Journal of Experimental Psychology: Copyright 2002 by the American Psychological Association, Inc.

Learning, Memory, and Cognition 0278-7393/02/$5.00 DOI: 10.1037//0278-7393.28.4.735

2002, Vol. 28, No. 4, 735–747

The Influence of Phonological Similarity

Neighborhoods on Speech Production
Michael S. Vitevitch
Indiana University

The influence of phonological similarity neighborhoods on the speed and accuracy of speech production
was investigated with speech-error elicitation and picture-naming tasks. The results from 2 speech-error
elicitation techniques—the spoonerisms of laboratory induced predisposition technique (B. J. Baars,
1992; B. J. Baars & M. T. Motley, 1974; M. T. Motley & B. J. Baars, 1976) and tongue twisters—showed
that more errors were elicited for words with few similar sounding words (i.e., a sparse neighborhood)
than for words with many similar sounding words (i.e., a dense neighborhood). The results from 3
picture-naming tasks showed that words with sparse neighborhoods were also named more slowly than
words with dense neighborhoods. These findings demonstrate that multiple word forms are activated
simultaneously and influence the speed and accuracy of speech production. The implications of these
findings for current models of speech production are discussed.

Current models of spoken-word recognition treat as axiomatic (Schacter, 1999; Woodworth, 1929) or compete with each other, as
the hypothesis that acoustic-phonetic input activates multiple pho- they do in models of spoken-word recognition. Using a tip-of-the-
nological word forms that compete among each other, thereby tongue (TOT) elicitation task, Jones (1989) presented definitions
affecting the speed and accuracy of lexical access during word to participants and asked them to retrieve the word (i.e., the target)
recognition (e.g., Luce & Pisoni, 1998; Marslen-Wilson & Zwit- that fit the definition. Along with the definition, a prime that was
serlood, 1989; Norris, McQueen, & Cutler, 2000). In contrast, the semantically, phonologically, or both semantically and phonolog-
influence of phonologically related words on the speed and accu- ically related to the target was presented. Jones (1989; see also
racy of speech production is unclear. Evidence supports the hy- Jones & Langford, 1987; Maylor, 1990) found that more TOT
pothesis that words with similar forms compete with each other states were elicited when a phonologically related prime was
during speech production, as well as the hypothesis that formally presented after hearing the definition of the target. The increase in
similar words facilitate speech production. The present experi- TOT states— or the decreased ability to retrieve the target
ments attempted to better describe the nature of the activation word—in the context of a phonologically related prime suggests
among phonological word forms; do phonologically related rep- that phonologically related words compete with each other during
resentations compete among each other or facilitate processing at speech production.
the word-form level during speech production? Work by Sevald and Dell (1994) also supports the hypothesis
During the retrieval of a phonological word form in speech that formally related words compete during speech production.
production, phonologically similar words may block each other Sevald and Dell showed that speakers had slower production times
to sequences of words with the same initial sounds (e. g., cat, cab,
Michael S. Vitevitch, Department of Psychology, Indiana University. can, cad) than to sequences of words with different initial sounds
This work was supported, in part, by Research Grants DC-0265801 (to (e. g., cat, bat, mat, rat). Together, these demonstrations of slower
State University of New York at Buffalo), DC-04259 (to Indiana Univer- and less accurate speech production in the context of phonologi-
sity and University of Kansas), and DC-00111 (to Indiana University), and cally related words suggest competition among formally similar
by Training Grants DC-00036 (to State University of New York at Buffalo) words during speech production.
and DC-00012 (to Indiana University) from the National Institute on
Alternatively, phonologically similar words may facilitate the
Deafness and Other Communication Disorders, National Institutes of
Health.
activation and retrieval of a lexical word form during speech
I thank Rochelle Newman and Kenneth Allendoerfer for their assistance production (e.g., A. S. Brown, 1991; Burke, MacKay, Worthley, &
in a pilot version of Experiment 1, Lynn DiLivio and Jacqueline Weaver Wade, 1991). Meyer and Bock (1992) showed that the targets used
for their assistance in Experiment 1, Luis R. Hernandez S. for his technical by Jones (1989) differed across conditions in the susceptibility to
assistance in Experiment 2, and Markus Damian for his assistance with the TOT states. When targets with equal susceptibility to TOT states
stimuli used in Experiments 3–5. I also thank David B. Pisoni for his were used across conditions in a TOT-elicitation task, phonolog-
generosity and support while I was a postdoctoral fellow in the Psychology ical primes did not interfere with the retrieval of the target word
Department at Indiana University. Finally, I thank Paul Luce, Holly Stor-
form; rather, phonological primes aided in, or facilitated, the
kel, Alice Healy, and two anonymous reviewers for their helpful comments
retrieval of the target word form. The results of another TOT-
and suggestions.
Correspondence concerning this article should be addressed to Michael elicitation task by James and Burke (2000) further support the
S. Vitevitch, who is now at the Spoken Language Laboratory, Department hypothesis that phonologically related word forms facilitate re-
of Psychology, 1415 Jayhawk Boulevard, University of Kansas, Lawrence, trieval. James and Burke presented participants with words like
Kansas 66045. E-mail: mvitevit@ku.edu indigent, abstract, and locate and then presented the question
735
736 VITEVITCH

“What word means to formally renounce a throne?” They elicited ical neighbors) and are, therefore, more likely to be accurately
fewer TOT states when the target word, in this case abdicate, was retrieved from the lexicon.
preceded by phonologically related rather than unrelated words, sug- Similarly, Vitevitch and Sommers (2001; see also Harley &
gesting facilitated retrieval of the phonologically similar target word. Bown, 1998), using the traditional TOT-elicitation task (i.e., no
Evidence from the cross-modal picture–word interference task prime words were used; see R. Brown & McNeill, 1966), found
also supports the idea that phonologically related words facilitate that more TOT states were elicited for words with sparse rather
speech production (e.g., Costa & Sebastian-Galles, 1998; Jesche- than dense neighborhoods. As in Vitevitch (1997), it was hypoth-
niak & Schriefers, 2001; Meyer, 1996; Schriefers, Meyer, & esized that words with many phonological neighbors receive suf-
Levelt, 1990). For example, Jescheniak and Schriefers (2001) ficient amounts of activation from formally related neighbors to be
presented pictures that participants had to name while a word that completely retrieved from the lexicon. However, words with few
was either phonologically related or unrelated to the picture was phonological neighbors do not receive sufficient amounts of acti-
presented auditorily. Jescheniak and Schriefers found faster nam- vation to be completely retrieved from the lexicon, resulting in a
ing times when the auditorily presented words were phonologi- TOT state. Together, these studies support the hypothesis that
cally related rather than unrelated to the to-be-named picture, phonologically related words facilitate the retrieval of, rather than
suggesting that phonologically related words facilitate the process compete with, target words during speech production.
of speech production. Two points in the previous studies examining neighborhood
Note that the tasks used in the previous experiments relied on density in speech production should be noted. First, Vitevitch
some form of sequential presentation of relevant stimuli, or prim- (1997), Vitevitch and Sommers (2001), and Harley and Bown
ing. A word (presented visually or auditorily) was either related or (1998) used methods that did not involve priming. That is, the
unrelated to a subsequently to-be-produced item. Several studies influence of phonological similarity on speech production was not
have pointed out the limitations of priming methodologies (e.g., examined by manipulating the formal relationship between a word
Bowles & Poon, 1985; Roediger, Neely, & Blaxton, 1983). Spe- and a subsequently presented and to-be-named item. Rather, the
cifically, the relationship between the prime and the target may ability of participants to retrieve words that had many neighbors
(consciously or unconsciously) induce task-specific strategies to was compared with the ability of participants to retrieve words that
use the prime as a cue to retrieve the target. The particular retrieval had fewer neighbors. Manipulating neighborhood density rather
strategy that is induced may facilitate or inhibit the retrieval of the than the relationship between prime and target words may provide
target and may not accurately reflect the strategy used during evidence regarding the influence of phonologically similar words
normal processing. on speech production that is less prone to task-specific strategies
Rather than relying on priming methodologies, an alternative (e.g. Bowles & Poon, 1985; Roediger et al., 1983). Second, evi-
approach can be used to examine the influence of phonologically dence from these studies examining neighborhood density in
related words on speech production. Namely, words that vary in speech production suggests that phonologically related items fa-
the number of formally related neighbors can be used as targets. cilitate speech production. To further examine the influence of
The number of words that are phonologically similar to a target phonologically related word forms, I manipulated neighborhood
word is a variable that is commonly manipulated in studies of density in several experimental tasks in the present set of experi-
spoken-word recognition and is referred to as neighborhood den- ments. The results of these experiments converge on the idea that
sity (e.g., Goldinger, Luce, & Pisoni, 1989; Luce & Pisoni, 1998; simultaneously activated word forms facilitate speech production,
Vitevitch & Luce, 1998, 1999; Vitevitch, Luce, Pisoni, & Auer, a finding that may be difficult for some models of speech produc-
1999). A word with many similar sounding words has a dense tion to account for.
neighborhood, whereas a word with few similar sounding words
has a sparse neighborhood. Experiment 1
Previous studies examining neighborhood density in speech
production have found evidence to suggest that phonologically The spoonerisms of laboratory induced predisposition (SLIP)
similar words facilitate processing (Harley & Bown, 1998; technique (Baars, 1992; Baars & Motley, 1974; Motley & Baars,
Vitevitch, 1997; Vitevitch & Sommers, 2001). Vitevitch (1997) 1976) was used to elicit phonological speech errors on words that
examined the neighborhood density characteristics of whole-word varied in neighborhood density. The SLIP technique elicits pho-
speech errors known as malapropisms (e.g., saying octane instead nological speech errors by activating two incompatible speech
of octave) that were collected by means of naturalistic observation plans. The competition between the two incompatible speech plans
(Fay & Cutler, 1977). The neighborhood density of the target and increases the likelihood of making a speech error when prompted
the error that was produced were compared with 10 samples of to produce a verbal response. Competition among speech plans is
words of comparable length and syntactic class that were randomly accomplished by instructing a participant to repeat to themselves
sampled from a computer-readable version of Webster’s Pocket word pairs that are rapidly presented on a computer screen. In each
Dictionary, which contains approximately 20,000 words (Nus- word pair, the initial phoneme of the first word is, for example, /p/,
baum, Pisoni, & Davis, 1984). The results showed that targets and and the initial phoneme of the second word is, for example, /b/
errors had fewer similar sounding words (i.e., sparser neighbor- (e.g., push–big, pig–bull, and pin–ban), strongly activating a p–b
hoods) than the words randomly sampled from the lexicon. It was speech plan. Occasionally, participants are cued (by a tone or
hypothesized that words that have sparse neighborhoods do not visual cue) to say a word pair out loud. In the word pair that must
receive sufficient activation to be accurately retrieved from the be produced, the initial phoneme of the first word is /b/, and the
lexicon, resulting in a malapropism. In contrast, words with denser initial phoneme of the second word is /p/, such as beach–palm.
neighborhoods receive sufficient activation (from more phonolog- This creates competition between the p–b speech plan that was
 PHONOLOGICAL NEIGHBORHOODS IN SPEECH PRODUCTION 737

activated by the preceding word pairs and the b–p speech plan that relevant conditions. In the present experiment, this resulted in eight or-
must be used to correctly produce the cued word pair. The com- thogonal stimulus conditions (high vs. low frequency, dense vs. sparse
petition between the two speech plans may result in the production neighborhood density, high vs. low neighborhood frequency) with 14
of peach–balm, an induced speech error, instead of the intended words per condition. Unless otherwise specified, all analyses were signif-
beach–palm. icant at p ⬍ .05. The initial consonants found in the eight conditions did not
differ significantly across conditions, ␹2(126, N ⫽ 112) ⫽ 146. Words in
Although other responses that differ from the intended word pair
the high-frequency conditions had a mean frequency of occurrence of 45.0
may also be produced, these responses are not counted as speech per million, and words in the low-frequency conditions had a mean
errors. Speech errors induced in elicitation tasks, such as the SLIP frequency of occurrence of 3.0 per million, F(1, 110) ⫽ 197, MSE ⫽ 380.
technique, are similar in kind to naturally occurring speech errors All of the frequency and neighborhood frequency counts in the experi-
collected in various error corpora (Cutler, 1982; Ferber, 1991; ments presented are based on Kučera and Francis (1967).
Stemberger, 1992). Furthermore, the errors produced in the SLIP As in the spoken-word recognition literature (e.g., Greenberg & Jenkins,
paradigm are not artifacts of proactive interference or confusions 1964; Landauer & Streeter, 1973; Luce, 1986; Luce & Pisoni, 1998),
in short-term memory (cf. Motley, 1986, and Sinsabaugh & Fox, neighborhood density was defined as the number of words that were
1986). The SLIP technique is simply one of many competing-plans similar to a target on the basis of the addition, deletion, or substitution of
techniques (Baars, 1992; Bock, 1996) that have been used to a single phoneme in the target item. For example, in the word cat [/kæt/],
examine phonological speech errors (Dell, 1984, 1986, 1990; the words scat [/skæt/], at [/æt/], hat [/hæt/], cut [/k∧t/] and cap, [/kæp/],
as well as other words found in the computer readable version of the
Levitt & Healy, 1985; Shattuck-Hufnagel & Klatt, 1979; Stem-
Webster’s Pocket Dictionary (with a familiarity rating of 6 or higher;
berger & Treiman, 1986), syntactic representations (e.g., Bock,
Nusbaum, Pisoni, & Davis, 1984) would be considered neighbors. Famil-
1986), and idiom blends (Cutting & Bock, 1997) in speech pro- iarity ratings of 6 or higher were used so that the stimuli and the estimate
duction. The advantage of using such techniques includes the of neighborhood density were based on words that were familiar to most of
precise calculation, rather than estimation, of actual error proba- the participants. The mean value for stimuli in the dense-neighborhood
bilities (Motley & Baars, 1976). These techniques also allow for condition was 24.86 neighbors and in the sparse-neighborhood condition
the control and manipulation of selected variables, such as word was 14.50 neighbors, F(1, 110) ⫽ 176, MSE ⫽ 1,938.
frequency, neighborhood density, and neighborhood frequency in Neighborhood frequency is the mean frequency of the neighbors of the
the present experiment. target word. Words in the high-neighborhood-frequency conditions had a
Dell (1988, 1990) and Stemberger and MacWhinney (1986) mean frequency of occurrence of 19.0 per million, whereas words in the
found more speech errors for rare rather than for common words in low-neighborhood-frequency conditions had a mean frequency of occur-
the English language, suggesting that word frequency affects rence of 4.5 per million, F(1, 110) ⫽ 274, MSE ⫽ 10. The mean and
standard errors of each variable for each condition are listed in Table 1.
speech production. Word frequency was manipulated in the
The 14 stimulus words in each condition were grouped to form seven
present experiment in an attempt to replicate these results. Neigh-
word pairs. Each word in the pair was similar in word frequency, neigh-
borhood density refers to the number of similar sounding neigh- borhood density, and neighborhood frequency and had minimal overlap of
bors a target word has and was manipulated to examine the the consonants or vowels; the frequency with which overlap occurred did
influence that the simultaneous activation of multiple word candi- not differ across conditions (F ⬍ 1). Furthermore, when the initial conso-
dates might have on speech production. Finally, neighborhood nants of each word in the pair were switched, either a real word or a
frequency refers to the mean frequency of occurrence of the pronounceable nonword was formed. There were no cases in which both
neighbors (Luce & Pisoni, 1998). Luce and Pisoni (1998) have words in the pair formed nonwords. Note that Baars, Motley, and MacKay
shown that this variable (along with density and frequency) influ- (1975), among others, have suggested that there may be a bias in the speech
ences the speed and accuracy of lexical retrieval during spoken- production system to output lexical items and prohibit the output of
word recognition. Neighborhood frequency was manipulated in the nonlexical items. The data they used to support this claim, however, were
present experiment to see if the frequency of similar sounding based on elicited speech errors in which both words in the pair formed
nonwords. It is unclear whether the so-called lexical bias affects word pairs
words also influenced lexical retrieval in speech production.
in which only one item in the pair forms a nonword when switched.
Furthermore, Dell (1986) found an interaction between lexical bias and
Method output cue deadline. Specifically the lexical bias was present when partic-
Participants. In all the experiments reported, participants were native ipants were cued to produce a response at longer deadlines (700 and 1,000
English speakers with normal or corrected-to-normal vision and no history ms), but it was not present when participants produced a response at a short
of speech or hearing problems as determined by self-report. None of the deadline (500 ms). A deadline of 600 ms (relatively short compared with
participants who took part in a given experiment took part in any of the that in Dell’s, 1986, study) was used in the present experiment to decrease
other experiments. In this experiment, 78 speakers from the State Univer- the lexical bias in error output.
sity of New York at Buffalo pool of introductory psychology students Additional words were paired to act as interference and filler word pairs.
participated in partial fulfillment of a course requirement. The data from 2 The interference word pairs were constructed according to the criteria
participants were not included in the analyses because of technical prob- described in Motley and Baars (1976): The first interference pair contained
lems that occurred during the experiment resulting in only part of the word-initial phonemes that were different from those of the targeted error,
session being recorded. or spoonerism, but resembled the spoonerism in all other respects. The
Materials. The stimuli consisted of 112 consonant–vowel– consonant second interference pair differed from the spoonerism in the initial pho-
(CVC) words. The mean familiarity rating for the words was 6.84 on the neme of the second word. The third interference pair differed from the
basis of a 7-point subjective rating scale of familiarity that ranged from 1 spoonerism in the initial phonemes of both words. In all cases, the inter-
(don’t know the word) to 7 (know the word and know its meaning). All of ference words were as similar in all other respects to the spoonerism as
the subjective familiarity ratings in the experiments presented here are possible. By way of example, the target pair name–life had the following
based on Nusbaum, Pisoni, and Davis’s (1984) study. The median value as interference pairs: same-strife, lake-fife, and late-nine. Across the two
was used in all of the experiments to equally divide the words among the lists and across the eight conditions there were no differences in word
738 VITEVITCH

Table 1
Mean Values by Condition for the Stimuli in Experiment 1

High frequency Low frequency

Dense Sparse Dense Sparse

High NF Low NF High NF Low NF High NF Low NF High NF Low NF

Variable M SE M SE M SE M SE M SE M SE M SE M SE

Freq. 61.9 1.6 21.5 0.7 58.9 1.4 32.9 1.4 7.6 1.1 3.3 0.8 3.3 0.5 2.4 1.6
Density 23.71 0.8 20.43 1.1 14.07 1.1 11.64 1.1 22.43 1.5 21.50 0.9 14.56 1.1 12.36 1.1
NF 24.8 0.4 7.0 0.3 22.0 0.4 4.2 0.4 16.3 0.4 5.9 0.3 17.8 0.4 3.6 0.4

Note. Freq. ⫽ word frequency in occurrences per million; NF ⫽ neighborhood frequency in occurrences per million.

frequency, neighborhood density, or neighborhood frequency for the in- there was disagreement between the initial and second coding were
terference pairs (all Fs ⬍ 1). The filler word pairs contained words that resolved by an independent judge who was naive in regards to the
were not part of the stimulus or interference pairs. Filler pairs were nature of the experiment. Following the scoring conventions in
randomly dispersed among the interference-stimulus groupings to com- Baars et al. (1975), a response was scored as a speech error if the
plete the experimental list. In total, 486 word pairs were presented to
response was either a complete or an incomplete reversal of the
participants: 56 stimulus pairs, 168 interference pairs, and 262 filler pairs.
initial phonemes of the word pair. As per Baars et al. (1975), some
Procedure. Participants were seated at a comfortable distance from a
Macintosh Centris 650 computer with a 13-inch Macintosh monitor used responses to the stimulus items were not counted as speech errors.
for stimulus presentation. The participants were instructed to repeat to These responses included errors not involving initial consonants,
themselves each pair of words that appeared on the monitor and to be producing a word not in the cued pair, failures to repeat any of the
prepared to say some of the word pairs out loud when periodically cued by words in the cued pair, and errors in which participants misread or
the computer. mispronounced a word (e.g., saying lamb for lame or tone for ton).
Each participant received all of the word pairs with 112 pairs being cued These errors accounted for less than 1% of the responses made by
for a response. Half of the cued-word pairs were the target stimuli (and are participants. There was no difference across conditions or across
available from Michael S. Vitevitch on request, as are all the stimuli), and lists (all Fs ⬍ 1) for these other kinds of responses.
half were filler items that were cued to prevent participants from noticing
Because a set of highly controlled, nonrandomly selected stimuli
a pattern in the stimuli. No word (or pair of words) was presented more
that almost exhausted the pool of possible items was used in this
than once.
Each participant received one of two lists that differed in the order of the and all the experiments that follow, only analyses of variance
stimulus (e.g., name–life on one list; life–name on the other). The word (ANOVAs) with participants as a random factor were conducted
pairs were also presented in different pseudorandom orders (to maintain (J. Cohen, 1976; Hino & Lupker, 2000; Keppel, 1976; Raaijmak-
interference-stimulus groupings) on each list. Each word pair appeared in ers, Schrijnemakers, & Gremmen, 1999; Smith, 1976; Wike &
the center of the monitor for 900 ms. An interstimulus interval (consisting Church, 1976). Because there were no significant main effects or
of a blank screen) of 100 ms separated each pair. interactions for list/word order (all Fs ⬍ 1), all further analyses
Periodically, participants were cued to repeat the previously presented were collapsed across lists. The lack of a difference for list/word
pair of words by a string of eight question marks that appeared on the order also suggests that the order of the words (or the ordering of
screen in the same location as the word pairs. The visual cue remained on
the initial segments in those words; see Levitt & Healy, 1985) was
the screen for 900 ms. A computer beep was presented 600 ms after the
not a major factor in eliciting speech errors in this experiment.
onset of the visual cue to encourage participants to respond rapidly.
Participants were encouraged in the instructions to repeat the word pairs The percentage of speech errors elicited for each condition is
prior to the onset of the auditory cue if possible. The auditory cue was used displayed in Table 2. A significant main effect of frequency was
only to encourage participants to repeat the word pairs quickly; it was not found, F(1, 75) ⫽ 14.3, MSE ⫽ 3; more speech errors were elicited
used as an exclusion criterion for responses. for low-frequency word pairs (31.9%) than for high-frequency
After the offset of the visual cue to respond, 200 ms elapsed before word pairs (16.4%). A significant main effect of neighborhood
another stimulus pair was presented. Responses from the participants were density was also found, F(1, 75) ⫽ 15.3, MSE ⫽ 3; more errors
recorded on high quality audiotape, using a microphone (Shure 5755; were elicited for word pairs from sparse neighborhoods (31.6%)
Evanston, IL) and a tape recorder (Marantz, PMD221; Aurora, IL) to be than for word pairs from dense neighborhoods (16.8%). No dif-
scored at a later time. The number of filler pairs that occurred between each
ference in neighborhood frequency, nor any significant interac-
cued pair (whether the cued pair was a stimulus or filler item) in the
tions were observed (all Fs ⬍ 1). The overall mean number of
practice session and in the experimental list ranged from 2 to 8 word pairs.
The experimental session was preceded by a short practice session con- speech errors in this experiment (24.2%) was within the expected
sisting of 20 pairs of words, 6 of which were cued for a verbal response. range of speech errors for this task (up to 30%; Motley & Baars,
1976).
Results
Discussion
The recorded responses were examined for speech errors. In-
trarater reliability (with at least 36 months passing between the The results of the present experiment showed that more speech
initial and second coding) was very high (97.6%). Cases in which errors were elicited for word pairs with low rather than high
 PHONOLOGICAL NEIGHBORHOODS IN SPEECH PRODUCTION 739

Table 2
The Rate of Speech Errors for Each Condition

High frequency Low frequency

Dense Sparse Dense Sparse

High NF Low NF High NF Low NF High NF Low NF High NF Low NF

% SE % SE % SE % SE % SE % SE % SE % SE

17.1 0.5 7.9 0.3 23.7 0.6 17.1 0.5 27.6 0.6 14.5 0.4 42.1 0.7 43.4 0.8

Note. NF ⫽ neighborhood frequency.

frequency of occurrence, replicating analyses of speech-error cor- each condition rules out the possibility that the observed effects
pora (Stemberger & MacWhinney, 1986) and studies using error- may be due to differences in the phonological segments in each
elicitation techniques (Dell, 1988, 1990). Although there was a condition. Furthermore, the stimuli in the present experiment had
significant effect of word frequency, no influence of neighborhood equivalent familiarity, word frequency, and neighborhood fre-
frequency was observed (cf. Vitevitch & Sommers, 2001). quency, focusing only on the influences of neighborhood density
More important, the results of the present experiment showed in speech production.
that more speech errors were elicited for word pairs with sparse
rather than with dense neighborhoods. The influence of phonolog-
ical similarity neighborhoods observed in the current experiment is Method
consistent with the results of Vitevitch (1997) and Vitevitch and Participants. Twenty-eight speakers from the Indiana University pool
Sommers (2001; see also Harley & Bown, 1998). Recall that of introductory psychology students participated in partial fulfillment of a
Vitevitch (1997) found that malapropisms tended to have sparser course requirement.
neighborhoods than comparable words randomly selected from the Materials. Ten pairs of highly confusable target segments (from Ex-
lexicon, and Vitevitch and Sommers (2001; see also Harley & periment 2 of Shattuck-Hufnagel, 1992) were used to select CVC words for
Bown, 1998) found that more TOT states were elicited in college- the tongue twisters in this experiment. Twenty tongue twisters, each
age adults for words that had sparse rather than dense neighbor- containing four words of comparable neighborhood density, were created.
Half of the tongue twisters consisted of words from sparse neighborhoods,
hoods. Together, these results suggest that multiple word forms
and the other half consisted of words from dense neighborhoods.
become activated during speech production. Furthermore, simul- Neighborhood density was computed as in Experiment 1. The median
taneously active word forms facilitate processing in speech pro- value was used to separate the words into stimuli with either dense or
duction rather than compete among each other. That is, lexical sparse neighborhoods. In the dense condition the mean number of neigh-
representations with many similar sounding neighbors receive a bors was 23.9 words, and in the sparse condition the mean number of
greater amount of activation than lexical representations with few neighbors was 15.4 words. The difference between the dense and sparse
similar sounding neighbors, supporting the more accurate retrieval neighborhood conditions was significant, F(1, 78) ⫽ 143, MSE ⫽ 1,453.
of the target word form for words in dense neighborhoods. To Although the stimuli differed in neighborhood density, the words in each
further examine how the simultaneous activation of multiple word condition did not differ in word frequency, F(1, 78) ⫽ 2.0, MSE ⫽ 0.83.
forms influences the accuracy of producing words, I selected The mean frequency of the items was 6.9 occurrences per million in the
dense condition and 10.7 occurrences per million in the sparse condition.
another set of words varying in neighborhood density for use in a
The two sets of words also did not differ in neighborhood frequency, F(1,
different phonological speech error elicitation task. 78) ⫽ 1.2, MSE ⫽ 0.1. The mean neighborhood frequency values of the
items was 16.3 occurrences per million in the dense condition and 14.0
Experiment 2 occurrences per million in the sparse condition. Finally, the words in each
condition were also equivalent in subjective familiarity ratings, F(1,
Experiment 2 attempted to replicate the results of Experiment 1 78) ⫽ 1.2, MSE ⫽ 0.2 (dense condition: M ⫽ 6.7; sparse condition:
with a different set of stimulus words and a different speech-error- M ⫽ 6.8).
elicitation task, namely the tongue twister task (Shattuck- Procedure. Participants were seated individually in a soundproof
Hufnagel, 1992). The tongue twister task, like the SLIP task, elicits booth (IAC Model 402; Bronx, NY) equipped with a 13-inch monitor
speech errors from participants by activating competing speech (Gateway 2000 Crystal Scan 1024 CRT; San Diego, CA) and a head-
plans (Baars, 1992; Bock, 1996). The tongue twister task differs mounted microphone (Shure SM-98; Evanston, IL). A 486 PC computer
from the SLIP task in that words are repeated rapidly rather than presented a prompt (“Please repeat the following words six times in a
presented rapidly. row.”) in the center of the monitor for 3 s. After the offset of the prompt,
a tongue twister was randomly presented in the center of the monitor for
The stimuli used in Experiment 2 were even more rigorously
12 s. Participants were instructed to repeat the tongue twister six times as
controlled than were the stimuli in Experiment 1. Although there quickly as they could. The tongue twister remained on the monitor for the
was no difference in the distribution of the initial segments across entire duration. Responses were recorded on a Sony (New York, NY)
conditions in Experiment 1, the stimuli used in the present exper- TCD-D8 tape recorder, using high quality audiotape, for later analysis. At
iment had equal numbers of initial segments in each condition. the end of 12 s, the prompt was presented on the monitor, and a new trial
Having equal numbers of words with the same initial phoneme in began.
740 VITEVITCH

A practice session using five pseudo–tongue twisters, each composed of dense neighborhoods. In Experiments 3–5, the influence of neigh-
four randomly selected monosyllabic words not included in the stimulus borhood density on the speed of lexical access during speech
set, familiarized the participants with the task. The responses from the production was examined.
practice session were not included in the final analyses.

Experiment 3
Results
A great deal has been learned from spontaneous and experimen-
The recorded responses of each participant were examined for tally elicited speech errors. Indeed, many models of speech pro-
accuracy. Intrarater reliability (with at least 24 months passing duction were developed to account for speech-error data (e.g.,
between the initial and second coding) was very high (94.4%). Dell, 1986, 1988; Fay & Cutler, 1977; MacKay, 1987; Shattuck-
Cases in which there was disagreement between the initial and Hufnagel, 1979). However, Levelt, Roelofs, and Meyer (1999)
second coding were resolved by an independent judge who was have argued that
naive in regards to the nature of the experiment. To maintain
consistency with Experiment 1, I did not score the responses of the models of lexical access have always been conceived as process
present experiment as perseverations, anticipations, or reversals of models of normal speech production. Their ultimate test . . . cannot lie
in how they account for infrequent derailments of the process but
the initial phonemes but only as speech errors. For example, if the
rather must lie in how they deal with the normal process itself. RT
tongue twister was peach balm bull pig but the participant re-
studies, of object naming in particular, can bring us much closer to
sponded “beach balm bull big,” (note that two words have incor- this ideal . . . [because] . . . object naming is a normal, everyday ac-
rect initial phonemes) this repetition was scored as a single speech tivity . . . [and] . . . reaction time measurement is still an ideal proce-
error. dure for analyzing the time course of a mental process. (p. 2)
Repetitions that were not correct but not counted as speech
errors (less than 1% of the responses made by participants) in- To meet the challenge that Levelt et al. (1999) have set for speech
cluded errors not involving initial consonants (e.g., peep for production research, I used an object-naming task (also known as
peach), substitutions of words other than those presented, failures a picture-naming task; Oldfield & Wingfield, 1965) in Experi-
to repeat one of the four words in the list, and errors in which ment 3 to examine how monosyllabic words varying in neighbor-
participants misread a word (e.g., made for mead or doze for dose). hood density influence the speed of lexical access during speech
There was no difference for these responses between conditions production. Given the facilitative effects of neighborhood density
(F ⬍ 1). observed in the previous experiments, it was predicted that partic-
For each density condition, there were 10 tongue twisters (re- ipants would more quickly name pictures illustrating words from
peated six times each). Thus, there were 60 opportunities to dense rather than sparse neighborhoods.
correctly repeat tongue twisters containing dense words and 60
opportunities to correctly repeat tongue twisters containing sparse Method
words. A significant difference in the number of errors elicited
between conditions was observed, F(1, 27) ⫽ 16.8, MSE ⫽ 154. Participants. Thirty-four participants from the same population sam-
More erroneous repetitions were elicited from tongue twisters pled in Experiment 2 took part in this experiment.
Materials. Forty-eight line drawings (Snodgrass & Vanderwart, 1980),
containing words with sparse neighborhoods (M ⫽ 12%, SEM ⫽
half of which illustrated words from sparse neighborhoods and the other
1) than from tongue twisters containing words with dense neigh- half of which illustrated words from dense neighborhoods, were used as
borhoods (M ⫽ 7%, SEM ⫽ 0.9). The overall mean percentage of stimuli in the present experiment. The words from sparse neighborhoods
speech errors in this experiment (9.5%) was comparable to the had significantly fewer neighbors (M ⫽ 6.8 words) than the words from
percentage of speech errors made in other tasks (Motley & Baars, dense neighborhoods (M ⫽ 19.4 words), F(1, 46) ⫽ 107, MSE ⫽ 1,887.
1976). Although the difference in neighborhood density of the two conditions
was significantly different, the differences in familiarity ratings, word
frequency, and neighborhood frequency were not, all Fs(1, 46) ⬍ 1. Words
Discussion from sparse neighborhoods had a mean familiarity rating of 6.9, a mean
frequency of 38.5 occurrences per million, and a mean neighborhood
In Experiment 2 a different error-elicitation task, different stim- frequency of 16.5 occurrences per million. Words from dense neighbor-
uli, and a sample of participants from a different university were hoods had a mean familiarity rating of 6.9, a mean frequency of 24.2
used. The results, however, are the same as those in Experiment 1: occurrences per million, and a mean neighborhood frequency of 17.6
More speech errors were elicited for words with sparse neighbor- occurrences per million. There was also no difference in the distribution of
hoods than for words with dense neighborhoods. The results of the initial phonemes used in each set of stimulus words, ␹2(13, N ⫽
Experiment 2 are also consistent with the results of other studies 48) ⫽ 9.27.
investigating the role of neighborhood density in speech produc- Procedure. Participants studied a booklet that on each page contained
tion (Harley & Bown, 1998; Vitevitch, 1997; Vitevitch & Som- the stimulus picture and the monosyllabic word that identified that picture.
mers, 2001). These results suggest that multiple word forms be- When participants were confident that they could use the given label for
each picture, they were seated in front of a Macintosh Quadra 950, with a
come activated simultaneously during lexical access and influence
17-inch Macintosh monitor, that was running PsyScope 1.2.2 (J. D. Cohen,
the accuracy of speech production. Furthermore, the multiple word MacWhinney, Flatt, & Provost, 1993), which controlled stimulus random-
forms activated in memory facilitate rather than inhibit or compete ization and presentation and the collection of response latencies. A
during lexical access in speech production. Word forms with many headphone-mounted microphone (Beyer-Dynamic DT109, Heilbronn, Ger-
neighbors receive a greater amount of activation than word forms many) was interfaced to a PsyScope button box that acted as a voice key
with fewer neighbors, facilitating the accurate retrieval of words in with millisecond accuracy. A typical trial proceeded as follows: The word
 PHONOLOGICAL NEIGHBORHOODS IN SPEECH PRODUCTION 741

READY appeared in the center of the monitor for 500 ms. One of the 48 back down to the phonological nodes, thereby increasing the
randomly selected stimulus pictures was then presented and remained activation of those shared phonological nodes. The activation that
visible until a verbal response was initiated. Response latency, measured is sent to the phonological nodes from similar sounding word
from the beginning of the stimulus, was triggered by the onset of the forms will further activate those phonological nodes, which will in
participant’s verbal response. Another trial began 1 s after a response was
turn increase the activation of the target word that is composed of
made. Responses were also recorded, on high quality audiotape, for later
those phonemes. The higher levels of activation that the target
accuracy analyses. No picture was presented more than once.
word receives from similar sounding words via the shared phono-
logical nodes will increase the probability that the target word
Results (being the highest activated representation) will be selected.
The tape-recorded responses of each participant were scored for Thus, in an interactive model of speech production, a target
accuracy. Only accurate responses were included in the repeated word with many phonological neighbors (i.e., a dense neighbor-
measures ANOVA for response latency. Errors included responses hood) will receive activation from many similar sounding words
that were words other than the given label (e.g., responding with via the shared phonological nodes. A target word with fewer
“sofa” instead of “couch”) and improperly triggering the voice key neighbors (i.e., a sparse neighborhood) will receive activation from
(e.g., by coughing or saying “uh”). A significant main effect of few similar sounding words via the shared phonological nodes.
neighborhood density was found, F(1, 33) ⫽ 8.3, MSE ⫽ 8,768. The difference in the number of similar words contributing to the
Participants responded to words from dense neighborhoods more activation that is sent to the target via the phonological nodes
quickly (716 ms, SEM ⫽ 9) than to words from sparse neighbor- results in words with dense neighborhoods being produced faster
hoods (739 ms, SEM ⫽ 11). There was no difference in error rates and more accurately than words with sparse neighborhoods.
between the two sets of words (F ⬍ 1), suggesting that participants Levelt et al. (1999) described a different model of speech
did not sacrifice speed for accuracy in making their responses. production, WEAVER⫹⫹, that has a strictly feedforward archi-
Words from dense neighborhoods were correctly responded tecture. That is, activation at the word-form level cannot spread
to 94.4% of the time, and words from sparse neighborhoods were “backward” to influence the activation of a lemma, nor can acti-
correctly responded to 94.0% of the time. vation among phonological segments spread “backward” to influ-
ence the activation of word forms. (The only “feedback” in
WEAVER⫹⫹ is indirectly through the speech comprehension
Discussion
system, which is not considered feedback in the traditional sense.)
The results of the present experiment show that words from Like most models of speech production, there are no lateral con-
dense neighborhoods are produced more quickly than words from nections between representations within a level. Levelt et al.
sparse neighborhoods, suggesting that multiple word forms do described how their model could—without lateral inhibitory con-
become simultaneously activated in memory and do influence the nections—account for inhibitory effects of phonologically similar
speed in addition to the accuracy of speech production. The results words observed in some speech production tasks (e.g., Sevald &
of the present experiment further suggest that multiple word forms Dell, 1994). The mechanism they described involved the weight-
activated simultaneously facilitate speech production. ing of recently activated syllable nodes in the (Luce choice)
The results of Experiments 1–3 suggest that many phonological decision rule. This resulted in inhibitory effects on subsequently
neighbors facilitate the accurate and rapid retrieval of word forms. produced words if they had similar syllable nodes. Note, however,
How might a model of speech production account for the facili- that this weighting mechanism accounts only for inhibitory effects
tative effects of simultaneously activated phonologically related on subsequently presented words. This mechanism does not ad-
words on speech production? Current models of speech production dress the issue investigated in the present set of experiments,
can be generally classified into one of two types of models: namely the influence of phonologically related words that are
interactive and feedforward models. An example of an interactive simultaneously activated during the retrieval and production of
model of speech production is described in Dell (1986), and an isolated words.
example of a feedforward model of speech production is described In section 5.2.1, Levelt et al. (1999) also discussed how
in Levelt et al. (1999). WEAVER⫹⫹ accounts for some facilitative effects observed in
In Dell’s (1986) interactive model of speech production (indeed, the literature. The facilitative effects they discussed, however, are
in most models of speech production) there are no lateral connec- facilitative effects among words that are semantically related. It is
tions between representations within a level. Without lateral con- unclear whether the mechanisms described in section 5.2.1 of
nections between similar sounding word forms, an interactive Levelt et al. would also apply to words that are phonologically
model of speech production can still account for the facilitative related. Furthermore, given the strictly feedforward architecture of
effects of neighborhood density in the following way. When the WEAVER⫹⫹ and the constraint that only selected lemmas be-
representation of a word form (cat) is partially activated by se- come phonologically activated (Levelt et al., 1991), it is unclear if
mantic information, the word form will partially activate the pho- multiple word forms that are phonologically related can even
nological nodes that constitute it (/k/, /æ/, /t/). (Note that in an become activated simultaneously. Levelt et al. (see also Roelofs,
interactive model other word forms may be partially activated by 1992) did suggest that multiple word forms may be activated in
semantic information, but for ease of explication we only follow situations in which two (or presumably more) lemmas are equally
the activation of cat.) The activated phonological nodes (/k/, /æ/, activated and selected. However, given the arbitrary relationship
/t/) will feed activation back to the word-form level to all the word between meaning and sound (e.g., Saussure, 1916/1966), it is
forms that contain those phonemes (e.g., hat, cut, cap). Those unlikely that these semantically related representations would also
phonologically related word forms will in turn send activation be phonologically related (e.g., sofa and couch). In its present
742 VITEVITCH

form, it is not clear that the strictly feedforward model, 1–3 were due to neighborhood density. Given the language-wide
WEAVER⫹⫹, can account for the facilitative effects of simulta- correlation between phonotactic probability and neighborhood
neously activated and phonologically related words observed in density and the effects of phonotactic constraints on speech pro-
Experiments 1–3. duction demonstrated by Dell et al. (2000; see also Motley &
WEAVER⫹⫹ might be able to account for the effects observed Baars, 1975), it is possible that the effects observed in Experiments
in Experiments 1–3 if (a) the effects observed in Experiments 1–3 1–3 were due to differences in phonotactic probability and not due
were due to phoneme frequency (i.e., phonotactic probability) to differences in the number of words simultaneously activated in
rather than neighborhood density, and (b) the phonological nodes memory. To rule out the possibility that phonotactic probabilities
in the model were sensitive to the frequency with which those were the source of the observed effects, this variable was con-
phonemes occur (nota bene, in its present form, frequency is trolled in the following experiments. Furthermore, only CVC
encoded only at the word-form level in WEAVER⫹⫹). words with the same initial segments were used as stimuli in each
Taking each of these points in turn, phonotactic probability condition. Given that the same initial segments and the overall
refers to the frequency that a particular segment or sequence of
frequency of occurrence of the segments in the words are equiv-
segments occurs in a given position in a word or syllable
alent, no difference between the two conditions should be observed
(Vitevitch, Luce, Charles-Luce, & Kemmerer, 1997). Vitevitch et
if differences among the phonological segments that constitute the
al. (1999) found a positive correlation between phonotactic prob-
words are the source of the effects observed in Experiments 1–3.
ability and neighborhood density. Common segments and se-
Alternatively, if a difference in the number of simultaneously
quences—those with high phonotactic probability—tend to be
segments and sequences that are found in many words. Con- activated similar sounding words (i.e., neighborhood density) is
versely, patterns with low-probability phonotactics typically occur the locus of the effects observed in Experiments 1–3, then effects
in words with sparse phonological neighborhoods. Work by Dell, of neighborhood density should be observed in the present
Reed, Adams, & Meyer (2000) suggested that the frequency with experiment.
which segments occur in the language (and within the context of
the experiment) influences the frequency of occurrence of certain
Method
errors elicited experimentally. It is, therefore, possible that the
results of Experiments 1–3 (which did not control phonotactic Participants. Twenty-five participants from the same population sam-
probability) were the result of the difference in frequency among pled in Experiment 2 took part in this experiment.
the phonological segments and sequences of segments, rather than Materials. Forty-eight monosyllabic words with a CVC syllable struc-
the difference in the number of similar sounding words. If the ture were used as labels for line drawings selected from Snodgrass and
results of Experiments 1–3 were indeed due to frequency differ- Vanderwart (1980) and Cycowicz, Friedman, Rothstein, and Snodgrass
ences among the phonological segments, WEAVER⫹⫹ might be (1997). All lexical characteristics were assessed in the same manner as in
Experiments 1–3. The words from sparse neighborhoods had a mean
able to account for the present findings if it is modified to weight
density of 11.72 neighbors, and the words from dense neighborhoods had
the activation of phonological segments as a function of their
a mean density of 21.38 neighbors. This difference was significant, F(1,
frequency of occurrence.
46) ⫽ 103, MSE ⫽ 1,102.
Given the role that phonological segments play in the interactive Although the difference in neighborhood density of the two conditions
account of the present results, it is also important to rule out the was significantly different, the differences in familiarity ratings, word
possibility that the observed effects are due solely to activity at the frequency, and neighborhood frequency were not, Fs (1, 46) ⬍ 1. Words
level of phonological segments. Recall that in the interactive from sparse neighborhoods had a mean familiarity rating of 6.9, a mean
account activation spread from the partially activated word node frequency of 11.5 occurrences per million, and a mean neighborhood
via phonological segments to phonologically related word forms frequency of 21.5 occurrences per million. Words from dense neighbor-
(and back again to increase the activation of the target). If the hoods had a mean familiarity rating of 6.9, a mean frequency value of 11.0
observed effects are due solely to activity at the level of phono- occurrences per million, and a mean neighborhood frequency of 20.5
logical segments, then an interactive model may not be required to occurrences per million.
account for the observed effects. However, if neighborhood den- To control the phonological segments used in the stimuli, I ensured that
sity effects are still observed when the frequency of the phono- equal numbers of words in each stimulus condition contained the same
logical segments (i.e., phonotactic probability) is controlled, then initial phonemes. Furthermore, the phonotactic probabilities of the words
only a model that allows activation to feedback from representa- in the two conditions were also equivalent. Phonotactic probability was
tions of phonological segments to representations of word forms calculated with the same two measures that have been used extensively in
other studies of phonotactic probability (e.g., Jusczyk, Luce, & Charles-
can account for these findings. To better determine the locus of the
Luce, 1994; Storkel & Rogers, 2000; Vitevitch & Luce, 1998, 1999;
neighborhood-density effect in speech production, and to adjudi-
Vitevitch et al., 1997). These two measures are the sum of the positional
cate between an interactive and feedforward account of the find-
segment probability and the sum of the biphone probability for the three
ings, the picture-naming task was used in Experiments 4 and 5. segments and two biphones in each word. The average probabilities for the
segments in the given word positions for the items with sparse neighbor-
Experiment 4 hoods was .144, and for the items with dense neighborhoods it was .151,
F(1, 46) ⬍ 1. The average probabilities for the biphones for the items with
In the present experiment the picture-naming task was used with sparse neighborhoods was .005, and for the items with dense neighbor-
stimuli that had equivalent phonotactic probability, word fre- hoods it was .006, F(1, 46) ⬍ 1.
quency, and neighborhood frequency but different neighborhood Procedure. The equipment used in this experiment was the same as
density, to ascertain whether the effects observed in Experiments that used in Experiment 3.
 PHONOLOGICAL NEIGHBORHOODS IN SPEECH PRODUCTION 743

Results The standard picture-naming task was modified in the following

manner. Instead of using a voice key triggered by a vocal response
The tape recorded responses of each participant were scored for to measure the response time, a buttonpress was used to collect
accuracy using the same criteria and type of analysis used in response times. Participants were instructed to view the picture
Experiment 3. A significant main effect of neighborhood density presented on the computer screen and retrieve the word used to
was found, F(1, 24) ⫽ 15.9, MSE ⫽ 7,662 such that words from label that picture. Participants were further instructed to press a
dense neighborhoods were responded to more quickly (795 ms, button on a response box as soon as they retrieved the correct word
SEM ⫽ 12) than were words from sparse neighborhoods (820 ms, and then say the name of the object out loud so that the accuracy
SEM ⫽ 12). There was no difference in error rates between the two of the response could be assessed. Cutler, Sebastian-Galles, Soler-
sets of words (both Fs ⬍ 1), suggesting that participants did not Vilageliu, and van Ooijen (2000) and McQueen (1998), for exam-
sacrifice speed for accuracy in making their responses. Words ple, have used similar modifications to standard word-recognition
from dense neighborhoods were correctly responded to 84.0% of and word-segmentation tasks to show that the effects they ob-
the time, and words from sparse neighborhoods were correctly served were due to differences in lexical access and not to differ-
responded to 85.8% of the time. Note that the responses in the ences in articulation.
present experiment were overall less accurate that those in Exper- We hypothesized that if the results of Experiments 3 and 4 were
iment 3. A number of factors—including different stimuli and due to differences in the ease of articulation between dense and
different participants—may account for the difference in error sparse words, then no difference in response times should be
rates between the two experiments. What is important, however, is observed in the present experiment because the speed with which
that there was no speed–accuracy trade off in the present results. articulation is initiated is not being measured. However, if the
number of simultaneously activated neighbors does influence the
Discussion speed with which lexical access occurs during speech production,
then words with dense neighborhoods should still be responded to
The results of the present experiment again showed that words more quickly than words with sparse neighborhoods.
from dense neighborhoods were produced more quickly than words
from sparse neighborhoods. The observation of a neighborhood- Method
density effect in the present picture-naming experiment is impor-
Participants. Twenty-five participants from the same population sam-
tant not only because it replicated the facilitative neighborhood
pled in Experiment 2 took part in this experiment.
density effect found in Experiment 3 with a different set of words Materials. The same stimuli used in Experiment 4 were used in the
and different participants but because the words in the present present experiment.
experiment were equivalent in phonotactic probability. In addition Procedure. The procedure and equipment used in Experiments 3 and 4
to having the same number of words in each condition with the were also used in the present experiment with the following exception.
same initial segments, the phonological segments that composed Instead of using the voice key to measure reaction time, participants
the words used in the present experiment were equivalent in their pressed a button on a response box with their dominant hand and then said
positional frequency and biphone frequency. By using words the name of the object aloud so that the response could be scored for
equivalent in these two measures of phonotactic probability, it was accuracy.
possible to rule out the possibility that the effects observed in the
previous experiments were due to differences in the phonological Results
segments that composed the words rather than the number of The author scored the tape recorded responses of each partici-
words simultaneously activated in memory. The results of the pant for accuracy using the same criteria used in Experiments 3
present experiment further suggest that multiple word forms be- and 4. In addition, responses were counted as incorrect if the
come simultaneously activated during speech production and that participant responded with the name of the object before pressing
these phonologically related word forms facilitate lexical access in the response button or failed to produce the name of the object
speech production. after pressing the button. Repeated measures ANOVAs showed a
significant main effect of neighborhood density, F(1, 24) ⫽ 7.9,
Experiment 5 MSE ⫽ 5,640. Participants responded more quickly to words from
dense (662 ms, SEM ⫽ 22) rather than sparse neighborhoods (683
The results of Experiment 4 clearly rule out the possibility that ms, SEM ⫽ 23). There was no difference in error rates between the
differences among the phonological segments composing the two sets of words (F ⬍ 1), suggesting that participants did not
words in each condition were the only source for the effects sacrifice speed for accuracy in making their responses. Words
observed in the previous experiments. The results of that experi- from dense neighborhoods were correctly responded to 89.2% of
ment do not, however, rule out the possibility that words with the time and words from sparse neighborhoods were correctly
dense neighborhoods were simply easier to articulate than words responded to 89.7% of the time. As in Experiment 4, the responses
with sparse neighborhoods. That is, the previously observed results in the present experiment were overall less accurate than those in
may not be due to differences in the time it takes to retrieve from Experiment 3 (but more accurate than those in Experiment 4). A
the lexicon words varying in neighborhood density. Rather, the number of factors—including the use of slightly different tasks and
differences may be due to differences in the ease with which the different participants—may account for the difference in error
musculature involved in producing words varying in neighborhood rates among the experiments. The more important thing to note
density can be moved. To evaluate this possibility a modified about the accuracy results is that there was no speed–accuracy
picture-naming task was used. trade off.
744 VITEVITCH

Discussion lexical retrieval process than the result of the articulatory processes
involved in speech production. Furthermore, the stimuli used in
The results from the modified picture-naming task used in the Experiments 4 and 5 that controlled phonotactic probability ruled
present experiment showed that words from dense neighborhoods out the possibility that the observed effects were due solely to
were responded to more quickly than were words from sparse processes at the level of phonological segments. Rather, the results
neighborhoods. This result replicated the neighborhood density of these and other experiments manipulating neighborhood density
effect observed in Experiments 3 and 4 with a different sample of in speech production (Vitevitch, 1997; Vitevitch & Sommers,
participants and a slightly different task. Replication of the 2001) suggest that phonologically related word forms facilitate the
neighborhood-density effect observed in Experiment 4 was espe- activation of target word forms by their interaction with shared
cially important because the stimuli used in that experiment as well constituent phonological segments. That is, the observed effects do
as the present experiment varied in neighborhood density but were not appear to be the result of processes among just word forms or
controlled for the initial phonological segments, familiarity, word among just phonological segments but are the result of word forms
frequency, neighborhood frequency, and phonotactic probability and phonological segments interacting with each another.
of the stimuli. Keeping phonotactic probability constant between Although the results and conclusions of this set of experiments
the two conditions was essential for ruling out the possibility that may appear incongruent with those of other studies (e.g., James &
the effects observed were due only to differences in the phono- Burke, 2000; Sevald & Dell, 1994; Yaniv, Meyer, Gordon, Huff,
logical segments that made up the words, a variable closely related & Sevald, 1990), they, in fact, are consistent within a broader view
to neighborhood density, rather than to neighborhood density of the speech-production system. For example, James and Burke
itself. (2000) found fewer TOT states when targets were primed with
The results of the present experiment also ruled out the possi- words that shared phonological segments with the target (e.g.,
bility that the effects observed in the previous experiments re- pellet primed the phonological segments /␧l/ in velcro). They
ported here were due to differences among the stimuli in the ease suggested that the phonologically related primes served to
of articulation. The present experiment used a buttonpress instead strengthen the connections between lexical items and their constit-
of a voice key to measure response time. Observing an effect of uent phonological segments, making it easier for the target word,
neighborhood density with this modified task further suggests that which shared many of these phonological segments, to be
differences in articulation between the two types of words were not retrieved.
the source of the differences in response times found in Experi- The findings of the present experiments, which examined si-
ments 3–5 (Cutler et al., 2000; McQueen, 1998). The results of the multaneous rather than sequential activation as in James and Burke
present experiment more strongly suggest that multiple word (2000), can easily be accommodated by the same model used to
forms that are phonologically related are simultaneously activated account for the findings of James and Burke, namely, node struc-
and facilitate the retrieval of words during speech production. ture theory (NST; MacKay, 1987). In NST there are bidirectional
connections between phonological segments and lexical represen-
General Discussion tations. (To be more precise, phonological segments connect to
syllable nodes in NST, however, for ease of explication I assume
The facilitative effects of simultaneously activated phonologi- some form of isomorphism between the syllable and lexical nodes
cally related word forms on lexical access in speech production for monosyllabic words like those used in the present experi-
that were demonstrated by means of both accuracy rates (see also ments.) The bidirectional connections between levels of represen-
Harley & Bown, 1998; Vitevitch, 1997; Vitevitch & Sommers, tation enable the model to account for the results of the present
2001) and response times (cf. Jescheniak & Levelt, 1994) in the experiments with the same interactive mechanism described ear-
present set of experiments are important for several reasons. First, lier. Thus, the same model that James and Burke used to account
manipulating the number of phonological neighbors for words for results of sequential activation in their experiments can also
presented in isolation provides a clearer picture of how phonolog- account for the results of simultaneous activation in the present
ically related words affect lexical access during speech production experiments.
than do experiments that rely on priming methods. Tasks that rely The results of Sevald and Dell (1994; cf. Sevald, Dell, & Cole,
on priming or the sequential presentation of words and that vary 1995) or Yaniv et al. (1990) are also consistent with the results of
the relationship between a presented item and a subsequently the present experiment when viewed from a broader context in the
to-be-produced item are potentially prone to task-specific strate- speech-production system. Specifically, Sevald and Dell (1994)
gies that may not reflect normal processing (Bowles & Poon, 1985; found competitive effects for the production of sequences of words
Roediger, Neely, & Blaxton, 1983). Although all experimental with the same initial sounds (e.g., cat, cab, can, cad vs. cat, bat,
tasks are prone to various task-specific strategies, including the mat, rat), suggesting that there was competition among phonemes
tasks in the present set of experiments, the consistent findings for placement in the representation of the word frame. Yaniv et al.
across different words, different task demands, different dependent (1990) also found inhibitory effects when the vowels in pairs of
variables, and different participants clearly speaks to the reliability CVC words were similar, suggesting that a lateral inhibitory mech-
of facilitative effects of phonologically related words in speech anism may modulate the motor programming of vowels during
production. speech production. Although the results of these studies propose
The consistent results of the present set of experiments also inhibition– competition between similar representations, rather
localized the source of these effects in the speech-production than facilitation as in the present set of experiments, these studies
process. The modified picture-naming task in Experiment 5 proposed such processes at different levels of representation (word
showed that the observed effects were more likely the result of the frame and motor programming levels) than that investigated in the
 PHONOLOGICAL NEIGHBORHOODS IN SPEECH PRODUCTION 745

present experiments. Vitevitch and Luce (1998, 1999; see also Pitt Brown, R., & McNeill, D. (1966). The “tip of the tongue” phenomenon.
& Samuel, 1995) found evidence for facilitation among phonolog- Journal of Verbal Learning and Verbal Behavior, 5, 325–337.
ical segments and competition among word forms in studies of Burke, D. M., MacKay, D. G., Worthley, J. S., & Wade, E. (1991). On the
spoken word recognition, suggesting that different processes may tip of the tongue: What causes word finding failures in young and older
operate at different levels of representation. The speech-production adults? Journal of Memory and Language, 30, 542–579.
Cohen, J. (1976). Random means random. Journal of Verbal Learning and
system may also operate with different processes at different levels
Verbal Behavior, 15, 261–262.
of representation. From this broader perspective, the present re-
Cohen, J. D., MacWhinney, B., Flatt, M., & Provost, J. (1993). PsyScope:
sults and those of Sevald and Dell (1994) and Yaniv et al. (1990)
A new graphic interactive environment for designing psychology exper-
are not at odds but serve to more precisely describe the speech iments. Behavioral Research Methods, Instruments, and Computers, 25,
production system. 257–271.
The results of the present experiments are also consistent with Costa, A., & Sebastian-Galles, N. (1998). Abstract phonological structure
interactive models of speech production. In discussing the work of in language production: Evidence from Spanish. Journal of Experimen-
James and Burke (2000) it was noted that the results of the present tal Psychology: Learning, Memory, and Cognition, 24, 886 –903.
experiments were consistent with the predictions of NST, an Cutler, A. (1982). The reliability of speech error data. In A. Cutler (Ed.),
interactive model of speech production. Furthermore, simulations Slips of the tongue and language production (pp. 7–28). Berlin: Walter
by Gordon and Dell (2001) have produced facilitative effects of de Gruyter/Mouton.
phonologically related neighbors in an interactive model of speech Cutler, A., Sebastian-Galles, N., Soler-Vilageliu, O., & van Ooijen, B.
production using normal processing parameters. (2000). Constraints of vowel and consonants on lexical selection: Cross-
The results of the present experiments are not, however, easily linguistic comparisons. Memory & Cognition, 28, 746 –755.
Cutting, J. C., & Bock, K. (1997). That’s the way the cookie bounces:
accounted for by strictly feedforward models (e.g., WEAVER⫹⫹;
Syntactic and semantic components of experimentally elicited idiom
Levelt et al., 1999). As discussed, it is unclear how multiple-word
blends. Memory & Cognition, 25, 57–71.
forms that are phonologically related can become simultaneously
Cycowicz, Y. M., Friedman, D., Rothstein, M., & Snodgrass, J. G. (1997).
activated in the current instantiation of WEAVER⫹⫹. In addition, Picture naming by young children: Norms for name agreement, famil-
the results of Experiment 4 and 5 ruled out the possibilities that the iarity, and visual complexity. Journal of Experimental Child Psychol-
observed effects were due to differences in the phonological seg- ogy, 65, 171–237.
ments or to articulatory processing. These results further constrain Dell, G. S. (1984). The representation of serial order in speech: Evidence
the modifications that could be made to WEAVER⫹⫹ to allow it from the repeated phoneme effect in speech errors. Journal of Experi-
to account for the data reported here. mental Psychology: Learning, Memory, and Cognition, 10, 222–233.
Finally, the results of the present experiments investigating Dell, G. S. (1986). A spreading-activation theory of retrieval in sentence
speech production demonstrated a facilitative effect of neighbor- production. Psychological Review, 93, 283–321.
hood density, which contrasts with the competitive effects of Dell, G. S. (1988). The retrieval of phonological forms in production: Tests
neighborhood density typically observed in investigations of of predictions from a connectionist model. Journal of Memory and
spoken-word recognition (e.g., Luce & Pisoni, 1998). This finding Language, 27, 124 –142.
Dell, G. S. (1990). Effects of frequency and vocabulary type on phono-
may further guide modeling efforts in speech production and
logical speech errors. Language and Cognitive Processes, 5, 313–349.
speech perception, especially those efforts that attempt to model
Dell, G. S., Reed, K. D., Adams, D. R., & Meyer, A. S. (2000). Speech
the interface between the two processes (e.g., NST; MacKay, errors, phonotactic constraints, and implicit learning: A study of expe-
1987). rience in language production. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 26, 1355–1367.
Fay, D., & Cutler, A. (1977). Malapropisms and the structure of the mental
References lexicon. Linguistic Inquiry, 8, 505–520.
Ferber, R. (1991). Slip of the tongue or slip of the ear? On the perception
Baars, B. J. (1992). A dozen competing-plans techniques for inducing
and transcription of naturalistic slips of the tongue. Journal of Psycho-
predictable slips in speech and action. In B. J. Baars (Ed.), Experimental
linguistic Research, 20, 105–122.
slips and human error: Exploring the architecture of volition (pp.
Goldinger, S. D, Luce, P. A., & Pisoni, D. B. (1989). Priming lexical
129 –150). New York: Plenum Press.
Baars, B. J., & Motley, M. T. (1974). Spoonerisms: Experimental elicita- neighbors of spoken words: Effects of competition and inhibition. Jour-
tion of human speech errors. Catalog of Selected Documents in Psychol- nal of Memory and Language, 28, 501–518.
ogy, 4, 118. Abstract obtained from Journal Supplement Abstract Gordon, J. K., & Dell, G. S. (2001). Phonological neighbourhood effects:
Service. Evidence from aphasia and connectionist modeling. Brain and Lan-
Baars, B. J., Motley, M. T., & MacKay, D. G. (1975). Output editing for guage, 79, 21–23.
lexical status in artificially elicited slips of the tongue. Journal of Verbal Greenberg, J. H., & Jenkins, J. J. (1964). Studies in the psychological
Learning and Verbal Behavior, 14, 382–391. correlates of the sound system of American English. Word, 20, 157–177.
Bock, J. K. (1986). Syntactic persistence in language production. Cognitive Harley, T. A., & Bown, H. E. (1998). What causes a tip-of-the-tongue
Psychology, 18, 355–387. state? Evidence for lexical neighbourhood effects in speech production.
Bock, J. K. (1996). Language production: Methods and methodologies. British Journal of Psychology, 89, 151–174.
Psychonomic Bulletin & Review, 3, 395– 421. Hino, Y., & Lupker, S. J. (2000). Effects of word frequency and spelling-
Bowles N. L., & Poon, L. W. (1985). Effects of priming in word retrieval. to-sound regularity in naming with and without preceding lexical deci-
Journal of Experimental Psychology: Learning, Memory, and Cogni- sion. Journal of Experimental Psychology: Human Perception and Per-
tion, 11, 272–283. formance, 26, 166 –183.
Brown, A. S. (1991). A review of the tip-of-the-tongue experience. Psy- James, L. E., & Burke, D. M. (2000). Phonological priming effects on word
chological Bulletin, 109, 204 –223. retrieval and tip-of-the-tongue experiences in young and older adults.
746 VITEVITCH

Journal of Experimental Psychology: Learning, Memory, and Cogni- Nusbaum, H. C., Pisoni, D. B., & Davis, C. K. (1984). Sizing up the
tion, 26, 1378 –1392. Hoosier mental lexicon: Measuring the familiarity of 20,000 words
Jescheniak, J. D., & Levelt, W. J. M. (1994). Word frequency effects in (Research on Speech Perception, Progress Report #10). Bloomington:
speech production: Retrieval of syntactic information and of phonolog- Indiana University, Speech Research Laboratory.
ical form. Journal of Experimental Psychology: Learning, Memory, and Oldfield, R. C., & Wingfield, A. (1965). Response latencies in naming
Cognition, 20, 824 – 843. objects. Quarterly Journal of Experimental Psychology, 17, 273–281.
Jescheniak, J. D., & Schriefers, H. (2001). Priming effects from phono- Pitt, M. A., & Samuel, A. G. (1995). Lexical and sublexical feedback in
logically related distractors in picture–word interference. Quarterly auditory word recognition. Cognitive Psychology, 29, 149 –188.
Journal of Experimental Psychology: Human Experimental Psychology, Raaijmakers, J. G. W., Schrijnemakers, J. M. C., & Gremmen, F. (1999).
54(A), 371–382. How to deal with “The language-as-fixed-effect fallacy”: Common
Jones, G. V. (1989). Back to Woodworth: Role of interlopers in the misconceptions and alternative solutions. Journal of Memory and Lan-
tip-of-the-tongue phenomenon. Memory & Cognition, 17, 69 –76. guage, 41, 416 – 426.
Jones, G. V., & Langford, S. (1987). Phonological blocking in the tip of the Roediger, H. L., Neely, J. H., & Blaxton, T. A. (1983). Inhibition from
tongue state. Cognition, 25, 115–122. related primes in semantic memory retrieval: A reappraisal of Brown’s
Jusczyk, P. W., Luce, P. A., & Charles-Luce, J. (1994). Infants’ sensitivity (1979) paradigm. Journal of Experimental Psychology: Learning, Mem-
to phonotactic patterns in the native language. Journal of Memory and ory, and Cognition, 9, 478 – 485.
Language, 33, 630 – 645. Roelofs, A. (1992). A spreading-activation theory of lemma retrieval in
Keppel, G. (1976). Words as random variables. Journal of Verbal Learning speaking. Cognition, 42, 107–142.
and Verbal Behavior, 15, 263–265. Saussure, F. d. (1966). Course in general linguistics (W. Baskin, Trans.).
Kučera, H., & Francis, W. N. (1967). Computational analysis of present- New York: McGraw Hill. (Original work published 1916)
day American English. Providence, RI: Brown University Press. Schacter, D. L. (1999). The seven sins of memory: Insights from psychol-
Landauer, T. K., & Streeter, L. A. (1973). Structural differences between ogy and cognitive neuroscience. American Psychologist, 54, 182–203.
common and rare words: Failure of equivalence and assumptions for Schriefers, H., Meyer, A. S., & Levelt, W. J. (1990). Exploring the time
theories of word recognition. Journal of Verbal Learning and Verbal course of lexical access in language production: Picture–word interfer-
Behavior, 12, 119 –131. ence studies. Journal of Memory and Language, 29, 86 –102.
Levelt, W. J. M., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical Sevald, C. A., & Dell, G. S. (1994). The sequential cuing effect in speech
access in speech production. Behavioral and Brain Sciences, 22, 1–38. production. Cognition, 53, 91–127.
Levelt, W. J. M., Schriefers, R., Vorberg, D., Meyer, A. S., Pechmann, T., Sevald, C. A., Dell, G. S., & Cole, J. S. (1995). Syllable structure in speech
& Havinga, J. (1991). The time course of lexical access in speech production: Are syllables chunks or schemas? Journal of Memory and
production: A study of picture naming. Psychological Review, 98, 122– Language, 34, 807– 820.
142. Shattuck-Hufnagel, S. (1979). Speech errors as evidence for a serial order
Levitt, A. G., & Healy, A. F. (1985). The roles of phoneme frequency, mechanism in sentence production. In W. E. Cooper & E. C. T. Walker
similarity, and availability in the experimental elicitation of speech (Eds.), Sentence processing: Psycholinguistic studies presented to Mer-
errors. Journal of Memory and Language, 24, 717–733. rill Garrett (pp. 295–342). Hillsdale, NJ: Erlbaum.
Luce, P. A. (1986). Neighborhoods of words in the mental lexicon. Doc- Shattuck-Hufnagel, S. (1992). The role of word structure in segmental
toral dissertation, Indiana University, Bloomington. serial ordering. Cognition, 42, 213–259.
Luce, P. A., & Pisoni, D. B. (1998). Recognizing spoken words: The Shattuck-Hufnagel, S., & Klatt, D. (1979). The limited use of distinctive
neighborhood activation model. Ear and Hearing, 19, 1–36. features and markedness in speech production: Evidence from speech
MacKay, D. G. (1987). The organization of perception and action: A error data. Journal of Verbal Learning and Verbal Behavior, 18, 41–55.
theory for language and other cognitive skills. New York: Springer- Sinsabaugh, B. A., & Fox, R. A. (1986). Reevaluating the SLIP paradigm:
Verlag. A research note. Communication Monographs, 53, 335–341.
Marslen-Wilson, W. D., & Zwitserlood, P. (1989). Accessing spoken Smith, J. E. K. (1976). The assuming-will-make-it-so fallacy. Journal of
words: The importance of word onsets. Journal of Experimental Psy- Verbal Learning and Verbal Behavior, 15, 262–263.
chology: Human Perception and Performance, 15, 576 –585. Snodgrass, J. G., & Vanderwart, M. (1980). A standardized set of 260
Maylor, E. A. (1990). Age, blocking and the tip of the tongue state. British pictures: Norms for name agreement, image agreement, familiarity, and
Journal of Psychology, 81, 123–134. visual complexity. Journal of Experimental Psychology: Human Learn-
McQueen, J. M. (1998). Segmentation of continuous speech using phono- ing and Memory, 6, 174 –215.
tactics. Journal of Memory and Language, 39, 21– 46. Stemberger, J. P. (1992). The reliability and replicability of naturalistic
Meyer, A. S. (1996). Lexical access in phrase and sentence production: speech error data: A comparison with experimentally induced errors. In
Results from picture–word interference experiments. Journal of Memory B. J. Baars (Ed.), Experimental slips and human error: Exploring the
and Language, 35, 477– 496. architecture of volition (pp. 195–215). New York: Plenum Press.
Meyer, A. S., & Bock, J. K. (1992). The tip-of-the-tongue phenomenon: Stemberger, J. P., & MacWhinney, B. (1986). Frequency and the lexical
Blocking or partial activation? Memory & Cognition, 20, 715–726. storage or regularly inflected forms. Memory & Cognition, 14, 17–26.
Motley, M. T. (1986). On replicating the SLIP technique: A reply to Stemberger, J. P., & Treiman, R. (1986). The internal structure of word-
Sinsabaugh and Fox. Communication Monographs, 53, 342–351. initial consonant clusters. Journal of Memory and Language, 25, 163–
Motley, M. T., & Baars, B. J. (1975). Encoding sensitivities to phonolog- 180.
ical markedness and transitional probability: Evidence from spooner- Storkel, H. L., & Rogers, M. A. (2000). The effect of probabilistic pho-
isms. Human Communication Research, 2, 351–361. notactics on lexical acquisition. Clinical Linguistics & Phonetics, 14,
Motley, M. T., & Baars, B. J. (1976). Laboratory induction of verbal slips: 407– 425.
A new method for psycholinguistic research. Communication Quar- Vitevitch, M. S. (1997). The neighborhood characteristics of malaprop-
terly, 24, 28 –34. isms. Language and Speech, 40, 211–228.
Norris, D., McQueen, J. M., & Cutler, A. (2000). Merging information in Vitevitch, M. S., & Luce, P. A. (1998). When words compete: Levels of
speech recognition: Feedback is never necessary. Brain and Behavioral processing in spoken word perception. Psychological Science, 9, 325–
Sciences, 23, 299 –370. 329.
 PHONOLOGICAL NEIGHBORHOODS IN SPEECH PRODUCTION 747

Vitevitch, M. S., & Luce, P. A. (1999). Probabilistic phonotactics and Wike, E. L., & Church, J. D. (1976). Comments on Clark’s “The language-
spoken word recognition. Journal of Memory and Language, 40, 374 – as-fixed-effect fallacy”. Journal of Verbal Learning and Verbal Behav-
408. ior, 15, 249 –255.
Vitevitch, M. S., Luce, P. A., Charles-Luce, J., & Kemmerer, D. (1997). Woodworth, R. S. (1929). Psychology (2nd Rev. ed.). New York: Holt.
Phonotactics and syllable stress: Implications for the processing of Yaniv, I., Meyer, D. E., Gordon, P. C., Huff, C. A., & Sevald, C. A. (1990).
spoken nonsense words. Language and Speech, 40, 47– 62. Vowel similarity, connectionist models, and syllable structure in motor
Vitevitch, M. S., Luce, P. A., Pisoni, D. B., & Auer, E. T. (1999). programming of speech. Journal of Memory and Language, 29, 1–26.
Phonotactics, neighborhood activation, and lexical access for spoken
words. Brain and Language, 68, 306 –311.
Vitevitch, M. S., & Sommers, M. S. (2001). The role of phonological Received March 28, 2001
neighbors in the tip-of-the-tongue state. Manuscript submitted for Revision received December 30, 2001
publication. Accepted January 21, 2002 䡲