks on reading=have continued to=refer tti; these studies Because of
=

--theit=iniporiance;= can-=, ei-

ShOuldifb-e=pointe o
Erdmann and ge- t worclen
ari&Shapefiwire:the-s rima ille rea
-recognition.17They=came=to this =s-conclusion After=firidia--- skilled
readets:-_couldirecognize_' wor at=were =to e fixation
poinr_that7iirdiVidtiat-lettert-Etrin- -7riot7 ret ize i-fan
could bcrecognized -evenw _en: eteersmereitouSnia =to exec-opt/6d
indiVidually. _ in 7-
It =seems= fallacious- to-- assume -_that--because- skilled readers can
recogniit a --WOrd-__-from--
itt_shape-And=:length=rinder-eirperintentall
conditions when lother-Cues7are:mitiing-7-4hat shape- andilength -are
the priMary cues_ adul tac: rely-On-nunderrnorinal-toriditiorit Secondly,
if Erdma-nn
_
Dodge are correct- abOut adults rising:word- shape
and -length-=as= the priMary Criet,-Which=lis :doubtful,- it cseeMs incorrect
to -assrim-e,-atmanyledUcators do, that these-ate the main cues children
use in learn_ing__ to rea-d,'Marchbanks and :Levin (12) :dernonstrated
_

that shape was the_ cue least_ used -by- children. Furtherindre,- it is
apparent- -that- a strategy-of learninglo -read using word shape and
length provides a poor-basis for transfer to-reading new words.
Cattell (6), in 1885,=pUblished a Study-which led to the present
belief that beginning readers use the whole word- in word recogni._
don. The-major finding in this study was that pnderatould recognize
a short common word in-slightly less time than it-takes to recognize
a single letter. There are several flaws- in this study which should
make the reader cautious about concluding that the results of this
study apply to children learning how to read. In this study Cattell
used a small number of highly educaied adults. He had them read
aloud as quickly as possible a= passage from Oulliver's Travels, spell
the letters contained in the words, and then read a passage = consisting
of 100 common nouns. The most serious error was that the time
to pronounce the words is confused with the time it takes to recognize
the words. Secondly, he used only skilled readers, and consequently,
the findings =-are- not valid for children.
In ttel lis
thus eliminatin lem in _the - other study where
the time to ronoMice e =word con un e wi e tfine to
recognize e =word: oun ex ure thne, two
unconnected letters or two unconnected wor could _recognized.
in he used adults but he nientiOn one- nine - year -old -boy in
the study Who was descr as being su "or in reading ability to
some of the adults.
The Cattell studies demonstrate= that = skilled teaders do not
engage in letter-by-letter processin en the time for
recognizing a-word would be the - sum =of=the time= necessary_ for
recognizing each of the letters. Manypeopie have interpreted Cattell's
results to mean that a skill er uses the entire word as the unit
for recognition. = =Tfi s interpretation =is not valid because Cattell's
. -
experiments were not designed to answer the= question= of what cues
are actually used by skilled readers in word recognition. It is possible
that Cattell's readers recognized_ just some of the letters in the word
and were able to correctly identify the word from a partial percept.
To infer = from these studies that naive readers use the entire word as
a cue in learning to read is an error, partly because naive readers were
not used in these studies.
Second, it is now known that naive readers tend to select a detail
rather than the entire word. Third, while-it is known that the adult
can perceive several letters together as a unit in word recognition, no
one knows at the present time when beginning readers perceive these
higher order units.
A higher order unit is a spelling pattern having invariant spell-
ing-to-sound correspondence. For example, a higher order unit might
be gh in words like rough or tough. Adults can recognize higher
 34 THEORETICAL- MODELS AND PROCESSES OF READING

order units which conform to English spelling rules even when they
are presented in nonsense words (9). The critical question is How
do beginning_ readers learn the higher order units?
In =order to study how beginning readers learn higher order
units, Gibson et al (10) gave kindergarteners and first graders a task
in which it was possible for them to learn patterns =of spelling. The
child was given -a set of eight cards.-=Four= of the cards =had words with
a higher order unit such as LACK, MUCE, DECK, an SOCK. e other
four car a wor such as = LA ttrcFr EEZ and so x -will no
gher order unit. The attls were- esen
r rzE), and the child= simply -point o one of the =cards -If he
pointed _to e card _wi
cx,he=was-told drat he e =to consisten
point to -flie correctca e
critninatin er uni was i
Gibson n at for some to nce-im ove
indicating a were learning nate an a tract
_incriminate
the common s llin
The- final topic w ich iscussed _relates to the = findings of
a= laboratorystudy testing the a ect o onic versus-look-say reading
training on transfer to tea in -newAvor n is study (11) kinder-
garten children were given = phonic - blend- training and= then were
randomly assigned to a = look- say, phonic, or control group k-say
training consisted of learning to-rea a list of words: e = letters
of these words= were = used in_ new combinations to rm-the words
used in =the transfer list. Phonic training consisted of learning letter
sounds. These letters were= used in the transfer list of words. The
control group got an irrelevant task to perforin. Following training, -
all the subjects were given the same list of-transfer words. Pint the
subjects were shown the words and were asked to read them-without
any help. Then they were given instruction and the number of trials
required for learning the entire = list was computed. The results in-
dicated that the phonic trained group was significantly better than
the other two groups -in number of words read without any help
and speed of learning the entire list. There was no significant differ-
ence between the look-say and control groups on either o_f these
measures, indicating that= look-say training did not provide a basis
for positive transfer to reading new words. A similar study (2) was
 Cues_ Used in Learning_ to Read," Journal of Edutational Psychology, 57
(1966), 337-340.
Samuels, S. J., -and _M.-_C. Wittrock. "Word Association Strength and Read-
ing-Acquisition." JournOl of Psychology, in -press. ._

Singer, _ H. "Conceptualization in Learning-co -Read," New---Frontiers in

_

College -AdultnReadirig, The -=15th -Yearbook, Nacional Reading Conference.

Milwaukee:- Nacional-Reading :Conference, 1966,- -116.132i =

Staats,s_-kW. LearningkLanguaga,-ind,COgriitioni_ Nei/ York: Holt, Rine-

harfind Mitistop,- 1968.1_ :-= _ =

25, Stevenson; and-=Correlatev in-_ Children's

Learnintandi_TrObleinSOlVing,",-,Motiographt-or-the=Society-ffor=-Researth=
= -======= =

26:-iaticrC,=-Tinker, -Teacking-Elementary=Rea4ing.--llew-York:
_ =

Appleton -Century- Crofts- 1962:===

27: Traiiirs-,=_ -111;-_C,_ Vettelittial"_Leafitine Review of Educational Research,__
AfteriCaii-Editcaticinal=Research,-1967.=-
Reactions to Modes of Word Recognition
JOANNA P. WILLIAMS
University of Pennsylvania

PROFESSOR SAMUELS has reported on a topic that is of fundamentai

importance:_ = Whatever one's ultimate definition- or criterion = of
"reading" = may be, word recognition -Must clearly be included as
one of the primary and major component skills. _The strategies that a
reader -actually utilizes in word recognition are-far = from == explicitly
delineated at the moment. Since = the question- is central to any
genuine understanding of reading, however, our current hypotheses
and rather tentatively= held = conclusions ought to be given further
serious examination.
Samuels has provided us with an = extensive list of the various
cues and strategies that may be used in word recognitionword
length, = initial letter, overall shape, contextand he has done an
admirable job of explaining how each one of these might-be em-
ployed as the effective cue by a beginning reader. His warning that
"children select the easiest cue for recognition" and that the easiest
cue is sometimes an "incidental detail" is very well taken. Moreover,
one should not expect =that other, more relevant cues will also be
picked up by the child. One of Samuels' own experiments illustrates
this point. Subjects learned to read= words printed in color more
easily than words printed in regular type; however, when color cues
were removed, they could not recognize the words that had previ-
ously appeared in color. In this case, the color cue was so salient that
it overrode the relevant cue. Taber and Glaser (9) did a similar study
in a programed instruction context, in which children learned 'a
paired-assbciates list of printed color names. The actual color
was the supporting cue and was faded out gradually over the series
of learning trials. Again, in this experiment, there was little evidence
that the children had in fact responded during training on the
basis of anything besides the color cue.
One of Samuels' major points was that in developing training
strategies, one must often choose between a criterion of speed of
38
WILLIAMS 89

initial learning and a criterion in terms-of-the ability to- transfer to

new, untrained material. Surely we would not be-satisfied -if our in-
structional- .gains were focused =Solely on the materials- actually -en-
countered in training. Samuels' data, 'indicating that:initial -train-
ing on a word _lisrof leads to the development_
of a-better- transfer strategy, are:. most= convincing. It appears -to -me--
that the - development of an effeetive---teaciiing__ strategy MU-st=-13e---one-
that_ is based- on:good answers to two questiOns: -1)=What--cuei Must-
the ---ch attend-= iti,order_=:=ta-'s olve the-cri terion --task*hith we
assume - will =be aitianifee-tatk:;4fid :1Y-how-caw wetensure-thatile-
Willattenthothosecue
also appreciate 7having Samnele-idescription_= art& etialuation of
the -_ early -work-done -by=_Gattell,-andfEby-_-Erdinanni:and-iDodge.- It_ is
well to be -reminded iso_i-than_y=COn-
ChiSiOnSi:and implications -have --136eliTittawni-=-actuaily-corisist
is important--_ to note,-: as --Samuels=points--:=out,---that--:a-='dernonStration
Of What a skillecfreader- ddesi.S_Ilti-tifecestatily: What 4-_-beginiterWill_
do. =But an-even more important tpoint,-:_-Lbelieve,
subject=naive--Or sophisticated-rabout-readingeafir__-doi--under--_'ter-- --
lain specified,_ artificial; arbitrary _laboratory-=conditions-,-- is- -not== at
all the same-as what _thatsubject --does do in am-ordinary- reading
situation. To- confound these two= questions Is very- misleading, and
it is tad-to -see:how often there -has_l_ beeri-_this =corifuSion. Not -_that
-=
this confounding- is limited toyesearch-_ in-leading, of -Course:- the
literature in- child development -similar_ instances in w_ hiCh
estimates- of normative or -average performance are -confused with-
data on maximal performance. _

I should like to turn now- to a dittussion.Of additional_data that

are -relevant to our discussion of word- recognition. -Before a child
can recognize a word, -he must have learned 0 sonic degree-how to
differentiate_ the written symbols, or graphemes; that he finds on the
printed page. Gibscin (.5) has- suggested 'that- improvement of- visual-
discrimination depends -on learning the -- distinctive featuret Of the
forms to be discriminated, -i.e., those dimenSions 'of difference that
distinguish the stimuli. Precise specification- of the critical- features
of letters of the alphabet, of course, will-be a difficult=task.
It is obvious in any case that the graphemic characteristics of
the word provide an important category of cues in word recogni-
 40 THEORETICAL MODELS AND PROCESSES OF READING

tion (12). One of my students, Margaret Ackerman, and I (1) did an

experiment concerned not with the nature of the critical features
themselves, but rather with the_ effectiveness of different training
methods in ensuring that attention is focused on the features, what-
ever they maybe.
In this experiment we compared simultaneous and successive
'discrimination tasks involving highly similar letters (b and d) and
dissimilar letters (s and b). First grade urban children were used as
subject& In simultaneous training, the two_ letters were presented at
the_same_time, and_ on every trial, the subject_was reinforced -for-
._respondirig to the same = one of these stimuli,_regardless of its spatial
position. In_ the successive discrimination problem, on the other
hand, only == -one letter = was presented-_on- each, =trial. The = subject
learned to press the left of two,response-butfons when b was pre-
sented, and the right button when s was presented. =The response
measure was the number of-trials required to reach-a criterion =of
nine consecutive correct response& We found that = with = highly simi-
lar stimuli (b and d), the-successive problem was less _diffic_ult= than
the simultaneous problem; while with -dissimilar stimulL(s and b)
the successive problem was more difficult. Because these findings
were unexpected, we replicated =the studyand got the same re-
sult& Our notion is that for the s-b comparison, distinctive cues are
easily identified (such as size), and so, in-simultaneous training, the
subject has from the start some basis for comparison and diff- rentia-
don. However, b and -dare are notoriously confusing to a first gradef,
and the cues that are to be used in the solution of this discrimination
problem must be developed during the training. It is possible that
simultaneous training in this -case= presents the subject with so much
information at- one time that .the identification of some critical
feature which can be used in comparison will be- hindered.
In a follow-up study, we found- that- the- same results held on a
paired-associates task in which trigrams were paired with color
names. This task, we felt, simulated the decoding part of the reading
task, in which the child must learn to associate phonemes with
their graphic representations. An approximation of the simultaneous
discrimination problem was attained by the use of the trigrams bad
and dab, which would -be paired, for example, with red and blue.
In the successive training condition, bab and dad were the critical
trigrams. Again, successive training proved superior. We are specu-
WILLIAMS 41-

lat: lg on the implications that might be drawn for the development

of workbooks and other practice materials.
Now let us turn from comparisons of training methods to more
descriptive data. Samuels mentioned a study of Marchbanks and
Levin (8) which assessed the relative importance of several graphemic
cues in children's matching responses. These investigators asked
middle-class children in kindergarten and first grade to select, from
a group of three letter and five letter "pseudowords," the one similar
to tharwhithhadititrbeen expose o eicr fri om
sight. They found thatthe specific letters are much more important
in determining recognition than is the overall shape of the word.
The beginning letter was-the most salient cue, followediby the final
letter.
Ellen Bluniberg, David Williams, and I (3) did the same type
of study with disadvantaged urban children to see what true non-
readers _would do. Our kindergarten sample had had no formal read-
ing training and had little or no knowledge of the alphabet. We
found that these children showed no preference f, r any of the cues;
they matched on-a random basis:Our _first grade sample, which had
had some reading training, behaved exactly as=did MarChbanks and
Levin's middle-class children. -That is, individual letters, especially
the initial letter, provided- the important cues.
What implications are there here for, instruction? In view of
these and other findings, there really seems to be no justification
for developing instructional Methods-or primer materials based on
the use of overall shape as the primary cue. Shape seems a poor
choice after reading training is begun, for when children know the
alphabet; individual letttrs become quite salient. Moreover, if one's
instructional strategy were to attempt to capitalize on .tendencies
seen before any instruction is given, shape would be a poor choice,
for there was no tendency- at all on The part of our nonreaders to
utilize this Cue.
We also tested some adults on this task. The data were quite
different from those of the children; the adults' choices were much
more complex. Surprisingly, although the task as presented seemed
wholly visual, half of the subjects reported some use of an "aural"
strategyrhyme, for the most part. The other half reported that
they used a purely visual strategy, and again surprisingly, -half of
these described their strategy as One in which they tried to match
 42 THEORETICAL MODELS AND PROCESSES OF READING

on the basis of overall shape. This had been the least salient cue
for the children.
These results remind us once more that ;t should not be
assumed thai adults and thiiciren behave in the same manner on this
type of visual matching task. Indeed, it -is most interesting to note
the fact that the mast widely used reading method over the past
thirty years (the look-say or whole word) has stressed identification
of words on the basis of overall shape and configuration. But it is
rchildren-wIro-sonretimes-show-this-strategrin-word----
recognition.
'The second general basis -for word recognition lies in the re-
lationship between the graphic characteristics and the nature of the
spoken -language. Early work focused on the correspondences be-
tween individual letters and the sounds they represented. However,
English spelling is quite irregular wh-en individual letters and sounds
are considered "Auk; fact suggested-that thinking of correspondences
in term_ s of single letters was not useful, which led in turn- to- a
search for a more appropriate unit. It appears_that clusters of letters
do have more stable relationships with sound patterns, and it has
been suggested that these "spelling patterns" are critical units for
perception (5). That is, while the letter c alone is not sufficient to
signal a consistent pronunciation, initial ca or ce does correlate well
with specific phonemic patterns.
Eleanor Gibson's hypothesis (5) that spelling-to-sound invari-
ance accounts for the fact that "pronounceable" pseudowords such
as glurch are more easily recognized in a tachiStoscopic exposure
than are unpronounceable pseudowords such as crurgl has not held
up, largely oti the buds of her own research showing that deaf chil-
dren behaVe in the same way as normals (6). This leaves us with
the question of what it actually is that makes certain "spelling pat-
tern?' or letter clusters more easily recognized than others. Perhaps
it will turn out after all not to depend heavily on the spoken lan-
guage.
Here may be an appropriate point -to bring up another experi-
ment that dealt with methods of training. Sometimes assumptions
have been made and accepted wholeheartedly with little or no ob-
jective data to support them. Such is the character of the recom-
mendation, made by Bloomfield (2) and echoed by almost all other
proponents of linguistics reading methods, that only simple, single
WILLIAMS 43

grapheme-phoneme correspondences be presented to the beginning

reader. Evidently this untested assumption seemed commonsensical
enough to go unquestioned until Levin and Watson (7) proposed
an equally commonsensical point of view. They argued that a child
must learn that there are variations in correspondences in English,
and that if he is presented with multiple correspondences early in
instruction, he will be more likely to develop a useful problem-
solving approach to the reading task (that is, a "set for diversity").
These contradictory points of view were tested in an experiment
in which fifth and sixth graders learned multiple correspondences
in a modified paired-associates paradigm (10). The two methods used
were: successive, where only one of two phonemes which mapped to a
particular grapheme was presented for the first half of the training
trials, and the other phoneme for the last half of training, and
concurrent,- where both phonemes associated with each grapheme
were presented on each of the six trials. When they were asked how
many sounds went with each form, subjects were much more ac-
curate at identifying as multiple correspondences the ones that had
been trained concurrently. Presumably, then, in attempting to read
new words, children would more readily identify such graphemes
as multiple and so try out more than one phoneme for them, thereby
making it more likely that they would succeed in reading the word.
Additional support of the "set for diversity" hypothesis came
from analysis of the errors on another test, in which the subject was
to identify all the phonemes that were associated with each
grapheme. Here, there were fewer omissions on concurrent items
than on consecutive ones. These effects- were large, and they re-
mained over two variations in the procedure (which equated first
the number of presentations of each type of correspondence and
secondly the strength at the end of training of the two successive
correspondences).
A second important question, of course, is how much of the mate-
rial presented in training was actually learned. In terms of this sim-
ple performance criterion, there were again differences in favor of
concurrent training, but here the effect was small and unstable.
Further work indicated that first-graders, tested early in their first
semester of reading instruction on a similar task (in which the stimuli
were homographs), showed a similar pattern. '
Admittedly, much more research is needed in order to apply
 44 THEORETICAL MODELS AND PROCESSES OF READING

such findings as these to actual instruction. For one thing, few pro-
grams teach letter-sound correspondences in isolation. Further work
is in progress, focusing on spelling patterns presented in the con-
text of words. At present, however, in the absence of sufficient data
on which to base a final decision, it would seem reasonable to provide
at least some variationsome kind of concurrent trainingwhen
presenting multiple grapheme-phoneme correspondences. Certainly
the data indicate that critical questioning of some of the time-
honored assumptions and recommendations is desirable.
There is a third general category of, cues that is used in word
-recognition: the context in which the word appears. That is, a reader
makes use of the information contained in the rest of the sentence
in his efforts to recognize the word. For example, in the sentence "He
. walked toward the . . . ," there are certain limitations,. of both a
grammatical and a semantic nature, on the words that can complete
the sentence in an acceptable fashion. Investigators-have =found that
it is indeed easier to recognize a word when it is presented in context
than when it is .presented alone. For example, Samuels has demon-
strated that recognition speed for the response item in a word pair
can be facilitated or interfered with by an appropriate selection of-
stimulus items.
More emphasis is being placed on context, or, to state it another
way, the search for cues when units larger than a- single word are
considered. Only a few years ago, psychologists were comfortable
with a very simple and narrow definition of reading, one that con-
centrated on what was distinctive about reading. Such definitions put
major emphasis on the decoding process, of course.
Today the focus has shifted. Decoding is necessary but not
sufficient, and other aspects of "reading"notably, of course, com-
prehensionhave been attracting attention (11). The emergence of
such interest undoubtedly reflects the very strong influence of
cognitive psychology. Reading now tends to be tied to information-
processing and other related concepts. Definitions also seem to be
growing more general and less focused on what is unique to reading.
One can reasonably describe skilled reading, I believe, as a process
in which the reader samples the cues on the printed page. Using
these partial cues together with previous knowledge both about
printed pages and about the world, the reader forms hypotheses (or
WILLIAMS 45

expectations), which are confirmed or disconfirmed by subsequent

samplings.
It is clear that whatever the definition of reading, the processes
involved for the proficient reader and for the beginner do not com-
pletely overlap. Language has a hierarchical structure, and whether
we are' considering the processing of speech or of text, the identifica-
tion of the "processing units" is of prime concern. It seems obvious
that there are different types of units and different levels of units
which depend in part on the characteristics of the-readerhis pro-
fidency, his purpose, and so forth. (They also depend on the nature
of the material to be .processed.) The specification = of these units,
intraword and beyond, how they are structured and how they are
identified and recognized, will continue to be a central focus of the
research in this area.
As would be expected, research interest has been shifting to
the adult, skilled reader. Certainly, many of- the issues arising from
cognitive theory are more. pertinent -to the reader who has some
proficiency, that is, an individual who is past the point at which
decoding skills have become fully established and automatized. More-
over, a good bit of work on the visual recognition processes has al-
ready been accomplished, and it is to be expected that there would
be a major push to do the same kind of initial "mapping out" with
respect to the other aspects of reading.
I am concerned -about this shift in interest, in that findings
may not relate very closely to the behavior of the beginning reader,
let alone to any ultimate concern with- instruction in- beginning
reading. Other recent interests in the psychology of reading also
seem to _be leading away from questions of literacy training. The
notion, for example, that speaking and writing are both derived
directly from underlying linguistic structures, as opposed to the
more common view that speech is primary and that writing is simply
(or complicatedly) a representation of speech, is most provoca-
tive (4). Such questions are tremendously fascinating and obviously
deserve study. But, again, what about the five year old? How will he
gain from this kind of research?
We cannot afford to lose interest in fundamental investigations
of the behavior of the beginning reader, investigations that focus
on the basic processes of word recognition that are clearly so crucial
 46 THEORETICAL MODELS AND PROCESSES OF READING

to the five year old. More work of the type that is so well represented
by Samuels' experiments is sorely needed. .1

REFERENCES
1. Ackerman, M. D., and J. P. Williams. "Simultaneous vs. Successive Discrim-
ination Trainin& as a Function of Stimulus Similarity," paper presented at
meetings of the American Educational Research Association, February 1969.
2. Bloomfield, L., and Clarence L. Barnhart. Let's Read: A Linguistic Ap-
proach. Detroit: Wayne State University Press, 1961.
3. Blumberg, E L., D. V. Williams, and J. P. Milliams. "What Cues Are Used
in Visual Word Recognition?" paper presented at meetings of the Eastern
Psychological Association, April 1969.
4. Francis, N. "Commentary on Chapters One Through Three," in H. Levin
and J. P. Williams (Eds.), Basic Studies on Reading. Harper and Row,
forthcoming.
5. Gibson, E. J., et al. "A De_ velopmental Study of the Discrimination of
Letter-Like Forms," Journal of Comparative and Physiological Psychology,
55 (1962), 897-906.
6. Gibson, E. J., A. Shurcliff, and-A. Yonas. "Utilization of Spelling Patterns
by Deaf and Hearing Subjects," in H. Levin and J. P. Williams (Edi.),
Basic Studies on Reading. Harper and Row, forthcoming.
7. Levin, H., and J. Watson. "The Learning of Variable Grapheme-to-Phoneme
Correspondences: Variations in the Initial Consonant PoSition," A Basic
Research Program on Reading, U. S. Office of Education Cooperative
Research Project No. 639. Ithaca, New York: Cornell University, 1963,
8. Marchbanks, G., and H. Levin. ''Cues by which Children Recognize Words,"
Journal of Educational Psychology, 56 (1965), 57-61.
9. Taber, J. I., and R. Glaser. "An Exploratory Evaluation of a Discriminative
Transfer Learning Program Using Literal Prompts," Journal of Educational
Research, 55 (1962), 508.512.
10. Williams, J. P. "Successive vs. Concurrent Presentation of Multiple Gra-
pheme-Phoneme Correspondences," Journal of Educational-Psychology, 59
-(1968), 309-314. 1
11. Williams, J. P. "From Basic Research on Reading to Educational Practice,"
in H. Levin and J. P. Williams (Eds.), Basic Studies on Reading. New York:
Harper and Row, forthcoming.
12. Williams, J. P., and H. Levin. "Word Perception: Psychological Bases,"
Education, 87 (1967), 515518.
Models of Perceptual Processes in Reading
JOHN J. GEYER
Rutgers University

FEW AREAS OF KNOWLEDGE offer more promise of providing new

insights into reading instruction and remediation than the consider-
able advances made in the psychology of perception during recent
years. The decade between Broadbent's Perceptionand Communica-
tion (20) and Neisser's Cognitive Psychology (116) saw growing
numbers of researchers publishing in newly founded journals and
coalescing into a new branch of psychology. Cognitive psychology
takes as its central function the -study of the processes by :which:
44
the sensory input is- transformed, reduced, elaborated, stored,
recovered, and used" (116:4). Since this function on the visual-verbal
dimension could serve as a definition of reading, it is clear that this
branch of experimental psychology= bears close watching by anyone
interested in understanding reading phenomena.
As early as 4964, Holmes and Singer used the trend toward
more basic research in reading as-the central theme of their review
artide for the Review of Educational Research (89) and in it specifi-
cally pointed to the importance of the lines of research which were
later to make up -the area of cognitive psychology. In 1966, the
present writer analytically reviewed this research literature and
from it suggested a model of perception in reading (62, 63). The
model and experimental evidence supporting it was reported the
following year to the International Reading Association (64).
This paper will extract from the growing literature in cognitive
psychology several studies which can offer insights into the reading
process. Classical reading theory will be briefly presented in its
historical context and the assumptions underlying this theory will
be- stated. Those assumptions will then be examined in light of
recent research and alternate models suggested. While a rich array
oLdetails are currently being actively debated, there are large areas
of agreement with which tosupport a generalized model as a replace-
ment for classical theory.
47
 48 THEORETICAL MODELS AND PROCESSES OF READING

Classical Theory
Historical background
The span of attention. Central to- the classical theory of per-
ception in reading is the hypothetical construct of the span of
perception, a concept which stems directly-from the span of attention
experiments- of early psychology. The second oldest- experiment in
experimental psychology, the span of attention was a lecture topic
in- the classes of Sir William Hamilton-(73:176-177) prior to 1859:
If you throw a handful of marbles on the floor, you will find
it difficult to view at once more than six, seven at most, without
confusion; but if you group them in twos, or threes, or fives,
you can comprehend ai many groups as you can units, because
the mind considers these groups only as units, it views them as
wholes, and throws their parts out of consideration.
Sir William's marble-throwing demonstration was developed by
Wundt, Cattell, and others into precise experimental piocedures
which became central to the theoretical disputes of the turn-of-the-
century. In Germany, Wuait and his pupils Dietze, Cattell, Hill,
and Jastrow worked to establish the limits of- the Umfanea problem
which the latter three were to bring across -the Atlantic with them.
Exner experimented on the "primary memory" and Richet on the
"elemental memory?" In England such giants as -Galton, Sully, and
Bain were concerned with the problem. Their techniques and
theories were reflected in France by the work of Billet, Henri, and
Simon who were the most successful of all at applying- the new
techniques to the new -field of- mental testing. To this latter group
of scientists, the measurement, of the span of attention was a method
by which they- could investigate and demonstrate individual differ-
ences.
In the United- States, the leader of the Wundtian school of
Structuralists was Titchener, assisted by, a host of students which
included Erdniann, Dodge, Huey, Hylan, and others. Their ap-
proach to psychology.was to determine the elemental units of the
mind, chiefly through introspection, and to build these units into
larger and larger blocks in a direct analogy to chemistry. Since the
span of attention appeared to remain constant under a variety of
GEYER 49

conditions, they considered it to be the elemental unit of attention.

Humpstone (91), for example, defines the span as "the ability to
grasp a number of discrete units in a single moment of attention and
to reproduce them immediately."
Galton's influence in the United States was represented chiefly
by J. Mc Keen Cattell and the Functionalists. Concerned primarily
with individual= differences = and mental testing, the work of this =

theoretical viewpoint was quickly and naturally applied to -the

problems of education. When Ja-val (93) established the fact that_the -
eyes in reading do not sweep, but move in a series of saccadic jerks,
the relationships between the rapid succession of discrete visual
images provided by the fixations and the brief tachistosCopic image
utilized in the man of attention experiments were recognized. The
tachistoscope became -the standard experimental instrument in the
search for the primary cue in word recognition.
The "total word picture" concept. One of the earliest and most
seminal experiments in the field of reading was conducted by Cat --
tell (31). The findings and implications of Cattell's experiment and
its influence on our thinking today can be shown by a brief quotation
from the definitive work of Woodworth and Schlosberg (149:101).
The use of the tachistoscope in experiments on reading goes
back to the pioneer work of Cattell (1885), who found that with
a very brief exposure under adequate illumination:
3.4 unconnected letters could be read.
2 unconnected short words could be read._
4 connected short words could _be read.
Since even the two short words, not to speak of the four-word
sentence, exceeded the letter spanand since, as Cattell found
in his reaction time experiments . .. a familiar short word could
be read as quickly as a single letterit was clear that the words
were not read by spelling them out. What could be the reader's
basis for recognizing a word? Cattellyeasoned that it must be the
"total word picture." His results were confirmed by Erdmann
and Dodge (1898), who found the span for unconnected letters
to be 4.5, while familiar words, even as long as 12-20 letters,
were correctly read from a single exposure of 100 ms. They
inferred that the "general shape of the word" must be the
primary cue for word recognition, with a few clear letters close
 50 'THEORETICAL MODELS AND PROCESSES OF READING

to the fixation point as supplementary cues. By "general word

shape" they probably meant the external configuration of the
printed word, while Cattell's "total word picture" covers also the
internal pattern of curves and vertical strokes . . . .
These conclusions of Cattell and of Erdmann and Dodge, al-
though much disagreed with, remain the basis of much of the ratio-
nale behind our views of perception in reading today. The "span of
attention" has become the "span of perception" when applied to
reading, but the basic concept and the assumptions underlying it
remain the same, viz., that it is-a- single, unitary act= of perception,
that all elements of the stimulus field are =perceived instantaneously
and simultaneously, and that those in= the center of the field- are
perceived dearly While the elements become less dear as they occur
farther out in the periphery. As the- elemental unit of perception
in reading is the word- or even the phrase, the letters are important
only in determining the word shape. Hence, the whole word method
of teaching reading (5:212):
The psychological rationale of the whole word method has
been demonstrated numerous times by laboratory studies of the
psychology of- reading. Cattell's study . . . reported hi 1885, is
a landmark .... If the limit for unrelated letters was only three
or four, the words obviously were not pezteived in terms of
letters. The experiment definitely proved that we do not read by
letters but by whole-word units.
Early disagreement with the "total word picture" concept. As
well accepted as the "total word picture" concept is,- most authors
find it necessary to qualify it to some extent due to petsistent experi-
mental findings which appear to contradict it. The studies which
require such reservations are discussed in several textbooks ,covering
perception in reading and will only be overviewed- here. The- ob-
jections to Cattell's concept of the "total word picture" as the basic
elemental cue began early. Zeitler (150) found, through a study of
errors made during tachistoscopic- presentations, that the capital
letters at the beginnings of the words were invariably reported
accurately and that the rest of the word reported bore certain
relationships to the letters within the stimulus word. He, there-
GEYER 51

fore, discounted the influence of word shape and suggested instead

that the subject was constructing a word from the letters he obtained
during the brief flash. Zeitler concluded
The apparent simultaneous apprehension of a word as a
whole is an illusion of assimilation, that the real apperceptive
process consists of successive apprehensions of the word, part
by part, by a movement of attention from left to right, in
which part of the letters of a word are passed over rapidly while
the attention fastens upon_ the dominating letters [translation by
Bonny (16:28); italics added by- present writer].

GOldscheider and Muller (68) had likewise held that the letters
were important and that the "deterniining letters" were the initial
letter, "ascenders," "descenders," and vowels. While certain words
might be read from word shape, they concluded from their-experi-
ments that such reading was not the rule. Crosland and Johnson (42)
attempted to rank the letters in terms of relative legibility and
found their results in general agreement with Goldscheider and
Muller. The determining= letters were thought to arouse an = auditory
image of the word, and-the word was then completed by means of
associations. =Huey (90), in the first comprehensive treatment of the
psychology of reading,- presenteit essentially the same- View and
further suggested that the speed of association was determined by
"expectancy."
Pillsbury (121:349-350) projected words= with omitted letters,
incorrect letters, or an x overprinted on one or more letters. His
five subjects were all mature readers and trained introspectionists
as well; in fact,- one was Titchener. His results indicated that
. . . there is a striking decrease in the percentage of recog-
nition [of disfigurements] as we proceed from the first letter to
the last throUghout the word . . . . This seems to indicate a
general tendency of the subject to read through the word- -from-
left to right (emphasis in- original), and thus to give the -first
letters in the word a more prominent_ part in the recognition of
the word as a whole.
Further evidence.of the same tendency-is offered by the fact
that in the words with more than one letter changed, whenever
 52 THEORETICAL MODELS AND PROCESSES OF READING

[only] one disfigurement was noticed ,. . it was the first altera-

tion that was noticed in the greater number of cases.

By presenting words to peripheral vision, Korte (98) found

that isolated letters could be read farther in the tail of the eye than
could words. He- also found that the longer the word, the closer it
must be brought to -a central point in order to be identified. Since
general outlines are discTiminable in peripheral vision where the
individual letters making up the- word are not, he concluded that
identification of a word = proceeds from identification of its letters
and that the "total optical form" was relitively unimportant.
That shape alone can provide useful cues was demonstrated by
Erdmann and Dodge (52), who found that words can be identified
at distances beyond the maximums for the determination of in-
dividual letters. This, as Woodw-orth (148:741) pointed out, does
nothing- to demonstrate that the same phenomenon occurs at the
normal reading distance. Grossart (70),_Kutzner (101), and Wiegand
(146), utilizing-the distance experimentshowed that as the words
were brought closer, more and more letters became identifiable and
that guesses might be made at any stage. The closer the word, the
-

more letters were identifiable, and these, together with some general
characteristics such as length, provided better and better guesses.
The early experimental findings tending to contradict the
"whole word" view could be extended considerably. That the con-
clusions arrived at during the last century by Cattell and by Erd-
mann and Dodge should have survived so many years in the face
of a large amount of contradictory evidence seems due to two pri-
mary factors. In the first place, these experiments went to the heart
of the extremely bitter theoretical warfare being waged between
the Structuralists and the Functionalists. When the introspective
base of Structuralist methodology was discredited by the rising tide
of Behaviorism, the Functionalists were left with the field. Secondly,
at the very time that the Functionalists needed explanations to dis-
-

count the experimental findings of- the Structuralists, these explana-

tions seemed to be offered by the Gestalt psychologists who enjoyed
an immense wave of popularity in the decades following World War
I. (For a thorough, if somewhat overly critical review of the influence
of Gestalt psychology on reading theory, see Diack, 46.)
GEYER
53

Classical reading theory.and its underlying assumptions

The concept of reading by word shape combined with an exten-
sive eye-movement literature provided this classic view of percep-
tion in reading. The eye moves in a series of saccadic movements.
Little or no vision takes place during the movement, but at each
fixational pause the visual field is flashed to the brain. Words and
phrases within these visual fields are identified primarily by their
general shape, and growth in reading is a function of how many
words the reader can "take in" during each fixation. That is, the
"span of perception" is thought of and measured as a spatial variable,
all parts being "taken in" and processed simultaneously. The span
becomes larger with training and practice because the reader learns
to "take in" and process progressively larger elementsfrom parts
of letters, to letters, to syllables, to words, and eventually to phrases.
It has been argued that it should be possible to perceive a page at
at a glance.
It is thought that approximately 100 msec. are consumed at the
beginning of each fixation in clearing an_ d refocusing time. Other
temporal portions of the fixational pause length are devoted to per-
ceiving and responding to the stimulus. The reading process is con-
ceived of as a series of tachistoscopic presentations flashed to the
brain by the saccadic movements of the eyea series in which both
the stimulus and the response are immediate and based on the whole
word or phrase.
Regressions are generally ascribed to "faulty oculomotor move-
ments," particularly the regression commonly found at the beginning
of the line, or to difficulty in trying to understand what hai been
read. For one reason or another, the reader 'Moves his eyes so that
his "span of perception" does not overlap his previous span, and he
must move his eyes again so as to include the missing words. At one
time there was a considerable dispute over whether "faulty eye-
movement patterns" were the cause or result of poor reading. Most
authorities agreed- that ". . . the number and duration of fixations,
and hence the speed of reading, are-limited by central, rather than
peripheral factors . . ." (119:506). Yet the central factors were vague
and peripheral measures lead to inadvertent peripheral interpreta-
tions. Woodworth and Schlosberg should have written ". . . the
54 THEORETICAL MODELS AND PROCESSES OF READING

speed of reading, and hence the number and duration of fixations

. . . ." The difference is not trivial.
This classic view of perception in reading with its emphasis on
the span .of perception was an outgrowth of span of attention con-
cepts and makes the same basic assumptions. Chief among these
assumptions are:
1. That perception is a unitary phenomenon, or, at least, that
sensory and perceptual processes are isometric. Thus Tinker
(137:478) concluded that ". . perception begins in the initial
part and is completed in the final part of a . . . fixation. These
two parts of a . . . fixation appear to be phases of a unitary whole."
As Hebb (85:26) has recently pointed out, this assumption was based
on the neurology of the turn of the century and had a purely
theoretical origin. It was stoutly maintained because to do otherwise
would have lent respectability to the method of introspection which
had suggested that a number of things may happen between the
objectively-measured
, stimulus and the objectively-measured re-
sponse.
2. That within the "single perceptive act" isolated by the span
of attention experiment, all elements of the visual field are per-
ceived instantaneously and simultaneously. This assumption is a
necessary correlate of the first, for if it were not true, the span- of
attention would not be unitary, nor necessarily isometric.
It will be argued in the remainder of the paper. that neither of
these assumptions is correct and that whatever vestiges of classical
theory remain should be honorably retired. That the first assumption
is false is attested to by a large Cody of evidence and overwhelming
agreement. With the first assumption false, the second becomes the
very complex focus of much current research. In the current vocabu-
lary of the field, this research is concerned with determining what
portions of the perceptual act employ sequential processes and what
processes are accomplished in parallel.

The Assumption of the Unitary Nature

of Perception
The fading memory trace
The memory afterimage. One of the most persistent intro-
spective reports made by subjects in tachistoscopic experiments is
 GEYU 55

that they saw all the elements but forgot some before they could
be reported. This report is so universal and so tenaciously believed
by the subjects, that even its introspective origin was not sufficient to
discard it altogether.1 That =such reports contradicted the assump-
tions underlying the span of perception and "total word picture"
views of reading was recognized, but the implications were - largely
ignored. Thus while Woodworth and Schlosberg (149) tell us on
page 104 that "as children advance in reading ability they become
familiar -with many word pictures, and their span for words- in-
creases much faster than their span for unconnected letters," they
also tell us on page 102 chat
. . . the visual impression received during a= brief= exposure
must be much more complete and detailed than is implied by
the phrase 'general = word shape' . . . . Even when a subject can
report only a few of the letters shown, he firmly believes he has
seen all of them during the actual exposure. He forgets some
before he can report them all because mere memory cannot re-
tain so many disconnected items.
Woodworth and Schlosberg's more detailed view of perception
in reading seems to imply immediate and simultaneous sensing of
all the parts of the visual field, but successive identification of its
component parts, limited in a brief exposure by immediate memory.
On page 507 they describe the reading act as a ". . . continuous
process in that the perceptual development of meaning goes on
steadily. Perhaps one can think of it as a continuous production
process, a machine into which the raw material- is tossed by the
shovelful."
One of the experiments which forced this more detailed view
was that done 13) Glanville and Dallenbach (66). In a definitive study
of the span of attention, these investigators exposed two rows of
letters across the visual field for an exposure length of 78 msec. Light
pre- and post-stimulus fields were used, a factor the importance of
which will be discussed below. Their results showed that the letters

1 Sperling (131) gave the frllowing as "representative examples" of experi-

menters who reported-this introspection: Bridgm (17), Cattell (30), Chapman (33),
Dallenbach (44), Erdmann andModge (52), Glanville and Dallenbach (66), Kulpe
(100), Schumann (129), Wagner (141), Whipple (143), Wilcocks (147), and Wood-
worth (118). The list could be extended considersbly.
fr

56 THEORETICAL MODELS AND PROCESSES OF READING

to the left of center in the upper row were most correctly reported.
Introspective reports given by their highly trained observers indi-
cated that most of the letters were reported from a "memory after-
image." Glanville and Dallenbach (66:225) concluded that:
. . the usual Jrder of report followed by the O's was to name
the letters from left to right, and those in the top row before
those in the bottom. This plan places those in the upper
row at an advantage over those in the lower. This theoretical
advantage is borne out by the objective results.
They also reported that the subjects could change the order of
_report at will.
Very similar results were found by Crosland (40) when he ex-
posed random- letters in series ranging from I to 9 for exposures of
100 msec. A falling shutter tachistoscope =of a type hich gave ap-
proximately equal lighting of the_fields was used His results showed
-that most letters reported =were from the = left field, and that the
curve of correct reports by letter position declined rapidly= from
left to right. The sharpest decline came between the fourth and
fifth positions. Since it was clear that these findings did not agree
with the doctrine of the day which held that perception of all ele-
ments was simultaneous and that vision was clearer in the center
and less clear on the periphery, Crosland explained these results as
indicating a "left-to-right mindedness."
When Crosland (41) extended his experiment to groups of good
and poor readers of elementary school age, he found the left-to-right
effect more pronounced in the good- readers than in the poor but
apparent in both groups when the letters were flashed to the center
of fixation. When the L ord was presented to the left of fixation, how-
ever, the poor readers reported more of the letters at the' ninth posi-
tion. Anderson and Dearborn (5 :281) find this to be evidence that
"good readers are attracted by the beginning of words. Poor readers
not so much so."
The experimental findings were further extended in an unpub-
lished experiment by Crosland and, presumably, Anderson (re-
ported in Anderson and Dearborn, 5:227-228), by repeating the
procedure with bilingual Jewish children. In reading English, these
children showed the same results as had been obtained before, but
 GEYER
57

in reading Hebrew, the results were reversed showing a "right-to-

left mindedness" in reading Hebrew.
That these results were contradictory to the current assump-
tions and beliefs concerning perception was recognized by Crosland,
and attempts were made to explain them in terms of laterality. When
results in further experimentation showed no laterality effects, Cros-
land suggested they were due to a "left-to-right =progression of at-
tendon" (4). The importance of these implications was not followed
up, however, and as is often the- case, it was-only when the issue
becime central to a theoretical dispute that its importance was recog-
nized.
The retinal locus controversy. In his very influential work, The
Organization of Behavior, Hebb (81) chose the operation of vision
in tachistoscopic presentations as a central issue in the theoretical
dispute of functional organization versus equipotentiality. Lashley
(102) had argued that the phenomenon of- serial order could not be
accounted- for neurologically in terms of associative chains, nor in
terms of the sensory control of movements. He felt there must be
some process whereby traces and motor neurons can -be scanned in
sequence. Hebb, noting that speed of reading ". . . depending on the
instantaneous recognition of larger and larger blocks of letters . . .
increases up to the age of 12 to 16 years, or perhaps even later . . .,"
argued for a structural differentiation of the retina due to training
in reading. He cited the results of- an experiment by Mishgin and
Forgays (111) which, in Hebb's- words, ". . . show directly that
reading does not train all parts of the retina in the same way, even
when acuity does not enter the picture."
In the experiment cited by Hebb, Mishgin and Forgays had
presented eight-letter English and liebrew words to bilingual sub-
jects for an exposure length of 150 cosec. The subjects were directed
to focus on the usual prestimulus target, and the words were pre-
sented to the left or right of this- focal point. The English words
were fotind to be more easily recognized when presented to the right
visual _field, the Hebrew words when presented to the left. These
findings were interpreted by the investigators, as by Hebb, as show-
ing differential training for the different areas of the retina with
resulting structural differentiation.
The results of Mishgin and Forgays' experiment had been
 THEORETICAL MODELS AND PROCESSES
OF READING
much more clear-cut for the
so Orbach (118) replicated theEnglish words than for the Hebrew,
experiment and extended it by an
the results for his Hebrew-speaking
which language had been learned subjects in terms of
bilingual Hebrew-English reading first. These findings indicated that
sented to the left field best if subjects perceived Hebrew pre-
learned. Hebrew had been the first
language
Forgays (60) tested whether or
the left and right loci not the differential results from
were due to training and/or maturation
testing 12 subjects in each school by
grade fifteen. He exposed grade from grade two through
three- and four-letter words
or right of fixation for 150 to the left
nisec. His results showed that the
number of words recognized total
increased-significantly with the increase
in. age and/or educational experience
the lowest grades little differentiationup to grade fifteen. While at
fields was apparent, the of the right:and left visual
amount -recognized to the left of
increased only to grade five. fixation
After grade seven, the right field
superiority was clear. These results
were, of course, interpreted to
support Hebb's position of differential
the retina. -training of the receptors of
That Forgays' results were due
not maturation was shown by Kimurato educational experience and
(95), who exposed 24 different
arrangements of-letters and geometric
in age but differing in educational forms to two groups similar
posed of 40 army-personnel with level. The first group was com-
grade eight. The second group an average educational level of
The differential effect was found was comprised of 23 college students.
to be more prominent in the stu-
dents than in the soldiers.
Other findings of this
be discussed below. experiment will
Some years prior to Mishgin and
son (3) had presented English and Hebrew Forgays' experiment, Ander-
lingual subjects. His presentations nonsense words to bi-
were to the central focus, and his
findings under this condition
were the reverse of those found by
*Mishgin and Forgays---that is,
"words" were recognized to the Anderson found more English
left of focus and more Hebrew to
the right. To explain these
results, Anderson postulated a "left-to-
right mindedness" in English-reading
sumably intended something similar subjects, by which he pre-
to his earlier "left-to-right"
progression of attention (4).
GEYER 59

Noting the above- differences resulting between presentations

to the central locus and those to either" side, Heron (86) reviewed
all previous work in the area, conducted` a comprehensive experi-
ment, and collected the folloWing facts: 1) If English letters are
exposed simultaneously to both sides of the focus, those on the left
are reported best; if to one side only, -those on- the right. 2) If the
subject knows where the letters will fall, the knowledge =makes no
difference in the report of the right field, but- improves the left.
3) Whenlour letters are presented in a square== pattern, subjects re-
port the upper left, upper right, lower left, and lower right in that
order, whether the group is to the left or right of fixation. 4) In a
series of letters, subjects report the beginnings better than the end-
ings. 5) When letters are double- spaced, the superiority of the left
field-in equidistant -presentations is greater. Putting -these facts -to-
gether, Heron postulated= -that differences- found =in= the "retinal
locus" experiments could be best= explained as due to "post-exposural
attentional processes" on the remaining neural traces. While Heron's
conclusions were carefully worded to avoid any hint of the Platonic
sensorium observing the eidola, they seem -to postulate what later
investigators refer to as a scanning of the memory trace.
Kimura (95) replicated Heron's findings that when letters are
presented in a square pattern the subjects report them in an 'UL, UR,
LL, LR order and that errors ificrease in that order. When the letters
were spread out horizontally and presented across the center of fixa-
tion, -the less educated soldiers tended to` report= the right-field ele-
ments and, consequently, reported. more of these correctly. The
students, however, tended to keep the left-to-right pattern and re-
ported more of the left elements correctly. Kitnura, therefore, settled
for a compromise position which acknowledged Heron's conclusions
but said that other things were operating as well.
Terrace (136) exposed words and nondirectional forms to the
right or left loci of 30 college students. His results agreed with the
previous findings that under these conditions words are best per-
ceived in the right field, but no difference was found for the non-
directional forms. Terrace agreed with Heron's attentional scanning
hypothesis.
In a series of experiments utilizing a variety of techniques,
Harcum and others (9,7 , 78, 79, 81) showed that readers of English
normally reproduce binary numbers- in a left-to-right order and
60 THEORETICAL MODELS AND PROCESSES OF READING

"exhibit superior tachistoscopic perception for elements at the left"

when these elements are projected.actoss the center of fixation. But
when the subjects were instructed to respond from right to left, they
showed a strong right superiority. When horizontal patterns of ten
circles with five blackened were presented to Israeli and American
students the results (78:368) indicated that
The basic difference between these observed groups ap-
parently is that the American "instructs" himself to perceive
and respond from-left to right, whereas, the Israeli "instructs"
himself =to perceive and to-respond- from right to left. If the
American is instructed by the experimenter to respond from
right to left, he shows right-superiority to the same degree that
the Israeli shows without instructions-from the experimenter
The perceptual process involves a spatial-temporal se-
quence whose directionality is influenced by factors probably
acquired through experience in reading.
Bryden (24) had found results similar to those of .Harcum and
Friedman when using nondirectional geometric forms. Requiriq
subjects to report from right to left-did not produce the correspond-
ing directional effects, however, when letters were =used as the stim-
ulus material. The major effect under these conditions was a reduc-
tion in the number of letters reported. His subjects frequently stated
that they first had to repeat the letters silently in a left-to-right
direction before they were able to report- them in the prescribed
direction. These_results indicated to Bryden that the trace systems
which are excited when letters are tachistoscopically exposed are
"polarized" in a left-to-right direction.
Bryden's work with- differing types of stimulus materials was
extended in experiments by Bryden- and others- (26, 27). These ex-
perimenters concluded that -left -field superiority could be best ex-
plained by postulating a sequential processing of a fading memory
trace but that differences in right -field superiority were strongly
affected by the familiarity of the stimulus. Subjects were able to
process nondirectional forms most flexibly, digits less flexibly than
forms, but with more flexibility than letters. They concluded -that
the inability to process letters flexibly resulted from reading ex-
perience.
GEYER 61

The relative inflexibility with which-alphabetic materials are

processed suggested that further insights into perceptual processes
could be gained by Utilizing various transformations of normal
letters. An early study by Wolfe (1939) reported that a "set to read
backwords" could be induced in 9 to 11-year-old children by present-
ing mirror reversals of words. Harcum and Finkel (77:224-225); com-
bining this technique with the retinal locus_ experiment, presented
English words and left-right mirror itnages of-English words to the
left and right of focus. Whileithe Words -were read best in the right
field, the mirror- ,images- Were read best-in- the left. Harcum and
Finkel concluded that these= results could be thought of as being
due to
. .. a temporal distribution- of _attention across the persisting
physiological traces of the projected Stimulus elements. The se-
quence of attending to the traces is assumed-to correspond to the
sequence of successive eye -fixations across- the visual_field with
the stimulus present, if eye movements could- be made. Thus,
having fixated on the center of =the target-field, in order to per-
ceive the word to-the left of the fixation cross, the observer
Must direct his scan all-the way to the beginning of the word at
the left, and then read-it-fro:1M left to right. This initial= tendency
to scan toward theleft conflicts with the_tendency to read toward
the right and perceptual accuracy suffers. When the word is
presented to the right of- the fixation cross, hoWeveri the be7
ginning of the word is near the fixatiOncross,-and the tendency
to scan from the beginning of the worditoward the end supple-
ments the tendency for the left-to-right scanning sequence.
Harcum's explanation of retinal locus effects being due to
conflicts in scanning tendencies receives some support from the
finding by Goodglass and - Barton (69)` that the superior perception
in the right visual field is- accompanied by a lower threshold. If a
subject must first scan to the beginning of a word, as in the case with
words in the left visual field, it would be expected that such scanning
would require additional time. In two rather extensive follow-up
studies, Harcum and Filion (76) and Harcum and Smith (82) ex-
tended their findings on the effects of various letter reversals and
found further evidence in support of a scanning hypothesis and
against the original structural hypothesis proposed by Hebb. In the
 62 THEORETICAL MODELS AND PROCESSES OF READING

same year, Hebb (85:25-26) rejected his earlier hypothesis and

suggested an explanation quite similar to Harcum's:
what common features do [the word FRANCE and its mir-
ror image] have, as visual configurations, that are not common to
other words: Not much, considering them as static wholes. But
as one moves his gaze from -left to right in -the upper word,
or right to left in- the lower, an identical series of- events oc-
curs . . .
Now suppose that perception of . . . the word FRANCE is
a serially ordered= activity-of mediating processes or cell assem-
blies, even when the object is recog. ized= at a single glance,
and that the orientation is= a function of serial order. What
change is required . . .? A change in the order of firing of cell
assemblies, only, and such "a = change is easily unde.rstood.
What I mean =to= emphasize here are the new possibilities
of explanation that open up when one separates sensory born
perceptual process^s, and recognizes that identifying the two had
a purely theoretical origin, and neurological to boot.
The Heron-Harcum model (1966)
The implications of = the research discussed above, as Hebb
recognized, were far-reaching. By making untenable the long held
assumption that perception is a unitary phenomenon, many old
questions were reopened and many new questions suggested. If
sensory and perceptual processes = are separate, then the response
cannot be used as a measure of sensory input. (In= the reading case,
the span of= perception as generally measured would be less related
=

to a visual span than to complex interactions of, processing systems.)

What interacting systems are involved? How is information processed
through these systems? What variables account for the retinal locus
effects?
Harcum (74) reviewing his own extensive research and the
related literature,. suggested a model which =postulated that sensory
processes established a fading memory trace of the stimulus elements
which were sequentally scanned until the trace faded. To account for
differences resulting from interactions of retinal locus and letter
orientation, Harcum postulated two post-exposural scanning ten-
dencies: 1) a left-to-right tendency learned through normal reading
 GEYER 63

experience, and 2) a tendency dictated by the material or experi-

mental procedure. When the two tendencies are in agreement, as in
words printed normally and presented to the right hemifield, max-
imum performance results. Under conditions =which bring the scan-
ning tendencies into conflict, however, performance suffers: A block
diagram of this model (prepared by this writer) is shown in Figure 1.

f2 SYSTEM PROGRAM

Visual All elements in- visual field sensed

Input simultaneouily.

Storage fading memory trace system of un-

knoWn parameters.

Sequential scanning of memory- trace

Output with- response occurring when suffi-
cient elements have been scanned.

FIGURE 1. Block diagram of Heron-Harcum model.

While the Heron-Harcum model suggested more questions than

it answered, a large body of evidence could be cited in support of
its general outlines. There is wide acceptance today that perception
is not a unitary process and that the sensory and response Phases are
more independent of one another than had long been assumed.
Recognition of stimulus elements continues after the cessation of the
physical stimulus. The sequential order- of perception and, con-
sequently, the response can vary independently of any variation in
the external stimulus. In pursuing the various lines of research
involved in the research on subception (Pierce 119), investiptors
have shown that certain factors long known to influence perception,
in fact influenced the recognition phase only. The factor of famil-
iarity, for instance, known to significantly influence thresholds for
64 THEORETICAL MODELS AND PROCESSES OF READING

visual presentations, has been shown to be entirely a response factor

which fully operated-in the tachistoscopic situation when no stimulus
material was actually presented (67).

The _AssumptionTh at All Elements of the Visual Field

are Perceived InstantaneOusly and Simultaneously
If perception is not a unitary phenomenon, -then it cannot
operate instantaneously and- the elements of the -viittal field are not
processed iiimiltaneottsly. -Visual perception= is: an event which- occurs
over time and entails complexprocesses. While this seems self-evident
today, it has-only-been a decade since the importance of the _Concept
was recognized (20i 130): --Yet; to discard: the assumption of the
instantaneous_and simultaneous nature: of overall perception does
not mean that_=thei assumption could not hold_ for :certain systems
and processes. Almost 41l-expetimenters to date have maintained this
assumption in regard to the sensory input phase of visualperteptimi.
The model-presented in_this paper(61, 62) and-the well-known model
by Sperling (1968) -differ primarilyconcerning this assumption. It is
an essential pOint: if we arerto- understand Cattell's 1885 findings
covering the differences in--processing letters and words with the
attendant implications for reading. Heron's "scanning" hypothesis
was clearly a response scanning-4 sequential processing of respOntes
from an established memory trace. Since he used random letters
as stimuli, his hypothesis was silent on the-question of how_ words
are scanned. Harcum and his coworkers haVe extended the scanning
hypothesis -to cover words- by postulating that each element within
the word is scanned in an appropriate direction. Hebb (85) ap-
parently- agrees with this position. While these investigators do not
specifically discust this quettion, they-Seem to imply that the memory
trace is established by "parallel" sensing and that their hypothesized
scanning takes place sequentially across the memory trace. According
to this view, presttmably; the subject scans the elements (e.g., letters)
of the Memory trace sequentially and responds when he has scanned
all of the elements making up a response-unit (e.g., a word).
Neither Heron nor Harcum- specify the time parameters of
either the hypothetized memory trace or the scanning thereof. If
this hypothesis were used to explain Cattell's findings that more
 GEYER 65

"letters" can be processed when organized into words, we would have

to assume either that the memory trace lasts longer for words than
for letters or that the scanning rate is faster for letters than for words.
The first of these seems improbable since a- memory trace is presum-
ably a physiological function which would not change to suit
differences in the, subjective organization of stimuli. The variable
scanning explanation seems somewhat more probable, but it leaves
much of the data unexplained. For example, it fails to account for
the persistent belief among subjects that they saw= all the individual
letters but could not remember them. Heron's subjects-(86)_reported
that they fixated -each letter in turn, although the tachistoscoPic
_
presentation was too short for any eye movement to = have =taken -place.
If every letter is "fixated," yet= some cannot be reported because
they_ are lost ftoin meniory, then the "fixating " -must= occur= = prior to
the memory trace.
The variable scanning rate hypothesis also is contradicted by
Heron's finding that double-spacing the stimulus elements increased
the differential effects shown in reading from the two visual loci.
This finding seems to suggest =a constant-scanning rate uninfluenced
by the spatial charicteristics of the field, since the farther the stimulus
elements are from center, the more pronounced the effects. The
finding by Goodglass and Barton (69) that words printed vertically
and flashed to the right locus have shorter thresholds than similarly
printed words presented to the left locus would seem also to suggest
that distance and direction from the center of fixation are important
.factors. This, in turn, argues against a variable scanning rate which
presumably= could compensate for these differences.
The Heron-Harcum model, then, =runs into considerable dif-
ficulty where applied to ihe often-replicated findings of eattell (31).
In the model su :4: ested by this writer (62, 63), these difficulties were
met by dropping the assumption of in,stantaneous and simultaneous
input and postulating the operation of two-scanning rates in visual
perception, viz., 1) a sequential response rate operating from a fading
memory trace, essentially as- postulated by Heron (86); and 2) an
entirely different scanning subsystem, similar in operation to that
postulated by Harcum and Finkel (77), but operating in the sensory
input phase and whose function -is to establish the memory trace.
The evidence for this hypothesis will be developed below;
66 THEORETICAL MODELS AND PROCESSES OF READING

Visual image and iconic storage

It has long been known that minimum tachistoscopic exposures
necessary to perception are not -good measures of perception time,
since the incoming information is I _ocessed for some time before
identification is'achieved (140). Any model dealing with visual per-
ception must be able to account for this phenomenon and the fading
memory trace of the Heron-Harcum modetwas one such mechanism.
In a series of ingeniously designed Ixperimenti, Sperling (131)
investigated the = fact= =that__ =observers in- tachistoscopic experiments
commonly report that they have =seen more elernerits than they can
remember. Using dim pre and post-stimulus-felds, he.found that his
subjects reported slightly over four random letters from-exposure
durations ranging from--15- to_ 500 msec. This constant number
reported over a wide range= of = durations is, of course; the phenom-
enon which led early-day experimenters to consider the span of. atten-
tion_as an elemental unit and-was =a major factor in the reasoning
behind the classic theory of perceptiOn in reading based on the span
of perception. Sperling's procedure of using dim pre- and post-fields
was essentially the procedure followed_ by the early investigators.
Sperling, having determined =the basic-"span of perception" of
his subjects. conducted a series of experiments utilizing the technique
of "partial report." In = these, `sev-eral= stimulus =elements ,would be
arranged in several rows (e.g., 16 letters arranged in -4= rows of 4
letters each) and the subject would_ be called upon to report a specific
row b_ y the sounding of =the tone. Sperling= found that-his subjects
were able to report any row called for even when the command tone
was not sounded until after the cessation of the light source. Since the
fields used were well beyond the memory span limitations of the
subjects, he concluded-that the subjects were reading the stimuli
after the light had ceased from a "visual image."
As Sperling's study sought to find some explanatiort for the
persistent reports of subjects that they can see more than they can
remember and report in tachistoscopic experiments, he interpreted
his findings as affording experimental confirmation of this report.
In his discussion of the data and in subsequent writings, he seems
to equate the "visual image" with the fading memory trace of the
Heron-Harcum Model. _Other investigators have agreed with this
 7

GEYER 67
0
interpretation. Mackworth (108,109), for example, equates Sperling's
"visual image" with her own use of the term and with Glanville and
Dallenbach's "memory after-image." But there are difficulties with
this interpretation when applied to normal reading conditions.
In the normal reading situation, all "fields" are equally lighted
at the level to which the ,eye is adapted. Tachistoscor ::! variations
from this luminance procedure create' difficulties in applying the
results to normal reading phenomenon. Figure 2 plots the data of
Sperling with data reported by _Mackworth (110), showing the effects
on number of elements reported by an interaction of exposure dura-
tion and various luminance schedules. In the area below 50 nisec.,
the luminance schedule is critical. 'The use of dark pre- and post-
stimulus fields at short exposure duration provides the subject with
a visual input equal to that provided by much longer durations.
Cattell's results (31), therefore, reflect, response variables such as
familiarity and contextual 'constraint, - and provide no data for
inferences regarding the "primary cue" in reading.
The effects on form recognition caused by varying the luminance
of tachistoscopic fields has been extensively investigated by Eriksen
and coworkers (53, 54, 55), who ascribe the effects to a combination

1
I

6
0......... P. cod Postfields dark, stioulue fad& 3 It. (Mesckwettlt, 1963)
1J
//
cs........ Pre. and Post 0.14. dims stimulus 0.14, 31 ft.d. (Spoiling, 1960)
410. Pre. and Pee4elds32 ftel: stimu!te Aeld 3 9 fa (Morkwonit, 1963)
2
i Hypothstaind scanning rote

./
O
1 I

20
i L i-i
40 60
I 1

60
I 1

100
I I

120
l'il 140
I

160
I I
110
I
200
I

Exposure duration (rim)

Ftotnm 2. Interaction of luminance and exposure duration in relation to

hypothesized scanning ram,

14-

.
 68
THEORETICAL MODELS AND PROCESSES OF READING

of luminance summation and reduction of

contrast. Deviations such
as those found below 50 msec. in Figure 2 seem to fall under
Block's
Law which states that at brief durations the length of
exposure and
intensity level of light are interchangeable in effect. Tha is,
ing a stimulus field at a higher intensity of illumination present.
than that to
which the eye is adapted has the same effect
as increasing the ex-
posure duration. The common finding- is that holding
exposure
duration constant, an increase in illumination of one log is required
to add one stimulus _t0 the report. Efron (50, 5.1) reported
that a
reduction in intensity by one log unit increased the latency
of the
visual process by 8.17 msec.
Sperling's "visuaLimagerwhile a genuine effect in the specific
tachistoscopic situation, cannot be the-"fading memory trace" found
in experiments using equally lighted fields nor can it be considered
a storage system in normal reading. In experiments employin4
stimulus fields unequal to the adaptation level, these
two phenomena
would be temporally confounded. Therefore, in
the discussions
below, "visual image" will refer to that portion of the fading
memory
trace attributable to luminance difference& The term "iconic
age" will be adapted from Neisser (1961) to describe stor-
the fading
memory trace produced by tachistoscopic conditions comparable
to
those found in normal reading. This usage may- prove confusing
since Neisser used the term "icon" to replace Sperling's
image." "visual

Characteristics of the icon

Of considerable theoretical interest are the temporal
character-
istics of iconic storage. Using tachistoscopic conditions
equated to
normal reading, what is the minimum exposure necessary to establish
iconic storage and how long does the icon remain in usable form?
The answers to these questions depend largely on the assumptions
employed in arriving at them. Most investigators of these
questions
have maintained the assumption that the icon is established
through
a parallel process and elements are processed sequentially from
the
fading trace (e.g., 71,108,109,116). This is the Heron-Harcum
model
with the additional assumption that the length of iconic
storage
varies with the luminance-duration parameters of the stimulus
presentation. Thus, longer 'exposure. durations produce
longer last.
 in icons enabling more elements to be _reported. While attractive,
IN.. model suffers from severe flaws-which will-be_discussed below.
Estoblishment of_ the icon. _The- minimum ekpasure duration
necessary -to transfer -elements- froM -the;visual field to iconic storage
is:an important theoretical question -for-it-would,specify- the procesies
acting during a- knoWn :initial- -portion-- of- visnal-perception-. More
importantly, "perhaps, it AvOuld_-provide significant- evidence as to
whether-these initial procetsti---atted--sequentiallyir in-parallel --If
_The tattle-, amount= rdlessfc& the-number-of
regardless
elements--re rtect,Ithit=*citi in keeping _ expected-
-'aetiOns-r fallel;syStein.--z-Ifi-additionalincrementiortime_were
required = for= eat --iadditiOnalelementireportid,--hoWeveri= Sequential::
,-pnicessing=-Would-s_be_inditated;==,:=--?
SincCS tlines-classie-feicp_erim-eng31yrevealedl-dramatically_
_ _

e_= at:-_prOtessingdoes_liot-iiecess4-rilyittopWithetheetisation- _

of -1C-al_f_stintuhiv;;additional==controlprOCedittes- :_=-Ivere
.
Vit-hPch-=-_,and--Sperling-- used assic-T_pro,
=

ceddre7o jt-t-_=-dernOn-
is--= -tote
stratedihy---Bakt-114;haslongbcen=aniiin _ aant-AC/0 for =the
_

y- isttal!piticessys-and---i-:cer tti.for,th-eoreti
beetiTithOrtitighly-ireview_ccl-,by-Kahneinan- -Canno- t-1)e_ ex-
lensiVeltdisCuised-like.-ImtheifollOwin paradigm t_cance-=to
this--pap0,'-the procedure consist's of= following a stimulus field With
r a field-- corn designed to-- destroy= the inform-a--

correct, immediately _=stimulusifield-Vi di- an adequate

masking-field should 'restilein-Ahetailit..,effects=asrthe=iiie_-_ofstiMultis:
fields-Mon-7i, -,7ted7eqttally: -and-rekattly=4t=t1i-e-levels;Of the
adiptation.-Inii-analysisiiasmorbeenemPirically-Arerified,
n --=
-and irethains an asSumptiO-._
-_Bakt--()iusingsingle_Tlettersi-Eand series of letets-'follonied-by
=_ .

-a bright-;flaih,:foundJhat1-01-Msec-.='_fret of-interferentewairnecestary--
for the ,rierteption-:of ,a= single letter and that an-_additional=_ if----insic.
was - required for each -letter:reported:in a-series. Since -his -Masking
stimulus: was brighter -than-his, stimulus fields, _this _
-estimate-
is--probably'slightly-exteisive.---Aveibach and-=Sperling:(8)-;__howeVer,
report estimate -11-e_use-of--"parts of as the-mask.
Sperling_(196S),2goinifOund_thatriq_Insec. of exposure was= re4Oited
 THEORETICAL MODELS AND I,AO CESSES =OF READING

= for each letter reported- and that the effect

was independent of the
_
number of letters in the field (from two to six),
The- question of sequential- versus parallel
processing would
seem settled were it not for 'the -careful work-
Warrington (96, 97). -These investigators of.Kinsbourne and
presented -tachistoscopically
singe letters-to:lowed:by-a masking field-Composed-of
a-statistically
random -pattern with all =fields illuminated equally.,-
All: subjects
achieved complete-identifications of"a-single letter -from_
a stimulus
preseritatianr_of:_2fis-mseciibucatEthis-exposUre :duration a_ relatiVely-
16*Ni-hie-fir&
Stitaulus-Jor:identifiCatiOn;:to-=--sbe: possible'._At progressiVely-Tonger
stimulus- durations the necessary
at interstiffitiluisinterValsleikned'Until=
durations s
_masking: was -no longer- effective
.
even-=-Whenpreiented-:anstantaneotitly after -_ the= inforniational=itini-
ultiS.Theitasking-ieffect-sivas:alLor,bone-ittak-tpretiliClo==fhei:areas_of-=
the visual -field = coveted;
settion_ndf:- e_-randOnr patternjust-
large enduglvt-6Cover=theit
e-niettetwas4teffectiVe
--a-mask--=arilvas--thez larger -itandiim_patterni=
e large ==randOth
ipattern from =which =the= section= had_ been:- cut-=-produced--no
ferente.=- -Truett
" _

-Kinsbotane:and- Warringonninveitigatedr_the
quettion-
quentiaVVerstis:==_
parallef--_-pracetsing_-_=_Isp_resentingT:4-=randll_ltletter
grOups to tviorottheir=subjeCti.
Their-results indicated
. that
the_threshold_ifor-thecOrrectlidentification
-___--- of all the=_
letters lir the_ taclustoscopic:grov _--
never= aceeded-that of- identi=
-ficationiof one -of the -igeoup-:by more titan= -cini6
_ tachistoscopic
step. In _most = cases,- :two -sandsl_threer-Jettersi=were-
simulfaneoitsly
identifiedat the-same-exposure duration as .
It apptaii-that_-andet-ftlie-presenc Conditions=
tandout--pei.;_
ceptioncisEinfluented by the aftercoming randOni pattern much
as- is-single perception:

In .their conclusion, Kinsbourne and_Warrington

very reason-
ably applied this finding= specifically against ". .
. serial = perception
of= the type postulated' in 'filter theory" (20). Their=
perfectly with what would be expected= from= the assumptionresults Weed
==

parallel perception, namely, a certain minimum= time of

= necessary to
the sensing of the visual field regardless of =the number
of elements
GEYER 71

present. Their finding, if left unchallenged, would constitute a severe

difficulty to any scanning hypothesis. However, the conflict between
these results and those reported by Baxt (12), Averbach and Sperling
(8), and Sperling (1963) requires that a closer examir,,tion be made.
The subjects used by Kinsbourne and Warrington were highly
practiced in tachistoscopic viewing, having previously been exposed
to the hundreds of presentations involved in the other sections of the
study. They knew that no more than two or three letters were to
comprise the stimulus field and that-the exposure duration was to be
extremely brief. These experimental conditions would be expected
to encourage a viewing strategy of "mending broadly" (discussed
in a section below) and the results obtained would apply only w
these particular conditions,
- By varying the amount of alphabetic material in the stimulus
field and the length of the exposure, Gilbert .(63)_ obtained masking
_
effects at-each of a series-of-exposure durations ranging up to 168
msec. The effects were a function of the namher of elements in the
field, the elements on the right being lost ii each case (Gilbert,
personal communication), as would be expected from the retinal
locus studies. Gilbert's data will be presented and discussed in detail
below. The important point here is that masking effects of alphabetic
material can be obtained from exposure durations many times as
long as 8 msec if the number of elements in the field is sufficiently
numerous and if the stimulus durations are randomized. Therefore,
while the careful illumination controls used by Kinsbourne and
Warrington give credence to their finding of 8 msec. as the minimum
exposure necessary for the processing of a single letter under the
illumination conditions of normal reading, their finding that two or
three letters require no additional time would not apply except
under the procedures they report. Scharf and Lefton (124), for- ex-,
ample, studying the effects of luminance and masking on-thresholds
for letter recognition, report 8 msec. per letter as the averafre rate of
acquisition for their subjects.
If 8 rnsec. is accepted as the exposure duration necessary for a
single letter to become immune, to masking effects under the illu-
mination-adaptation conditions present in normal reading, and if a
similar period of time is required for each additional letter, then
the Heron-Harcum model- runs into difficu:ty. Certainly identifica-
 72 THEORETICAL MODELS AND PROCESSES OF READING

dor, of a letter cannot be accomplished in 8 msec., but something

happens which prevents further interference of a masking nature.
Sperling (1963), using his estimate of 10 msec. (the difference is
trivial), suggested a model wherein the visual field is transferred
in
parallel to iconic storage and the subject "reads out" from the icon
until it is "erased" by the mask. But -this requires a reading rate of
100 elements a-second. Therefore, the model has been dropped
(Sperling, 1967).
Another difficulty is the constant 4-or 5 element report given
by subjects from-a wide range of stimulus durations where a dark-
light-dark tachistoscopic procedure without -masking is employe&
This finding, the classic "span _of= attention," is consistent across
groups and generationi (31, 40, 131). Since Sperling's experiment
showed that this procedure actually provides long viewing durations,
what subsequent systems limit the report well below
average memory
span?
The hypothesis advanced at the beginning of this section would
seem to account for thete phenotnena. Once the assumption of in-
stantaneous and simultaneons input is- dropped, the rate of 8 (or 10)
msec. per letter becomes that rate at-which letters are transferred to
iconic storage. Masking would- interrupt -this process but would =not
effect those letters already in storage. These elements, would be
read out at the rate of silent speech until the iconic storage faded._
Duration and nature of the icon. Very little can be confidently
stated concerning the parameters -of iconic storage. Neisser (116)
estimated its duration as from one to two seconds, but this estimate
includes variations in lutninance schedules: Mackworth (108; 109),
using equally-lighted fields, estimated storage as "about one second."
A storage limit of one second would explain the ''span of attention"
effect, since 4 or 5 letters a second is a reasonable silent eading
rate.
Geyer (62) postulated iconic storage as the basis for eye -voice span
in oral reading am! predided that the temporal- span would'remain
relatively constant in smooth reading at -a span of approximately one
second This prediction was confirmed from data generated by eight
subjects reading three passages of varying difficulty. The overall mean
was 1004 msec. and individual- subject means for all passages com-
bined ranged from 904 to 1088 msec.
Even less is known of the nature of iconic storage. Introspection

s,
GEYER 73

identifies it as being auditory in nature (90): Severatstudies showing

greater confusability from letters which sound alike cciuld -be inter-
preted to support this position. Sperling (1967), for example, reports
that when 3 letters make up the stimulus, it makes little difference in
performance if the letters sound alike or sound different. Perfor-
ma- nce deteriorates for more than 3 letters, however, if the letters
sound-alike. Sinmmemory spaircan typically _reproduce seven letters
without confusion, it is difficult to ascribe thew findingS tOlimita-
dons of subvocal reheatsal. _Taylorand-fPcisner(19-68), on the _ other_
harid,- have presentedvidence:for vituatitorageiprior.torhe-fverbal
storage,ialthough them:use of- dark= interpolated-intervals makes their
datacliffitultro- understand ip _the present-context. The:question at
preteht is=tocr complex. for_fruitful
for -_analysis =here.

Viewing_
The inalysik,presented:2a the :retinal- locus _--eitperi7---_-
mentatiori can -provide-scime7 insight into epossible-__Strategies-
availible.to-Subjectsiirtrachittostopit__experimenW--Reaclingf_can-_,be-
thoughtof a-proceis=brWhichespatiaiiequentei are-zzeiniVerted- to-
temporalrsequekes. -accomplishedlby=processing4r=at least
three levels loosely _ideritifiecf_With=tensbry;tecOgnitional- and motor
processei.-DireCtionalityAs--initiatelk at rhe-:sensory- leirei hut
-parently determinedPby -the-Ilesireds: sequence of _the.;finalE:reporL_
Thus, if a:subject- knOWS=before-the----expOstire the -sequence.-iiilithith
the response' will -be -made, -heseats-che---inaterial- in the- direction-
whiCh produCet- that sequence. Internal =-recognition: takes- _place in
the sequence, and -the externalFresporie is -made-airetogniticin
occurs, -or from a_storage of a covert response-. Ifrhessame-Secfnence
is maintained= from= to output, -the systenisroperattatinakiMuin
efficiency. If a sequence of external- ieiponse---is-required--- which is
counter_ to the-direction, of scanning, however,- the __-most effective
strategy-would-be-to store the totarinternalretponseiand_the report
in the-reverse_ direction. This would limit == the= final report: to back-
ward memory span,_ while -preserving a SeMblance of:the eirectional
effects_ imposed by _the -fading: icon.
Long prattice has apparently= made the English-reading subject
more proficient witl-theleft4o4ightScan-direction. 'Thus, -while he
shows right field= superiority-wheri:reqUited to respond= to _nondirec-
 74
THEORETICAL MODELS AND PROCESSES OF READING

tional material in a right-toleft order, his right superiority

is not-as
great as his left field superiority when his response is in the
normal-_
order (9, 76,-79). Given-a choice, he responds in the
normal left-to-
right pattern. If he isinitructed to-respond-froth
right-to-left before
the exposure, he shows a marked-right field superiority.if the
instruc-
dons are given- after the exposure -but before the response, his
accuracy is-__reduced, and a:slight left :stiperiOrity is shown_ (79).
-Bryden --(24)
alsolound-that-withnondirectional rciatcrial-the lateral_
sUperioritywas*ffinctitiii-of the ordercol rep:in.-1kt with--:Alphabetic--
_Materialsi_theleffett Of_frequirin t=t6-left--re ft Was to -ratlike
of ==letters- reporte- teserving "the left _field
superiority._7_Brydett's Subjetts-:frequently reportedi_ that in order: to
report:theletters_in-:iirighp!toleft direction, they first:hadilo-repeat
them tOlitheMselves
One-sensory
-strategy_ seemvpossible- _z-certain
istostopic7-c-onditions,- classic ittnann'(129):_a04
Wagner: (141)11O7seem-_to,in tate --tsubjects On-attend t0'1-a;-to
stimulus: This __is the strat :tending: broadly "_ = _mentioned
above in relation -to-1- fife studies =of = Kinsbourne fan& Warrin-
4.- "
Wagner; -fot-itaiiiple;(ixs"----- -a long -Sefieti:CCUntonnettedletters_
fort-dilratiori of 1001insec.-fan = urged -is-aUbjects-to_attend-
broadly
to the whole row. His its ults showed that subjects repc ted letters
which = had occurred throughout-_-e-i_roir;;although'itost-,letteis_
repOrted Were-from:botkendi and theitenter.=Several fattors in
classic experiments= -need-diSCUSSion.li-The: fact that Wagner-_=had
AO-
Urge_:_his-rhighly_piacticed-subjetts to ="` -itterid==bioi-dly" Significant.
That this proredureAs hoc- strategy with large arrays. of
alphabetic materials is clear= om ttie =studies presented--abOve. The
reason -that- it -iS,ficit-zthe-i normal
procedure_-_ii- apparent 'from the
_

retfirti obtained. _If-- the_ row-c_ of' letters- had dztOmprised-SeVeral

'Words,
the subjetts-presumablyivould have- perteivect the first two or dace
letters of thefiritt-worka-Scattering of
few--lettersfrom-the_center
words;ancithe laStrivo-or three-letters froni the final word certainly
a poor_ strategy for :reading. Whetthe=Vitual field -contains
-more-
alphabetic material than = can be processed= in a _brief -flask_ the_beit
procedure would seem- -to -be Mitt:1*i from left to right in Order to
perceive as mucitof the;materialf as possiblein- a meaningfullwar--
and -thit isapparently the proeedure that subjects -follow
typically.
GEYER 75

When the stimulus field is small= and the presentation very brief,
however, ex_ Perienced subjects might develop the "attending broadly"
technique.

The. Geyer Model (1966)

Thrtnodel
Figure 3 presents a block diagram =of =an initial heuristic model
of reading developed-by Geyer (62) from an analysis of the litetature
similar to that presented above. The Model does not -purport to be a
complete representation; such obvious processes as feedback systems,
for instance, are not represented. The systems presented in thefigure
and described below are those hypothesized as functiming between
The Moment in time when the stimulus_ is presented and when it is
reported, and the order in whi are ptesented is the order of
their functioning in processing a en stimulus. =In an ongoing
perceptual task such as reading, the system= would = tend toward an
overall steady state based on the "channel capacity" of- the slowest
system.
Sensory! input (scanning) system. A = major hypothesis of= the
model is that visual input is sequentially ordered within the= fixes
tional pause. In reading English, the input is from left to right at -a
rate of approximately 8 cosec. per_letter space. It this hypothesis is
=

accurate, then the major ftmction of = the =saciadic movements = would

be to keep the eyes in a Imition where the covert or attentional
scanning is within the retinal area of fine discrimination and input
could ..ate place during any time = segment of the fixational pause.
The input of importance to this_ model is that portion of the
visual environment central to the reading act of the moment. Other
types of visual information may be processed simultaneously through _
other systems. The contents of the visual = field have many character-
istics, e.g., presence versus absence, shape, color, size, location and
brightnes.s. There is growing evidence that some of these character-
istics are treated separately by the visual system (34, 57, 58,125). The
sequential input hypothesis relates to the -"figure" and not to the
"'ground:* .

Sensory organizational systems. Immediately upon input from

 76 THEORETICAL MODELS AND PROCESSES OF READING

PROGRAM

Visual Sequential input at rate of approx.:

Input imately 8 msec. per element.

Coding Conversion to icon, including some

higher order organization.

Iconic Icons independently stored for

Storage proximately-one second.

=internal =Verbal _responses at approximate rate

nse: Of 250msec. Oerluriit

Verbal Store limit of seven., seconds when

Storage needed. Indefinite 'storage possible
through rehearsal.

= Verbal_ -_ Processing rate m oral reading is

Oiktpiit approximately 300 row. per unit.

FIGURE 3. Block diagram of Geyer model of oral reading.

the above s; ,P.n, organizational and associational processes would

begin. From till:, point in time, the development of meaning would
continue until well after the processing of a given word was com-
pleted. The logic of the model, however, requires that prior to
GEYER 77

transfer to the next subsequent System certain transformations be

completed. -In the case of unconnected letters or -digits, relatively
little transformation would be required;for -conversion- to icons; In
reading words,_ however; the letters woul&betransforined to higher
order units (presumably- phonemic) prior to storage. Ii should be
specifically noted_ that in this= model-, the stimulus_ elements are
scanned and-undergo some processing --prior to_ their _transfer to the
iconic storage _system.
Iconic itoroges_system. Once the- elementS__ol the stimulus= are
organize& into units-_-(eg.i- = =these -units --
livered--to an _iconic'sicra 0=iysten.--Siehi-tniCr eita s-k tessibl elofs
responsetfornapproxinatelyone_second,__hut=iUno=reipOritei- stnade
'the _unitslz.are lott._ This _Storagez-is_t_thought to=be__identical td,that
which Glanville and-Dalletibith----(66yittidied=ailke-,_!Meittory- after-
image. "- -Sperlines --"_viSualAmage!!--,-(04-4Oul&-zitidtide =

Storage-but --_

to linninanCe-variablerartifictual-toKthit4nodeFsinterthefda not-
_ --z
occur in normal- reading _ --primarrfunctioir ofithis=system: is -to
-----
act as a -temporalizrbuffer _t_weenT-_-seticory,a-itd__respCitisey'phases-___to
_-
allow integrationzof=thedifferingiprocessingrates.
Internal -Tesponw-syttem.The-pottihle----ketporiset,Matie:=-
- -dnring
_
_
reading are -many= and _-_complex-:and--would-tmvolve-Naried -systems.
-0f-interest to this:paper-arethe--verbalfrespOnses-zrequired in:tachis%
totcopir eicperiments--_aridz-nitistil-not=a11;_--teading.-=:Theinteractions
bettveen the rate_ at which-tresponses:_can'ibe niade,=-__the- elements
making- up response -units,------andilllielength-Utikteticont-temain in
storage are seen as-the-Major fattors-sgoverning speed of reading-and-
eye-MoVement charatteristics; = _ _

Several ingtigatort,Aiting-Auite_-dissimilar_egporimtntalisitua,-
tions, have -pot tiilatedza -5= or evicyde: per second:Central nervous
system response rate (Pladcive11,14).-Lansing;-SchWartz;and Lindsley
(103) have pOstiilated_ a physiolOgical-b-a-sis-= for such a respante-rate,_
centering around the action of the -ascending reticular =activating
system- on cortical resolution and the -transmission of incoming
messages. As- recognized by Davis (1963); these postulated response
rates agree quite well- with the -long known (but little underttood)
fact that the mean -pause length in normal silent readihg-remains
 78
THEORETICAL MODELS AND PROCESSES OF READING

remarkably constant at around 250 msec. and

-reading rate are a function of wide that differences in
variances in number of fixations
used (Stone, 1941;_ Tinker, 1951; Walker, 1933).
It seems entirely possible :that the physiological
Heron's "scan" rate -(identified here _with- the response rates,
system under discus-
siori),--and the relatively:
constantirfixatiorial pause -in -- reading
functionsA_related -phenoniena.- Lichtenstein, White,fliand
are
(105:544),-stu dying- the Terce d on' oUler_ ies =o f bref flashes
= at Aariousfret receerttieiti _ d eliVe red
unci the- flashes -VereitoUited--=
at a-stable5=to_ per- second=at ea Tecceritrici
ju -e=tubjecti-
occurring ee elovea
as at the -most- distant- int of:-
The -to-tiduded-- that
The resultt of Ve hich,
should- currencacnce
duration-iunits = i_assoaate
thost itOriceptioni -thereiss_theibitir okAacit
one basic ,icar-Ttate =is_operant system.
torecognietie-vossibiltie -em =reialtXo -stu y
that-morelhanne-,baiia=leid-rate-zma
____ operating
neouslyiand-that_each_'s-cteratema tom
entvistial_ ftMc-clan 'Ur 'set- of-=ciperations.--

Secondary (internal responsos4totage_ te -the-covert

verbaLresponse :-name =_Cif- the :stiinuluxix trantferrecrito-_
sccondizstoragesystait
-AJiiii-e-t4p-atign_OU-setferaiXecoriclis=bfit
with-- limited!- Unit=ca ci -reading_ overt--_-res -mei_ could
accompany th-responses-_or, eit--requir efinternal
response _be stored'monten tardy until the
responses could=
be -made. -Thit- system, -en,t=co --o
-ax a= tem - buffer
between ,the-te-Co=gnitionaVarid-TradOt--- in a way analogous to
the- operation of the_ini -=storage_ tem operating between the
sentoryiandirect_ignitional-
`Considerable_rese" has--been devoted to delineating
the thit sy:teitil under the rubric of -short terin-
_

memory:(1)-. _Questions
currentlybeingclebatedinclude whether _the
of storage auditory in_ nature-(47, 48,115),
whether coding= is byleininticoricouitieltitaMeters,,(10, 38i- 141i-7
145) =and whether the _
--processes-are decay, interference; or
 GEYER 79

displacement (Conrad, 1967; Murdock, 1967; Schonfield and Don-

aldson, 1966). Much= of this research confounds the two storage sys-
.tem.T hypothesized- in this paper and in some cases Sperling's "visual
image" is further confounded. A clear delineation of the interacting
systems would seem to be necessary before their properties can be
idently sorted out.
confconfidently
response system. No atteinpt will be made in this paper
to analyze -the properties of the external response beyond acknowl-
edging its existence.
The model in operation
The model can be differentiated into three processing systems
associated with sensory, recognitional, and motor processes. Inter-
spersed between the three systems are two storage systems which make
smooth information processing possible by acting as buffers for the
differing operational rates of the processing_ system& These differing
rates are thought of as being based on physiological function&
The model does not purport to represent "the = visual system,"
many well-known functions of- that system are not included and the
last two components of the -Model are not directly part of vision. The
model is concerned specifically= with reading and the discussion of its
I,
operation below will be limited to three ftuations: 1) the tachis-
toscopic situation where the response =unit .;:id stimulus element are
the same, 2) the tachistoscopic situation where the response = unit is a
.combination of stimulus elements.and 3) normal oral reading.
Condition 2. The response mit and stimulus element are the
same. Under the experimental condition where the response unit
and stimulus = element= are the same, = as with digits or unconnected
letters, the effects of the initial scanning= system would be apparent
only at the briefest exposure durations. If a subject can process four
digits from iconic storage, it would only be a the exposure durations
allowing the input of less than = four digits that the initial scanning
system would be the limiting system.
The broken line on Figure 2 shows the hypothesized scanning
rate, if the digits were presented so that there was no necessity to
scan to the beginning of the arra,. For each 8 msec. of exposure
duration, one digit would be added to iconic ,torage where it would
be available for one second:= If a response were made every 200
 80 THEORETICAL MODELS AND PROCESSES OF READING

msec.,_ four or five digits could be reported- before the icon faded.
Additional increments of -time added to the stimulus presentation
would add additional- digits to the report up to a maximum of
approximately 40 -to 48-_msec. For longer= exposure durations, the
limiting factor would be the interaction, of response -rate and the
_duration of the iconic storage. -Additional increments_ of exposure
time would not add to the report until the duration was long enough
to allow the first digit to be processed before the exposure was
terminated-----a duration_ of -apprOxiinatelys_ 250: msec.. approx-
imately 50-Lk:v=250
_

conttAnt_iepqrt:of- four or :five dig4t;Athougli--the--isul4ett-_ Might =

indeed remember haVing-''seen7ftfiore atithelongerdurition;-

= _ With a -large number of- in the stimulus field, longer:
exposure diatatiorti ccialclreiultirt-Smallerrepares. -2t_sitbjctiscanning_
12 digits during- areXpnsureoU100==msec, foreic-aMple,_-wonidJeadi
in many rii6redjgrs-cha-t(he =could-proast and the -Superfluous time
Spent-inscianingf-would,leisen:the;time available to_praceSs,thei-con.
Since the--final=digit- wotil&be=available-_-foralmostA00sInsec.-_aftei-_-
_thefirst-digitladed,- it thiewh- e-fead=oitt from-the=fadirig icon These
effects are commOrily_rep-orted_ in the_Aiterattire.
For each F-subject
.
_
there, would be an optimal field Size_ which
would be just:large enough- _ta permit maximum use of iconic
_

storage :11M. not _so-large-as-to: encourage-,overicanning. :Ma-ckworth

_(i58) reported that -individual= subjects- Varied_ in their optimal
field size but in--each-caie_it-was one element larger than the-subject's
t
largest rdport. _field' _size= and
exposure duration are-lindeperidently--_Variek there -would be an
optimat=fieldsiZe-foreach,duration andyicer_Veriaw Fraisseand-Battro
_
(64 have reported data in agreement with this -deduction -froin -the
model.
Condition 2, The responie-univis a _com-bination-_ofstimulns-
elements; All thel effectStlestribed in the section above would' be
expected to apply to the situation where : the response unit is a
combination of stimulus element:S., In this situation, :however, the
number of stimulus elements-which would be scanned effectively
Would be much larget. Itaisubject:stanned 121etters_zin 100 insec.=,
there would he little -change from the :digit example above, if the
letters were unthnnected. If the letters combine make short
GEYER 81

familiar words, however, they could all- be processed from iconic

storage in- the three or four responses ncessary and possible.
In a section above, an experiment by Gilbert (65) was = cited as
evidence That masking effects can be= obtained following-relatively
long durations if the amount of material in the field is "adequite.
Gilbert. tested 64 college students = on their ability = to read - short,
familiar words in contexts ranging from a single = word to five word
sentences from masked tachistoscopic presentations of varying dura-
tion as well as from an unulasked presentation of 84 msec.2 In keep-
ing with his-ex imenta ur ses Gilbert's data were-teported in
termr---of per cent correct. _ more interest here, is the average
number of -words = =reported from_ =each= condition and the data have
been so convertedThis is legitimate in terms of the scanning hy-
pothesis only if all subjects =invariably reported the-words from left
to right. Gilbert (private communication) has stated-tht such was
the case. The= converted data are presented in Table 1.

TABLE
-oLEAN NUMBER-0P WORDS-CORRECTLY REPORTED

- Number of
words in Exposure duration (ms.)
stimulus field 84 126 168 210 252 84e

One word .81b = .98 .99 1.00 -1.00 1.00

Two-Words 1.73e lz 1.96 1.98 2.00 2.00 . 1.97
Three words 1.93 = '2.81 2.89 2.95 2.90 2.88
Four zwords 2.25 3.35 3.42 3.71 3.67
Eire words 1.80 3.20 3.32 3.60 4.0/ 3.76
a This- presentation was unmasked; all others were masked.
'b Above the dark line are figures for reports limited by the number of words
available in the stimulus.
e Below the dark line are figures for reports limited by the amount of time
available for input.
Note. The figures in the crosshatched -boxes represent two important excep-
tions in the data.
2 Gilbert used procedures yielding time measures in 24ths of a second. These
have been converted td milliseconds here to facilitate comparison -with other
studies.
 82 THEORETICAL MODELS AND PROCESSES OF READING

An examination of Tible 1 shows two_main effects, viz., reports

whichmere limited by the number of words available in the stimulus,
and reports limited by the amount =of time available for input. The
first of these effects is obvious inthe data. The second effect, that
of time limitation, is derived from the unmasked presentation. Since
at each field condition the unmasked reports were the same as the
maximum masked reports, it is assumed that the visual image result-
ing from the luminance conditions of the experi ent was of sufficient
duration to make the systems subsequent to ut =the == limiting
factor in the unmasked presentations If this assn ption is tenable
then the masked presentations =which=yielded smaller reports must
have been limited by the time available for input prior to the mask.
The five-word-252 cosec. cell =is an= important exception which will
be discussed 1.:elow.
Several interesting effects can be seen in Table =1 as divided by a
darkened line. Under the masked condition, additions:of tithe in--
creased the reports until the maxima possible were readied. Adding
words to the fields also increased the reports up to the four-wor
field, but the addition of _a fifth word= was detrimental. This-is the
optimal field phenomenon_ discussed above. In this case the optim
field was four words which is just larger than the maximum reporte
(with one exception discussed below).
All cells in Table L with two important exceptions,= are either
time- or word-limited. The line separating these two effects forms a
diagonal across the table. The agreement hetween this line and the
hypothetized rate of action of the initial scanning system can be seen
graphically in Figure 4 which plots the data- from Table- I with the
estimated system_ s-parameters. Except for one point, the one and
=

two-word fields were too word-limited to shoW the action of the

initial scanning system with the durations-employed. With more
words in the field, however, the points follow the estimated line of
the initial scanning system= until they taper off as they approach the
limitations imposed by the interaction of the iconic Storage and
recognition systems. Those points, then, which were not limited by
the number of words in the field are in essential agreement with the
"prediction" of the model.
,
Two cells in Table I are of particular interest. The model
would predict that at some point the= variables of time and number
 Slinifircul

dotted line An caies Scanning =rate

nse imitations 1

-_
..aA AXiagilaMMui-ii;

Iteratur
 licatidri

tioiL.
___---

_._._
ten tn
taa ner- iieif-
tanite iiilei te

.LATadiiitoscopkStudyof
-dsolotka _
ono a '-= 19 ---;

GSPh:OlogbfthiReUna
ttention a
ustic n usions
(19641:75 83
matnanr7

=7_
 ucationa
--_----_- - _=

ntn tatio
______-- ------
 -A a %/LIM, _164 a:. _a 110 I" &&&&&& 411Ly7r1V4UVIlLys iV14l11J11511111, _ JUILY7144 111G4pC7_1_._-_

41_8. -Orbach;I="getinatiLbetii 23:aiTactor =in Recogniti-on'of A/tit-101r Perteive

-N91'43;77American 9uin-al 4= Tsych-Ol96,-:fi4=Lisow555.-50
119, ZPieree;_j,-11-"Tliterininan -of7= h-reshold:forlFo-rm "_=.1' chaogicallulietini-
-----sazff(1963 ;-491-407"
120. and -eadm -te:, an elnfonnition_-,
elliSyitetiliTedinicatloitinakz367:(1947),=_497.516._--
121: :-PillsbuilieW;:11-; eidifirbFWordi-:--A-=-;Stucly-1 in
_Psychology, ' (1897),----31-5-493.---
122.- -Ildsner,_ inm iate-= eino equenua
B
123; Sal tzina and d_ -W:_R- Garner eaction nue-4i a &attire -roU_Span-
-of Attention; -= oifriiii/o th-oloit);725--(19-48),:=227=24r _

-124,-- Scharf,_ B; and -L _Lefton "Forward and = Backward 11lasking =as a Function=
of Illumina_n_ce,__ p- aper presented EaiteriNy-eh-ologicatAiiititiation0968=
-125, etacontiait,'_ journal o,
z-kriin'efitaliTsyekology--,-,7V(1966);-=_82z
126,- iSchilletc=_P;-
ournal of Ex irinentill;P*holdly_;;66_141963)386=39
127. Schotifierd,: an
_Infraseries Variation;' _Canadian_ Journal of =1201(196W 21-8-227f-_---:
128.- _Schintaiin-,--sF,=mieEkliefinling --lioli=lhichitabeii-Jtii ort-iii==beiMoineti-__

Psychologie,
,-i8thintiann; is serken indigiu-rteiViZeitieh t -ItirsP*4460;08=_0_9217_,
1922). 205,224
130: -_=ItiklarPeteepticirW 0-,'!_iPsychelo
= 6C(1957Nao6=3
1;-s
*Acilogiedl-FM0frogid:hi441(1960) .
132. iSprigie;X.---_1:;:fEffe-ctic11iffetential--I'rainir_ig=_pir
----nititinalteiholds77otirnarbtExperimentat P.i.y_ch6W 1959 427:2,
133; Taylor; j,&=:mfett_i)::fiSe(6UAis6ciatedl;Wof uration Thresho
=Perceptual and otor:=Skilk 956)03
184: Taylor;- J. =- e-ahing-iz-Yrequeney;-an Visual
I n
-19UrnaVO e_iiinent-alrErteliology;:5k(1958); -29=3$=
135. TaIlori
Canadian Journal to -ch79/9-107= (1953),-,42-
 cultural deman
wever _inappropriate
seem, em oniy ones
articular circumstances. It must kCe
eenou o accomrnodate-man
it altmative,
cient or more sOCiallyadelita
has had Many experiern
from w CMS
problems, with zest and con ence ct s out his
)robletns, his own world animals who have re.
ceived adequate stimulation and a ection in infancy will ee out
actto

m_ e,Ear ssta
ctors=e4
thitie=sator=
,,---- --_---_-i------_--------r- 1
_

_
 RETICAL----MODELS-ANDPROCESSESrm-

-=--===--- --- -
 =---_-

= =-=

rocess o cation
 w e- =rea m -
-movement= =recor
wit outa eetin e-on r
r_conee

un era le
u

icu
-as a- reader atten er
sense rson
i scanay

.andzinore-:= rune_ o-
-,beenifounctcleficierit-in-ie oltren e eabili to-gauge _-:
-reality-and4 eiize--= -thitritft iti=ipiirci-priatei
_,
_
'reeted-Tacticri-
4-;=ltkiisey: -)liamremarked -=: at rreaders-_-- --
e ess re isticii- leitegiiateso
____
pntelitrat-readers;-vhile-
 _
- - .
 I

114 ThEORETICAt. MoDELs ANb PROCESSES OF READING

the research front, such measures would be invaluable in sharpening

up the picture of the precise relationships between affective fadors
and chool learning. Until such time, We must remember that learn-
ing to read, or even school -learning .as whole, is not an end in
itself, but a meanS toward greater self expression through successful
coping with problems and the invention-of new problems with which
tochallengethedevelopingorganism;,
The role-of the intellect is to enrich rIther than curb the
-
en1otions, to direct their ecpression toward goals emerging from
viable knowledge and cultural ideals (6! 13)
If wewant our children to be intellectually literate, perhaps we -

should &mcentrate on m4ing -them emotionally souid as the most

efficient route to oufdual- objective. Perhaps we should worry less
about Johnny's reading ability, and more about Johnny.
---
REFERENcES - I
Abrams,J. C. "A Study of Certain Personality Characteristics of Nonreaders
and Achievers," unpublished dcctori1 dissertation, Temple University 1955.
2. Anderson, Dorothy J. "Manifet -AnxietfLevel andReadhig-Achievement:
4 A Study of the Relationshipi between the Manifest Anxiety Level and the
Reading Achievement of Fifth and Sixth Grade] pys," unpublished doctoral
dissertation, New York University, 1964.
3. Athey, Irene J., and J. A. Holmes. ReadingSucccss and Personality Charac-
teristics in Junior Higk School Students. Berkeley: University of California
Press, 1969
4. Barber, Lucille IC.. "Immature Ego DevClcpment as a Factor in Retardd
Abilitjto Read," unpublished doctoral dissertation, University of Michigan,
1952.
5. Bauei, Edith B. "The Interrelatedness of Personality d Achievement in
Reading," unpublished doctoral dissertation, University-of California, 1956.
6. Bid, J. E. "Emotional Factors in the Treatment of Rding Difficulties,"
Journal of Consulting Psychology, 9 (1945), 125431.
7. Blackhain, G. J. "A Clinical Study of-the Personality Structures and Adjust-
ments olPupils Underachieving and Overachieving in Reading," unpub.
t fished doctoral dissertation, Cornell University, 1955.
8. Bodwin, R. F. "The Relationship between Immature Self.Concept and
Certain Educational Disabilities," unpublished doctoral dissertatin,Mich-
Igan State University, 1957.
9. Boulse, Louise M. "Emotional and Personality Problem of a Group of
Retarded Reader,," Elemintarj English,-32 (1955), 544.548,
 Ici n-, o;
eve lolbm en

= ,Row--
2 a ___ure:stin=_:ticlus: -in _Ournal-fo
otiOnal= _ncationnUAssociation 19
-___ e:IInderathieiltit Giftedf one,'_'-
a_t-etio_tta - Children ; 1955)247=24 ----:-
._- _rituo_wi-____ "A _ntparativeStutly-o n erachielrewIt_IormalAChievers,- _
and 1 ers-i_it,=; --din--"Aapith itsertatiOn-i-lolva--
State _JJfiiitraity 19
ttitudes = toward Reading An
2C --GrOff,-=_IF.:I_-'_n!Childreiter, Attitudes eir-Criticaljtead,--
,-= _ing:_-:Abilities--:_in'=Fati:_:Ccinterit, T_=ype Materiali,"_ -- JoUrnal_01=-_-Educational
--i_ isearch, 55.196 2), 313-31
=

25. Guilford, J.____P. "The= ThterFates of :Intellect," American:Psychologiet, 14

(1959),A69479._ = = -- _-_-=__ i - ---=-- : =- _- -- _-
26; Hallotk, -G:. A. "AttitUdinal- Factors--Affecting-Achieveniefit hi:Reading,"
unpublished -doctoral_dissertation; -Wayne:Stitt -University, -_= 1958.- _-- = :-
27. Healy, Ann-X. "Effects-of=Changing4jhildren!sAttittidaToward Reading," --
Ekmentary-English,-42-(1965)-,r26947t --_= _- -_ _ _
tv 51. Peter,-_-Suter
I 'cal Factors n Remedial_ Rey
dlisseYdriitlglon,UniveriitfofRochester,
E "The Mea.stnetnent of xi t to Youn Children ==
and 'nship to Later ReadingMAachnifiees;nen1,- 11
usertahon un1V-C Connection,- 1967 .
a
of Sahe-nt_ Ch era eristia ofdPuint7b5

rs and _Academic ent in 4797

SO ekoblY
S"
=Olive
nd"-
425430.

Boys with Low Reading Ability and ReaTonably Nonnal Intelligence," 1

unpublished doctoral dissertation,_Texas State College. 1960.
58. Shapiro, M. A. "Relationships among Extraversicin, Neurotiden, Academic
Reading Achievement, and Verb-al Learning," unpublished doctoral dis-
sertation, Rutgers -The- State Univers4 1968.
59. Shatter. Florenc. "An Investigation of =the= Effectiveness of a Group Therapy

;
- -

1 1$ THEORETICAL MODELS AND PROCESSES OP READING

f
Program, Including the Child and His Mother, for the Remediatin of
Reading Disability,' unpublished doctoral dissertation, New York Univer
sity 1956.
60. Shrodes, Caroline. "Bibliotherapy A Theoretical and Clinical Experimental
Study," unpublished doctoral dissertation, University of alifornia, 1949.
61. Simula, V. L. "An Investigation of the Effects of Test Anxiety and Per-
ceptual Rigidity upon Word Recognition Skill of Second Grade Children,"
unpublished doctoral dissertation, indiiia University, 1964.
62. Singer. H. "Subitrata.Factor Reorganization Accompanying Develpment in
Speed and Power of Reading at the Elementary School Ievel" report of a
Cooperative Resarch Project to U. S. Office of Education,-Contract No.
2011; 1965.
63. Smith, D. E. P., and Patricia M. Carrigan. The Nature 0/Reading Dir-
- ability. New York: Harcourt, Brace 1959.
64. Sopis, Josephine F. "The Relationship of Self-Image of a Reader to Reading
Athievement,"unpublithed doctoral dissertation, New York University
1965.
65. Spache, G. D. "Personality Characteristics of Retarded Readers as Measured -

j by the Picture-Frustration Study," Educational and P.sychologicalMeasure.

tnent, 14 (1954), 186.192.
66. Spache, G.D. "Personality Patterns ol Retarded Readers," Journal of
Educational ilesearch,i0 (1957), 461.469
67. Stratemeyer, Florence B.Perspective on Action in TeacherEducation.-
Washington: Americaii Association of Colleges for Teacher Education, 1965.
= =68. Tabarlet, B. E. "A Study of 14enta1 Health Status of Retarded Readers,"
unpublished doctoral dissertation, Louisiana State University 1958.
69. Tolleri, Gladys S. "Certain Aspects of the Self-Evaluations Made by Achiev-
Ing andRetarded Readers of Average and Above Average Intelligence,"
j
unpublished doctoral dissertation,- Temple University, 1967.
70. Tuddnhans, Read D. "The Nature and Measurement of Intelligence," in
Leo Postman (Ed.), Psychology in the Making. New York: Alfred A. Knopf,
j
1963.
71. VanZandt, W. "A Study of Some Homelamily.Coniniunity Factors Related
to Children's Achievement in Reading in an Elementary School," unpub.
hued doctoral dissertation, Wayne State Universit 1963.
72. Vehar, Mary A. "Extraversion, Introversion, and Reading Ability," Reading
Teacher, 21 (January 1968), 357.360.
73. Vernon, Magdalen P. The Movements of the Eyes in Reading, Medical
Research Council Report of Committee upon the Physiology of Vision,
-
Special Report Series, No. 142. London: H. M. Stationery Office, 1930.
74. Wattenberg, W. W., and Clam Clifford. "Relation of Self-Concepts to
Beginning Achievement In Reading," Child Development, 35(1964), 461:467.
75. White, L W. "Motivation Reconsidered; the Concept of Competence,"
= Psychological Review, 66 (1959), 297-333.

L
76. Witkin, H. A. "Origins of 'dye S e," in Constance Sheerer (Ed.).
Cognition: Theory, Research,, Promise. New York: Harper and Row, 1964.
Witty; Pr= A. "Rea_ding= Success= and Emotional Adjustment," Elementary
English Review;27 (1950),
Zimmerman, Irla L., and . . ebrand. "Personality Characteristics and
Attitudes = toward Achievement of and Poor Readers," Journal of
Educational_ Research, 59 (1965), 28.30.
--
=SKOP"
- V

V
f

V
V

122. THEORETICAL MODELS AND PROCESSES OF READiNG V

V
for the older children to teach the younger children spontaneously j V

throughout the dayfl helping with an unfamiliar word, helping with

V

I :
spelling, etc. We have known for years that the best way to learn
V
anything is to have to teach it. 2)4 curriculum built around areas of j V

V
V
interest rather than content subjects. As far as reading is concerned, I
V

there is no formaF reading period but reading goes on informally V

throughout the-day; The teacher tries to spend some time each day V V
V

with each child; either Vin(jj,jdliy or in a sinai! group, keeping.

dose check on ch child's progress-in reading.i) A slow, relaxed,
anxietyjree introduction to reading. The child has an opportunity V

to take about four 'years, from the age of five to nine, to break the
code. Then seems to be no pressure put on him by the teacher-or

apparently by the parent to show significant progress in codeThreak-

ing by Christmas of first grade, as is true for many of our first graders. V

It seems obvious that this natural pressure-free learning-to-read

process in which older peers play a major role provides children
with a learning environment that is highly conducive to' expressing
openly and coping with feelingL - -
V
'he final suggestion for those teachers who are concerned about
V
V
children's feeling in reading is to reexamine the literature concern-
ing individualized reading In the early 1960's there was an obvious
trend toward an increase in individualized reading-programs, but
j
V

in the last few years that trend seems to have reversed. We-hear little
about individualized reading and see kw examples of itin the class- i
room. V

As far as feelinp in reading are concerned, the most important

element in an individualized reading choice program r
of
reading material. When the reader is able to choose what he wants
to read, we maximize the thinces that material will have high per. V V

sonal meaning for him. We also maximize the chances that 'reading
V

experience will be involved rather than superficial, thus increasing

V
the reader's opportunities to practice "good" comprehension skills.-
The literature concerning individualized reading reveals one
t irrefutable impression: children who go through an individualized
I reading program seem to read significantly more and seem to enjoy
reading more than those who go through a basal program. How many
teachers ever consider the effect on children of continually going
through a reading program in which they have a minimum of

F
MUSKOPF 123

control over what they read? Free choice responds to personal feelings
and, therefore, would seem to be an important element of any read-
ing program.
Summary
We live in a society in which =the pervasive effects of external
control on humans must be counteracted by encouragement toward
spontaneity, open expression =and= recognition = of basic feelings and
emotions.- The _major eo r -was _to su st several
ibilities or treatin a.nd mod fy ng =a reading program
ools to-enable - children == to = express an recognize tllese feelings
er an repress

REFERENCES

1. Besse% Harold. Human Development Program, Human Development Train-

ing Institute, 7255 Girard Aveilue,-La Jolla, Calif., 1969:
2. Yeomans, Edward. Education = For Initiative and Responsibility, National As-
sociation of Independent Schools, 4 Li re, Boston, Mass., 1968.
 Reading as Cognitive Fu: toning
RUSSELL G. STAUFFF.R
Universie; of Delaware

FOR MANY YEARS authorities have contended that reading is an in-

tellectual process akin to thinking. Much=speculation has been done
aboutEthe-concepts-and cognitiVi skills-used= in this process. Huey_
said-that when reading was done-for the attainment of the reader's
purposes, it became= = excellent - .practice_ in the higher -thought-pro-
cesses. The feeling-for values an oosing of the= relevant' re-
quires, he said; a mental_discipline ' blden practice =in the
training of judginent"-(5:363-364 ). Wein on to say that "real
reading"-= whereby the reader actively an s -pathetically follows
the ins-and-outs of an author's intentions; his fidelity to truth, his
accuracy and method - ":: Cannot but--train the mind to modes of
functioning that are similat=to his"-(1.5:365). By so doing it "acquaints
one with The= more= effective ways of thinking, and= develops them in
the reader . . ." (15:365).
Apparently not all _of= the variability_ of attainment in reading
is attributable to conditions within the learner. Some of it resulti
from differences existing among authors as reflected in their thiliking
and their writing. Both conditions, author and, reader differences,
are encompassed by a host of variables which affect success in compre-
hension and concept learning. If, as Bacon said, writing maketh the
accurate man, then the degree to which authors observe the disci-
plines of accuracy permits an imitation by readers that should be
cognitively productive. It would be dangerous to assume, however,
that reading the classics resulted in automatic acquisition of cognitive
skill by-products.
The admonishing of many scholars about the proper business
of schools, and particularly the teaching of reading, has until only
recently been negated largely by teaching practices that produce non-
thinking parrots and word callers. Many so-called authorities in
reading have been writing glowing accounts of the need for efficient
reading and the use of higher mental processes and at the same time
STAUFFER 125

attaching their names to reading instruction materials that cultivate

passive mastery. Recent advances in the understanding of concept
attainment and the 'maturation of- thought processes have great
promise and may offset such compromising.
Reading is a mental processa dynamic, active process and it
can be taught that way. TeaChing strategies that can elevate cognitive
effectiveness are being developed. As a result we need no longer wait
for the= superior children = to discover the strategies of thinking by
slowly and painstakingly examining the writing of scholars to deter-
mine their modes of =functioning and in the-meantime see the less
able either drop out or become semiliterate robot&

Language =and= Thought

The role of language in cognitive devewpment is undoubtedly
significant It is largely because of language - and = symbolization= that
most forms of cognitive functioning become possible. In this regard,
however, Piaget has been quite clear about the relationship between
language and intellectual operations.
The source of intellectual operations or cognitive functioning
resides in the sensorimotor period in =which the intellectual instru-
ments consist of percepts and movements (25: Chap. 3). Even though
objects and events are experienced only in their perceptual im-
mediacy, elementary forms of coniervation and operative reversibility
are to be found. For instance, recognition of the permanency of
objects represents a first invariant-Thus the changes that occur in
intellectual operations when language is acquired show that language
is not fully accountable for the transformations.
The beginning of representation and of representative schema-
tization or symbolic functioning in general appears at about the
same time as language. Symbolic or imaginative play is a source of
personal cognitive and affective representation -that is contempora-
neous with the appearance of language but is independent of it.
By the end of the sensorimotor period, thought with its roots in
action, has permitted a child to overcome initial perceptive and
motor egocentrism. He can attain practical aims by activity that
yields success cognitively speaking and pleasure effectively speaking.
Deferred imitation, or imitation that occurs in the absence of a
 126 THEORETICAL MODELS AND PROCESSES OF READING

model, and mental imagery, or the -sonorous imagery of- a voice

represent two other possible:links between sensorrinotor -behavior
and representative behavior. --
Sinclair-de-Zwart, alinguist on Piaget's staff, uses an aptillustra-
.; tion to distinguish between symbols: as signals and -as_:signs. The
former leads to what istignified-m-uth-astheipart-leads:th the-whole.
In addition they are usually pertonal.rSigns;-__on- the other hand, are
distinct =signifiers that are arbitrary and can form systems For
instance, a child pushing edgeof _a-:box-_-ancl= saying
''inetrie--illuttratet-si; =
ers. e;s e resemblance
with the objeet--(Cat)-arid--
Such --jets= of practical ;intelligence accompanied -by-words
guage)_ are thus ai__part -_oUamutti--largerzprocestis-cOnstittiting_intel-
._ lectULoperatiOns._Even--,thoilgk-lauguage-_ii,netessatily-interperional
ancLa--systenv of,arbitrarrsignsjhiefreiis the --indiVidtfirfromTthe-
immediate, itis-kpatticAiLaiqornr
Motor- schemata seem to7-be:ritlinportance_tiot only from the -_begin,_
ning ofsintelleetuali-operatio-riebuttontiniteirii_develop And structure
thought-rUP-tO--ther constitution of -fintiallOgic. --"It----itrlierinitsiblel-to-
Crindudei'!:Writes:Piaget_!'that=thOught=precedes_lafigUage and that
-language confines Itselfl_toiprofriuridly=transfOrMingthOttghtby--help-
ing-it attain i tt:formS of equilibrium by means otia_triore- advanced
schematization and a more mobile abstrattion'! (25: 9_1=92).:
While it appears-0-r be true _tilit-lintuagelandithelikermake pos- --

sible mOst---of the complex -formtotcognitisie-fUnctioning,- it is not the

source of all ,.:Oordinationt._ School learning:and-reading instruction
cannot oVerl, ioksuth truths. Many operations are basically coordina-
tiani among tetions -before they are trarisposedinto language:and the
operatiOns of thotight. Children can classify collections of ribjeets or
striate objects before they can do so linguistically.
The question is also whether-language is sufficient in and of it-
self to give rise to these- formal] operations cx nihilo, or-whether,
on the contrary, its role is limited to allowing the fulfillment_
of structuring which -originates from the systems of concrete
operations and, therefore, from the well-springs of action itself
(25:95).

Undoubtedly language permits symbolic condensation and social

STAUFF/1 127

regulation and the integration of actions into simultaneous-systems,

but also it is linked with actions in continuous reciprocity. Sinclair-
de-Zwart, after summarizing her studies on linguistic development
and research on deaf and blind children, = said results confirmed
Piaget's views that "... language is not the source of logic, but is on
the contrary structured by logic" (28:325).

Cognitive -Development
.
Cognitive di op m en 1s t a=c o ntin:uo uv_ con_ truct: o nof m teue.t =-
tual operations at eVolveto als1 ean mobile:systemization
essentially direct 'tot-yard- ibriinn;- tellectual-operations, or
the organization v int _activity, are rooted -in- action- and
are always a part tan of operatiOnsior structured wholes:At all
levels uf:deteelopinent-thereareconstant--- ctions-Cominon-
. . . action presupposes' a_precipita factor: a physiologic:4,
affective, of_ intellectual need; (In, elattet case, the need -ap-
pears in,the-guise of a question or a=probIem). ._. -inin addition
to the -constant functions, there- are the variable structures
(25:4-5).

The constant functions assure transition from oneleriod of develop-

znent to another and the variable structures _provide the organiza-
tional forms of- mental activity, intellectually as well as=affective'''.
The developing child acquires CoMplek Sets-of 'earnings based
on disCrimination, perception, transposition, and generalization. In
so doing he acquires Concepts and =a set of appropriate behaviors.
Concepts and proceises enable a child,- according to his capacities,
to cope with his environment, to organize his mental activities along
two dimensions-----intraliersonal and interpersonal (social), and -to
form attitudes that are emotional and intellectual.
The stages 'of intellectual development as described by Jean
Piaget provide a schematic description of developmental cognitive
changes that occur through time. His:stages are open-ended enough
to allow for the fact that children show different levels of ability,
knowledge, and skills as- a function of the rate and quality of the
learning experiences they encounter.
The first two years of life are described as the iensorimotor
 128 THEORETICAL MODELS AND PROCESSES OF READING

period. The infant, using the- inherent reflexes of his biological

endowment, interacts with his environment. The interplay of in,
ternal and external conditions through stimulation = and response
characterizes the normal = development of infants. Gradually from the
maze of undifferentiated, unreflective, and = unspecified experiences
the child attains rudimentary know_ ledge. To= accomplish this, he
Ile
establishes =a differentiation of himself from objects; he
localizes = himself in space; he establishes a nnmg awareness of
cause- and effect, of time an s (27:215 : = =By time most
children are eight to twelve- mon hey have shown intention
or a activity.- or t e mtention
now = begins to influence eitin teratt i OS: "ThiS =is =a big stride=intel=
leetually, arEawarentii4 _-'meanvend_Atlationshi =child
tnetipewith=_tht_physiolEati-di=stitialEcooipleititiesTo
end-of=thiarta
environment -symbolically =and-- conceptually can already-Invent
solotions-in bir-Minditathet!than-aerin- einfonifbftrial'ancLerrce
It is apparent,- then,= that in = the__first =fivo years- of_ -lifethildren
live _in a worl -albeit a _concrete:W-6r series Of -Situations.
ThejnteraCtiOn-- that irgoirign-on between, ild and-- physkal
and _Social:World=permits5linn- to =septaze h iMself--_from his__envirdn-
Inept as Well as-to_-realiie=o-at the environment-has certain properties
of space, location,-perthineike,--and-causalityiAncreasingly; he is7able
to operate symbolically` -by:clissa--iorigroups.7He_'cari tell= that a dog,
Silver, is a-Member of- thezclog_ family but -cannot deal with the cat=
egory animal.
The next-five-years-of life Piaget describes as the-preoperational
phase. In thii phase language_ plays" an increasingly important ink
arthe child acquires _concepts through-a_complexItt of processes. To
attain concepts be has to become increasingly sensitive -to objects in
his concrete world. He-has-to learn not only that-they-exist but, also,
that they have many characteriitics and attributes. In addition he
sees that diverse items_ can be organized into classes or categories
(Silver, collie, dog, animal, vertebrate) and that language can facil-
itate as well as direct the process.
At the sensorimotor period, a giant intellectual Stride notcu was
the influence of intentionnr purpose, the-sensing of mcOns-end-rela-
tionships. At the operational stage an equally significant stride is
STAUFFER 129

made as the- sensory -motor infant becomes, through symbolic func-

tioning, a manipulator of representations. The act of symbolic func-
tioning -is the result of= the generalized capacity to differentiate
between signifierssymbols which stand for something, and signif-
icatesthe objects. This representatianal intelligence through its
possession of symbolic functionii.g,sets the stage for =the upper limits
of cognition and the manipulation of reality.
At this stage, though, conceptualization is dominated by the
world of percept& The poten of h attributes to a large degree
deterinines the = concepts orm et calls this the_-prec-oncept
period because children rst- eve concepts primarily. They can
grasp the fact that-peaches an ars are f but cannot distinguish
between different pears. Or, -they_ canr_recOgnize-that certain_ Very
different things belong together: daddy's watch, daddy's chair,
daddy's haL -
In the four to seven = years of age -phase, increased symbolic
functioning is possible. Signslinguistic- signs--are acquired from
the social surroundings and are socially shared. Words-are the com-
monest signs of our codified and socially shared linguistic system.
The private signifiers or symbols- as well_ as =the early use of
linguistic signs provide the focus for:Piaget's saying that children are
egocentric They lack generally the ability to take =the role of another
person and to treat their own thought processes as the object of
thought. Over and over again Piaget indicates that it is in the context
of social interaction as a member of a learning group that a child,
forced to take cognizance of the ideas of others and forced to become
increasingly cognizant about his own thoughts, their reliability and
validity, emerges as a sociocentric objective scholar.
In addition to grasping the function of images. and signs as
signifiers, the child learns' to use them as- anticipative mediators of
future actions. Starting with imitative images which serve as antic-
ipative schema, the child begins to- direct future action. He begins
to evoke acts and deeds in thought, as opposed to actually carrying
them out in reality. This ability to anticipate, to look ahead, to con-
jecture, to speculate leads to the ability to hypothesize, to deal with
variants and covariants, and to test logically. In the life of a learn_ er
this ability signifies advance of the utmost importance. It is the
pattern of inquiry which George Kneller defines as, ". . to analyze
 130 THEORETICAL MODELS AND PROCESSES OF READING

the problem and to consider ways of dealing with itthat is, to set up
hypotheses" (22:42). Now the learner is becoming more reflective
and less impulsive. He is beginning to want proof, to suspend judg-
ment, to think of information as tentative and relative. Rather
Than seizing on the first idea-that occurs to him, he pauses (suspends
judgment) to note whether or not there- are better ways or Other
alternatives.
Another cognitive advance that occurs at this stage is the ability
to- use numbers; not only-to order things- in uanti 'but
also to see _that relationships can exist =on a nunreri
such= as the = number syStem-has pro ies - properties
that are agreed upon by mathematician& w o _can produ-ce
a sum deals with an abstractiori roperties Of
mathematics. Interestingly u- h; Piaget makes it quite dear that
the understanding of numbers oes hot learning numerals.
The latter part= of=tbe preoperational"stage- fnds_children= Making
judgments largely on asis o rtia* 1 and immediate percePtions
and/or on the basis of o jective s an They judge by the
way
things look and -usually in-terms=of just one of a number of
relevant
dimension& Even so, three= fundamental = operations= can be deter-
mined. They can think- in =terms =of classe& When ptesented with a
group of circles or squares, they can classify the items on the basis
of roundness. They can think in terms of relationships; i.e., Mr.
Jones is the father of Ralph, Mt Jones is = bigger than Ralph, and
Ralph is the oldest of three children. They can think in terms of
quantity or by handling number concept&
In the concrete operations period the thought of the
seven to
eleven age group is more like that of the adult in= that they = think
more in logical terms. Operations is used by Piaget to refer to mental
acts or imminent acts of an internalized nature and taking place in
the mind. These mental acts represent a process of interaction
and
development. whereby new syntheses are formed by discovery. At-
tributes are noted, objects are classified, and categories determined.
These syntheses are real in the sense that they not only have a loca-
tion in time and space but also that they take place in the minds of
human beings. In the process of cognitive growth through
discovery
and synthesis the individual is merely the neural medium 'in the
resynthesis of cultural elements.
 STAUFFER 131

Three significant operations described by Piaget are reversibility

as in arithmetic (2 + 3 = 5 or 5 3 =-- 2), classification or the
organization of objects into classes (desk, chair, table = furniture),
and serration or arranging ideas along a spectrum of increasing values
(2, 4, 8, 16, 32): In brief at this stage the child is able to treat objects
as alike (desk-chair) (furniture) even though different, to note that
they can be in more than one dass and that come classes can be sub-
ordinate to others, and to count one =item as first -and == another =as
second.
In addition the child has =to - understand the concept of conserva-
tion. In other= words, he has to see at certain properties= of objects
such as quantity can remain invariant even in ce of certain
changes. For instance, two= circles with a diameter of six inches re-
main alike even though one is cut into quarters and the other is cut
into thirds. Cutting the circles_ doesn't= alter= the amount or quantity
of the circle.
To arrange items in-a series along a continuum a child must
grasp the principle of transitivity. He must _understand such ordering
whereby he recognizes that if A is larger than B, and `B is larger than
C, then A is larger than C.
In this concrete operations stage, even though the child's think-
ing may be logical and systematic, his thoughts= are limited to the
direct experiences he has had. When he has no direct experience, he
tends to reason by analogy to something he has experienced. In this
regard, the crucial element may be verbal ability, as well as physical
activity and social interactions with verbal ability, acting as =a support
to help a child overcome the influence of his visual perceptions.
While training designed to increase the appropriate vocabulary may
facilitate the development of logical thinking and help resolve the
perceptual-cognitive conflict, it is "equilibration" or self-regulation
that takes on greatest significance. To permit a child ". . . to learn
an appropriate answer without making certain that he can retrace
F his steps, or arrive at the same result in another way, is to encourage
the erection of a verbal superstructure that may crumble under even
minimal cognitive stress" (1:123).
The fourth stage or that of formal operations is the time when
abstract thinking develops. The child enters this stage at about the
beginning of adolescence. Now he begins to grasp the ability to deal
 THEORETICAL MODELS AND PROCESSES OF READING

with the possible without reference to the actual; he begins to

grasp the complexity of human knowledge by learning how to con-
struct theories and make logical deductions about their consequences
without the need for empirical evidence. As J. McV. Hunt puts it,
". . . instead of observation directing thought . . . the adolescent's
thought directs his observing" (17:355). In all this, language, or repre-
sentational thought, plays an important role, but Piaget is of the
opinion that ability to use language to express logic is an outcome
of activity, and that attempts to improve a child's logic by teaching
him in the-use of language-is apt not to-be very successful.
It seems then = that language development as a -part of maturation
or all round mental capacity influences much' of the child's progress,
from thought that is predominantly = perceptual and = intuitive to
thought that is more conceptual and logical. A child's verbal ac-
=

commodation to a learning experience is helpful but it will produce

last_ ing effects only if, through further-self-regulation; generalization
to other tasks has resulted. It is= not enough jtist to have had an
experience, even verbally, unless it affects a child's way of organizing
his experiences.
Finally, as Piaget has declared,_the key factors in the transition
from one level of thought to the next involve maturation, social'
interaction, physical activity, and, moss t important, the process of
equilibration or self-regulation.
Children's cognitive actions and = interactions, intellectually and
affectively, as described by Piaget, suggest that reading ability to the
degree that it is in nature represents similar potential. It
remains for us to account for the reading-thinking processes in as
definitive and astute a manner as he did. Reading tasks Structured
carefully both syntactically and semantically may reveal the "how"
of the reading-thinking act and the "why" of different strategies for
attainment and assimilation. Developmental stages may be deter-
mined and reading materials prepared to foster growth in subtle
and mobile skill acquisition and functioning.
Piaget's theory of cognitive development advanced over the
past half-century reflects an empiricist-idealist base and has only
recently met with wide acceptance in the United States. He starts
from the central postulate that action (motor) is adaptive and is the
STAUFFER 133

source from Which mental operations emerge. Intelligence is viewed

as an adaptation, an organizing activity whereby there occurs a pro-
gressive balancing of increasingly complex forms under the impact
of experience. His work in cognitive development and critical think-
ing now represents a major influence upon basic research- in psy-
chology and education.

Concept = Learning
Contept attainment -is now generally-considered =a =part of the
psychology of- l--earning and the development of cognitive processes
(12:37). Attention is focused on the logical form of concepts with
studies of concept attainment = generally based on inductive methods
and the strategies used (6:37). Strategies differ from person to person,
from discipline to discipline, and from one level of sophistication to
another. While many = concepts are acquired by discovery, learning
through discrimination, abstraction,= differentiation, hypothesis gen-
eration and testing, and - generalization, many more are acquired
through school learning and/or reading. In the latter, the concepts
are learned by means of criterial attributes presented and the relat-
ing of= them to established ideas (2). Thus the acquisition of concepts
can be accomplished inductively by concept formation or deductively
by concept assimilation. Even so, there is a considerable likelihood
that the learner must use much of= the same processes of concept
formation even when appropriate contexts are presentee d as in con-
cept assimilation.
At the Wisconsin Research and Development Center for Cogni-
tive Learning (20), Concept is treated as a superordinate category of
which all concepts are instances and differentiated from other pro-
ducts of learning such as facts, principles, and problem- solving skills.
A concept is referred to as having four characteristicsdefinability,
structure, psychological meaningfulness, and utility. Four bases of
defining concepts are identified in terms of perceptible defining
properties, semantically, operationally, and logical or numerical
relationships or axioms. Structure is determined by the form in
which the concepts are experienced. Psychological meaningfulness
refers to the phenomenological or idiosyncratic nature of concepts
 134 THEORETICAL MODELS AND PROCESSES OF READING

and the internal representations held by an individual. Utility is re-

lated to the use of a concept which is determined primarily by how
well an individual has formed a concept.
Bruner (6) -in his Study of Thinking defined strategies as regu-
larities in decision-making and indicated that they provided the
basis for making inferences about the mental processes involved in
concept learning. He identified four strategies under the selection
paradigm and two in the reception paradigm Under the former= he
defined the strategies as 1 simultaneous scanning,-2) successive=scan-
ning, :3) conservative focusing, an ocUs gambling Under =the
latter he labeled them as 1) wholist an partist. Byers'== study =(7)-
done = at the =Wisconsin Center designated e strategies used as one
of three variants of a conservative focusing ttrat or as one of two
variants of =a focus gambling grate
Byers found that practice modi e e probability =with which
subjects used various straw 'es. Use o -conservative focusing strat-
egies increased while use -of= others= = decreased.= Attempts at instruc-
tion in the use of strategies (21:32) showed that subjects were readily
taught the-conservative focusing strategy and did better than those
not taught. On the other hand they = could = not be taught to use focus
gambling strategies Consistently and tended toward use of conserva-
tive focusing strategies:
Concepts are utilized when reading. The meanings of previously
learned concepts and propositions are perceived and dealt with and
the acquisition of new concepts is facilitate& The cognitive processes
of concept attainment and concept assimilation are used most likely
both in simple and more complex varieties of reading.
In addition = it may be noted that insofar as the central role of
cognitive variables are concerned the- distinction between forma-
tion and utilization on the one- hand, and problem solving on the
other, becomes less definitive- (16, 18, 19). This is true of concepts
acquired by discovery as well as by meaningful reception learning. -It
is true of simple as well as complex problem solving.
Reading is a form of problem solving in much the same way as
is concept development. All threeconcept attainment, problem
solving, readingare active cognitive processes of seeking relation-
ship to, differentiating from, and reconciling with, existing ideas,
and the processes therefore overlap in many ways. Some of the princi-
STAUFFIrit 135

pal ways are hypothesis-generating and testing, -_-abstracting and

generalizing. The efficient reader-reads with a purpose, abstracts in,
formation, tests its value, and then accepts- or rejects.
Singer (29, 30), after= reviewing the literature- on con_ ceptualiza=
tion and reading_behavior,_andin an examination of--the variance-in
the substrata- theory of reading, -said- that the formation' and-:=Use of
concepts _needed' to= he- aCcounted-- for as it enteiredfintd=the -develop-
ment -arid dynamics --ageneral reading-
using ia-battery__43f-dnucaltests fol-CoriceOtforthation_rndci*paiinu
" achieving= readers" -and - '-'nonreaders; " = concluded that capacity for ==
Ccinceptualiiation,was:Spedifitally=telatedttoreading-fa
As is alreadritatek to-;read-=re-cptirest airea =to- -employ- the
concepts he-7has,-=acquired-=_and-_proVideS-opportuni six attain= new= =
concepts and remodel-old-ones. -Theatt of iaCquitingconi:eptszbeing ==
so Much -like-that Of --problem=solvintantt *turn-like-reading,- The
-analysis- of _concept acquisition SuggestiVe--abourways=
of analyzing the-reading-thinking 'act =-

Critical Thinking- and Teaching Strategies

Even though Huey (13), in 1913; stated- the case for critical
reading in a professional text =on the teaching of reading, his account
did not have the impact on thinking about = reading as did Edward
L Thorndike's "Reading as Reasoning" report (Thorndike con-
cluded in 1917 (35:329) that
. . . understanding a paragraph is like solving a- problem in
mathematics. It- consists in selecting the right elements of the
situation and putting them together in the right relations, and
also with the right amount_ of weight or influence or force for
each. . . . all under the influence of the right mental set or pur-
pose -or demand.
More recently Ennis (9, 10) has-said that critical reading is the
use of critical thinking in the act of reading. Russell (26) main-
tained a similar position and defined critical thinking as a three
factor ability. He included an attitudinal factor of questioning and
suspending judgment, a functional factor -of logical inquiry and
problem solving, and a judgmental factor in terms of some norm
or consensus.
 136 THEORETICAL MODELS AND PROCESSES OF READING

A significant study by Taba (34) was concerned with teaching

strategies and thought processes. Her multidimensional analysis of
classroom transactions in terms of measurable changes in levels of
thinking had several advintages. Results showed- that children can
learn to make inferences to generalize, and to make logical assump-
tions if they receive systematic instruction. The enormous influence
of teacher behavior on the of=students was most impressive.
The Productive Thinking- program of Covington -and-- Crutch-
field-=.(s)_ slitiived that_ instructed children= =were`_more- willing- and
able Witiaketide- of- the cognitive find' Stilt es _to
both=creative problem,solvin r=an -to= discerriing7-,reflective-readirig.
They developed _a- general = problem gOlving-progratn= of= =16= self.
contained =problem?Aolving:iepis es; teitive- - problem solving
strategies were taught as-well=as_a-Enum o strategies.
A icomprehenSive study--i-done =at -Ohin State= University had _ as
the triajtn%purpose-to ndetermine-whether or =rot= children in "the ele-
Mentarygrades could be taught to read critically. This:required the
deVelopthent of-an extensive :Operational definition--of_critical =read-
ing and the identification-of=specific Skills. It alsorequired the de-
velopMent- of=anlobserv-ation ifittrument-fOr Classrootn use Bloom's
approach (5): to ways-of-Ordering-knowledge-influenced the develop-
ment of a classification_ System for -teacher'S verbal behavior and
Guilford's structure (14) of the intellett proved useful in determin-
ing =the separate types of thinking Of the pupils. Twelve teaching
units were developed and a test labeled the Ohio State:University
Critical Reading Test (37); Results = indicated that teaching critical
reading is= feasible to children of both -sexes and that achievement
is influenced by intelligence,general reading ability, and personality,
and that teaching skill, especially :the ability 'to-ask- questions and
interpret pupil responses was a key .faCtor (36). In this latter respect
it is interesting to note that Gallagher (13) showed how the questions
a teacher asks determined the kind of thinking the student did.
Ennis (9, 10), in- a project on critical .thinking, set as his goal
to contribute to knowledge about what critical reading is and about
when it can be taught. The report covering the first phase dealt
with deductive logic in adolescents. He defined logic as that part
of critical thinking which deals with whether a conclusion follows
necessarily from the premises that are offered in Support of it. Of the
ST.AUFFER 137

types of logic th t exist he studied two: conditional logic and class

logic. In general- ne concluded that progress in mastering class logic
and conditional logic could- be made although instructional time
required varied.
Suchman (32, 33), has been experimenting with the teaching of
strategies and tactics of scientific inquiry to children and devised a
method known as inquiry training. He states that inquiry is "the
pursuit o& meaning" and= his major- emphasis appears to be on the
means by which- knowledge is acquired. While the training- in-
creased the numer of valid questions asked,-he found no- significant
differences between two groups.
A longitudinal study to learn the nature and direction of changes
from freshman to senior year in critical= thinking ability, in attitudes
of stereotypes and dogmatism, and in traditional value orientation
was done at Michigan State University. There were marked changes
in critical thinking ability, attitudes, and values from the freshman
year to the senior year with the changes in critical thinking of great-
est magnitude occurring during the freshinan year (24): It m4,0 .t be
concluded-that students come to college with the hope that centers
of higher learning are- also centers that foster and require critical
thinking. However, by the end of the freshman year they have dis-
covered apparently that conformity and intellectual bondage win the
higher grades and please the professors.
Berlyne (4:19) concentrated on directed thinking an_ d defines its
function as "to convey us to solution of problems." In so doing
directed thinking involves both epistemic. behavior and symbolic
behavior. Such thinking is launched by a "felt difficulty," a problem,
a questiOn, a conflict, uncertainty, or lisequilibrium and is in turn
motivated- thereby. The native propensity of the mind to ask "why"
from age three on and its compelling force are still far from under-
stood psychologically or physiologically, but there seems to be little
doubt about the potency of the desire for equilib..ium in the func-
tion of learning and thinking and their responsibility for adaptive
change.
Strategies of thinking, of problem solving, bf concept attain-
ment must be learned and therefore can be taught. Similarly, strat-
egies for effective reading-thinking must be learned and can be
taught. It is significant and directive to-note that a basic operational
 138 THEORETICAL MODELS AND PROCESSES OF READING

mechanism in problem solving and in concept formation is a prob-

lem, or a question or a mental set,- and that a question well asked
can be half the answer. Similarly it is the purpose of a reader that
determines not only his rate of reading but also the nature and
depth of his achievement. At the same time the most essential ped-
agogical skill is the teacher's resourcefulness in the art of questioning.
As Piaget has said, "It is so hard not to talk too much when question-
ing." The unequivocal-role of "the question" in learning as well as
in the directing of leaniing is a noin-t that hasinot.been sufficiently
appreciated. U_ ndoubtedly:there can be no learning without-a prob-
lem and it is in this regard that the studies reviewed here are highly
suggestive.

Conclusion
Sam A. Fleming (//: "4) said
Dedication and clear thinking are= our need today. It is ironical
that in this period of so much 'learning . . . a- specter of igno-
rance should hang over us . . . that there should be iiresolution
about many of the true values whiCh are fundamental to our
way of life.

Miiih the same might be- said about what is reading and how to
teach reading. Universal agreement can be obtained supporting the
conclusion that comprehension is the invariant condition of read-
ing. Almost universal agreement can be obtained regarding the
conclusion that reading is a process akin .to thinking. Some- few
agree that if the first two premises are true then reading should be
taught as a thinking process (31).
This review of theories and practices concerned with cognitive
functioning and development and its possible relationship with
reading per se and with the teaching of reading may prove helpful if
it will stimulate research that will define ways and means to more
effective reading. Specific thought processes involved in reading for
various purposes and with varied materials should be identified.
There seems little doubt about the nature of the strategies in-
volved in reading-thinking acts. Increase in task complexity most
likely requires cognitive functioning that ranges in complexity from
STAUFFER 139

stage to stage of maturity similar to the stages Piaget and others have
declared. Like related intellectual tasks such as concept attainment,
reading requires of the reader problem solving ability that is logical
and mobile.- Undoubtedly, too, achievement is influenced by a
reader's intellectual potential, his attitudes and values, his intra-
personal and interpersonal dimensions, and the teaching and testing
to which he is exposed. Even though at the college level a renewed
attempt at critical thinking is made by college freshmen, pedagogical
demands do not foster similar changes across the four-years.
Children deal with means-ends relationships as early as the sen-
sorimotor period. They learn to deal with. variables selectively and
to act reflectively in the preoperational period. They make decisions
on logical terms through immanent acts of =an internalized nature.
This they do by discovery and synthesis as they resolve perceptual-
cognitive conflicts of the operational period. Finally at the logical
stage they can construct theories and make sound deductions without
the need for empirical evidenceAnd_in_all this, language (oral or
written), or representational thought plays a= highly significant role.
If reading is akin to thinking and represents a means of generat-
ing predictive systems, and if science is- a search for relationships,
then the science of critical reading 'requires that such actions be
central. We cannot be satisfied with passive reading accomplished
through passive processes and directed by passive teaching and least
of all at the college level. We must require that reading be a think-
ing act and teach it that way. If thought has its roots in action then
reading does too within a developmental interactionist theory of
cognitive activity.

REFERENCES

1. Almy, Millie, Edward Chittenden, Paula Miller. Young Children's Think-

ing. New York: Teachers College, Columbia University, 1966.
2. Ausubel, David P. Educational Psychology, A Cognitive View. New York:
Holt, Rinehart and Winston, 1968.
3. Berlyne, D. E. "Condition of Prequestion and Retention of Meaningful
Material," Jouinal oj Educational Psychology, 57 (1966), 128-132.
4. Berlyne, D, E. Structure and Direction in Thinking. New York: John Wiley,
1966.
5. Bloora, Benjamin et al. Taxonomy of Educational Objectives Handbook I:
Cognitive Domain. New York: David McKay, 1956.
 140 THEORETICAL MODELS AND PROCESSES OF READING

6. Bruner, Jerome S., Jacqueline J. GoodnOw, and George A. Austin. A Study

of Thinking. New York: John Wiley, 1956.
7. Byers, J. L "Strategies and Learning Set in Concept Attainment," unpub-
lished doctoral dissertation, University of Wisconsin, 1961.
8. Covington, Martin V. "Training for Creative Understanding in Reading:
Some Experimental Evidence," in Russell G. Stauffer (Ed.), Reading and
the Cognitive Processes, Highlights of the 1966 Preconvention Institutes.
Newark, Del.: International Reading Association, 1967, 13-22.
9. Ennis, Robert. "A Concept of- Critical Thinking," Harvard Educational
Review, 32 (Winter 1962), 81-111.
10. Ennis, Robert. "Critical Thinking_Readiness in Grades 1-12," Cooperative
Research Project No. 1680. Ithaca. N.Y.: Cornell University, April 1962.
11. Fleming, Sam M. -"The Specter of Ignorance," Phi Kappa Phi Journal,
Winter 1969, 9.16.
12. Gagne, Robert M. Conditions of Learning. New York: Holt, Rinehart and
Winston, 1965
13. Gallagher, James J. Teaching the Gifted Child. Boston: Allyn and Bacon,
1964.
14. Guilford, J. P. "Intellectual Factors in Productive Thinking," in Mary
Jane
Ashner and Charles-L--Bish- (Eds.), -Phiductive Thinking in Education.
National Education Association, 1965.
15. Huey, Edmund Burke. The Psychology and Pedagogy
of Reading. New
York: Macmillan, 1913.
16. Hunt, Earl B. Concept Learning: An Information Processing Problem.
New
York: John Wiley, 1962.
17. Hunt, J. McV. Intelligence and Experience. New York: Ronald
Press, 1961.
18. Kendler, Howard H., and Tracy S. Kendler. "Vertical and
Horizontal Pro-
cesses in"Problem Solving," Psychological Review, 69 (1962), 1-16.
19. Kendler, Tracy S. "Learning- Development and Thinking," in E.
Harris
(Ed.), Fundamentals of Psychology: The Psychology of Thinking, Annals
the New York Academy of Science, 90 (1960), 52.56. of
20. Klausmeier, Herbert-J., et al. "Strategies and Cognitive Processes
in Concept
Learning," final report, Project No. 2850, Contract O. E. 5.10.143.
Washing-
ton, D. C.: Bureau of Research, U. S. Department of Health, Education and
Welfare, 1968.
21. Klausmeier, Herbert J., Chester W. Harris, and W. Wiersma.
"Strategies of
Learning and Efficiency of Concept Attainment by Individuals and
Groups,"
Cooperative Research Project No. 1442, University of Wisconsin
1964.
at Madison,
22. Kneller, George F. Logic and Language of Education.
Wiley, 1966. New York: John
23. Kress. Roy A. "An Investigation of the Relationship between
Formation and Achievement in Concept
Reading," unpublished doctoral dissertation,
Temple University, 1955.
STAUFFER 141 i

24. Lehmann, Irvin J., and Paul L. Dressel. "Critical Thinking Attitudes Values
in Higher Education," Office of Education, U. S. Department of Health,
Education and Welfare, Cooperative Research Project No. 590, Michigan
State University, 1962.
25. Piaget, Jean. Six Psychological Studies, David Elkind (Ed.). New York:
Random House, 1967.
26. Russell, David H. Children Learn to Read. Boston: Ginn, 1961.
27. Sigel, Irving E. "The Attainment of Concepts," in Martin L. Hoffman and
Lois Wladis Hoffman (Eds.), Review of Child Development Research. New
York: Russell Sage Foundation, 1964, 209-248.
28. Sinclair-de-Zwart, Herinina. "Developmental Psycholinguistics," In David
Elkind and John H. Flavell (Eds.), Studies in Cognitive Development: Essays
in Honor of Jean Piaget. New York: Oxford University Press, 1969, 315-336.
29. Singer, Harry. "Conceptual Ability in the Substrata Theory of Reading,"
unpublished doctoral dissertation, University of California at Berkeley, 1960.
30. Singer, Harry. "Substrata-Factor Theory of Reading: Theoretical Design for
Teaching Reading," in J. A. Figurel (Ed.), Challenge and Experiment in
Reading, Proceedings= of the International Reading Association, 7, 1962.
New York: Scholastic Magazines, 226232.
-31. Stauffer, Russell G. Directing Reading-Maturity-as-a Cognitive Process. New
York: Harper and Row, 1969.
32. Suchman, J. Richard. "Inquiry Training in the Elementary School," Science
Teacher, 27 (1960), 42-47.
33. Suchman, J. Richard. "A Model for the Analysis of Inquiry," in Herbert
J. Kliusmeier and Chester W. Harris (Eds.), Analyses of Concept Learning.
New York: Academic Press, 1966.
34. Taba, Hilda, Samuel Levine, and Freeman F. Elzey. "Thinking in Elemen-
tary School- Children," U. S. Office of Education Cooperative Research
Project No. 1574, San Francisco State College, April 1964:
35. Thorndike, Edward L. "Reading as Reasoning: A Study of Mistakes in
Paragraph Reading," Journal of Educational Psychology, 8 (June 1917),
323-332, .
36. Wolfe, Willavene, Charlotte S. Huck, and Martha L. King. "Critical Read-
ing Ability of Elementary School Children," Office of Education, U. S.
Department of Health, Education and Welfare, Project No. 5-1040, Contract
No. OE-4-10.187, Ohio State University, 1967.
37. Wolfe, Willavene, et al. The Ohio State University Test of Critical Reading
Ability. Columbus, Ohio: Ohio State University Press, forthcoming.
 Reactions to Reading as Cognitive Functioning
Roy A. KRESS
Temple University

corrriNunqc STUDY of the development of cognitive-processes has re-

vealed- much- of significance about _childten!s_learning_in general
and has many implications for their learning of the reading process
and their growth toward real maturity -in reading. However, dis-
satisfaction continues with the current status of reading instruction.
Perhaps the key lies, at least partially, in these statements from
Stauffer's conclusion:
_Universal agreement can be Obtained supporting the conclusion
that comprehension is the invariant condition of reading. Almost
universal agreement can beRibtained regarding the conclusion
that reading is a process akin to thinking. Some few agree that
if the first two premises are true, then reading should be taught
as a thinking process.
The implication seems strong that there is a divergence between the
product which is expected and the process by which it might be
reached. It is as if the child were expected to apply in something
called reading cognitive leamigigs which he has acquired in other
situations, without guidance in cognitive functioning as an essential
ingredient of his reading- instruction.
Characteristically, the learner has been askedto master a series
of skills and abilities, understandings and attitudes and, by applying
these appropriately, to readto- deal thoughtfully with printed
material. Such an expectation seems unrealistic in terms of what
is known about "ng and thinking. The problem appears to be
that the c. :4 five func toning which is necessary for the mastery
and abilities as not been considered. It is the thesis of
this paper that comple, thinking processes are involved in the
acquisition of the-components of reading ability and in decisions
about their appropritte application. Unless a child has thought his
way to a particular ability, it is unlikely that it will ever be his
142
 KRESS 143

to use spontaneously when it is called for. Certain of the basic under-

standings which are necessary for successful reading, fortunately,
seem to .arise -from children's own thinking activities with little or
no structured guidance from either parent's or teachers. From con-
sideration of his own observations about labels on cans and packages,
his name- on artides of clothing or toys or envelopes, and myriad
other experiences with visual symbols of language- which are already
his at -the aural-oral level, the child usually generalizes' that -these
*ism"( symbols arfb another_way_ofrepiesattingthe things and people
and ideas which make up his world. He may not be so- fortunate,
however, in independently thinking his way to all the other- things
he needs to be a mature reader at his level of- general development.
Therefore, an example of how he might be helped to think his way
to a serviceable word analysis skill or a particular comprehension
ability seems in order.
Word-form analysis skills
Before a child can become proficient in any word analysis task
which requires association of a visual- representation (letur or
letter combination) with- a particular speech sound or combination
of speech sounds, he must have made that associatioq. jHow can he
most effectively accomplish this task? It is obviously possible for him
to acquire the association, at one level, on a simple conditioning
basis. He can be told repeatedly, in a variety of ways, that a partic-
ular visual symbol is likely to represent a particular speech sound
and he can usually be counted on to respond with one member of
the associated pair when he is given the other as a stimulus. How-
ever, this stimulus-response type of- conditioning frequently fails
to function spontaneously in a real word recognition situation. On
the other hand, if the child thinks his way to the association, it be-
comes a part, of him which operates appropriately across a broad
spectrum of situations. His thinking his way to this kind of associa-
tion results from his observations of known words in which- he hears
and "feels" the speech sounds as he listens to and says the words. He
perceives the visual likenesses and the phonetic likenesses in the
words as he develops the ability to abstract them from the total
words. Ultimately he generalizes that when he sees a particular visual
pattern in a word, it is likely to represent a particular speech ele-

tfi
 144 THEORETICAL MODELS AND PROCESSES OF READING

ment. It is the teacher's responsibility to see that when he has

sufficient raw material, in terms of known words, his observation is
guided so that he does perceive the likenesses and his thinking guided
by appropriate questions so that he is led to generalize. A further
responsibility is that of assuring adequate practice in the application
of the generalization to reduce it to the level of virtually automatic
response.
Although a- child of the Quinn family does not come with- a
built-in knowledge that qu- usually represents a- blend- of the -con-
sonant sounds normally associated with k and w, he has, as- soon as
he knows his family name, a first step toward that knowledge. As
he learns to read the words quack, quart, queen, quick, and quit,
for example, he is building up the body of raw material which will
allow him to discover this knowledge. Through his work in auditory
perception and discrimination, he will have learned to recognize that
all these words start the same way his last name does. To assure that
henow observes that-all- these -known words-have, also, the same
visual pattern, -the same letters, at the beginning, the teacher can
structure the 'situation. Placing all the words -in a list in which the
qu's appear in exact vertical alignment and asking him to see in
what way they all look alike forces the focus of his attention to the
basic similarity. He reads them all;- he knows they all begin_
egin with
qu-. Now he can be questioned about what he would expect of an-
other word with qu and led to generalize that it will probably have
the sounds with which his latt name begins. He can be confronted
with other words (in which all other elements are known) with
which he can test out his generalization. Whether or not he knows,
at this point, that these blended sounds are normally represented by
k and w is actually immaterial. He still has a serviceable analysis
skill. When he has the proper resources in terms of other observa-
tions and generalizations, he will be able fo add the usual representa-
tion to his generalization. The important point is that he has been
helped to observe raw materials of language with which he is already
secure, has abstracted similarities from these' raw materials, and
arrived at his generalization on the basis of his thinking. He has
learned the word analysis skill by using his thinking processes and
It is, therefore, a part of himto serve him as he needs it.
 145

Comprehension abilities
The development of any one of the commonly listed reading
comprehension abilities must proceed along similar lines. As an ex-
ample, how can a child most effectively learn to respond to and in-
terpret metaphors? The essential process that he must learn is to
recognize when an author is labeling someone or something in a way
that implies a similarity to something else and to figure out what the
intended similarity is To learn this process he.must again think his
-way-by observing aiuMfialrzitiy anguage situa ions w is are-fami-
liar to him_ , which he already understands. In these he can see that
the label is ridiculous if it is interpreted literally. He knows per-
fectly well that there is no -intention = to say his sister is a= rodent
_

with a long, skinny tail when he says,- "She's a fad" Since she is a
human being, albeit a rather distasteful one to him at the moment,
it is patently impossible for her to be some other kind of animal.
The likeness that he is implying may be only the distastefulness to
him of both his sister and a rat, nothing more specific. When he uses
the metaphor, he knows what he means. In the same fashion, he has
learned to know what his mother means when she says, "Don't storm
at me! I didn't break it." He tells someone he got an icy stare or a
cold shoulder with no sign of shivering.
The fact that the child has learned to interpret and to use
certain metaphoric language in- slang, in name calling, and in a
variety of other ways does not, however, guarantee that he is pre-
pared to interpret new instances of metaphor which he meets. Of
course, even recognition of the need for nonliteral interpretation
demands an awareness of the literal meaning. Without this, the
child might not realize that a literal translation was not intended,
could not actually hold up logically. Beyond this, he must have ob-
served many similar experiences with the particular kind of met-
aphor and observed the signals inherent in them. He must have
generalized his observations to the degree that he knows the essential
similarity intended can be a physical characteristic or something
else. It can be based on some reality about the compared object or
on some myth about it. It can- reflect the personal attitudes of the
author of the metaphor towarchhe things being compared. All this
 146 THEORETICAL MODELS AND PROCESSES OF READING

he cannot learn at once. He needs to think his way through one kind
of situation at a time.
The effective teacher, therefore, must again help him by struc-
turing the raw materials for his observation and analysis. .One group
of metaphors f;miliar to him may all rest on some physical likeness
in the compared objects (beanpole, giant, midget, squirt). Working
with these, he can be helped to see that this is one avenue of ap-
proach to the meaning of a metaphor how do these things look or
sound or feel alike ? From another group of familiar metaphors he
may abstract another kind of similarity =and generalize some other
possibilities for arriving at .an interpretation. Only after a series of
such experiences in thinking his way through a variety of types of
metaphoric situations can.he generalize at the level which will make
him master of the process of interpreting all metaphors.
Conclusion
It-is not-too surprising-thatin-considerations -of cognitive func--
timing in reading far more attention has been given to dealing
with specific ideas in reading materials than to learning the compo-
nents of reading ability. Early investigations of children's concepts
followed the same patterndealing with the question of what con-
cepts they had and had not developed to the neglect- of the processes
of concept formation and attainment. The day is gone, however,
when attention can legitimately be given to the products without full
study of the processes by which they are produced. The child's
ability to deal cognitively with reading materials will be deter-
mined by the degree to which he has learned reading as a cognitive
process.
Theoretical Models of Reading:
Implications for Teaching and Research
HARRY SINGER
University of California at Riverside

Arrrrovnirthe-term-"model"---has-multiple-meanings, Kingston (59)

defines a scientific model as an analogue of the thing being modeled
and states that a scientific model is useful only when it is isomorphic
or similar to something, but such a model is not a "synonym for
theory." Lachman (62) agrees that a model is a separate system from
a formal theory and asserts that, nore than one model generally func-
tions for a theory. Models are categorized by Lachman into four
types:1) representational, a new way of thinking about objects and
events, e.g. the model for conditioning theory; 2) inferential, a
system- of relationships and rules by which theoretical symbols are
manipulated to arrive at new relations, e.g. rules for inferring one
sentence from another; 3) interpretational, an explanation and test
of a theory in terms of a model, linking theory to experiment, as in
the establishment of empirical definitions of a theory; 4) pictorial
visualization, reproduction of a theoretical construct in.terms of a
visual image, a first step towards theory construction or useful at
least as a didactic technique.
For the field of reading, Robinson (73) has suggested that con-
fusion might be reduced if models would besubgrouped into three
categories: models representing 1) = theories or procedures of teach-
ing, 2) processes utilized or mobilized in reading, and 8) skills and
abilities required for reading attainment. This tripartite division
would provide suitable categories for the teaching machine model,
i.t.a. model, linguistic model, substrata-factor model, and mixed
dominance model identified by Holmes and Singer (52). But Robin:
son's tripartite division is not mutually exclusive. For example, the
teaching machine model, although obviously a member of the teach-
ing category, could also be placed into -the process category, espe-
cially when learning or acquisition of responses is emphasized. Also,
147
 148 THEORETICAL MODELS AND PROCESSES OF READING

in the construction of a program for the teaching machine, an hier-

archical model of skills and abilities has to be followed implicitly
or explicitly. Nevertheless, this tripartite division is a useful one
not only for differentiating the categories but also for understanding
their dynamic relationships. For example, the teaching machine
model contains a program based upon a logical organization of
skills and abilities, which through a process of learning becomes in-
corporated into the mental repertoire of a group of individuals, and
this product of learning can be empirically and statistically con-
structed into a model of knowledge, abilities, and capacities that
can be mobilized for attaining speed and power of reading (44, 51,
53, 79, 88). Such a model =identifies= the essential variables and
processes that the average individual can mobilize in response to a
printed verbal stimuli; the model also depicts the general organiza-
tion or interrelationships among these vark.oles. These definitions
of models. will be used in reviewing the literature.

Review of the Literature

Since 1960, a large number of models have been explicitly and
implicitly formulated (20, 52,57,78,94, 101). Many of these models
can be organized under the rubric of a logical organization of skills
and abilities or processes for teaching purposes (18,29,40,73, 77, 99).
Some can be categorized as process= models based on psycholinguis-
tic theory (39, 75, 94), or neurological theory (98). Some models
deal with a part of the process of reading, such as- perceptual
(32, 33), cognitive (58, 90, 99), or an interaction between physiologi-
cal and cognitive processes (21, 22), while others are intended to be
comprehensive (18, 66, 99, 104). Although the substrata-factor
theory of reading was intended to be a comprehensive theory, models
representing it are a cross section of factors that could be mobilized
at a particular grade level for a- particular criterion of reading.
While all of the models listed may be based upon some empir-
ical data, only a few have been empirically or statistically con-
structed (21, 22, 44, 51, 53, 61, 79).1

1The basic assumptions underlying these models were made explicit by

Holmes and critically evaluated by Sparks and Mittel (103) and Raygor (71).
SINGER 149

From this extensive list of models, several will be presented in

detail. Because models of reading must and do change qualitatively
and quantitatively to fit developmental changes in the reader (56,
88), the -set selected for detailed presentation tends to span the
developmental continuum from kindergarten or the initial stage of
reading to the college level or the mature stage of reading develop-
ment. Although not comprehensive at each level nor continuous,
the models tap segments of the developmental continuum which can
be described in the following way (94):
An average child's receptive, mediational, storage, and oral
subsystems- for processing and responding to spoken language
are fairly well deVeloped before he systematically starts" to form
his subsystems for decoding, comprehending, and encoding
responses to printed language. Consequently, the popular strat-
egy in teaching- the child to read is to have him learn to recon-
struct printed messages into spoken language through use of
vocal, subvocal, or even inner speech so thai-he can then -corn-
prehend printed messages with his subsystems for spoken lan-
guage. In the process, intermodal communication subsystems
between auditory and visual systems are developed which are
'necessary for transfer of meaning from one modality to the other
(45).
Individuals taught through an oral method might continue
to subvocalize or use inner speech when reading silently (27),
but they can learn on their own or be taught through sensory
feedback mechanisms to suppress subvocalization (42). Although
a non-oral method of instruction could be used to teach children
to read silently from the very beginning of reading instruction
(15), formation of oral reconstruction or at least recoding sub-
systems are necessary for oral reading. However, a reader who
has attained maturity in both oral and silent reading has not
only developed subsystems for both of these types of reading
but can minimize or suppress his oral reconstruction and re-
coding subsystems when reading silently. He also learns to re-
organize his mental organization as he shifts from one reading
task or purpose to another (47,84,85,88,92).
Peripherally and centrally determined changes in eye-move-
Some critiques of the theory and research based on it (17, 19, 20, 24, 25, 70,100,
were evaluated or answered by Singer (87, 95).
101)
150 THEORETICAL MODELS AND PROCESSES OF READING

ment behavior accompany development in reading (65). By

grade nine, the average individual has attained maturity in func-
tional oculomotor efficiency and accuracy in targeting familiar
printed stimuli in reading (34). At first grade level, eye-move-
ment behavior on primary grade material consists of two fixa-
tions per word, each lasting about seven-tenths of a second, and
one regression or backward eye-movement about once for every
two words. As children learn to perceive words, associate mean-
ing to them, process information, and formulate appropriate
responses, their eye-movement behavior also tends to improve.
At the college level, eye-movement behavior is- more rhythmic
(one regression for every two lines of print), broader in span of
perception (one and one-fourth words per fixation), and rela-
tively rapid in pause duratioh (one quarter second per fixation).
The developmental curve of span and pause duration in reading
is relatively steep in grades one-four, tends to level off from
grades five-ten, has another upward spurt at grade ten, and then
_levels off again, but rhythmic growth_continues all_the_way to
college (14).
As individuals progress_ through the grades, perceptual
processes tend to decrease in relative importance while meaning
factors tend to increase. Systematic changes also occur in general
mental organization of factors underlying speed and power of
reading (44,48,52, 53, 81, 88).

Trends in Research in Reading

In 1964, two trends in research in reading were discerned: 1) a
a trend towards more basic research in reading or the explanation of
reading phenomena in smaller and smaller units and 2) a trend to-
wards construction of theoretical models to "represent the processes
at work in the subsystems or causal chain of events that come to focus
in the reading act" (52:127). These trends, of course, are interrelated
or complementary: researchers can use a model as a guide for deter-
mining what variables and relationships need to be studied in greater
detail and can hypothesize where more basic units fit into the the-
oretical structure (102). The discovery of more basic units, such as
the fractionation of fixation time into stabilizing, seeing, and process-
ing time (35, 36, 37 . 481, would. of course lead to a modification of the
SINGER 151

structure of a model based on eye-movement behavior. Holmes and

Singer (52) pointed out that such closely interdependent and mu-
tually directed theory construction is likely to foster creative, produc-
tivity in a field of study. Essential for the attainment of this ideal
relationship between a theoretical model and research is the formu-
lation of a theory on which the model is based in such a way that
testable hypotheses can- be derived from it. Testable hypotheses are
not inherent within a theory or model alone, as Sparks and Mitzel
(108) and Clymer (20) imply, but instead, testable hypotheses result
from an interaction between the theory or model and the imagina-
tive, logical, critical, and knowledgeable Mind of the researcher. The
aim of this review of theoretical models of reading is to summarize
a selected set= of models and= draw implications from them for re-
search and teaching. =Of course, other researchers, drawing upon their
own frames of reference and theoretical premises, are likely to draw
additional implications.

Selected Models
Model of conceptual response to printed words
For one subsystem of reading, instruction could be= organized to
develop= a conceptual response to printed words. A teaching model
for such imtruction is depicted in Figure 1. The model indicates that
the related "materials of thinking," consisting of percepts, images,
memories, information and feeling tone, are organized through =the
process of concept formation and are linked through a linguistic
form to a printed word stimulus that represents a-class of objects?
The rationale for the model has been drawn from several the-
oretical formulations. The definition of a concept and concept
formation is consistent with Russell's formulation (76:117, 248-249):
Concepts develop out of related perceptual experience.
.. ." They are "the means by which a child or adult represents
anything to himself and thereby creates a readiness to respond
2 The original version of this model and its explanation have already been
published (91). In the present version, the concept of lioguistic form (72) has
been- added to the model.
 152 THEORETICAL MODELS AND PROCESSES OF READING

Percepts

4
Images

4 Linguistic
Memories Concept Formation Printed Word
Form (Objects 1, 2, 3, ... n)
t
Information
4
Feeling tone
FIGURE 1. A subsystem of related elements mobilized as a conceptualized
response to a printed word stimulus that represents a class of objects. (Modified,
after Russell, 76.) .

with a- particular type of behavior?' The process of concept

formation involves "discrimination . . . plus generalization = or
response to common elements in object or situation, the per-
cepts, memories and images are integrated- into a concept.
A "linguistic form," as deft by by Reed (72:849), "is a linking
of a unit of- meaning to a physical representation in terms of a con-
ventional system such as speech or writing."
A conceptual response in reading can then be defined as an
intermodal communication system of ideas, percepts, memories and
images which are mobilized in response to value determined .pur-
poses Of the individual and the Stimulus demands of -the printed
word (2,47, 89).
This model may also be explained- in terms of the localization
theory of neurology (68): engrams are developed in specialized
areas of the brain which correspond to various components, such as
kinesthetic, auditory and visual images, and feeling tone for words
and objects. When a class of objects is- presented in close temporal
continguity to printed and spoken words representing the objects, a
phase sequence of cell- assemblies (43) could result in which the as-
sociated elements or attributes abstracted from a class of objects
communicate with or "teach each other" (13). This communicational
or transactional- system (22) yields maximal interfacilitation in re-
sponse to the printed word. .
SINGER 153

Implications for teaching

A strategy for developing a conceptual response to printed words
can be initiated in the beginning stage of reading development and
could be continued with appropriate- modifications throughout
reading instruction. The objective at the initial stage is merely to
present printed and spoken words in close contiguity with each
other and their corresponding referent class of objecti so that mate-
rials of thinking related to a particular class of objects can= be as-
sociated with each other and with the referent - object. A variety of
games and rhytnes can be devised for this purpose. For example, a
kindergarten teacher has Children' recite a refrain of "Barnyard,
barnyard, what's in the barnyard?" as she flips over a picture showing=
another animal. in the barnyard. In this lesson, only the = printed
word barnyard would haVe to be added to the game to facilitate
development of a conceptual response to the printed= word.
Another strategy that could be followed at the primary level is
to-have a class of objects, such as a:set of toy cars with the word
car printed on the side:of each car, serve as the instructional objects.
The children will manipulate-the cars, talk about theni, learn some-
thing about their characteristics, perhaps trace over the printed word
or participate in- putting labels on the cars, and discover what is
true about all the cars. Through the processes of Visual perception
and discrimination, manipulation of- objects and action on the ob-
jects, and tactual - kinesthetic- perception of the stimulus features
of the printed word with emphasis on the word's left-to-right order,
the common elements in each modality could be abstracted and
generalized. Then, on a subsequent occasion, the printed word
stimulus alone might elicit the conceptualized response representing
these experiences. With the original labels removed from the objects,
this would provide a test of the attainment of a concept.
As lessons progress, children could develop flexibility in mental
organization by practice in switching -from one conceptual response
to another. For example, two groups of objects could be intermixed
which had previously been experienced separately. The printed
words representing the objects can be used as a basis for grouping
the objects. At successive grade levels, -a more appropriate procedure
would be 'followed. For example, a conceptual response to the word
154 THEORETICAL MODELS AND PROCESSES OF READING

tree might be developed through Udry's award-winning book (106),

A Tree is Nice. At the high school level, the concept of courage can
be classified and enriched through a unit in which students read
and interpret a variety of books, all on- the theme of courage, such
as Profiles in Courage and The Red Badge of Courage (13). Thus, as
individuals progress through the reading curriculum, they would be
stimulated to develop concepts that could be mobilized as media-
tional responses to printed words (45).
Implications for research
The developmeht of such a mediational response system would
have several effects on learning to read and on performance in read-
ing. For example, if a conceptual response system for printed words
is formed at an early age, the concepti would serve as advance or-
ganizers to facilitate subsequent. learning ,and retention (4). One
hypothesis is that the rate of learning would probably be greater
because categorization- of stimuli- would effectively reduce the per-
ceptual and cognitive -load- (79, 89). Indeed, grouping of stimuli ac-
cording to categories enhanced transfer in spelling (31), and is
likely to do so in reading.
Although development of a conceptual response system could
and probably does occur to some degree in many pupils without
instructional intent or curricular organization for- this purpose, an
hypothesis to be tested experimentally -is that a deliberate instruc-
tional strategy would facilitate or accelerate development-of a con-
ceptual response system. Afore specifically stated, the hypothesis
is that individuals who had experienced and benefited from a con-
ceptual response strategy are more likely to respond appropriately,
within the limits of their stage of mental development (55, 69), to
a printed word intended to represent a concept than are individuals
whose educational experience had been more in the direction of
learning to recognize a particular label as representing a particular
object. If the conceptual response curriculum is effective, then it is
also likely to produce more powerful and more rapid readers. The
rationale for the better comprehension is dependent upon the as-
sumption that a higher level of performance in power of reading
draws more upon mobilization of conceptual' than upon labeling
responses. The explanation for the increased rate of reading is that
SINGER 155

speed of reading depends, in part, not only upon level of thought,

but also on flexibility in organizing and reorganizing working sys-
tems, including mediational responses. Therefore, curricular ex-
periences in shifting from one conceptual response to another in re-
lation to printed words could augment flexibility and result in an
increased rate of reading (63, 64, 65).
A conceptual response system is only one subsystem of more
comprehensive models for attaining speed and power of reading.
One of these models has been constructed at the fourth grade level.

Fourth grade model

The model depicted in Figure 2 was constructed in answer to
the statistical question: what are the minimum number of variables
which tend to be associated with a maximum degree of individual
differences in speed and power of- reading?
The resulting model, explained in greater detail elsewhere (52,
51, 79, 80, 93, 94), consists of four systems for power of reading and
three systems for speed of reading. The systemkfor pt.wer of reading
are 1) mental age or-reasoning in context, 2) suffixes or morphemic-
analysis, 3) vocabulary in isolatign or word meaning, and 4) matching
word sounds or word recognition. The systems for speed of reading
are 1) mental age or reasoning in context, 2) auding vocabulary or
word meaning, and 3) phrase-perception discrimination or visuo-
motor perceptual systems.*
Each system is associated with related subsystems. For example,
in the word meaning system for power of reading, the associated sub-
systems are mental age, suffixes, and word recognition in context.
Word recognition in context, in turn, is associated with prefixes,
spelling recognition, and spelling recall. The number next to each
predictor variable is its relative weight or contribution to the variance
in it associated criterion or subcriterion. All of these systems and

3 Comparison of this empirically constructed fourth grade model with a

model by Ruddell (75) representing a logical interrelationship of psycholinguistic
research revealed considerable overlap. However, some psycholinguistic subsys-
tems which appear to be missing from the fourth grade model are structural
and semantic markers, storage systems for classifying sentences and semantic
aspects of words, and syntactical systems (94).
 SPEED 100% uon
THE READING ACT1----). POWER 100%
atat
SOME BASIC ELEMENTS
COMMON TO BOTH PHASES
(WITH CONTRIBUTIONS TO VARIANCE)
LEVEL 1 LEVEL 5 .LEVEL LEVEL 111 LEVEL S LEVEL I
mentot Age-32.7, Mental Age 42.6 24.6 &lento! Age 29.7 Menlo, Age
(Reasoning
in Content) Inn
Chronological 13.3 on Context)
COnCeptual 13.3 ,Age
cAbility
25.0 SuffMes 273 Suffixes
Audio,
( Residual-442
-- 9.0 (Motteemie
yocabulory Aiding Memory 333 Analysis)
(Wert for Stories
Messing) Residual 490
16.0 Prefixes 26.0 in :streYn O
Word yo
26.3 Word (Word
Recognition
in Context 31.2 Ftecognition
in Context Meaning)
Stalling 4 29.6 Spelling
Recognition Recognition
Word
. 19.2 Residual
Perception 49.5
Discrimination 362 LIP 47:e aggro
Phrase Viand Verbal ft9
Abstraction 42D Sounds
meption-35.4 in Words
Discrimination
26.4 MIT:mu
(Visuo-motor
7erceptuel 192 Residual (RWeegnitlen)
Systems) 25.9 Residual

Rtsidual--24.11 -1L6 .Suppressor -Like Effect

(Phonics. Consonants) 0
Probabty: Probably:
Functional Oculomotor Attitudinal Factors
Breadth and Conceptualization
Not Accounted Fat 22.9 psEythetticiseynencyhtema,shing of Experience '

Methodologicot Factors 10.7 Not Accounted For- 0

SpAtedbilitofy 1

Processing Mobilizers rxf

Visual Stimuli Reorganization of
Note:All values ecpressed in per cent Verbal flexibility %Mobilities
FIGURE 2. Flow chart to show the results of the substrata analysis of speed and power of reading at the fourth
grade level.
SINGER 157

subsystems are also structurally or functionally associated with each

other.
Some factors which might have been depicted in the model,
had variables for them been included in the statistical analysis, are
listed near the bottom of the model. Among the factors for power
of reading are attitudes and values (2); breadth and conceptualization
of experience (79); methodological factors (I, 10, 16, 93); mobilizers
(3, 46); and reorganization of subabilities (44, 88). For speed of read-
ing, additional factors could be functional oculomotor efficiency
(34); psychosynchromeshing ability (46, 88); speed =of processing
visual stimuli (38); and verbal flexibility (79, 88).
The structure of the fourth grade model= is consistent- with -the
definition that reading is an audiovisual verbal processing skill of
symbolic reasoning (53). Furthermore, the factors shown in the model
support the major theoretical premise that reading ability consists
of two interrelated components, speed and power of reading. Under-
lying or associated with each of these components is a functionally
organized hierarchy of complexly interwoven systems. These systems
are selected and mobilized into a working system of functional com-
munication network in response -to the interacting and changing
purposes of the reader and demands of the reading task. For example,
the Model indicates that the factors associated with speed overlap,
but are not identical with, those associated with power of reading.
Hence a shift from speed to power necessitates a- reorganization of
systems and subsystems.
The model also suggests that a reader organizes and reorganizes
systems and subsystems mobilized for satisfying the changing de-
mands of the stimulus task within the purpose of reading for speed
or power. For example, in reading along a particular sentence, a
reader must retrieve and mobilize systems for recognizing words and
phrases, next link the recognized words or phrases to their corre-
sponding meanings, then integrate the accumulated meanings, and
subsequently utilize various cognitive processes for inferring, inter-
preting, and inductively or deductively arriving at conclusions or
solutions to problems. Furthermore, it can be inferred from the
model that two individuals could obtain the same speed or power of
reading but with qualitatively or quantitatively different working
systems.
 158 THEORETICAL MODELS AND PROCESSES OF READING

Conditioning the reader's mental performance are his attitudes

and values. Whether the reader merely skims, or attains superficial
comprehension, or strives for his greatest depth of understanding
(49) is determined not only by his cognitive but also by his
conative
and affective systems. Furthermore, influencing his cognitive style
in both acquisition of mental systems for reading and for performance
in reading are the individual's biological- support system (21, 22)
and his environmental interactions (23).4
Implications for teaching
Objectives. The model indicates that the- objeceves of instruc-
Lon should- encoMpass both speed and power -of reading. Since
speed and= power of reading do have some overlapping fdctors, im-
provement in both of these components is likely to ensue froth
instructional emphasis on their shared factors. However, specific
development- of each of these components is necessary because the
organizational pattern for speed is lifferent from =power of reading.
Also, quantitative and qualitative differences in the general working
systems for these major components= of= reading necessitate develop-
ment and practice in- mobilizing appropriate working systems for
each component. Furthermore, since flexibility in mental organiza- I
tion is likely w facilitate general reading ability, children could
benefit -from- practice in switching from power to speed_ of reading,
and-vice versa. Analogous to the procedure for developing accuracy
and efficiency in arithmetic (11), children might read something first
for accuracy or power of reading and then for speed of reading. For
example, if an individual is taught to analyze a word into constituent
phonemic or morphemic elements, he might subsequently be given
practice in recognizing and associating the meaning of the word
quickly. This alternation strategy for developing systems and sub-
systems for spud and power of reading could be instituted in the
curriculum at the onset of formal reading instruction.5
4 Parts of the dynamics of the model agree with the "search model" and
other models of reading briefly mentioned and d'mussed in The Reading Process
(57: 85-115).
5 The substrata factor theory postulates that speed and power of reading
are
separate but interrelated components. Although research conducted over forty
years ago first established that speed of reading is a separate component of
SINGER 159

Curriculum. The model indicates that reading attainment con-

sists of a complex organization of systems and subsystems. The scope
of the reading curriculum ought to reflect this complexity by en-
compassing at least all the educationally modifiable elements depicted
in the model, including those suggested as additional factors. All of
these educational components also should be incorporated into an
hierarchically structured curriculum, projected downward to the
primary grades, so that by the fourth grade level the factors depicted
in the model could be realized.
Grouping. Grouping for instruction or providing_ for individual
differences should be determined not just according to overall level
of performance in power or speed of reading as is the current practice
in some, if not most schools, (5), but also according to level of perfor-
mance in each system and subsystem. Simply stated: grouping for
instruction can be best justified =on the basis of diagnosed instruc-
tional needs. Tests or interviews (104) could be used for this
diagnosis.
Diagnosis and evaluation. The model, as a guide for diagnosis
and evaluation, sensitizes -the teacher to relevant variables and their
interrelationships. The to -her can use this frame of reference to
observe manifest deficiencies and, formulate a diagnostic hypothesis.
Appropriate tests can then be administered to determine whether
this diagnostic hypothesis is tenable. For example, if an individual
has at least normal mental ability, can pronounce words, but does
not immediately know the meaning of some words, the next question
is whether the individual has a system fOr morphemic analysis and
its level of development in relationship to his other systems and sub-
systems. If the word can be broken down into meaningful elements,
but the individual does not do so, the teacher might administer a
test on prefixes, roots, and suffixes (54).
For evaluation, the teacher can administer an entire battery of
tests to the whole class, then construct class and individual profiles
to determine strengths and weaknesses in the curriculum and in
general reading ability (30), schools have not systematically taught for nor eval-
uated attainment of this objective (85, 96). Indeed, only recently did Robinson
(73), acting on the research results of the substrata-factor theory of reading,
expand Gray's well-known and influential teaching model by incorporating rate
of reading as one of its components.
 160 THEORETICAL MODELS AND PROCESSES OF READING

individuals in the class. When such a battery of tests was adminis-

tered to a precocious reader, age five and one-half, who could read at
the fourth grade level, her pattern of strengths and weaknesses in
reading were clearly revealed by the resulting psychographs, depicted
in Figures 3 and 4 (82).
Thus, the fourth grade model has implications for teaching. Of
course, the model is only a first approximation. Further research is
likely to lead to the construction of a more comprehensive model
-Standard Score Scale
35 40 45- 50 55 60

Ito. MMee.

Vim.
02 Power

Ind. Ver.

OS Suffixes

06 Prefixes

07 Vocab. 1601.

14 Mental Age

15 Word Recoil.

11 Match Sor-nde

19 Mend Sounds
20 Cons. Phonics

26 Spell. Recog.

27 Spell. Recall

Xeys e 16 Most Powerful Readers

o a 16 Least Powerful Readers

+ = Marion's Standard Score*

35 40 45 ...50 55 60

Ficuaz 3. Psychograph of selected substrata elements for Power of Read-

ing. The psychograph compared the average standard sources of the upper and
lower 27 per cent of 60 fourth graders, separated on the dependent variable,
Power of Reading. The groups are compared on variables that underlie Power
of Reading at the fourth grade level, as shown in the model, Figure 2.
Marion's
scores were superimposed on the psychograph to show how her abilities compare
with these groups. (Marion did not take the Prefixes or Spelling
of time limitations.) tests because
SINGER 161

Standard Score Scale

35 40 45 50 55 60

No. Nese

01 Speed

Ind. Var.

03 Aud. We:cab.

04 Aud. Memory

09 Vis. Abstr.

12 Concept Ab.

13 . Cbron. Age

14 Mental Age

15 Word Recog. ......

23 Phrase Pere.
..... "
24 Word Pere.

26 Spell. Reeog.

Keys cox 16 Slowest Readers

* Is 16 Pastest Readers

rs Marion's Scores

35 40 45 50 55 60

FIGURE 4. Psychograph of selected substrata elements for Speed of Reading

The psychograph compares the average standard scores of the upper and lower
27 per cent of 60 fourth graders, separated on the dependent variable, Speed of
Reading. The groups are compared on variables that underlie- Speed of Reading
at the fourth grade level, as shovm in the model, Figure 2. Marion's scores were
superimposed on the psychograph to show how her abilities compare with these
groups. (Marion did not take the Spelling Test because of time limitations.)

which would have even more implications for teaching. The model
also has other research implications.
Research implications
The fourth grade model is a byproduct of confirming the
prediction that the major premise of 'the substrata factor theory of
reading could be generalized from the college to the fourth grade
level (79). Further research would determine whether the scope
 162 THEORETICAL MODELS AND PROCESSES OF READING

of the model could, in fact, be expanded through the assessment and

statistical analysis of- certain perceptual, attitudinal, and psycho-
linguistic variables (2, 32, 75). -

A wealth of research hypotheses could be- formulated through

suppositional thinking in viewing .1c factors depicted in the model.
For example, what if more systematic development were given to
morphemic analysis (meaning of prefixes, roots, and suffixes) in grades
one to three, would the quantitative kvel of the factor be increased?
Would the group's level of power of reading be accelerated? Or what
if instruction in context clues were- to incorporate -Mc-
Cullough's more comprehensive list (66), would this modify the role
played by word recognition in context in the model? Would a gram-
matical factor, such as semantic or syntactic markers precipitate from
a substrata analysis, if systematic instruction in these were included
in the curriculum and tests - for them constituted part of the compre-
hensive test battery? The model cotild thus serve as a cognitive guide
for experimental research and for evaluating the effects of experi-
mental instruction.
Since the sample -on which the fourth grade model was con-
structed had been taught by a-variety of Methods and materials, the
model represents the consequence ofa general method of instruction
(79, 80). But the hypothesis that sistematic differences in methods
and materials would lead to different models and that these models
might be, more appropriate for some groups of children than for
others also needs to be tested. The first part of the hypothesis could
be tested by devising- experimental curricula, such as one with
emphasis on each segment of the decodingmeaning continuum
(16, 93). The second part of the hypothesis could be tested by sub-
grouping children in each curriculum according to their presumed
strengths and weaknesses for benefitting from the particular cur-
riculum. A comprehensive battery of tests administered- to each
curricular group at the end of each grade level would provide data
to answer the question of whether differentials in teaching models
or stimulus inputs would be reflected in statistically determined
models and whether these models would tend' to converge or remain
significantly different throughout the grades (93).
Differential inputs, at least in the primary grades, have produced
SINGER 163

different psychological models for reading. For example, Buswell

(14), using an eye-movement camera, compared a phonic method
with a content or meaning method. The results indicated that the
phonic method tended to promote left to right sequence and -word
pronunciation, while the meaning method emphasized concern for
the content, slower development in word recognition, and rhythmic
eye-moirement behavior. Agnew (1) also found that =the phonics-
emphasis group was superior to -the nonphonics-emphasis group on
word, recognition and oral reading, but on silent reading =the two
groups were equal. These results su:Kest- that a model for oral read-
ing would place more weight on phonics, but a model for silent read-
ing would give less weight to phonics ability. HOWever, thelypothesis
that these models might converge in later grades has to be tested."
Whether children might benefit more from an auditory or a
visual emphasis in input modality depends on whether or not they
have- a limitation in either modality. Children with certain sensory
and perceptual deficiences or modality preferences might not be
handicapped -in a curriculum that did- not tap their weaknesses or
nonpreferred modality (9, 28).7 But, even if children do not have
developmental lags in perception or other perceptual difficulties in
relation to methodological emphases nor encounter a mismatch
between stimulus input and modality preference, a particular meth-
odological emphasis might still be detrimental to some children's

6 Holmes and Singer (52) predicted at the third grade level of Durkin's study
that her early readers would = converge -with the control group when both groups
had reached the eighth grade. The amvergence of these groups may have been a
function of the methods of instruction, which were not controlled in Durkin's
study. However, the *diction applied only to the criterion used for assessing
reading achievement since the two groups could attain the same performance
level through different patterns of= reading abilities. For example, there could
have been persistent differences between the groups in rate or fluency in reading.
Since Durkin di4 not utilize a comkehensive battery of tests and various criteria
of readinTability, her data cannot answer these questions.
7 Although the intent of the First Grade Study was to test similar hypotheses,
admitted weaknesses in the overall design precluded testing them (10). However,
some of the individual studies in this group might lend themselves to substrata
analysis of their data that would provide at least a tentative answer to the above
questions.
 THEORETICAL MODELS AND PROCESSES OF READING

reading attainment if the method did not stimulate development of

all the necessary components or subsystems for attaining speed and
power of reading (12, 93).
Whether other- known groups, such as slow vs. fast and least vs.
most powerful readers, differ significantly in their patterns associated
with general reading ability-or only in their rate of development
of these patterns, might also be ascertained. Evidence at the high
school level suggests that the patterns of the knowniroups, although
similar in certain subsystems, are sufficiently different- to _warrant
differentiated _curriOnla.iiientas ati-ififtt=grade-, a differentiated
curritulum necessary. -to to:i-meet individual differences in -age of
=

onset of reading readiness (97): Duringithe intermediate grades, the

evidence indicates that the structure of-the syStems _associated with
speed and power of reading-sof:a represeniative group undergo not
only quantitative, but also qUalitatiVei changes, as shown in Figure
5 (81, 84). (The model for speed-for reading_appears as Figure 1 in
Singer's article in Part Two of this publication.)
One of:these changes provides _some- support for the hypothesis that
there is a Ainesthetic-auditory-visual gradient shift accompanying
development
. of general- reading - ability (88).= Whether further reor-
grinnation occurs in the general- Working systems or pattern_ of factors
related to power of reading at the junior and senior high School levels
is a_question that has not yet been answered; although some cross-
sectional studies have been reported at these advanced leVels.
Model of substrata-factor patterns accompanying development in
power of reading
Figure 6 depicts predictors Of power of reading at the elementary
(84), high school (53), and college levels (44). Since neither test bat-.
teries it these levels nor subjects were identical, the following com-
parison of predictors at these leveis can only suggest hypotheses for
a longitudinal investigation:
1. During the intermediate grades, word Meaning and reasoning
in context systems (meaning of affixes, verbal artalogies and relation-
ships) tend to increase in their proportion of variance associated with
power of reading. But, from the sixth grade through the college level,
the quantitative contributions of these systems to power of reading
 SINGER 165

Grade 3 Grade 4 Grade 5 Grade 6

Visual Verbal 1,; _Visual Verbal , Visual Verbal

16 % - Meaning
Visual Verbal
36%
Meaning '774-Meaning 18%-Meamng

Auding 7% Auding 6% Auding 8%

Meaning of
Affixes
, Meaning of
Affixes
., Meaning of
`4"-Affix61-7--- 174

Visual Verbal_ pry. Power

Conceptualization of_
AuditoryVisual Reading
Verbal = -
Conceptualization

Word Recognition Word Recognition= Ward Recognition

1 'In Context36Vet-ln Context--= 13%-ln Context- 14%

Matching Sounds - _ Matching Sounds _

-In Words-14 7 In -Words -20%

Blending Word
Elements 9%
77

Syllabication =

Consistency 7
Variance 66 71% 69% _73%
Explained:

Variance Not 29% 31% 27%

Mounted Fars ......... .........

Utah 100% -100% 100% 100%

.
Fiume 5. Diagrammatic Model= of the= Substrata-Factors Accompanying
Development of Power of Reading in Gradet 3 to 6. The model shows for each
grade the first level -substrata factors and their, per =cent contributions to
variance in Power of Reading. The variance not accounted for is probably
attributable to Attitudinal Factors, Verbal Flexibility, and Mobilizer& Harry
Singer, 1964. \
-ten& to remain-=constant:- _At -_t,ne =higher-- grade: levels; they =predict
about_half_-rif individual =differences -in of reading._Whether
these-- results-=, reflect rate of maturation or -instructional ,emphasis -.or
--=
some interaction-4A maturation-rind= ifistrtic don inwoul-cr- haVe- to be
_tested --via ran' experimental-pm-gram-at -the -=--jUnior and senior high
*chool
2.-M-eastires- of audi4were not_ includedrit the ollege level; but
_were= at:the -=elementary and-ibigh :s-school= -_-Although an auding
factor did-nOuptecipitate--,rit-thriixthfgrada_leVel;iit doeSritzthelovier
grades and at the _high school= level.--At= the sixth grade_ leV4i_a-udirig
 I ,
'
, I
1

1
I 1 I
1

' I

414''71"A
IIIII,,],
SINGER 167

seems to shift from an auditory-visual factor at a concrete or associa-

tional level to auding at a -more abstract level (81, 92). This result is
consistent with the hypothesis that throughout the develOpmental
continuum reading is an "iudiovisual verbal processing of
symbolic reasoning" (5I, 53).
Although a gradient shift to VistialModality dominance probably
occurs about the sixth grade level for relativelY easy material, when
confronted at higher grade levels by difficult materials the reader is
likely to _mobilize heith=vitual arid-auditory systems-iri_order to max-
imize his cues for solVingproblenis in word recognition, word-mean-
ing, or reasoning context (42, 88i 92).,
3. A system of phoneme7grapheme relationship, such as "match-
ing sounds in words," does not precipitate after-the sixth= grade kvel
as a first level predictor for= power = of reading, even though variables
for tapping this syttein had been _included in the test battery at the
high_schodl and college levels. However, readers at these advanced
levels logically must go through some of -word=recOgnition.
The- interpretation for this apparent paradox must be that -la word
recognition system, as such, does not precipitate =at the= first level of
analysis because at the advanced -leVels of reading individual- dif-
ferences in power of reading are primarily attributable= to factors
other than word recogn ition.
Implications for teaching-
The corresponding- implications for teaching are the following:
1. Emphasis, in- general, in teaching and evaluation of reading
ability should reflect the relative magnitude of the contributions to
variance of the systems associated with power of reading. Therefore,
approximately half of -the curricular time beyond the sixth grade
could justifiably be devoted to development of word meaning and
reasoning in context. Of course the- teacher must adapt the curric-
ulum to the wide range of individual differences that exist = in each
classroom.
Although direct teaching of vocabulary and affiv_its is likely to
develop an individual's mental dictionary, a broad educational cur-
riculum would contribute to this goal and would also tend to develop
other appropriate subsystems and processes, such as retrieval and-
integration of stored information. Included- in such- instruction
 168 THEORETICAL MODELS AND PROCESSES OF READING

should be the meaning and use of structural and semantic markers

(75,94).
2. Mobilization of auditory processes into an individual's work-
ing system is a function of his interacting purposes and the difficulty
of the materials. This mobilization it likely to occur when the per- .1
ceived reading task is difficult. Teachers, aware of this dynamic
process, could structure reading assignments appropriately. If teach-
ers try to develop silent-reading with a minimum of auditory pro-
cesses or subvocal _Speech, they should' use -relatively= easy materials
and only require a surface level= response-be-mute difficult materials
and -a purpose of reading for depth of understanding are likely to
mobilize both visual and auditory systems AlternatiOn of difficulty
of materials and purposes would_ probably deVelop necessary verSa-
tility in Mobilization of theie systems._
= 3. Although certainzsubSystenif for word recogniticin coritinneto
improve throughout the developmental continuum (53, 105), the
word recognition system is, in-general, well developed by the sixth
grade level. _Currieular emphasis at the&dVanced- leVelS of inttruction
should therefore be= on word meaning and reasoning in context
.
systems. Of course, instruction in word, recognition will still be
necessary" for-those individuals who have not yet developed an ad-
equate wordittognition System.
Implications for research
Present knowledge of the development of reading ability is based
essentially on cross-sectional studies. Although these studies provide
some inforMationz_on development of reading ability, they must still
be_folloWed by & longitudinal investigation. However, only a few
longitudinal studies of reading, limited in scope or duration, have
been reported. These have-been studiet of functional oculoirk zor
efficiency (34) and other aspects of eye-movement behaVior (13),
acquiiition of a particular Methodological approach to reading (67),
early readers (26).zand the effect of various_methoddlogical emphases
in the primary grades (74). z

What is very much needed in the field of reading is a longitu-

dinal- investigation of reading= development that encompasses each
domain related to reading achievement Perhaps such an investiga-
tion will arrive at a conclusion similar to Bayley's on the growth of
intelligence (7, 8:807): "intelligence . . . appears to be a dynamic
SINGER 169

succession of developing functions with the more advanced and

complex functions in the hierarchy depending on the prior maturing
of the earlier, simple ones."
Also, the relationship of stages in development of logical think-
ing (55, 69) to acquisition and performance in various systems and
subsystems of reading behavior- should be investigated. For example,

READING

Figure 7 not included due to copyright

restrictions

FIGURE 7. A learning structure for the basic skills of reading. (From The
Conditions of Learning by Robert M. Gagne. Copyright @ 1965 by Holt, Rine-
hart and Winston, Inc. Reprinted by permission of Holt, Rinehart and Winston,
Inc.)
 170 'THEORETICAL MODELS AND PROCESSES OF READING

Piaget's stages of mental development may explain the change in

factor patterns. that occur at the sixth grade level : the average sixth
grader may be passing from the concrete to the more abstract, formal
stage of mental development.
- Until comprehensive, longitudinal data are adduced on the
empirical development of= the hierarchical structure associated =with
speed and power of reading, the curriculum constructor must= fill in
the gaps in knowledge with a curriculum based both upon empirical
data =and logical analysis. Figure 7 depicts an outline of a logically
determined hierarchial structure for reading (29) which has not yet
been developed into a= complete teaching model. Objectives for
such a teachi ig model may= be based _upon= the comprehensive set
of categories constructed =by= Spache (99). Categories were formulated
for= word recognition, word meaning, and reasoning in context, cor-
responding to all the cells in the semantic layer of Guilford's model
(41) of= intelligence. All the Operationa =of cognition, memory, diver-
gent thinking, convergent thinking and evaluation and all the
products of intelligence, consisting of units, classes, relations, syitems,
transformations, and_ implication are = represented in= these objectives
for leading. These objectives are shown= in charts' in Figures = 8a,= 8b
and 8c. -
Spache's categories can also be used for selection of a test battery
to evaluate and plot in a longitudinal study the hierarchical develop-
ment of systems and subsystems associated with attainment of speed
and power of reading. A by-product of this investigatior would be
the construction
.
of a comprehensive developmental model of read-
ing.$

Interrelationship Among Types of Models

Models of- teaching, skills and abilities, and= processes involved
in reading are interrelated. Depicting these interrelaionships is the
s Comprehension objectives_ similar to Spache's, but corresponding to Bloom's
taxonomy of cognitive and affective dimensions, have been formulated by Barrett
(6). Although both sets are logical formulations, Spache's does have some em-
pincallrbased rationale. However, both sets_ of objectives ought zo be considered
in the formulation of an hierarchically structured, comprehensive curriculum
in reading.
pm1011 frr 4ht 'N,Ss rtfr ro+ file I rn1,1,11

I
Mental Processes Unit Class Relations Systems Transformations
Cognition (recogni- Recognition of Recognition of differ- Recognition *word Recognition Of phonic Recognition of word
tion of information) printed word as ence among letters. form and letter de- characteristics, i.e. endings (plurals, or
such words and numbers tails; as little initial and final roots and affixes)
Sounds
Memory (retention Recall of specific Recall of random Recall by word form Recall of complete Recall of word end-
of information) word forms, e.g. words; letters and 'h and details (sight word * combined ings (s, ed, ing)
own first name numbers in isola- vocabulary)' clues: e.g. initial root or affix
house number tion (reading signs) sound, sentence
pattern and form
Divergent Production Neologima,lwerd , Recognition by dia. Recognition, y base Spontaneous phonic Experiments with
(ilogicalmeative form errors tinctive detail as g Word Within larger, and structural gen- derived and base
ideas) in dog; capital compound Words eralizations words
letters
Convergent product, Recognition by pie Recognition by sheer by sen- Recognition by word Recognition of de-
don (conclusions. toiled clues, shape length, as grand- 'Itence pattern; Word families, common rived words by base
inductive thinking) of sign, etc. mother: or.by syllables, roots. etc. plus ending, or root
unusual shape; as plus affixes
elephant
Evaluation (critical Comparisons Of differ- Discrimination of Discrimination by Discrimination by Discrimination of
thinking) ences and Similari- words, numbers. 'form, details and form, details and base and derived
ties in gross form letters pattern phonic elements words by endings,
and sentence letter sounds, MM.
pattern' melee pattern, root
and affixes

I Figures ga, ilk and ec are reprinted from Toward letter Reading by George Spache, Champaign, Ill.: Garrard Publishing Co., 1963. By per-
mission of the author.

Ficuat 8a. Reading Behaviors in Symbolic Content (Words. I-ettere)1

"PERMISSION TO REPRODUCE THIS COPY.
RIGHTED MATERIAL HAS BEEN GRANTED
BY
Garrard Publishing
.
TO ERIC ANO ORGANIZATIONS OPERATING
UNDER AGREEMENTS WITH THE,US OFFICE
OF EDUCATION, FURTHER REPRODUCTION
OUTSIOE THE ERIC SYSTEM REQUIRES PER
MISSION OF THE COPYRIGHT OWNER "
 Mental Processes Unit Class Relations Systems TransforMations Implications
Cognitiea Recognition that Recognition of ReCognidem'Of
(recognition of printed word word function Recognition of im- Recognition of in. Recognition of
word associations piicit comparison herent meaning word connota-
information) has meaning naming, action, and relationships, ' In hierarchies of in roots, affixes tions, figures of
descriptive, pro- synonyms. words (temperal. and inflectional speech and pres-
ncminal. con- '
forms.
' onomi91m)', Spatial. itatensitY) ence of tone
nective alliteration.'
rhyme ' and mood.
'
Memory Recall 'specific ", Recalling specific Recall the luso- Recall of base
(retention of Recall of meaning Recall of specific
word meanings meaning thru word and mein-
"
'
datedsvisid of base, ord or connotations and
information) (denotations) recognition of (sPrniin eta) ing, recall of of root and
, function. e.g.. and thus recall meanings of fig-
other rleirrees In afixes. ures of speech.
this word is meaning of the hierard4;
'
a thing gives toed' 4.1riarvifted
strongest
through strong.
Divergent Secure meaning Recognizing ex- Using meinitinks
(ogical. creative from context Using trend of Recognizing words Free association to
amples of tut- to recall
ideas) usual or nt... hierarChy, to containing news connotations. i.e.
by inference, Meaning recognize com, ',IA huilated,
Vela usage, e.g., daydreaming or
Mae West parison'and thus, condensed or re-
'
Imagery; recog-
derive meaning lated 'affixes and nising new figures
for given Word: roots. of speech, new
acting differences connotations, 0
"*iiieaning,due allegories.
Convergent Meaning from con- Relating specific askwitatO
(cond unions. in- text by sentence 'Flynn a group Of Use Of structural Using connotation.
meaning to ante- ,ward oynloym, Words. formulates analysis es,'d ues tone and figura-
ductive thinking) structure (apposi- cedent.' to accent etc.) in context tentative
'

tive. explanation. and contextual' to meanings: ' tive language to

and thniderive dictionary; study interpret context.
contrast. defini- duesta function meaning intended of deriration.
tion) of the word for given Word:
Evaluation Critical reaction to Reacting to au- Dtscrihniaiting Reacting critically Reacting critically Detection of au-
(critical auther). dinice thor's usage, to differenies among to shades of to shades of
thinking) of words use of slang. thor's use of cm.
synonyms meaning. to meaning Am- O
colloquialisms. tionally.toned
antonyms; me- exaggerition,, parted by ,varlit. words, analysis of
etc.
Ogniring irony hyperbole:* dour in affixes. figurative lan-
and weasel. guage, tone or
mood.

FIGURE 8b. Reading Behaviors'in Semantic Content of Words

 1111, 1
11,

1100/11...,11,1.0014,4.4., . r
N.', I 1,14,411,1w, r o,

Mental Unit Class Relations Systems Transformations Implications

Processes
Cognition Recognition that , Recognition of Recognition of Recognition of types Underline key Recognize that
(recognition of word has meaning sentence as cons- paragraph mean- of relationships words of there are impli.
' information) piete thought ing (literal idea within structure paragraph cations in
of paragraph) of paragraph author's main
idea
Memory Recall specific Recall of thoughts Comprehend main Summarize facts of Combine recall Choose possible
(retention of word meanings of sentence e- idea is summa- paragraph in own With own implications
information) verberations) tion of sentences words with dui associations from given
(reverberation) attention to alternates
structure
Divergent Produc- Meaning Selecting implied"' Choosing implied Analyze author's Construct rebus Amplify author's
tion (logical, context by Meaning of main idea reasons for of paragraph; implications and
creative ideas) inference sentence structure offer new titles ideas in tree
for paragraph association
Convergent Produc- Mt. nr frees Combining idea' Evolving main Idea Categorize structure Choose anions; al- Suggest future
tion (conclusions, structure of COO., into literal as extension of of paragraph; ternate titles or applications of
inductive text (i.e. appeal. meaning of topic sentence outline it statements of author's ideas
thinking) tive sentence) sentence main idea
Evaluation Acceptance or Acceptance or rejec. Acceptanceer,rejec- Look for fallacies Identify author's Check author's
(critical rejection of tion of meaning tion of !Min' idea, in lOgic, appeals' viewpoint and background as
author's diatoms of sentence, as as fact or opiniOnr to reader's ant... Purpose; compare basis for view-
Waldo* fact.opinion cheek authors tit)** overgener- ! with other view- point; react to
sources; compare tOieloni; omit. points; explore ! author's value
Witit'Ortexperi- lions, distortions the ultimate judgments; ex-
eniii and heliefs', outcomes of amine author's
acceptance of basic assump-
author's ' tions and infer.
viewpoint axes from these

FIGURE 8c. Reading Behaviors in Semantic Content (Meanings. Ideas)

IsO
 174 THEORETICAL MODELS AND PROCESSES OF READING

sequential scheme constructed by Strang (104). This scheme is shown

in Figure 9.

0
Emu 9. The diagrans indicates. the- interrelationships among the individ-
ual (0), the _classroom- reading situation (S) the individual's -responses-to the
situation_ (R),-__therfresulting_memory--ttatet_gy-amb
th-e_iddiViduars -perception
_(1!) of theneit-sittiation(Reprinted-frcknmThelieadintProCeis
-and Rantifi-
tations."-Iiiinvitational Addfesses, 1965--;Newark,-Delaware: Internationallt.ead-
ing--AsiociatiOn, -I965,-49-74. By permistion-Uf atichor_and--the -publisher.)
. The individual student is represented-by 0 and= consists of a
psychological model of the products, -prerequisites and processes
within the reader. Included in= this model-are physical, physiological,
mental, attitudinal, knowledge;self-concept, desire to read=in short,
all those interacting systems and subsystems within the individual
that function between stimulus input and response -output. Instruc-
tion is next concerned with S or the classroom situation. its decor,
lean.-..ag atmosphere, and types of reading material, which mutually
interact with 0. -The resultant= of this interaction is 1?, the student's-
responses to the situation. This dynamic interaction leaves a trace, T,
which is stored within an individual's nervous system. This trace then
influences the individual's perception, P, of the next situation, which
completes the cycle.

Summary
After a review of the = literature on =deli of reacting develop-
ment,- several were selected for more detailed presentation and for
drawing implications for teaching and researi.h. These models
covered the developmental range from kindergarten to the college
level.
All of the empirical models selected for detailed =presentation
are byproducts of research to test hypotheses of the substrata-factor
theory of reading. They provide some cross-sectional evidence on
psychological factors associated with development of speed and power
of reading. Although this cross-sectional evidence tends to support
SINGER 175

the hypothesis that the structure of skills and abilities associated with
speed and power of reading is hierarchically developed and orga-
nized, a longitudinal study is necessary to fill =in the developmintal
gaps and to reveal individual patterns of reading development.
If a longitudinal investigation is undertaken, important -in
such an investigation would be the formulation and utilization
of a teaching model that is based upon a theory of the hierarchical
structure of learning (29) and a comprehensive = set of objectives,
such as those =constructed by Spache (99) and Barrett (6). =In such
an investigation we would seek to construct a comprehensive pat-
tern of the "common routes =to maturity around which individuals
vary," determine how these common = routes; interact with= the per-
sonality of the individual, his environmental influences, and his
biological support systems. If we can also ascertain whether =or not
variations in educational procedures have durable effects upon the
developmen4 structure, and= functioning of these common routes,
then we would =be in a position to construct-a comprehensive theory
along the lines outlined by Strang (104). This theory would then
serve as a cognitive map for improving development of speed and
power of reading.

REFERENCES

1. Agnew, Donald C. The Effect of Varied Amounts of Phonetic Training on

Primary Reading, Duke University Research Studies in Education, No. 5.
Durham, N.C.: Duke Univitsity, 1939.
2. Athey, Irene J. "Reading-Personality Patterns at the Junior High School
Level," unpublished doctoral dissertation, Universby of California at
Berkeley, 19.5.
3.- Athey, Irene J., and Jack A. Holmes. Reading Success und a Personality
Value-Systems. Syndrome: A Thirty Year THEN and N_OW =Study at the
Junior High School Level, Final Report on U.S. Office of Education Con-
tract No S-248, Bureau of Research No. 5-8027-2424, 1967.
4. Ausubel, David P. "The Use of Advance Organizers in the Learning and
Retention of Meaningful Verbal Material," Journal of Educational Psy-
chclogy, 51 (1960), 267-272.
5. Balow, Irving H. "Does Homogeneous Grouping Give Homogeneous
Groups?" Elementary School Journal, 63 (1962), 28-32.
6. Barrett, Thomas C. 'Taxonomy of Cognitive and A.:ective nsions of
Reading Comprehension," in chapter 1, Theodore Clymer, "What is
Reading? Some Current Concepts," in Helen M. Robinson (Ed 8, buoy&
176 THEORETICAL MODELS AND PROCESSES OF READING

lion and Change in Reading Instruction, The 67th Yearbook of the

National Society for the Study of Education, Part 2. Chicago: University
of Chicago Press, 1968, 19-23.
7. Bayley, Nancy. "Consistency and Variability in the Growth-of Intelligence
from Birth to Eighteen Years," Journal of Genetic Psychology, 75 (1949),
165-196.
8. Bayley,acancy"On the Growth of Intelligence,"---American-Psychologist,
10 (1955), 805-818.
9. Bond, Guy L. The -Auditory and Speech Characteristics of Poor Readers,
Teachers College Contributions to Education, No 657, 1935.
10. "Bond;-_ Guy _and -Robert -Dykstra.:-=''The- Cooperative Program zin =First
Grade-Reading-Instruction," Redding _Research 'QUarterly; 2 -(1967),_:5442,
11. Brownell,_William A.-41Lite, Aceuracy,,and Processes in Learning,7 abridged
by Theodore:L.- Harris and W. E SChwann (Eds.),ISelected-Readings _on
the Learning Procers. Ne*--Yo-rk:-_Okford UniVersity =Press,-: 1961, 388-400,
1Z Burt, Cyril, and R.-13:164- !Teaching Backivard Readers," British
Journal' EducationaVPsyckology, _16 (1946), 116-132. --_ z

13. Burton, =Dwight= L.,--and Steven-Dinning; A iScholaltk Literature Unit:

Courage.-New York:= &holistic:Book Services, 1960.
14. Buswell, Guy T. Fundamental Reading:Habits: A= Study of Their Develop-
ment, Supplementary- Educational = Monographs. Chicago: University of
Chicago Press,1922. _

15. Buswell,fGuyiT.'''Non;dral-Reading4 Study of Its Use -in the Chicago

PUblic Schools," Supplementary Educational Monographs; No 60,-1945.
16. Chall, Jeanne.: Learning- to Reading: The Great Debate. New York:
McGraw-Hill, 1967.- = =
17. Carroll, Jahn- "Comment on-the Holmes-Singer Substrata,Factor =Theory
Of Reading," in Richard Braddikk- (Ed.),_ Research in the Teaching of
English, Official = Bulletin ofjNational Council_ of Teacheri of English,
Fall 1968.
18. Cleland, Donald L. "A Construct of Comprehension," in J. A. "Figure?
(Ed.), Reading and InqUiry; Proceedings, 10, 1965. Newark, Del.: Inter-
national Reading Association, 59-64,
19. _Clymer, Theodore. Conimenti-on=-Harry-Singer's article, "Substrata-Factor
Evaluation of a Precociout = Reader, -Age -Five and One-Half," Reading
Teacher,_16-(1965),: 296.
20. Clymer, Theodore. 'What-Is_Reading1 Smile Current Concepts," in Helen
M. Robinson (Ed.), InnoVation and _Change in -Reading Instruction, The
67th Yearbook of the National:Society for-the Study of Education, Part 2.
Chicago: Univertity of= Chicago = Press, 1968, 7-29.
21. Davis, Frank R.i Jr; "Speed of -Reading and =Rate of RecoVery in Certain
PhysiolOgicil Subsystems," unpublished doctoral dissertation, University
of California at Berkeley, 1963,
2E Davis, Frank R. Jr.- "The Substrata-Factor Theory of Reading. "-Una!,
PhysiOlogy as a Factor -in Reading," in J-. A. Figure! :(Ed.), ha/ 'lent
SINGER

of Reading Through Classro-oef Practice, Proceedings, 9, 1964. Newark,

Del.: -International- Reading Association, 292-296.
23. Davis, Frank R. "Environmental_ Influences-on Substrates of- Achievement
Leading to Learning Disability," -Journal of Learning Disabilities, 2 (1969),
155-159;
24. Dliker, Sam-: Review of -"The _Substrata-Factor Theory: Substrata Factor
Differences- Underlying 4 eadingl Ability -in --Known-Groups," Journal of
Developmental Reading; 5_ -(1962),_ 195;197. =

25. Dukes., Sam."-'Applicability_ of -the, SUbstrata-Fattor Theory to Reading in

the Elementary -_School,"- in =- Patrick= Groff = (Ed.); ."ReSearch Critiques;'
31-__(1964)?,921=924.-5-_=-
-26. Durkin, ieDolOtes;-__Ghildrea=Whd-Read-Ectily;-NeW=York: Teachers- College
Prest,-Coltitabia4ThiVersity, = _ = -_=

27. -Edfeldt,-z-Ake=111;-=-Silerit_-speeck---and--Silent= Reading."-Chicago:-,University-i__-

of= Chicago _

Fen-drick,11.- Visual Characteristic's ef J:!o- or -Rea ders; -Teachers-College-Con-_,

=--,:tributions-Ao=Education,No.:656,-A935.t =

29._ -Gogne;-,Robort. _CoriditiOns- :of Learnirig, =New= York: and

Whitton-, =- -1
30. Gates,= "An_Experitnentalfend- Statisticals-St-udy_OUReading==and
= Reading_ Tests; journalerEddeatiortall'iyekblegy;-12-z(1921)-;:303414----
-31.--Gates?,-Arthur-I.,Generalizationi-:andi==-Transfer-
-Teicheran-College,ColnafibialThiVeriity, ==z -=

Geyer, John= J: =- Perceptual -SySteM-ssini_R-eadin-gt A Ternporal_-gye-Voice

Span Constant, Final---Report, Contriet---No._=0.E.---_540-362,' CO-Operative
-Research-Program_ -OffiCe-f-oViEdnaitiini;-- U.S.-=Departmenti of Health-,
r--Education, and Welfare_1966.- =

33. Gibson, -Eleanor J. et- a1. -"The -Role- of---GrapheMePhoneine:Cortet-pen:

denceln the =Perception of Words,"-American-_--Jourrial 44--P.tycholg_-
o Y0--75--
(1962), 554-570. z _

Gilbert, Luther C. "Functional _Motor Effitiency of the Eyes and its Rela-
tion to Reading," University of California Publications in Education, 2
(1953), 159!232. _

35. Gilbert, Luther_C. "Genetic Study of ye-Movements in Reading," Elemen-

tary School Journal, 59 (1959), 328-335.
36. Gilbtrt, Luther' C. "InflUenCe of Interfering Stimuli on Perceptio-n of
Meaningful -Material," California Journal of Educational Research? 10
(1959), 15-23.
37. Gilbert, Luther-G. "Saccadic Moventerits as a Factor in -Visual Perception
in Reading," Journal of Educational Psychology, 50 -(1959), 1511
=
38. Gilbert,- Luther C. 'Speed of Processiq Visual= Stimuli and Its Relation
to Reading," journal of Educational Psychology, 55 (1959), 8-14; _
39. Goodman, Kenneth. "A Psycholinguistic View of Reading Compiehenlion,"
in G. B. Schick and Merrill M. May (Eds.), New Frontiers in College4dult
 178 THEORETICAL MODELS ANb PROCESSES OF READING

Reading, The Lth Yearbook of the National Reading Conference. Mil-

waukee, Wis.: National Reading Conference, 1966, 188-196.
40. Gray, William S. "The Major Aspects of Reading," in Helen M. Robinson
(Ed), Sequential Development of Reading Abilities, Supplementary Educa-
tional Monographs, No. 90, 1960.
41. Guilford, Joy P. "Three Faces= of Intellect," American Psychologist, 14
(1959), 469-479.
42._ Hardytk-,7C.,IL. Petrinovich, and D. Ellsworth. '-'Feedback ofSpeechhfusde
Activity-,-During_ silent- Reading: Rapid--_F.xtiriction,'T Scienre, 154=-(1966),

-43.-_=1-1614--David0.-Orgnitition-upftBehatiior.=
Holiribila-ck_Ai-="Facrors,UnderlyintsMajcii-=ReadinglDiiabilities at the
ColletelLe-tel,"-Genetic'/34-chology=Monotraphi,;491954)4,95;-_
--- 45. Hanes, Jack A._ "Theikain-sand-,-theReading=Procesi-,"--Clarititontollegi;
---LivinvTwenty4econ-d- Yearliaok_--_of -:the-ClaremOnt_-_
-Reading-Conferenc4-__Claremont,- Cal: -_,-Curri-Culumsl:aboratory__1957=49-67=--
46. - Califor--
PtibliCationt=id Ediscatiou,1241959);-213-29E- ----_-_- =

47.= Holnies--Jack'-=-A-; StibittitaFactdrf-Thearr New =

btperimental-EVidente-in ET-AlletrFigitrel-:(Ed), New -Frontieti _in zRad,
ing,-_,Proceedingsjor theInternatiOnal-ReadingrAsiociaticin 5,11960;_-New
_ York:_Scholastic=Magatir.e41_15=121-_= _ z

48. Holmes, Jack A-. Personality the Disabled- Reader,"

Journal-of =Developmen-tal_sReading,-:4-(1961), 111;122f
49. Holmes,_JaCk A.!'Speed,=Comprehension-and:Power in Iteading,"-in:Emery-
P=Bliesiner--and_-Ralph_ C. Staiger4E4S.);Probletris, Programs, and Projeets
in College -Adult -Reading, The -11th Yearbook_cofr_ the National= Reading
_
Conference-. MilWaukee,_Wit.:-National- Reading ConferenCe, 1962,-6-14.
50. Holmes, Jack_--k- "Basic -Assnitiptioni Underlying-the t=_SUbstrati-Factor
Theory," Reading -RESeatch- QUarterlY,-1=(1965)
-51. Hohnesi_ Jack A.,, and Harry= Singer. "The Substrata-Factor Theory: &lb-
strata-Factor_ Differences 1.1nclerlying_ Reading z Ability in Known-Groups
at the High School:Level," Final _Report Covering: Contricts__No. 538 SAE
8176_and
--
538A. _SAE 8660,-U-S. Office of- Education, -Department of Health,
Education and= Welfare, 1961-._ _= _ _

52. Holints-,=_Jacr A., and --Harry- Singer. "TheoretI-al Models and Trends_
Toward- -Morel-Basic Research id-Reading," Rivieto of Educational
search, 34 (1964), -127-155. -

53. Holmes, Jack A., and Harry Singer. Speed and Po-wer in Read:ng in High
School, U.S. Department of _Health, Education and Welfare, Office of
Education, - Bureau of Research and Development, Catakig
No. F8 5.230: 30016, 1966.
54. Holmes, Jack A., and Harry Singer. The Language Perception Test Series,
Elementary Battery:Pebble Betch;-,Cal.: Psychological-Educational Services,_
1966:
e-and
eretice
 _ . _= _ _ -_=
_
=
= _ ===, -
-=- -=
=
--=-- -1
=== ==_
=== _s=

=- ====-
,_
=

=====-==-=

==
=

-=

=
ommuni
 _

_ f--:____ __ j_ --__---- _=-_:--,-_

-i

_
RETICAL -2MODELS OCESSESi= 0
zn-_
, =
MODELS
 -_---,----z---

--== - , ----_--- -_--

ins tenvincreasts-
at ikticitia
wino
_
_
THEORETICAL =MODELS:- AND =PROCESSES= OF_ -READING
RMALPERCENTILE-SC
RETICAL:ISIODELS-rc AND = PROCESSES :OF READIN
 rea in
ance situations as we 'eve w
 197

us,--_-_ii4 2=---=--_-----;----_
-;-iHebb,=,TheOiteptizition-af--hehavior.--Ntw-Yotk:=JohttiWiler-_& Sons, -_

- Inc., f1949,--_p.-;335-_--_----,:,_,_ ---------_

=_13:-=----Ja= alinei,--ipersoniiti _-an-cU-Spellsng4kiwy,JTriirts4rOf-_7c4iiToii_- --_-- -

=- _ -----:r---';Publications iniEl==ir!ui ticp,71959.-,-Yclume4-2;;N'oc=4;qi- i
9jack=A1Holfes!Pert6ralty=Chatatterittits--oLhe=DiabletReadefi" ----s-_-__

--.---- foutnal-of:Developroehpleading;I#,Tress._,_ ___

-- ----- 4- --- -__---__
10. r -usiwt.;,pp:3-95,f,-,
 Yf ING
iiiitersIty of- California itt Riv
ivilopnied owerio
 f:-Moditied test

peed of Ret
owerof:Rtaditi

:11iMusic-al-Aptittide
.-Pitclkdis-crimin-a 6-
-discrimination
- Toms- intensity= disciiininaton
-

eviations
--c
-
 SINGER

equation -that mould=itend--to

gr-ediction_:=4
-rectil hie ity__ _

in_ the b iv arat t r elati oni h ip:betwe en e ach -predictor itheregrets ion-
equation_withtheicrite ionIneasure;:bothirecti i ear
in-

coefficients ---Ivere = computed = - between= _each variable - and _Speed of

Readintat!each- gindeleVel;Wchi-ScfnareAetti -t=eath:retulting-Tpair
_ of bivariate relationshi revealed _no significant differences .7-There----
---,--- -_ _-- .-
-:--fore,===the-standarctiencit, ate--Effore essionietinationn-f_-,
irstu 14. nu- throughout _grange io

CoMputations were performed on- iherTBM 7090 Computer at the Univer-

sity of Califoitia,llerkeley,z4 meins offRSCAT Program -written by Mkt-
yamafor the Carnegie- Holmes - Reading Project.` =
8 Correlation Matricerforeach grade level appear in the-final report (Singer,
1965).
 cent contribution made by ictor to the variance of the
criterion is then computed from is multiple ession equatiort.
- Holmes extended-the technique to variance of each
predictor that could be= psychoeducationally analyzed from, the= re-
maining pool= of predictors by _eliminating= themajor = criterion = and
finding which of the remainin -variables in e correlation matrix

The results of the -Doolittle-Analysis= for grades three

through six appear in Table 'The =table shows the order of selection
*A detailed presentation of the theory, assumptions, and statistics, written by
Professor Holmes, appears elsewhere in is issue. Mr. Price S er modified the
substrata analysis program for the IBM 7090Cornputex at the University of Cal-
ifornia, Berkeley, and assisted in all statistical analysis for this investigation.
10 A paper on developmental model for Power of Reading was read in an
invited main address at the National Reading Conferenie, Dallas, December 3-5,
1964 (Singer, 1964). A final report to the U.S. Office of Edutation on the con-
plete project was due July 31, 1965.
 ___=Comps-itation of shrunken
Valiance N
increment_
Grade,Seq.' iubstratetfactor tnames (VyZ) )th _ 113

Speed of_word-perceptiiii .6593 1.0045' .5811

-Visual verbal meaning _ : _ ;58301_ _ 1.0091-- _ 1.5883= -A117-- .6418
-=D6t.flgure_ ground was. = -.0256= -.5575-
affix-et and roots .0178= ;5398 1.0183-: -'- .5495- . 4505- .6712_ _
revertals= .0188_1_ =- 5 _-- 1.0230 _

w s run en multiple correlation coe cient, e co=

efficient from which chance error as been remove 1931;

where
208 THEORETICAL MODELS AND -PROCESSES OF READING--

RESSION3 UATIOW_ ATAANDI ER NTRIBUTIONUM

- IN-SPEED OPfREADING FOR_ RADES IMRE THROUGH
SINGER 209
,
The regression equation for each grade level can = be visually
represented as a set of predictors for Speed of Reading. These pre-
dictors are shown in Figure -1 as =a diagrammatic model of the develop-
ment of Speed of Reading in -.es three throught six.
 11 The small, negative contributions to variance -Word = Reversals at
grade three, Spatial Relations at grade four, and Atidito-- Verbal Abstraction at
grade fife are interpreted =as mtaning that these factors a as-slatistical sup-
ipantowi-hilikceh purifi As such, they augment_ the = combined contributiots io_var-
the other predictors Make to S of Reading tbin, 1957; Holmes,
 Variance
explained 47%
Vat since not
accounted toe: 53%
Total: I00%
FIGURE aiding
11,1
three tO siit: The isOdel shows fOr 1112bIltrata Ilzactors I

SPeed ed'iteadiPg.'717he variance 11'9!

Processing VisttalSilmt4 htob*

I
'"11"""111,111,111'1111,1,111,1111:111111,,!:11:11,1,11,11:,:,,:11::111'd1,11,',1',
THEORETICAL -- MODELS AND-PROCESSES OF READING
READING
 _

__-_-r_e---,-

_
 FIGURE DeveloPmental integration onel an tem for* 3 e 0 rea ing in ea through six The
shows that Recognition of Affixes and accounts n s_ ed of Reading at the,third
grade level, but only indirectly through o at 6 e 'eve
 _-_Stitani -_ n-clusions =_ _-

-- -_---

e _major_d= eve- lop--:---nen ta hypothesis-i n f n; e- factor

=substrata=sa
then of-readitiOv*leite inisteringtelettedi--battoy_Of
Varia_ es-:tt-approkirtiate E, --: ti i In--_eatkgraddi_three-throngh__
sift. SubStrati-'an e fes nit irgricor rel ati n _m tr ite s_ C ori -
_

fi the 'statisticall
is atet_ ems= at Auantitative- and_-
izitionachan -et:in--in-strata= -c-tors_-are,,in- ci;;_aiSoCiatediivith
, ___ ----__- -- - _ _e:-

eVen-p-Ment--_ht-S ed---ol=Ret A theoret -ical=model;_co_nstructed

Is

to= depic-t--th-ese su trata- ctor- anges,-i -6-rettale J=the= following

detelnpinentaliitren =1-1_ ne =_workitigyitoto=for attaining
Speed,oUlteadingiundergne0=a'Aeveloprn-entaLshift- fronva 'predom-
inance of =visual _perceptitalabilities---it-=-the :third: grade-lo a more
eiliiitableerganitation-ofivisual_perceptitaLand-w_ordnieining factors
 te;=accura a -rOcesalinAtarnint--_-A_
Selected reddings-"on7t -the process
Neit-Nork:=OitfordAtniversi*Preit196L-Tp;_=388,4
nAvts,--F.7R.=The-sUbitratazfactortheOryOf-Ereadinghuman_Thysiology: as a-_-factor
j;-A.= Figurel;,;(4d),4niprotienfintol_--ra-ding_s-throtigiielasir-o-orn--
=practice.- =Neivirk;:Dellilire:7International---Reading:fAls-oCiation;-E _-1964-. --Pp; _-

-292-296.
trEi-ENTit; MitinaUforDitrrell,Stillivan reading- capacity
and teadintathievementzteiti.--Ne*Noit:_-_-:-WorldiBook;_193T-7
=and --
GATES, Themannalof direetions_tydritheiGateilReading-=:SurVey-_.: ew-Yorkt
Teachers College, ColtunbiaUnivertityr13ureau-ofzPublications,=1953;--- _:-=_
Thesmanual-o or the_fGatis_ReadingzSuroey-.,New-Nork:_
-Teacliera_College,-"ColumbiacUniVersity"--Bureati-ofiPublicitions, -1958;:
funitional-tribtor effidency_off-theEeyea,anthits_z_relationr_to-reading.
University Of Xduentiorf,-1953,:2,--159-232;
-GILBERT'''. speed --of- protessing:iiinal-stirrinli---and- to reading.
Journairef'Educational-Psychology195%-15-,18.44-_f_
itotAtEs, J. a.- Factors _underlyingfinajor--reading-diiabilitiei,at-- the college level;
Unpublished- doctoral_ -dissertation, -University :Of talifornia,-_Berkeley, 1948.
HOLMES, J. ti.__The-Substrata-faetor_theory of reading. Berkeley:- California Book
1953. (Out of-print) _ _
-
 A Theory of Language, Speech, and Writing'
DAVID WREED
University of California at Berkeley

RECENT AWARENESS that the science of linguistics may lie =of consider-
able relevance to research in elementary reading=instruttion could
produce-Very unfortunate== effects=if-lingdittitt :Shotild-"oe :mitunder-
stodd as giving-Wholehearted-support to phonic methods oUteaching=
reading. It is true that during--the4hirties; fortiesanct-fifties of this
century Mostilingdists_ emphasized th-e-spriinarytiattire of speech and
the secondary nature of writing. A feiv were-incautious enough,to
state that writing is-merely_ a,secondary-tepresentation Or identifica=
tion of speech. Many -non- specialists, in their efforts to apply the
Yr' findings-Of-linguistid to the accomplishment -of practical ends -con-
cluded= that speech is-the language. It demonstrated shortly
that these -Snore extreme-positions aboutthoprimacy of:speech over
writing are patently abstird.
Linguistic statements, like other scientific statements, have to be
understood in terms of the:_historical context that produced them.
During most of the nineteenth century and until about 1925, the
primary interest of linguists was in_ historical linguistics -and in the
reconstruction of prehistoric stages of descendant languages. This
interest naturally led investigators to-the written- records of earlier
periods is their primary-Source of information. Around -1925 Several
new interests combined to shift attention for writing to speech.
Ainong these were interest in recording - and analyzing the very large
number of unwritten languages in the world, interest in more ac-
curate descriptions of modern languages that have systems of written
1Tix., paperz,as presented at an International Reading Assodation Con-
ference on Linguistics and Reading held in Detroit, May 1965, and published in
Brother Leonard Courtney (Ed.), Highlights of the Pre-Convention Institutes,
Institute VI, Linguistics and Reading. Newark, Delaware: International Reading
Association, 1966, pp. 4.25. The first part of the paper was also = published in
Elementary English 42, December 1965, 845-851. The second part of the paper
was delivered at the Annual Convention of the National Council of Teachers
of English, Boston, November, 1965.
219
 220 THEORETICAL MODELS. AND PROCESSES OF READING

representation, and interest in the comparison of dialects of living

languages. All of these interests fostered the development of phonetics
_
structural linguistics as more exact sciences. It Was entirely
naLaral that linguists would seek, during this period, to make' sup-
portable generalizations about the primacy of speech over writing.
It was observed, for example, that even in historical linguistics it is
necessary to interpret written records in terms of speech sounds be-
fore a systematic description of the historical development of the
language can be given.
-= Although all the linguistic interests of recent decades have con-
tinued to= the present -day, it= has =come - to =be- recognized increasingly
since the late fifties that phonologically based grammars are incapable
of producing adequate analyses of syntax. All of the newer theories of
grammar, including the transformational, have in common the
procedure from syntax to speech-and writing rather than the other
way around. And, as soon = as one views speech and writing as the
end products of grammar rather than as the starting point for gram-
matical analysis, it is inevitable that he will view the relationship of
speech and writing to one another in i new light.-

The- Primacy of Speech Redefined

Speech is certainlY primary and writing secondary in at least the
following respects: (1) Speech in some form is probably more than
a half million years old; writing is no more than one percent this
age. (2) Every contemporary human society has one or more lan-
guages, but of the thousands of such now in existence, a large ma-
jority have no accepted system of writing.- (3), The experience of
individuals with unimpaired hearing, nervous systems, and vocal
organs, who grow up in a society that has a system of writing, parallels
the experience of the race, in that speech is acquired first and largely
without conscious effort, whereas understanding of the writing
system comes much later as a result of extended and intensive study.
(4) In societies that have writing all. normal adults have a working
command of the spoken language, but only a minority have a com-
parable command of the written language.
These. statements might be summarized by saying that speech
generally enjoys a chronological primacy over writing. But it is
 221

interesting to note that even this chronological primacy- is not a

necessary condition in the relationship of speech to writing. One who -

is- totally deaf may be expected to learn to read and write before he
learns to speak. Indeed, he May grow up as a deaf muteone who
never learns to speak and is thus exactly =comparable to the adult
illiterate; who never learns to-read and write. In the-very- interesting
case -of Helen- Keller, who- was rendered deaf and- blind by scarlet
fever before she was-two-_yearsold, the primary system -of representing
the_Englishianguage-maSEa_tattilelystenii jni-which,symbols identify=
ing language are formed -by illefingers_of=orie=persorr_ in =contact with
the hand of another. Braille:becaMeltheEseCondary:systeM,- compar-
able-in functionito_-readingiand-writing-ifor most people,- and speech,
which Mks Keller -never- acquired _perfectly,- was---merely a_ tertiary
system-for_repr sentingErigkish. _

In-short, the_observed:relationships betuieen speech and writing

will-not _justify-our saying--that ispeethzis the-language or that writing
_ _-
is merely- a-,secondary representation:of ,speech- If _speech--were the
langtrage, then deaf mutes _would-beipeople who have=never learned
a languagein-spite-of the obvious lact_that some of them read and
write English, otherS Germari,-otherarenCh; andsoforth.-We-should
also be at- a -loss to explain -how skilled readers-Can, within a given
period of time, identify throtigh reading, _a string-Of linguistic forms
that cannot possibly be uttered by the Most rapid speaker in a com-
parable period of time.

The Limits of Phonemic-Graphemic Correspondence

Reappraised
It is also easy to prove, in terms that would satisfy the most
exact physical scientist or mathematician, that neither writing nor
speech is a direct representation of the other. No set of rules of
phoneme-grapheme correspondence, however numerous or compli-
cated, can tell us how to pronounce all graphic configurations or how
to spell all spoken forms. As an example of the former impossibility,
consider <read> or <lead>. To read the first of these, we need to
know whether it consists of verb + present, or verb + past. To-read
the second, we need to know whether it is a verb, or a noun. As an
example of the other kind of impossibility, consider a homophone
 222 THEORETICAL MODELS AND PROCESSES OF READING

like /red/. One cannot write this correctly unless he knows -it to be
adjective or, alternatively, verb + past.

The Whole Word Fallacy in Reading

I have devoted so much time to refutation of what I shall call
the phonic fallacy,= because I believe that linguistics is popularly
understood to support it. However, it would be most unfortunate
indeed =if the preceding =remarks were= understood to give aid and
comfort to supporters- of what =I shall call the "whole word fallacy."
This position is more ridiculous and more dangerous by far than the
phonic fallacy. =It is more ridiculous, because =it =is based on a more
fundamental misunderstanding of the interrelations between lan-
guage, speech, and writing than is phonics. It is=more dangerous
because it has been used for a longer period as one of the methods
of the basal readers, which have been employed in the vast majority
of instances in teaching American school=children to read.

Reading and Understanding

In its-most extreme, or, shall I= say, crudest, form, the whole
word position is that the English system of writing represents whole
words which are conceived of, not as linguistic forms, but as some-
thing called "meanings!' Reading, then, is viewed as a process of
identifying meanings from their written representations. If this
hypothesis were valid,- a child confronted with a string of graphic
configurations like <The dog ate the meat.> could demonstrate
his ability to read by uttering "The canine quadruped devoured
k
the flesh," or some other suitable paraphrase. Obviously, if our hy-
pothetical child were able to perform such feats consistently, we .1
should be forced to conclude that he could- read, but, at the same
time, we should also have to say that he had not directly demonstrated
his ability to read, but rather his ability to understand written
Englisha more complicated skill that is based on the ability to
read. In fart, the only adequate response to <dog>, if one is to
demonstrate that he can read it, is to identify the linguistic form it
represents by producing the conventional sign for it- in some other
system of representationas, for example, by uttering the syllable
/d3:g/
 223

The whole word fallacy, then, is based on a confusion between

reading and understanding. Reading is the identification of linguistic
forms from strings of written configurations that represent them, as
evidenced by producing the conventional signs for the same linguistic
forms in some other system of representation. Understanding a pas-
sage of writing is the identification of the meining of the linguistic
forms that have been identified from the writing, as evidenced by
producing, in any system of representation,_ =the= conventional signs
for different linguistic forms that-will be judged by native users =of
the language to have the same meaning. -Anyone who has learned to
read can read many sentences whose meanings = are=almost completely
unknown to him. I shall= now do so: "The-dieory of- functions= of a
complex variable deals with the differentiability of- complex func-
tions, analytic continuation, the residue theorem, and conformal
mapping." In reading this sentence, I had no difficulty whatever in
identifying all the linguistic formsboth the words and the syntactic
structuresbut I would find it impossible to produce a paraphrase
that would be remotely satisfactory to a mathematician. That is, I
read this sentence but I don't understand it
Before turning to more specific matters, =I should like to dear
up two possible misunderstandings that may derive from my effort
to distinguish between reading and understanding. First, let me say
that I think that understanding is of vital importance. Ifone assumes,
as I do, that any normal or above average child= should be able to
learn to read within two years, then most of the reading that he does
from the third grade to the end of his formal education and beyond
ought to have as its purpose the improvement of his understanding.
But I also believe that to introduce this second goal before he has
learned to read can only confuse him and delay his progress in learn-
ing to read.
Second, there may be those who challenge the notion that dif-
ferentlinguistic forms are judged by native users of the language to
have the same meaning. Isn't it true, they may ask, that two formally
differeht statements never have the same meaning? This objection
may be categorically disposed of: -No, it is not true. If it were true,
human communication would be impossible. Consider the case of a
college student who is asked on an examination to discuss the mean-
ing of Hamlet's third soliloquy or any other passage from literature.
If the objection expressed- above were valid, the only correct answer
 THEORETICAL MODELS AND PROCESSES OF READING

to the question would be a verbatim and letter-perfect quotation of

the passage. Consideration of this and similar examples may lead to
the formulation of a revised and weaker hypothesis to the effect that,
although two formally different statements always have different
meanings, there is a wide range of variation =in the degree of differ-
ence between statements. Although I believe in the validity of the
qualification, its attachment to the-false generalization will not enable
us to account for examples of the type cited aboVe, because, even
though some paraphraset of Hamles_ third soliloquy- might:be
judged superior to others; all paraphrases would kill be- judged
inferior to a yerbatim quotation. In=shott, I believe that the correct
hypothesiS aboutmeaning is that some formally= different statements
have the same meaning, whereat;namont-thote that dO not, there is a
Wide range of variation in _the degree of difference in Meaning. Only
with some such assumption = can we_ continuen_tia=grade_examinatiOns.
Probably some of yOu agree with my Conclusions about the
inadequacy of both the phonic-arid the Whole word methods of teach-
ing elementary reading, but would hasten tonpoint out-that= these are
only two among a number of methods of = reading instruction em-
ployed in the basal readers. I am sorry to say _that this argument
reminds me of Stephen Leacock's description of an upright woman
who raised her daughter according to the best Christian principles,
then taught her Mohammedanism to make sure. The phonic and the
whole word methods are contradictory. If one is right; the other must
be wrong. To use both is to spend at least part of one's time-teaching
wrongly, so as to mislead and confuse the learner. -If, as I hope I have
demonstrated, both methods are baied on fallacious theories, to use
the two methods is to spend all of one's time teaching wrongly. What
is needed at the outset is a better theory of the interrelationship of
writing and speech.

Linguistic Forms Variously Represented

in Different Symbol Systems
In the theory I have been c -king to develop, the term linguistic
form is of critical importance. It is certainly not original to me,
having been used at least as early as in Bloomfield's Language in
1933. Bloomfield defined linguistic forms as the grammatical units
 REED 225

of language, consisting of morphemes (the smallest meaningful units)

and what he, called taxemes of order, selection (primatily of form
classes and constructions), phonetic modification (by which the form
of morphemes -is modified when they enter into combination with
. other morphemes), and modulation (i.e., intonation). Probably no
linguist today would analyze grammar in strict Bloomfieldian terms.
Phonetic modification, for example, would -be analyzed not as part of
the grammar itself, but as part of the transition from language to
speech, known as= morphophonemics; and modulation I% Jaid have to
be defined abstractly enough to permit its actualization- as either
intonation or punctuation. The significance of Bloomfield's analysis,
however, lies not in such details but in the recognition =of =a linguistic,
or grammatical, level, of which phonemic and_ graphemic representa-
tions make up no part. In the = discussions= that have raged over Bloom-
field's chapter on " Meaning, " = the placement of__that chapter in the
total book has been generally overlooked. It= =comes - immediately after
the chapters devoted to phonetics and phonemics and immediately
before the chapter devoted to morphology. Moreover, although many
Bloomfieldians have sought= to avoid = meaning iltogether in their
linguistic analyses, Bloomfield is explicit on the point that meaning,
regardless of how resistant it may prove to analysis, is an indispens-
able feature of every= linguistic form.
A linguistic form, then, is a linking of _a unit of meaning to a
physical representation in terms =of= a convention_ al system such as
speech or writing. Although some linguists have preferred to think
of linguistic forms as physical units that signify meanings that are
themselves non-linguistic, and others have thought of linguistic forms
as semantic units that are identified by nonlinguistic physical con-
figurations, I am of the =opinion that it is preferable to think of
linguistic forms as simultaneously having semantic and physical
features, neither of which is paramount. This may be explained by
the analogy of a- coin, on- which "heads" represents the meaning,
"tails" the physical representation. It would be absurd to speak of
either side of the coin as the "real" coin, in contrast to the other
face which is merely a secondary concomitant. To improve the
analogy, we might imagine a country in which several nationai banks
are authorized to strike coins. There is a prescribed national stahtlard
for the size, weight, and metallic content of each denomination of
 226 THEORETICAL MODELS AND PROCESSES OF_ READING

coin, but each bank may place its own design on the two faces, so long
as _these are always the same for the denomination struck by one
bank and different for different- dmominations. -We actually have
something like this system, since the designs on our coins=are- changed
from time to time. In any -event, --without laboring the analogy
further, the variant-designs on the-obVerseface of a-given -denomina-
tion Would-correspond-;_to the variant meanings-that may be atsociated
with a given linguistic _form, _whereas -== the Variant designs on ---the_=
reverse face would correspond toVarianCr_
variant-representations
hnguistic-fonn in speech, ivnung ta eSt4fCbrieCt=association-
of a written symbol with a spoken symbol is ,accomplished through
identification of the liriguistit-form-rather than,themeining,In the=
saMemay-thatiishOwn4 corn with a picture of Monticello on it one
know that= there are other coini-ofjhesaine-Value,nickel..-
=

that- bear a-torresponding-ipitture ofi-a-:_buffalo.-No--knotvledge Would'

be _required of ,what appears on --the-coppositejace, of either Coin,
thotigh__the ordinary User of American coins would have
such knoWledge. _

Further -clarification of-the concept, linguistic form, may be

secured by coMparing ittoa-relatectbut different concept, commonly
termed the "nonsense syllable?' I Must first comment on-the singular
ineptness with which theseunitshave been named-, The term syllable
suggests that they are phonological units of a particular length, yet
they may equally well -be graphic- configurations, and their length
may range- from lets:- than a -syllable to many syllables. The term
t nonsense indicates that- they- are devoid of meaning, but anyone who
has encountered "burbled" or "snicker-snack" -in "Jabberwocky"
knows that this is not true. A better term than nonsense syllable
Might be-pseudo linguistic form, since the term is meant to cover
graphic or phonological cOnfigurations- or sometimes graphic and
phonological forms that are associated with one another,-all of which
are likely to be assigned to -a common area -of- meaning by native
users of the language, but have in-common the fact that they do not
represent any current linguistic form in a given language; Many non-
English graphic configurations like <gough> do not correspond to
anything in speech, since they have no -linguistic form to serve as
intermediary. We may imagine- <gough> to be pronounced /gu: /, C

/rub /g3:f/, /gag, etc., indifferently, on the analogy of through,

f
 227

bough, cough, and tough. Likewise, many non-English phonological

configurations like /fo:t/ have no graphic correspondent forms. We
may think of /fo:t/ as beginning- with <f> or <ph >,- with its
vowel spelled <0 . e>, <oa>, and so- forth. Furthermore, the`
semantic featu_r_e_of_a pseudo linguistic form is usually much less
precise than that of a true linguistic form. It seems likely that many
forms which finally attain linguistic status pass through a transitional
stage as pseudo linguistic forms.

Linguistic Forms Regulany Represented

in Any Given Symbol System
If both writing and speech are actualizations of linguistic forms
and if neither is a representation of the other, just where does this
leave us as regards the teaching of reading? In order to proceed, we
need to take into account two apparently contradictory characteris-
tics of representational systems: they are always absolutely regular
and absolutely arbitrary in terms of the units represented. We are
familiar with statements, that have often-been ridiculed by linguists,
to the effect that English is not a phonetic language. If this statement
has any meaning at all, it should mean that the linguistic forms of
English are not consistently represented in speech by phonological
symbols. Given the knowledge that there is- a linguistic form in
English that means "canine quadruped," I might sometimes repre-
sent it as /ho:rs/, sometimes as /nO:son/, and sometimes as /nui:ni.
If this kind of- irregularity existed, spoken =communication in Eng-
lish would clearly be impossible. Parallel to the statement that
English is not a phonetic language is the often-heard allegation that
the English system of writing is highly irregular. Once again, if this
statement were at all meaningful; it should mean that the linguistic
forms of English are not consistently represented by graphic sym-
bols. I might represent the linguistic form that means "canine quad-
ruped" as <horse>, <notion>, or <moon>, as the spirit moved
F
me. If this kind of irregularity existed, written communication in
English would be impossible. In short, where the question of regu-
larity is concerned, the only hypothesis that makes sense with re-
ference to English speech and English writing is that both are
absolutely regular representations of the English language. It does
 228 THEORETICAL MODELS AND PROCESSES OF READING

not follow that either is a regular representation of the other, al-

though, as will be pointed out as soon as we have considered the
arbitrariness= of representational systems, since English writing is
alphabetic, there is a degree of correlation between spoken and writ-
ten symbols for the same linguistic forms.

Linguistic Forms and Learning to Speak

and to Read
Readers of linguistic literature during the last forty years are
probably more familiar with the concept that representational sys-
tems are arbitrary than that they are regular. There is ,nothing in-
trinsic in the linguistic form that means "canine quadruped" that
=

makes it essential for me to pronounce it /da:g/ and spell it <dog>.

Except for the arbitrary conventions of= English,=I might always pro-
nounce it /mu:n/ and spell it <tali>. It follows that the learning
of linguistic forms, which, I must remind you, consist of the link-
ing of a meaning or meanings With an arbitrary but' wholly regular
representational configuration, must at first consist of laborious trial-
and-error imitation and memorization. This is what normal children
=

do at the stage of learning to speak their native language, or what

deaf children do in learning to read their native language. Gradually
the child comes to recognize certain regularities within the grammati-
cal system and the representational systemthat, for example, there
are a limited number of different sounds in speech and that they
occur only in certain patterns. From this point on, his progress in
learning improves, because he can call into play higher mental facul-
ties than mere imitation and memorization.
The position of one who has already learned the grammar of a
language through one of its representational systems and then turns
to learning a second representational system for the same language
he has been learning is quite different from the situation described
above. The normal child who has learned to speak English and is
proceeding to the study of reading and writing is in this second p ci-
tron. He can begin to learn to associate graphic configuration wi,.
linguistic forms that he already knows. Only after he has learned to
read and write his own stock of linguistic forms should he be en-
couraged to make use of reading to learn further linguistic forms.
 229

How can the construction of reading materials take advantage

of the fact that the child is already a native speaker of English? In
the briefest possible terms, a set of rules can be devised, ordered from
the most specific rules, that account for single morphemes, to the
Most general rules, such that, given the spelling of English mor-
phemes, their spoken representations may be arrived at. The list of
rules will be long, and it is not necessarily being urged that= all or
even -any of these rules should be taught to -the child. It is urged,
however, that elementary reading materiais should be structured in
terms of the rules. This principle does not prescribe any particular
structure: Two organizations, of which one is the exact reverse of
the other, might both be structured in terms of the rules. Once the
rules have been set forth, however, -it- will =be the responsibility of
psychologists and classroom teachers to determine which organiza-
tion, deriving from the rules, is most conducive to karning. On
this question, the linguist, as linguist, cannot be expected to have
a valid opinion.

Linguistic Forms and Other Units of Language

If we are to come to a better understanding of the complex
relationship between speech and writing, we need to characterize
as accurately as possible those particular linguistic forms that are
actualized as English sounds and English letters. The phoneme
of descriptive linguistics is entirely inappropriate for this purpose.
Insofar as it may be useful at all, it is as _a unit of speech rather than
of language. Phonemes must bear a direct relationship to the sounds
that one speaker of English utters, and they cannot, by their very
nature, account for the morphological structure of his language or
for the relationships between his speech and that of other speakers
of the language, including that of speakers of other dialects. Graph-
emes, as units of writing rather than of language, are also inappropri-
ate for similar reasons.
Let us consider, first, how morphological structure is a factor
in the determination of linguistic forms and therefore affects the re-
lationship between speech and writing. The words metal, rebel, civil,
Mongol, and cherub have the same vowel phoneme in their second
syllables in most, if not all dialects of English. Yet in English spelling
 230 THEORETICAL MODELS AND PROCESSES OF READING

the vowel of these second syllables is represented by- five different

letters. If we try to describe this situation in terms of direct phoneme-
grapheme correspondence,'we shall need five rulesone each to ex-
plain the spelling of the unstressed vowel in each of these- words.
There are undoubtedly thousands of English words that present
comparable problems, so that the student of English reading who 1
follows strictly the method of phoneme-grapheme orrespondence
will have to memorize the spelling -of thousands of apparently iso-
lated-eXamples. If, however, -we= -consider = the following - words- that
are derived-froth-those theioriginalilistmetallic, repellion, civil-
ian, Mongolian,- and chernbie-,4-- is obVious thati- they contain five
different vowels that arein= a consistent relationship with the spell-
ing. If we assume that metal, for example_ , consists of the= same lin-
guistic forms (which would generally be called= inorphophonemes),
regardless_ of whether metal has /4/ in its second syllable when it
occurs. alone or // in its:second-Syllable when it occurs as part of
metallic, we shall be in a position-to account rdatively_sirnply for the
spelling of the unstressed vowels in these and thousands of other
words. In terms of practical applications. of this principle it is sug-
gested that, wherever-other considerations (such as unfamiliarity of
the derived form) do not militate against it, the longer derived
form ought to be taught first, to exemplify the basis for the spelling,
before the shorter base form is taught
Consider, next, such pairs of words as defy, defies; rely, relied,
and any number of other words_ that are spelled with <y> in final
position, but <ie> in medial position; In such cases it is dear that
English pronunciation signals the linguistic identity of the forms
/difiy/ and /rilay/, even though these forms-are spelled differently
as bases and as part of derived forms in conventional English writ-
ing. The method of phoneme - grapheme correspondence, which can-
not take account of the morphological identity involved, will re-
quire the memorization of long lists of words in which <ie>
spells the diphthong /ay/ of defies rather than the vowel /i:/
of relieve. Recognition of a linguistic form (in this case a morpho-
grapheme) that occurs in defy, rely, and derived words that contain
these and similar elements would obviate this difficulty. The practi-
cal application of this principle would be to teach base forms
like defy and rely first; then to teach a simple rule by which deriva-
 231

tives are formed from such bases, rather than attempting to teach
derivatives as independent, unrelated forms. The correct procedure
A
has always been followed by reasonably intelligent teachers, but it is
important to obseive that it is an outright contradiction of the
method of phoneme-grapheme correspondence.
From what has been noted up to this point, it should be obvious
that the linguistic forms that are represented in = English speech
and writing consist in part of morphophonemes that are different
from the phonemes employed by most, it not- all, speakers in cer-
tain words, and in part of morphographemes that are different from
the graphemes employed =in conventional English spelling in cer-
tain words. It remains to consider how= dialect difference is a factor
in the determination of linguistic form and therefore affects the
relationship between speech and (writing.
Let us consider examples, presented first in their = written
form: 1) <cant>, meaning = "to tilt or slant," or, . alternatively, "a
jargon," 2) <can't>, the contraction of can and not, and 3) <Kant>,
the German philosopher. The first- is pronounced /kmnt/ in most
dialects of English and America; the second- is pronounced /Itx-nt/
in the majority of American dialects, but /Ica:nt/ in Standard South-
ern British and a few American dialects; in the third form, these
vowels are reversed for many speakers, so -that the pronunciation
/ka:nt/ prevails in America, but /kwnt/ is the= only pronunciation
given by Daniel Jones for Standard Southern British. Clearly there
is a linguistic difference among these three forms, and English writ-
ing represents this fact by employing three different spellings for the
forms. In the commonest varieties of British and American speech,
however, only two different pronunciations are employed, Britain
equating forms 1 and 3, America equating forms 1 and 2. English
writing spells all three forms differently, in order to show that they
con_ tain three different diaphonemes. Diaphonemic spelling enables
an American reader to associate the "broad a" pronunciation with.
<Kant>, and the British reader to associate it with <can't>.
The opposite of this situation may be illustrated by the forms
pronounced /wayz/, /sayz/, and /ayz/ (that is, the verb-forming
suffix.) The first and second are spelled <wise> and <size> on
both sides of the Atlantic. The verbal suffix is spelled <ize> . in
America, but <ise> in England. The pronunciation of the suffix in
 232 THEORETICAL MODELS AND PROCESSES OF READING

both countries reflects the linguistic identity of the form, and the
different spellings are a matter of variation in written dialects. All
'`It ree forms contain the same diagrapheme, as indicated by their
pronunciations.
We can now say that the linguistic forms that are represented
in English speech and writing consist in part of diaphonemes that are
different from the phonemes employed =by some speakers in some
words and different from the phonemes employed= by= other speakers
in other words, and in part of diagraphemes that are different from
the graphemes employed in conventional English spelling in certain
words. Combining this conclusion with the one previously reached,
we might say that English spelling is "dia-morpho-grapho-phonemic,"
but we= cannot say that the_linvistic =forms represented in speech
and writing are qdiamorphographophonemes," or the like. Actually
the linguistic fornis represented in speech and writing include some
diaphonemes, = some diagraphemes, some morphophonemes, some
morphographemes, and, in all those cases that linguists have thus far
emphasized, where there is a one-for-one correspondence between
phoneme, linguistic form, and grapheme, some mere graph-
ophonemes. We need= a= name for the class of which these various
"emes" are members, and= =we cannot continue to call it merely
"linguistic form," because there are many other linguistic firms, such
as "noun," that are not represented directly by phonemes and graph-
emes. Under these circumstances, I shall venture to coin a new term,
Unguent, to signify the smallest linguistic unit, such as actualized by
phonemes in speech and by graphemes in writing. The term is
formed on the analogy =of proton, electron, neutron, etc. in physical
science, and is similar to such terms as phonon, morphon, lexon, and
semon, that are used in stratificational grammar.
The abstract nature of such a concept as "the English language"
cannot be overemphasized here.- Just as one cannot meaningfully
say, "I drive the American automobile," so he ought to recognize
that when he says, "I speak the English language," he is employing
an oversimplified statement to express some such notion as, "I em-
ploy a system of speech that is derivable by a list of rules from an
abstract structure known as the English language." In terms of the
analogy, a given dialect is like a particular make of automobile, say
Buick, and my idiolect is perhaps analogous to a particular 1949
 233

Buick sedan I once drove. If we wishjo discuss the American auto-

mobile in the abstract, we may employ such terms as battery, spark-
plugs, brakes, etc., each of which may be thought of as analogous to
morphophoneme, diagrapheme, etc. Linguons, in these terms, cor-
respond to automobile partsj without reference to the *make of car
or the function of the part.
I should like to test the linguon hypothesis against a particularly
recalcitrant body of data: words that contain one of the linguons2
//a://, //3://, or / /3//. These words are spelled with <a>
<au(gh)>, <aw>, <o>, or Gough>, and pronounced with three,
two, or one low back vowel phoneme in differeui dialects, with
considerable variation, in America at least, in the incidence of these
phonemes in different words. Ir. order to make the body of data more
manageable, only monosyllables containing no more than one post-
vocalic consonant have been studied. Such an arbitrary restriction
may, of course, either simplify or complicate the rules that must be
adduced -to account or sound-to-spelling correspondences. Any the-
ory of these correspoLiences must, ultimately, account for all the
data.
It may be asked how one can determine what linguons a word
contains, since linguons are often not identical to English graphemes
or to the phonemes employed in a particular idiolect. This is a
particular instance of the general question, how does one formulate
a hypothesis, to which the answer is that no- one knows. In this
particular case, one would, of course, assemble the words under study
in their graphic representations along with their principal pronuncia-
tions as indicated in dictionaries, linguistic atlases, and other avail-
able sources. He would then postulate the linguon constituency of
each word in a manner that seemed likely to permit the statement of
linguon-phoneme and linguon-grapheme correspondences in terms
of the smallest number of rules. One can never be certain that he has
found the best formulationonly that one particular formulation
is better than another. Thus, it might seem natural to suppose that
the words for, nor, Thor, tor, and war contain the linguon //3://,
since all are pronounced with /3: / in dialects that have this phoneme.
If we postulate this linguon in these words, however, we shall require

2 Linguons will be written between double virgules.

 234 THEORETICAL MODELS AND PROCESSES OF READING

the following rules to convert the linguon into the proper phonemes
and graphemes:

*(1) //a:// /3:/3

(2) Ho:// > <a> [w_r]
(3) //3:// > <o>
Of these rules, (1) is required for other words with //3://, and
so its statement here does not represent any addition to the required
set of rules. Rules (2) and (3), however, are not needed for any other
words, and may be described as due to our decision to assign //3://
to the words in question. If we had postulated //3// in for, nor,
Thor, tor, and war, we should have needed the- following rules to
convert the linguon into the proper phonemes and graphemes:
(4) //3// /3:1
(5) //3// 50>
Rule (5) is required for other words with //3//, but rule (4)
is necessitated by the assignment of //3// to the words in question.
Thus, we see that it will be better to assume that for, nor, Thor, tor,
and war contain //3// rather than //3://, because to do so will
enable us to convert linguons into the proper phonemes and graph-
emes with one fewer special rule.
Some 225 monosyllabic words containing no more than one
postvocalic consonant phoneme, but containing one of the three
vowel linguons in question have been studied. Of these 16 contain
//a://, including ma, pa, calm, palm, par, tar, and ten other words
with postvocalic r. Another 131 contain //3//, including pop, top
(and 11 others with postvocalic p), pot, tot, (and 12 others with
postvocalic t), botch, notch, Scotch, cock, bock (and 10 others with
postvocalic k), cob, bob (and 10 others with postvocalic b), pod, Tod
(and 8 others with postvocalic d), dodge, lodge, hodgepodge, tog, cog
(and 7 others with postvocalic g), Tom bomb, Don, John (and 5
others with postvocalic n), Poll, doll (and 3 others -with postvocalic
1), tong, bong (and 8- others, with postvocalic ng), doff, golf, scoff, off,
$ Rule (1) should be read, "The linguon long open o is actualized as the
phoneme long open o; rule (2) as, "The linguon long open o is actualized as the
grapheme a in the environment following the linguon w and preceding the
linguon r."
 235

Goth, moth, wroth, toss, boss, (and 3 others with postvocalic s), posh,
bosh, josh, gosh, of, Oz, tor, nor, for, and Thor; watt, squat, watch,
squad, wan, wash, squash, was, and war. The remaining 78 contain
//3://, including taught, caught, naught, aught, Maugham, and
Vaughan; ought, bought (and 4 others with postvocalic t), Sauk, auk,
baud, daub, Maude, Paul, maul, Saul, pause, and gauze; hawk, pawed,
cawed (and 6 others with postvocalic d); pawn, dawn (and 4 others
with postvocalic n), shawl, yawl, awl, paws, taws (and 6 others with
postvocalic z), (Bryn) Mawr, paw, taw (and 12 others with final
//o://); talk, chalk, (and 3 others with postvocalic k), tall, call (and
8 others with postvocalic 0.
It is found that 21 rules are needed to convert the vowel linguons
of these words into their correct English graphemes. Of these 21
rules; 18 relate to the linguon / /3: //, two relate to / /3/ /, and one
relates to / /a: //. The rules follow:
3
(1, 2) //3:// -4 <au> [taught, caught] A
(3) //3:// --) <ou> [X +
(4) //a:// <au>
(5) //3:// -+ <awe> [awe]
(6-10) //3:// -4 <aw> [shawl, yawl, awl, Mawr, hawk]
(11,12) //3:// -4 <aw> [
(13-16) //3:// -+ <au> [Paul, maul, Saul, haul]
(17) / /a: 1 I -) <a> [.._1]
(18) //3:// -> <au>
(19) //3 // -) <a> [w--]
(20) Ho // -* <o>
IL
(21) //a:// --) <a>
These are ordered rules, which is to say that they must be
applied largely in the order in which they have -just been presented.
Specifically, rules (1-4) must be applied in that order with reference
to one another and must precede rule (18): Rule (5) must precede
rules (6-12) and rule (18). Rules (6-10) must precede rules (7-18).
Rules (11-12) must precede rule (18). Rules (18 -16) must precede
rules (17-18). Rule (17) must precede rule (18), and, finally, rule
(19) must precede rule (20).
- s /ow is a linguon represented by <gh> and /0/ (i.e., "zero"). /1+// is
a linguon representing morpheme boundary.
 236 THEORETICAL MODELS AND PROCESSES OF READING

Turning to the conversion of these vowel linguons into the

phonemes of my idiolect, I find that 20 rules are needed. Since I
havt only two vowel phonemes, /a:/ and /3:/, in this phonetic area,
I need ozly two rules for //a:// and one rule for /13://. Seventeen
rules are required, however, to explain which words with //3// are
pronounced with /a:/, which with /3:/.
(1) na://-* /3:/ [
(2-7) Ho // /a: / [doff, Goth, posh, bosh, josh, gosh]
(8-15) //3 // /3: / [mock, dog, log, fog, hog, gone, on, golf]
(16-18) //3 // /3:/ jr, ng, F-J5
(19) //3//-4/a:/
(20) //- it -*
The-number of rules necessary_ to make these conversions is due
in part to the mixed character of my idiolect. The rules make clear
something that I had not previously knowh: In words that I learned
in my native dialect area, rules_(16-19) plus one additional rule
(18a)//3// /3:/ account for all of these //3// words
except mock, gone, on, and. golf. My pronunciation of mock is prob-
ably a folk-etymological back formation from mawkish, and golf
could probably be accounted for in a study that includes rules for
consonant,linguons by a rule that deletes / /1/ / in certain phonetic
environments, after which golf is accounted for by rule (18). Only
gone and on seem to have been anomalies in my native dialect. Thus,
if the effects of dialect mixture are eliminated, only 7 rather than 17
rules are required to account for Hall in my native dialect. Standard
Southern British and similar dialects of American English would
require only rules (16, 20). Northern Middle West dialects in which
//3// becomes /3:/ before //r//, otherwise /a: /, would require
only rules (16, 19, 20). Dialects like those of Western Pennsylvania
and some parts of the Far West, that have only one low back vowel
phoneme, would require only three rules, probably of the form,
(1) //a:// /3/, (2) //3:// /3/, plus rule (20) from above. 5

Only nine rules are needed to convert the vowel letters of these
words into the correct linguons:

5 //F// represents any voiceless fricative.

6 This rule is to be read, "Replace linguons with phonemes."
 237

(1) <ou> //a://

(2) <o> //3 //
(3) <au> //a://
(4) <aw> //a://'
(5, 6) <a> --> //a:// [__1k,._11]
(7, 8) <a> -9 //a // qui)
(9) <a> //a://
It should be noted that, in a system that converts consonant
graphemes int? linguons, rule-(8) will be unnecessary, since <qu>
will already have been converted into //kw//.
Fifteen rules are needed to convert the vowel phonemes of my
idiolert into the linguons of English in these words:
(1, 2) /3:/ //a:// [ma, pa]
(3'6) /3:/ /13 // [mock, gone, on, golf]
(7) /3:/ //3:// [(Bryn) Mawr]
(841) //o // [__g, __ng, ___F]
(12) /3:/ //a://
(13,14) /a: /-* //a:// _im]
.(15) /a:/ > //3 //
Rule (3) is necessitated by the mixed nature of my idiolect, and
rule (6) would not be required in a study that included rules for con-
sonants, since it would come under rule. (11) after the deletion of
/1/. Standard British English would require only rule (7), rule (10),
and a rule of the form /_../ --) //_.// to indicate the same relation-
ships. The simplest form of Northern Middlewest dialect would
require rules (7), 10), and (12-15)a total of six rules. Dialects like
that of Western Pennsylvania would require a much longer list
of rules, that will not be specified here, in order to sort the one
phoneme /a/ into three linguons.
It remains to inquire about the relative difficulty of converting
writing to language to speech in these words, on the one hand, as
opposed to converting speech to language to writing on the other.
The following table indicates the number of rules required to per-
form these operations in the case of my idiolect, of my native South
,Midland dialect, of the simplest Northern Middlewest dialect, and of
Standard Southern British:
 238 THEORETICAL MODELS AND PROCESSES OF READING

Graph. Ling. Graph. Phon. 0 Ling. to Phon. to

to Ling. to Phon. to Phon. Ling. Graph. Graph.
Reed 9 + 20 = 29 15 + 21 = 36
S. Micll. 9 + 10 = 19 13 + 21 = 34
N. Midw. 9 + 6 = 15 6 + 21 = 27
British 9 + 2 = 11 3 + 21 = 24

Column 3 is the total of columns 1 and 2, and column 6 is the

total of columns 4 and 5. It will be observed that the two total
columns indicate that it is always easier to move from writing to
speech than it is from speech to writing. The differences are greatest
for the relatively "pure" dialects. Since elementary school children
ordinarily have much less dialect mixture than adults, the figures
for the pure dialects assume increased importance. Of course, we
cannot know at this stage of our research whether these relationships
will hold when a larger segment of the language is studied.
Superficially these figures suggest the intuitive reaction that it
is easier to learn to read than to write. This interpretation will hold,
however, only in the normal situation of a subject who knows the
spoken language and is trying to learn the written language. If we
assume a formerly deaf person who can read and write and is trying
to learn the system of English speech, the same figures suggest that it
is easier to learn to speak than to comprehend spoken English. One is
reminded of Abe Burrows' statement, "I speak French, but I don't
understand it."
In conclusion, what has been presented here is what I take to be
a set of facts about the relations of speech and writing to a small seg-
ment of the English language. What the implications of these facts
may be for the organization of reading materials and for classroom
procedures in reading instruction will have to be hypothesized by
psychologists and reading experts and tested in the laboratory and
the classroom. These are not matters upon which linguists may
safely pontificate.
Psycho linguistic Implications for a
Systems of Communication Modell
ROBERT B. RUDDELL
University of California at Berkeley

THE ROLE of- psycholinguistics in studying language skills learning is

undoubtedly more powerful than either that of linguistics or psy-
chology considered separately. Although the linguist has, offered a
description of language competency through possible systems for
describing and generating sentences and the psychologist has pur-
sued learning theory from various viewpoints, the psycholinguist
is interested in exploring the psychological reality of linguistic de-
scriptions. In brief, the viewpoint of the psycholinguist as described
by Miller is -that of accepting "a more realistic conception of what
language is" (3). A major goal of the psycholinguist which may be
realized in -the distant future is the development of a theory or
theories of language performance.
The relationship between psycholinguistics and reading in-
struction is apparent if one views the former discipline as develop-
ing an understanding and explanation of language processing and
the latter as having its central focus on the enhancement of the
ability to decode and comprehend language. This relationship is
even more obvious as one considers the definition of reading as
a complex psycholinguistic behavior which consists of decoding
written' language units, processing the resulting language counter-
parts through structural and semantic dimensions, and interpreting
the deep structure data relative to an individual's established ob-
jectives.
A central problem, however, in attempting to relate the find-
ings of linguistics and psycholinguistics to the language skill of
reading is ironically that of communication. The many dimensions

1 Reprinted from Psycholinguistics and the Teaching of Reading, Kenneth

S. Goodman and James T. Fleming, editors. Newark, Delaware: International
Reading Association, 1969, 6148.
239
 240 'THEORETICAL MODELS AND PROCESSES OF READING

of these two disciplines have caused the reading specialist and read-
ing researcher to ask how the multitude of individual components
are related to one another and in turn to language skills develop-
ment. The purpose of this discussion is to provide an overview of
selected linguistic and psycholinguistic variables related to decoding
and comprehending language, to briefly examine their psycholog-
ical reality, and in summation to incorporate these variables into a
systems of communication model.
Transformational and semantic theories have proposed that lan-
guage may be viewed on several levels. The -first level is considered
to be the surface structure and encompasses morphemic and synactic
structures which are realized in the form of the graphemic. morpho-
graphemic, phonemic, and morphOphonemic systems. It is at this
level that reading- instruction considers the decoding process. The
second level consists of structural and semantic readings which make
provision for processing language for interpretation. The various
transformational and rewrite rules and the structural reading, as
well as an individual's mental dictionary of semantic readings, are
considered to be incorporated into this level.
The third and least-understood level consists of the deep struc-
ture of language where it is hypothesized that the syntactic and
semantic components of the language are integrated for language
interpretation and stored in memory. This article will initially
examine the decoding process, representing- one dimension of the
surface structure level. Next, the comprehension process encompass-
ing the syntactic dimension of surface structure, the structural and
semantic readings, and the deep structure will be considered. A
minor emphasis will be given to the role of affective mobilizers
and cognitive strategies in language processing. And finally, a sys-
tems of communication model will be presented to summarize the
discussion relative to reading and language skills processing.

The Decoding Process

One of the central tasks of early reading instruction is that of
discovering the nature of the correlation between printed units and
their oral counterparts. Instructional approaches have placed vary-
ing degrees of emphasis on a variety of decoding units. These include
RTJDDELL 241

careful control' of "regularities" and "irregularities" in grapheme-

phoneme correspondences, notably vowels; spelling sound units
which are related to an intermediate level unit known as the mor-
phophoneme; and -a phonologically based unit known as the vocalic-
center group which closely 'appioximates the syllable and in certain
instances the smallest significant meaningful language unit or mor-
pheme.
Most reading programs place some degree of emphasis on these
various units at some point in the program although the exact
structure and sequencing of these units may not always be obvious.
Nevertheless, decoding skills have been taught successfully by plac-
ing special emphasis on one or a combination of these units. Per-
haps a -more scientific statement would -be that children have learned
to decode while being instructed through these various approaches.
The latter statement leaves open the possibility that in some manner
children are independently able to arrive,at an optimal decoding
unit depending upon their own cognitive strategy and the particular
decoding approach used. But the key question at this point asks
what research evidence is available to support a particular perceptual
unit or units leading to decoding skill development in reading in-
struction. Parallel questions not considered in this discussion ask if a
variety of units should be considered in the instructional program,
at what point in a developmental sequence should they be intro-
duced for maximum utilization, and is there any relationship be-
tween specific linguistic units and learner characteristics.
Grapheme-phoneme correspondences. The recommendation
that initial words be introduced on the basis of grouped grapheme-
phoneme consistencies has been proposed by Soffietti (47), Fries (12),
Smith (49), Hall (20), and Bloomfield (4). These individuals have
expressed the opinion that the inconsistencies of the English orthog-
raphy place a limitation on the acquisition of sound-symbol corre-
spondences as- presently developed in widely used reading textbooks.
Although the results have been inconsistent in investigations varying
the degree of emphasis on sound-symbol"correspondences and related
generalizations, some early studies have revealed superior results
for phonic emphasis at early grade levels, particularly in word recog-
nition (3, 26, 48). More recently the work of Hayes (22), Rud-
dell (40), Hahn (19), Tanzer and Alpert (51), Mazurkiewicz (29),
 242 'THEORETICAL MODELS AND PROCESSES OE READING

and Downing (9) have lent support to the value of greater consis-
tency in the introduction of sound-letter correspondences. Addition-
ally, the consistent replication of research findings discussed by
Chall (6) also supports the logical expectation that an approach to
decoding which helps the child grasp the nature of the English
writing code would be of value.
From the standpoint of information transfer the research by
Samuels and Jeffrey (43) emphasizes the value of sound-letter cor-
respondence units. In their research psuedo letters were designed
to- represent English phonemes and kindergarten subjects were
taught to decode on the basis of sound-letter correspondences and
on the basis of "whole words." The findings indicated that subjects
taught by the first- method were more effective in transferring their
skills to "new words" than were those subjects taught by the second
method. The emphasis on individual correspondences appears to
provide a lower error rate and more effective decoding skill than
does attention to word identification based on single featues.
In a later study the same researchers (23) replicated aspects of
the study with similar findings. However, they attributed
their results in part to one 'aspect of the experimental treatment
which taught the subjects to blend phonemes represented by the
psuedo letters into words. These findings are similar- to those of
Silberman (44) in that subjects were unable to transfer correspon-
dence information to new words unless they had received phonic-
blend instruction. The findings may be interpreted to suggest that
sound blending places the phonemes in a natural sound-unit con-
text constituting a =more elaborated decoding unit which is of value
in transferring sound-letter correspondence information to new letter
patterns and words.
Morphographemic-morphophonemic correspondences. If a
decoding program is to account for the nature of the English writ-
ing system, it is necessary to consider spelling units or letter pat-
terns which provide for prediction of sound correspondences beyond
the grapheme-phoneme correspondence level. Venezky (52), Ward-
haugh (53), and Reed (38) have discussed this concern with reference
to the morphophoneme. This unit represents an intermediate unit
between the phoneme and morpheme and may be thought of as a
sound-spelling pattern unit or morphophonemic-morphographemic
RUDDELL 243

system. For example, in considering the words "extreme" and "ex

tremity" one might point to the second e grapheme and note that
there is little regularity in its representation of a given sound. How-
ever, when one considers these two words on the morphophonemic
level, a very regular pattern is immediately obvious. In the altern-
ations extreme-extremity, obscene-obscenity, supreme-supremity, one
observes a consistent shift in the sound value (/iy/ to /i/) with the
addition of the suffix -ity. Although some = reading programs have
developed a few alternationssuch as that found in the final e
marker (sit /i/, site /ay/)very little consideration has been given
to detailed study and research in this area
Venezky has emphasized that a distinction needs to be made
between spelling-sound patterns based on the spelling system and
those based on phonological habits. In the first case children prob-
ably need to be taught the generalization that the letter c represents
/s/ when followed by e, i, or y (e.g., city/sitiy/) and represents /k/
otherwise (e.g.; cat/kwt/). However, the generalization that final
consonant s is pronounced as /z/ following voiced sounds (e.g., dogs
/z/), as /a z/ after /s, z, j/ (e.g., buzzes /a z/) and as /s/ in all
other contexts (e.g., hops /s/) is phonological in nature. For this
reason the native speaker will automatically produce this change and
there would seem to be little need to teach it.
An intensive research effort is needed to examine the contribu-
tion of the value of morphophonemic generalizations for reading
instruction. One basic question might explore the possible advan-
tage of near simultaneous introduction of contrasting letter patterns
representing different but consistent vowel values (e.g., bat, bate),
in contrast to sequencing grapheme-phoneme correspondences on
the basis of "consistent" vowel correspondences (e.g., bat, mat), as
is the case in many recently published reading programs. The re-
search by Levin and Watson (28), although limited in scope, has
demonstrated possible value in establishing a "set for diversity" in
decoding and may be interpreted to lend some support to the former
consideration.
Vocalic-center group, morpheme. Hansen and Rodgers (21)
have posited that a linguistic unit identified as the vocalic-center
group provides for high transfer value in decoding. This unit de-
fined as "a vowel nucleus with 0.3 preceding and 0-4 following con-
 244 THEORETICAL MODELS AND PROCESSES OF READING

sonants" (39) is phonologically rather than semantically based. In

most cases, however, this unit would parallel that of the syllable as
defined by the lexicographer. The rationale for considering such a
unit is that phonological segmentation is of greater significance
than morphological segmentation for the early reader. Rodgers has
reported one experiment in which children were asked to repeat
disyllabic words (e.g., toas -ter) and bimorphemic words (e.g., toast-er)
after the investigator. Their errors were found to favor redivision
-along the syllabic or phonological rather than along morphological
breaks. It should be pointed out, however, that many words classified
along phonological boundaries (e.g., quick-ly) will also be classified
in an identical fashion along morphological boundaries (e.g.,
quickly).
Other research, notably that of Gibson and her colleagues (11),
has explored the presence of a higher. order unit formed by grapheme-
phoneme correspondences. This research has demonstrated that chil-
dren in the early stages of reading- have developed higher-order
generalizations which provide for decoding pseudolinguistic letter
patterns following English spelling expectancies. The children ap-
peared' to perceive regularities in sound-letter correspondences and
transfer these to decoding unfamiliar trigrams even though taught
by what the researchers refer to as the "whole word" approach.
Additional work by Gibson, et al. (13), has demonstrated that
adult subjects perceive pseudolinguistic trigrams more easily when
they follow English spelling generalizations or pronounceable units
(e.g., BIF) than when they are less pronounceable (e.g.; IBF) or
more meaningful (e.g., FBI). Because the task of the reader is that
of decoding written letter patterns and transferring them- into oral
counterparts, pronounceable letter combinations would seem to be
of significant value. On the other hand, meaningful trigrams, such as
FBI, require the reader to work with three units rather than one.
It was noted that the latter type of trigram was more easily recalled
than the pronounceable unit. The ease in recall of the meaningful
unit was attributed to known and exhaustible storage categories
while the pronounceable trigram syllables would call for an ex-
tremely large number of categories and be more difficult to retrieve.
The researchers concluded that pronotinceability was the better
grouping principle for reading. This conclusion lends support to the
 RUDDELL 245

validity of the previously discussed vocalic-tenter group and in cer-

tain instances the corresponding morpheme.
At this point the discussion has considered several decoding
units and their psychological real value for developing decoding
skills. It would appear that the following units are psychologically
real decoding units as used by early readers: grapheme-phoneme
correspondences; morphographemic-morphophonemic patterns; and
vocalic-center groups and in some cases corresponding morphemes.
Upon-initial-examination-the above. units appear to be mutually
exclusive. This condition may not be -so obvious, however, when
operationalized in the instructional program. The great majority of
linguistically influenced programs which attempt to control for
sound-letter correspondences do not teach correspondences in isola-
tion. For the most part, this learning is accomplished through initial
consonant substitution, final consonant- substitution, and vowel sub-
stitution contrasts. For- example, the matrix in- Figure 1 accounts for
emphasis on initial consonants 13/13/ and m/m/ in context; medial
vowels a/m/ and i/i/ in context; mrophophonemic pattern contrast
with a vowel shift from /x/ to /ey/; and utilizes the vocalic-center
groups and corresponding morphemes.

FIGLIRZ 1. Commonalities In Decoding Units

-at -ate -it
b- bat bate bit
m- mat mate mit
This example greatly oversimplifies the discussion but serves to
illustrate the operational economy which is possible in teaching
various decoding generalizations. Most programs, however, attempt
to place specific emphasis on a particular unit of analysis by con-
trolling letter-sound relationships with substitution of correspon-
dences in initial, medial, and final positions.
It is thus possible to view the decoding process as establishing
an understanding of the relationship between grapheme-phoneme
correspondences, which form the larger morphographeme and mor-
phophoneme units which, in operational form, can in turn formulate
the pronunciation of the vocalic-center group and, in some instances,
the corresponding morpheme.
246 THEORETICAL MODELS AND PROCESSES OF READING

The Comprehension Process

In examining the process of comprehension the two general
areas of relational meaning and lexical meaning are of primary
concern. With the foimer, one is concerned with the importance of
structural relationships in sentences; and with the latter, the im-
portance of semantic considerations realized through denotative and
connotative meanings and nonlinguistic signs.
Research related to the comprehension process has been
prompted by the extensive sentence knowledge which the English
speaker possesses. For example, he can recognize grammatical from
nongrammatical sentences, The car struck the tree versus The struck
tree the car; comprehend different sentences having the same con-
stituent structures, John is eager to please versus John is easy to
please; identify ambiguous sentences with identical surface struc-
ture, They are frying- chickens versus They are frying chickens;
and understand sentences with similar meaning but possessing dif-
ferent surface structures, The girl struck the robber versus The
robber was struck by the girl. Additionally, the English- speaker can
comprehend as well as generate a phenomenally large number of
novel sentences. These facts alone suggest that language production
and comprehension must be characterized by a rule governing nature.
But what evidence is present which will provide for the validation
of language generalizations proposed by language scholars?
Relational meaningsurface structure.. Recent psycholinguis-
tic research has sought to explore the psychological reality of surface
structure constituents or the way in which language patterns tend
to "chunk" into syntactic categories. Glanzer (16) has shown that
pseudosyllable-word-pseudosyllable patterns are more easily learned
when the connecting word is a function word (e.g., of, and) than
when it is a content word (e.g., food). This finding supports the view
that the resulting constituent group is a more natural word group
and thus more easily processed.
The work of Johnson (24) dealing with a paired associate learn-
ing task has shown that adult subjects make -a larger -number of
recall errors between phrases (e.g., The valiant canary . . . ate the
mangy cat.) than within phrases (e.g., The . . . valiant . . . canary,
etc.). This finding suggests that phrases may operate as psycholog-
 RUDDELL
247

ically real units. The experiment of Fodor and Bever (11) also sup-
ports this.contention. In their investigation, a clicking noise of brief
duration was made as a sentence was read. Regardless of the place-
ment of the click (e.g., during T word occurring immediately befc
or after a phrase boundary), the subjects indicated that the c'
occurred at the phrase boundary. Thus their conclusion supports tbc
viewpoint that perceptuai units correspond to sentence constituents
as designated by the linguist.
The .recent work of Ammon (1) has revealed that third-grade
and adult subjects require more time to= process and respond to
phrases. Suci et al (50) reported similar findings, thus providing ad-
ditional support for sentence constituents as- Meaningful processing
units.
Relational meaning-- deep structure. The transformational
theory has proposed that sentences are processed from the surface
structure level to an underlying or deep structure for comprehension
purposes. This deep structure is realized through transformational
2
and rewrite rules and is then integrated with the semantic com-
ponent to convey Meaning.
The work of Miller (32) has demonstrated that when subjects
are asked to transform sentences from one form into another (e.g.,
active affirmative to passive or active affirmative to passive negative),
a positive relationship is present between transformation time and
the comAexity of the transformation. This finding supports the con-
tention that transformations possess psychological reality in thit the
greater the number of transformations the greater is the distance
between the surface and deep structure of a sentence.
Mehler (30) has shown that after subjects have been asked to
memorize a series of complex sentences varying in grammatical type,
they tend to recall the sentence but in a simpler grammatical form.
For example, a sentence in the passive may be r_ ecalled in its active
form. These findings suggest that a recoiling of the sentence has
occurred and that the semantic form is maintained but -the deep
syntactic marker indicating the passive form has been forgotten.
The role of transformations in sentence comprehension has also
been demonstrated in the research of Gough (18) and Slobin (46) .
These researchers have shown that sentence comprehension varies
in increasing difficulty (speed in determining truth value of sentence)-
248 THEORETICAL MODELS AND PROCESSES OF READING

in the following orderactive affirmative, passive, negative, and pas-

sive negative. Thus, the available evidence does give suprvAt to the
reality of deep sentence structure. Additional support will be derived
from the discussion of short- and long -term memory presented later
in this p-.per.
Lexical meaning. It should be obvious at this point that this
discussion of surface and deep structure has placed little emphasis on
the role of lexical meaning. Some evidence is present in the pre-
viously discussed work of Gough (18) and Slobin (46) to suggest
the importance of this language component. It is of interest to note,
for example, that passive sentences were comprehended with greater
ease than negative sentences even though the former are thought to
be syntactically more complex.. This unexpected finding may be at-
tributed in part to -the semantic difference between the passive and
the negative and to the semantic similarity b....wteen the passive and
the active. In instances requiring a true or false determination, nega-
tive sentences seem to be difficult to comprehend. Slobin has- empha-
sized that not only is syntax important in- comprehending sentences
but semantic considerations must also be accounted for. His research
has shown that the differentiation- in difficulty between active and
passive can largely be eliminated by clarifying the role of nouns in
the subject and object positions. This clarification can be accom-
plished by reducing the possibility of semantic reversibility( e.g.,
Reversible: The girl struck the boy. The boy was-struck by the girl.
Nonreversible: The boy picked the apple. The apple was picked by
the boy.) Such findings suggest that much more is involved in sen-
tence understanding than relational meaning.
One would expect structure words o play an important role
in narrowing possible semantic alternatives in the sequence of a
sentence context. For example, the word the not only cues a noun
which follows but may also clarify or emphasize the semantic nature
of the noun (e.g., The dog was in our yard versus Some dog was in
our yard). Miller (32) and Miller et al (34), demonstrated that words
in context following a -similar grammatical pattern are perceived
more accurately than when in isolation. These findings suggest that
the contextual constraint serves to narrow the possible range of ap-
propriate words. Additional support for the importance of context
in narrowing semantic possibilities is found in the research of Good-

.0
RUDDELL 249

man (17). He has shown that although children may be unable to

decode words in isolation, they deal successfully with the same words
in a running context. Research by Ruddell (41) has shown that
reading comprehension of fourth grade children is significantly
higher on passages utilizing basic high-frequency patterns of their
oral language structure in contrast to passages using low-frequency
and more-elaborated construction. These findings may be interpreted
to support the importance of contextual associations which provide
sufficient delimiting information to enable a. child to determine the
semantic role of a word and further to recognize and comprehend it
in the sentence.
The importance of the connotative dimension of word mean-
ing obtains support from the research by Samuels (42). Fifth and
sixth grade subjects were- found to perform significantly higher in
comprehending a reading passage containing. words of high asso-
ciation value than a control group reading a passage containing low
association value words.
Although effort is being made in developing a semantic theory
which parallels the- previously discussed deep structure, progresi has
been understandably slow because of the extremely complex nature
of relating the semantic and structural components. Katz and Fodor
(25) have characterized the form of a semantic theory as linguistic
description minus grammar. They have-postulated that a semantic
component in language serves =to assign meaning to each sentence
through semantic markers. For example, semanticists have con-
structed semantic categories such as objectnonobject, animatein-
animate, humannonhuman, and- male female. A semantic marker
(37) such as male represents the content of words like -man, boy, or
father in contrast to words like car, truth, and girl. In some respects
this approach resembles the game of 20 questions which provides for
narrowing the definition of the meaning under consideration. A
sequence of such semantic markers constituting the dictionafy would
thus provide a semantic reading and define-theconceptual content
of words. By then utilizing a set of projection rules, the readings of
lexical items would be integrated with the grammatical relationships
as indicated by the deep structure to derive the semantic charac-
terizations of sentence constituents. Postal (37) expresses the view
that such characterizations will explain semantic properties such as
 -0 ./ inn.1141A.Y. fOra4-
tence context. To use an example from Katz~
atz and Fodor, the sentence
"The bill is large," can mean a sizeable debt or the unusual size of .a
bird's beak. To know that the sentence is ambiguous is only a first
step toward the understanding of its meaning. The meaning diffi-
culty is resolved .in a larger verbal context such as "Oh, I see you
bought a new dress," or "My, what an unusual bird." The ambiguity
may also be accounted for in a nonlinguistic fashion if one is pur-
chasing clothing at a store or visiting a zoo. To describe rules, how-
ever, which will define the larger verbal context and nonlinguistic
meaning represents an enormous task for the psycholinguistic
researcher.
Short- and long-term memory. The importance of memory in
language processing is also of significant concern as one considers
surface and deep structure. Miller and Chomsky (35) have proposed
that a short- and long-term memory are operative in language proc,:s-
sing. It is further proposed that the limited short-term memory deals
with the less complex surface structure of sentences while the long-
term memory handles the more involved, deep structure of sentences.
Miller (32) has demonstrated that subjects have great difficulty
in processing sentences containing self-embedded structures (e.g.,
The rat that the cat that the dog worried killed ate the malt, etc.) in
contrast to right-recursive sentences '(e.g.,This is the dog that worried
the cat that killed the rat. that ate the malt, etc.) Because the deep
structure of these sentences is identical, Miller attributes this vari-
ation in difficulty to the heavy demand placed on the short-term
memory by the surface structure of the self-embedded sentence.
It would thus seem logical that because of the limited short-
term memory (33), a deep language structure and a long-term mem-
ory component are essential for information processing over running
discourse. The previously cited work of Mehler may be interpreted
to support this viewpoint in that complex sentences presented to
subjects were recalled in a simpler form. It would thus seem that
after a sentence is processed in the deep structure, the underlying

264 THEORETICAL MODELS AND PROCESSES OF READING

.show similarities and differences, but all point t, a selective, tenta-

tive, anticipatory process quite unlike the process of precise, se-
quential identification commonly assumed.
Let's take a closer look now at the components the reader
manipulates in this psycholinguistic guessing game.
At any point in time, of course, the reader has available to him
and brings to his reading the sum total' of his experience and his
language and thought development. This self-evident fact needs to
be stated because what appears to be intuitive in any guessing is
actually the result of knowledge so well learned that the process of its
application requires little conscious effort. Most language use has
reached this automatic, intuitive level. Most of us are quite unable
to describe the use we make of grammar in encoding and decoding
speech, yet all language users demonstrate a high degree of skill
and mastery over the syntax of language even in our humblest and
most informal uses of speech.
Chomsky (3) has suggested this model of sentence production by
speakers of a language:

Semantic Deep Surface

transformations phonological
RUDDELL 251

meaning is retained with little regard for the structure. It would also
appear that the deep underlying structure is basic to comprehending
sentences.

Affective and Cognitive Dimensions

Affective mobilizers. A system of communication must in some
manner account for an individual's interests, attitudes, and values
which become operationalized as his objectives and goals. As an
individual confronts a verbal task, his motivation reflected in his
persistence and drive is extremely important. This viewpoint is sup-
ported by Durkin's research (8) which has identified the preschool
reader as an individual who is serious and persistent, possesses the
ability to concentrate, and is of a curious nature. In a study of
achieving and nonachieving elementary school readers, Kress (27)
has reported that the former group demonstrated more initiative in
exhausting solutions and was found to persist in probleth solving
under changing conditions. The research of Piekarz (36) has indi-
cated that the high-level reader, in contrast to the inferior reader,
provides significantly more responses in interpreting a reading pas-
sage, a trait thus indicating greater involvement and participation.
The- high -level reader is also more objective and impersonal in syn-
thesizing information sought. The research of Athey (2) has demon-
strated .the importance of value systems as mobilizers for reading at
the junior high school level.
One would also expect life objectives to influence an individual's
store of concepts and in turn his semantic dimension of language
processing. The reality of this view is reflected in functional varieties
of language. The lexicon of the organic chemist varies markedly
from that of the newscaster, and both in turn differ from that of the
farm laborer.
The affective mobilizers operationalized as the objectives and
goals of the individual would thus be expected to influence language
processing at the surface-structure level, through the structural and
semantic readings, and at the deep-structure zvel.
Cognitive strategies. As an individual participates in the- com-
munication act, he is constantly required to perceive and organize
experiences. He must develop a symbol-processing system which will
252 THEORETICAL MODELS AND PROCESSES OF READING

provide for conceptualization of experience. Bruner et al. (5) have

shown that a cognitive strategy is of basic importance -to the con-
ceptualization process. Kress (27) concluded from a concept forma-
tion study of elementary school children that achieving readers were
superior to nonachievers in versatility and flexibility, ability to draw
inferences from relevant clues, and ability to shift set when new
standards were introduced. From an extensive review of research on
conceptualization, Singer (45) concluded that an important dimen-
sion in comprehending language consists of changing, modifying,
and reorganizing a previously formed concept.
Thus the cognitive strategy of an individual is considered to be
operationalized as a process of evaluating the adequacy of informa-
tion, data gathering, hypothesis building, organizing and synthesiz-
ing data, and hypothesis testing. Additionally, the utilization of these
factors must be guided by a constant awareness of the need to ,shift
one's strategy to account for other approaches to problem solutions.

A Systems of Communication Model

Although care must be exercised in attributing psychological
reality to a competence model or linguistic description, the -research
presented in this discussion lends some *degree of support to the
reality of surface structure, language processing through structural
and semantic readings, deep structure, short- and long-term memory,
and the importance of affective mobilizers and cognitive strategies.
Thus, in a limited degree, evidence is present to suggest the general
nature of a model of performance.
The language model in Figure 2 has been formulated in order to
integrate the previous discussion and express relationships between
the various psycholinguistic factors involved in the communi-
cation process. On the extreme left of the model the basic com-
munication skills are identified. The rectangular line encloses a
hypothetical representation of the systems involved in encoding and
production processes of speech or writing and the decoding and com-
prehension processes of listening or reading.
Near the bottom of the rectatigle," the affective mobilizers and
the cognitive strategies are noted. The mobilizers represent indi-
RUDDELL 253

SYSTEMS OF COMMUNICATION MODEL

SURFACE STRUCTURE STRUCTURAL AND I
SEMANTIC READINGS I DEEP STRUCTURE

Contest feedback for verifying

Morphophonemic 4.( 4 4 decoding information
Spout'
Phonemic
T,/ Feedback for
USfening , evaluating rimming
rit discourse
Denotative Z
meaning i/4<long term
iL , Structural Memory >
Connotative I
/leonine
Short
Term
Memory 7
NoLinguistis
Mooning
Syntactic

Tronsfaimationot
and
Short Rewrite rules
Term
flooSng Memory N t...2_,,,. .
Semantic Memory
Grophemie
Contest feedback for verifying
syntactic information Feedback for
Writing Ir 4, O evaluating running
IMorpnographernic w. discourse

Lk..\`.

key Primary Communication Line swwwswmisisi

SMendory Conitunic0fron Una -----
Feedback Line 4
Pict= 2. Systems of Communication Model.

vidual interests, attitudes, and' values and become operationalized as

goals and objectives in the communication setting. The strategies
represent' an individual's approach to the language-processing task
as determined in part by his objectives. The vertical arrows are
entered at key parts of the model from the affective and cognitive
dimensions and indicate that during language processing the reader
is constantly interacting with each phase of the communication
model. These aspects of the model would enable the individual to
shift his attention, for example, to the structural aspects of a sentence
in order to obtain added relational data to determine the specific
semantic dimension of a given word.
The model becomes more meaningful as the act of reading is
 254 THEORETICAL MODELS AND PROCESSES OF READING

traced through the various dimensions. In the early stages of reading,

the child, encounters the English writing system. The objective of
the instructional program is to help him understand the nature of
the code. This objective may be accomplished by establishing the
relationship between the graphemic and morphographemic systems
and the phonemic and morphophonemic systems, respectively. The
development of cognitive strategies. should lead the child to examine
alternatives to decoding words. For example, after an unsuccessful -
attempt to utilize the grapheme-phoneme relationships in decoding,
it may be necessary to examine clues at the morphographemic-morpho-
phonemic level or w utilize the context feed-back from 'the inter-
pretation derived froth the deep structure of the sentence. The more
advanced reader may move directly from the morphographemic level
through the morphophonemic system to the morphemic level= or
directly from the morphographemic level to the morpheme with a
minimal use of the morphophonemic system (10).
The organiied sound patterns- directly involve the morphemic
system. At this point the short-term memory is effected, and the
syntactic system begins- to "chunk" the language units through the
constituent structures for transformational and rewriting purpos.
Following the transformation and rewriting of the sentence in its
most basic form, the semantic aspect of the model is encountered,
and the meaning of the various morphemes are considered through
a semantic reading utilizing the denotative, connotative, and non-
linguistic dictionary components. The semantic and structural mean-
ings are then meshed through the semantic interpretation or projection
rules, and the meaning is established. Simultaneously, the appropri-
ate semantic markers and structural markers are attached, and the
semantic and structural contexts are placed in long-term memory.
If at the point of semantic projection some difficulty is encountered
and the sentence appeais to be ambiguous, the reader may return to
the morphophonemic level or the syntactic level to verify the surface
structure realization for a specific morpheme or constituent struc-
ture, respectively.
As a new sentence appears in the running context, the com-
munication is processed in the same fashion. Previous informathm,
which has been stored in the long-term semantic and structural
memories, is available for mobilization to the semantic-interpreta7
RUDDELL 255

tion level to aid in evaluating the running discourse relative to the

objectives established by the reader.
Just as one has examined the act of reading which involves the
factors inside the rectangular line, one could proceed through the
model in a similar fashion for the decoding and comprehension
processes of listening. The encoding and productive processes of writ-
ing and speech can be examined by 'starting with the long-term
semantic and structural memories, progressing to the semantic inter-
pretation or projection, and moving from right- to left through the
model.
In conclusion, the field of psyc.holinguistics holds significant
promise in developing an understanding of language processing
which in turn should generate valuable knowledge about the read-
ing process and other communication skills. This result is suggested
in the overview of research presented in this discussion providing
support for the psychological reality of selected linguistic units.
Tormulations such as those realized in the systems of communication
model should enable reading researchers and-specialists to work more
profitably toward a theory of reading.

REFERENCES

1. Ammon, P. R. "The Perception of Grammatical Relations in Sentences:

A Methodological Exploration," Journal of Verbal Learning and Verbal
Behavior (in press).
2. Athey, Irene J. "Reading-Personality Patterns at the Junior High School
Ltvel," unpublished doctoral disiertation, University of California at
Berkeley, 1965.
S. Bear, D. E. "Phonics for First Grade: A Comparison of Two Methods,"
Elementary School Journal, 59 (1958), 394-402.
4. Bloomfield, L. Language. New York: Holt, Rinehart and Winston, 1933.
5. Bruner, J., J. Goodnow, and G. Austin. A Study of Thinking. New York:
Wiley Press, 1956.
6. Chall, Jeanne S. Learning to Read: The Great Debate. New York: McGraw-
Hill, 1967.
7. Clifton, C., I. Kurca, and J. J. Jenkins. "Grammatical Relations as Deter-
minants of Sentence Similarity," Journal of Verbal Learning and Verbal
Behavior, 4 (1965), 112-117.
8. Durkin, Dolores. "Children Who Learn to Read Before One," Reading
f Teacher, 14 (1961), 163.166.
 256 THEORETICAL MODELS AND PROCESSES OF READING

9. Downing, J. A. "The t.T.A. (Initial Teaching Alphabet) Reading Experi-

ment," Reading Teacher, 18 (November 1965). 105-110.
10. Edfeldt, A. W. Silent Speech and Silent Reading. Stockholm: Almgrist and
Wiksell, 1959.
11. Fodor, J. A., and T. G. Bever. 'The Psychological Reality of Linguistic
Segments," Journal of Verbal Learning and Verbal Behavior; 4 (1965),
. 414-420.
12. Fries, C. C. Linguistics and Reading. New York: Holt, Rinehart and
Winston, 1963.
13. Gibson, Eleanor J., et at "A Comparison of Meaningfulness and Pronounce-
ability as Grouping Principles in the Perception and Retention of Verbal
Material," A Basic Research Program on Reading. Ithaca, Ncw York: Cornell
University, Cooperative 'Research Project No. 639, 1963.
14. Gibson, Eleanor, J., H. Osser. and Anne D. Pick. "A Study of the Develop-
ment Grapheme- Phoneme Correspondences," A Basic Research Program
on Reading. Ithaca, New York: Cornell University, Cooperative Research
Project No. 639
15. Gibson, Eleanor J., et al. "The Role of Grapheme-Phoneme Correspon-
dences in the Perception of Words," American Journal of Psychology, 75
(1962), 554-570.
16. Glanzer, M. "Grammatical Category: A Rote Learning and Word Associa-
tion Analysis," Journal of Verbal Learning and Verbal Behavior, I (1962),
31-41.
17. Goodman, K. S. "A Linguistic Study of Cues and Miscues in Reading,"
Elementary English, 42 (19.'..5), 639-643.
18. Gough, P. G. "Grammatical Transformations and Speed of Understanding,"
Journal of Verbal Learning and Verbal Behavior, 4 (1965), 107.111.
19:-Hahn, H. T. "Three Approaches to Beginning Reading Instruction,"
Reading Teacher, -19 (May 1966), 590-594.
20. Hall, R. A., Jr. Sound and Spelling in English. Chilton Books, 1961.
21. Hansen, D. N., and T. S. Rodgers. "Exploration of Psycholinguistic Units
in Initial Reading," in Goodman (Ed.), The Psycholinguistic Nature of the
Reading Process. Detroit: Wayne Stale University Press, 1968, 59-102.
22. Hayes, R. B. "ITA and Three Other Approaches to Reading in First Grade,"
Reading Teacher, 19 (May 1966), 627.630. -I
23. Jeffrey, W. E., and S. J. Samuels. "Effect of Method of Reading Training
on Initial Learning and Transfer," Journal of Verbal Learning and Verbal
Behavior, No. 3, 7 (1967), 354-358.
24. Johnson, -N. F. "The Psychological Reality of PhraseStructtire Rules," Jour-
nal of Verbal Learning and Verbal Behavior, 4 (1965), 414-420.
25. Katz, J. J., and J. A. Fodor. "The Structure of a Semantic Theory," Lan-
guage, 39, No. 2 (1963), 170-210.
26. Kelly, Barbara Cline. 'The Economy Method Versus the -Scott, Foresman
Method in Teaching Second Grade Reading in the Murphysboro Public
Schools," Journal of Educational Research, 51 (1958), 465469.
RUDDELL 257

27. Kress, R. A. "An Investigation of the Relationship Between Concept Forma-

tion and Achievement- in Reading," unpublished doctoral dissertation,
Temple University, 1955.
28. Levine, H., and J. Watson. "The Learning Variable Grapheme-to-Phoneme
Correspondences," A Basic Research Program on Reading. Ithaca, New York:
Cornell University, Cooperative Research Project No. 639, 1963.
29. Mazurkiewicz, A. J. "ITA and TO Reading Achievement Whta Method-
ology Is Controlled," Reading Teacher, 19 (May 1966), 606-610.
30. Mehler, J. "Some Effects of Grammatical Transformation on the Recall of
English Sentences," Journal of Verbal Learning and Verbal Behavior, 2
(1963), 346-351.
31. Miller, G. A. "Some Preliminaries to Psycholinguistics," American Psychol-
ogist, 20 (1965), 15.20.-
32. Miller, G. A. "Some Psychological Studies of Grammar," American Psychol-
ogist, 17 (1962), 748.762.
33. Miller, G. A. "The Magical Number Seven, Plus or Minus Two," Psycho-
logical Review, 63 (1956), 81-97.
34. Miller, G. A., G. A. Heise, and W. Lichten. "The Intelligibility of Speech
as a Function of the Context of the Test Material," Journal of Experimental
Psychology, 41 (1951), 329-335.
35. Miller, G. A., and N. Chomsky. "Finitary Models of Language Users," in
R. D. Luce, R. Bush and E. Galanter (Eds.), Handbook of Mathematical
Psychology, 11. New York: Wiley Press, 1963.
36. Piekarz, Josephine A. "Getting Meaning From Reading," Elementary School
Journal, 56 (1956), 303 -309..
37. Postal, P. M. "Underlying and Superficial Linguistic Structure," Harvard
Educational Review, 34 (1964),_246.266.
38. Reed, D. W. "A Theory of Language, Speech, and Writing," Linguistics
and Reading, Highlights of the Pre-convention Institutes. Newark, Dela-
ware: International Reading Association, 1966; 4.25.
39. Rodgers, T. S. "Linguistic Considerations in the Design of the Stanford
Computed Based Curriculum in Initial Reading?' Institute for Mathemat-
ical Studies in the Social Sciences, Technical Report No. Ill, usoz Grant
Number 0E5-10-050, 1967.
40. Ruddell, R. B. "A Longitudinal Study of Four Programs of Reading Instruc-
tion Varying in Emphasis on Regularity of Grapheme-Phoneme Correspon-
dences and Language Structure on Reading Achievement in Grades Two
and Three." University of California, Berkeley, 1968. Supported by the
tisoz, Projects 3099 and 78085.
41. Ruddell, R. B. "The Effect of the Similarity of, Oral and Written Patterns
of Language Structure on Reading Comprehension," Elementary English.
42 (1965), 403-410.
42. Samuels, D. J. "Effect of Word Associations on Reading Speed, Recall, and
Guessing Behavior on Tests," Journal of Educational Psychology (in press).
43. Samuels, S. J., and W. E. Jeffrey. "Discriminability of Words and Letter
258 THEORETICAL MODELS AND PROCESSES OF READING

Cues Used in Learning to Read," Journal of Educational'Psychology, 57,

No.- 6. (1966), 337.340.
44. Silberman, H. F. Exploratory Research on a Beginning Reading Program,
Technical Memorandum, No TM895/100/00. Santa Monica: Systein Devel-
o t Corporation, 1964.
45. Singer, H. "Conceptualization in Learning to Read: New Frontiers in
Co Um...Id:At Reading," in G. B. Schick and M. M. May (Eds.), Fifteenth
Yearbook of the National Reading Conference. National Reading Confer-
ence, Milwaukee, 1966, 116-132.
46. Slobin, D. I. "Grammatical Transformations and Sentence Comprehension
in Childhood and Adulthood," Journal of Verbal Learning and Verbal
Behavior, 5 (1966), 219-227.
47. Soffietti, J. P. "Why Children Fail to Read: A Linguistic Analysis," Harvard
EducationalReview, 25 (Spring 1955), 63-84.
48. Sparks, Paul E., and L. C. Fay. "An- Evaluation of Two Methods of Teaching
Reading," Ekmentaty School Journal, 57 (April 1957),-385-390.
49. Smith, H. L, Jr. Linguistic Science and the Teaching of English. Cambridge:
Harvard University Press, 1959.
50. Suci, G. J., P. R. Ammon, and P. Gamlin "The Validity of the Probe-
Latency Technique for Assessing Structure in Language," Language and
Speech, 10 (1967), 69-80.
51. Tanyzer, H. J., and H. Alpert. "Three Different Basal- Reading Systems
and First Grade Reading Achievement," Reading Teacher, 19 (May 1966),
636-642.
52. Venezky, R. F. "English Orthography: Its Graphical Structure and Its Rela-
tion to Sound," Reading Research Quarterly, 2 (Spring 1967), 75-106.
53. Wardaugh, R. "LinguisticsReading Dialogue," Reading Teacher, 21
(February 1968), 432441.
%
Reading: A Psycho linguistic Guessing Garnel
KENNETH S. GOODMAN
Wayne State University

AS SCIENTIFIC-understanding develops in any field cit study, preexist-

ing, naive, common sense notions must give way. Such outmoded
beliefs clutter the literature dealing with the process of reading.
They interfere with the application of modern scientific concepts of
language and thought to research -- in reading. They confuse the
attempts at application of such concepts to solution of problems in-
volved in the teaching-and learning of reading. The very fact that
such naive beliefs are based on- common -sense explains their persis-
tent and recurrent nature. To the casual and unsophisticated ob-
server they appear to explain, even predict, a set of phenomena in
reading. This paper will deal with one such key misconception and
offer a more viable scientific alternative.
Simply stated, the common sense notion I seek here to refute
is this:
"Reading is a precise process. It involves exact, detailed, se-
quential perception and identification of letters, words, spelling pat-
terns and large language units."
In phonic centered approaches to reading, the preoccupation is
with precise letter identification. In word centered approaches, the
focus is on word identifications. Known words are sight words, pre-
cisely named in any setting..
This is not to say that those who have worked diligently in the
field of reading are not aware that reading is more than precise, se-
quential identification. But, the common -sense notion, though not
adequate, continues to permeate thinking about reading.
Spache (8) presents a word version of this common sense 'View:
"Thus, in its simplest form, reading may be considered a series of
word perceptions."
1 Paper read at the American Educational Research Association, New York,
February 1967, and published in the Journal of the Reading Specialist, May 1967.
Reprinted with permission of the author and publisher.
"PERMISSION TO REPRODUCE THIS COPY-
259 RIGHTED MATERIAL HAS BEEN GRANTED
BY
College Reading
Association
TO ERIC AND ORGANIZATIONS OPERATING
UNDER AGREEMENTS WITH THE US OFFICE
OF EDUCATION FURTHER REPRODUCTION
OUTSIDE THE ERIC SYSTEM REOUIRES PER
MISSION OF THE COPYRIGHT OWNER"
 260 THEORETICAL MODELS AND PROCESSES OF READING

The teacher's manual of the Lippincott Basic Reading (6) incor-

porates a letter by letter variant in the justification of its reading ap-
proach: "In short, following this program the child learns from the
beginning to see words exactly as the most skillful readers see them
. . . as whole images of complete words with all their letters."
In place of this misconception, I- offer this: Reading is a selec-
tive process. It involves partial use of available minimal language
cues selected from perceptual input on the basis of the reader's
expectation. As this partial information is processed, tentative deci-
sions are made to be confirmed, rejected, or refined as reading pro._
gresses.
More simply stated, reading-is a psycholinguistic guessing game.
It involves aii interaction between thought and language. Efficient
reading does not result from precise perception and identification of
all elements, but from skill in selecting the fewest, most productive
cues necessary to produce guesses which are right the.fiist time. The
ability to anticipate that which has not been seen, of course, is vital
in reading, just as the ability to anticipate what has not yet been
heard is vital in listening.
Consider this actual sample of a relatively proficient child read-
ing orally. The reader is a fourth grade child reading the opening
paragraphs of a story from a sixth grade basal reader (5).
"If it bothers you to think of it as baby sitting," my father said,
"then don't think of it as baby sitting. Think of it as homework.Tart
of your education. You just happen to do your studying in the room
where the baby brother is sleeping, that's all." He helped my mother
with her coat, and then they were gone.

hoped,4 a
So education it was t.I opaatits dictionary and picl:ed out a
a PH. He
word that sounded good. "p111/1INZpl_sAlYn/ 4- yelled. Might
what it means 1. Phizo 2. Phiso /soophical
as well study wend mat: -first. "Philesepkiesi: showing calmness
his 1. fort 2. future S. futshion
and courage in the face of ill fortune." I mean I really yelled it. I
guess a fellow has to work off steam once in a while..
4

GOODMAN 261

He has not seen the story before. It- is, by intention, slightly
difficult for him: The insights into his reading process come primiaily
from his errors, which I choose to call miscues in order to avoid value
implications. His expected responses mask the process- of their at-
tainment, but his unexpected responses have been achieved through
the same process, albeit less successfully applied.- The-ways that they
deviate from the expected reveal this proCess.
in the common sense view that I -am rejecting, all deviations
must -be_ treated_aserrois. Furthermore, it _must be assumed= in this=
=

view that an error either indicatet- that:the- reader does not kilo*
something or that he hat-been- "careless" in ihe application of hit
knowledge.
-For_ekample; his substitution Of:the:for your in the first para-
graph of the sample_ must mean =that -he ,_careless, since he has
already read your_anct the-cotrectly-rin---theAiery:same sentence. The
implication_ is=that we must _teach _him= toibemore_ Careful,- that_ is to
be more rreite in idetnifyitig each-word.orietter.
But now feet-take the view that I-haVei-uggested. What sort of
info..matitin :could haVe led to tentatively_ deciding onike in this
situation and not or refiningithis decision?_ There obviously
is no graphic relationship -between_your and= the.lt-may be =of course,
that he picked =up the in the fif-lis visual field; But; there
is ar important non-graphic relationship -between_ the and -your.
They both haVe thesame grammatical fUnctidn..: they are-, in:my term-
inology, noun marker-s._Bither the reader anticipateda noun marker
and supplied one paying no attention to graphic information or he
Used your as a-igrammatical Signal ignoring its graphic shape. Since
tite tentative choice the disturbs neither= the meaning nor the gram-.
mar of the-passage, there_ is-no_reasoni to -reject and-correct it. This
explanation appears to confirmed "by two similar miscues =in the
next paragraph. A and his are-both substituted for the Neither are
corrected. Though the SubstitutiOn-of,:hisichanget the meaning, the
peeuliar idiom used in this dictionary definition; "in the face of ill
fortune apparently has-little meaning to- this-_feader anyway.
The conchisitin this = time is that-he is -using _noun :Markers_ for
grammatical, as well as:graphic, information in reaching- his tenta-
tive conclusions. All together in reading this ten:page story, he made
twenty noun Marker substitution& six omissions and two insertions.
 .1p

262 THEORETICAL MODELS .AND PROCESSES OF READING

He corrected four of his substitutions and one omission. Similar mis-

cues involved other function words (auxiliary verbs and prepositions,
for example). These miscues appear to have little effect on the mean-
ing of what he is reading. In spite of their frequency, their elimina-
Lion would- not substantially improve= the
on more precise identification of each word might cause this reader
to stop seeking grammatical information and use only graphic
information.
The substitution of hoped for opened could again be regarded
as careless or imprecise identification of letters. But, if we dig be-
yGnd this common sense explanation, we find 1)= both are verbs and
2) = the words= have key graphic similarities. Further, there may be
evidence of the xeader's bilingual French-Canadian background here,
as= there is in subsequent miscues (harms for arms, shuckled for
chuckled, shoose for choose, shair for chair). The correction of this
miscue may involve an immediate rejection, of the tentative choice
made on e.e basis of a= review =of the graphic stimulus, or it may
result =from recognizing that it cannot lead to the rest of the sentence,
"I hoped a dictionary .. ." does not make sense. (It isn't decodable).
In any case, the reader has demonstrated the process by which he
constantly tests his guesses, or tentative choices,-if yoti prefer.
Sounds is substituted for sounded, but the two differ in ending
only Common sense might lead to the conclusion that the child does
not pay attention to word endings, slurs the ends or is otherwise care-
less. But, there is no consistent similar occurrence in other word end-
ings. Actually, the child has substituted one inflectional ending for
another. In doing so he has revealed 1) his ability to separate base
and inflectional suffix, and 2) his use of inflectional endingi as gram-
matical signals or markers. Again he has not corrected a- miscue that
is both grammatically and semantically acceptable.
He for I is a pronoun for pronoun substitution that results in a
meaning change, though the= antecedent is a. bit vague, and the in-
consistency of meaning is not easily apparent.
When we examine what the reader did with the sentence "Might
as well study word meanings first," we see how poorly the model of
precise sequential identification fits the reading process. Essentially
this reader has decoded graphic input for meaning and then encoded

a,
 GOODMAN 263

meaning in oral output with transformed grammar and changed

,vocabuldiy, but with the basic meaning retained. Perhaps as he en-
coded his output, he was already working at the list word which
followed, but the tentative cho...e was good enough and was no_ t
corrected.
There are two examples, in this sample, of the reader working
at unknown words. He reveals a fair picture of his strategies and
abilities in these miscues, though in neither is he successful. In his
several attempts at philosphical, his first attempt comes closest In-
cidentally, he reveals here that, he can use a phonic letter-sound
strategy when he wants to. In subsequent attempts he moves away
from this sounding out, trying other possibilities, as if trying to find
something which at least will sound familiar. Interestingly, here he
has a definition of sorts, but no context to work with. Philosophical
occurs as a list word a number of times in the story. In subsequent
attempts, the child tried physica, physicacol, physical, philosovigul,
phizzlesovigul, phizzo sorigul, philazophgul. He appears to move. in
concentric circles around- the phonic information he has, trying
deviations and variations. His three unsuccessful attempts at fortune
illustrate this same= process. Both- words are apparently unknown to
the reader. He can never really identify a word he has not heard.
In such cases, unless the context or contexts sufficiently delimit the
LW" word's meaning, -the reader is not able to get meaning from the
words. In some instances, of course, the reader may form a fairly
accurate definition of the word, even if he never recognizes it (that
is matches it with a known oral equivalent) or pronounces it cor-
rectly. This reader achieved that with the word typical which occurred
many times in the story. Throughout his reading .he said topical.
When he finished reading, a check of his comprehension indicated
that he knew quite well the meaning of the word. This phenomenon
is familiar to any adult reader. Each- of us has many well-defined
words in our reading vocabulary which we either mispronounce or
do not use orally. - *

I've used the example of this youngster's oral reading not be-
cause what he's done is typical of all readers or even of readers his
age, but because his miscues suggest how he carries out the psycho-
linguistic guessing game in reading. The miscues of other readers

G.
GOODMAN 265

and any matching or coded signal which results is a kind of by-pro-

duct.
In oral reading, the reader must perform two tasks at the same
time. He must produce an oral language equivalent of the graphic
input which is the signal in reading, and he must also reconstruct the
meaning of what he, is reading: The matching in Chomsky's inter-
pretation model is largely what I prefer to call a recoding operation.
The reader recodes the coded graphic input as phonological or oral
output Meaning is not normally involved.to any extent. This recod-
ing can even be learned by someone who doesn't speak the language
at all, for example, the bar - mitzvah =boy may learn to recode Hebrew
script as chanted oral Hebrew with no ability to understand what he
is chanting; but when the reader engages in semantic analysis to re-
construct the meaning of the writer, only then is he decoding.
In oral reading there are three logical possible arrangements of
these two operations. The reacter_may recode graphic input as oral
language and then decode it. He may recode and decode simultane-
ously. Or,--he may decode first and = then encode the meaning as oral
output.
On the basis of my research to date, it- appears that readers who
have achieved some degree of proficiency decode 'directly from the
graphic stimulus in a process similar to Chornsky's sampling model
and then encode from the deep structure, as illustrated in Chomsky's
model of sentence production. Their oral output is not directly re-
lated to the graphic stimulus and may- involve transfortnation in
vocabulary and syntax, even if meaning is retained. If their-compre-
hension is inaccurate, they will encode this changed or incomplete
meaning as oral output.
The common misconception is that graphic input is precisely
and sequentially recoded as phonological input and then decoded bit
by bit. Meaning is cumulative, built up a - piece- at a =time in this
view, This view appears to be supported by studies of visual percep-
tion that indicate that only a very narrow sp in of print on either side
of the point of fixation is in sharp focus at any time. We might dub
this the "end of the nose" view, since it assurstes that input in reading
is that which lies in sharp focus in a straight line from the end of the
nose. Speed and efficiency are assumed to come from widening the
span taken in on either side of the nose, moving the nose more

fp.
 266 THEORETICAL MODELS AND PROCESSES OF READING

rapidly or avoiding backward movements of the eyes and

nose, which
of course must cut down on efficiency.
This view cannot possibly explain the speed with which the
average adult reads, or a myriad of other constantly occurring phe-
nomena in reading. How can it explain, for example, a highly pro-
ficient adult reader reading and rereading a paper he's written and
always missing the same misprints. Or how can it explain our fourth
grader seeing "Study word meanings first" and saying, "Study what it
means"?
No, the "end of= the nose" view of reading will not work. The
reader is not confined to iriforniation hereceives from a half inch of
print in clear focus. StUdies, in fact, indicate that children with severe
visual handicaps are able to learn to_ read as well as normal "children.
Readers utilize not one, but three kinds of information siMulta-
neously. Certainly Without graphic input there would be no reading.
But, the reader uses Syntactic and semantic information as well He
predicts and anticipatti On the basis of this information, sampling
from the print just enough_ to confirm his- &els of what's coining, to
cue more semantic and syntactic information. RedundanCy and se-
quential constraints in:language, which -the reader reacts to, make
this prediction possible. EVen the blurred -and shadowy images he
picks up in the peripheral area of his visual field May help to trigger
or confirm guesses.
Skill in reading involves not greater precision, but more ac-_
curate first guesses based 4 4 better sampling techniques, greater
control over language structure, broadened experiences and increased
. conceptual development Al the child develops reading skill and
speed, he uses increasingly fewer graphic cues. Silent reading can ,r
then become a more rapid and efficient process than oral reading, for
two reasons: 1) the reader's attention is not between decod-
ing and recoding or encoding as oral output, and 2) his speed is
not restricted to the speed of speech production. Reading becomes a
more efficient and rapid process than listening, in fact, since listen-
ing is normally limited to the speed of the speaker.
Recent studies with speeded up electronic recordings where dis-
tortion of pitch is avoided have demonstrated that listening can be
made more rapid without impairing comprehension too.
Though the beginning reader obtiously needs more graphic
GOODMAN 267

information in decoding and, therefore, needs to be more precise

than skilled readers, evidence from a study of first graders by Good-
man (4) indicates that they begin to sample and draw on syntactic
and semantic information almost from the beginning, if they are
reading material lw if& formedlanguage.
Here are excerpts from two primer stories (1, 2) as they were
read by a first grade child at the same session. Ostensibly (and by
intent of the authors) the first, from a second preprimer, should be
much easier 'than the second, from a third preprimer. Yet she en-
countered problems to the point of total confusion with the first and
was able to handle exactly the same elements in the second.
Note, for example, the confusion of come and here in "Ride In."
This represents a habitual association in evidence in early reading
of this child. Both come and here as graphic shapes are likely to be
identified as come or here. In "Stop and Go," the difficulty does not
occur when the words are sequential. She also substitutes can for and
in_ the first story, but encounters no problem with either later. Stop
stops her completely "Ride In," a difficulty that she dosen't seem
to know she has when she reads "Stop and Go" a few minutes later.
Similarly,-she calls (ride) run in the first story, but gets it right in
the latter one.
Though there are miscues in the second story, there is a very
important difference. In the first story she seems to be playing a
game of name the word. She is recoding graphic shapes as phono-
logical ones. Each word is apparently a separate problem. But in
"Stop and Go" what she says, including her miscues, in almost all
instances makes sense and is grammatically acceptable. Notice that
as Sue becomes better known she become_s Suzie to our now confident
reader.
A semantic association exists between train and toy. Though the
child makes the same suf: :thud= many times, nothing causes her to
reject her guess. It works well each time. Having called (train) toy,
she calls (toy) too (actually it's an airplane in the pictures), not once,
but consistently throughout the story. That doesn't seem to make
sense. That's what the researcher thought too, until- the child spoke
of a "little red too" later in retelling the story. "'What's a 'little red
too," asked the researcher. "An airplane,',!,_ she replied calmly. So a
train is toy and a plane is a too. Why no? But, notice that when toy
 RIDE IN
Run
Ride in, Sue.- Jimmy said, "Come here, Sue,
Run
Aide in here.
__too
Look at my .tertrain.
Come here
Here- I sew, Jimmy. See it go.
Can Come tky
And here I Look-4t mylit/tle train go."
toy
Sue said, "Stop the-tom._
Corm
Stop ithers, Jimmy."
-toy
Jimmy _said,- "I_ can -stOpi the- -train.
toy
See the trine stoP."-
_ _ = too.
:Sue said,'"_Lo-ok--at -20 toy.
Joy., _

'tit in the-44k.:
: too
_ _

-Sed my little red-tny.

-toy=
It can-ride in the train:"
--
= toy=

Jimmy Said 'See the teak go.

Look at it go."
Stizie too
Stie said,- "Look at- my- little red ky.
-toy-
-See-it go for a train ride."
Suzie too
See said, "My little red 467!-
said too
JmunyAmr toy is not here.
It is not in theireinr
toy_
Stop the taudn, Jimmy.
too
Stop it and look for my 4ey."
GOODMAN 269

occurred preceding train, she could attempt nothing for train. There
appears to be a problem for many first graders when nouns are used
as adjectives.
Common sense says go back and drill her on come, here, can,
stop, rale, and; don'aerher go to the next-book which she is 61) -
viously not ready to read.
But the more advanced story, with its stronger syntax, more
fully formed language and increased load of meaning makes it pos-
sible for the child to use her graphic cues more effectively and supple-
ment them with semantic and syntactic information. Teaching -for
more precise perception with lists and phonics charts may actually
impede this chiles reading development. Please notice, before we
leave the passage, the effect of immediate experience on anticipation.
Every one of the paragraphs in the sample starts with "Jimmy said"
or "Sue said." When the reader comes to a line starting Jimmy, she
assumes that it will be followed by said_ and it is not until her :ex-
pectition is contradicted by subsequent input that she regresses and
corrects her nfiscue.
Since they must learn to play the psycholinguistic guessing game
as they develop reading ability, effective methods and materials used
by teachers who understand the tides of the game, must help them
to select the most productive cues, to use their knowledge of lan-
guage structure, to draw on their experiences and concepts. They
must be helped to discriminate between more and less useful avail-
able information. Fortunately, this- parallels the processes they have
used in developing the ability to comprehend spoken language.
George Miller (7) has suggested ". psycholinguists should try to
formulate performance models that will incorporate -... hypo_ thetical-
information storage and= information processing components that can
simulate the actual behavior of language users."
: like to present now my model of this psycholinguistic guess-
irg game we call reading English. Please- understand that the steps
do not necessarily take place in the sequential or stretched cut-form
they are shown here. [The model appears on page 272.].
1. The reader scans along a line of print from left to right and
down the page, line by line.
2. He fixes at a point to permit eye focus. Some print will be
 270 THEORETICAL MODELS AND PROCESSES OF READING

central and in focus, some will be peripheral; perhaps his

perceptual field is a flattened circle.
3. Now begins the selection prOcess. He picks up graphic cues,
guided by constraints set up through prior choices, his lan-
guage knowledge, his cognitive styles, and strategies he has
learned.
4. He forms a perceptual image using these cues and his antici
pated cues. This image then is partly what he sees and partly
what he expected to see.
& Now he searches his memory for related syntactic,, semantic,
and nhonological cues. This may lead to selection of more
graphic cues and to reforming the perceptual image
& At this point, he makes a guess or tentative choice consistent
with graphic cues. Semantic analysis leads to partial decoding
as far as possible: This meaning is stored in short-term
memory as he proceeds.
7. If no guess is possible, he checks the recalled perceptual in-
put and tries again. If a guess is still not possible_, he takes
another= look at the text to gather more graphic cues.
& If he can make a decodable rhoice, he tests it for semantic
and grammatical acceptability in the context developed by
prior choices and decoding.
9. If the tentative choice is not acceptable semantically or syn-
tactically, then he regresses, scanning from right to left along
the line and up the page to locate a point of semantic or syn-
tactic inconsistency. When such a point is found, he starts
over at that point. If no inconsistency can be identified, he
reads on seeking some cue which will make it possible to
reconcile the anomalous situation.
10. If the choice is acceptable decoding is extended, meaning is
assimilated with prior meaning, and prior meaning is ac-
commodated, if necessary. Expectations are formed about in-
put and meaning that lies ahead.
II. Then the cycle continues.
Throughout the process there is constant use of long- and short-
term memory. -

I offer no apologies for the complexity of this model. -Its faults

G00141AN 271

lie, not in its complexity, but in the fact that it is not yet complex
enough to fully account for the complex phenomena in the actual be-
havior of readers. But such is man's destiny in his quest for knowl-
edge. Simplistic folk lore must give way to complexity as we come
to know.

REFERENCES

1. Betts, Emmett A. "Ride In," Time to Play,-Second Preprimer, Betts Basic

Readers (3rd -ed.), LanguagesArti Series. -New= York: American_ Book, 1963.
2. Betts, Emmett A, and Carolyn M. Welch.="Stop and Go," All In A Day,
Third Preprimer, Betts Basic-Readert Nevi-York: 4merican Book, 1963.
3. Chomsky, Noam. Leeture at Project Literary, Cornell University, June 18,
1965.
4. Goodman, Yetta M. College of_ Education, Wayne State University, doctoral
study of development of reading_in firtt grade children.
5. Hayes, William D. "My Brother is a Genius," Adventures Now and Then,
Book 6, Betts Batic Readers (3rd ed.), Etat:nett:A. Betts and Carolyn M. Welch.
New York: American.Book, 1963, 246.- 7

6. McCracken, Glenn, andCharles C WalcuttzBasii-Reading; teacher's edition

for the preprimer and prither. Philadelphia: B._ Lippincott, 1963, *I ,

7. Miller, George A.-"Some Prelimintries to Psycholinguistics," American

.
Pay-
chologist, Vol. 20, No. 18, 1965.
8. Spathe, George. "lading-In The Eleinentary SChool. Boston: Allyn and
Bacon, 1b64, 12.
 MATttIAL
TO It A FLO* CHART OF GOODMAN'S
MODEL OF READING

Ln SIM M 1.= Mt...Len

eV/ MOVIdert ed eat LW 12rstety m AL D OM AMATOInt Of 4K.CACt

Asa (Imafele To011t er:e ASS 5IADIX CAS Of
1151lie TO;OS TM MD OTT Of AM COMMO.tevOtcCICAL.
'MM) NTIVIAtte STOTACTIC AND MOAT%
SteCTOL AteCTS Of war

wax am MINX

MAIN YAM NNW

LDxtAltt
NI*t roar V
OIC
.1C
WP) as 11/01/0t

acme
rosstsst

wavy ms wutan
TITS Intl: 'Mt
01101CL Ia SteXTIC twins et vsre Met KV CO
AM Sert Ante ACCIASIATTO 05551115 AND
CddLIT MUNK ' STORIO TM Lde
TIM Med
L. 1.0 011;110 MIMS 00.

"iglu
most TTO5 LOTT
MD If PILL

Editin's Note: This chart is an updated version of the model described in Goodman's article. It is taken from a report on "The
Jo II) on ;et.rwaoanto.4 In ?).e I c T
 The Reading
Reading Competency Model
RICHARD L. VENEZKY
University of Wisconsin at Madison
AND

ROBERT C. CALFEE*
Stanford University.

WHAT FOLLOWS is a sketch of the reading process, in model- form,

for a competent reader. This idealized individual is assumed to be
in full possession of his sensing systems, speaks, has an average IQ,
and is emotionally stablethat is, is normal in the fullest sense of
the term. Competency in reading is defined by two factors: overall
reading ability as measured by a general reading test which, we will
assume, taps basic reading skills,
.
and the w-o ratiothe ratio of
.
comprehension of written materials to that of oral matenals. (It .is
assumed that this ratio can be determined accurately for any given
population .1) A competent reader is one whose reading scores and
w-o ratio approach thatl of the general adult -population. Tor the
present model; silent reading will be assumed, although subvocaliza-
tion will be a functional component. The input to the model is
printed materials, in the English language, like this paper; the out-
put is a vagary called understanding; the
.
model describes a pro-
cedure for generating output from input.
Central- to the reading process-As high speed (by human stan-
dards) visual scanning, dual-processing, and the search for the largest
manageable unit (um). The scanning process is directed by 1) gen-
eral know3edge such as the nature of written materials, properties
The work reported het,: was supported in part by the Wisconsin Research
t and Development Center for Cognitive Learning at Madison, under U.S. Office
of Education Grant No. 5-10.154.
'We will not br concerned here with the experimental procedures for deter-
1:-
mining the w-o ratio, nther than to state that a reliable test must deal with both
familh.r and unfamiliar material and have some compensation for the differences
in reading and speaking rates. (Whether or not subvocalizers should be discrim-
C
t: t inated against in such a test is a moot point.)
,..,..4

r 273

v_
274 THEORETICAL MODELS AND PROCESSES OF READING

of the real or imaginary world, language habits, common sentences

types, and 2) immediate knowledge, obtained from the material
being read. General knowledge will be stored in the Integrated
Knowledge Store (ms) while immediate knowledge -will be in the
Temporary Knowledge Store (ms). (While these could be labeled
long term memory and short term memory, *there are associations
invoked by these latter terms which we would like- to avoid a_ t
present. Furthermore, the Temporary Knowledge Store will need a
longer persistence than is generally_attributed to short term mem-
ory.) Two forms of immediate knowledge must be distinguished:
a) that which is deduced from the material= being read- and which
therefore creates logical sets (expectancies) for particular words,
phrases, or ideas; and b) that 'which is most immediate to where
the eye is processing at any time, that is, intra-sentence, intra-phrase,
or intra-word data that aid in the search for Lmu's. The deductions
drawn from what is being read may become part of- the In without
overt processing on the part of the reader; the other class of imme-
diate knowledge just mentioned will obtain this status only by overt
effort. Thus, the spelling of a word may be encountered dozens of
times; yet, unless the reader consciously attempts to retain its graphic
form, he will not in general have facilitated his ability to spell it
(He may, however., be able to spell the word for other reasons. All
that is being sail here -is that there is little or no facilitation of
spelling from reading because information like spelling is used only
to locate Lmu's and is discarded as soon as it is no longer needed.)
Two major forms of processing generally take place simulta-
neously: syntactic-semantic integration of what has ju:s. been scanned
and forward scanning to locate the next Lmu. All of this is directed
by the ControrUnit which mediates between the various stores and
the ongoing prolesses. Information passes from the scanning and
integrating into the Scratch Pad Store (srs), from whence it may be
routed to other stores. Information retrieved from the other stores
also passes into the Scratch rad Store, to be used in processing.
The control unit, or hOmunculus, is chiefly a convenience for
di:zussion. It is analogous to a computer monitor system which me-
diates between running programs and peripheral devicesschedul-
ing jobs, locating routines for performing specific functions, and
controlling the flow of data into and out of the machine. Integration
 _

r
 276 THEORETICAL MODELS AND PROCESSES OF READING

begin to subvocalize hen the rnatria1 leomes difficu1 perceptually

or conceptua11y Furthermore, the works of Calfeeet al (1969) and
Venezky eta! (1968) show that adults are capable o1 applying a
great number ofietteisound generalizations, even thoughthcy can
not verbaIize the generalizations themselves.
Cenal to the entire scaiiiingproths is attentk)n.tagging each
- item scannedletters, letter-units, words, phrasesis marked during
forward scanning with a tag which indicates, for integration, how
much attention musbe paid to the item (It is assumed that-items
requiring-no special attention receives airotag.that is, ire un
marked.) The tag indicates that-the particular iteinhas not- been
identified yetor thedegree of i4entificton made. (For example,- 4
a=strange proper noun will be identified m forrd scanning as a
proper noun and may even be classed by typepersbnal name,
street,- etcyet still be marked for special ttntion.) intgratio
involves: i)identification of taggeditems, 2) relating newly acquired i
matrial with ,what is already stored, and 3) predictingwhat should 3
bescannedfornext = -=
Scanning, therefore; fscentral t the reading process, and in
a ba'ance between two processes integration and forward
-
scanning, the former being strongly dependent upon the latter. if
the material requires excessive attention auring integration, ethr
because of perceptual or conceptual difficulty, forward scanning is I
restrained. The balance between integrationand scanning, however,
also be controlled by the intent of the Teadcr: reading 'light'
4
novels or other materials requires primary attention to forward
4 scanningwith little integration. Digestion of dense writings, on the
other haii4, requires extensive integration with limited forward
I scanning.z Poorer readers may not be able to shift the balance be-
tween these two processes for different materials and goals. -.-

Forward scanning sek out word and phrase breaks, and asks,
along the way, "Is there anythin unfamiliar in the terrain?" If so,
4 tags are assigned to indicate the degree of strangeness. Sarching,
however, ii always oriented towards the biggest digestible gulp j
4
the largest manageable uniC The information stores referenced dur-
-3 For the competent reiIer, fotward scanning always occurs. The degreeof

forward scanning is marked by the distance betwc-what is being scanned and

what is being integrated.

-,----
S
 _

_ -_-_--_-_:=-:=_;- = ----_

_
VENIMKk---AND:CA

to CtyiIi ma
to account

F7_atitittiti
ILETICAL_MODELSI-AND--PROCESSUfbUREAD
estionu_
 -_-_,i-:-- --_ --_ -_-- --__--- i__---- _ss;- --- _ :---- -__===_.=-s_ :- -__-_i_
_
---- _-__---- _----f c-_ _- -=-
_ _-_- -_ s=

_
_ = fr- =_--- --=-_ _- -.,_ -_- ___-z_:_ _in -_ --_ -___ - r__-- ---_ --_ --__s_
-___=-_-n,

_
 nrmem_
ezau We=

t
_

sr,__-=t,_ ,
a-tarage- and retneva TrotessecareAnvo
don.
 ,
i
286 111LORETICAL MODELS _AND PROCESSES OF READING

3.- Common words vs. uncommon words -

a. n we show a major difference between -recognition of com-

mon words and uncommon words
_.
in terms of the recognition
procedures?

concern .ere with remedial _reader -who n-has

pro on _r_ segmenting
.. -sou ucin mdmatio
-__ a MUM an appreciation e it
----_, ---- doesn't at_ discover
.- _
amnia abilities processes or skills model
cts-tre_=im- rtant reading. ,cannot.
___ ,., .
probably e
-
e- -rem
_ ation
uesuons,- o
roc
remedial
tent-reit-
e integration-
_ r____,_______-_, -__ __
_
 adison,

nitrncticiiircat

UcatlOni
esear_
 ature"--b

Educationalf-Testink_
erne

i.Oate !mime on slexia an

ei
_

. --
THEORETICAL MODELS AND PROCESSES_ OF READING
THEORETICAL MODELS AND =PROCESSES OF' READING
 _

300 THEORETICAL --MODELS AND PROCESSES_OF READING

reader is hi est fortficifitheskill have listed as --

lion of printed wordsfrom -zwhatever cuevare-avada_le,'=an usually
this skill in turn djends the other
---_---
iti-_-_p_artictilat: earning:ipitternilktoriespondente betWeem
=
=
letters and sOunds. The ability of the adult reader-716_apptehend, _
_1 -I meaning quickly is an advanced form of skill 7, and his .;a i _-to-
thinkz-abontWhat herfel&ii-=iiii- Vane ortiv-0
_ The "great debate" about-how shOuh
=
_
a=debate:a tat eorder tn_ which-the -shou
-rda&towar learning- each -of the ikils. Few
mature reading involves
skills should tar
are e.
skills s ou
they have been listed;
should be an early ent-
via-letter-sou-it relations.
believe that the s
ollowing o

nnte -w

are $11
prehen

ild sltuld learn to

_-=
 J.
CARLL 301

4
This latter -view is held by thcse who argue that there should
be an early emphasis on getting the meaning from print, and thit
the child should advance as quickly as possible toward theword-rec-
ognition and meaning-apprehension capacities of the mature reader.
Skills 2, 3, and 5 are intioduced only iftr the child has achieved
considerable progress- towards mastery of skills 4, 6, 7, and 8.
-: These are the two main views about the process of teaching-read-
-
ing. If each one is taken quite strictly. and seriously, there can be
clear differences in the kinds of instructiOnal materials and
procedures that are used. It is beyond our scope to discuss whether
the two methods differ in effectiveness. We would emphasize, rather,
methods may differ ineffectiveness from child to child. Further-
1- more, it is possible to construt othr reasonable orders in which the .

various components of reading skill can be introduced to the- child.

There is currently a tnidency to interlace the approahej distin-
guished above in such a way that the childcn attain rapid sight
I recognition of words at the same time that he is learning letter-
- sound correspondences that willhelp him "attack" words that he
-
- does not already know. -

For the child who is having difficulty in learning to read, it may

be necessary to determine exacily which slills are causing most
- difficulty. The dyslexic child may be hung upon the acquisition-of
just one or two skills. For example, he may be having particular
trouble with skill 3the recognition and discrimination of the let-
-
ters of the alphabet, or with skill 2the dissection of spoken words
into component sounds. On determining what skills pose obstacles
for a particular child, it is usually necessary to give special attention
to those skillswhile capitalizing on those skills which are easier for
- the child to master.

Uncertainties and Research Problems

I The above description of the nature of the reading process is

based on the findings of nearly threequarters of a century of research.
A good deal is known about reading behavir, yet there are many
I questions that have not been answered with precision and certainty.
W shall list the most important of these.

-
 3O2
THEORETICAL MODELS AND PROCESSES OF READING

Questions about the mature =reading= process

1._= =How== does the individual's ability to recognize= words -
stantaneously develop?_ What =cues= for= word recognition _are most
portant? =How =and when does awareness of _spelling= clues and inner
speech representation recede, if at all? What is the extent of the sight
vocabulary of the mature read& (It- should be= noted that most
studies of Word recognition processes have been conducted with
adults; there js need for evelopmen ,studies inewhicl word rec-
ogn Won* .ocesses would e investi gated over different cl rono 6
e leve s.

inferenti _= processes?
7.at ire _ e ma or sources of_ individual differences in rate
and accuracy of comprehension in mature readers?
Questions about certain components _of reading shill as they =affect
=

learning
1: In what way= does knowledge of the spoken language interact
_
with learning-to read? What kinds and amounts-of competence are
des irable_before the child undertakes any given task in learning to
read?
2. What is the nature of the ability to discriminate sounds in
the spoken,language and to diisect words-in terms of these sounds?
HoW does it develop and what role does it play in the beginning
reader's learning of letter-sound correspondences? How can this
ability be taught?
 3. Hoiv do children learn to recognize and discriminate alpha-
betic letters- in their various forms? When children have difficulty
with letter recognition, how can = these = difficulties = be overcome?
4. = How do children learn the left-to-right principle in orthogra-
phy, both as applied to individual words and to the order of words
in continuous tex e ther children with special difficulties in
learning this component of reading skill? :
5 y w at are e most useful and functional patterns of
ndence in = =what

ere one best_order in which intr uce e com

of_ reading s in e learning process, or are ere different orders
depending upon aracteristio =of = individual = children or-grouprof
children? If so; how can thesees_e- individual -or group characteristics
be determined?
3. On the assumption that there is an optimal ordering of skills
for any given child, how- much = mastery of a given skill is deiirable
before another skill is introduced? =
 VIRGINIA =$ROOK
ew York UnIversit

c a racte r istic ahar e t eseA eelatti- vitietov _obvious-

= =in the than In th_e iother ti4oAat behaviors-
-p= neitt-anslitLTiof=authniatie-res ez=a uence
tiernede-,zt tinfulati-ont at con 4nts tfie= subject: The =reader= 40a
--not_zMerely-zregardz=-I-- tian zimmediate)
theSsage-._11ef
____
must7int-en tPrea t ei ;aftentioe-
tO-_itatueauifig,jf= e a e to respond =to Its -contents What
a=phraeAikts-_ ay-zattentoirler might -imeariAn==thiaantekt-haaitot=
received ,thoughEorexperinentaLresearch,==but-itwould lc se en
t:vbe oHun rhentaFim rtanteE-t6=anrunderstanding if=vh_ at_ t e
readingsroCesa is allz-za =

-Let42sTconsiderbrieflystMe-Cif_the-more.-obvioUs waysin-_which
intention= affectareading:-__ :)--AS,rio-ted,---if=a-:-=persdn_lOoks--idly_=at
page_ of text, with no intentionto read, he may recognize only those
few wordsAanivhicbhis--gaze-:marreit;==2)-:Art :effiCient__ reader may
read_ textlor Meaning-atiUpAo--4,800letteralier- ininUte, _but at -that
rate he is very likelynotitO-notice_minor spellingl_errors_Or omissionsf
therefore,_-this--Agure certainly-di:its-not:titan:that each letter is
actually looked- at, -even =though- coMpreherition scores may-be very
z

high. 3) In fatt,- if the-taskrequireathe _reader to look at or near each

1Supported by RICH and HD R0111D04213.01.

 HOCHBERG AND- BROOKS

-lettert-for-example if he it= checking--= for imisspellingt or !broken-

letters,- atittproofreadingtheg-his=reading rate will slow down and
hittomprehension_score drop.- Here,r-Ahin, _ate-twO -clearly
_with=-intentionito_ ettracr meaningful -con=
-arable- reading-taski:1-one=wi
-tentifrontitheprinted-paget--: the -= other tvithtintentionft-opay:attention
-to theletters-;and-the Spelling of-the =Words that_InAke =up the printed-
page-._These--and-fother-lask.v Set either =by_the=goals given- the- reader;-__
by the relative difficulty or unfamiliarity of the materiaL affect
-z-the---reader visual -_isearch rformincesja-_-=2
= I -___t-otirte; -eatu tire- ffititherat- as- lightly
as e_AlirzTA0-iatteil etter.-_,Instea
e aviOrit- etertnitie t angm einan iiiripose
b what he -is 4-ta in ===at:on-e intitastin efit
aiot erlookin -ar a mosr eazic et e ;--=
-z eneraLite be aiiorsJnvolving_
= _

iverse
t
Skills 1scai n =an; _be- itrarc t
:--0

-abilit einoe_diii:-rininatezlettert=in;= etter= o , _ea _

fp oneme_z_cottes _entes_;_an eneral;--_

on-en MI ht-_--xpeCt=thattt o te-s s =are= to`_some degree = neces= ---
sary to e=:atquisitiOnT of higher onessome-=_zininimnin--=-'agility
letter iterithinatidif and e==*i- 40ord;rett--- itiorrit-=iieeded
er -Tor- readingfor-meamn totee iii-,At-;-seenii2-
treniely-likely=that every practiced reader has =had -toz-acquiret
= At fethet t a geir=hil- tat e err o_ Wa_ rd litetaty,- and t ha- r h e
--can=fallrbackuporthen2whenAhetaik-requiretit;=foretample,_:- he
can dissect - some -unfamiliar=-Word-or-fpastage--into: its-:comp:Went
letterts- and -the-n =putt loWertkillt are
probably used en-=- tbe= skilled reader= -reads
nOrmally, that iS;_When= heAtteiflifsleyetn_locketrievelingnittit :meaning
from-The -printed--_pagein: as= natural and-tontent=oriented= a_ way as
he= uses -h itn-ears-to-= to-=tpeech,-- he -prObably looks at the text
infreq uently,- compareokta When= he :prOofreadtor compared= to a. less__ _

Skilled reader. He -satiripl the text-in- order to -deVelop_ hypotheses _

about what the_inett-Aring_of=sythbols consists of,- and- to-test those

expectations at appropriate =places- further along -=in- the text.- This;
of toUrse, sounds -more= like= what- we think of a_ s "skimming" than
what we consider to be reading, but no hard distinction is really
 listenin
eat

.F,cannoti=r

_=res
attended
ef-thatthe !s
_requires_
 210CHBERG- AND BROOZS

filter- at-all- in=-orderAci- explain-selectivefattentionAn -listening;_

steadi-_Werieed: to,considermore- closely-i what is---reallyi
= a- ttenti ve-W*46nm to speech:
The study- ofivhich- sounds-=listeners- confuse
_andivhieh they-- can,-distriMinate-i-lends somesupport =___thelproposi;
Aion- that zwatdo_ not norma4do muCh-with-_-the-rawphoniccelethents-
ofizspeech_ 00. respond, intteadi=sto--lOrger_p_sc-Itet-cifl
--SoundS-,-judgiUt them to; onesphoneme_or anotherzontthelbases-or-
_-- certaini;distinctive--2-i atur tinethie_ :attires,* i
those -which = distinguish _the _listener s actions,
for example, ,ton keititit The perception WO 401,111
--accom lis rocess-= 13510
that_=is by resIo
corres ridin features _of= one's lans =to - articulate -sT= ech _ - - - =_
eisser tenni
iisame_Jou ine-las3listenin loun ds ln active
subject e,to nerattri
-r- eneratin
ma eior hratet
sentences _in _lengtlr -depending -_conte-nt-o e language-
itua4inte ner un 61 =SS=
listens checking =the ttsin dt-=re uendift-o_:_con
-Or revisehis hypotheses :- Speech sounds -vere-nor antici
=this-lathiOn-,--iiot -stor er =-
speech = structure,, r_esentjowmu materiaLto-retaini -
memory- span -bite =:saticl-independent
niaterial;_iiitcannOt= =be _recalled: =e-kplaitt =what -happens
infihe -two-chafiritiexperinients-i -Without Ieconrie to = =any Concept: of
allter:-__-theaubjeCt -makes :actiVe;anticipatrirrrespontet Ao -the initial-
phonemes_-thati-he-ireteivesin---the--channeLto which- he is attending
and -stores the results of-hittesting- of those expectations Sounds that
were uttered- ofilthelionatterided:channel __may :1*-::briefly- stored-
unrelated sounds, andlvilli-lisually-lade'firom-inemory, while the_sub;
ject reporti the message received on the -attended _channel As we
should expect- =froin- our _analysis- of the sitnation, the =experiment
_

cannot be perfOrmed-_ unless: the --verbat material- presented is

nized into redundant-and predictable sequences (19). This-argriMent
is presented in more detail elsewhere (16). What is- important to us
 THEORETICAL MODELS AND PROCESSES OF READING

here =is_thatiflistining-to=_speech-rests:lipon anticipating-: the results_

_ _ _

Sampled=redundant= trains= oUsourid;= then- _something = like

in-and
thii:hypOthesii=testing
listening-and- speaking 43roce-sses.
_
increadirig,---Which rooted

_tee.Uslufw-turn-to-theikills that-the subject must_ have

from= his ex-perienCes- in-- looking at= which -also= antedate
4y tin lity_-_-years:hisi acquisitiOri=of=feadirig-=abilities;=Betatiseout_:00___-
_

-regittersfine:-detaiForily-vithiiiEa very small -= region _of the_retina_ (the = -

fdveziyi--Ave---_-;mutt-leatwiabot succession of=
atitetitkliffeithedirediani__ e 6_Xet oo
Ing -at scenes-mustlocair4 by a temporal =se itelke_o patterned itimu,,-
ation = But =whereasA m e 1stat:on ae r in -Ordeit
anOsiest-ol gui a tiMWo it emtttirioMent's-----:
stimulation -,wilLbnn the e subject -nvieWing=aliorifia 6--
expectations e Metter,==h-t his=learnedionie_
=about= t :ex a
(14;_17); The wide periphery of the =retthiitiVhith-=islo- In acuity
and t ere relin-I eta can est=
an intimation-o=w e_O_ _rver=MoVet
his -6=t-641-6Mt lin=oUthre=liiiita use-suc_ ft es
-3cieutect-7bOac- ca icintowerizen whose end-
pOints'art decided bef ore the ;;=saetadesart
-ballisticinovenienti e contents -of each lanceis- Wa
an answer 46--a-zqUestiottAbOdwhat=Will- some specific-part
6_ 1-the _perpheral ly4icu ec Uttne it'b rOught=toAielo Veu-
Wt_doinot-=know -too-_-_mudy-abou- tthe:-strategie-s4rid- tactiarof
free oye-movenent- deploymenCin--= triewirig=icend,=:biit-- it seem.vsafe-
la say-th-at the_-major pOtentiallyinfOrmative -parti:s-Of--thi-_sCenes_ are
scanned. This= = making= whatever -snail= excursions are
necessary - to filliri=detaiLlieedecttoftlicide-whafit being-looked
There is, in---generali=a---relativelY =high number of-largexaccades, a
greaudeal of=fecursiOn-,-incl=nothinfrat- all-I-that:should:Make-it easy
for a= subject to keep-precise- trat1C:of the= temporal order in -which_
any search of the visual=field- has-beeri=exeCdted, that is, no mapping
of spatial order into temporal -- sequence. But -this is just -the task
that faces the child when he first-learns to -read: he -has to put to-
gether letteri into words- by a sequence of small adjacent fixations.
 etter '_

2- Adjustment to this task- is not always satisfactorily achieved. Thus Taylor

and Robinson (1963) found that inefficient early oCular motor activity` may result
in final motor habits which inhibit efficient reading after- the original causes
(difficulty with word identification and recognition) no longer exist.
 510 THEORETICAL MODELS AND PROCESSES OF READING

single fixation -that- is made in-_ the_ -course==of the treading- rprocesS.
Many-years agOi-Cattell- (5,-6)_shoWddrthat- the =exposure Irequirect=to
identify each of-atachistoicopically_-presented sevoft_unrelated letteri;-_
increased -as ==the -number=of-letters--=,intreased. The-=_-exposure:_'thirie
-reqnired--_tkyrecbgriiie-soinefaniihaelirord--0--phrase however, -no-
longer -than that for= a shigle-letterArriother4ords.,-- the- skilled= reader
_Inch ngr_up t =iv hat, e-'-iinskilled:_zeider have- to- _-

-by mUltipleffikatiorit;-,Since--theni=it haiAleetr,amplyi-shOwn

identify -by=
can-_,reco ne-_words: and shapes _loth which: they -:are-
tire)thifei=thaitaretieeded4ar_niaterial=44i_
ihar 3 ere
er can 111 diVef=ati thief,
text however __t at- --=:--=seact etteriny
erent rbcdsses- eent-affede__
ere-are=reasons=-to :e e-c miliarit -occur =
cause7ti zsu leersfire$ onse= ;mor-e=--rea respon
en-= etter-;-==tv-dr raie-evettAv en,on
--r_
ot_t_
__
= senso til*Jhatilt.es nse_=isatttia lr resent=
_ _ _ _ _
e -can-
tterreea miliar=matetia speech -Aoun
t_m_ottlitex it=t -e-ratiarigeineht_ito etters,4)
moul otteni-Iv killed
:rea haSAdenti ezttiMu us,--ai=being-t- azfainiliar-*Orkfhecair---
=-10fittateralli= ividuanetteiiidie
upon to do =so- assuming`he can _spelhit), regardless fli5lthenuniber
---ot-letteri involve nsuppOrt:=tifshi..1-=-ttt --eition,1:147-e--,-shottldi' note- --
thatit-is-=poSSi 163= tiii_Tdenionstratceffects;o familiarity= on = tachisto- _
scopic --word-reco ition-AindeftenditionOir*hich=_ thost;effeCts= are -

almost certainly-dne: to:thedifferentiali=airail units

bylvhich ,the-__Stringt of=letters-7may=b-e-=enttidednd-=s-Ubsequently-re,
generated, and not dite differeritiatenSOry processing= (12,
Andithe demonstration :_by:Graf_=-and--=-Torrey--(11)--that-COmpreheri-
s- ion scoreizarei higher for -44,Tritteri-iimatetiatilhatis broken -at major
linguittic boundaries -than-7-=are-n_torriprehenSion:-:Scoret=_fOr--- Material-
broken at -minor -boUndaries!might_ be interpreted = as the---effect of
greater= ease Of=Storhig:_the" in-aerial= in the ifdtrier -condition.-
= By responding_ to= the few -features foveal vision
with an entire- word tor--phrise, -the_ skilled-==reader, then; has largely
relieved himielf -=of th-e:netessity -of -looking-_=closely at- the -text. He
 therefore needs to fixate only those parts of the array, further along
the page, that will enable him mmake new guesses and to check his
previous ones.The better the reader, the more widespread can be
the fixations by which he samples tile textjust so=long as the text
provides him with contextual redundancy and as long as his task
permits him to attend =to the meaning or content, rather than to
spelling or t6 individual letters.
The question then becomes, what guides the skilled reader's fix-
ations? How does e decide_w text in question_ is, without
look ng area
0 -COWS skilled
cations of = =what he will _Snd when 1e looks= er alon
_are_based in= art _on e.s tax; _arid =ineiniii -6f* at=
morder to- Oranything=less==
than etterty=lettetirea n suffice, =so redundancy;
_ mine --=_Our---presentri
meddle of =long r=s-e-ntenteils= close
to 100 percent. the 'Estelle17-
ere -reader=caii4ornifilatean --ec eritt-=
-:neednot,100 =atea :letter
enough
ows--isia -utitht=constraints Aan a &tbfittikea__
profitable -guess at oW-rmuch=lurther along he should next in-
Ordertoitestzt ese= ate=riew=ones.
been-,foi i Eth a r r ea n eirir s-aft-lnluet tedby_=asiotiative-fac t Ott
and -language habits based on syntactic nework003,14 , which
is =what -we= would expect _if thesetgare==-only-intettait--
teste 1- ternm
iantio fixation-:-Cso =for
search guidance- (13)_ Unlike e--rlittener;,_ oWever,i the reader not
restricted =to -_previous_==contentias---:the of extrapolations
because hezzcan,:iii,a-dditiow=use4hei=inforinationigiven_iivperipheral
select the places
at:Whichlie-shOuld-seektuCcessivel-stithultisinput.--Thus-, at -the very
leiSti-hshould--be-_-_able =to =aritiCipate,- through=_the-nse of interword
Spaces_ that_ appear in the petipheral_viiion, ;where:lie_ Must-look" in
order to-fixate -the 1210st-infoinative-portionsiol*ords (that is, their
termini), and which --words-are-likely-10- be= short functors. =Call such
determinants _of- fikation -PSG 'for peripheral :search guidarice,_ The
practiCed_rtader must move his _eyes -under the- coMbined control of
cso and
 312 THEORETICAL MODELS AND PROCESSES OF READING

_ The beginning_ reader, _who': really= must-look at most -or all of

the _=letters,__,probably MakeS,,,little --use- Of --,PSG. He --is, -therefore:_less =

hanipered--than_ is the-better reader when- the --- intormatiow available ---

in peripheral -=vision --is :interfered--with:for example, by= Belling in the
interwo-rd-spaces-so
_ -asf ta-make-the wordE,boUndaries,indiscriminable
-_ When, viewediperipherallf(15)-,Readers--Vith`!tunnel=_viiion,'!- that -is,
lacking peripheral vision, have_-verylow-reading= ates., But, the
coVerysof---What:-actuallregOides-skilledreadefs_z:fikations---andf-,hoW
-_-,tite-sean-ibeimproved!callrforrUorethanta,siM e, erhotistratiOniall
the rtance Of peripheral Nision;-=_ - possible
which-rent :=:_eXplorint er=_Nelson=and = Murray - -=
lanzeris-- resenc t -erea erfcine-matitallyLWit
-_--=text =istr=0 ith,4-iSirttUlate- ation-- -at
vision containing four or five etters;_ appears Iskceisively-Jits-iari-Ous
cesfin_ithe,--1exL This --roceditrelirOvi-es-=the -- c- reader _with-:=4n=
=c-MitpW-e:Saniple_so = The enttvariablesiiw-thit
inv_eitigationiare=theie:i ow=t eisam eiiteta_
oi%7t e=simulated---
_ _

Ifixationpint-s--ar-e,pitietite rOugh-oUtt e--i-tektr anct-fthe4in-ottrit =

of information,-_ if is-AfartiMitt -e-=_peripherp-Of=the-

reader fforni---_at-roi_41-eaf r- eadizi--tate; we -niq_--assttrrie-

that we-have simulated his s- ampling procedure In other words, we
_ -
haire=given= hith,ithe=i-inforinatioii==he-=,tiormally= employs= in =terms of
the =-
=
= Thit=pittire -of skilled=readint_itifntasIceetsiverextrIpOlatio ns,
not atiacCurate;
picture, it why= it:appears-as though really,skilledfreaders
_
are processing- a= tremendous -attfattit =of lijnformation-- per second;:
whereas, faet,= they are= not=theysmerely-iknow--ai great- deal-Aout
the language- anctabout-_-Writers.--If=we ate -toiattem pi_ to_ teach readers
to make -such_ ektrapolatiOns-ziand= to _use= theMOve-ihalL have to keep
two things-in mind:-=first,-Ave-imUstifit-thertext=-which we Zare-r-teach;
ing them to- anticipate-_,=to anticipations z that theit speech habits
already, provide; second-,---we_-_cannot'expect -them-to- transfer such
skilled- reading habits-to less-i idiosyncratic -text until-the appropriate
habit& of-speeth and-InoWledge- have-been- built -up.
Btit, are we talking -about "skimming"-or "reading?" In view of
the fact that visual perception almostalways- entailS- a- procedure of
HOCHBERG AND BROOKS

REFERENCES
1. -Betti,- A;-_--112.e-ading:-;Pereepttal :Learning," =Edupition.--__ItidianapOlict
=

2.- Brandt,:IL-F. The P.tych-ology of Seeing.; New York: Philosciphital Library,

J945. -z-- --
Bniadb-ent,1). E-; Perception-and Communicatio- n: London: -Pergamon;r1958.
4.: Buswell,-G._T.-Hoto-People-Look at-Pictures.- Chicago: Unhieriity of- Chicago
Press, 1985. _ _

5. Cattell, _J. McK. "The Time I_ t Takes

toiSee_and-Name-Objecie Mind, 11
(1180), 6365.1 _ s
6. Cattell-,J.-McK. "The-Inertia of _theEye:andSiain- ,"-_-Brain, 7-(1880), 295-312.
7. Cherry, E. G. 'ISOme_Experitrientsron the Recognition:6f Speech,- with_ :_One
and Two _Ears," J. Acoust. Soc.-Amer., 25:0953VS75-979. =
8._ Clay,_M. -"AiSyntactic Analysis -of -Reading-Xrrors," Journal-of Verbal L- earn-
ing and Verbal Behavior, 7-(1968), 4344311 _ --=
9. Deutsch,_ J. A., and D: Deutsch. "Attention: Some Theoretical Contidera-
tions,"- Plychological -Review, 70 (1963),' 80.90.-
10. Goldiamond, I. "Indicators of Perception: L Subliminal Perception, Sub,
 THEORETICAL= MODELS AND PROCESSES-_OF READING

_ _
ception, Unconscious_ Perception: An= Analysis- in Terms-oflosychophysical
Indicator-Methodology," -P.rychOlogiee&Balletini-=55(1950),
-11. -Graf, R., an d-1.1 "W-_ Torrey. "Perceptioii-of Phrase Structure-in Written
Language," -== Proceedings- of-=the -14th fAnntiaUConvention-Tof the Ameriedn
e_PsychOlogicl- Association,_1966,_83-8t.---
12.-_ Hayes,--W.4._Robinscin;-_and_ I.,-Brown.--''An'Effect-- of -Paitxperience,on
Perception: An_Artifitt,"Anierican-PsycholOgis4-1611961), 420-(Abstract): -_
-zHenle.:M:_ "The F...iperiinental Investigation_ oU_PastEsieriences as-- a-rDeter--
minantiof Visual-TFOrni_Per curnal-,of Ex- --rirnental-Psychology,
1
14.= :MY
Th _ =in Perce lion.
instoir-1969i_309431
nsponents:o lationaianctilEi lort
=sear -acting,-

16;_ ttentio_Organtitioian nac6un .rgostofsk_y

d.VAttintioniFContem- =1The 1EidrAnal evYo =45pleton-
ntay.Graftsairtkonin
-17. Hochliecg _J:' Units= of= Perceptual_Analysis, "in -J ---Handbook-
of Psychology 4.ngle* Mice=
18.-_ Judd, asw enV -A -Study of elyarious--
Ty plementatEdticatiotta _on6 a -=2.3':
--19. "Decisions_-_:Goncerning-
-Re* tt
--- Journal:of =Ex-fierimental=psychkgyi=18=(19__ 260-265.=-
20. Liberman, A..-.-_-M--;-="Somefiteittltsiio on S J.
= Acoust.- Soc.- rAmer.,--2941901),1117-1
_ _ _
- _

21. Moray,_;11.,_And-A-__Mcf_Tayleir.--__''The-Affect OURedundanCtin-Shado*ing

TiVo---,MessageS,"-Langnage-Sidliie&G 16116_53V376.335.
-22.-=-Neisser-,A_GognitivePsychotogy-liewlforki-Appletein.Centu_ry-Crofts, 1967,
29. Rosenberg, =S. "Associatire=-Fattorin- the Ricall'Of-IConnected -Diatourse,"
-Psychonomic- -Sciencti=4-1960)5=54.
24.- -Samuels, S.- J, "Effect:of -Wcird--Atici-ciatiOnsCOSSReading:Speed_and- Retail,"
Proceedings of --the_ 74tkArinual_ronnintiori of- the American Psychological
Association,51966.
Solomo n-, R. L,- and- H. _Howes, "Wok =Frequency,- Personal-Values, and
Visual--Duration Thresholds," Psychologicallteviete,i58- (1951), -256471
Taylor, S. -E., ind_11.-A. Robinion;_ "The itelitiOnship _Of the- Ocular-Motor
Efficiency of the -Beginning_ Reader- to-Ilislatcess-in -Learning-to -Read,"
-paper presented: at the American =Educational-Research Association Con
February 13.16, 1963.
Treisman, A._ M. "Monitoring and-Storage of Irrelevant Messages -in Selec-
tive Attention,"lournal of-Verbal Learning and-Verbal- Behatior, 3(1964),-
449;459.
 -PERMISSION TO REPRODUCE
RIGHTEO MATERIAL THIS COPY.
BY HAS BEEN GRANTED

Eleanor- J. Gibson
TO ERIC AND ORGANIZATIONS
UNDER AGREEMENTS OPERATING
, OF EDUCATION WITH THE US OFFICE
FURTHER REPRODUCTION
OUTSIDE THE ERIC SYSTEM REOUIRES
PER
Learning to Reads MISSION OF THE COPYRIGHT
OWNER"

Experimental psychologists examine the process by which a funda-

mental intellectual skill is acquire&
ELEANOR J. GIBSON

EDUCATORS and the public have exhibited a keen interest in the teach-
ing of reading ever since tree _public education became a fact '.(1).
Either because of or despite-their interest,- this most important sub-
ject has been remarkably susceptible to the influence of fads and
fashions and curiously unaffected by disciplined experimental and
theoretical psychology. The psychologistt hive traditionally pursued
the study of verbatlearning by means of experiments with nonsense
syllables and the likethat is, materials carefully divested of useful
information. And the educators, who found little in.this work that
seemed relevant to the classroom, hlie stayed with the classroom;
when they perfonned experiments:the method was apt to be a gross
comparison of classes privileged and unprivileged with respect to the
latest fad. The result has been two cultures: the pure scientists in the
laboratory, and the practical teachers ignorant of the progress that
has been made in the theory of human learning and in meth_ ods of
studying it.
That this split was unfortunate is dear enough. True, most
children do learn to read. But some learn to read badly, so that
school systems must provide remedial clinics; and-a small proportion
(but still a large number of future citizens) remain functional illiter-
ates. The fashions which have led to classroom experiments, such

Reprinted from Science, 148 (1965) 1066.1072 with the permission of the
author and the American Association for the Advancement of Science. The
author is senior research associate in r; Aology at Cornell University. This
article is adapted from a paper read at a c) ,ference on Perceptual and Linguistic
Aspects of Reading, sponsored by- the Committee on Learning and the Educa-
tional Process of the Social Science Research Council and held at the Center for
Advanced Study in the Behavioral Sciences, Palo Alto, California, 31 October
1963. -PERMISSION TO REPRODUCE THIS COPY.
RIGHTED MATERIAL HAS BEEN GRANTED
BY
315 American Association
Agi.,,as Advancement of
TO ERIC AND ORGANIZATIONS OPERATING
UNDER AGREEMENTS WITH THE US OFFICE
OF EDUCATION FURTHER REPRODUCTION
OUTSIDE THE ERIC SYSTEM REOUIRES PER.
MISSION OF THE COPYRIGHT OWNER'

Copyright 1965 by the American Association for the

Advancement of Science
516 THEORETICAL MODELS AND PROCESSES OF READING

as the "whole word" method, emphasis on context and pictures for

"meaning," the "flash" method, "speed_ reading," revised alphabets,
the "return" to "phonics," -rid so on, have done little to change the
situation.
Yet a systematic approach to the understanding of reading skill.
is possible. The psychologist has only to treat reading as a learning
problem, to apply ingenuity in theory Construction and experimental
design to this fundamental activity on which the rest of man's educa-
tion depends. A beginning has recently been made in this-direction,
ands -it -can be expected that a-nuMber of theoretical and experimental
studies of reading will be forthcoming (2).

Analysis of the Reading Process

A prerequisite to good research on reading is a psychological
analysis of the reading process. 'What is it that a skilled reader has
learned? Knowing this (or having a pretty good idea of it), one may
consider how the skill is learned, and next how it could best be
taught. Hypotheses designed -to answer all three of these questions
can then be tested by= experiment.
There are several ways of characterizing the behavior we call
reading. It is receiving communication; it is making discriminative
responses to graphic symbols; it is decoding graphic symbols to
speech; and it is getting meaning from the printed page. A child in
the early stages of acquiring reading skill may not be doing all these
things, however. Some aspects of reading must be mastered before
others and have an essential function in a sequence of development
of the final skill. The average child, when he begins learning to read,
has already mastered to a marvelotis extent the art of communication.
He can speak and understand his own language in a fairly complex
way, employing units of language organized in a hierarchy and with
a grammatical structure. Since a writing system must correspond to
the spoken one, and since speech is prior to writing, the frame work
and unit structure of speech will determine more or less the structure
of the writing system, though the rules of correspondence vary for
different languages and writing systems. Some alphabetic writing
systems have nearly perfect single-letter-to-sound correspondences,
but some, like English, have far more complex correspondence be-

7
 GIBBON
317

tween spelling patterns and speech patterns. Whatever the nature of

the correspondences, it is vital to a proper analysis of the reading
task that-they be understood. And it is vital to remember, as well,
that the first stage in the child's mastery of reading is learning to
communicate by means of spoken language.-
Once a child begins his progression from spoken language to
written language, there are, I think, three phases of learning to be
considered. They present three different kinds of learning tasks, and
they are roughly sequential, though_there must be considerable over-
lapping. These three phases ate: learning to differentiate graphic
symbols; learning to decode letters to sounds ("map" the letters into
sounds); and using progressively -high -order units of structure. 1 shall
consider these three stages in order and in some detail and describe
experiments exploring each stage.

Differentiation of Written Symbols

Making any- discriminative response to printed characters is
considered by some a kind of reading. A- very 'young child, or even a
monkey, can be-taught -to point to a patch of yellow color, rather
than a patch of blue, when the printed characters YELLOW are pre-
sented. Various people, in recent popular publications, have seriously
su:ested teaching infants to respond discriminatively in this way to
letter patterns, implying that this is teaching them to "read." Such
responses are not reading, however; reading entails decoding to
speecn. Letters are, essentially, an instruction to produce a given
speech sound.
Nevertheless, differentiation of written characters from one an-
other is a logically preliminary stage to decoding them to speech.
The learning problem is one of discriminating and recognizing a
set of line figures, all very similar in a number of way's (for example,
all are tracings on paper) and each differing from all the- others in
one or more features (as straight versus curved). The differentiating
features must remain invariant under certain transformations (size,
brightness, and perspective transformations and less easily described
ones produced by different type faces and handwriting). They must
therefore be relational, so that these transformations will not destroy
them.
318 THEORETICAL MODELS AND PROCESSES OF READING

It might be questioned whether learning is necessary for these

figures to be discriminated from one another. This question has been
investigated by Gibson, Gibson, Pick, and Osser (3). In order to trace
the development of letter differentiation as it is related to those
feitures of letters which are critical for the task, we designed specified
transformations for each of a group of standard, artificial letter-like
forms comparable to printed Roman capitals. Variants were con-
structed from each standard figure to yield the following 12 trans-
formations for each one: == three= degrees of transformation from line
to curve; five transformations of rotation or reversal; two perspec-
tive transformations; and two topological transformations (see Fig. 1
for examples). All of these except the perspective transformations
we considered critical for= discriminating letters. For example, con-
trast v and I); c and u; o and c.

Standard Line to curve NKr Rotation- Slant back Close

Standard Line to curve 45 Rotation Slant right Brit*

FIGURE 1. Examples of letter-like figures illustrating different types of

transformation.

The discrimination task required the subject to match a standard

figure against all of its transformations and some copies of it and to
select only identical copies. An error score (the number of times an
item that was not an identical copy was selected) was obtained for
each child, and the errors were classified according to the type of
transformation. The subjects were children aged 4 through 8 years.
As would be expected, the visual discrimination of these letter-like
GIBSON 319

forms improved from age 4 to age 8, but the slopes of the error curves
were different, depending on the transformation to be discriminated
(Fig. 2). In other words, some transformations are harder to dis-
criminate than others, and improvement occurs at different rates
for different transformations. Even the youngest subjects made
relatively few errors involving changes of break or- dose, and among
the 8-year-olds these errors dropped to zero.- Errors =for perspective
transformations were very numer- ous among 4-year-olds and still nu-
merous among 8-year-olds. Errors for rotations and reversals started
high fbut dropped to nearly zero by 8 years. Errors for changes from
=

line to curve were relatively numerous (depending on the number of

changes) among the youngest children and showed a rapid drop
among the olderalmost to zero for the 8-year-olds.
The experiment was replicated with= the same transformations
of real letters on the 5-year-old sroup. The correlation between con-
fusions of the same transformations for real letters and for the letter..
like forms was very high (r = +.87), so the effect of a given trans-
formation has generality (is not specific to a given form):
What happens, in the years from 4 to 8, to produce or hamper
improvement in discrimination? Our results suggest that the hildren
have learned the features or dimensions of difference which are
critical for differentiating letters. Some differences are critical, such
as break versus dose, line versus curve, and rotations and reversals;
but some, such as the perspective transformations, are not, and must
in fact be tolerated. The child- of 4 does not start "cold" upon this
task, because some of his previous experience with distinctive fea-
tures of objects and pictures will transfer to letter differentiation.
But the set of letters has a unique feature pattern for each of its
TABLE 1
NUMBER OF ERRORS MADE IN TRANSFER STAGE
BY GROUPS WITH THREE TYPES OF TRAINING

Type of training
Group Standards Transformations Errors
El Same Different 69
E2 Different Same 39
.C Different Different 101
320 THEORETICAL MODELS AND PROCESSES OF READING
.

Members, so learning of the distinctive features goes on during the

period we investigated.
If this interpretation is correct, it would be useful to know just
what the distinctive features of letters are. What dimensions of =dif-
ference must a child learn to detect in order to perceive each letter
as unique? Gibson, Osier, Sthiff, and Smith (4)iinvestigated this ques-
tion.-Our method was-to draw up a chart-of the features of a given
set of letters (5), test to seelwhich of these letters were most frequently
confusecli.by prereading-children; arid - compare the errort in the-re=
sultingi"conftiiiOrt=matriezWith=theise predicted by-the feature chart.
K_tetrof distinctive features fOr letters intistilie-relational in the
sense that each feature presents a contratt-whichE is invariant Under
certain 'transformations,- and it Mutt-. yieldsa= unique pattern for -each
letter. The set must also be reasonably economical. Two- feature

FIGURE 2. Error curves showing roe :4 improvement in discriminating four

types of transformation.

lists which satisfy these requirements for a specified type c--e were
tried out against the results of a confusion matrix obtained with the
same type (simplified Roman capitals available on a sign-typewriter).
Each of the features in the list in Fig. 3 is or is not a character-
-i GIBSON
321

Features A B C E K I. N.0
Straight segment
Horizontal + + +
Vertical
Oblique / +
+ + ++ +. +
Oblique \ + + +
Curve
Closed +
Open vertically
Open horizontally + +
Intersedion ++ ++ +
Redundancy
+
Cyclic change
Symmetry
Discontinuity
++ + ++ + ++
Vertical
+ + + + +
Horizontal ++
FIGURE 3. Example of a "feature chart." Whether the features chosen are
actually effective for disaiminating ktten must be determined by experiment.

istic of each of the 26 letters. Regarding each letter one asks, for
example, "Is there a curved segment?" and gets a yes or no answer.
A filled-in feature chart gives a unique pattern for each letter. How-
ever, the number of potential features for letter-shapes is very large,
and would vary from one alphabet and type font to another. Whether
or not we have the right set can be tested with a confusion matrix.
Children should confuse with greatest frequency the letters having
the smallest number of feature differences, if the features have been
chosen correctly.
We obtained our confusion matrix from 4-year-old children,
who made matching judgments of letters, programed so that every
letter had an equal opportunity to be mistaken for any other, without
bias from order effects. The "percent feature difference" for any two
letters was determined by dividing the total number of features
possessed by either letter, bat not both, by the total number possessed
by both, whether shared or not. Correlations were then calculated
between percent feature difference and number of confusions, one
for each letter. The feature list of Fig. 3 yielded 12 out of 26 positive
significant correlations. Prediction from this feature list is fairly good,
 $22 THEORETICAL MODELS AND PROCESSES- OF READING

in view of the fact that features were not weighted. A multidimen-

sional analysis of the matrix corroborated the choice of the curve-
straight and obliqueness variables, suggesting that these features
may have priority in the discrimination process and perhaps develop-
mentally.- Refinement of =the feature list will take these facts into
account, and other methods of validation will be tried.

Detecting Distinctive Features

If ye are correct in thinking that the child comes to discriminate
graphemes by detecting their distinctive features, what is the learn-
ing process like? That it is perceptual learning and need not be
verbalized is probable (though teachers do often call attention to
contrasts between letter shapes). An experiment by Anne D. Pick
(6) was designed to compare two hypotheses about how this type of
disirimination develops. One might be called a "schema" or "proto-
type" hypothesis, and is based on the supposition that the child
builds up a kind of model or memory -image of each letter by repeated
experience of visual presentations of the letter; perceptual theories
which propose that discrimination occurs by matching sensory ex-
perience to a previously stored concept or categorical model are of
this kind. In the other hypothesis it is assumed that the child learns
by discovering how the forms differ, and then easily transfers this
c=s knowledge to new letter-like figures.
Pick employed a transfer design in which subjects were pre-
sented in step 1 with initially confusable stimuli (letter-like forms)
and trained to discriminate between them. For step 2 (the transfer
stage) the subjects were divided into three groups. One experimental
group was given sets of stimuli to discriminate which varied in new
dimensions from the same standards discriminated in stage 1. A
second experimental group was given sets of stimuli which deviated
from new standards, but in the same dimensions of difference dis-
criminated in stage 1. A control group was given both new standards
and new dimensions of difference to discriminate in stage 2. Better
performance by the first experimental group would suggest that dis-
crimination learning proceeded by construction of a model or mem-
ory image of the standards against which the variants could be
matched. Conversely, better performance by the second experimental
group would suggest that dimensions of difference had been detected.
 323

The subjects were kindergarten children. The stimuli were

letter-like forms of the type described earlier. There were six stan-,
dard forms and six transformations of each of them. The transforma-
tions consisted of two-changes of line to curve, a right-left reversal, a
45-degree rotation, a perspective transformation, and a size trans-
formation. Table 1 gives the errors of discrimination for all three
groups in stage 2. Both experimental groupi performed significantly
better than the control group, but the group that had familiar trans
formations of new standards performed significantly better than the
group given new transformations of old standards.
We infer from these results that, while children probably do
learn prototypes of letter shapes, the prototypes themselves are not
the original basis for differentiation. The most relevant kind of train-
ing for discrimination is practice which provides experience with the
characteristic differences that distinguish the set of items. Features
which are actually distinctive for letters could be emphasized by
presenting letters in contrast pairs.

Decoding Letters to Sounds

When the graphemes are reasonably discriminable from, one
another, the decoding process becomes possible. This process, com-
mon sense and many psychologists would tell us, is simply a matter
of associating a graphic stimulus with the appropriate spoken re-
sponsethat is to say, it is the traditional stimulus-response par-
adigm, a kind of paired-associate learning.
Obvious as this description seems, problems arise when one
takes a closer look. Here are just a few. The graphic code is related
to the speech code by rules of correspondence. If these rules are
known, decoding of new items is predictable. Do we want to build
up, one by one, automatically cued responses, or do we want to teach
with transfer in mind? If we want to teach for transfer, how do we
do it? Should the child be aware that this is a code game with rules?
Or will induction of the rules be automatic? What units of both codes
should we start with? Should we start with single letters, in the hope
that knowledge of single-letter-to-sound relationships will yield the
most transfer? Or should we start with whole words, in the hope that
component relationships will be induced?
Carol Bishop (7) investigated the question of the significance of
324 THEORETICAL MODELS AND PROCESSES OF READING

knowledge of component letter-sound relationships in reading new

words. In her experiment, the child's process of learning to read was
simulated by teaching adult subjects to read some Arabic words. The
purpose was to determine the transfer value of training with indi-
vidual letters as opposed to whole words, and to investigate the role
of component letter-sound associations in transfer to learning new
words.
A three-gage transfer design was employed. The letters were 12
Arabic characters, each with-a one-to-one letter-sound correspon-
dence. There were eight consonants and four vowels, which were
combined to form two sets of eight Arabic words.- The 12 letters
appeared at least =once in both sets of words. A n_ ative speaker of the
language recorded on tape= the 12 letter-sounds and the = two sets of
words. The graphic fonn of each letter or word was printed on a
card.
The subjects were divided into three groupsthe letter training
group (L), the whole-word training group (W), and a control group
(C). Stage 1 of the experiment was identical for all groups. The sub-
jects learned to pronounce the set of words (transfer set) which would
appear- in stage 3 by listening to the recording and repeating- the
words. Stage 2 varied. Group L listened to and repeated the 12 letter-
sounds and then learned to associate the- individual graphic shapes
with their correct sounds. Group W followed the same = procedure,
except that eight words were given theni to learn, rather- than letters.
Learhing time was equal for the two groups. Group C spent the same
time-interval on an unrelated task. Stage 3 was the same for the three
groups. All subjects learned to read -the set of words they had heard
in stage 1, responding to the presentation of a word on a card by
pronouncing it. This was the transfer stage on which the three groups
were compared.
At the close of stage 3, all subjects were tested on their ability
to give the correct letter-sound following the presentation of each
printed letter. They were asked afterward to explain how they tried
to learn the transfer words.
Figure 4 shows that learning took place in fewest trials for the
letter group and next fewest for the word group. That is, letter train-
ing had more transfer value than word training, but word training
did produce some transfer. The subjects of group L also knew, on
GIBSON 325.

7
40
6
... 5
o
is: 4

9-
I -11-1-1 1 t t I I -t I I I t- t t -I 11-

0 2 4 6 8 10 12 14 16 18 20 22 24
Trio Is
FIGURE 4. Learning curves on transfer task for group tra ned originally
with whole words (W), group- trained with= single letters (L), ant - control group
(q.
the average, a greater number of component letter-sound correspon-
dences, but some subjects in group W hadAearned all 12. Most of
the subjects in group L reported that they had tried to learn by using
knowledge of component correspondences. But so did 12 of the 20
subjects in group W, and the scores of these 12 subjects on the trans-
fer task were similar to those of the letter-trained group. The subjects
who had learned by whole words and had not used individual cor-
respondences performed no better on the task than the control sub-
jects.
It is possible, then, to learn to read words without learning the
component letter-sound correspondences. But transfer to new words
depends on use of them, whatever the method of original training.
Word training was as good as letter training if the subject had ana-
lyzed for himself the component relationships.

Learning Variable and Constant Component

Correspondences
In Bishop's experiment, the component letter-sound relation.
ships were regular and consistent. It has often been pointed out,
326 THEORETICAL MODELS AND PROCESSES OF READING

especially by advocates of spelling reform and revised alphabets_ (8),

that in English this is not the case. Bloomfield (9) suggested that the
beginning reader should, therefore, be presented with material care-
fully programed for teaching those orthographic-phonic regularities
which exist in English, and should be introduced later and only
gradually to the complexities of English spelling and to'the 'fact that
single-letter-to-sound relationships are open variable. But actually,
there has been no hard = evidence to suggest that transfer, later, to
reading spelling-patterns with more variable component correspon-
dence will be facilitated= by beginning =with only constant ones. Al-
though variable ones may be harder to learn in the beginning, the
original difficulty may be compensated for by facilitating laser learn-
ing. -

A series of experiments directed by Harry Levin (10) dealt with

the effect of learning variable as opposed to constant letter-sound
relationships, on- transfer to learning new letter-sound relationships.
In one experiment, the learning material was short lists of- paired -
associates, -with a word written in artificial characters as stimulus and
a triphoneme familiar English word as response. Subjects (third-grade
children) in one group were given a list which contained constant
graph-to-sound relationships (one-to-one component correspondence)
followed by a list in which this correspondence was variable with
respect to the medial vowel sound. Another group started with a
similarly constructed variable list and followed it with a second one.
The group that learned lists with a variable component in both
stages was superior to the other group in the second stage. The results
suggest that initiating the task with a variable list created an expecta-
tion or learning set for variability of correspondence which was trans-
ferred to the second list and facilitated learning it.
In a second experiment, the constant or variable graph-sound
relation occurred on the first letter. Again, the group with original
variable training performed better on the second, variable list. In a
third experiment adult native speakers of English and Spanish were
compared. The artificial graphs were paired with nonsense words.
Again there was more transfer from a variable first list to a variable
second list than from a constant to a variable one. Variable lists were
more difficult, on the whole, for the Spanish speakers, perhaps
, GIBSON 327

because their native language contains highly- regular letter-sound

relationships.
A "set for diversity" may, therefore, facilitate transfer to learn-
ing of new letter-sound correspondences which contain variable
relationships. But many questions about how -the code is learned
remain to be solved, - because the true units of the graphic code are
not necessarily single letters. While single-letter-sound relations in
English are indeed variable, at other levels of structure regularity
may be discovered.

Lower- and Higher-Order Units

For many years, linguists have been concerned with the question
of units in language. -That language has a hierarchical structure, with
units of different kinds arid levels, is generally accepted, though the
definition of the units h not easily reached. One criterion of a unit is
recodabilityconsistent mapping or translation to another code. If
such a criterion be granted, graphic units must parallel linguistic
units. The units of the writing system should be defined, in other
words, by mapping rules which link them to the speech code, at all
levels of structure.
What then are the true graphic units? What levels of units are
there? Exactly how are they mapped to linguistic units? In what
"Chunks" are they perceived? We must first try to answer these ques-
tions by a logical analysis of properties of the writing and speech
systems and.the correspondence's between them. Then we can look at
the behavior of skilled readers and see how units are processed during
reading. If the logical analysis of the correspondence rules is correct;
we should be able to predict what kinds of units are actually pro-
cessed and to check our predictions experimentally.
Common sense suggests that the unit. for reading is the single
grapheme, and that the reader proceeds sequentially from left to
right, letter by letter, across the. page. But we can assert at once and .

unequivocally that this picture is false. For the English language, the
single graphemes map consistently into speech only as morphemes
that is, the names of -the letters of the alphabet. It is possible, of
course, to name letters sequentially across aline of print ("spell out"
1

328 THEORETICAL MODELS AND PROCESSES OF READING

a word), but that is not the goal of a skilled reader, nor is it what he
does. Dodge (11) showed, nearly 60 years ago, that perception occurs
in reading only during fixations, and not at all during the saccadic
jumps from one fixation to the next. With a fast tachistoscopic ex-
posure, a skilled reader can perceive four unconnected letters, a very
long word, and four or more words if they form a sentence .(1 2).
Even- first graders can read three-letter words exposed for only 40
milliseconds, too short a time for sequential eye movements to occur.
Broadbent (13) has pointed out that speech, although-it consists
of a temporal sequence of stimuli, is responded to at the end of a
sequence. That is, it is normal for a whole sequence to be delivered
before a response is made. For instance, the sentence "Would you
give me your ?" might end with any of a large number of
words, such as "name" or "wailer or "wife." The response depends
on the total message. The fact that the component stimuli for speech
r.c1 reading are spread over time does not mean that the phonemes
or letters or words -are processed one at- a time, with each stimulus
decoded to a separate response. The fact that o is pronounced differ-
ently in BOAT and BOMB is not a hideous peculiarity of English which
must consequently be reformed. The o is read -only in context and is
never responded to in isolation. It is part of a sequence which con-
tains constraints of- two kinds, one morphological and the other the
spelling patterns which are characteristic of English.
If any doubt remains as -to the unlikelihood of sequential pro-
cessing letter by letter, there is recent evidence of Newman (14) and
of Kolers (15) on sequential exposure of letters. When letters forming
a familiar word are exposed sequentially in the same place, it is al-
most impossible to read the word. With an exposure of 100 milli-
seconds per letter, words of six letters are read with only 20 percent
probability of accuracy; and with an exposure of 375 milliseconds
per letter, the probability is still well under 100 percent. But that is
more than 2 seconds to perceive-a short, well-known. word! We can
conclude that, however graphemes are processed perceptually in
reading, it is not a letter-by-letter sequence of acts.
If the single grapheme does'not map consistently to a phoneme,
and furthermore, if perception normally takes in bigger "chunks"
of graphic stimuli in a single fixation, what are the smallest graphic
units consistently coded into phonemic patterns? Must they be whole
GIBSON 329

words Are there different levels of units? Are they achieved at dif-
ferent stages of development?

Spelling Patterns
It is my belief that the smallest component units in written
English are spelling patterns (16). By a spelling pattern, I mean a
cluster of graphemes in a given environment which has an invariant
pronunciation according to the rules of English. These rules are the
regularities which appear when, for instance, any vowel or consonant
or cluster is shown to correspond with-a given prounuciation in an
initial, medial, or final position in the spelling of a word. This kind
of regularity is not merely "frequency" (bigram frequency,- rrigram
frequency, and so on), for it implies that frequency counts are rele-
vant for establishing rules only if the right units and the right re-
lationships are counted. The relevant graphic unit is a functional
unit of one or more_ in a given position .:111in the word,
which is in correspondence with a specified pronunci .tion (17).
If potential regularities exist within wordsthe spelling pat-
terns that occur in regular correspondence with speech patterns
one may hypothesize that these correspondences have been assim-
ilated by the skilledreader of English (whether or not be can
verbalize thezules) aird have the effect of organizing units for percep-
tion. It follows that strings of- letters which are generated by the
rules will be perceived more easily than ones which are not, even
when they are. unfamiliar words or not words at all.
Several experiments testing this prediction were performed by
Gibson, Pick, Osser, and aammond (18). The basic design was to
compare the perceptibility (with a very short tachistoscopic ex-
posure) of two sets of letter-strings, all nonsense or pseudo words,
which differed in their spelling-to-sound correlation. One list, called
the "pronounceable" list, contained words with a high spelling-to-
sound correlation. Each of them had an initial consonant-spelling
with a single, regular pronunciation; a final consonant-spelling hav-
ing a single regular pronunciation; and a vowel-spelling, placed be-
tween them, having a single regular pronunciation when it follows
and is followed by the given initial and final consonant spellings,
repectivelyfor example, cL/tt/c.x. The words in the second list,
330 THEORETICAL MODELS AND PROCESSES OF READING

called the "unpronounceable" list, had a low spelling-to-sound cor-

relation. They were constructed from the words in the first list by
reversing the initial and final consonant spellings. The medial vowel
spelling was not changed. For example, GLURCK became CKURGL.
There were 25 such pseudo words in each list, varying in length
from four to eight letters. The pronunciability of the resulting lists
was validated in two ways, first by ratings, and second by obtaining
the number of variations when the pseudo words were actually pro-
nounced.
The words were projected on a screen in random order, -in five
successive presentations with an exposure time beginning at 50 milli-
seconds and progressing up to 250 milliseconds. The subjects (college
students) were instructed to write each word as it was projected. The
mean percentage of pronounceable words correctly perceived was
consistently and significantly greater at all exposure times.
The experiment was 7.-.tcr repeated with the same material but
a differentlifdgment. After the pseudo word was exposed, it was
followed by a multiple-choice list of four items, one of -the correct
one and the other three the most common errors produced in the
previous experiment. The subject chose the word he thought he had
seen from the choice list and recorded a number (its order in the
list). Again the mean of pronounceable pseudo words correctly- per-
ceisied significantly exceeded that of their unpronouncable counter-
parts. We condude from these experiments that skilled readers more
easily perceive as a unit pseudo words which follow the rules of
English 'spelling-to-sound correspondence; that spelling patterns
which have invariant relations to sound patterns function as a unit,
thus facilitating the decoding process.
In another experiment, Gibson, Osser, and Pick (19) studied the
development of perception of grapheme- phoneme correspondences.
We wanted to know how early, in learning to read, children begin
to respond to spelling-patterns as units. The experiment was de-
signed to compare children at the end of the first grade and at the
end of the third grade in ability to recognize familiar three-letter
words, pronounceable trigrams, and unpronounceable trigrams. The
three-letter words were taken from the first-grade reading list; each
word chosen could be rearranged into a meaningless but pronounce-
able trigram and a meaningless and unpronounceable one (for ex-
GIBSON 381

ample, RAN, NAR, RNA). Some longer pseudo words (four and five
letters) taken from the previous experiments were included as.well.
The words and pseudo words were exposed tachistoscopically to in-
dividual children, who were required to spell them orally. The
-first-graders read (spelled out) most accurately the familiar three-
letter words, but read the pronounceable trigrams significantly better
than the unpronounceable ones. The longer pseudo words were sel-
dom read accurately and were not differentiated by- pronunciability.
The third-grade girls read all three-letter combinations with high
and about equal accuracy, but differentiated the longer pseudo
words; that is, the pronounceable four- and five-letter pseudo
words were more often perceived correctly than their unpronounce-
able counterparts.
These results suggest that a child in the first stages of reading
skill typically reads in short units, but has already generalized cer-
tain- regularities- of- spelling-to-sound correspondence, so -that three-
letter pseudo words which fit the rules are more easily read as units.
As skill develops, span increases, and a similar difference can be
observed for longer items. The longer items involve more complex
conditional rules and longer dusters, so that the generalizations must
increase in complexity. The fact that a child can begin very early
to perceive regularities of correspondence between the printed and
spoken patterns, and transfer them to the reading of unfamiliar
items as units, suggests that the opportunities for discovering the
correspondences between patterns might well be enhanced in pro-
graming reading materials.
I have referred several times to levels of units. The last experi-
ment showed that the size and complexity of the spelling patterns
which can be perceived as units increase with development of read-
ing skill. The other levels of structure, both syntactic and semantic,
contain units as large as and larger than the word, and that percep-
tion of skilled readers will be found, in suitable experiments, to be a
function of these factors is almost axiomatic. As yet we have little
direct evidence better than Cattell's original discovery (12) that
when words are structured into a sentence, more letters can be ac-
curately perceived "at a glance." Developmental studies of percep-
tual "chunking" in relation to structural complexity may be very
instructive.
332 THEORETICAL MODELS AND PROCESSES OF READING

Where does meaning come in? Within the immediate span* of

visual perception, meaning is less effective in structuring written
material than good spelling-to-sound correspondence, as Gibson,
Bishop; Schiff, and Smith (20) have shown. Real words which are
both meaningful and, as strings of letters, structured in accordance
with English spelling patterns are more easily perceived than non-
word pronounceable strings of letters; but the latter are more easily
perceived than meaningful but unpronounceable letter-strings (for
example, sim is perceived accurately, with tachistoscopic exposure,
faster than IBM). The- role of meaning in the visual perception of
words probably increases as longer strings of words (more than one)
are dealt with. A sentence has two kinds of constraint, semantic and
syntactic, which make it intelligible (easily heard) and memorable
(21). It is important that the child develop reading habits which
utilize all the types of constraint present in the stimulus, since they
constitute structure- and are, therefore, unit - formers.- The skills
which the child should acquire in reading are habits of utilizing the
constraints in letter strings (the spelling and morphemic patterns)
and in word strings (the syntactic and semantic patterns). We could
go on to consider still superordinate ones, perhaps, but the problem
of the unit, of levels of units, and mapping rules from writing to
speech has just begun to be explored with experimental techniques.
Further research on the definition and processing of units should
lead to new insights about the nature of reading skill and its attain-
ment.

Summary
Reading begins with the child's acquisition of spoken language.
Later he learns to differentiate the graphic symbols from one an-
other and to decode these to familiar speech sounds. As he learns to
decode, he must progressively utilize the structural constraints which
are built into it in order to attain the skilled performance which
is characterized by processing of higher-order unitsthe spelling and
morphological patterns of the language.
Because of my firm conviction that good pedagogy is based on
a deep understanding of the discipline to be taught and the nature
of the learning process involved, I have tried to show that the psy-
GIBSON 333

chology of reading can benefit from a program of theoretical analysis

and experiment. An analysis of the reading taskdiscriminatory and
decoding aspects as well as the semantic and syntactical aspectstells
us what must be learned. An analysis of the learning process tells us
how. The consideration of formal' instruction comes only after
these steps, and its precepts should follow from them.

REFERENCES

1. See C. C. Fries, Linguistics and Reading (Holt, Rinehart, and Winston, New
York, 1963), for an excellent chapter on past practice and theory in the
teaching of reading.
2. In 1959, Cornell University was awarded a grant for a Basic Research Project
on Reading by the Cooperative Research Program of the Office of Education,
U.S. Department of Health, Education, and Welfare. Most of the work
reported in this article was supported by this grant. The Office of Education
has recently organized "Project Literacy," which will promote research on
reading in a number of laboratories, as -well as encourage mutual under-
standing between experimentalists and teachers of, reading.
3'. E. J. Gibson, J. J. Gibson, A. D. Pick, H. Osser, J. Comp. Physiol. Psychol.
55, 897 (1962).
4. E. J. Gibson, H. Osser, W. Schiff, J. Smith, in A Basic Research Program on
Reading, Final Report on Cooperative Research Project No. 639 to the
Office of Education, Department of Health, Education, and Welfare.
5. The method was greatly influenced by the analysis of distinctive features of
phonems by R. Jacobson and M. Halle, presented in Fundamentals of Lan-
guage (Mouton, The Hague, 1956). A table of 12 features, each in binary
opposition, yields a unique pattern fol. all phonemes, so that any one is
distinguishable from any other by its pattern of attributes. A pair of pho-
nemes may differ by any number of features, the minimal distinction being
one feature opposition. The features must be invariant under certain trans-
formations and essentially relational, so as to remain distinctive over a wide
range of speakers, intonations, and so on.
6. A. D. Pick, J. Exp. Psychol., in press.
7. C. H. Bishop, J. Verbal Learning Verbal Behay. 3, 215 (1964).
8. Current advocates of a revised alphabet who emphasize the low letter-sound
correspondence in English are Sir James Pitman and John A. Downing.
Pitman's revised alphabet, called the Initial Teaching Alphabet, consists of
43 characters: some traditional and some new. It is designed for instruction
of the beginning reader, who later transfers to traditional English spelling.
See I. J. Pitman, J. Roy. Soc. Arts 109, 149 (1961); J. A. Downing, Brit.- J.
Educ. Psychol. 32, 166 (1962); , "Experiments with Pitman's initial
334 THEORETICAL MODELS AND PROCESSES OF READING

teaching alphabet in British schools," paper presented at the Eighth Annual

Conference of International Reading Association, Miami, Fla., May 1963.
9. L Bloomfield, Elem. Engl. Rev. 19, 125, 183 (1942).
10. See research reports of H. Levin and J. Watson, and H. Levin, E. Baum,
and S. Bostwick, in A Basic Research Program on Reading (see 4).
11. R. Dodge, Psycho!. Bull. 2, 193 (1905).
12. -J. McK. Cattell, Phil. Studies 2, 635 (1885).
13. D. E. Broadbent, Perception and Communication (Pergamon, New York,
1958).
14. E. Newman, Am. J. Psychol., in press.
15. P. A. Kolers and M. T. Katzman, paper presented before the Psychonomic
Society, Aug. 1963, Bryn Mawr, Pa.
16. Spelling patterns in English have been discussed by C. C. Fries in Linguistics
and Reading (Holt, Rinehart, and Winston, New York, 1963), p. 169 ff.
C. F. Hockett, Basic Research Program on Reading (see 4), 'has made
an analysis of English graphic monosyllables which presents regularities of
spelling patterns in relation to pronunciation. This study was continued by
R. Venezky (thesis, Cornell Univ., 1962), who wrote a computer program for
obtaining the regularities of English spelling-to-sound correspondence. The
data obtained by means of the computer permit one to look up any vowel
or consonant cluster of up to five letters and find its pronunciation in
initial, medial, and final positions in a word. Letter environments as well
have now been included in the analysis. See also R. H. Weir, Formulation
of Grapheme-Phoneme Correspondence Rules to Aid in the Teaching of
Reading, Report on Cooperative Research Project No. 5.039 to the Office of
Education, Department of Health, Education and Welfare.
17. For example, the cluster Cu may lawfully be pronounced as an F at the end
of a word, but never at the beginning. The vowel cluster non, pronounced
/A/ ( /ej /), may occur in initial, medial, and final positions, and does so with
nearly equal distribution. These cases account for all but two occurrences of
the cluster in English orthography. A good example of regularity influenced
by environment is [c] in a medial position before I plus a vowel. It is always
pronounced /S/ (social, ancient, judicious).
18. E. J. Gibson, A. D. Pick, H. Osser, M. Hammond, Am. J. Psycho!, 75, 554
(1962).
19. E. J. Gibson, H. Osser, A. D. Pick, J. Verbal Learning Verbal Behay. 2, 142
(1963).
20. E. J. Gibson, C. H. Bishop, W. Schiff, J. Sinith, J. Exp. Psycho!., 67, 173
(1964).
21. G. A. Miller and S. Isard, J. Verbal Learning Verbal Behay. 2, 217 (1963);
also L. E. Marks and G. A. Miller, ibid. 3, 1 (1964).