Under the theoretical assumption that lexical ambiguity is not a homogeneous phenomenon,
but rather that it is subdivided into two distinct types, namely homonymy and polysemy, the
present study investigated whether these different types of lexical ambiguity are psychologi-
cally real. Four types of ambiguous words, homonymous words (e.g., pen), polysemous
words with metaphorical extensions (e.g., eye), polysemous words with a count/mass met-
onymic extension (e.g., turkey), and polysemous words with a producer/product metonymic
extension (e.g., Dali), were used in a cross-modal sentence-priming lexical decision task.
Overall, the theoretical distinction between homonymy and polysemy was reflected in the
results of the present study, which revealed differential processing depending on the type of
ambiguity. 2001 Elsevier Science (USA)
Key Words: lexical ambiguity; homonymy; polysemy; metonymy; metaphor; cross-modal;
sentence priming.
INTRODUCTION
The research reported in this article was supported by an MCRI (Major Collaborative Research Initia-
tive) grant from the Social Sciences and Humanities Research Council of Canada (412-95-0006) awarded
to Gonia Jarema (Universite de Montreal), director and coprincipal investigator; Eva Kehayia (McGill
University), coprincipal investigator; and Gary Libben (University of Alberta), coprincipal investigator.
I am grateful to Shari Baum, Brendan Gillon, and Eva Kehayia for their insightful contribution and their
helpful suggestions. I also thank the reviewer for helpful comments and suggestions. This work was
completed while the author was at the Department of Linguistics, McGill University.
Address correspondence and reprint requests to Ekaterini Klepousniotou, School of Communication
Sciences and Disorders, McGill University, 1266 Pine Ave. West, Montreal, Quebec, H3G 1A8 Canada.
E-mail: eklepo@po-box.mcgill.ca.
1
Homography is the case where two words have the same written form but different pronunciation
and meaning. For example, wound 1 is the past tense form of the verb to wind, and wound 2
means a sore/cut. Homophony is the case where two words are pronounced the same but they have
different written forms, as, for example, red which refers to a color and read which is the past
tense form of the verb to read. Finally, homonymy is the case where both the pronunciation and
written form of two words are the same but they have distinct and unrelated meanings, as, for example,
pen 1, which means a writing device, and pen 2, which means an enclosure.
205
0093-934X/01 $35.00
2001 Elsevier Science (USA)
All rights reserved.
206 EKATERINI KLEPOUSNIOTOU
dArcais, 1989). On the other hand, theoretical linguistics has focused mostly on the
semantics of lexical ambiguity. According to theoretical linguistic accounts, lexical
ambiguity is not a uniform phenomenon. Traditionally, two kinds have been distin-
guished. Homonymy, in which a lexical item accidentally carries two (or more) dis-
tinct and unrelated meanings, such as punch 1, which means a blow with a fist,
and punch 2, which means a drink; and polysemy, in which a single lexical
item has several different but related senses, such as mouth meaning both organ
of body and entrance of cave (Cruse, 1986; Lyons 1977).
Two criteria have been proposed for the distinction between homonymy and poly-
semy (Lyons, 1977). The first criterion has to do with the etymological derivation
of words. Words that are historically derived from distinct lexical items are taken to
be homonymous. It is generally taken to be a sufficient condition of homonymy that
the lexical items in question should be known to have developed from what were
formally distinct items in some earlier stage of the language. In practice, however,
the etymological criterion is not always decisive. One reason is that there are many
words whose historical derivation is uncertain. Another reason is that it is not always
very clear how far back we should go in tracing the history of words (Lyons, 1977).
The second criterion for the distinction between homonymy and polysemy has to
do with the relatedness/unrelatedness of meaning. The distinction between hom-
onymy and polysemy seems to correlate with the native speakers feeling that certain
meanings are connected and that others are not. Generally, unrelatedness in meaning
points to homonymy, whereas relatedness in meaning points to polysemy (Lyons,
1977). However, it seems that relatedness of meaning is not an all-or-nothing
relation, but rather a matter of degree. In a large number of cases, there does not
seem to be an agreement among native speakers as to whether the meanings of the
words are related. So, it seems that there is not a clear dichotomy between homonymy
and polysemy, but rather a continuum from pure homonymy to pure polysemy
(Lyons, 1977).
In this article, the term homonymy is consistently used to refer to words that have
the same pronunciation and written forms, but distinct and unrelated meanings, as,
for example, bank 1, which means a financial institution, and bank 2, which
means the side of the river. The term polysemy is consistently used to refer
to a single word that has several related senses, as, for example, chicken, which
means the animal and the meat of that animal.
are literal. For example, the ambiguous word chicken has the literal basic sense
referring to the animal and the literal secondary sense of the meat of that an-
imal.
Thus, consistent with the observation that homonymy and polysemy are relative
concepts, it seems that some types of metaphorically motivated polysemy are closest
to homonymy. On the other hand, metonymically motivated polysemy seems to be
at the other end of the continuum, further away from homonymy (Apresjan, 1974).
Nunberg (1979), continuing the work of Apresjan (1974), further observed that
the changes of meaning in metonymic polysemy are not accidental, as in the case
of homonymy, but systematic, or regular, as Apresjan (1974) called them. He
posited that these changes of meaning can be explained by means of a function,
which he called the referring function (RF), that has the general interpretation x
for y.
The RF is a linguistic process that makes it possible to use the same expression
to refer to distinct categories of things. The idea is that when we cannot point at the
referent itself, we can identify it by pointing at something, called the demonstratum,
which stands in a certain unique relation to the referent (Nunberg, 1979). The hearer
is expected to be able to identify the relation that we have in mind. So, for example,
we can say Washington voted for Bill Clinton and leave it to the hearer to identify
the relation, namely place for people (using the referring function, which has the
general interpretation x for y, so that Washington really refers to the inhabi-
tants of Washington).
Using the notion of the RF, several types of changes of meaning (or shifts of
meaning) have been identified, such as count/mass alternations (e.g., lamb),
container/containee alternations (e.g., bottle), producer/product alternations (e.g.,
Picasso), and place/people alternations (e.g., Washington), to name just a few.
These shifts of meaning, which seem to hold cross-linguistically to a great extent,
are systematic in nature and, thus, in a way, predictable and productive (Copestake &
Briscoe, 1995).
Having identified the different types of lexical ambiguity and the relations that
hold among the multiple meanings of ambiguous words, the question arises as to
how these meanings are stored. In traditional accounts, words are assumed to be
understood by selecting their intended sense from a short exhaustive list of potential
senses which are stored in the mental lexicon (Clark & Gerrig, 1983). These models
of lexical design are often called Sense Enumeration Lexicons (SEL), which make
two assumptions: (a) all possible meanings for each word are listed in the mental
lexicon and (b) the right meaning is selected among these enumerable meanings. So,
under the SEL account, homonymy and polysemy are represented in the same way.
Each meaning of a given ambiguous lexical item (irrespective of whether it is homon-
ymous or polysemous) is stored separately in the mental lexicon (Kempson, 1977;
Weinreich, 1966). Within its specification, there is information about the syntactic
category of the item (i.e., CAT) as well as its broader meaning category (i.e., GENUS)
(Pustejovsky, 1995). For example, one possible partial listing in a SEL of a homony-
mous word (e.g., pen) and a polysemous word (e.g., lamb) appears below:
Homonymy Polysemy
pen1 lamb1
CAT count noun CAT count noun
GENUS writing device GENUS animal
pen2 lamb2
CAT count noun CAT mass noun
GENUS enclosure GENUS meat
208 EKATERINI KLEPOUSNIOTOU
On the other hand, the generative lexicon account, which holds that the senses of
homonymous words are listed separately in the mental lexicon, whereas the senses
of polysemous words are derived from the basic sense, which is stored in the lexicon,
by means of a lexical rule, would predict that there is differential processing de-
pending on the type of ambiguity that the given word exhibits.
On the other hand, studies that support the multiple (or exhaustive) access model
(e.g., Lucas, 1987; Onifer & Swinney, 1981; Seidenberg et al., 1982; Swinney, 1979;
Tanenhaus et al., 1979) are interpreted as supporting the modularity position (Fodor,
1983). Under the strongest form of this model, all the meanings of an ambiguous
word are activated upon its presentation. The activation occurs regardless of context
or meaning frequency. It is after the initial activation of all meanings that the context
plays a role in determining which of the meanings is the appropriate one (Simpson,
1984, 1994).
Looking at the results of all the processing studies collectively, it seems that they
point toward a compromise between pure selective and pure exhaustive access of
meanings of ambiguous words. Simpson (1984, 1994) suggested that the results of
the studies on lexical ambiguity processing could be better explained by having a
hybrid model where all meanings are activated, but the degree of activation would
be sensitive to the relative frequency of the meanings and the context in which the
ambiguous word occurs. This kind of hybrid model, called the reordered access
model, has been proposed by Duffy et al. (1988) and has found empirical support
from studies using eye movement data (e.g., Duffy, Morris, & Rayner, 1988; Rayner,
Binder, & Duffy, 1999; Rayner & Frazier, 1989; Rayner, Pacht, & Duffy, 1994).
According to the reordered access model, the meanings of an ambiguous word are
always exhaustively accessed, but prior context affects the access process by increas-
ing the availability of the contextually appropriate meaning without affecting the
alternative meaning.
The psycholinguistic studies have focused mainly on homonymy (i.e., words that
have the same pronunciation and written form, but distinct and unrelated meanings,
as, for example, bank). Regarding polysemy (i.e., a single word with multiple,
related senses, as, for example, chicken), however, there is only scarce and, at
times, indirect evidence about the representation and processing patterns. In fact,
polysemy raises psycholinguistic issues of a different nature which have received
little attention. It poses questions about the nature of the representations of words
with multiple, related senses in the mental lexicon and investigates whether multiple
senses are represented in a way similar to multiple meanings (i.e., each one sepa-
rately).
There is some evidence that homonymous and polysemous words are processed
differentially. In particular, polysemous words were found to require shorter fixation
times (Frazier & Rayner, 1990) than homonymous words in reading tasks. On the
basis of these findings, Frazier and Rayner (1990) suggested that there are different
implications for the processing of these two types of ambiguous words. In the case of
polysemy, since the multiple senses are not incompatible with one another, immediate
selection of one sense may not be necessary for processing to proceed. In the case
of homonymy, on the other hand, the meanings of the word are mutually exclusive.
Therefore, one meaning must be selected before further processing, and this is time-
consuming. In other studies, it was found that ambiguous words with related mean-
ings are processed faster than words with unrelated meanings (Azuma & Van Orden,
1997). In addition, it was found that the various senses of polysemous words are
interdependent (Williams, 1992) and it may not be possible to suppress them even
in incongruent contexts (Williams, 1992). These findings point toward differential
representations for ambiguous words, depending on whether they are homonymous
or polysemous.
RESEARCH QUESTIONS
of polysemy), are accessed differentially and whether they are stored differently in
the mental lexicon, the present study addresses the following questions, using a cross-
modal sentence-priming lexical decision task: (1) Is lexical ambiguity a homogeneous
or a heterogeneous phenomenon depending on the type of ambiguity? (2) Are homon-
ymous words accessed differently from polysemous words? (3) If polysemy is indeed
the result of a lexical rule, does this affect on-line lexical processing? (4) Are meta-
phorical words accessed the same way as metonymous words?
In accordance with the generative lexicon approach, it is predicted that the partici-
pants reaction times during on-line word recognition in a priming paradigm will be
affected by the type of ambiguity of the target word. In particular, it is expected that
polysemous words will show greater priming effects than homonymous words. The
process of sense creation that takes place in polysemy, and the lexical rule which is
involved in it, is assumed to be less time-consuming than the process of sense selec-
tion that takes place in homonymy. In particular, in the case of polysemy only the
basic semantic value of the word is needed for further processing. On the other hand,
in homonymy one meaning has to be selected before further processing, and this is
time-consuming. In addition, the fact that the meanings of polysemous words are
related in a systematic way should lead to a processing advantage over homonymous
words that have unrelated meanings. Therefore, the processing of polysemy is ex-
pected to be affected differentially, leading to faster reaction times. Finally, it is
expected that within polysemy there will be differential priming between metonymy
and metaphor. Metonymically ambiguous words are expected to show greater prim-
ing effects than metaphorically ambiguous words. Although metaphorical senses are
still quite related in meaning, they tend to be more lexicalized than metonymic senses.
As a result, they are expected to be less sensitive to priming effects.
EXPERIMENT
Method
Participants. Forty-five native speakers of English, with an average of 22 years of age (range: 20
35) and an average of 16.5 years of education (range: 1525), participated in a cross-modal sentence-
priming lexical decision task.
Materials. Three types of target real words were used: (1) the critical (primed) ambiguous words
(W; e.g., panel), (2) control words matched for type of ambiguity (CAW; e.g., mold), and
(3) control words matched for frequency of occurrence (CFW; e.g., trail) [see the Appendix for a
full list of the sentence primes and the target words along with their frequency and familiarity ratings
as well as a list of the filler sentences and nonwords (see below)].
Two types of control words were used because of the peculiarities of ambiguous words. In particular,
the frequency of ambiguous words in standardized lists is collapsed across meanings and does not neces-
sarily reflect the real frequency of each meaning separately. Furthermore, studies have shown that words
with multiple (or many) meanings are accessed differentially from words with one (or few) meaning
(e.g., Forster & Bednall, 1976; Jastrzembski, 1981; Millis & Button, 1989; Rubenstein et al., 1971).
Thus, by using only frequency control words we would control for printed frequency, but not for meaning
frequency and number of meanings. Therefore, it was important to include a second set of control words
that matched the critical ambiguous words for ambiguity type and meaning frequency/familiarity. In
order to assess the familiarity/frequency of each meaning of the ambiguous words, a familiarity/fre-
quency judgment task was designed (see below).
Thus, the target words comprised 40 ambiguous words, 40 frequency of occurrence control words, and
40 type of ambiguity control words. The ambiguous words were divided into four conditions, according to
the type of ambiguity they exhibited: (1) 12 items in the Homonymy condition (i.e., homonymous words,
such as pen a writing device and an enclosure), (2) 8 items in the Metaphor condition (i.e.,
polysemous words with metaphorical extensions, such as eye organ of the body and hole in
a needle), (3) 10 items in the Metonymy condition (i.e., polysemous words with the count/mass met-
onymic extension, such as turkey the animal and the meat of that animal), and (4) 10 items
in the Name condition (i.e., polysemous words with the producer/product metonymic extension, such
212 EKATERINI KLEPOUSNIOTOU
as Dali the person and a painting of that person). All the ambiguous words used had the
same orthographic and phonological forms but different meanings.
Frequency control words (CFW) were chosen from standardized frequency lists (Kucera & Francis,
1967) and were matched for frequency and length with the critical ambiguous words. Although all target
words were of relatively low frequency, targets in the Homonymy condition (e.g., pen) were the most
frequent with a mean frequency of 55; targets in the Metaphor condition (e.g., leg) had a mean
frequency of 46, whereas those in the Metonymy condition (e.g., turkey) had a mean frequency of
11; and targets in the Name condition (e.g., Picasso) were the least frequent with a mean frequency
of 4 (see the Appendix).
Ambiguity control words (CAW) were chosen from a familiarity/frequency judgment task that was
given to a different set of 40 subjects. The familiarity/frequency judgment task was composed of a list
of sentences that included an underlined word that was either a homonymous word (e.g., pen), a
polysemous word with metaphorical extension (e.g., leg), a polysemous word with the mass/count
metonymic extension (e.g., turkey), or a polysemous word with the producer/product metonymic
extension (e.g., Picasso). A total of 40 homonymous words, 23 metaphorical polysemies, 34 met-
onymic polysemies with the mass/count extension, and 28 metonymic polysemies with the producer/
product extension were used in the familiarity/frequency judgment task. Two sets of sentences were
created (i.e., two lists) biased toward the two meanings of the ambiguous words. Each ambiguous word
appeared only once in each list. Subjects rated the two lists with an interval of 3 weeks between them.
Subjects had to decide how often the underlined word is used with the meaning that it conveys in the
given sentence. They were provided with a scale from 0 to 4 and they were given explicit instructions
with examples. If a word is never used with the specific meaning, it should be rated 0. If a word is very
often used with the specific meaning, it should be rated 4. The numbers 1, 2, and 3 should be used to
indicate intermediate stages of usage with the specific meaning. Words within each ambiguity type that
received similar ratings for both their meanings, and also had similar length, were then chosen as ambigu-
ity controls. For example, in the Homonymy condition, race received the ratings 3.54 for the compe-
tition in speed meaning and 2.74 for the human race meaning. Its ambiguity control word was
date, which received similar ratings for its two meanings, namely 3.54 for the time meaning and
2.69 for the fruit meaning (see the Appendix).
Each of the three target word conditions was presented under two meaning bias conditions: (1) the
priming sentence was biased toward the primary meaning of the ambiguous word and (2) the priming
sentence was biased toward the secondary meaning of the ambiguous word. Thus, the same sentence
primes were used for a given ambiguous word (W) and its control words (CFW and CAW). Sentence
primes were constructed trying to exclude word associates of the ambiguous words, so that the whole
context would bias/prime one of the meanings of the ambiguous target words. This was decided by the
author and two native speakers, but there was no other measure of context bias (see the Appendix).
An equal number of filler sentences followed by legal nonwords was also used. Filler sentences were
of equal length and similar syntactic complexity to the sentence primes. Nonwords were constructed by
changing the first or second letter of real words, and they were derived from all the different critical
word categories used in the experiment (except proper names; see the Appendix).
Procedure. Testing was conducted on a Power Macintosh computer using the PsyScope 1.2 program
for stimulus presentation. The participants heard a sentence, read by a female voice, priming one of the
meanings of an ambiguous word. The ambiguous word was not incorporated in the sentence primes. At
the offset of the sentence, and at a 0 interstimulus interval (ISI) [to ensure maximum priming of each
meaning (Simpson, 1984, 1994)], a string of letters (either a real word or a legal nonword) was presented
on the computer screen for lexical decision. The participants were asked to decide whether the target
is a real word of English by pressing the designated yes/no keys on the computer keyboard. Reaction
times in milliseconds were measured from the onset of the target. Priming was measured as the time
to recognize the primed ambiguous word relative to its control words. Items were presented in random
order, in three lists, and counterbalanced across subjects in order to avoid repetition of the sentence
primes and the targets. Each list comprised 240 stimuli, half of which were nonwords.
RESULTS
Only correct responses to word targets were analyzed. Prior to statistical analysis,
errors and outliers (2 standard deviations) were removed. These comprised 12.1%
of the data. It should be noted that 6.36% of the errors and outliers were observed
in a single type of ambiguous words, namely the Name condition (i.e., words with
the producer/product metonymic extension, such as Dali), which was also the
LEXICAL AMBIGUITY PROCESSING 213
TABLE 1
Mean Reaction Times in Milliseconds and Standard Deviations in Parentheses
for All Conditions
Target type
Homonymy 564 (78) 578 (87) 538 (74) 552 (84) 526 (84) 543 (70)
Metaphor 574 (98) 549 (117) 584 (108) 576 (78) 513 (66) 525 (88)
Name 629 (141) 647 (159) 649 (129) 630 (133) 634 (163) 634 (130)
Metonymy 614 (106) 587 (74) 620 (102) 626 (125) 511 (85) 518 (68)
least frequent one. The mean reaction times in milliseconds, as well as the standard
deviations, for each condition can be seen in Table 1.
The data were subjected to a 2 (Meaning bias) 4 (Ambiguity type) 3 (Target)
repeated-measures ANOVA for both subjects (F1) and items (F2). The analysis re-
vealed no main effects of Meaning bias and no interactions with the Meaning bias
variable.
Statistically significant main effects for Ambiguity type [F1(3, 105) 40.45,
p .0001; F2(3, 216) 28.75, p .0001] and Target type [F1(2, 70) 43.85,
p .0001; F2(2, 216) 20.87, p .0001] were revealed (see Fig. 1). Furthermore,
there was a significant Ambiguity type Target type interaction [F1(6, 210)
9.76, p .0001; F2(6, 216) 5.84, p .0001], indicating that the mean differ-
ences among Target types are not constant across the various levels of Ambiguity
type (see Fig. 1).
Post hoc comparisons using the NewmanKeuls test (p .05) revealed that ambig-
uous words (i.e., W) showed significantly faster reaction times (RTs) than the control
frequency words (i.e., CFW) and the control ambiguity words (i.e., CAW) (p .05)
FIG. 1. Mean reaction times (RTs) for control ambiguity word (CAW), control frequency word
(CFW), and ambiguous word (W) target conditions for each type of ambiguity.
214 EKATERINI KLEPOUSNIOTOU
DISCUSSION
The research presented here addressed the issue of whether lexical ambiguity is
a unitary phenomenon or whether it is a heterogeneous phenomenon depending on the
type of ambiguity. Overall, the results supported the hypothesis that type of ambiguity
involves differential processing in the mental lexicon. In particular, faster reaction
times and greater priming effects were observed for metonymous words than for
homonymous words indicating different processes in the recognition of ambiguous
words depending on the type of ambiguity.
LEXICAL AMBIGUITY PROCESSING 215
It is conceivable that differences in context bias could have played a role in the
observed results. However, an attempt was made to avoid lexical associates of the
critical ambiguous words in the sentence primes and construct sentences that biased
the ambiguous words contextually. It could also be claimed that frequency effects
have played a role in the present results since the critical stimuli differed in printed
frequency (i.e., homonymous words had a mean frequency of 55, metaphorical words
had a mean frequency of 46, whereas metonymous words had a mean frequency of
11). However, all critical categories of words, namely homonymy, metaphor, and
metonymy, are considered to be in the low (e.g., Rubenstein, Lewis, & Rubenstein,
1971) or in the medium-low (e.g., Grainger, 1990) range of printed frequency. In
addition, studies with ambiguous words have shown that in lexical decision tasks,
high (frequency above 150) and low frequency (frequency below 30) words produce
identical polysemy effects, where polysemy refers to number of meanings
(Hino & Lupker, 1996). In particular, it was found that similar polysemy effects
were observed for both high and low frequency words, and the interaction between
frequency and polysemy was not significant. Thus, it seems unlikely that the pres-
ent findings are due to frequency effects.
The results of the present study cannot be accounted for by the SEL approach.
According to this theoretical approach, each meaning of an ambiguous word, irre-
spective of whether it is homonymous or polysemous, is stored separately in the
mental lexicon. As a result, both homonymous and polysemous meanings are repre-
sented in the same way. The SEL approach, therefore, predicts similar processing
patterns in order to access meaning for both types of ambiguity. Furthermore, similar
processing patterns would be expected within polysemy as well, that is, between
metonymy and metaphor, since all meanings are assumed to be separately stored.
The results of the present study, however, indicate differential processing patterns
between homonymy and polysemy. Thus, the prediction of the SEL approach, namely
that all ambiguous words are processed similarly, is in direct contrast to the present
results that show differential processing depending on the type of ambiguity.
On the other hand, according to the generative lexicon approach, homonymy and
polysemy are represented differently in the mental lexicon. Homonymous words have
their different senses represented separately in the lexicon and are, therefore, under-
stood by selecting their intended sense from a (presumably exhaustive) list of poten-
tial senses (i.e., the process involved is sense selection). In the case of polysemy,
however, only the basic sense with a general semantic value is stored in the lexicon.
The extended senses are generated (presumably on-line) from the basic sense by
means of a lexical rule (i.e., the process involved is sense creation).
The results of the present study may be interpreted as consistent with the generative
lexicon approach. In particular, it was found that metonymous words showed signifi-
cantly greater priming effects and were processed significantly faster than homony-
mous words. Under the generative lexicon account, metonymous words have a single
representation in the mental lexicon that has a basic semantic value. Thus, having a
single representation, metonymous words avoid any issues of ambiguity resolution
that might compromise the activation process. Since only one meaning is stored,
there is no competition among meanings for activation. As a result, the activation
process is not compromised by extra processing that is caused by the necessity of
selecting a single meaning of the ambiguous words when more than one has been
activated (Simpson, 1994). Thus, metonymous words are expected to show the sig-
nificantly faster reaction times that were found in the present study.
Homonymous words, on the other hand, have all their senses stored separately in
the mental lexicon. Context is useful in deciding which of the meanings will be
selected, but it is not actively involved in the on-line creation of senses. In addition,
216 EKATERINI KLEPOUSNIOTOU
since two or more meanings of a homonymous word are competing for activation,
the process of activation will possibly be compromised by the process of ambiguity
resolution. It has been shown that even when the context is biased toward one mean-
ing, all the meanings of a homonymous word are initially activated (e.g., Onifer &
Swinney, 1981; Seidenberg et al., 1982; Swinney, 1979). The initial activation of all
meanings probably reflects comprehension processes that are involved in the retrieval
and consideration of the several meanings of an ambiguous word (Onifer & Swinney,
1981; Swinney, 1979). The slower reaction times that were observed in the processing
patterns of homonymous words could thus be explained.
Turning to polysemy, there seem to be differences in the processing of metonymy
(e.g., turkey) and metaphor (e.g., eye), even though the results did not always
reach significance. As can be seen in Fig. 1, there was a tendency for metonymous
words to be processed faster and receive more priming than metaphors. Furthermore,
there was a tendency for metaphors to be processed faster and receive more priming
than homonyms. If homonymy and metonymy are assumed to be at the two ends of
lexical ambiguity (i.e., from separately stored to rule generated), then metaphor seems
to lie somewhere in the middle. In some cases, the multiple meanings of metaphorical
words are felt to be quite related, whereas in other cases, they are further apart and
the relatedness in meaning is not so obvious. These results are consistent with the
observations made by Apresjan (1974) that metaphorically motivated polysemy, un-
like metonymically motivated polysemy, seems to be quite unconstrained. Metaphori-
cal words seem to be in a transition phase from generated senses to separately stored
senses. In the former case, the representation of metaphorical words would resemble
the representation of metonymous words. In the latter case, their representation would
resemble the representation of homonymous words.
The present findings that metonymous words are accessed faster than homonymous
words give further support to the findings of Frazier and Rayner (1990) that words
with multiple senses showed shorter fixation times than words with multiple mean-
ings. In addition, they also provide support to studies that have found that ambiguous
words with related meanings are accessed faster than ambiguous words with unrelated
meanings (Azuma & Van Orden, 1997).
The present findings also have important implications for the nature of the mental
representations of ambiguous words. They point toward a dichotomy in the nature
of the mental representations, depending on the type of ambiguity that the words
exhibit. Homonymous words appear to have several, distinct mental representations,
one for each meaning. Polysemous words, on the other hand, and in particular met-
onyms, appear to have a single mental representation. This representation would be
specified for the basic sense of the polysemous word. Furthermore, it should specify
the lexical rule, or lexical rules, that would operate on the basic sense in order to
derive the extended meanings.
The obtained results are consistent with psycholinguistic models of representation
and processing that allow for differentiation between homonymy and metonymy. One
such model is that proposed by Bierwisch and Schreuder (1992). In their model, the
representational system of lexical meanings is called the semantic form. Homony-
mous words are associated with more than one semantic form, thus having multiple
representations. Polysemous words, on the other hand, are associated with just one
semantic form, thus having a single representation. The different conceptual interpre-
tations of polysemous words are the result of lexical rules that bring in different
background knowledge that determines the specific interpretation.
Overall, the present data provide experimental support for the theoretically moti-
vated differentiation of lexical ambiguity into homonymy and polysemy. They pro-
vide evidence that homonymy and polysemy, especially metonymy, rely on distinct
LEXICAL AMBIGUITY PROCESSING 217
APPENDIX
Sentence primes are presented followed by the targets that appeared after them.
First the sentence biasing the primary meaning of the ambiguous target word (W) is
given followed by the sentence biasing the secondary meaning. The primed ambigu-
ous word (W) appears with its printed frequency in parentheses followed by the
familiarity ratings for each meaning in another set of parentheses. The ambiguous
control word matched for familiarity (CAW) is given with the familiarity ratings for
each meaning. The unambiguous control word matched for printed frequency (CFW)
is given with its printed frequency. Finally, the filler sentences followed by the non-
words are given.
Homonymy
He consulted a group panel (31) (2.67/2.36) mold (2.38/2.36) trail (31)
of experts.
The office is divided
in two.
All the snow has spring (127) (3.59/2.51) bridge (3.00/2.54) hotel (127)
melted now.
In the mountains, we
refilled our can-
teens.
The wrestler knocked punch (5) (2.81/2.23) cricket (2.87/2.66) altar (5)
him down.
For the party, I pre-
pared a drink.
I have the papers in file (81) (3.05/2.41) seal (3.05/2.46) watch (81)
my office.
The carpenter
smoothed the wood.
John likes sports. racket (5) (3.44/2.23) mole (1.92/1.36) photo (5)
I heard a sudden
noise.
Harry is training every coach (24) (3.49/1.89) lock (1.15/1.05) leather (24)
day.
The passenger left
from the station.
He goes to every foot- fan (18) (3.64/3.56) lap (3.67/3.13) lamp (18)
ball game.
This room is very hot.
Sally brought me a plant (125) (2.85/1.11) ball (3.59/2.54) farm (125)
nice gift.
Sue works in a fac-
tory.
Larry went to the suit (48) (3.72/2.31) pen (2.82/1.39) poem (48)
mens store.
Jenny went to her law-
yer.
218 EKATERINI KLEPOUSNIOTOU
The athletes were fast. race (103)(3.54/2.74) date (3.54/2.69) unit (103)
Europeans belong to
the same family.
Cathy checked her bank (83) (3.62/1.18) page (3.69/1.03) ship (83)
account.
Tom was fishing from
a rock.
The cottage has run- tap (18) (2.79/1.13) deck (3.31/0.95) brick (18)
ning water.
I heard a man knock-
ing at the window.
Mean: (55) (3.27/2.13) Mean: (2.92/2.02) Mean: (55)
Metaphorical polysemy
Laura has a slim fig- waist (11) (2.00/1.69) sole (2.38/1.00) shaft (11)
ure.
The skirt is too tight
to button.
My dog is happy. tail (24) (2.72/1.18) mouth (2.51/1.15) motel (24)
I went to the back of
the airplane.
He cannot read small eye (122) (3.28/2.00) skin (3.26/1.79) blood (122)
print.
He threaded the nee-
dle.
I do a lot of cycling. leg (58) (3.92/2.67) arm (3.74/2.31) stone (58)
The table is broken.
I put on a scarf. neck (81) (3.74/1.85) tooth (3.59/1.95) train (81)
The cork got stuck.
I bit myself by acci- lip (18) (3.77/1.03) nose (3.74/0.97) rod (18)
dent.
This cup is chipped.
I burnt myself with tongue (35) (3.74/1.41) shoulder (3.82/1.63) bench (35)
the food.
The dog ate my shoe.
This bird is beautiful. wing (18) (3.31/3.21) heel (2.95/2.77) tray (18)
Airplanes travel fast.
Mean: (46) (3.31/1.88) Mean: (3.25/1.70) Mean: (46)
Mass/count metonymic
polysemy
I put one in my mar- olive (5) (2.36/1.79) willow (1.31/1.28) canon (5)
tini.
I like it on my pizza.
I baked one for lunch. potato (15) (3.54/2.36) pear (3.36/2.31) anchor (15)
I ate some mashed
with gravy.
It is sour but I like it. lemon (18) (3.54/0.88) fig (2.79/1.79) angel (18)
I put some on my fish.
The hunter killed one. rabbit (11) (3.10/2.28) plum (2.33/1.95) violin (11)
The chef made a stew.
I bought one to make cabbage (4) (3.10/1.07) apricot (2.85/2.28) gadget (4)
salad.
I had some in my
salad.
I eat a green one apple (9) (3.51/2.62) elm (2.15/1.29) rack (9)
every day.
I made a delicious pie.
LEXICAL AMBIGUITY PROCESSING 219
Producer/product met-
onymic polysemy
I read about a famous Webster (5) (2.46/1.33) Elgar (2.51/1.34) Jason (5)
lexicographer.
I borrowed a good dic-
tionary.
He is a well-known Orwell (2) (2.92/1.11) Raphael (2.77/2.23) Antony (2)
novelist.
I bought a novel.
The newspaper writes Chomsky (1) (3.03/1.33) Boccelli (3.08/1.56) Benoit (1)
about a linguist.
I gave him a linguis-
tics book.
He is a great tenor. Pavarrotti (1) (2.97/1.23) Hemingway (3.08/1.21) Frederic (1)
I bought a classical
music CD.
I met him in an art gal- Dali (1) (2.79/2.23) Wagner (2.97/1.33) Emil (1)
lery.
I bought a painting.
He is a sculptor. Rodin (1) (2.72/2.15) Wordsworth (3.28/1.05) Jacob (1)
I own a sculpture.
I met an artist. Degas (1) (2.74/2.08) Strauss (2.95/1.21) Franc (1)
I acquired a famous
piece of art.
He is a famous com- Vivaldi (1) (3.23/1.33) Matisse (2.85/2.44) Leopold (1)
poser.
I listened to an opera.
He is a classical Shakespeare (30) (3.36/1.33) Mozart (3.31/1.15) Franklin (30)
writer.
I read a classical play.
He is a poet. Coleridge (1) (2.90/1.26) Rembrandt (2.87/2.54) Emmanuel (1)
I bought a poetry col-
lection.
Mean: (4) (2.91/1.53) Mean: (2.97/1.60) Mean: (4)
REFERENCES
Bierwisch, M., & Schreuder, R. (1992). From concepts to lexical items. Cognition, 42, 2360.
Bredart, S., Brennen, T., & Valentine, T. (1997). Dissociations between the processing of proper and
common names. Cognitive Neuropsychology, 14, 209217.
Clark, H. H., & Gerrig, R. J. (1983). Understanding old words with new meanings. Journal of Verbal
Learning and Verbal Behavior, 22, 591608.
Copestake, A., & Briscoe, T. (1995). Semi-productive polysemy and sense extension. Journal of Seman-
tics, 12, 1567.
Cruse, D. A. (1986). Lexical semantics. Cambridge, UK: Cambridge Univ. Press.
Duffy, S. A., Morris, R. K., & Rayner, K. (1988). Lexical ambiguity and fixation times in reading.
Journal of Memory and Language, 27, 429446.
Fodor, J. A. (1983). The modularity of mind. Cambridge, MA: MIT Press.
Forster, K. I. (1976). Accessing the mental lexicon. In R. J. Wales & E. C. T. Walker (Eds.), New
approaches to language mechanisms. Amsterdam: North-Holland.
Forster, K. I., & Bednall, E. S. (1976). Terminating and exhaustive search in lexical access. Memory &
Cognition, 4, 5361.
Frazier, L., & Rayner, K. (1990). Taking on semantic commitments: Processing multiple meanings vs.
multiple senses. Journal of Memory and Language, 29, 181200.
Grainger, J. (1990). Word frequency and neighborhood frequency effects in lexical decision and naming.
Journal of Memory and Language, 29, 228244.
Hino, Y., & Lupker, S. J. (1996). Effects of polysemy in lexical decision and naming: An alternative
to lexical access accounts. Journal of Experimental Psychology: Human Perception and Perfor-
mance, 22, 13311356.
Hogaboam, T. W., & Perfetti, C. A. (1975). Lexical ambiguity and sentence comprehension. Journal
of Verbal Learning and Verbal Behavior, 14, 265274.
Jastrzembski, J. E. (1981). Multiple meanings, number of related meanings, frequency of occurrence,
and the lexicon. Cognitive Psychology, 13, 278305.
Kempson, R. M. (1977). Semantic theory. Cambridge, UK: Cambridge Univ. Press.
Kucera, H., & Francis, W. N. (1967). Computational analysis of present-day American English. Provi-
dence, RI: Brown Univ. Press.
Lucas, M. M. (1987). Frequency effects on the processing of ambiguous words in sentence contexts.
Language and Speech, 30, 2546.
Lyons, J. (1977). Semantics (Vol. 2). Cambridge, UK: Cambridge Univ. Press.
Marslen-Wilson, W. W. (1987). Functional parallelism in spoken word-recognition. Cognition, 25, 71
102.
Martin, C., Vu, H., Kellas, G., & Metcalf, K. (1999). Strength of discourse context as a determinant of
the subordinate bias effect. The Quarterly Journal of Experimental Psychology, 52, 813839.
McClelland, J. L. (1987). The case for interactionism in language processing. In M. Coltheart (Ed.),
Attention and performance XII: The psychology of reading. Mahwah, NJ: Erlbaum.
Millis, M. L., & Button, S. B. (1989). The effect of polysemy on lexical decision time: Now you see
it, now you dont. Memory & Cognition, 17, 141147.
Nunberg, G. (1979). The non-uniqueness of semantic solutions: Polysemy. Linguistics and Philosophy,
3, 143184.
Onifer, W., & Swinney, D. A. (1981). Accessing lexical ambiguities during sentence comprehension:
Effects of frequency of meaning and contextual bias. Memory & Cognition, 9, 225236.
Paul, S., Kellas, G., Martin, M., & Clark, M. B. (1992). Influence of contextual features on the activation
of ambiguous word meanings. Journal of Experimental Psychology: Learning, Memory and Cogni-
tion, 18, 703717.
Pustejovsky, J. (1995). The generative lexicon. Cambridge, MA: MIT Press.
Rayner, K., Binder, K. S., & Duffy, S. A. (1999). Contextual strength and the subordinate bias effect:
Comment on Martin, Vu, Kellas, and Metcalf. The Quarterly Journal of Experimental Psychology,
52, 841852.
Rayner, K., & Frazier, L. (1989). Selection mechanisms in reading lexically ambiguous words. Journal
of Experimental Psychology: Learning, Memory and Cognition, 15, 779790.
Rayner, K., Pacht, J. M., & Duffy, S. A. (1994). Effects of prior encounter and global discourse bias
on the processing of lexically ambiguous words: Evidence from eye fixations. Journal of Memory
and Language, 33, 527544.
LEXICAL AMBIGUITY PROCESSING 223
Rubenstein, H., Lewis, S. S., & Rubenstein, M. A. (1971). Homographic entries in the internal lexicon:
Effects of systematicity and relative frequency of meanings. Journal of Verbal Learning and Verbal
Behavior, 10, 5762.
Schreuder, R., & Flores dArcais, G. B. (1989). Psycholinguistic issues in the lexical representation of
meaning. In W. Marslen-Wilson (Ed.), Lexical representation and process. Cambridge, MA: MIT
Press.
Seidenberg, M. S., Tanenhaus, M. K., Leiman, J. M., & Bienkowski, M. (1982). Automatic access of
the meanings of ambiguous words in context: Some limitations of knowledge-based processing.
Cognitive Psychology, 14, 489537.
Semenza, C., & Zettin, M. (1989). Evidence from aphasia for the role of proper names as pure referring
expressions. Nature, 342, 678679.
Simpson, G. B. (1981). Meaning dominance and semantic context in the processing of lexical ambiguity.
Journal of Verbal Learning and Verbal Behavior, 20, 120136.
Simpson, G. B. (1984). Lexical ambiguity and its role in models of word recognition. Psychological
Bulletin, 96(2), 316340.
Simpson, G. B. (1994). Context and the processing of ambiguous words. In M. A. Gernsbacher (Ed.),
Handbook of psycholinguistics. San Diego: Academic Press.
Simpson, G. B., & Kang, H. (1994). Inhibitory processes in the recognition of homograph meanings.
In D. Dagenbach & T. H. Carr (Eds.), Inhibitory processes in attention, memory and language.
San Diego, CA: Academic Press.
Simpson, G. B., & Krueger, M. A. (1991). Selective access of homograph meanings in sentence context.
Journal of Memory and Language, 30, 627643.
Swinney, D. A. (1979). Lexical access during sentence comprehension: (Re)consideration of context
effects. Journal of Verbal Learning and Verbal Behavior, 18, 645659.
Swinney, D. A. (1991). The resolution of indeterminacy during language comprehension: Perspectives
on modularity in lexical, structural and pragmatic process. In G. B. Simpson (Ed.), Understanding
word and sentence. Amsterdam: Elsevier.
Tabossi, P. (1988). Accessing lexical ambiguity in different types of sentential contexts. Journal of
Memory and Language, 27, 324340.
Tanenhaus, M. K., Leiman, J. M., & Seidenberg, M. S. (1979). Evidence for multiple stages in the
processing of ambiguous words in syntactic contexts. Journal of Verbal Learning and Verbal Behav-
ior, 18, 427440.
Weinreich, U. (1966). Explorations in semantic theory. In T. A. Sebeok (Ed.), Current trends in linguis-
tics (Vol. 3). The Hague: Mouton.
Williams, J. N. (1992). Processing polysemous words in context: Evidence for interrelated meanings.
Journal of Psycholinguistic Research, 21, 193218.