You are on page 1of 16

Psycholinguistics and Literature

Author(s): Eugene R. Kintgen


Source: College English , Mar., 1978, Vol. 39, No. 7 (Mar., 1978), pp. 755-769
Published by: National Council of Teachers of English

Stable URL: https://www.jstor.org/stable/375698

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/375698?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

National Council of Teachers of English is collaborating with JSTOR to digitize, preserve and
extend access to College English

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
College
Vol. 39, No. 7 * MARCH 1978 g"

English
EUGENE R. KINTGEN

Psycholinguistics and Literature

ALTHOUGH MY TITLE mentions psycholinguistics and not cognition, I would


begin by arguing for what might be called a 'cognitive' approach to literary
'cognitive' as it is used in cognitive psychology: "what has always been co
to perception, learning, and-thinking ... a concern with the process of kno
an empirical approach to epistemology."1 I think it would be useful to stu
empirically-how readers perceive literary works, how they come to know t
and how they think about them. The primary emphasis of such an approac
the how: what are the actual processes readers perform in understanding a
Do all readers employ exactly the same procedures, or is there a pool of o
tions from which each reader selects a subset? How-and how much-does the
text affect what the reader does? A hardly less important emphasis is on the
empirical: autobiographical anecdotes about encounters with literature may be
interesting in suggesting directions for research, as does introspection, but neith
in itself provides the kind of information necessary for rigorous inquiry, simp
because there is no guarantee that the processes of understanding (as opposed
the results of understanding) are open to introspection. Who can tell by intro
spection what it is about a sound that allows it to be identified as a /p/ rather than
a /k/? Finally, there is an emphasis on the plural in readers: what should
studied is not how a particular reader on a particular occasion approached a pa
ticular text, but rather how that series of events is related to similar events involv-
ing different readers, different texts, and different times. The aim of a cognit
approch to literature, then, would be to specify what kinds of things happen
when readers read.
This is clearly no modest undertaking, even when stated in such imprecise

'Barry F. Anderson, Cognitive Psychology (New York: Academic Press, 1975), pp. xi-xii.
Eugene R. Kintgen is an associate professor of English at Indiana University, long interested in
stylistics, who has fairly recently become interested in the processes by which a reader per-
ceives a literary text and constructs a representation of it in his/her mind.

755

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
756 COLLEGE ENGLISHI

terms,
terms,and
and
I would
I would
hesitate
hesitate
to attempt
to attempt
it, were it,
it not
were
for it
a very
not obvious
for a very
and very
obvious
simple
simpleobservation
observation
thatthat
everybody
everybody
knows but
knows
tries but
to ignore,
tries at
toleast
ignore,
in practice:
at least in
the
therepresentation
representation of a of
literary
a literary
work in
work
a reader's
in a mind
reader's
is different
mind isfrom
different
the repre-
from th
sentation of the literary work in the outside world. The external text consists of
marks on a page, but the internal representation must be in the form of bio-
chemical residues, electrical impulses, whatever-the exact terms are not clear,
nor are they important for my purposes. What is important is that the most basic
process constitutive of literature (in any sense worth discussing) is one of transla-
tion wxhereby some set of external stimuli (to put it baldly) is translated into or
given an internal representation. Literary studies have generally assumed that these
internal representations are veridical and therefore constant across readers: what
happens to one reader happens, "within a range of inconsequential variation," to
any other reader.2 When the range of variation becomes consequential, it is pos-
sible to discover where various readers have made errors or allowed their pre-
conceptions to lead them astray.3 Of course, readers may respond differently to
the same work, but the difference is in the reaction, not in the internalized text
being reacted to. But we have only to turn to everyday experience to realize that
different people translate differently; if thy did not, there would be only one
translation of The lliad, Beowulf, and so forth. And if this seems to be exploiting
an equivocation in my use of 'translation', consider William F. Battig's comments
on "Within-Individual Differences in 'Cognitive' Processes": "The most con-
sistent finding throughout the research reported herein on within-individual dif-
ferences is that wherever we have looked for them, we have never failed to find
them in considerable magnitude."4 This suggests that those who believe that all
readers read the same way are likely to be mistaken, and that it would be worth-
while to study just how readers do read.
I am using reader as a handy generic term for 'apprehender of literature',
covering the same semantic ground as Holland's longer and fancier 'literant'.5
This does not mean that I propose to discuss the mechanics of the reading process,
though that is in itself an interesting if baffling topic; instead I would like to con-
sider the process (es) by which readers perceive and manipulate the linguistic
aspects of the text to arrive at a final conception of the work as a w^hole. And this
brings me back to mv title, for the findings of psycholinguistics, particularly that
branch of psycholinguistics which studies how people perceive and understand
language, can suggest useful directions for such research. The aim of this branch
of psycholinguistics is to construct a truly mentalistic theory of linguistic ability,
a theory specifying not only what a language user must know to understand

2Stanley Fish, "Literature in the Reader: Affective Stvlistics," NLH 2 (1970), rptd. in Self-
Consuming Artifacts (Los Angeles: University of California Press, 1972), p. 410. Fish no longer
agrees with this view (see "Interpreting the Variorum," Cl, 2 [19761, 465-85), and even in the
original statement he was referring to 'informed readers'. I use the quote because it expresses a
common sentiment.
3See I. A. Richards, Practical Criticism (New York: Harcourt, Brace, & World, 1939), for a
well known example.
4In Inforniation Processing and Cognition: The Loyola Symposium, ed. Robert L. Sos
(Hillsdale, N.J.: Lawrence Erlbaum, 1975), pp. 224-25.
5Norman Holland, "The New Paradigm: Subjective or Transactive?" NLH, 7 (1976), 336.

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
Psycholinguistics and Literature 757

language-competence-but
language-competence-but also also
how how
that knowledge
that knowledge
is used-performance.
is used-performance.
Happily
Happilyfor
forour
our
purposes,
purposes,
it isiteasier
is easier
to study
to study
the perception
the perception
of language
of than
language
its than its
production,
production,sosomuch
much
of the
of the
research
research
in this
infield
thishas
field
concerned
has concerned
such questions
such as
questions as
how
how aahearer
hearerperceives
perceives
the the
phonological
phonological
stratum
stratum
of language,
of language,
how syntax
how is syntax is
processed,
processed,how
howthethe
meaning
meaning
of a of
sentence
a sentence
or a paragraph
or a paragraph
is represented
is represented
mentally, mentally,
how
how (and
(andhow
howwell)
well)thethe
phonological,
phonological,
syntactic,
syntactic,
and semantic
and semantic
aspects of
aspects
lan- of lan-
guage
guageare
areremembered,
remembered, andand
howhow
the information
the information
presentpresent
in sentences
in sentences
is used in is used in
problem
problemsolving
solving
or or
related
related
to other
to other
information.
information.
Naturally,
Naturally,most
mostof of
thethe
psycholinguistic
psycholinguistic
conclusions-if
conclusions-if
there can
there
be said
cantobe
be said to be
any
any 'conclusions'
'conclusions'at at
thisthis
stage
stage
of research-cannot
of research-cannot
be applied
be applied
directly directly
to the to the
perception
perceptionofof literature.
literature.
OneOne
reason
reason
for this
foristhis
thatismost
thatpsycholinguistic
most psycholinguistic
experi- experi-
ments
mentsrequire
require thethe
subject
subject
to respond
to respond
as quickly
as quickly
and accurately
and accurately
as possible
astopossible
the to the
stimulus
stimuluspresented,
presented, unlike
unlike
the the
normal
normal
reading
reading
situation
situation
in whichinthe
which
readerthehasreader has
leisure
leisuretotocontemplate
contemplate whatwhat
is being
is being
read (though
read (though
this 'normal
this 'normal
reading situation'
reading situation'
may
may bebemore
more ananidealization
idealization
thanthan
an everyday
an everyday
occurrence).
occurrence).
Another Another
reason is reason is
that
that these
thesestudies
studiesareare
conducted
conductedin a non-aesthetic
in a non-aesthetic
setting:setting:
the experiments
the experiments
em- em-
phasize
phasizespeed
speedandandaccuracy,
accuracy,not not
truthtruth
and beauty,
and beauty,
and do and
not encourage
do not encourage
the dif- the dif-
ferent
ferentkinds
kindsofofreading
readingeacheach
of usofis us
capable
is capable
of, especially
of, especially
the two the
Ingarden
two calls
Ingarden calls
the
the 'aesthetic'
'aesthetic'andandthethe
'preaesthetic'.6
'preaesthetic'.6 Thus Thus
there may
there bemay
special
beprocesses
special processes
de- de-
pendent
pendenton,on,ororconstitutive
constitutive of, the
of, leisurely,
the leisurely,
aesthetic
aesthetic
perception
perception
of literature
of literature
which
whichpsycholinguistic
psycholinguistic experiments
experiments will overlook.
will overlook.
Even so,Even
the findings
so, the findings
of the of the
psycholinguists
psycholinguists cancan
be be
extremely
extremely useful useful
for research:
for research:
they canthey
provide,
can meta-
provide, meta-
phorically, directions for research, and at times, can suggest the actual processes
in literary understanding. Equally important, they can suggest what cannot be
happening in reading.
The most basic level of linguistic perception is the phonological, and though
this seems the most remote from literature, it is here that we find the clearest
demonstration of the disparity between external stimuli and their internal repre-
sentation, for there is a startling lack of isomorphism between acoustic cues-
what can be measured with instruments-and perceived speech sounds.7 The
same cue may be perceived as two different speech sounds in different contexts:
a burst of energy at 1400 cps. is perceived as a /p/ when followed by the cues
for /i/ or /u/, but as a /k/ when followed by the cues for /a/. Quite different
cues may be heard as the same sound, the second formant frequencies associated

6See The Cognition of the Literary Work of Art, trans. by Ruth Ann Crowley and Kenneth
R. Olson (Evanston: Northwestern University Press, 1973), especially chapter 4, "Varieties of
the Cognition of the Literary Work of Art."
7For good introductions to this field, see Helen S. Cairns and Charles E. Cairns, "Phonological
Performance," Psycholinguistics, A Cognitive View of Language (New York: Holt, Rinehart
& Winston, 1976), pp. 112-44; A. M. Liberman, "The Grammars of Speech and Language,"
Cognitive Psychology, 1 (1970), 301-23; Michael Studdert-Kennedy, "The Perception of
Speech," Current Trends in Linguistics, XII, ed. Thomas A. Sebeok. (The Hague: Mouton,
1974), 2349-85; Kenneth N. Stevens and Arthur S. House, "Speech Perception," Foundations of
Modern Auditory Theory, II, ed. J. V. Tobias (New York: Academic Press, 1973), 1-62;
Robert A. Cole and Brian Scott, "Toward a Theory of Speech Perception," Psychological Re-
view, 8 (1974), 348-74; and C. J. Darwin, "The Perception of Speech," Handbook of Percep-
tion, VII: Speech and Language, ed. Edward C. Carterette and Morton P. Friedman (New
York: Academic Press, 1976), 175-226.

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
758 COLLEGE ENGLISH

with
with/b/,
/b/,for
for
instance,
instance,
ranging
ranging
from about
from600
about
to 2700
600cps.
to depending
2700 cps. on
depending
the fol-
lowing
lowingsound.
sound.Acoustic
Acoustic
cues cues
may exist
mayindependently
exist independently
even wheneventhe speech
when
sounds
soundsthey
theycuecue
do not;
do not;
if theifvocalic
the vocalic
segment segment
of a stop of
consonant-vowel
a stop consonan se-
quence
quenceisis
progressively
progressively
shortened,
shortened,
there is
there
neveris
a time
neverwhen
a time
the stop
whenalone
the is st
heard.
heard.Instead,
Instead,
thethe
vowel
vowel
is heard
is heard
to get to
shorter
get shorter
until it disappears,
until it disappears,
leaving be-
hind
hindonly
only
a non-linguistic
a non-linguisticchirp,chirp,
even though
even though
the cue that
the was
cueperceived
that was as perce
a
stop
stopwhen
when followed
followed by aby
vowxvel
a vowxvel
is still is
intact.
still To
intact.
complete
To the
complete
symmetry,the some-
symm
times
timesthethelack
lackof of
a cue
a cue
can be
canperceived
be perceived
as a speech
as asound:
speech
splice
sound:
a silent
splice
interval
a sile
into
intoa atape
tapeof of
slitslit
between
between
the s the
and the
s and I and
the
hearers
I and perceive
hearerssplit,
perceive
not slit
split,
with n
a silent interval.8
The results of phonological studies thus emphasize the one basic point that
there is no simple relationship between an acoustic cue and its perception as a
language sound. Consequently there must be a system or grammar relating acous-
tic cues to perceptual units. The value of this research for literary studies is pri-
marily metaphorical, providing a clear and simple demonstration that one can-
not hope to account for perceptual categories by studying merely physical cate-
gories. In literary terms, this means that attention to the text itself, no matter
how detailed and sophisticated, does not ipso facto provide any information
about how a reader actually perceives the text. Textual units are not necessarily
mental ones, and while the two may in some instances correspond, the extent of
that correspondence is an empirical question, not something to be stipulated in
advance. It might be noted in passing that much of the traditional literary crit-
ics' dissatisfaction with stylistic, especially linguistic stylistic, studies, stems from
just this problem: the textual units identified and studied by the stylisticians have
no relation to the perceived units the critics habitually use.9 Thus it is unlikely
that any advances in the techniques of describing texts (say, text grammara) will
make stylistics any more interesting or useful to critics: what is needed instead
is some way to relate the textual categories uncovered by textual analysis to the
perceptual categories used by the critics.
When we turn from the perception of sounds to the perception of larger lin-
guistic units, we came upon studies whose relevance to literary perception is
more straightforward. Since George A. Miller's famous 1956 paper, "The Mag-
ical Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Pro-
cessing Information," it has been accepted as a working hypothesis that only
seven items can be retained in active, short term memory; if more items are to be
remembered, they must be either recoded into larger units, rehearsed aloud or
silently (as repeating a telephone number), or entered into long term memory.'0
Since many sentences are longer than seven words, and thus cannot be retained
fully in short term memory while they are being processed, there must be some
way to segment them and recode the segments into higher-level units. The most

8A. M. Liberman, F. S. Cooper, K. S. Harris, P. F. MacNeilage, and M. Studdert-Kennedy,


"Some Observations on a Model for Speech Perception," Models for the Perception of Speech
and Visual Form, ed. W. Wathen-Dunn (Cambridge, Mass.: M.I.T. Press, 1967), pp. 68-88.
9See, for example, Teun van Dijk's Some Aspects of Text Grammars: A Study in Theoretical
Linguistics and Poetics (The Hague, Mouton, 1972).
'0Psychological Review, 63 (1956), 81-97.

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
Psycholinguistics and Literature 759

obvious
obvious candidates
candidatesfor
forsegmentation
segmentationwould
would
naturally
naturally
be the
be the
syntactically
syntactically
de- de-
fined
fined constituents
constituentsofofthe
thesentences
sentences
themselves.
themselves.
Peter
Peter
Herriott
Herriott
nicely
nicely
illustrated
illustrated
the
the utility
utilityof
ofsuch
suchsyntactic
syntactic structure
structure eveneven
in the
in the
absence
absence
of referential
of referential
meaningmeaning
by
by combining
combiningnonsense
nonsensestemsstems wvith
wvith English
English inflections
inflectionsand and
function
function
wordswords
to to
form
form strings
stringswith
withthethestructure
structure of of
English
English phrases
phrases
but but
no lexical
no lexical
meaning
meaning
(sim- (sim-
ilar
ilar to
to "The
"TheJabberwocky"),
Jabberwocky"), strings
strings thethesamesamelength
length
without
without
the structure
the structureof of
phrases,
phrases, andandstrings
stringsthethesame
same length
length in in
which
which the the
order
order
of items
of items
was random-
was random-
ized.
ized. Not
Not surprisingly,
surprisingly,the thestrings
strings were
wereeasiest
easiest
to memorize
to memorize whenwhenthey they
corre- corre-
sponded
sponded mostmostclosely
closelytotoEnglish
English phrases,
phrases, suggesting
suggesting thatthat
subjects
subjects
process
process
non- non-
sense
sense most
mostefficiently
efficientlywhen
when they
they cancan
organize
organize it into
it into
the the
kindskinds
of phrases
of phrases
they they
expect
expect inin English.1"
English.1"
But
But whether
whethersubjects
subjectsalways
always segment
segment meaningful
meaningful sentences
sentences
into into
the constitu-
the constitu-
ents
ents defined
definedby bylinguistic
linguistic analysis
analysis is another
is another question.
question.
N. F.N.Johnson
F. Johnson
required
required
subjects
subjects totomemorize
memorizesentences
sentences andand
thenthenrepeat
repeatthem them
later.
later.
Naturally,
Naturally,
not all
not theall the
sentences
sentences were
wereremembered
remembered perfectly,
perfectly, andandby bystudying
studyingthe the
Transitional
Transitional
ErrorError
Probability-the
Probability-theprobability
probability that
thata given
a given wordword is misremembered
is misremembered whenwhenthe pre-
the pre-
vious
vious word
wordisisremembered
remembered correctly-it
correctly-it is possible
is possible to gauge
to gauge
the cohesiveness
the cohesivenessof of
multiword
multiwordunits.
units.Cohesive
Cohesive units
units will
will
be be
those
those
with with
low low
TEPsTEPs
between
between
the mem-
the mem-
bers,
bers, and
and aalarge
largeTEP
TEPwill
willsignal
signal
thethe
boundary
boundary
between
between
two two
units.
units.
Johnson
Johnson
found
found that
thatthetheunits
unitsdefined
definedbyby
TEPs
TEPstend
tend
to correspond
to correspond to the
to surface
the surface
struc-struc-
ture
ture as
as predicted
predictedby byphrase
phrase
structure
structurerules.12
rules.12
When
When
a slightly
a slightly
different
different
version
version
of of
the
the task
task isisused,
used,however,
however,the
the
results
resultsareare
somewhat
somewhat
different.
different.
R. A.R.Kennedy
A. Kennedy
and and
A. L. Wilkes had subjects memorize sentences and then decide whether two
words presented as a test occurred in the same order in the original sentence.
By submitting subjects' average answering time to hierarchical clustering anal-
ysis, they provided another kind of phrase structure profile for the sentences,
and concluded that the structures discovered by the clustering analysis "do not
follow in any detail the surface structure of the sentence types.""3 W. J. M.
Levelt presented subjects with sentences masked by noise and asked them to
write down what they had heard. His analysis combined transitional error prob-
abilities with hierarchical clustering analysis, and concluded that the major units
so identified often correspond to linguistically defined constructions, but that the
smaller units often do not.'4
One of the factors that apparently prevents agreement between linguistic anal-
ysis and subjective segmentation is the length of the constituents. Edwin Martin
pioneered a remarkably simple approach to what he called "subjective phrase
structure"-he asked subjects to indicate where they thought sentences should
be segmented-and concluded that "the conventional constituent structure anal-

"Peter Herriott, "Phrase Units and the Recall of Grammatically Structured Nonsense," The
British Journal of Psychology, 58 (1967), 237-42.
12Neal F. Johnson, "The Psychological Reality of Phrase-Structure Rules," The Journal of
Verbal Learning and Verbal Behavior, 4 (1965), 469-75.
'3R. A. Kennedy and A. L. Wilkes, "Functional Structure in Sentences: A Performance
Analysis," The Quarterly Journal of Experimental Psychology, 23 (1971), 215-24.
'4W. J. M. Levelt, "Hierarchical Chunking in Sentence Processing," Perception and
Psychophysics, 8 (1970), 99-103.

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
760 COLLEGE ENGLISH

ysis is not
not generally
generally reliable
reliable or
or even
even valid
validfor
forthe
theaverage
averagecollege
collegesophomore."1
sophomore.
Z. S. Bond
Bond and
and John
John Gray
Gray used
used the
the same
sameapproach
approachand andfound
foundthat
thatthe
theverb
ver
tended toto be
be grouped
grouped with
with the
the subject
subjectwhen
whenthethesubject
subjectwas
wasrelatively
relativelyshort,
short,
three words
words oror less,
less, but
but with
with the
the object
objectotherwise.16
otherwise.16J. J.E.E.Martin
Martinand
andhis
hisasso
ass
ciates allowed
allowed their
their subjects
subjects to
to rehearse
rehearsesentences
sentencesaloud
aloudbefore
beforeindicating
indicatingwher
wh
breaks should
should come,
come, thus
thus emphasizing
emphasizingthe thepossibility
possibilityofofphonological
phonologicalrecoding.
recodin
Their results
results were
were similar
similar toto Bond
Bond and
andGray's:
Gray's:segmentation
segmentationisismore
morelikely
likelyatat
constituent boundaries than within boundaries, but which constituent bounda-
ries are chosen depends on the length of the constituents. They suggest that the
phonological phrases subjects identify are subjectively real, the units actually
used in sentence processing; the linguistically defined surface structure and even
phonological structure are more abstract in this sense.17 Further evidence for the
reality of subjective phrase structure comes from an experiment by George Suci,
who demonstrated that subjects pause in different places when reading material
aloud; they learn material faster if it is presented to them in units defined by
their own pausing than if it is presented in units of the same size that do not
correspond to their own pause units, even if those units are determined by an-
other reader's pauses.18 Phonological phrases thus appear to be both subjectively
real and subjectively defined: what is a phrase (and thus useful) for one reader
may not be for another.
The results of these studies are by no means unambiguous, due in part to the
different methods used and the different demands made of the subjects; in par-
ticular, studies requiring subjects to remember something add another variable-
memory-to those which require only judging a presented stimulus. What they
all suggest, however, is that there is a subjective component in the basic segmen-
tation of incoming stimuli, so that even in short sentences-and most of the sen-
tences used in these studies were less than ten words long-the constituents iden-
tified by linguistic analysis do not necessarily correspond to the units people use
in processing them. Another conclusion that can be drawn from all these studies
(but emphasized only by Suci) is that there is a certain amount of variation
among subjects, all of whom segment longer linguistic constructions to process
them, but not all of whom segment them in the same way. For literary studies
the moral is clear: if we cannot assume that people agree either with each other
or with linguistic analysis in 'the segmentation of such seemingly well structured
entities as sentences, there is no reason to assume that readers will agree with
each other or with any formal analysis in their segmentation of literary works.
In fact, we should expect just the opposite, and (without pushing Suci's conclu-

'5Edwin Martin. "Toward an Analysis of Subjective Phrase Structure," The. Psychological


Bulletin, 74 (1970), 153-66.
16Z. S. Bond and John Gray, "Subjective Phrase Structure: An Empirical Investigation," The
Journal of Psycholinguistic Research, 2 (1973), 259-66.
17J. E. Martin, Barbara Kolodziej, and Joseph Genay, "Segmentation of Sentences into
Phonological Phrases as a Function of Constituent Length," The Journal of Verbal Learning
and Verbal Behavior, 10 (1971), 226-33. Hereafter this journal will be abbreviated JVLVB.
'8G. J. Suci, "The Validity of 'the Pause as an Index of Units in Language," JVLVB, 6
(1967), 26-32.

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
Psycholinguistics and Literature 761

sions
sionstoo
toofar)
far)
might
mighteveneven
anticipate
anticipate
that thethat
units
the
oneunits
readeronefinds
reader
most finds
con- most con-
genial
genialwill
will
bebeuseless
useless
to another.19
to another.19
But
Butititmay
may well
well
be that
be that
hearers
hearers
who segment
who segment
sentencessentences
differently differently
finally put finally pu
them
themback
backtogether
together to form
to form
the same
the sentential
same sentential
units: after
units:
all, to
after
understand
all, to aunderstand
sentence
sentenceit it
is is
necessary
necessaryto reconstruct
to reconstruct
certain certain
basic relationships
basic relationships
(subject-verb,
(subject-verb
verb-object,
verb-object, modifier-head)
modifier-head) no matter
no matter
what pieces
whatonepieces
startsone
with.starts
And there
with.isAnd there
some
someevidence
evidence that
that
the the
clause
clause
is theis
salient
the salient
unit of sentence
unit of processing.
sentence processing.
David David
Caplan
Caplanpresented
presentedsubjects
subjects
with with
sentences
sentences
constructed
constructed
so that a test
so that
word awas
test word w
always
alwaysthethesame
same
number
numberof syllables
of syllables
from the
fromend,the
but end,
was either
but was
in the
either
final or
in the final
the
thepenultimate
penultimate clause
clause
(Now(Now
that artists
that artists
are working
are working
in oil, prints
in oil,
are rare
prints
and are rare an
Now
Nowthatthat
artists
artists
are are
working
working
fewerfewer
hours, hours,
oil prints
oilare
prints
rare). The
are test
rare).
wordThe test word
(oil in the examples) was recognized more quickly when it was in the final
clause, suggesting that comprehension works clause by clause, with the most re-
cent clause being the easiest to recall.20 Similarly, Robert J. Jarvella interrupted
spoken discourse and asked subjects to recall as much of what they had heard as
possible. The interruption always followed a seven word clause, but sometimes
this clause was an independent sentence and sometimes it was in the same sentence
as a previous six word clause. The final seven word clause was always well re-
called, but the prior six word clause was recalled much better when it was in the
same sentence as the final clause, presumably because after clauses are processed,
they are somehow retained until all the clauses for one sentence are available,
when they are all combined.21
Clauses and sentences, then, apparently have some kind of perceptual reality;
but how are the clauses themselves understood? Since the advent of transforma-
tional-generative grammar, with its insistence on two levels of linguistic descrip-
tion, an underlying level representing the meaning of a sentence and a surface
structure representing the phonological structure, this question is really two: first,
how is the surface structure of a clause determined; and second, how can that
surface structure be related to the appropriate deep structure? John Kimball has
provided an answer to the first question by providing "Seven Principles of Sur-
face Structure Parsing in Natural Language."22 This involves an internalized
phrase structure grammar which describes the allowable configurations for an
English clause and seven principles which relate a given sentence to the grammar.
The answer to the second question is more problematical. The first proposed
answer was the 'Derivational Theory of Complexity', which assumed that to
understand a sentence, a hearer 'undid' the transformations that had been applied
to the deep structure to produce the surface structure. Transformational com-
plexity thus lead to perceptual complexity: a sentence to which five transforma-
tions had applied should take more time to understand than one involving only

1'In S/Z, for instance, the segments Barthes chooses for comment are sometimes words,
sometimes phrases or clauses, sometimes sentences, and sometimes longer units.
20David Caplan, "Clause Boundaries and Recognition Latencies for Words in Sentences,"
Perception and Psychophysics, 12 (1976), 73-76.
2tRobert J. Jarvella, "Synitactic Processing of Connected Speech," JVLVB, 10 (1971), 409-
16.
22John Kimball, "Seven Principles of Surface Structure Parsing in Natural Language,"
Cognition, 2 (1973), 15-47.

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
762 COLLEGE ENGLISH

four.23
four.23 Counterexamples
Counterexamples were
were quickly
quickly
found,
found, and and
it additionally
it additionally
became
becc
that
that the
the DTCDTCwould
wouldbebe unwieldy
unwieldy in in
an an
actual
actual
linguistic
linguistic
situation,
situation,
wherewha
often
often processes
processesa asentence
sentence as as
it it
is being
is beingheard
heardso that
so that
a reply
a reply
can be
can
g
soon
soon asas the
thespeaker
speakerstops:stops: howhow would
would thethe
hearer
hearer
know,know,before
before
the end
theof
end
t
tence,
tence, what
whattransformations
transformations to to
'undo',
'undo',
andand
howhow couldcould
theythey
be undone
be undon
quic
enough?
The answer to both these questions now seems to be that the hearer uses a set
of heuristic strategies which allow him to guess directly at an underlying struc-
ture for a sentence while he is still hearing it. Thomas G. Bever has suggested how
such a set of heuristics, or 'perceptual strategies' would work.24 The hearer has
an internalized lexicon which provides not only semantic and form class informa-
tion about each word, but also tells what kinds of constructions it can occur in.
The verb adorn, for example, has to be followed by a direct object, while see
can take either a simple direct object or a sentential complement. Hearers use
this structural knowledge to predict what the structure of incoming strings will
be and then to segment them into clauses consisting of actor-action-object-modi-
fier sequences. When the verb is simple in the sense that it occurs in a limited
number of structures, such as intransitives (arrive) or transitives that require
nominal objects (fix), the prediction is also simple. When the verb can enter
into a large number of different structures, as with see or think, there is more un-
certainty about what might follow; the hearer then has to entertain a number of
hypotheses about the following structure and postpone segmentation until more
information is available. In either case, the first actor-action-object-modifier
sequence identified is taken to be the main clause unless there is some indication
of subordination, such as a subordinating adverb or conjunction; these subordinate
clauses are considered modifiers. These strategies suggest why The horse raced
past the barn uwvn is more difficult to comprehend than The horse that 'was raced
past the barn won. In the former, the first N-V-N sequence (the horse raced past
the barn) is tentatively identified as the main clause; when there is no way to
relate won to this sequence, the whole sentence must be reanalyzed. In the second
sentence, the relative clause is clearly marked by the subordinator that, which
prevents an incorrect initial segmentation.
Various other strategies, including semantic ones, are proposed to account for
other types of constructions such as noun phrases, and to guess at the semantic
relations between the constituents of clauses and sentences. But the basic insight
of the theory of perceptual strategies is that a hearer must reconstruct the
syntactic and semantic relations of the underlying structure of a sentence very
quickly on the basis of a surface which often does not explicitly represent them;

23For good reviews of this approach, see Judith Greene, Psycholinguistics: Choinsky and
Psychology (Harmondsworth: Penguin, 1972), pp. 107-16; Jerry A. Fodor, "Current Ap-
proaches to Syntax Recognition," The Perception of Language, ed. David L. Horton and
James L. Jenkins (Columbus: Charles E. Merrill Publishing, 1971), pp. 120-39; and Jerry A.
Fodor, T. G. Bever, and M. F. Garrett, The Psychology of Language (New York: McGraw
Hill, 1974), pp. 320-28.
24Thomas G. Bever, "The Interaction of Perception and Linguistic Structures: A Prelimi-
nary Investigation of Neo-Functionalism," Current Trends in Linguistics, XII, ed. Thomas A.
Sebeok, (The Hague: Mouton, 1974), 1159-1233.

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
Psycholinguistics and Literature 763

to do this he must utilize operations different from linguistic transformations. The


formal operations of linguistic grammars thus offer one way of relating under-
lying structures to surface structures, but they are not the operations a hearer
normally uses. W. C. Watt has offered a persuasive explanation of this disparity:
a linguistic grammar attempts to provide the most economical way of relating all
the deep structure/surface structure pairs in the language, where economy is
measured in terms of the number of operations in the entire grammar. The most
highly valued grammar will be the one with the fewest transformations, even if
the derivation of each individual sentence requires the application of a relatively
large number of transformations. For the hearer, however, the relevant economy
is the one relating deep and surface structures: the most highly valued grammar
for him is the one that requires the fewest operations (which might not be only
transformations) relating the deep and surface structure for each sentence, even
if that implies a larger total inventory of operations.25 Linguistic elegance is thus
no guarantee of psychological utility.
Now it seems that literary works, like sentences, have both an underlying and
a surface structure or representation, and that the two are not the same (for if
there were only one level to a work it would have to be the surface, and in that
case there would be no need for either teachers or critics). The underlying struc-
ture of a literary work is vastly more complex than that of a sentence, of course,
and there are probably several interrelated levels of representation rather than a
single one, but in any case there must be some way for the reader to penetrate
to these more abstract levels of the work from the surface he initially encounters,
and it seems not unlikely that literary strategies akin to Bever's perceptual
strategies are used for this purpose. Stanley Fish has suggested just this-that
readers use a series of 'interpretive strategies' to understand a literary text-but
he has not suggested what, precisely, these strategies consist of. In his earlier work
he argued that all readers used the same strategies, but more recently he has sug-
gested that different readers may use different strategies, and that only those using
the same strategies-members of the same interpretive community-will be able
to understand each other.26
This variation is precisely what we would expect from the psycholinguistic
analogue. Evidence from a number of studies (which I have cited in an earlier
article27) suggests that different hearers may use different strategies in under-
standing what they hear. An extremely interesting example of this is an experi-
ment by Lila R. and Henry Gleitman concerning the understanding of three-part
compounds such as lighthouse keeper and light housekeeper.28 They constructed
a series of three-part compounds reproducing the arrangements (and presumably
the relationships) found in well-known examples but using other words (e.g.,
bright, house, bird, and kill) to form such compounds as bright housebird, house-
bird kill, brightbird house, and so forth. They asked their subjects to provide
paraphrases for these compounds: for housebird kill they expected something

25W. C. Watt, "Mentalism in Linguistics II," Glossa, 8 (1974), 3-40.


26See footnote 2.
27Eugene R. Kintgen, "Reader Response and Stylistics," Style, 11 (1977), 1-18.
28Lila R. Gleitman and Henry Gleitman, Phrase and Paraphrase (New York: Norton, 1970).

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
764 COLLEGE ENGLISH

like
like'a'aslaughter
slaughter of birds
of birds
that live
thatinlive
the house';
in thefor house';
bright for
housebird,
bright 'ahousebi
colorful
bird
birdthat
that lives
livesin the
in the
house'.
house'.
The findings
The findings
were astounding:
were astounding:
when they presented
when th
their
theircompounds
compounds to graduate
to graduatestudents,
students,
13% of the13%responses
of the responses
deviated from deviate
what
they
theyexpected
expected by bylinguistic
linguistic
analogy,
analogy,
but when butthey
when
presented
they presented
the same compound
the sa
to
to aagroup
group of ofsecretaries,
secretaries, the deviation
the deviation
was 62%.wasThe62%.secretaries
The secretaries
continued to co
have
havea ahigh
high 'error'
'error'raterate
even even
after after
tutelage tutelage
and when andtheywhen
merely they
had merely
to chooseh
between
between paraphrases
paraphrases instead
instead
of supplying
of supplying
them. The them.
Gleitmans
The Gleitmans
concluded that con
the
theconception
conception of aof monolithic
a monolithiccompetence
competence
shared by shared
the speakers
by the of speakers
a dialect
is
is false;
false;ininitsits
place
place
theythey
proposepropose
a core competence
a core competence
shared by shared
all the speakers
by all thof
aa dialect
dialect surrounded
surrounded by aby penumbra
a penumbra
or semigrammar
or semigrammar
which onlywhichsome speaker
only s
have.
have.Different
Different speaker/hearers
speaker/hearers of the of
samethedialect
samemay dialect
thus have
mayquite
thusdifferen
have q
inventories
inventories of of
perceptual
perceptualstrategies.
strategies.
The
TheGleitman
Gleitman experiment
experiment is especially
is especially
interesting
interesting
because, unlike
because,
mostunlike
other
psycholinguistic
psycholinguistic experiments,
experiments, it requires
it requires
the creative
the use
creative
of linguistic
use of abilities:
linguistth
compounds
compounds usedusedcouldcould
all beall
understood
be understood
by analogyby toanalogy
well knownto well
patterns,
known bu
they
theythemselves
themselves were were
unique.
unique.
The results
The results
support support
Fish's postulation
Fish's postulation
of interpre-
tive
tivecommunities,
communities, and and
emphasize
emphasize
how muchhowvariability
much variability
among individuals
amongeithe ind
in
in linguistic
linguistic ability
abilityor the
or use
theofuse
that
ofability
that isability
uncoveredis uncovered
by a slight complication
by a sligh
of
of the
thelinguistic
linguistic materials
materials
used. used.
This would
This suggest
would thatsuggest
whenthatthe material
when the is a
complex
complex as as
literature,
literature,
we should
we should
expect expect
a wide variation
a wide variation
between individua
betwe
readers indeed.
Another similarity between the Gleitman's compounds and literature is that
many of them lacked any referent in the real world: who has ever seen a house
birdfoot, a house eatbird, or a wash housebird? To understand these compounds,
the subjects had to rely entirely on their linguistic form and ignore the real
world. This is quite different from what people normally do: most language use
is transparent in that language is used to convey information of some sort (not
always or solely about the outside world), and in such circumstances most people
abstract the information in the message and forget its form. In a famous experi-
ment, Jacqueline Sachs had subjects read a paragraph containing a test sentence,
then interrupted their reading immediately after the test sentence, eighty syllables
afterward, or one hundred sixty syllables later, and presented them with either the
test sentence in its original form, a version of the test sentence in which the voice
was changed, a version which was a close paraphrase, or a version of it which
differed semantically. Immediately after the test sentence, subjects were able to
distinguish all types of changes well, but after eighty or one hundred sixty syl-
lables they could identify only the semantically changed versions as changes: by
that time they had forgotten the form and the exact wording of the original, but
they remembered what it meant.29
The information thus gained is apparently integrated with other incoming in-
formation and with what is already known to form a unified whole, and the
individual bits that originally composed the whole may be forgotten. To test

29Jacqueline Sachs, "Recognition Memory for Syntactic and Semantic Aspects of Con-
nected Discourse," Perception and Psychophysics, 2 (1967), 437-42; see also her "Memorv in
Reading and Listening to Discourse," Memzory and Cognition, 2 (1974), 95-100.

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
Psycholinguistics and Literature 765

memory
memory for for
thethe
specific
specific
sentences
sentences
in whichininformation
which information
was presented,
wasJohn
presented,
D. John
Bransford
Bransford andand
Jeffrey
Jeffrey
J. Franks
J. Franks
constructed
constructed
sentences containing
sentencesfour
containing
separate four separ
ideas
ideas(e.g.,
(e.g.,
TheThe
rockrock
whichwhich
rolled rolled
down thedown
mountain
the mountain
crushed thecrushed
tiny hut atthe tiny hut a
the
theedge
edgeofofthethe
woods)
woods)
whichwhich
could be
could
analyzed
be analyzed
into four simple
into four
sentences
simple
eachsentences e
expressing
expressing oneone
ideaidea
(The (The
rock rolled
rock down
rolledthe
down
mountain.
the mountain.
The rock crushed
The rock
the crushed t
hut.
hut.The
Thehut
hut
waswas
tiny.tiny.
The hzlt
Thewas
hzlt
at the
wasedge
at the
of the
edge
woods).
of the
These
woods).
simple These sim
sentences,
sentences, known
knownas ones
as ones
because
because
each expressed
each expressed
one idea, could
one beidea,
combined
could in-
be combined
to
to twos
twos(e.g.,
(e.g.,
TheThe
rockrock
crushed
crushed
the tiny
thehut)
tiny
and hut)
threesand
(e.g.,threes
The rock
(e.g.,
crushed
The rock crus
the
thetiny
tinyhuthutat the
at the
edge edge
of theofwoods).
the woods).
Subjects were
Subjects
read some
wereones,
readtwos,
some and
ones, twos,
threes,
threes,andandthen
then
laterlater
presented
presented
with either
withsentences
either sentences
they had actually
they heard,
had actually hea
sentences
sentences they
theyhadhad
not heard
not heard
but which
but combined
which combined
ideas they hadideas
heard,
theyorhad
sen- heard, or
tences
tencesexpressing
expressing
ideasideas
different
different
from any
from
theyany
had heard
they (e.g.,
had heard
The rock
(e.g., The ro
crushed
crushed the
the
huthut
at the
at edge
the of
edge
theof
mountain).
the mountain).
The subjects
Thewere
subjects
most confident
were most confide
that they had heard fours-all of the ideas combined in one sentence-even
though they had never actually heard a four. Similarly, they had difficulty dis-
tinguishing threes they had heard from those they hadn't heard, but less trouble
distinguishing between old and new ones and twos. This indicates that they in-
tegrated the information they heard into the largest possible units, and then
tended to think they had heard that whole unit or something closely approximat-
ing it. Only when there was a sharp distinction between the whole and the test
sentence (as in the case of the ones and twos) could they tell what they had
actually heard.30
This study suggests that memory is not merely abstractive in the sense that it
abstracts or extracts information from a sentence and files that information away
for future reference, but rather constructive or integrative in that the informa-
tion from different sentences is integrated into a semantic whole. Such integration
can of course be greatly influenced by the context of the linguistic materials
used. Psycholinguistic experiments try to insulate their linguistic materials from
any context to force subjects to rely on their linguistic abilities instead of sup-
plementing these with other skills. Some recent studies, however, have argued
that since linguistic abilities do not normally operate in a vacuum, it is useless to
study them in a vacuum: language comprehension requires both linguistic and
nonlinguistic abilities working together.
In some circumstances, of course, the language used supplies its own context,
as with the sentences used by Bransford and Franks. In other circumstances, even
the very simple context supplied by a title can have a large effect on compre-
hension. Consider the following paragraph (the slashes mark phrase boundaries
determined by student ratings):

WITH HOCKED GEMS FINANCING HIM/ OUR HERO BRAVELY DEFIED ALL SCORNFUL LAUGH-

30John D. Bransford and Jeffrey J. Franks, "The Abstraction of Linguistic Ideas,"


Cognitive Psychology, 2 (1971), 331-50; see also John D. Bransford, J. Richard Barclay, and
Jeffrey J. Franks, "Sentence Memory: A Constructive versus Interpretive Approach," Cogni-
tihe Psychology, 3 (1972), 193-209, and John D. Bransford and Nancy S. McCarrell, "A
Sketch of a Cognitive Approach to Comprehension: Some Thoughts about Understanding
What it Means to Comprehend," Cognition and the Symbolic Processes, ed. Walter B.
Weimer and David S. Palermo (Hillsdale, N.Y.: Lawrence Eribaum, 1974), pp. 189-229.

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
766 COLLEGE ENGLISH

TER/
TER/ THAT
THAT TRIED
TRIEDTO
TOPREVENT
PREVENTHIS
HISSCHEME/
SCHEME/YOUR
YOUR
EYES
EYES
DECEIVE/
DECEIVE/
HE HE
HADHAD
SAID/
SAID
A
NOT A TABLE CORRECTLY TYPIFIES THIS UNEXPLORED PLANET/ NOW THREE STURDY SIS-
TERS SOUGHT PROOF/ FORGING ALONG SOMETIMES THROUGH CALM VASTNESS/ YET M
OFTEN OVER TURBULENT PEAKS AND VALLEYS/ DAYS BECAME WEEKS/ AS MANY DO
ERS SPREAD FEARFUL RUMORS ABOUT THE EDGE/ AT LAST/ FROM NOWHERE/ WELC
WINGED CREATURES APPEARED/ SIGNIFYING MOMENTOUS SUCCESS31

In isolation such a highly metaphorical passage is difficult to comprehend,


though the individual sentences offer no particular problems. What is diffic
somehow relating the sentences to each other and to some situation. A s
title-"Christopher Columbus Discovering America"-provides a key to un
standing the metaphorical references and to relating the different sentenc
each other. It does so by actuating what we know about history, by encoura
an interaction between the linguistic material and information stored in ou
memories. The actual mechanics of the process are not clear, but James Doo
and Rebecca Mullet have determined that presenting the title before the pa
graph increases the amount readers can recall about the paragraph; whi
presenting the title afterward has no such facilitating effect: apparently the
text affects the original process of understanding rather than the later proce
recall.32
Now a title is only the simplest kind of context; much literary education con-
sists of supplying more complex contexts-historical, social, religious, literary,
linguistic-expected to help the reader in his/her interaction with literary works.
These contexts are presumably not (or not entirely) unsystematic; instead they
are combined into unified conceptions, such as 'the Petrarchan sonnet', The
Elizabethan World Picture, and so on. Such conceptions bear a striking re-
semblance to what psychologists have begun to call 'scripts', 'frames', or
'schemata', which provide prototypes of situations.33 These prototypes guide our
perception of new situations and relate them to familiar patterns in much the
same way that Bever's perceptual strategies guide our perception of new sen-
tences. Thus there is a script or frame for a room which leads us to expect such
things as a floor, a ceiling, walls, windows, etc. If the room contains a bed and a
dresser, we might switch to a more specific script (for bedroom) and expect a
night table, an alarm clock, and other things, but not-emphatically not-a stove
or a dishwasher. Similarly, in entering a restaurant, we might expect waiters or
waitresses, a menu, a various kinds of food, etc. but not racks full of tennis
rackets along the walls, which is what we would expect from a sporting goods

31The text is one used in D. James Dooling and Roy Lachman, "Effects of Comprehension
on Retention of Prose," The Journal of Experimental Psychology, 88 (1971), 216-22.
32D. James Dooling and Rebecca Mullet, "Locus of Thematic Effects in the Retention of
Prose," The Journal of Experimental Psychology, 97 (1973), 404-6.
33The impetus for this research came from the necessity of representing human knowledge
in some coherent form for computer use. The classic article is Marvin Minsky's "A Frame-
work for Representing Knowledge," The Psychology of Computer Vision, ed. P. Winston
(New York: McGraw Hill, 1975), pp. 211-77. See also Benjamin J. Kiupers, "A Frame for
Frames: Representing Knowledge for Recognition," Terry Winograd, "Frame Representations
and the Declarative-Procedural Controversy," Roger C. Schank, "The Structure of Episodes in
Memory," and the other papers in Representation and Understanding: Studies in Cognitive
Science, ed. Daniel G. Bobrow and Allan Collins (New York: Academic Press, 1975).

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
Psycholinguistics and Literature 767

store.
store. Our
Oursurprise
surpriseatata dishwasher
a dishwasher
in in
thethe
bedroom
bedroom
or tennis
or tennis
rackets
rackets
in thein the
restaurant
restaurantisisexplained
explainedbyby the
thefact
fact
thatthat
thethe
frames
frames
we are
we using
are using
makemake
no provision
no provision
for
for them;
them;they
theyare
areanomalies.
anomalies. Clearly,
Clearly,such
such
frames
frames
or scripts
or scripts
are extremely
are extremely
usefuluseful
in
in making
makingsense
senseout
outofofthe
theworld
world weweencounter;
encounter;if we
if were
we were
without
without
themthem
all of all of
our
our time
time would
wouldbebespent
spentinintrying
tryingto to
organize
organize
our our
perceptual
perceptual
world.world.
One
One type
typeofofscript
scriptespecially
especially
relevant
relevantforfor
literary
literary
studies
studies
is the
is 'story
the 'story
schema'
schema'
or or
story
story grammar
grammarfirstfirstdeveloped
developed in in
psychological
psychological research
research
by David
by David
Rumelhart.34
Rumelhart.34
Story
Story grammars
grammarsare areinteresting
interesting notnot
merely
merelybecause
because
theythey
can describe
can describe
narrative
narrative
structures-similar
structures-similargrammars
grammars stemming
stemming more
moreor less
or less
directly
directly
fromfromPropp's
Propp's
Mor- Mor-
phology
phology of
ofthe
theFolktale
Folktaleare
are
well
well
known-but
known-but
because
because
theythey
claimclaim
to represent
to represent
readers'
readers' internalized
internalizedexpectations
expectations
about
about
thethe
formform
of narratives.
of narratives.
Readers
Readers
develop
develop
these
these grammars
grammarsfrom fromtheir
their
own
ownexperience
experience
of stories
of stories
and and
fromfrom
what what
they they
know know
about
about causality
causalityand
andpermissable
permissableor or
reasonable
reasonable
sequences
sequences
of actions;
of actions;
they they
use use
them
them toto guide
guidetheir
theirperception
perception
of of
stories
stories
by by
directing
directing
attention
attention
to what
to what
is likely
is likely
to
to be
be important,
important,organizing
organizingwhat
whathashas
already
already
occurred,
occurred,
predicting
predicting
what what
is to is to
come,
come, and
andsignalling
signallingwhen
whensome
some
partpart
of of
thethe
story
story
is complete.
is complete.
Unfortunately,
Unfortunately,
these
these grammars
grammarshave haveonly
only
recently
recentlybeen
been
proposed,
proposed,
and and
so are
so still
are fairly
still fairly
rudi- rudi-
mentary:
mentary:Rumelhart's
Rumelhart'soriginal
original
grammar
grammar consisted
consistedof eleven
of eleven
phrasephrase
structure
structure
rules
rules (nine
(nineofofthem
thempaired
pairedwith
with
rules
rules
forfor
semantic
semantic
interpretation)
interpretation)
such such
as as
story
story -- setting
setting+ +episode,
episode,episode
episode
- event
- event
+ reaction,
+ reaction,
reaction
reaction
? internal
? internal
response
response ++overt
overtresponse.
response.Even
Even refined
refined versions
versions
of the
of the
grammar
grammarincorporating
incorporating
transformations
transformationscan canhandle
handleonly
only simple
simplenarratives,
narratives,
but but
theytheyshowshow
greatgreat
promisepromise
in
in explicitly
explicitlylinking
linkingthethestructure
structure of of
a tale
a tale
withwith
a reader's
a reader's
perception
perception
of it.35
of it.35
As
As II suggested
suggestedabove,above,the
thepsychologists'
psychologists' scripts
scripts
(or schemata,
(or schemata, or frames)
or frames)
are are
akin
akin toto something
somethingwell wellknown
known in in
literary
literary
studies:
studies:
the the
context
context
or convention
or convention
(which
(which isis just
justa aspecial
specialkind
kindofof context).
context).TheThepsychologists'
psychologists' unique
unique
contribution
contribution
is thus
thus not
notin inrecognizing
recognizing thetheexistence
existence of such
of such
contexts,
contexts,but inbut
studying
in studying
how how
these
these contexts
contextsaffect
affectperception.
perception. As As
teachers
teachers
we explain
we explainand perhaps
and perhaps
exemplify
exemplify
the
the tradition
traditionof ofthethepastoral
pastoral
elegy,
elegy, anticipating
anticipating
thatthat
it will
it will
help help
our students
our students
un- un-
derstand
derstand and andappreciate
appreciate "Lycidas,"
"Lycidas," butbutwe we
are are
not not
veryvery
clearclear
on precisely
on precisely
how. how.
Or
Or wewe talk
talkabout
aboutexpectations,
expectations, especially
especially frustrated
frustrated
expectations,
expectations,
withoutwithout
being being
able
able to
to account
accountfor fortheir
theircontent
content or or
eveneven
existence,
existence,
except
except
on anonadan hocadbasis.
hoc basis.
By By
insisting
insisting that
thatsuch
suchknowledge
knowledge is systematically
is systematically organized
organized
and interacts
and interacts
with what
with what
we
we experience
experienceininananorderly
orderlyfashion,
fashion, psychology
psychologypoints
points
a promising
a promising
direction
direction
for for
literary
literary studies.
studies.
The
The psycholinguistic
psycholinguisticstudies
studies
I have
I have
reviewed,
reviewed,
however
howeverbriefly,
briefly,
are thus
are thus
usefuluseful
for
for literary
literarystudy
studyinintwo
twoways.
ways.First,
First,
they
they
provide
provide
a general
a general
framework
framework
for re-
for re-
search
search into
intothe
thecognition
cognition ofof
literary
literaryworks.
works.
All All
of these
of these
studies
studies
emphasize
emphasize
over over
and over the interaction between the perceiver and what is perceived; the
stimulus can be studied in isolation only at the risk of ignoring what makes it
valuable for human beings, just as sound can be studied acoustically but language

34David E. Rumelhart, "Notes on a Schema for Stories", in Bobrow and Collins, pp. 211-36.
35The transformations are suggested in Jean M. Mandler and Nancy S. Johnson, "Re-
membrance of Things Parsed: Story Structure and Recall," Cognitive Psychology, 9 (1977),
111-51. A good general review is Gordon H. Bower, "Experiments on Story Understanding
and Recall," The Quarterly Journal of Experimental Psychology, 28 (1976), 511-34.

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
768 COLLEGE ENGLISH

cannot.
cannot. TheThe
lack lack
of isomorphism
of isomorphism between the between
external stimulus
the external
and its internal
stimulus an
representation
representation arguesargues
that the that
readerthemust
reader
have some
must sorthave
of grammar
some sortor pro-
of gra
cedure
cedure forfor
constructing
constructing
an internalan representation
internal representation
of a literary text.ofThis
a literary
gram- tex
mar
marpresumably
presumably includes
includes
general schemata
general for schemata
approachingfor literary
approaching
texts which literar
specify
specify whatwhat
kindskinds
of things
of tothings
look for,
to as
look
well for,
as moreas specific
well as operations
more specific
or
heuristic
heuristic strategies
strategies
which which
allow theallow
readerthe
to segment
readerthe totext
segment
into unitsthe
small
text in
enough
enough forfor
processing
processing
and to relate
and to these
relate
segments
theseto each
segments
other. Thetonumber
each other.
and
andelaborateness
elaborateness of theofschemata
the schemata
are probably are
a function
probably of education
a functionand train-
of educa
ing,
ing,sosothat
that
an original
an original
'poem' schema
'poem'may schema
later bemay
replaced
later
by abesetreplaced
of schemataby a se
related
related to to
specific
specific
kinds kinds
of poems, ofbased
poems,
on form
basedor content
on form (or both).
or content
Similarly,(or bo
the
theperceptual
perceptual strategies
strategies
may become maymorebecome
refined more
or, inrefined
some extremeor, cases,
in some ex
calcified.
calcified. TheThegeneral
general
aim of aim
cognitive
of cognitive
literary studies,
literary
based on studies,
the psycho-
based o
linguistic
linguistic analogy,
analogy,
would would
thus be tothusdetermine
be to first
determine
wvhat schemata
first and
wvhatheuristics
schemata
are
areavailable
availableto readers;
to readers;
then what
thensubset
whatof these
subset
is available
of theseto a particular
is available t
reader;
reader; then
then
how how
a reader
a reader
decides which
decides
to use
which
one a particular
to use one occasion;
a particular
and, oc
finally,
finally, howhow
theythey
are actually
are actually
used. used.
Those
Those who
whochoose
choose
to pursue
to pursue
this rather
this
ambitious
rather program
ambitious
will find
program
the w
psycholinguistic
psycholinguistic literature
literature
useful inuseful
a secondinway:
a second
it suggests
way:
specific
it suggests
questions spec
that
thatcancan
be asked
be askedabout about
the processing
the processing
of literary language,
of literary
and provides
language,
specific
and pr
methods
methods forfor
studying
studying
these questions.36
these questions.36
For instance, For
Natalie
instance,
Waterson Natalie
has sug- Wat
gested
gested that
that
much much
normalnormal
perception
perception
of language ofis accomplished
language is by accomplished
means of b
what
what she
she
callscalls
phonological
phonological
skeletons-schemata
skeletons-schemata
of intensity patterns,
of intensity
classes ofpatter
vowels and consonants, and types of release-xvhich are 'fleshed out' by the
hearer's knowledge of contextual, pragmatic, and schematic constraints.37 Do
readers or hearers employ such skeletons in their perception of poems, especially
well defined shorter forms, such as the sonnet? On a higher linguistic level, we
might use Suci's methods to study how readers segment literary texts, and then
try to discover how they reintegrate these segments a la Bransford and Franks.
Recent research in this paradigm suggests that readers can distinguish between
sentences they have actually read and those conveying information they have
read better than hearers can.38 This might suggest that hearing a poem or a play
is more different from reading it than we would expect. And finally, recent
studies in the comprehension of sentences indicates that general terms are given
concrete instantiations which are better recall cues for the sentences than the
general terms originally presented. Thus bottle is a better recall cue for The
container held the cola than container, even though the latter appeared in the
sentence and the sentence and the former did not.39 Presumably something

36For a guide to this general area, see the annotated bibliography on "Psychology, Language
and Literature," by David Bleich and Eugene R. Kintgen forthcoming in Style.
37Natalie Waterson, "Some Views on Speech Perception," Journal of the International
Phonetic Association, 1 (1971), 81-96.
38Paul W. Flagg and Allan G. Reynolds, "Modality of Presentation and Blocking in Sen-
tence Recognition Memory, Memory and Cognition, 5 (1977), 111-15.
39See Richard C. Anderson and Andrew Ortony, "On Putting Apples into Bottles-A Prob-
lem of Polysemy," Cognitive Psychology, 7 (1975), 167-80; Richard C. Anderson, James W.
Pichert, Ernest T. Goetz, Diane L. Schallert, Kathleen V. Stevens, and Stanley R. Trollip,
"Instantia'tion of General Terms," JVLVB, 15 (1976), 667-69, and the references cited there.

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms
Psycholinguistics and Literature 769

similar,
similar, notnotconfined
confinedexclusively
exclusively to to
imagery,
imagery,happens
happens
in reading
in reading
literature.
literature.
But But
how
how does
does ititwork:
work:what
whatis is
thethe
range
range
of of
instantiation
instantiation
for for
a give,n
a give,n
term? term?
Do allDo all
readers
readers share
sharethe
thesame
samerange?
range? How
Howdo do
readers
readers
choose
choose
one one
particular
particular
instantia-
instantia-
tion,
tion, if
if indeed
indeedthey
theydo?
do?
This
This series
seriesofofquestions
questions(which
(whichcould
couldbe be
extended)
extended)
is typical
is typical
of the
of kind
the kind
of of
study
study II am
amadvocating,
advocating,the
the
study
study
of of
howhow
readers
readers
perceive
perceive
and comprehend
and comprehend
literary
literary works.
works.Such
Sucha astudy
studyseems
seems to to
meme to be
to central
be central
to the
to study
the study
of literature
of literature
simply
simply because
becausewithout
withoutthe the
operations
operations it explores
it explores
therethere
couldcould
be nobeliterature.
no literature.
This
This is
is aa point
pointthat
thatshould
shouldbebe
stressed:
stressed:
ignorance
ignoranceof various
of various
traditions
traditions
may affect
may affect
the
the perception
perceptionof ofliterature,
literature,but
butit does
it does
notnot
disable
disable
it. Without
it. Without
the mechanisms
the mechanisms
of
of perception,
perception,however,
however,nono
amount
amount
of of
background
background
information
information
couldcould
con- con-
stitute literature for the reader: the interaction between reader and work is the
one necessary ingredient in the transaction. And to study that interaction, I think
it would be profitable to adopt not only some of the conclusions and methods of
psycholinguistics, but also its perspective: that the aim of inquiry is not a char-
acterization of a language or a text, but of how people understand and use that
language or text.

al{rarf st zVM
AJqZ?WmEW f
eC E?3%%

/y

This content downloaded from


139.135.38.55 on Sun, 21 Jan 2024 10:46:44 +00:00
All use subject to https://about.jstor.org/terms

You might also like