J O U R N A I -O F V E R t s A L L E A R N I N ( ; A N D V E R B A L B ! . H A v r o R2 3 , 3 9 - 6 6 ( 1 9 8 4 )
RecognizinW
g o r d s ,P i c t u r e sa, n d C o n c e p t sA
: C o m p a r i s o no f
L e x i c a lO
, b j e c t ,a n d R e a l i t yD e c i s i o n s
Junrrs F. Kri.or--rMount Hoh'oke Collepe
ANf)
Mnnv C. Porren
M u s s u c h u s e l l sI n s t i t u t e o J T e c h n o l o g y
A s e r i e s o f f i v e e x p e r i m e n t s a d d r e s s e dt h e q u e s t i o n o f w h e t h e r p i c t u r e s a n d t h e w o r d s
t h a t n a m e t h e m a c c e s sa c o m m o n c o n c e p t u a lr e p r e s e n t a t i o n .I n t h e f i r s t t h r e e e x p e r i m e n t s
t h e p r o c e s s i n go f w o r d s i n t h e / e - r l t a 1t l e t i s i o n / z r . r tw a s c o m p a r e d w i t h t h e p r o c e s s i n go f
p i c t u r e d o b j e c t s i n a f o r m a l l y a n a l o g o u st a s k w h i c h w e c a l l e d l h e o b j e c t d e t i s i o n t r i , r f t T
. he
results showed that the lexical and object decision tasks produce approximately similar
r e s p o n s el a l e n c i e sa n d a r e s i m i l a r i n t h e i r s e n s i t i v i t yt o a s e t o f e x p e r i m e n t a l m a n i p u l a t i o n s
( e . g . . f r e q u e n c y e f f e c t s . i n t e r f e r e n c ee f f e c t s . s e m a n t i c f a c i l i t a t i o n f r o m r e l a t e d w o r d s o r
p i c l u r e s ) . I n t w o a d d i t i o n a l e x p e r i m e n t s t h e p r o c e s s i n go f w o r d s w a s c o m p a r e d w i t h t h a t
o f p i c t u r e s i n a m i x e d r e u l i t - t t l e t i . s i o nl a s l i n w h i c h a d e c i s i o n a b o u t w h e t h e r a w o r d o r
p i c t u r e r e p r e s e n l sa r e a l t h i n g i s t o b e m a d e i n d e p e n d e n to f t h e s u d a c e f o r m . T h e r e s u l t s
i n d i c a t e d t h a t s u b j e c t s w e r e u n a b l e t o m a k e a m o d a l d e c i s i o n so f t h i s s o r t l t h e r e s p o n s e
l a t e n c i e si n r e a l i t y d e c i s i o n w e r e m a r k e d l y l o n g e r t h a n t h o s e i n e i t h e r a p u r e l e x i c a l o r p u r e
o b j e c t d e c i s i o n a n d t h e r e w a s l i t t l e c o n c e p t u a lt r a n s f e ra c r o s sr e p e t i t i o n so f d i f ' f e r e n ts u r f a c e
f b r m s . O v e r a l l . t h e r e s u l t s o f t h e f i v e e x p e r i m e n t s s u g g e s tt h a t t h e m a j o r c o m p o n e n t i n a
l e x i c a l o r o b j e c t d e c i s i o n i s a f b r m - s p e c i f i c m e m o r y r e p r e s e n t a t i o no f t h e w o r d o r v i s u a l
obiect.
Does a common conceptual representation underlie the process of recognizing a
picture of an object and the word that
names it? Comparisonsof pictures and
words in tasks that require conceptual understanding show few differences in res p o n s e st o t h e t w o s u r f a c e f o r m s , s u g T h i s w o r k w a s s u p p o r t e di n p a r t b y C o n t r a c t M D A
903-76-C-0342from the Advanced Research Projects
Agency. We thank the laboratory at Scientific Applic a t i o n s , I n c . . f o r t h e i r a s s i s t a n c ei n t h e e a r l y s t a g e so f
this workl Jonathan Grudin, Janell Schwieckert. Kit
Oldham. and Jill Schusler Merves fbr their assistance
i n t e s t i n g s u b j e c t sa n d a n a l y z i n g d a t a r J o s e p h C o h e n .
Ellen Reese. David Rosenbaum.and Gair Snodgrass
for helpful comments on the manuscriptl and especially Barbara Faulconerfor her creative and artistic
et'forts in producing the nonobjecls used as stimuli in
these experiments. Requests fbr reprints should be
sent 1trJudith F. Kroll. I)epartment of Psychology and
Education.Mount Holyoke College.South Hadley.
Mass.01075.
gestingrelianceon a common code for both
( e . g . . B a n k s & F l o r a , 1 9 7 7 ;G u e n t h e r &
K l a t z k y , 1 9 7 7 ;P o t t e r & F a u l c o n e r , 1 9 7 5 ;
Potter, Valian, & Faulconer, 1977).In tasks
that do not clearly require conceptual or
semantic access, however, comparisons of
pictures and words reveal processing differences which presumably reflect reliance
on a surface level of representation(e.g.,
Durso & Johnson, 1979:Paivio, 1975,
1 9 7 8 ) .T h e e v i d e n c ei s t h u s c o n s i s t e n tw i t h
a t h r e e - c o d em o d e l ( e . g . , N e l s o n , R e e d &
M c E v o y , 1 9 7 7 ;P o t t e r , 1 9 7 9 ;S n o d g r a s s ,
1 9 8 0 , 1 9 8 4 )i n w h i c h t h e r e a r e m o d a l i t y specific representationsfor words and pictures as well as an abstract conceptualcode
common to both. Which codes are used
seemsto depend on the task.
T h e r e s e a r c h d e s c r i b e dh e r e i s a n a t tempt to examine more closely the relation-
39
0022--537
l/84 $3.00
Copyright O 1984 b) Academic Press. Inc.
All rights of reproduction in an_vlbrm reserved
KROLL AND POTTER
ship between surface and conceptual representations of words and pictures of objects. The research strategy is to use the
lexical decision task together with an analogous task developed for pictured objects.
In the lexical decision task (e.g., Coltheart,
Davelaar, Jonnason, & Besner, 1977:
Forster & Chambers, 1973; Frederiksen &
Kroll, 1976; Rubenstein, Lewis, & Rubenstein, 1971) subjects decide whether a
string of letters forms a real word. Logically, a lexical decision does not require retrieval of the concept associated with a
word; recognition of the surface lexical
form should suffice. There is abundant evidence to suggest,however, that lexical decisions are influenced by semantic factors.
For example, semanticallyrelated words
prime each other (Meyer & Schvaneveldt,
l97l), and at least part of the priming effect
appears to be automatic in that it operates
over a brief time interval and is unrelated
t o s u b j e c t s ' e x p e c t a t i o n s( F i s c h l e r , 1 9 7 7 ;
Neely, 1977). Lexical decisions are also influenced by word concreteness(Day, 1977.
James, 1975), a factor unlikely to be representedat the lexical level.
Lexical decisions are not based on conceptual or semantic representationsalone,
however. There are large effects of word
frequency in lexical decision (Frederiksen
& Kroll, 1976;Scarborough, Cortese, &
Scarborough, 1977), a factor presumably
relevant to the surface form (although measures ofword frequency are also correlated
with conceptual familiarity as seen in Experiment l). Also, lexical decision time is
influenced by the degree of word-likeness
of nonwords, suggestingthat subjects are
responsiveto surface orthography (Rubenstein, Richter, & Kay, 1975; Schulman &
Davison, 1977).
The goal of the present researchwas first
to develop an analogous decision task for
pictures (a task which we call the object
decision task') and then to use it as a way
of directly comparing the role of form-specific and conceptual representations in
w o r d a n d p i c t u r e p r o c e s s i n g .T h e o b j e c t
Lexical Decision
Object Decision
NONOBJECT
jox
,-..1
piono
\ 'Ihs
:c<
plonk
F t c . l . E x a m p l e s o f t h e m a t e r i a l su s e d i n t h e l e x i c a l
a n d o b j e c t d e c i s i o nt a s k s .
decision task is formally analogous to the
lexical decision task; subjects decide
whether pictures of objects represent real
things. In the lexical decision task, using
nonwords that conform to the orthographic
rules of English ensuresthat the subject
consults the lexicon before making a decision. Similarly, we attempted to construct
pictures of pseudo-objects(or nonobjects)
that represented a high approximation to
real objects. The procedure for generating
pseudo-objectsis describedin detail in Experiment l. Examples of the materialsused
in the lexical and object decision tasks are
shown in Figure l
The first section of this paper (Experiments 1-3) consistsof a set of experiments
designed to look for possible parallels between the lexical and object decision tasks.
If lexical and object decisions rely on the
s a m e c o n c e p t u a l r e p r e s e n t a t i o n ,t h e n t h e
two tasks should be influenced in similar
ways by experimentalmanipulationswhich
are thought to affect conceptual processing.
The second section of the paper (Experiments 4 and 5) consistsof two experiments
in which the two tasks are mixed into a
single reality decision task. To the extent
that lexical and object decisions rely on
c o n c e p t u a l p r o c e s s i n g ,a n d t o t h e e x t e n t
that words and pictures of objects share a
common conceptualrepresentation,mixing
the tasks should produce little disruption in
overall performance relative to conditions
WORDS, PICTURES, AND CONCEPTS
4l
in which the two tasks are performedsep- plete set of nonobjects is included in Aparately.
pendix A along with the results of a rating
SscrloN l: CoupenrNc LEXTcAL AND
OsrEcr DscrsloNs
In the first section three experiments are
described in which lexical and object decisions on correspondingwords and pictured
objects were made by separate groups of
subjects.Experiment I was designedto determine whether subjects could make object decisions in roughly the same amount
of time and with the same degree of accuracy as they make lexical decisions. In Experiment 2 the question was whether the
introduction of meaningful distractors in
the other surface format would produce
equivalent interferencein the two tasks. Finally, in Experiment 3 semantic priming effects were compared in the two tasks. To
anticipate the results of this section, the
pattern of data for lexical and object decisions looked remarkably similar across the
three experiments.
Experiment l: Baseline Comparison of
Lexical and Object Decisions
In Experiment I separategroups of subjects were asked to make either lexical or
object decisions. The two tasks were made
as comparable as possible by using words
in the lexical decision task that named the
objects pictured in the object decision task.
Method
Stimulus materials. The materials in the
object decision task were line drawings of
120 real objects. These were divided into
two sets of 60 each which were paired with
the same set of 60 nonobjects. All of the
objects had one-word names. The nonobjects were line drawings of closed figures
with an object-like appearance, created by
tracing parts ofdrawings ofreal objects and
regularizing the resulting figures. The com-
scale in which a group of subjects rated
each nonobject on how similar it appeared
to a real object.
The materialsin the lexical decision task
were the 120words that best named the objects and 120pronounceablenonwords, divided into two equivalent sets. The words
and nonwords varied in length from three
to eight letters. The words and pictures
used were chosen from a larger set of items
in which words and pictures had been
equated (as a group) for visual threshold
(Potter & Faulconer, l9l5). Drawings were
excluded if previous naming data (Potter,
Note l) had indicated any substantial disagreement among subjects as to what the
pictured object should be called.
For each surface form there were two
versions of the stimulus materials, each
consisting of 60 positive and 60 distractor
items. Only the nonobjects were repeated
in the two versions of the object decision
task. The positive items in each version (the
words and the pictures of real objects) were
equated for length and frequency of the
word names. The resulting items were categorized as being of high or low frequency,
according to whether the word label was
ranked above or below a value of 35 times
per million in the Kudera and Francis (1967)
word count. The mean frequency of the
high frequency words was 122 times per
million and that of the low frequency words
was 10 times per million. A large number
o f s e m a n t i c c a t e g o r i e sw e r e r e p r e s e n t e d ,
including animals,food, clothing, furniture,
tools, vehicles,and musical instruments.
The two versionsof the materialsthat were
constructed for each task were also counterbalanced so that instances of all categories were present in both forms.
Apparatus and procedure. The subject
was seatedat a table 0.5 m in front of a rear
p r o j e c t i o n s c r e e n . A r e c t a n g l e( 1 6 . 5 c m
high x 25.4 cm wide) was marked off in the
center of the screen as the fixation area. A
responsebox with two buttons (one for the
KROLL
AND POTTER
was
)eJ response,one for the ncrresponse)
positionedat a comfortabledistancein
front of the subject. Slidesof the stimulus
materialswere projectedusing a Kodak
carouselprojector equippedwith a Lafayette tachistoscopic
shutter.The exposure
duration was 500 milliseconds.A Marietta
digitaltimer was usedto measureresponse
time (to the nearestmillisecond)from the
onset of the stimulus. Each subject received 140trials of which the first 20 were
practice. In the middle of the experiment
the subjectwas given a short break.
ln lexicaldecisionsubjectsweretold that
they would see a brief flash containinga
row of letters.Sometimes
the row of letters
would form a real English word and sometimes it would not. Subjects were instructedto pressthe yes buttonif they saw
a word, and the ruobutton otherwise.Subjects wereencouraged
to guessif they were
not sure about their decisionand to try to
respondon the basisof their first impression. ln object decision, subjectswere
given similar instructions:"Press the yes
button if you seea picture of a real object,
and the no button otherwise."In both the
lexicaland objectdecisiongroups,subjects
were told to respondas quickly and as accuratelyas possible.
Subjects.Twelve subjectsperformed
eachtask. Within eachtask subjectswere
randomly assignedto one of the two versions of the stimulusmaterials.The subjects were collegestudentswho were paid
$3.00for their participation.All subjects
were screenedfor normal or corrected-tonormal visual acuity and for English as
their native language.
TABLE I
MEIN RespoNssTrvss (msec)ro MerE Lrxrcal
,cNoOe.rrcr DECrstoNS
asour Wonns ANDPrcruREs
oF OerecrsrN ExpgttrusNrI
Decision
Lexical
Object
Yes
632(.04)'
6 1 7( . 0 5 )
No
703(.03)
676(.03)
Nole. Error rates are given in parentheses.
< .001, and slightly less accurate, F(|,22)
: 5.34,p < .05.There were no interactions
between the modality of the task (word or
picture) and the type of response. Hence,
recognizing an object seems to be no rnore
difficult than recognizing a word. Superficially, the existenceof only 26 letters in the
alphabet might have led one to expect that
the algorithms for accessing lexical
memory would be far simpler and faster
than those used to recognize a novel
drawing. On the other hand, our visual
system has developedto recognizeobjects,
not words, so perhaps the facility of object
recognition should not be surprising.
According to the view that similar patterns of reaction times and errors imply
s i m i l a r u n d e r l y i n g p r o c e s s e s ,t h e s e d a t a
can be taken as support for reliance on a
common conceptual representation in processing words and pictures (Snodgrass,
1984).The argument is weak, however, because similar response patterns in lexical
and object decision could be explained in a
number of other ways. For example, it is
impossible to equate the distractors in the
two tasks, and it is known that subjectsrespond faster to both positive and negative
items when the distractors in a lexical decision task are less word-like. Thus, the
Resultsand Discussion
similarity between the two tasks could conMean responsetimes and error rates for ceal a true difference that was masked by
the lexical and object decisiontasks are compensating differences between the nongivenin Thblel. The responsetimes(RTs) words and nonobjects.
and error ratesfor the two taskswere comIn addition to the overall analysesof RTs
parable;none of the differencesbetween and errors, a number offiner-grain analyses
the two taskswas significant.Overall,pos- were performed to see whether lexical and
itiveresponses
werefasterthannegativere- o b j e c t d e c i s i o n s w e r e i n f l u e n c e d b y t h e
sponsesin both tasks,F(l,22) : 31.86,p same variables. These additional analvses
WORDS. PICTURES. AND CONCEPTS
were restricted to the set of positive items
in each task.
Frequency effects. Since the positive
items were chosen to represent a range of
frequencies (measured by the frequency of
the word's name), it was possible to obtain
a rough estimate of the frequency effect in
each condition. This post hoc analysis of
frequency effects was accomplished by
comparing the mean RTs and errors for
items of relatively high frequency (over 35
per million in the Kudera and Francis, 1967,
word count) with those for items of relatively low frequency (less than 35 per million). An analysis of variance was performed on yeJ responses as a function of
frequency class (high or low) and task (lexical or object decision). The result was a
significant frequency effect in both tasks
F ( 1 , 2 2 ) : 2 2 . 1 6 ,p < . 0 0 1 . T h e f r e q u e n c y
effect was smaller than that typically found
in lexical decision tasks (here it was 35 milliseconds in lexical decision and 24 milliseconds in object decision), although this
was probably a result of the crude division
of frequency classes and the omission of
very low frequency items since few picturable objects fit into that category. Of
greater interest is the fact that the size of
the frequency effect did not interact with
task; the small frequency effect was reliable
for both lexical and object decision even
though the measure of frequency was a
measure of frequency of printed words.
The fact that both tasks appear to be influenced by familiarity in similar ways is
consistentwith the common-code model, in
that it suggests that at least part of the frequency effect is due to conceptual frequency rather than to lexical or visual frequency per se. On the other hand, it is
likely that word frequency is correlated
with object frequency, in which case, the
two frequency effects could each reflect
m o d a l i t y - s p e c icf i p r o c e s s i n g .
Rank-order correlation.If lexical and object decisions are both based on access to
a common concept, then the relative availability of particular concepts should influ-
ence each type of decisionin the same way;
concepts which represent familiar ideas
should be recognized more rapidly than
concepts which represent less familiar
ideas. To see whether this was true. a rankorder correlation was computed on the 120
positive item meanscollapsedacrossthe
two versions of each task. The resulting
correlation was significant, r, : .41, p <
.025, supporting the idea that similar processes, and perhaps the same conceptual
representation,underlie lexical and object
decisions. Again, although the obtained
similarity between the two tasks is consistent with the common-code model, it could
also be due to the similar degree of familiarity of the name and the appearance of
given objects. Only if it can be shown that
the words gain in functional frequency
when one sees their referents, and vice
versa, could one be sure that the correlation is attributable to retrieval of a common
code at the time o[ decision.
Are pictures named in ob.ject dec'ision?
One possibility for the correlation between
the two tasks that we have not yet considered is that the same surface form. the object's name, is used in both decisions.If an
object must be named, or its verbal code
accessed,before an object decision can be
made, then object decisionsshould take app r o x i m a t e l y 2 0 0 - 3 0 0 m i l l i s e c o n d sl o n g e r
than lexical decisions because that is the
additional time it takes to attach a verbal
label to a picture, compared to a written
w o r d ( C a t t e l l , 1 8 8 6 ;P o t t e r & F a u l c o n e r ,
1975; Smith & Magee, 1980).This idea is
consistent with dual-code models such as
Paivio's (1978) in which conceptual information is primarily accessedvia the verbal
system. This predicted difference was not
supported by the data just presented; lexical and object decisions took about the
s a m et i m e l o p e r f o r m .
We have argued that the similarity between lexical and object decisionsmay just
be a resemblance;the two tasks may involve different processeswhich happen to
finish in the same overall time. The next set
KROLL
AND POTTER
of experimentsconsiderssome alternative
proceduresfor determiningwhether the
overall similarity of lexical and object decisionsfound in Experiment1 was just a
superficialresemblanceor a true reflection
of the use of the sameunderlyingconceptual code.
Experiment2: Filtering Pictures or Words
In Experiment I the time and accuracy
of decisionsabout words and pictured objects appearedsimilar.This similaritydoes
not necessarilyimply accessto a common
representation;
lexicaland objectdecisions
could rely on modality-specificrepresentations which are processedsimilarly.
Thesepossibilitieswere testedin Experiment 2 by asking subjectsto filter out the
other modality.Considera subjectwho is
participatingin a lexical decisiontask with
instructionsto respondyes to words andno
to nonwords.What if a picture of an object
were presentedas a distractor?An object,
like a word, has meaning.If retrievalof a
conceptualrepresentationis part of the
basisfor lexicaland objectdecisions,then
rejection of real items in the wrong modality might be difficult, just as it is difficult
to ignore meaningfulwords in Stroop-type
interferencetasks (Rosinski,Golinkoff, &
Kukish, 1975;Smith & Magee,1980).Picturesofnonobjects,however,shouldbe rejected quickly in a lexicaldecisiontask becausethey are neither words nor meaningful. If lexical decisionsare basedon
modality-specific
codes,however,thenpicturesof real objectsand nonobjectsshould
both be relativelyeasyto filter out (i.e., to
reject as not a word). The sameargument
holdsfor the complementary
casein which
an object decisionis performedand words
or nonwordsappearas distractors.
Method
The stimulusmaterials,apparatus,and
procedurewereidenticalto thosedescribed
for Experimentl, with the followingexceptions.The 60 negativetrials in eachexperiment were now divided into three cateso-
ries of 20 items each: same-form distractors, other-form real items, and other-form
distractors. The other-form real items were
always taken from the version of the positive set not used for that subject. That is,
if cat was seen as a positive item in lexical
decision, then the picture of a cat was not
used as a distractor. The materials were
counterbalanced across subjects so that the
entire set of items appeared as positive targets and as distractors.
Subjects. Twelve subjects performed
each task. Within each task subjects were
randomly assigned to one of the two versions of the stimulus materials. The subjects, recruited from the same subject pool
as that for Experiment 1, were paid $3.00
for their participation. All subjects were
screened for normal or corrected-to-normal
visual acuity and for English as their native
language.
Resultsand Discussion
Mean response latencies and errors in
performing the lexical and object decision
tasks are shown in Table 2. For both tasks,
distractors of the other modality were extremely easy to reject. The mean response
time for both other-form real items and
other-form distractors was approximately
225 millisecondsfaster than the time to reject same-form distractors. In addition,
there were no errors on either type ofotherform stimuli in either of the tasks. Clearly,
subjectswere able to filter out items in the
irrelevant form. There was, however, a substantial reduction in the speed and accuracy
with which the same-form distractors were
rejected. An analysis of variance comparing lexical and object decisions in Experiment 2 yielded no significant differences between the tasks, F(1,22) < l. A
separate analysis of the negative trials revealed a highly significant effect of the stimu l u s t y p e o n R T s , F ( 2 , 4 4 ) : 3 3 1 . 0 1 ,p <
. 0 0 1 ,a n d e r r o r s ,F ( 2 , 4 4 ) : 3 8 . 8 9 ,p < . 0 0 1 .
The same-form distractors were significantly slower and more likely to produce
errors than were the other-form stimuli.
45
WORDS, PICTURES,AND CONCEPTS
TABLE 2
MEaNRtspoNsl Tttrlts (msec)ro MerE Lgxrcat. nNDOsrscr DnctstoNsesout Worus ANDPrcruREsoF
Oerpcrs wHL,NDrsrRActonslNcluDp sotH Sunrecs Fonvs (ExrenrrvtnNr'
2)
No
Same-form
distractor
Lexical
Object
5 8 6( . 0 1)
624(.04)
Other-form
real
7 s 7( . t s )
7 6 0( . 1 0 )
523(.00)
s44(.00)
Other-form
distractor
(.00)
-s20
(.00)
-5.s5
Nole. Error rates are given in parentheses.
There was no suggestionof an interaction
between task and pattern of RTs (or errors)
for negative trials.
Frequency effects. As in Experiment l,
the positive items in each task were categorized into high and low frequency
classes.An analysis of variance performed
on )eJ responsesas a function offrequency
class (high or low) and task revealeda small
but significant frequency effect, F(|,22) :
11.67,p < .01. The magnitude of the frequency effect was 25 millisecondsin lexical
decision and l3 milliseconds in object decision. The comparable effects in Experiment 1 were 35 milliseconds in lexical decision and 24 milliseconds in object decision.
The diminished frequency effect in Experiment 2 is consistent with a peripheral
filtering explanation in which it is assumed
that an initial judgment was made on the
basis of the surfaceform of the stimulus. If
there was a mismatch on the basis of surface features(a possiblebasis for peripheral
filtering), then the item was quickly rejected. If the surface features matched,
however, there was an additional stage of
memory access as in normal lexical or obj e c t d e c i s i o n . I f s u b j e c t s s o m e t i m e sr e sponded positively simply on the basis of a
modality match (as the increase in false
positives to same-form distractors suggests), then the positive trials would have
consisted of a mixture of fast responses
without lexical or object access(and hence
without frequency effects) and slower res p o n s e sw i t h a c c e s s .T h e r e s u l t w o u l d b e a
dilution of the frequency effect of the kind
observed.
An examination of the error Rfs for the
same-formdistractorssupportedthis explanation. The average RT for the l5% false
positives to nonwords in the lexical decision was 523 milliseconds,which was faster
t h a n t h e 5 8 6 - m i l l i s e c o n da v e r a g c R T f o r
true positives and almost identical to the
average RT for correct rejection of picture
distractors, 522 milliseconds. Similarly, in
the object decision task, the averageRT for
the lUVofalse positives to nonobjects was
563 milliseconds,again faster than the 624millisecond average RT for true positives,
and close to the 550 millisecondsfor rejection of verbal distractors. Thus. when
other-form stimuli are mixed with sameform distractors in both lexical and object
decisions there appears to be a change in
decision strategy to allow for peripheral filtering.r
Since the conditions of the present experiment permit a change in decision criteria. these data do not constitute a strong
test of conceptualinvolvement in lexical
and object decisions,other than to indicate
that access to a conceptual representation
is neither automatic nor mandatory under
these conditions. A more direct wav to exI A similar result has been reported
by Scarboro u g h . G e r a r d , a n d C o r t e s e ( 1 9 8 4 )i n a b i l i n g u a l l e x i c a l
d e c i s i o n t a s k . I n t h e i r t a s k . s u b j e c t sw e r e t o l d t o r e s p o n d p o s i t i v e l y t o o n e l a n g u a g e( e i t h e r E n g l i s h o r
S p a n i s h )a n d t o r e j e c t t h e o t h e r l a n g u a g e .T h e i r r e s u l t s
showed that there were no frequency effects on the
time to reject the distractor language. Like the present
r e s u l t s ,t h e r e s u l t so f t h e S c a r b o r o u g he t a l . s t u d y s u g g e s t t h a t s u b j e c t s w e r e a b l e t o s u c c e s s f u l l yf i l t e r t h e
distractor stimuli even when they shared surface characteristics.
KROLL AND POTTER
amine the conceptualcontributionto lexical and objectdecisionsis to seewhether
both tasks benefit equally from semantic
priming.If the semanticprimingeffectdis(1971)
coveredby Meyer and Schvaneveldt
for lexical decision is due to lexical organization and to associationsspecific to
words, then we would not expect equivalent facilitation for pictures of related objects.
mantically related but not highly associated; (4) the materials were presented tachistoscopically;and (5) the response was
vocal rather than manual. The details of
these changesare described below.
Stimulus materials. Positive trials consisted of pairs of real items (words in lexical
decision, pictures of objects in object decision). Half of the real items were semantically related and half were unrelated. The
unrelated pairs were formed by randomly
Experiment3: SemanticPriming
scrambling members of related pairs. Of the
In Experiment3 the Meyer and Schva- r e l a t e d p a i r s , o n e - q u a r t e r w e r e s e l e c t e d
neveldt(1971)semanticpriming paradigm from norms of free association (Postman &
was usedin the objectand lexicaldecision Keppel, 1970),such that they were highly
tasks.If semanticpriming is the result of associatedas well as semantically related.
accessto a generalconceptualmemory The criterion for high association was that
systemthat is commonto words and pic- the second item of each word pair had to
turedobjects,then we mightexpectsimilar be a primary response to the first item with
priming effects in both tasks. If semantic a relative frequency of .20 or greater. The
priming is due exclusivelyto lexical orga- r e m a i n i n g t h r e e - q u a r t e r s o f t h e r e l a t e d
nization and word-specificassociations, i t e m s w e r e s e m a n t i c a l l y r e l a t e d b u t n o t
then we might only expect to find a se- h i g h l y a s s o c i a t e d . S e m a n t i c r e l a t e d n e s s
manticpriming effectin the lexical decision was determined by subjective ratings taken
task. Furthermore,if only associativere- from an independent group of subjects. A
giveriseto semanticfacilitation, description of these ratings and the actual
lationships
then we might expect to find a larger se- stimulus pairs used are given in Appendix
mantic priming effect for words that are as- B. An example of the distinction between
sociativelyrelated to one another than for "semantically related and associated" and
wordsonly relatedsemantically
or concep- "semantically related only" can be seen in
tually. To test this hypothesisin Experi- the following word pairs: dog-cat, arm-leg
ment 3 we comparedthe priming effect for (associatedas well as related) versus cowpairs of items that were relatedonly by horse, apple-banana (semanticallyrelated
virtue of membershipin the samesuperor- but not highly associated).Unrelated pairs
dinate semanticcategorywith the effect of were formed by interchanging items within
sharingthe same superordinatecategory the associated pairs or the semantically reand also beinghighlyassociated.
lated only pairs. Since items forming associated pairs may also differ from semantiMethod
cally related items in a variety of ways
The materialsandprocedureweresimilar (e.g., they are often words of higher freto those describedfor Experimentl. The quency), the two types of unrelated pairs
majorchangeswerethat (l) two itemswere provide a good control for item-specific difpresentedsimultaneously
on a singletrial, ferences.
one abovethe other,for both tasks;(2) the
Negative trials consisted of mixed pairs
subject had to decide whether both items of real and distractor items (words and nonwerereal words(in lexicaldecision)or pic- words in lexical decision, pictures of obturesofreal objects(in objectdecision);(3) j e c t s a n d n o n o b j e c t s i n o b j e c t d e c i s i o n ) ,
half of the positivepairswere semantically and pure pairs of two distractors (two nonrelatedand associated.and half were se- words in lexical decision. two nonobiects
WORDS, PICTURES, AND CONCEPTS
47
TABLE 3
in objectdecision).Half of the mixedpairs
MEIN RnspoNsETIMES(msec) ro MaxE Llxtcel
position
of
containeda real item in the top
eNo Ougct DrctstoNs por. Rglerr,l eNo
the stimulusarray and half containeda real
oF
UNneL,rrpl PAIRSoF WoRDs AND PICTURES
item in the bottom position. Each of the
OsrEcrs tN ExptttugNr 3
three types of negativetrials occurred
Unrelated
Related
equallyoften.
pairs
pairs
Diff
Decision
Two versionsof the materials,eachconl8
840(.03)
822(.03)
Lexical
sistingof 96 pairs, were constructedsuch
49
856(.07)
807(.05)
Object
rein
both
that eachpositiveitem appeared
No/e. Error rates are given in parentheses.
lated and unrelated conditions but in difversion
of
within
a
ferent versions.Thus.
the materialsno item was repeated.Of the the items as they could recall from the ex96 pairs in each version48 were positive periment, followed the lexical or object depairs (both membersreal) and 48 were neg- cision task.
Subjects. Thirty-two college-age stua t i v ep a i r s ( 3 2 m i x e d p a i r s , l 6 p u r e d i s pairs).
dents were each paid $2.50 for participatractor
pairs
procedure.
Stimulus
tion. Half the subjects were randomly
Apparatusand
and
a
three
presented
field
of
in one
were
chosen to perform each decision task. AII
(Scientific
Prosubjects had normal or corrected-to-normal
one-halffield tachistoscope
appeared
The
items
visual acuity and were native English
totypeModel N-1000).
horizontally
centered
speakers.
one above the other,
within the visual field. The distancebeResultsand Discussion
tweenthe two itemswas approximately1.2
Decision latenciesfor positive responses
cm. A secondfield containeda fixationdisplay consistingof two line segmentsindi- to related and unrelated stimulus pairs are
cating the position of the two stimulus shown in Thble 3 for both the lexical and
items. A differentfixation field was usedin object decision tasks. The overall pattern
the word and pictureconditionsto accom- of results is similar for both tasks: related
modate differencesin the vertical visual pairs were judged to be real words or pictures of real objects more rapidly than unangleof the word and picturedisplays.
r e l a t e dp a i r s ,F ( l , 3 0 ) : 3 0 . 5 5 ,p < . 0 1 .T h e
individual
in
an
was
tested
Each subject
magnitude of the semantic facilitation efmin30
lasted
approximately
that
session
utes and consistedof two blocks of 48 fect, although significant for both lexical
trials, precededby a block of 24 practice and object decisions,was noticeably larger
trials using different stimulusmaterials. f o r p i c t u r e s ( 4 9 m i l l i s e c o n d s ) t h a n f o r
Subjectswere instructedto respond as w o r d s ( 1 8 m i l l i s e c o n d s )F, ( 1 , 3 0 ) : 6 . 6 3 , p
quickly as possibleby sayingyes if both < .05. Although there were somewhat more
itemswere real words (in lexical decision) errors in object decision (5.9%) than in lexor picturesof real objects(in object deci- icaf decision (3.0%) the difference was not
. h e i r statisticallyreliable. In addition, there were
s i o n )a n d b y s a y i n gn o o t h e r w i s e T
spokenresponsesactivateda voice key no reliable differences in errors between the
(ScientificPrototypeaudiothresholddetec- related and unrelated conditions in either
tion relay,761G), which stoppeda counter t a s k , F ( I , 3 0 ) < l .
One way in which the priming of pictures
(ScientificPrototypeModel N-1002)which
had beenactivatedat the onsetof the stim- and words might differ is in the role of word
ulus display.Reactiontime was measured associations.If the primary mechanismfor
lexical priming is associated(as reflected in
to the nearestmillisecond.
A surprisememory task, in which sub- a s s o c i a t i o n n o r m s ) , w h e r e a s t h e p r i m a r y
jects were askedto write down as many of mechanism for picture priming is semantic
KROLL AND POTTER
48
relatedness,that differencecould account
for the smaller priming effect for words,
since only one-quarterof the related pairs
were also associated.To evaluatethis hypothesis,the reactiontimes and errorsfor
positive items (related and unrelated)in
both taskswere comparedfor the two types
of relatedpairs (semanticallyrelatedand
associated
vs semanticallyrelatedonly).
Thesedata are shown in Table4. It is important to rememberthat the unrelated
;lairs were constructedby scramblingrelated pairs of the sametype: related associated pairs were recombinedto form unrelated"associated"pairs,
semantically
related pairs were recombined to form
unrelated pairs using the same items. Inspectionof the data in Table 4 revealsno
effect of associationon the magnitudeof
the semanticpriming effect, either for
words or for objects. Overall, there was a
significanteffect of associationvalue on
reactiontimes,f(I,30) : 24.95,p < .01,
a n d o n e r r o r s ,F ( 1 , 3 0 ): 6 . 4 4 ,p < . 0 5 .
There was also a significantinteractionbetweentask (lexicalor object decision)and
associationvalue in the reactiontime analysis,F(l,30) : 12.9,p < .01.The effectof
associationwas entirely due, however,to
the fact that both related and unrelated
combinationsof the words used in associatedpairsproducedfasterreactiontimesin
lexical decisionthan did the words in the
semanticallyrelatedonly conditions.(This
differencein materialswas not significant
in the objectdecisiontask.)The strikingresult is that despitethe lareedifferencebe-
tween the subgroups of words, the magnitude of the semantic relatedness effect was
virtually identical (15 milliseconds for associated and related pairs, l8 milliseconds
for semantically related only pairs). The hypothesis that words might be primed differentially by associatesthus received no support from the present experiment.
The fact that the words that enter into
associative relationships were accepted
more rapidly in lexical decision,which their
conceptual counterparts in object decision
showed no such benefit, suggeststhe presence of a factor correlated with associative
strength that influences a stage ofword recognition but not object recognition. The
failure to find an enhancement of the semantic priming effect due to association
replicates a previous result of Fischler
(1977) for the lexical decision task. It further supports the idea that the effect of semantic facilitation is not within the lexicon
(where word-specific associations might reside), but in an amodal conceptual system.
Negative /rials. Reaction times and errors for the three types of negative trials
were compared for the lexical and object
decision tasks. Recall that there were negative trials in which both items of the pair
were distractors (pure negatives) and other
trials in which one item was a distractor and
the other was real (mixed negatives).The
position of the real item in mixed negative
trials was balanced so that it appeared on
the top of the display half of the time and
on the bottom for the remaining trials. The
results are shown in Table 5. In both tasks
TABLE 4
M p a N R E s p o N s ET I M E s ( m s e c ) r o M A K E L E X T C A La N o O g r E c r D E c r s r o N sA B o u r W o R D Se u o P r c r u t E s o p
Osiecrs THer Ans Htcnly AssoctarEo es WElr- es SguaNttceLLy RrlerEo ColtpetEo ro Wonos aNo
Prcrunrs or Oerncrs THAT ARE ONr-y SpveNrrceLLy RrLersD (ExpERTMENT
3)
S e m a n t i c a l l yr e l a t e d
and associated
Decision
Lexical
Object
Semantically related
only
Related
Unrelated
Diff.
Related
Unrelated
Diff.
776 (.02)
804 (.02)
7 9 1( . 0 1 )
843(.04)
l-5
39
8 3 8( . 0 3 )
808(.06)
856(.04)
860(.07)
l8
52
Note. Error rates are given in parentheses
WORDS. PICTURES. AND CONCEPTS
TABLE 5
(msec)ro Rs:r.cr Drsrnecron PernsrN Lrxrcal eNl Oerscr DEcrsroN(ExpsnluENr3)
MEaNRssporvsrTttr.tps
Type of distractorpair
Decision
Lexical
Object
Top real,
bottom distractor
r 0 7 8( . 2 8 )
9 9 r( . 1 6 )
Top distractor,
bottom real
Both
distractors
9 7 5( . 1 5 )
9 6 5( . 1 7 )
953(.06)
892(.04)
the time to reject mixed pairs was longer as foflows: 1087milliseconds(287aerror) to
than the time to rejectpure distractorpairs, rejectmixed negativeswith the real item on
and the time to reject mixed pairs with the top, 904 milliseconds(8Voenor) to reject
real item on the top was longer than the mixed negativeswith the real item on the
time to reject mixed pairswith the real item bottom. and 884milliseconds(3Voerror\ to
on the bottom. This overall patternwas sig- reject pure negatives.The main difference
nificantfor both RTs, .F(2,60): 28.47,p < betweenour lexical decisionresults and
.01,and errors,F(2,60): 17.17,p < .01. their resultsis in the magnitudeof the seThe mixed conditionswhich producedthe mantic priming effect; we found an lS-millongestRTs alsoproducedthe mosterrors. liseconddifferencebetweenrelatedand unA post hoc Newman-Keuls test revealed relatedword pairs, while they found an 85significanterror differencesfor the pure vs milliseconddifferencebetweenrelatedand
mixed comparisononly; the two types of unrelated pairs. (Meyer & Schvanemixed trials were not statisticallydifferent. veldt'sactualresultswere940milliseconds
The patternof negativeresponseswas (9Voerror) for the unrelatedpairs and 855
slightly different in lexical and object de- milliseconds(6% error) for the related
pairs.) The longerRTs and higher error
cisionsas revealedby both RTs, F(2,56):
3.64,p < .05,and errors,F(2,56): 3.29, ratesfor both the relatedand unrelatedconp < .05. In lexical decision,there was a ditionsin their experimentmay have conclear effect of the position of the real item tributed to a large semanticpriming effect
in mixednegatives:
when the real word ap- in their study than in ours. An important
pearedin the top position of the array, lex- differencebetweenthe two studies,howical decisiontimeswere longerand lessac- ever, is that the words in our experiment
curatethanthey wereon trialsin which the werenecessarily
concretenouns(objectlareal item appearedin the bottom position: bels).To the extentthat word concreteness
q(,56) : 7.41,p < .01for RTs, andq(4,56) influenceslexicalretrieval(James,1975;
: 4.56,p < .05 for errors.In objectdeci- Kroll, Supraner,& Merves, Note 2), the
sion, the differencebetweenthe two posi- smallereffectof semanticfacilitationin this
tions of the real object in mixed negatives experimentmight be attributableto the fact
wasnot significantfor RTs, q(3,56): 2.08, that concretewords are often accessed
p > .05, or for errors, q(2,56)< l, sug- morerapidlythanabstractwords,and thus
gestinga differencein processingorder for are lesssusceptible
to priminginfluences.
picturesand words.
Two other studieshave reported small
Overall,the patternsof resultsfor the effects of semantic facilitation (Sperlexicaland objectdecisiontaskslook quite ber, McCauley,Ragain,& Weil, 1979)and
similar and bear a strong resemblanceto word frequency (Scarborough,Gerard, &
Meyer and Schvaneveldt's(1971)original Cortese,1979)with picturablenouns.The
results.The patternof negativeresponses Sperberet al. study examinedsemantic
that Meyer and Schvaneveldt
obtainedwas priming effects in word- and picture-
KROLL AND POTTER
naming.Their resultswere strikinglysimilar to ours: the magnitudeof semanticfacilitation for words (19 milliseconds)was
much smallerthanthat for pictures(51milliseconds).An examinationof the relatednessratingsincludedin Appendix B suggestsa possiblereasonfor the differencein
the magnitudeof the priming effect in lexical and object decisions:overall,pictures
were ratedas beingmore closelyrelatedto
one anotherthan words (seeThbleB-l).
The greaterperceivedrelatednessfor pictures might have producedgreaterfacilitat i o n ( i . e . , f a s t e rR T s ) i n o b j e c td e c i s i o n
when the pair of pictureswas actuallyrelated,and possiblysomeinterference(i.e.,
slowerRTs)whenthe pairswereunrelated.
The net resultwould be an increasein the
magnitudeof the semantic priming effect
for picturesrelativeto words. The data in
Table3 offer some supportfor this kind of
analysisin that RTs were fasterfor pictures
than for words on relatedpairs, but the reverse was true for unrelatedpairs. A different explanationfor differential patterns
of semanticprimingfor picturesand words
has recentlybeen given by Huttenlocher
and Kubicek (1983).They proposethat semantic priming affectsthe speedof perceptual identification of pictures but not
words. These two explanationsare not
easilydistinguished
by the presentdata.
The resultsof the incidentalmemorytask
given at the end of the experimentwill be
presentedin a later sectionof this paper.
Summaryof Resuhs
In Experimentsl-3 consistentparallels
were found betweenthe lexicaland object
decisiontasksunder conditionsin which
the two tasks were performedby separate
groupsof subjects.The overall response
times and pattern of errors were similar for
the two tasks(Experimentl), the apparent
easein filtering out distractorsof the other
modalitywas the samefor both tasks(Experiment 2), and there were effects of semantic priming for both lexical and object
decision(Experiment3). The only distinct
differencesbetween the tasks were the size
of the semantic facilitation effect (it was
larger for pictures than for words) and the
pattern of negative responses (which suggested that pictures were processedin parallel while words were processedserially).
It was argued that reliance on a common
conceptualrepresentationin lexical and object decisions would produce the kinds of
parallel results obtained here. While the observed parallels are consistent with a
common locus of representation for words
and pictured objects, they do not, when
considered alone, provide a critical test of
the common-code hypothesis. For one
thing, the presentation of pictures and
words blocked between subjects might
have permitted the use of independent
strategiesfor each surface form. These
strategies might bear some resemblance to
each other without requiring the use of the
same representation.
The next section of the paper addresses
another prediction of the common-codehypothesis, namely, that mixture of the two
surface forms should not have a disruptive
influence on the processing of either surface form.
SecrroN 2: Rner-rry DBcrsroNs pon
Mtxrunss oF WoRDSaNo PrcruRe,s
In the second section two experiments
are described in which the lexical and object decision tasks are mixed into a single
reality decision task. If the results of the
first three experiments reflect access to a
common conceptual representation for
words and pictured objects, then forcing
subjects to treat pictures and words independently of their surface form should not
radically alter the pattern of results.
Experiment 4 provided a test of the
common-code hypothesis by mixing the
pictures and words that had been presented
separatelyin Experiment l. ln Experiment
5 the question asked was whether the facilitation that typically results from repetition
of words in lexical decision (e.g., Scarborough et al., 1977) would also hold across
51
WORDS, PICTURES, AND CONCEPTS
surface forms (from words to pictures and
vice versa). The results of this section revealed changes in performance when pictures and words were mixed and little evidence for facilitation across surface forms.
E E X P E R I M E N T lP: U R E
I expenruexr 4: urxeo
o
q)
o
UJ
=
F
Experiment 4: Comparing Words and
Pictures in Reality Decision
IJJ
a
z
o
In Experiment 4 subjects were asked to
decide whether individual words, pictures
of objects, pseudo-words, and pictures of
nonobjects were real or not real when they
were presented in a mixed sequence.
uJ
z
ut
=
WORDS
Method
Stimulus meterials. The materials were
identical to those described in Experiment
1. To construct the sequencefor the reality
decision task. the materialsfrom the lexical
and object decision tasks were randomly intermixed and divided into versions so that
the total length of the experiment was identical to Experiment 1. The result was four
versions of reality decision, each consisting
of 60 positive and 60 negative trials. Additional constraints were that an equal
number of each type of stimulus appearsin
each version and lhat any given item appears in a given version in one surface form
o n l y ( i . e . , a n i t e m p r e s e n t e da s a w o r d
would not be repeated as a picture within
that version of the materials). The same
procedure was used to construct practice
lists.
Apparatus and proc'edare. Except for a
change in instructions alerting subjects to
the mixture of words and pictures, the apparatus and procedure were identical to
those described in Experiment l.
Subjects. Twenty-four subjects performed the reality decision task, onequarter of whom saw each of the four versions of the stimulus materials. The subjects, recruited from the same subject pool
used in Experiment l, were paid $3.00 for
their participation. All subjects were
screened for normal or corrected-to-normal
OBJECTS
YES
NONWORDS
NONOBJECTS
NO
F r c . 2 . M e a n r e s p o n s et i m e s ( m i l l i s e c o n d s t)o a c c e p t
words and pictures ofobjects and reject nonwords and
nonobjects in Experiment I (pure lexical and object
decision tasks) and in Experiment 4 (mixed reality decision task).
visual acuity and for English as their native
language.
Results and Discussion
Mean responsetimes for reality decisions
are shown in Figure 2 where they are directly compared to the data obtained for
pure lexical and object decisions. Two aspects of the reality decision results distinguish them from the earlier lexical and object decision results. First, the overall response latencies were substantially longer
in the mixed task. Second,the averagetime
to accept a picture of an object as real was
approximately 47 milliseconds longer than
the time to accept a word as real, F(1 ,23)
: 50.9,p < .001,in the reality decision
task. The difference between words and
pictures held only for real concepts, howe v e r ; b o t h p s e u d o - w o r d sa n d n o n o b j e c t s
were rejected equally slowly. The average
increasein time to reject both types of distractors was approximately 150 milliseconds in reality decision compared to the
previous pure conditions. The large and
similar increase in time to reiect the two
52
KROLL
AND POTTER
types of distractors might be taken to suggest the operation of a response deadline
not present under pure conditions.
An analysis of variance performed on
RTs and errors in reality decision showed
a significant interaction between the response class (whether an item was real or
a distractor) and the modality of the item
( w o r d o r p i c t u r e ) ,F ( | , 2 3 ) : 5 . 5 9 ,p < . 0 2 5 .
A post hoc Newman-Keuls test supported
the description of the results we have already offered: real words were accepted
more rapidly than pictures of real objects,
q ( 2 , 2 3 ) : 5 . 1 2 ,P < . 0 1 , b u t t h e r e w a s n o
difference in time to reject the word and
picture distractors,q(2,23) < l. In addition,
the difference between real words and
pseudo-wordswas significant, q(3,23) :
17.7, p < .01, as was the difference between
r e a l o b j e c t sa n d n o n o b j e c t s ,q ( 3 , 2 3 ): 1 3 . 0 .
p < .01.
The error rates were low in reality decision (less than 57o) but there were some
reliable differences between conditions for
the few errors that were made. Words were
judged slightly more accurately than pict u r e s ( 2 . 7 a n d 3 . 4 % , r e s p e c t i v e l y )b u t t h e
difference was only significant when the
real items were analyzed separately,
F ( 1 , 2 3 ) : 4 . 1 3 , p < . 0 5 , w i t h 1. l c / o e r r o r s
for reaf words. and 4.37oerrors for real objects. An overall analysis of variance of errors also revealed a significant interaction
between response class (real or distractor)
and modality (word or picture), F(|,23) :
4.84, p < .05, such that real words were
judged more accurately than nonwords,
q(4,23) : 3.6, p < .05, but real objects and
nonobjectswere judged with equal accuracy.
Frequenc,v effec'ts. An analysis of frequency effects, analogous to the one performed on the pure data of Experiment 1,
revealed a frequency eff'ect of larger magnitude in reality decision (45 milliseconds
in Experiment 4 compared to 30 milliseconds in Experiment l). The frequency effect was also similar for pictures (43 millis e c o n d s )a n d w o r d s ( 4 6 m i l l i s e c o n d s ) .
The reality decision data make it clear
that decisions about words and pictures
were not based solely on a conceptual repr e s e n t a t i o ni n t h i s t a s k . R e l i a n c e o n a
common representation would not have
produced the substantial increase in responsetime under mixed conditions that
we found in Experiment 4. The overall increasein RT in reality decision is similar to
a result reported by Meyer and Ruddy
(Note 3) in a mixed bilingual lexical decision task: when German and English words
were presented in mixed orders for lexical
decision, there was an increaseof approxi m a t e l y 6 0 m i l l i s e c o n d sc o m p a r e d t o t h e
correspondingtimes when German and English words were presented separately.
Scarboroughand Gerard (Note 4) have also
reported a similar result for a mixed bilingual lexical decision task with Spanish and
English words. Interestingly, in the Scarborough and Gerard study, the time to respond to pseudo-words was markedly
longer in the mixed conditions than in the
pure conditions, like the results of Experiments l and 4 in our study.
It is important to note that the overall
increase in decision time in the reality decision task cannot be attributed to a change
in expectation resulting directly from the
mixing of pictures and words. When pictures and words are mixed in tasks that require conceptual access (e.g., categorizat i o n ) , t h e r e i s n o t a c o n s i s t e n it n c r e a s ei n
r e s p o n s et i m e ( P o t t e r , N o t e l ) . T h e i n crease in responsetime here suggeststhat
subjects cannot easily make use of a
c o m m o n c o n c e p t u a l r e p r e s e n t a t i o nw h e n
asked to verify the existence of a concept
in different surface forms.
T h e i n c r e a s ei n r e s p o n s el a t e n c i e si n
reality decision, when comparedto the
pure conditions of Experiment I, suggested
that a single conceptualrepresentationwas
not the sole basis for making a lexical or
object decision. To the extent that a lexical
decision or object decision depends on a
surfacerepresentationof a word or picture,
one would exnect some overall increase in
WORDS. PICTURES. AND CONCEPTS
response time when the two tasks are
mixed. The differential increasein latencies
for picture decisions in the context of
reality decision, however, suggeststhat the
overall increase in decision time does not
reflect a simple addition of time attributable
to switching between the two surface representations.In addition, the time to judge
both types of distractors was long (the
overall difference between yes and no responseswas 65 milliseconds in Experiment
1 , a n d 1 4 2 m i l l i s e c o n d si n E x p e r i m e n t 4 ) .
T h e r e j e c t i o n l a t e n c i e sf o r p s e u d o - w o r d s
and nonobjects were equivalent, but there
were a few more errors on pseudo-words.
What is the basis for judging pictures and
words in reality decision'lThe long decision
times to reject both types of distractors and
the increased magnitude of the word frequency effect for both words and pictures
(relative to Experiment 1) is consistentwith
an extended memory search in which negative decisions are made by default on the
basis of a deadline. What of the asymmetry
between picture and word decisions? The
r e s u l t i s p o t e n t i a l l yi m p o r t a n t s i n c e i t r e p resents the first instance in these experiments of an absolute difference for pictures
and words. Why should pictures take almost 50 millisecondslonger to judge as real
than words? If subjectswere implicitly
naming the pictures an even larger difference between picture and word latencies
would be expected; the obtained difference
was small relative to the 200-300 milliseconds additional time required to access a
picture's name (Potter & Faulconer, l9l5).
In addition, if the magnitude of the word
frequency effect can be taken as an index
of the degree of memory search, then the
almost identical frequency etfect for pictures and words suggestssimilar memory
searchprocesses.It will be seen in the next
experiment that the puzzling asymmetry
between words and pictures does not replicate. Thus it is concludedthat the apparent difference between pictures and
words in this experiment probably reflected
sampling error.
53
What is clear about the results of Experiment 4 is that the mapping of mixed surface forms onto a single decision class results in increased decision latencies.
Whether this increaseis simply due to
greater uncertainty or to distinct differencesbetween picture and word processing
remains to be seen. The similarity between
reality decisionsand mixed bilingual lexical
decisions suggeststhat the increased decision time in each is probably not due to
specific differences in surface features (pictorial vs alphabetic) but to a separationof
two representationalsystems (imaginal vs
lexical).
Experiment 5: Repetition Facilitation in
Reality Decision
Finding that lexical and object decisions
are disrupted by a mixture of surface forms
underscores the role of form-specific
m e m o r y r e p r e s e n t a t i o n sf o r t h e s e t a s k s .
This conclusiondoes not rule out a conceptual component in the decision process,nor
access to a common concept following a
form-specificdecision. ln Experiment 5 evidence was soughtfor conceptualaccessby
examining the nature of repetition effects
within and across surface forms. lf reality
decisions involve some contact with the
concept common to both surface forms,
then repetitions across surface forms
should produce some facilitation. lt was expected that repetitionswithin surfaceforms
would produce facilitation, since Scarborough et al. (1977)have shown reliable repetition facilitation when words are repeated
in lexical decision. Previous attempts to
demonstrate across-form repetition effects
have failed (e.g., Scarboroughet al.,1919),
but the failures may be attributable to the
use of tasks in which pictures and words
are not processedto the same degree (e.g.,
naming).
Method
StimLrlusmaterials. The materials were
identical to those describedfor Experiment
4. The design of the repetition sequences,
KROLL AND POTTER
however, required that eight separateversions of the materials be constructed.
Within each version there were four blocks
of 64 trials each, plus an occasional filler
item. Each block contained eight instances
of each type of stimulus (words, objects,
nonwords, and nonobjects) presented
twice. Nonwords and nonobjects were always presented in exactly the same form
on the second presentation.Words and objects were presented in the same form in
half of the repeated trials, and in the other
form in the remaininghalf of the trials.
Within each block of trials half of the repetitions occurred after a lag of 3 trials and
half after a lag of 10 trials. In a few instancesit was necessaryto add a few filler
items to extend the length of the block to
a c c o m m o d a t ea l l o f t h e r e p e t i t i o n c o n straints. The number of filler items varied
between two and three per block. The eight
versions of the materials that were constructed representeda counterbalancingof
format and order of presentation within a
repetition sequence (i.e., word-word,
word-object, object-word, object-obj e c t ) . a n d l a g ( i . e . , 3 o r l 0 t r i a l s ) .A n a d d i tional constraint was that no version contained more than two instances of a given
c o n c e p t . A p r a c t i c e l i s t c o n s i s t i n go f 2 4
trials and 3 filler trials was constructed to
representthe structure of the experimental
lists.
Apparatus and proc'edure.The apparatus
was identical to that described fbr Experiment 3. The procedure was similar to the
one described fbr the reality decision task
in Experiment 4 except that subjects were
told that they could expect to see some repeated trials. Following the reality decision
task. subjectswere given an unexpectedrecall task.
S u b j e c t s . T h i r t y - t w o c o l l e g e - a g es t u dents were each paid $2.50 to participate in
the experimenl. The subjectswere randomly assignedto one of the eight versions
of the experiment. All subjects had normal
o r c o r r e c t e d - t o - n o r m a lv i s u a l a c u i t y a n d
were native English speakers.
zro
$
o
E
=
z
'UU
F
UJ
(/) 690
z
fr oeo
z
Lll o/U
=
o^.JJR?l'o"r*.1?"*3{?,o*
r".1?t"oll?,on.
suRFAc
E
G" r"it^t"?"FBI") (saME
)
F t c . 3 . M e a n r e s p o n s et i m e s ( m i l l i s e c o n d s t)o a c c e p t
words and pictures of objects on the original presentalion and after repetitions in the same or different
sudace form. (Experiment 5)
R e s u l t sa n d D i s c ' u s s i o n
Overall, there was a significantrepetition
effect on response times for real items,
F ( 1 , 3 1 ): 3 8 . 1 9 ,p < . 0 0 1 ,a n d f o r d i s t r a c t o r s , F ( 1 , 3 1 ) : 8 9 . 0 5 ,p < . 0 0 1 .T h e a v erage repetition effect was 25 milliseconds
for the real items (713 milliseconds on the
first presentation, 688 milliseconds on the
s e c o n d ) a n d l 0 l m i l l i s e c o n d sf o r t h e d i s tractors (858 milliseconds on the first pres e n t a t i o n ,7 5 7 m i l l i s e c o n d so n t h e s e c o n d ) .
T h u s , r e a l i t y d e c i s i o n s ,l i k e l e x i c a l d e c i sions, benefit fiom multiple exposures of
the same items. The main focus of this experiment. however, concerned the relative
magnitude of the repetition etfect across
and within surface forms. Mean response
l a t e n c i e sf o r r e p e a t e d r e a l i t y d e c i s i o n s
about words and pictured objects are
shown in Figure 3. The data shown in
Figure 3 make it quite clear that the repetition effect, for both words and pictured
o b j e c t s , w a s s u b s t a n t i a l l yl a r g e r w i t h i n
form than acrossforms. An analysisof variance performed on these data indicated a
significant interaction between the type of
presentation(original or repeatedtrial), the
WORDS. PICTURES. AND CONCEPTS
surface form of the original presentation
(word or picture), and the surface form of
the repeatedpresentation(word or picture),
F(1,31)
26.82,p < .001. Post hoc
Newman-Keuls tests showed that the repetition effect was significant for both
within-form conditions; repeated words
were judged more rapidly on repeated
t r i a l s , q ( 5 , 3 1 ) : 7 . 8 9 ,p < . 0 1 , a s w e r e r e p e a t e dp i c t u r e s , q ( 8 , 3 1 ) : 1 1 . 4 7 ,p < . 0 1 .
Across surface forms there was a small
but significant repetition effect for word decisions that had been preceded by picture
d e c i s i o n sa b o u t t h e s a m e c o n c e p t , q ( 5 , 3 1 )
: 4.18, p < .05, but no repetition effect for
picture decisionsthat had been precededby
the correspondingwords, q(2,31) : 0.0.
The smaller repetition effect across surface
forms than within surface forms suggests
that the major factor contributing to repetition facilitation is surfacelevel rather than
conceptual processing. This conclusion is
supported by the results of other repetition
studiesshowing little facilitation across surface modality (Scarborough, et al., 1979),
a c r o s s a b i l i n g u a l ' s t w o l a n g u a g e s( S c a r b o r o u g h , G e r a r d , & C o r t e s e , 1 9 8 4 ) ,o r
across unrelated contexts (Carroll & Kirsner, 1982). The small repetition effect obtained when words followed pictures may
reflect conceptual facilitation. Recall that
objects produced a larger effect of semantic
facilitation in Experiment 3 than words did,
and a slightly larger repetition effect than
words did in the present experiment. Thus
pictures may produce more conceptual activation than words, which in turn might
produce an asymmetric repetition effect.
Alternatively, some small part of the repetition effect may be attributable to practice
( r e p e a t e dt r i a l s a l w a y s c o m e a f t e r i n i t i a l
p r e s e n t a t i o n s ) ,s o i t i s d i f f i c u l t t o k n o w
what to make of this cross-modality effect.
There was no overall effect of lag, f( I ,3 1)
< 1, nor was there any interaction between
lag and the repetition effect, F(1,31) :
2 . 4 4 ,p > . 0 5 .
An analysis of errors indicated that error
rates were low throuehout the conditions of
55
the experiment (average error rate was
2.8%). There were a few more errors made
on the initial presentations(3.3%) than on
r e p e a t e dp r e s e n t a t i o n s( 2 . 2 7 o ) ,F ( 1 , 3 1 ) :
5 . 4 ,p < . 0 5 .
Repetition effects for distractors. The
largest repetition facilitation in the experiment occurred for the repeated distractor
trials. The time to decide that a distractor
was not real was facilitated by repetitions;
on initial presentationsthe mean time to reject distractors was 858 milliseconds, and
on repeated trials it was 787 milliseconds.
The repetition effect was highly significant,
F ( I , 3 1 ) : 8 9 . 0 5 ,p < . 0 0 1 .F o r t h e d i s t r a c tors there was an effect of lag, F(1,31) :
4.24, p < .05, such that repeateddecisions
after a lag of three trials were approximately l0 millisecondsfaster than repeated
decisionsafter a lag of ten trials. There was
also a significant main effect of distractor
m o d a l i t y ,F ( 1 , 3 1 ) : 5 . 2 5 , p < . 0 5 , s u c h t h a t
pseudo-words were rejected more rapidly
than nonobjects (813 and 832 milliseconds.
re spectively).
The overall error rate for distractors was
5.4Vc.There were no significant differences
in error rate for any of the experimental
conditions.
Data from the surprise recall task will be
reported in the general discussion.
The repetition eJfectand reality decision.
The results of Experiment 5 show that
reality decisionsare facilitatedby prior presentationsof identical words and pictures.
Although the results of Experiment 4 sugg e s t e d t h a t r e a l i t y d e c i s i o n sw e r e s o m e what different from separate lexical and obj e c t d e c i s i o n s( o v e r a l lr e s p o n s el i m e s w e r e
longer in the mixed conditions and there
was a word advantagefor real items), the
finding of a strong repetition effect, largely
attributable to facilitation of the identical
perceptualforms, is consistentwith previous studies of the repetition effect in lexical decision. In addition, the asymmetry
between positive responsesto words and
pictures was marginal or nonexistent in the
conditions of the present experiment.
WORDS. PICTURES. AND CONCEPTS
actions between surface forms and processing could provide such definitive evidence? Probably the strongest case can be
made for comparisonsinvolving across and
within-form transfer. lf the processingtask
under consideration is based on a single
idea or concept underlying both words and
objects, then there should be little consequence for processingwhether the surface
forms are mixed or blocked. That is. the
u l t i m a t e f o r m o f r e p r e s e n t a t i o n ,r e s u l t i n g
from having seen a picture of an object and
the nature ofprocessing that representation
(be it facilitatory or inhibitory), should also
result from having seen the word that
names the pictured object. In addition, if
processingis based on a common concept,
it should be possible to mix not only pictures and the words that name them, or
words in one language with words in another, but all three, with little or no decrement in performance.
Where do these considerations leave us
i n i n t e r p r e t i n gt h e r e s u l t s o f t h e e x p e r i ments we have presentedhere? The results
of Experiments I and 2, showing parallels
between the two tasks, may be regardedas
n e c e s s a r yb u t n o t s u f f i c i e n t e v i d e n c e f o r
the conceptual model. The results are also
compatiblewith a dual-codemodel (e.g..
Paivio, 1978), as long as a reasonable ac"coincidental"
count of the
similaritiescan
be given. Likewise, the finding of semantic
facilitation in Experiment 3 for each of the
t a s k s i s c o n s i s t e n tw i t h a c o m m o n - c o d e
model, but not definitive support for one.
A stronger test of semantic facilitation efi e c t s w o u l d b e t o l o o k a t p r i m i n g a c r o s s .a s
well as within. surface forms. Vanderwart
(1984) has recently performed such a test
for lexical decisions. ln her study, lexical
decisionswere precededby picture or word
primes that were related or unrelated to the
target words. She found strong priming effects for both types of primes and interpreted her results as support for reliance on
a common conceptual code in lexical decision.
We have said that the strongestevidence
fbr or againsta common-code model should
57
be obtained when surfaceforms are mixed.
The reality decision task, in which lexical
and object decisions are mixed, provides
such a test. The results of the two reality
decisionexperiments(Experiments4 and 5)
show clear evidence for at least partial reliance on form-specific codes: in both experiments decision times were elevated relative to the pure decision times of the first
experiment, and in Experiment 5 there was
little evidence (except for the case in which
words were precededby pictures)for a repetition effect across forms. While these experiments suggest reliance on a form-spec i f i c r e p r e s e n t a t i o n ,t h e y d o n o t a l l o w a
precise evaluation of the degree to which
the task involves a conceptualcode as well.
Semantic Proc'essingin the Lexic'al
Decision Task
What constitutes good evidence for one
r e p r e s e n t a t i o n a lm o d e l o v e r a n o t h e r ? A t
the broadest level an answer to this quest i o n r e q u i r e s a c o n s i d e r a t i o no f t h e a s sumptions made in using a particular task
as well as the assumptionsunderlying particular experimentalmanipulations.For example, we were motivated to devise the object decisiontask as a way to study semantic processing for pictured objects
b e c a u s ew e a s s u m e dt h a t t h e a n a l o g o u s
l e x i c a ld e c i s i o nl a s k w a s s e n s i t i v et o c o n ceptual factors, an assumption which has
recently been challengedas being too
s t r o n g ( e . g . . C a r r o l l & K i r s n e r , 1 9 8 2 ;K o r i a t , l 9 8 l ) . A l t h o u g h l e x i c a l d e c i s i o n sc a n
be facilitated by semantically related
p r i m e s , K o r i a t ( 1 9 8 1 )h a s s h o w n t h a t t h e
degree of semantic relatednesshas no effect on the magnitude of the semantic
p r i m i n g e f f e c t , s u g g e s t i n gt h a t t h e p r i m e
may influence a decision process after the
target is presentedand partially processed,
rather than biasing the course of target processingitself. Thus initial processingcould
be form-specific before the hypothesized
common code is retrieved. On this view. we
would not have been surprised by our res u l t s , s h o w i n g p r i m a r y d e p e n d e n c eo n
f o r m - s p e c i f i c r e p r e s e n t a t i o n s ,n o r w o u l d
KROLL AND POTTER
we have been surprised by the results reported by Scarborough et al. (1984)
showing that bilingual subjects seem to
keep their two languagesseparatein a lexical decision task.
One proposal concerning lexical decisions is that there are two basesfor a lexical
decision, one based on perceptual recognition, and a second based on meaning (Jacoby & Dallas, 1981).To the extent that a
lexical (or object) decision is based on
meaning, one would expect to find similar
results for pictures and words and for
words in two different languages.To the extent that a lexical (or object) decision is
basedon what Jacoby and Dallas call "perceptual fluency" one might expect to see
little evidence of conceptual transfer from
one form to the other.
Inciclental Rec'allJoltow,ingLexic'al,
objec't,and Reatitl,Dec.isions
A s a f i n a l w a y o f e v a l u a t i n gt h e l e v e l o f
processingin the present decision tasks, we
looked at incidental memory performance
following completion of the reaction time
tasks in Experiments 3 and 5. The results,
like those reported by Carroll and Kirsner
(1981)and Jacoby and Dallas (1981). show
a dissociation between the effect of certain
variableson lexical or object decisions and
later recall.
In E,xperiment3 separategroups of subjects performed lexical or object decisions
in which pairs of related or unrelated items
had to be judged simultaneously.There was
asemanticrelatednesseffectforbothtasks,
although the magnitude of the effect was
larger in the object decision. ln addition, an
analysis of the contribution of associative
strength to the relatedness effect yielded
negative results; the magnitude of the relatedness effect was as large for pairs related only by virtue of membership in the
same superordinate category as for pairs
that were also highly associated.The incidentalrecall data, shown in Table6, tell a
very different story. For both the lexical
and object decision tasks, episodic memory
f o r w o r d s a n d o b j e c t sw a s e n h a n c e db y t h e
presenceof an rzssociativerelationshipbeyond semantic relatedness.An analysis of
variance on these data revealed a significant effect of task, F(I,30) : 12.21,p <
.01, with object decision producing superior recall to lexical decision (19 and 12%,
respectively). There was also a significant
effect of relatedness over all conditions,
F ( I , 3 0 ) : 3 6 . 0 8p
, <.01, withberterrecall
for items appearing in a related context
than for those appearingin an unrelated
context (21 and 1070, respectively). of
greatest interest, however, was the significant effect of association,F(1,30) : 87.4,
p < .01, and the significant interaction between the effect of relatednessand associa t i o n , F ( 1 , 3 0 ) : 1 5 . 3 8 ,p < . 0 1 . F o r i n c i dental recall following both lexical decision
und object decision' memory for items ap-
was greatlvenflli1t'.t.l
hanced by ::1"::1-pairs
the presence of an associative
relationship, q(2,30) : 8.38, p < .01.
Memory for items appearing in pairs that
were semantically related only was no
better than memory for items appearing in
unrelated pairs. These data show that the
associationvalue played an influential roie
in episodic memory for words and pictures,
although it did not seem to influence semantic priming in the reaction time task. ln
addition, although we found an overall picture superiority effect (recall was superior
following object than lexical decision), the
associationeffect did not interact with the
modality of the task. The independenceof
the picture superiority effect and the semantic relatedness effect in memory has
been reported in a number of other studies
in which incidental memory tasks follow
lexical decisionor naming of words and pictures (vanderwart, l9g4; Kroll & potter,
Note 5). lt suggeststhat pictures and words
share the same conceptual representations,
since the relatednesseffects presumablyreflect the nature of the relationships represented in conceptual memory. Superior
memorv for pictures is more likelv to be a
59
WORDS, PICTURES, AND CONCEPTS
TABLE 6
MraN PEnceNr.rce
Recall FoLLowtNG
Lnxtcel on Os:EcrDEcrsroxrN Exppntttnnl3
Semantically related
and associated
Lexical
Object
Semantically related
onlY
Related
Unrelated
Diff.
Related
Unrelaled
Diff
27
32
6
t5
2l
t7
9
l6
5
l4
4
2
consequenceof some aspect of the surface
representation for pictures, than a reflection of semantic encoding.
In E,xperiment 3 we also looked at
memory for real items when they appeared
in the context of mixed negatives.Because
overall recall was so poor (less Ihan 47c)
these data were not analyzed in greater detail. The fact that recall was poor supports
the conclusion that filtered items received
shallow processing.
The incidental recall data collected following the repeatedreality decision task in
Experiment 5 also produced results that differed from the reaction time data. The data
are given in Table 7. The recall measure
does not distinguishbetween the initial and
repeated presentation of a stimulus. With
the simple recall task we used, it was impossible to tell which of the two stimulus
presentationsthe subject was recalling.The
measure is thus a reflection of conceptual
recall as a function of the modality of the
stimulus presentationson the initial and repeated trials. The aspect of interest here of
these data is that the form of the repetition
influenced episodic memory in a different
way than it influenced reality decision latencies. Reality decisions were facilitated
by within-form repetitions, but not significantly by across-form repetitions. Recall,
however, was best following mixed repetitions, intermediatefollowing repetition of a
picture, and worst following repetition of a
word. The effect of the form of the repetition was significant, lq'(l,31) : 30.08,p <
.01. Post hoc Newman-Keuls tests revealed significant differences between pure
word and picture repetitions, q(2,31) :
4 . 2 2 , p < . 0 1 , b u t n o t b e t w e e nt h e t w o
m i x e d r e p e t i t i o n s ,S Q , 3 l ) : 2 . 2 5 ,p < . 0 5 .
The mixed repetitions produced better
memory than the pure repetitions in all
comparisons except those between pure
picture repetitions and words followed by
pictures.
Overall. the results of the incidental
memory task show that recall of words and
pictures was influenced by conceptual factors in similar ways; memory for both surface forms was a function of semantic relatednessand associationvalue of the context in which they were seen. The finding
that mixed repetitionswere remembered
more often than identical repetitions is
more ambiguous. It could mean that pictures and words were ultimately processed
to a conceptual level, or alternatively,that
two separatecodes were independentlyret r i e v e d . F i n a l l y , t h e i n d e p e n d e n c eo f t h e
picture superiority effect and conceptual
processing suggeststhat the source of superior memory recall for pictures lies in the
nature of the surface representation for pictures rather than in an elaborated conceptual representation.
TABLE7
MsaN PenceNrace Rscall FoLLowrNG REaLrlv
DEcrstoNs ttr ExpgrlrvtENr 5
Percentage
Within-formrepetitions
Word-word
Picture-picture
Across-formrepetitions
Word-picture
Picture-word
l5
.J
27
3l
60
KROLL
AND POTTER
CoNcr-usroNs
This paper reports a set of experiments
which compare the processingof words and
pictures in lexical, object, and reality decision tasks. The results show that object
decisionsare similar to lexical decisions
when the tasks are performed separately,
but that mixing the two tasks into a single
reality decision task greatly reducesprocessingspeed. The results do not offer unambiguous support for either the dual-code
model or the common-code model. They do
suggest,however, that there is an important
form-specific component in lexical, object,
and reality decisions, and they point to an
important qualification of the logic whereby
semantic priming effects are automatically
taken as evidence for reliance on concept u a l p r o c e s s i n g .B e c a u s e t h e s e d e c i s i o n
tasks appear to be sensitive to both formspecific and conceptual factors, further
studies comparing pictures and words in
such tasks may help to determine the precise nature of form-specific memory representationsand their relation to concepts.
A p p E N n r xA : N o N o e l E c l R a r t N c s
A groupof 100undergraduates
wasgiventhetask
of rating the degree to which the nonobjecls used in
o u r s t u d y r e s e m b l er e a l o b j e c t s . S u b j e c t st o l d t o r a t e ,
u s i n g a 7 - p o i n t s c a l e ,a n o n o b j e c t a s a " l " i f i t l o o k e d
very much like a real object and as a "7" if it looked
nothing like a real object. lntermediatevalueswere to
b e a s s i g n e dt o i n t e r m e d i a t ej u d g m e n t s . T h e c o m p l e t e
set of 88 nonobjects was reduced and printed on a
singlepage. Subjects placed their ratings next to the
picture of the nonobject.
The average ratings for each of the nonobjecls is
g i v e n i n T a b l eA - l . T h e a v e r a g er e s p o n s et i m e t o r e j e c t
each nonobject is also given for the pure object decision experiment(Experiment l) and for the reality dec i s i o n e x p e r i m e n t ( E x p e r i m e n t 4 ) . P i c t u r e so f t h e n o n objects are provided in Figure A-1. The first 60 pict u r e s a r e o r d e r e d a c c o r d i n g t o t h e r e s p o n s et i m e s
obtained in Experiment l. Note that only the firs1 60
p i c t u r e s w e r e u s e d i n o u r e x p e r i m e n t s ;t h e r e m a i n i n g
28 pictures were either discardedor used in practice
s es s i o n s .
T o s e e w h e t h e r t h e r e w a s a r e l a t i o n s h i pb e t w e e n t h e
nonobject ratings and rejection latencies in Experiments I and 4 a set of correkitionswas computed. ln
the pure condition of the object decision in Experiment 2, the correlation between reiection time and
rated object similarity was /' .31,p <.025. In the
mixed condition of the reality decision in Experiment
4. the correlation between the rejection time and rating
w a s r : - . 5 2 , p < . 0 1 . T h e s i g n i f i c a n tn e g a t i v e c o r relations suggest that nonobjects that resemble real
objects were more difficult to reiect as distractors in
object decision.
AppENorxB: SsunNrrc
Rp,larpoNEssRerrNcs
In Experiment 3 related pairs were formed by
pairing coordinate members of the same superordinate
category. Unrelated pairs were formed by repairing
related items with items from a different superordinate
category. The items were then rated by an independent
group of subjects. The reasons for obtaining ratings
for the related and unrelated pairs were (l) to check
the degree to which the related and unrelated pairs
were discriminable(since we had paired them intuitively) and (2) to determine the influence ofthe surface
form of the items in a rating task.
An initial set of 64 related and 64 unrelated pairs
was rated for sematic relatedness. Subjects were instrucled to rate the pairs on a 7-point scale on which
"l"
a rating of
meant closely relatedand "7" meant
unrelated. Each of four different forms of the scale.
corresponding to all possible combinations of word
a n d p i c t u r e p a i r s . w a s a d m i n i s t e r e dt o 2 5 s u b j e c t s .T h e
100 subjects who participated in the study were stud e n t s i n a n i n t r o d u c t o r y p s y c h o l o g y c l a s s .T h e y w e r e
randomly assigned to one of the four versions of the
m a t e r i a l s :( l ) w o r d p a i r s ; ( 2 ) p i c t u r e p a i r s ; a n d ( 3 ) a n d
( . 1 )m i x e d p i c t u r e - w o r d p a i r s .
T h e o v e r a l l r e s u l t s o f t h e r e l a t e d n e s sr a t i n g s a r e
s h o w n i n T a b l e B - 1 . W i t h r e g a r dt o t h e d i s c r i m i n a b i l i t _ v
of the related and unrelated pairs, it was clear that
subjects agreed with our intuitive categorizations. The
distributions of ratings for related and unrelaled pirirs
w e r e a l m o s t n o n o v e r l a p p i n g .A n a n a l y s i s o f v a r i a n c e
of these data also indicated a significant effect of the
m o d a l i t y o f t h e r a t e d i t e m s , F ( 3 . 3 7 U ) - 1 3 5 . 4 .p <
.001. and a significant inleraction between the mod a l i t y a n d s e m a n t i cr e l a t e d n e s so f r a t e d i t e m . F ( 3 . 3 7 t t )
: 1 6 . 9 3 ,p < . 0 0 1 . T h e i n t e r e s t i n g r e s u l t w a s t h a t
picture pairs,overall,were ratedas beingmore closely
related to one another than word pairs or word-picture mixtures. A comparison of the pure word and
pure picture ratings showed that the higher perceived
r e l a t e d n e s sf o r p i c t u r e s h e l d f o r b o l h r e l a t e d , q \ 2 , 3 7 8 )
: 1 5 . 2 .p < . 0 1 . a n d f o r u n r e l a t e d p a i r s . q ( 2 . 3 7 8 ) :
U . 6 .p < . 0 1 .
The rating data we have described served to validale
our intuitions about related and unrelated word and
picture pairs. ln addition, however, we observed an
unexpected interaction, with the surface form being
rates such that pictures were consistentlyrated as
more closely related to each other than words. One
6l
WORDS. PICTURES. AND CONCEPTS
TABLE A.I
MEIN Osrsct StulLanlrv RATrNcs r.ox e SEr or 88 NoNonlncrs AND Conr.espoNorNcRsspoNsn Truss to
Rugcr Tuoss NoNog.rrcrs es Drsrnecrons UNDERPURE(ExpERIMENT1) AND Mtxm (Exrrr.rnrNr 4)
DgctsloN CclNnlrloNs
Mean RT
Mean RT
Nonobject
b
Mean
rating
3.6
5.9
5.3
5.5
5.6
5.6
1
8
9
l0
ll
t2
13
l4
l5
l6
t7
18
l9
20
2l
22
l. /
25
26
27
28
29
30
4.9
2.8
5.9
5.2
o./
3.6
3.9
4.3
4.9
5.1
3.8
3.'7
J.O
5.2
4.8
3.3
4.0
Expt I
557
564
56,5
586
59'1
-598
602
602
605
609
616
6t'7
621
625
625
625
627
637
643
645
645
6-50
652
657
658
660
664
666
666
6'7|
Expt 4
153
712
749
748
767
826
833
715
810
875
823
965
975
167
ti00
810
885
857
844
915
961
732
786
879
80s
876
819
859
892
Nonobjecl
3l
33
34
35
36
37
38
39
40
4l
11
Mean
rating
5.2
,5.9
4.4
1.7
4.tt
3.9
3.1
5.6
) . J
5.9
2.9
4.0
44
45
46
2
A1
l
I
46
49
-s0
5t
52
53
54
5-5
56
57
58
59
60
5.5
3.0
3.3
3.4
5.6
3.6
5.3
3.1
2.8
Expt 1
Expt 4
673
680
680
682
682
683
688
689
690
690
694
695
712
713
719
121
722
12s
745
746
750
751
763
764
789
79'7
803
804
810
8r9
808
733
932
t076
8r8
806
1 0 3I
190
862
762
840
794
810
894
896
971
866
889
926
844
884
813
922
883
876
785
I 030
921
869
958
Mean RT
Nonobject
Mean
rating
6l
62
1
OJ
6
6
64
65
66
67
68
69
70
7l
72
73
1t
l5
16
77
18
79
80
8l
82
83
84
85
86
87
88
Expt I
Expt ,1
A
2.9
3.1
4.3
3.7
1.6
4.6
4.4
2.5
1.6
3.3
,5.0
2.4
3.4
802
4.8
5.1
4.5
6.1
5.9
3.7
2.3
Nota. RTs are in msec
possible interprelation of these results is that pictures
can be related on a greater number of dimensions than
their word labels. Rating of words may represent a
prototypical view of the concept, while ratings of picTABLE B-I
Mr.c.N Rrr-ererNEss RATINGSFoR 64 RELATEDAND 64
UNnEI-argo Perns as a FuNcrtoN op rug MooeLttv
op PnEsENte'rlox
Modality of
presentation
Pure
Words
Pictures
Mixed
Version '
Version,
Mean
Related
palrs
Unrelated
pairs
2.35
1.75
6.42
6.08
2.45
2.86
2.66
6.41
6.57
6.52
Note. Twenty-five subjects contributed ratings to
each of the four versions of the materials. The pairs
"1"
were rated on a 7-point scale, with a rating of
"7"
being unrelated.
being closely relatedand
tures may also representa particular senseof the
meaning of the concept. Interestingly, observers were
able to find some similarities even among unrelated
picture pairs that were easily discriminated from related pairs. An alternative dimension for picture ratings may have been scenic coherence as well as semantic relatedness. For example, when presented with
a picture of a duck and a picture of a kite, the observer
might have judged whether the two things could be
v i e w e d i n t h e s a m e s c e n e( e . g . , a p a r k s c e n e ) ,a s t h e s e
two things might.
The materials selected for use in Experiment 3 are
shown in Table B-2. The first 6 items shown for each
version of the materials were highly associated pairs
taken from the Postman and Keppel (1970) norms for
word association. The criterion for high association
was that the second item of each pair had to be a
primary response to the first item with a relative frequency of .20 or greater. The remaining l8 items for
62
KROLL
@
b
1
2
ss
/\
4
q\5
U
3
\
AS=.
x.
\9
@
5
6
\^\Y)
'fl.,
I
$,
\D| /
10
11
t
-
&
0r/
13
14
w
4
16
17
n
25
<./;
.-)
27
26
/)
/(
ll
w
P
28
29
30
n
v
w
31
32
33
g{N
e
\)
N
12
34
35
-f,*L
t\t( l.1
a
E\
%r"
15
a
18
{l_A
\)
Hk3
19
20
@
M
@
22
23
24
Frc. A-l.
*q
4bz
9
@
@
[r--\
@
21
ft.
tr'r-)
,"\
=.
7
l
/t^5
\
tt\
^
\L-\\
\\5f-\
\nvd_
f/lr.vA
'r r"r'.1
\-nn l
AND POTTER
,f
lK
'\r
36
,l4q
€
@
37
;;:l
'?aa
40
i-,
1 P +
i rt ,"'
\""1
43
39
NN
\\[\\
l$r$'
^41
@
42
I
ffi
44
45
rA-l
,F.:
46
;-b-.
"'iJ
47
48
W
P i c t u r e so f 8 8 n o n o b j e c t su s e d a s d i s t r a c t o r s t i m u l i i n t h e o b j e c t d e c i s i o nt a s k . T h e f i r s t
WORDS. PICTURES.AND CONCEPTS
ffi
49
@
P
@
50
51
a-
ffi
53
52
tr
&
<a
58
wl
6
61
-#
4\,+
"z
59
60
G
.
-t<--'
q
73
74
75
,1,*
ffi
76
77
B
_iD-ffi
79
80
62
63
4tt"
l^
-,4
:4X].
82
i-1
\:
78
ffi\
{$q
81
'a')
t7
2 ,
"?)
ire
tfr.tQ)
83
84
(
-
h\
s$
i/ \)
/a\,
85
86
87
N./
@
€tr
@
65
66
88
Q6J
A
,#eP
-\-'r
I
A
w
<1/_/
56
55
6
@
54
fhm
w
N
^&
63
0.q
VN
68
67
l^
WM ffi\
l_,/
70
71
@JiNr
72
60 nonobjects (with a single exception noted in Table A-l) were used in the experiments reported
here.
Ftc. Al-Continued.
64
KROLL
AND POTTER
TABLE B-2
Merunrels Usnu rr.rExpEruugNt3: Posrlrvp Rnlerno lrplrs ONly
Related and Associated Pairs
VersionB
VersionA
Associative
strength
Pair
43.4
52.2
56.0
28.6
34.0
20.0
leaf-tree
s o c k s- s h o e s
thread-needle
hand-gloves
lruck - car
coat-hat
Pair
dog-cat
chair-table
door-window
arm-leg
ostrich -feather
shirt - tie
Associative
strength
64.2
43.0
40.0
32.0
29.0
23.0
SemanticallyRelatedbut N o t H i g h l y A s s o c i a t e dPairs
VersionA
Version B
Pair
cake-pie
apple-banana
kettle - coffeepot
globe-map
kangaroo-monkey
stove-toaster
carrot- corn
shovel- wheelbarrow
sheep-pig
church-house
harp-drum
pear- strawberry
snail -frog
zebra-elephant
butterfly- spider
lion-giraffe
fork-knife
boat-anchor
Subjective
relatedness
Pair
1.52
2.00
1.36
1.56
2.56
2.48
2.60
2.80
2.20
3.00
2.80
chicken-turkey
cow-horse
television- radio
hammer- screwdriver
comb-brush
bread-cheese
book-pencil
kite-balloon
bicycle - rollerskate
ear-eye
piano-violin
1 1/
slingshot-gun
doll-teddybear
igloo-tent
stairs- ladder
deer-rabbit
barn-windmill
pliers -file
2.84
2.64
2.44
2.10
1.40
2.20
each version were semantically related pairs selected
from the larger set of related items that had been rated
for subjective relatedness.
REpgRpNcgs
BaNrs. W. P., & Flona, J. Semantic and perceptual
processes in symbolic comparisons. Journal of
Erperimental Psychology : Human Perception and
PerJbrmance, 1917, 3, 278-290.
C e n n o l l , M . , & K I R s N E R ,K . C o n t e x l a n d r e p e t i t i o n
effects in lexical decision and recognition
memory. Journal oJ Verbal Learning and Verhal
B e h u v i o r , 1 9 8 2 ,2 1 . 5 5 - 6 9 .
Subjective
relatedness
t.44
1.96
1.64
1.92
1.24
2.40
2.96
2.72
2.76
1.92
2.04
2.80
1.88
2.08
1. 8 0
2.36
3.56
3.00
Cerrgt-r-, J. M. The time it takes to see and name
objects.Mind, 1886.11, 63-65.
C o L l ' u r e n r , M . , D e v r l e e r , E . , J o N N e s o N ,J . T . , &
B E S N E RD
. . Access to the internal lexicon. In S.
Dornic (Ed.), Attention and performance VI.
Hillsdale, N.J.: Erlbaum, 1977.
D e v , J . R i g h t - h e m i s p h e r el a n g u a g e p r o c e s s i n g i n
normal right-handers. Journal of Experimental
Psychology: Humon Perception and PerJorm a n c e . 1 9 7 7 .3 . 5 1 8 - 5 2 8 .
Dunso, F. T., & JosNsor, M. Facilitation in naming
and categorizing repeated pictures and words.
JournaL oJ Experimental Pscl-hology: Human
Learning and Memory, 1979, 5, 449-459.
WORDS. PIC-|URES. AND CONCEPTS
Frscsr-pn. I. Semantic facilitation without association
in a lexical decision lask. Memon & Copnition.
1 9 7 7s,, 3 3 s - 3 3 9 .
FonsrEn,K. I., & CuRustns, S. M. Lexical access
and naming time. Journul o.f Verbal Learning and
Verbul Behavior, 1973. 12, 627 -635.
FnnolrrxsnN, J. R., & Knoll, J. F. Spelling and
sound: Approaches to the internal lexicon.
Journal of Erperimentul Ps1-chology:Humon Pert e p t i o n u n d P e r J b r m a n c e ,1 9 7 6 , 2 , 3 6 1 - 3 7 9 GulN'tttER. R. K. & Klerzrv. R. L. Semantic classification of pictures and words. Jourrutl o.f Exp e r i m e n t a l P s y c h o l o g y :H u m u n L e u r n i n g a n d
M em o r y - ,1 9 1 7 ,3 , , { 9 8 - 5 1 1 .
HuttENt-ocsen, J., & KustcEK. L. F. The source of
relatedness effecls on naming latency. Journal o.l'
E r p e r i m e n I a l P s 1<' h: o I o g t : L e u r n i n g , M em o r y ,
a n d C o g n i t i o n , 1 9 8 3 .9 , . { 8 6 - 4 9 6 .
J e c o s v , L . L . , & D e n . , q s , M . O n t h e r e l a t i o n s h i pb e tween autobiographical memory and perceptual
l e a r n i n g . J o u r n a l o f E x p e r i m e n t u l P s 1 , < ' h o l o g: , G e n e r a l , 1 9 8 1 ,1 1 0 , 3 0 6 - 3 4 0 .
J , r v E s . C . T . T h e r o l e o f s e m a n t i ci n f o r m a l i o n i n l e x ical decisions. Journal of Experinental Psycholog,t: Hunan Perception and Performance,
1 9 7 5 . 1 ,1 3 0 - 1 3 6 .
KrnsNEn. K., & Srrlnu, M. Modality effects in word
identifi cation. M emory & C o gnit ion, 1974, 2, 637640.
Konrer. A. Semantic facilitation in lexical decision as
a function of prime-target association. Memory& C o g n i t i o n , 1 9 8 1 , 9 ,5 8 7 - 5 9 8 .
K u e r r e , H . , & F n a N c t s , W . N . C o n r p z t a t i o n a la n a l American English. Providence,
l sis o.fpresent da-uR.l.: Brown Univ. Press. 1967.
MEyrn, D. E., & ScsveNEvpr-or. R. W. Facilitation
i n r e c o g n i z i n gp a i r s o f w o r d s : E v i d e n c e o f a d e pendence between retrieval operations. Journal o.f
Exp er i me nt al P s1,
c ho lo g-,-, I 97 l, 90, 221-234.
Nrnt-y, J. H. Semantic priming and retrieval from lexical memory: Roles of inhibitionless spreading activation and limited-capacity attention. Journal oJ
E -rp er i m en t a L P s1-c h o I o g."-: G en er a l, 1977, 106,
114
1<1
NllsoN. D. L.. Rpso. V. S., & McEvov. C. L.
Learning to order pictures and words: A model of
sensory and semantic encoding. Journal of Erp e r i m e n t a l P s . v c h o l o g l -H: u m a n L e a r n i n g a n d
Memory, 1977. 3, 485-497.
Petvto, A. Neomentalism. Canadiun Journal of Psyt holo gy, 1975, 29, 263 -291.
Pervro, A. A dual coding approach to perception and
cognition. in H. L. Pick and E. Saltzman(Eds.),
Modes oJ'perceiving and processing information.
Hillsdale, N.J.: Erlbaum, 1978.
PosrrueN. L.. & Krppgl. G. Norms or u,ord associr
llon. New York: Academic Press. 1970.
Porrrn, M. C., & FeulcoNen, B. A. Time to understand pictures and words. Nature (London) 1975,
253,437-438.
65
Porren. M. C. Mundane symbolism: The relations
among objects,names, and ideas.In N. R. Smith
& M . B . F r a n k l i n ( E d s . ) , S , v n b o l i c. f u n c t i o n i n gi n
childhood. Hillsdale, N.J.: Erlbaum, 1979.
PorrER, M. C., ValreN, V V, & FeulcoNln, B. A.
Representation of a sentence and its pragmatic implications: Verbal. imagistic, or abstract'! Journal
of Verbal Learning and Verbal Behayior, 1977.
16, l-12.
R o s r N s r r , R . R . , G o r - r N x o r - r ,R . M . . & K u r r s H . K . S .
Automatic semantic processing in a picture-word
interference task. Child Dey'elopment. 1975. 46.
247 -253.
R u s E N S r e r NH, . . L E w r s .S . S . , & R u s E N s r E r NM
, . A.
Evidence for phonemic recoding in visual word
recognition. Journal of Verbal Learning and
Verbal Behavior, 197l, 10, 615-657.
R u s E N s r E r N ,H . . R r c u r * , M . L . , & K e y , E . J . P r o nounceability and the visual recognition of nonsense words. Journal o.f Verbal Learning and
V e r b a l B e h a v i o r , 1 9 7 5 ,1 4 , 6 5 1 - 6 5 7 .
Scannor.oucs, D. L., CoRTESE,C., & Sc.cnuoRoucH.
H. S. Frequency and repetition effects in lexical
memory. Journal of Experimental Ps,tchologt:
H u n t a n P e r c e p t i o n a n d P e r f o r n a n c e , 1 9 7 7, 3 ,
t-17.
S c e R . s o n o u c HD
, . L., Gsreno, L., & Conrpsn, C.
Accessing lexical memory: The transfer of word
repetition effects across task and modality.
Menort & Cognition, 1979,7, 3-12.
S c a n s o r o u c H . D . L . , G E R A R DL, . , & C o n r r : s E ,C . I n dependence of lexical access in biligual word recognition. Journal of VerbaL Learning and Verbal
Behavior. 1984. 23. 84-99.
S c u u l u a u , H . G . , & D A V r s o N ,T . C . B . C o n t r o l p r o p erties of semantic coding in a lexical decision task.
Journal ofVerbal Learning and Verbal Behayior,
1 9 ' 7 7 ,7 6 , 9 t - 9 8 .
Slttru, M. C., & Mecrs, L. E. Tracing the time
course of picture-word processing. Journal of Erperimental Psycholog-,-: Generul, 1980, 109, 373392.
SNoocr.ess, J. G. Toward a model for picture and
word processing. In P A. Kolers, M. E. Wrolstad, & H. Bouma (Eds.), Processing of visihle
LangLrage.New York: Plenum, 1980. Vol. 2.
SNoocness, J. G. Concepts and their surface representations. Journal o.f Verbal Learning and
V e r b a l B e h a v i o r , 1 9 8 4 ,2 3 , 3 - 2 2 .
S p n n s n r ,R . D . , M c C A U L r v , C . , R a c a r r , R . , & W s r t - ,
C. M. Semantic priming effects on picture and
w o r d p r o c e s s i n g .M e m o r y - & C o g n i t i o n , 1 9 1 9 , 1 ,
3 3 9- 3 4 5 .
VaNostwant, M. Priming by pictures in lexical decision. Journal of Verbal Learning and Verbal Behavior. 1984. 23. 67 -83.
W a x n n N , C . , & M o n r o N , J . T h e e f f e c t so f p r i m i n g o n
picture recognition. British Journal of Ps,tc h o L o g y, 1 9 8 2 ,7 3 , l l 7 - 1 2 9 .
66
KROLL
AND POTTER
Rsnr,Rg,Nce,
NorEs
P o r r r n , M . C . U n p u b l i s h e dm a n u s c r i p t ,M a s s a c h u s e t t s I n s t i t u t e o f T e c h n o l o g y ,1 9 7 5 .
2 . Knoll, J. E, SupnaNtr, A., & Mrnvls, J. S. texi c a l a c c e s sf o r ( o n ( r e l e a n d a b s t r a c t n , o r d s . U n published manuscript, Mount Holyoke College,
l 982.
MEyen, D. E., & RuDDy, M. G. Bilingual u'ordre<;ognition: Organizcttion dnd retrierdl of alternative lerical codes. Paper presented at the
meeting of the Eastern Psychological Association,
Aprll 1974.
4. ScansoRoucH, D. L., & Gsnarl, L. Language
protessing in bilinguals. Paper presented at the
meeting of the Eastern Psychological Association,
April 1981.
5. Krorr, J. F., & PorrER, M. C. Can pictures and
v,ords prime each other? Paper presented at the
eighteenth annual meeting of the Psychonomic Society, November 1977.
( R e c e i v e dM a v 1 9 8 2 )