Inhibitory priming effects in auditory word recognition: When the target’s
competitors conflict with the prime word.
Sophie Dufour and Ronald Peereman
University of Bourgogne & C.N.R.S, Dijon, France
L. E. A. D. - C. N. R. S.
Université de Bourgogne,
21065 Dijon Cedex,
Several studies indicate that the number of similar sounding words that are activated
during recognition is a powerful predictor of performance on auditory targets. Words with
few competitors are processed more quickly and accurately than words with many
competitors. In the present study, we examined the contribution of the competitor set size in
determining the magnitude of the inhibitory priming effect. The data show that the priming
effect is stronger when word targets have few competitors. This result supports the view of
direct competition between lexical candidates.
Keywords: Phonological priming; Lexical competition; Competitor set size
It is now widely accepted that as listeners attend to a spoken word, phonologically
similar words are activated and affect the speed of lexical access (see Frauenfelder & Peters,
1998, for a review). The influence of lexical competitors is assumed in most contemporary
models of auditory word recognition (Luce, Pisoni & Goldinger, 1990; Marslen-Wilson,
1987; McClelland & Elman, 1986; Norris, 1994) although the precise mechanism by which
competition is supposed to arise still remains controversial. In some models, such as TRACE
(McClelland & Elman, 1986) and Shortlist (Norris, 1994), lexical competition is due to intra-
level inhibition that operates between activated lexical entries. In contrast, in the Cohort
(Marslen-Wilson, 1987; Marslen-Wilson, Moss & Van Halen, 1996) or the NAM models
(Luce, Pisoni & Goldinger, 1990), competitors have no direct influence on the activation level
of word candidates. Lexical competition takes place only at the decision stage of recognition
where the existence of close competitors slows down the process of discrimination among
Up to now, most studies have examined the magnitude of the inhibition effects as a
function of the size of the competitor sets. Several studies indicate that words with few
competitors are processed more quickly and accurately than words with many competitors
(e.g. Luce & Pisoni, 1998; Vitevitch, 2002). Although simulation work (Frauenfelder &
Peters, 1998; Luce & Pisoni, 1998) indicate that models both with and without inhibitory
links can capture the effect of the competitor set size, a few studies suggest that an intra-level
inhibition account could constitute a more appropriate theoretical approach (Norris, McQueen
& Cutler, 1995; Vroomen & de Gelder, 1995). Using a word spotting experiment, Norris et al.
(1995) showed that it was harder to detect a target word (e.g. STAMP) embedded in a
bisyllabic nonsense sequence (e.g. STAMPIDGE) when the second syllable of the nonsense
sequence activates a large number of competitors. Using a cross-modal repetition paradigm,
Vroomen and de Gelder (1995) found smaller priming effects in lexical decision for visually
presented targets (e.g. MELK) when the second syllable of the auditory non word prime (e.g.
MELKAAM) activated many competitors. Both observations suggest that competition occurs
between lexical candidates during spoken word processing. The present study was undertaken
to provide more direct evidence that competition occurs between word candidates by
examining how the influence of a particular lexical competitor of the target word varies as a
function of the size of the candidate set.
An interesting way to study competition processes consists in measuring target
performance after the presentation of one of its lexical competitors. This would appear to
make the priming paradigm an interesting methodological tool for the study of lexical
competition during target recognition, since a competitor is explicitly presented and its effect
on performance can be measured. Although previous reports of phonological priming have
provided conflicting results (Radeau, Morais & Dewier, 1989; Slowiaczek, Nusbaum &
Pisoni, 1987; Slowiaczek & Pisoni, 1986 ), more recent studies controlling both the amount of
overlap between primes and targets and the contribution of strategic factors have shown that
the time taken to identify a target word is delayed when it is preceded by a phonologically
related prime, namely by one of its competitors (Dufour & Peereman, submitted; Hamburger
& Slowiaczek, 1996, Monsell & Hirsh, 1998; Slowiaczek & Hamburger, 1992). Importantly,
this result seems to be restricted to the case where primes and targets share onsets1 (Monsell
& Hirsh, 1998; Radeau, Morais, Segui, 1995). When the same kind of overlap occurs at offset
facilitation tends to be found (Dumay, Benraïss, Barriol, Colin, Radeau & Besson, 2001;
Monsell & Hirsh, 1998; Radeau, Morais & Segui, 1995; Slowiaczek, McQueen, Soltano &
Lynch, 2000), though the effect in the lexical decision task appears to be modulated by
strategic bias (Norris, McQueen & Cutler, 2002).
As discussed by Slowiaczek and Hamburger (1992), inhibitory priming effects can be
readily explained within the framework of models that postulate intra-level inhibition. For
example, although the TRACE model (McClelland & Elman, 1986) has not yet been
developed to simulate priming effects, all activated word units inhibit each other as a function
of their activation levels. The more a word is activated, the more it sends out inhibition to
other candidates. According to Slowiaczeck and Hamburger, inhibitory priming effects are
occasioned during target presentation. Since the prime word is re-activated by the phonemes
that it shares with the target, its level of activation increases sufficiently to strongly compete
with the target word. Although such an account of the inhibitory priming effect is generally
adopted (Monsell & Hirsh, 1998; Slowiaczek & Hamburger, 1992), an alternative view would
be to consider that conscious identification of the prime word causes inhibition of the target
word, therefore slowing down its subsequent identification. This possibility has been
envisaged by Segui and Grainger (1990) to explain inhibitory priming effects in the visual
modality with unmasked primes. Such an alternative account seems, however, difficult to
reconcile with models such as Cohort (Marslen-Wilson, 1987; Marslen-Wilson et al., 1996)
and NAM (Luce et al., 1990) that do not incorporate mechanisms allowing the inhibition of
the target word during the processing of the prime word. We will return to this explanation
when discussing the results of the present study within the framework of models that assume
The assumption that the competition effect results from intra-level inhibition naturally
leads one to predict that during the processing of the target, the prime itself will also receive
inhibition from the pool of candidates from which it is re-activated. As a consequence, inter-
word competition models predict that the amount of inhibition exerted by the prime on the
activation level of the target word should vary as a function of the number of other
competitors activated by the target. In particular, one might expect that a prime should be
more competitive when the target has few other competitors since, in this case, it will receive
inhibition from a smaller pool of candidates, and thus should act as a stronger competitor of
the target. Conversely, models such as Cohort (Marslen-Wilson, 1987; Marslen-Wilson et al.,
1996) that assume that competitors have no direct influence on one another predict that the
inhibitory power of a prime should not vary as a function of the competitor set size. These
predictions will be examined in a naming task using targets having either a large or a small
competitor set size.
Competitors were defined sequentially in terms of candidates that are aligned with the
onset of the target word (Marslen-Wilson, 1990). Because some recent data suggest that
words are quickly deactivated once they no longer match the incoming signal (Frauenfelder,
Scholten & Content, 2001), only the words that diverged the latest from the target word were
counted. As an example, consider the French target word CURE /ky?/ primed by the word
CUBE /kyb/. The target CURE /ky?/ has only two competitors (CUBE /kyb/ and CUVE
/kyv/) one of them serving as prime. The hypothesis of intra-level inhibition predicts that
during the processing of the last phoneme of the target CURE /ky?/, the prime CUBE /kyb/
should be strongly re-activated since it will receive inhibition from only one other competitor
(CUVE /kyv/). In contrast, if we consider the target word CAVE /kav/ primed by CASE
/kaz/, the prime CASE /kaz/ will be re-activated from a larger set of competitors including the
target CAVE /kav/, but also numerous other lexical candidates such as CAGE /caj/, CAPE
/kap/, CANNE /kan/, CAR /ka?/, CALE /kal/, CACHE /ka?/ etc. As a consequence, the
prime CUBE /kyb/ should receive less inhibition from the competitors of the target word and
should inhibit target word recognition more strongly than the prime CASE /kaz/. Inhibitory
effects should thus be of greater magnitude when the target word has few competitors2.
Forty–two students at the University of Bourgogne participated in the experiment for
course credits. All were native speakers of French and reported no hearing or speech
Fifty-six monosyllabic target words, three phonemes in length, were selected from
VOCOLEX, a lexical database for the French language (Dufour, Peereman, Pallier & Radeau,
in press). Half of them had a large number of competitors (mean: 12.57; range: 6-17). The
other twenty-eight words had a small number of competitors (mean: 3.18; range: 1-5)3. Both
sets of targets were matched for word frequency (mean frequency of 76 per million for the
words with many competitors, and 125 for the word with few competitors (F(1,54)=1.03,
p>.20). As the targets differed on their initial phonemes, a delayed naming task was
performed by 18 additional participants to assess latency differences resulting from the ease
of articulation and the triggering of the voice key. Participants were presented with the
auditory target words and waited until the appearance of a visual response cue (occurring
either 1800, 2100, or 2400 ms after target offset) before repeating the targets. Mean latencies
were 392 ms and 388 ms for targets with few and many competitors, respectively. The
difference was not significant in both the by-subject and the by-item analyses (both p>.20).
For each target, two monosyllabic primes of three phonemes in length were selected.
Because it has been observed that primes sharing initial phonemes with the targets produce
more inhibition in the naming task when they are of lower frequency than the targets (Radeau,
Morais & Segui, 1995), the primes were on average less frequent than the targets (see also
Eberhard, 1994). The one shared the first two phonemes with the targets (e.g. RIDE /?id/-
RICHE /?i?/), the other, used as a control, had no phoneme in common with the targets (e.g.
DALLE /dal/- RICHE /?i?/). For targets having many competitors, the mean frequencies of
related and control primes were 12 and 14 respectively (F<1). For targets with few
competitors, the corresponding values were 22 and 15 respectively (F(1,54) = 1.34, p>.20).
The prime-target pairs are provided in the Appendix.
Because each target was paired with two different primes (related and control), and no
subject was presented with the same target twice, two experimental lists were created. Each
list included the 56 target words. Half of them were preceded by a related prime, the other
half by a control prime. The lists were counterbalanced in such a way that each target was
preceded by the two types of prime. In order to achieve a low proportion of related prime-
target pairs (25%), 56 filler trials without any relation between the primes and the targets were
included in each list.
The stimuli were recorded by a female native speaker of French on a digital audio tape
recorder. The items were digitized at a sampling rate of 44 kHz with 16-bit analog to digital
recording. The participants were tested individually in a quiet room. The presentation of the
items was controlled by a personal computer. RTs were collected via a voice key connected to
the computer. The primes and the targets were presented over headphones at a comfortable
sound level. An interval of 50 ms (ISI) separated the offset of the prime and the onset of the
target. The participants were asked to repeat the target as quickly and accurately as possible.
The subject's response and the onset of the prime of the following trial were separated by 2
seconds of silence. The naming latencies were measured from the onset of the target to the
subject's response. The participants were tested on only one experimental list and began the
experiment with a block of 16 practice trials.
Because of technical problems, one participant was excluded from the analyses. For
each subject, both reaction times (RTs) longer than 1200 ms and those greater than 2.5
standard deviations above and below the participant’s overall response time were removed
from the latency analyses (2.35%). Incorrect responses were also removed from the analyses.
Because few errors occurred (0.22%), analyses were performed on RTs only. Analyses of
Variance (ANOVA) by participants (F1) and by items (F2) were conducted with prime type
(related, control) and competitor set size (large, small) as variables. Because of the intrinsic
variability in word duration, RTs were corrected by subtracting the duration of each word
from the RT for the word. Mean RTs in each condition are presented in Table 1.
<Insert Table 1 about here>
The analysis showed a main effect of prime type that was significant both by subjects
(F1(1,40)=10.98, p<.01) and by items (F2(1,54)= 17.18, p<.001). Targets preceded by related
primes (273 ms) were responded to more slowly than targets preceded by control primes (262
ms). The main effect of competitor set size failed to reach significance by subjects
(F1(1,40)=2.71, p=.11) and was not significant by items (F2(1,54)= 0.16, p>.20). The
interaction between prime type and competitor set size was significant both by subjects
(F1(1,40)=6.19, p<.05) and by items (F2(1,54)= 4.74, p<.05)4.
Planned comparisons were conducted to assess the effect of priming within each
competitor set size. A priming effect was observed only for the targets occurring in a small set
of competitors. Responses to targets were 21ms slower when they were preceded by the
related primes in comparison with the control primes. This effect was highly significant both
by subjects (F1(1,40)=16.18, p<.001) and by items (F2(1,54)= 19.99, p<.001). No priming
effect was observed when the targets occurred in a large set of competitors (F1(1,40)=0.05,
p>.20); (F2(1,54)= 1.93, p=.17).
The present study was designed to examine whether variations in the competitor set
size of target words lead to variations regarding the influence of one of their lexical
competitors (the related prime). Given the assumption that inhibitory priming effects can be
explained in terms of intra-level inhibition, we predicted that a word prime would have a
greater inhibitory influence on target word recognition when the targets themselves have few
competitors. In accordance with this prediction, RTs on word targets were slower when
preceded by a phonologically related prime only for targets having a small set of competitors.
No priming effect was observed when the targets had a large set of competitors, indicating
that the inhibitory influence of a prime word was considerably reduced when it was itself re-
activated among a large pool of candidates. As we have mentioned above, inhibitory priming
effects are generally accounted for by reference to lexical activation produced by the target.
However, in the framework of the TRACE model, inhibitory effects could also be
conceptualised as resulting from inhibition of the target word during conscious identification
of the prime. In this perspective, the larger priming effect observed for targets with few
competitors could be interpreted as follows. By construction, related prime and target words
have a similar number of competitors. Prime words should suffer more from lexical
competition when similar to many other words, and they should therefore cause less inhibition
of the lexical competitors including the target word during prime processing. Consequently
target recognition would be less delayed than when the prime word has few competitors. To
summarize, the results strongly suggest that the lexical competitor set size plays an important
role in determining the effectiveness of a prime word in inhibiting target word recognition.
Similar conclusions have been drawn from recent studies on visual word recognition
(Perea & Rosa, in press; Van Heuven, Dijkstra, Grainger and Schriefers, 2001). Although
these studies used nonword primes rather than word primes and reported facilitatory priming
effects rather than inhibitory effects, they also indicate that the competitor set size is an
important factor in determining the magnitude of the priming effects. Facilitatory priming
effects were found to be greater for words with few competitors than for words with many
competitors. Interestingly, this finding was captured by the IA model (McClelland &
Rumelhart, 1981) which, like TRACE, includes an inter-word competition mechanism.
Through this mechanism, the more competitors a target word has, the more inhibition it
receives, thus cancelling out the facilitation resulting from the pre-activation of the target
word during prime presentation.
Interestingly, the observation of an inhibitory effect only for targets having a small set
of competitors seems to challenge the recent view that the inhibition observed with an initial
overlap reflects a surprise effect (Pitt & Shoaf, 2002). Examining the emergence of biases by
comparing the magnitude of the priming effects at various points during the experimental
session, Pitt & Shoaf reported an inhibitory effect only for targets occurring at the beginning
of the experiment. No inhibition was observed when the targets occurred at the end of the
testing session. It was therefore suggested that priming effects are contaminated by response
biases developed by participants when they become aware of the presence of related pairs (see
also Goldinger, 1999). Moreover, it was argued that inhibitory priming effects are due to
participants’ surprise that arises when they encountered the first related trials. We believe
however, that the present inhibitory effect cannot be attributed to participants’ surprise. First,
it was observed only for targets having a small set of competitors and it seems unlikely that
the surprise effect is modulated as a function of the competitor set size. Second, our
experimental setting also included related trials in the training session (4 related trials out of
16) so that any surprise effect (if it occurs) should be manifested during the training session
and not during the experimental session. Finally, for each competitor set size, analyses as a
function of presentation order were performed in order to examine how the inhibition
developed over the course of the experiment. RTs to control and related trials were analysed
for each participant according to their position in the experimental list such that the first three
trials were labelled "time window 1", the second three "time window 2", the third three "time
window 3" and the last three "time window 4"5 ( see Goldinger, 1999; Slowiaczek, McQueen,
Soltano & Lynch, 2000, for similar analyses). As can be seen in Table 2, no significant
inhibition was observed at the beginning of the experiment in the large set of competitors
(F(1,40)=0.13) as would have been expected if a surprise on the first related trials was
primarily responsible for the inhibitory priming effects. Moreover, analyses performed for the
small set of competitors showed that the observed inhibitory priming effect was still present at
the end of the experiment and was of similar magnitude across the testing session
(F(3,120)=0.24). It should also be noted that RTs on related trials increased but did not
decrease between the beginning and the end of the experiment, an observation that again
suggests that the present inhibition was neither caused by a surprise thought to occur on the
first related trials nor blocked by response biases that participants might develop in order to
maximize response speed on related trials (Pitt & Shoaf, 2002).
<Insert Table 2 about here>
A potentially important factor related to the competitor set size is phonotactic
probability. Words phonologically similar to many other words generally include frequent
components (e. g., phonemes, biphones). As suggested by Vitevitch and colleagues (Vitevitch
& Luce, 1998, 1999; Vitevitch, Luce, Pisoni & Auer, 1999), pre-lexical processing could be
facilitated by high probabilistic phonotactics, an observation compatible with various models
such as TRACE(McClelland & Elman, 1986), Shortlist (Norris, 1994), and Parsyn (Luce,
Goldinger, Auer & Vitevitch, 2000). Because related primes and targets were similar in terms
of competitor set sizes, it could therefore be the case that the differential priming effects were
a function of probabilistic phonotactics of the prime. Such a hypothesis seems, however,
difficult to reconcile with the data. Indeed, although processing is facilitated for pseudowords
with high phonotactic probability, performance is actually impaired for words with high
phonotactic probability and large numbers of competitors. Thus in the case of words, the
disadvantage of having numerous competitors outweighs the advantage of having high
phonotactic probability. It is therefore unlikely that the prime words used in the present study
were processed more easily when they had numerous competitors.
Let us now briefly consider how models that do not postulate inhibition between
competing words can account for the present results. Examples in this class of models include
the Cohort (Marslen-Wilson, 1987; Marslen-Wilson et al., 1996) and the NAM (Luce et al.,
1990) models. In these models, competitors exert their influence only at the decision stage of
recognition by delaying the process of discrimination among lexical candidates. For example
in the Cohort model, recognition occurs when the difference in activation between the target
word and its most highly activated competitors reaches a fixed value. Assuming that primes
are still activated at target onset, inhibitory priming effects can be predicted because the pre-
activation of a competitor will delay the moment at which the target word can be reliably
identified. However, we believe that the lack of connections between lexical nodes makes the
Cohort model unable to explain the present findings. Indeed, because competitors have no
direct influence on one another, the influence of a prime word on target word recognition
should remain constant whatever the number of candidates that are activated by the target.
Like the Cohort model, NAM does not include inter-word inhibition mechanisms, but
the decision about the identity of the word depends on the ratio between the activation level of
the target word to that of all other competitors. Although this model was not initially designed
to deal with priming effects, an instantiation of NAM (PARSYN; Luce et al.,2000) succeeds
in showing inhibitory priming effects. However, the network also includes inhibitory links so
that it remains actually unclear how the model might behave in the absence of inhibitory
connections. Interestingly, the WEAVER production model (Levelt, Roelofs & Meyer, 1999)
which also excludes intra-level inhibition and uses the same decision rule as NAM, was able
to simulate inhibitory priming effects by having the activation level of the prime be
differentially weighted in the decision rule used to select the target. Applying such a
mechanism in NAM should permit a previously presented competitor (the prime) to compete
more strongly with the target word during target processing, thus producing an inhibitory
priming effect. Moreover, because targets having a large set of competitors are already subject
to a greater degree of competition, the additional competition caused by the related prime
might be too small to be empirically detectable, thus accounting for the lack of inhibitory
priming effect when targets have a large set of competitors. Simulation work is nevertheless
required in order to verify these predictions.
Unlike the models discussed above, the Distributed Cohort Model (DCM; Gaskell &
Marslen-Wilson (1997, 1999, 2002)) depicts competition as interference between multiple
distributed representations. As in TRACE, competitors in the DCM directly affect the
activation level of the target word. In this model, the activation of a word is considered to be
inversely related to the distance between the output produced by the network and the expected
pattern of activation, so that a small distance (i. e., a better match) indicates a high degree of
activation. Interestingly, DCM deals with the effect of competitor set size. It has been shown
that the more competitors a word has, the larger the distance between the output produced by
the network and the expected pattern of activation is (Gaskell & Marslen-Wilson, 1997). In
addition, simulations (Gaskell & Marslen-Wilson, 1999) indicate that DCM captures the
modulation of the repetition priming effect as a function of the competitor set size (see
Vroomen & de Gelder, 1995; Gaskell & Marslen-Wilson, 2002). However, more studies are
required to determine whether or not both the inhibitory priming effect and its modulation as a
function of competitor set size can be accounted for by the model.
In conclusion, the present study indicates that the number of words that are activated
during target word presentation is an important factor in determining the size of inhibitory
priming effects. We believe that this finding supports the view that intra-level inhibition is a
crucial mechanism in accounting for inhibitory priming effects (Slowiaczek & Hamburger,
1992) and that it provides additional constraints for the modeling of auditory word
We thank three anonymous reviewers who provided useful comments on earlier versions of
this manuscript. Correspondence should be addressed to Sophie Dufour, L. E. A. D. - C. N. R.
S., Université de Bourgogne, Pôle AAFE, BP 26513, 21065 Dijon Cedex, France. Email:
Dufour, S. & Peereman, R. (submitted). Lexical Competition in phonological priming:
Assessing the role of phonological match and mismatch lengths between primes and
Dufour, S., Peereman, R., Pallier, C., & Radeau, M. (in press). VoCoLex: A lexical database
on phonological similarity between French words. L’Année Psychologique.
Dumay, N., Benraïss, A., Barriol, B., Colin, C., Radeau, M., & Besson, M. (2001). Behavioral
and electrophysiological study of phonological priming between bisyllabic spoken words.
Journal of Cognitive Neuroscience, 13, 121-143.
Eberhard, K.M. (1994). Phonological inhibition in auditory word recognition. In D.
Dagenbach & T. H. Carr (Eds.), inhibitory processes in attention memory and language.
(PP. 383-407). San Diego, CA: Academic Press.
Frauenfelder, U. H. & Peters, G. (1998). Simulating the time course of word recognition: An
analysis of lexical competition in TRACE. In J. Grainger & A. Jacobs (Eds.).
Symbolic/Connectionist approaches to Human Cognition.LEA.
Frauenfelder, U. H., Scholten, M., & Content, A. (2001). Bottom-up inhibition in lexical
selection: Phonological mismatch effects in spoken word recognition. Language and
Cognitive Processes, 16, 583-607.
Gaskell, M.G., & Marslen-Wilson, W.D. (1997). Integrating form and meaning: A distributed
model of speech perception. Language and Cognitive Processes, 12, 613-656.
Gaskell, M.G., & Marslen-Wilson, W.D. (1999). Ambiguity, competition, and blending in
spoken word recognition. Cognitive Science, 23, 439-462.
Gaskell, M.G., & Marslen-Wilson, W.D. (2002). Representation and competition in the
perception of spoken words. Cognitive Psychology, 45, 220-266.
Goldinger, S. D. (1999). Only the Shadower knows: Comment on Hamburger & Slowiaczek
(1996). Psychonomic Bulletin and Review, 6, 347-351.
Hamburger, M. B., & Slowiaczek, L. M. (1996). Phonological priming reflects lexical
competition. Psychonomic Bulletin & Review, 3, 520-525.
Levelt, W.J.M., Roelofs, A., & Meyer, A.S. (1999). A theory of lexical access in speech
production. Behavioral and Brain Sciences, 22, 1-75.
Luce, P. A., Goldinger, S. D., Auer, E. T., & Vitevitch, M. S. (2000). Phonetic priming,
neighborhood activation, and PARSYN. Perception and psychophysics, 62, 615-625.
Luce, P. A., & Pisoni, D. B. (1998). Recognizing spoken words: The Neighborhood
Activation Model. Ear & Hearing, 19, 1-36.
Luce, P. A., Pisoni, D. B., & Goldinger, S. D. (1990). Similarity neighborhoods of spoken
words. In G. T. M. Altmann (Ed.), Cognitive models of speech processing:
Psycholinguistic and computational perspectives (PP. 122-147). Cambridge, MA: MIT
Marslen-Wilson, W.D. (1987). Functional parallelism in spoken word recognition. Cognition,
Marslen-Wilson, W.D. (1990). Activation, competition, and frequency in lexical access. In G.
T. M. Altmann (Ed.), Cognitive models of speech processing: Psycholinguistic and
computational perspectives (PP. 148-172). Cambridge, MA: MIT Press.
Marslen-Wilson, W.D., Moss, H.E., & Van Halen, S. (1996). Perceptual distance and
competition in lexical access. Journal of Experimental Psychology: Human Perception
and Performance, 22, 1376-1392.
McClelland, J.L., & Elman, J.L. (1986). The TRACE model of speech perception. Cognitive
Psychology, 18, 1-86.
McClelland, J.L., & Rumelhart, D.E. (1981). An interactive activation model of context
effects in letter perception: Part 1. An account of basic findings. Psychological Review,
Monsell, S., & Hirsh, K.W. (1998). Competitor priming in spoken word recognition. Journal
of Experimental Psychology: Learning, Memory and Cognition, 24, 1495-1520.
Norris, D. (1994). SHORTLIST: a connectionist model of continuous speech recognition.
Cognition, 52, 189-234.
Norris, D., McQueen, J., & Cutler, A. (1995). Competition and segmentation in spoken word
recognition. Journal of Experimental Psychology: Learning, Memory and Cognition, 21,
Norris, D., McQueen, J., & Cutler, A. (2002). Bias effects in facilitatory phonological
priming. Memory and Cognition, 30, 399-411.
Perea, M. & Rosa, E. (in press). Repetition and form priming interact with neighborhood
density at a brief stimulus-onset asynchrony. Psychonomic Bulletin and Review.
Pitt, M.A. & Shoaf, L. (2002). Revisiting bias effects in word-initial phonological priming.
Journal of Experimental Psychology: Human Perception and Performance, 28, 1120-
Radeau, M., Morais, J., & Dewier, A. (1989). Phonological priming in spoken word
recognition: Task effects. Memory and Cognition, 17, 525-535.
Radeau, M., Morais, J., & Segui, J. (1995). Phonological priming between monosyllabic
spoken words. Journal of Experimental Psychology: Human Perception and Performance,
Segui, J., & Grainger, J. (1990). Priming word recognition with orthographic neighbors:
Effects of relative prime-target frequency. Journal of Experimental Psychology: Human
Perception and Performance, 16, 65-76.
Slowiaczek, L. M., & Hamburger, M. B. (1992). Prelexical facilitation and lexical
interference in auditory word recognition. Journal of Experimental Psychology: Learning,
Memory and Cognition, 18, 1239-1250.
Slowiaczek, L.M., McQueen, J., Soltano, E.G., & Lynch, M. (2000). Phonological
representations in prelexical speech processing: Evidence from form-based priming.
Journal of Memory and Language, 43, 530-560.
Slowiaczek, L. M., Nusbaum, H. C., & Pisoni, D. B. (1987). Phonlogical priming in auditory
word recognition. Journal of Experimental Psychology: Learning, Memory and
Cognition, 13, 64-75.
Slowiaczek, L. M., & Pisoni, D. B. (1986). Effects of phonological similarity on priming in
auditory lexical decision. Memory and Cognition, 14, 230-237.
Van Heuven, W. J. B., Dijkstra, T., Grainger, J. & Schriefers, H. (2001). Shared
neighborhood effects in masked orthographic priming. Psychonomic Bulletin & Review,
Vitevitch, M.S. (2002). Influence of onset density on spoken word recognition. Journal of
Experimental Psychology: Human Perception and Performance, 28, 270-278.
Vitevitch, M.S., & Luce, P. A. (1998). When words compete: Levels of processing in
perception of spoken words. Psychological Science, 9, 325-329.
Vitevitch, M.S., & Luce, P. A. (1999). Probabilistic phonotactics and neighborhood activation
in spoken word recognition. Journal of Memory and Language, 40, 374-408.
Vitevitch, M.S., Luce, P. A., Pisoni, D.B., & Auer, E.T. (1999). Phonotactics, Neighborhood
Activation, and lexical access for spoken words. Brain and Language, 68, 306-311.
Vroomen, J., & de Gelder, B. (1995). Metrical segmentation and lexical inhibition in spoken
word recognition. Journal of Experimental Psychology: Human Perception and
Performance, 21, 98-108.
1. Pitt and Shoaf (2002) recently claimed that the inhibition observed with an initial overlap is
the result of participants’ surprise when they encounter the first related prime – target pairs.
We will return to this suggestion in the discussion section when considering the data of the
2. Of course, the target word will also receive more inhibition when occurring in a large
competitor set, and recognition should therefore be slower. However, the inhibitory influence
of the competitors on the target word is expected to occur for both related and unrelated
prime-target trials. What is critical is that for related trials, the influence of the strongest
competitor (i. e. the word serving as prime) will be reduced due to the existence of numerous
3. The number of competitors corresponded to the number of words that have the same length
and that share the first two phonemes with the target word. Consequently, related primes and
target words had identical numbers of competitors. When considering the exact definition
used in the Cohort Model (Marslen-Wilson, 1990) — the total number of words that share the
first phonemes with the target independently of word length—, the average number of
competitors was 96 in the small set of competitors and 653 in the large set of competitors
(F(1,54)=51,09; p<.001). Using the definition proposed by the NAM model (Luce, Pisoni &
Goldinger, 1990) — words that can be generated by substitution, addition or deletion of a
single phoneme at any position in a target word— , the average number of competitors was 20
in the small set of competitors and 35 in the large set of competitors (F(1,54)=51,04; p<.001).
4. The interaction between prime type and competitor set size was also significant both by
subjects (F1(1,40)=4.83, p<.05) and by items (F2(1,54)=4.74, p<.05) when RTs were
measured from the onset of the target words.
5. In order to have an identical number of trials for each time window, two control and two
related trials occurring in the middle of the experiment were deleted for each participant.
Table 1: Mean Reaction Times (in ms) for Related and Control primes as a function of
competitor set size. (The Standard Deviations are given in parentheses).
Competitor set size
Small254 (98) 275 (99)
Large 269 (93)270 (96)
Table 2: Mean Reaction Times (in ms) for Related and Control primes as a function Time
Window for each competitor set size.
First233 259 -26
Second 253 275-22
Third 259274 -15
Second 273269 +4
Fourth 283 281+2
27 Download full-text
Appendix: Experimental stimuli used in the Experiment
Few Competitors Many Competitors