 |
Branding Laws · Internet Marketing · eMarketing · Internet Advertising · Online Branding |
|
Branding Laws · Internet Marketing · eMarketing · Internet Advertising · Online Branding |
|
Masked Priming of Word and Picture Naming: The Role of Syllabic Units
LUDOVIC FERRAND AND JUAN SEGUI
CNRS and Universite´ Rene´ Descartes, Paris, France
JOURNAL OF MEMORY AND LANGUAGE 35, 708–723 (1996)
ARTICLE NO. 0037
Masked Priming of Word and Picture Naming: The Role of Syllabic Units
LUDOVIC FERRAND AND JUAN SEGUI
CNRS and Universite´ Rene´ Descartes, Paris, France
AND
JONATHAN GRAINGER
CNRS and Universite´ de Provence, Aix-en-Provence, France
A series of word, nonword, and picture naming experiments is reported using the masked
priming paradigm with very brief prime exposures. In Experiment 1 naming latencies for both
bi- and trisyllabic words were faster when preceded by primes that corresponded to the first
syllable than when preceded by primes that contained one letter more or less than the first
syllable. Experiment 2 showed syllable priming effects with bisyllabic nonword targets in the
naming task. Experiment 3 failed to observe such syllable priming effects with word and nonword
targets in the lexical decision task. Finally, Experiment 4 replicated the syllabic priming effects
using pictures as targets. These results suggest that the syllable represents a functional unit of
output phonology in French. q 1996 Academic Press, Inc.
Reading aloud printed words and naming (e.g., Evett & Humphreys, 1981; Forster &
common objects are two basic human activi- Davis, 1984) involves presenting very briefly
ties that involve a complex combination of a prime stimulus followed immediately by a
perceptual and motor skills. Cognitive psy- given target stimulus. This technique, referred
chologists strive to describe the processes and to as the masked prime paradigm (due to the
representations underlying such skilled behav- use of visual masking procedures) greatly reior
by breaking down the overall task into duces prime visibility and thus the likelihood
simpler subcomponents (e.g., perceptual pro- that subjects detect a relationship between
cessing, lexical retrieval, articulatory pro- prime and target stimuli. Since the detection
gramming). The major difficulty with such an of a relationship between the prime and target
endeavor is the development of adequate ex- during the course of an experiment might bias
perimental and theoretical tools. One very subjects’ responses in some way, reducing the
promising experimental technique developed likelihood that this will occur therefore rerecently
in studies of visual word recognition duces the possibility that observed priming effects
are due to such strategic biases. The present
experiments apply the masked prime tech-
The research reported in this article was partially sup- nique to the study of word and picture naming
ported by a post-doc grant (COGNISCIENCES) to Lu- in an attempt to determine the role of syllabledovic
Ferrand from the Centre National de la Recherche sized units in both of these cognitive activities. Scientifique (CNRS). We thank Manuel Carreiras, Ken
Forster, Linda Wheeldon, and three anonymous reviewers
for their comments on an earlier version. We also thank MASKED PRIMING IN WORD
Madeleine Leveille´ for programming expertise. A prelim- AND PICTURE NAMING
inary report of this study was presented at the Experimental
Psychology Society meeting, The University of Bir- Recently, Ferrand, Grainger, and Segui
mingham, UK, July 1995. Address correspondence and (1994) reported a series of word and picture
reprint requests to Ludovic Ferrand, Laboratoire de Psy- naming experiments using a masked priming chologie Expe´rimentale, CNRS and Universite´ Rene´ Descartes,
28 rue Serpente, 75006 Paris, France. E-mail: paradigm with prime exposures brief enough
Ferrandl@idf.ext.jussieu.fr. to prevent identification. Ferrand et al. (1994)
0749-596X/96 $18.00 708
Copyright q 1996 by Academic Press, Inc.
All rights of reproduction in any form reserved.
SYLLABLE PRIMING 709
found chance-level performance when sub- sensitive to orthographic prime–target overjects
were asked to judge whether the masked lap, whereas primed picture naming is priprime
word was nominally the same as the marily sensitive to phonological overlap. In
target, suggesting that very little precise infor- order to capture these results, Ferrand et al.
mation about the prime was available for con- (1994) presented a simple model of picture
scious identification. This absence of aware- and word naming. This model introduces a
ness was taken as clear evidence for the auto- distinction between an orthographic and a
maticity of the processes under study. Even phonological lexicon and between sublexical
in such impoverished prime presentation con- orthographic units and sublexical phonologiditions,
there is evidence that high-level (e.g., cal units (see also Ferrand & Grainger, 1992,
lexical) representations are activated during 1993, 1994; Grainger & Ferrand, 1994,
prime processing and can subsequently affect 1996). Sublexical units activate in parallel
word and picture naming. both orthographic and phonological word
Using this masked priming technique, Fer- units, whereas picture representations will
rand et al. (1994) examined the type of codes send activation only directly to the phonothat
are generated under masking conditions logical lexicon. The final articulatory output
when prime stimuli are words or pronounce- is assumed to depend both on the activity of
able strings of letters. In particular, we demon- whole-word units and on sublexical orthostrated
that the prior presentation of the same graphic units. Due to the hypothetical timeword
prime facilitated both picture and word course of information flow in this model (denaming
independently of target frequency. termined by its architecture), when reading
Furthermore, similar effects were obtained us- printed words aloud, the articulatory output
ing primes that were pseudohomophones of will receive activation first from sublexical
the picture and word targets (see also Ferrand, orthographic units. With picture targets, on
1995, for an extension of these results to digit the other hand, the articulatory output will
naming). Given that these results were ob- receive activation first from whole-word
served in conditions that prevent conscious phonological representations. Thus, speeded
identification (briefly presented forward and word naming will be dominated by sublexibackward
masked primes), they are consistent cal orthography, whereas picture naming
with recent evidence showing that phonologi- will be mainly influenced by activity in the
cal information about a written word becomes phonological lexicon. The dissociation in
available rapidly and automatically (see Be- priming effects obtained with picture and
rent & Perfetti, 1995; Ferrand & Grainger, word targets can be given the following in1992,
1993, 1994; Grainger & Ferrand, 1994, terpretation. A briefly presented pseudoho-
1996; Lukatela & Turvey, 1994; Perfetti & mophone prime can activate the correspondBell,
1991). ing phonological word unit in memory, thus
Furthermore, Ferrand et al. (1994) found giving rise to facilitation in picture naming,
a marked dissociation in the priming effects but will result in an even stronger activation
obtained with picture and word targets. Pic- of the corresponding sublexical orthographic
ture naming was strongly and reliably facili- codes since these are closer to the input.
tated by the prior presentation of a masked Since according to our model, in word nampseudohomophone
prime, whereas a masked ing the articulatory output receives informaorthographically
related nonhomophonic tion from these sublexical orthographic
prime failed to facilitate picture naming. On codes before it receives information from
the other hand, word naming was facilitated lexical codes, it is the orthographic overlap
by the prior presentation of both a pseudoho- between prime and target that will mainly
mophone prime and an orthographically re- influence the time to name written word tarlated
prime. These results suggest that the gets. It should be pointed out, however, that
masked priming of word naming is primarily the stimuli used by Ferrand et al. (1994) were
710 FERRAND, SEGUI, AND GRAINGER
only monosyllabic words compared to the tial role of syllables in naming by directly
bi- and trisyllabic words to be used in the comparing monosyllabic and polysyllabic
present experiments. words have yielded inconsistent results. For
The present research examines more pre- example, Klapp, Anderson, and Berrian
cisely the nature of the representations sub- (1973) reported that response latencies in a
serving the observed facilitation effects. In word-naming task were longer for two-syllaparticular,
this research examines the possi- ble words that for one-syllable words. Howble
role of sublexical phonological units, ever, other studies have reported no effect of
namely syllables, as functional units in word number of syllables on initiation of word namand
picture naming, an area that has received ing (e.g., Forster & Chambers, 1973; Frederiklittle
attention as yet. This point is important sen & Kroll, 1976).
given that in our previous pseudohomophone On the other hand, two recent studies
experiments we used only monosyllabic (Meyer & Schriefers, 1991; Tousman & Initems
making it impossible to distinguish be- hoff, 1992) have reported that the naming of
tween lexical and syllabic effects. The pres- a target word or of a target picture is facilitated
ent study was designed to find out if the pho- by the prior presentation of clearly visible or
nological codes that are generated from a audible primes that corresponded to the first
masked prime are structured syllabically at syllable of the targets. For example, Tousman
some level in the information processing hi- and Inhoff (1992) reported evidence for the
erarchy. We also want to know at what lev- syllable’s role in bisyllabic word naming.
el(s) of processing syllabic codes are in- They showed that the naming latencies of the
volved in word and picture naming (if they word GLUCOSE, for example, were faster
are involved at all). when preceded by the prior presentation (for
250 ms) of a prime that corresponded to the
THE SYLLABLE’S ROLE IN NAMING first syllable (e.g., GLU) compared to the prior
Over the past decade the syllable has come presentation of a neutral prime (e.g., —). In
to enjoy a privileged status as a critical unit Meyer and Schriefers’ (1991) picture–word
in phonology both from a linguistic (e.g., interference study, subjects named bisyllabic
Clements & Keyser, 1983; Fudge, 1969; picture names while hearing distractor words
Kaye, 1989; Selkirk, 1982) and a psycholin- that shared the first or the second syllable with
guistic (e.g., Cutler, Mehler, Norris, & Segui, the picture names or were unrelated to the
1986, 1992; Levelt & Wheeldon, 1994; picture names. Both types of related dis-
Mehler, 1981; Mehler, Dommergues, Frau- tractors facilitated picture naming in comparienfelder,
& Segui, 1981; Meyer, 1990; Segui, son with unrelated distractors and under cer-
1984; Segui, Dupoux, & Mehler, 1990) per- tain timing conditions. Distractors sharing the
spective. However, studies examining syllabic first syllable facilitated naming responses
effects in word naming in English have when presented 150 ms before, at, or 150 ms
yielded rather inconsistent results. For in- after picture onset, whereas distractors sharing
stance, Jared and Seidenberg (1990) hypothe- the second syllable facilitated the responses
sized that if words are broken down into sylla- only when presented at or 150 ms after picture
bles, then emphasizing the syllable should re- onset. These results suggest that polysyllabic
sult in faster reaction times than when the words are encoded syllable-by-syllable from
word is presented as a whole. They observed beginning to end.
that presenting the first syllable (for 250 ms) One problem in interpreting the results of
followed by the second resulted in equivalent the above research using long onset asynchronaming
times when compared to the whole nies is the possible use of predictive strategies
word condition and concluded that the syllable by subjects on noticing the relationship beis
not a functional unit in word naming. Sev- tween certain prime–target pairs. Since in the
eral other studies that have explored the poten- present study we are primarily interested in
SYLLABLE PRIMING 711
the highly automatized processes involved in base their responses on the generation of a
normal word or picture naming, it is essential syllabified phonological representation.
to avoid any contamination by strategic fac- In the following experiments, we used the
tors. One means of removing this possibility is masked priming technique combined with the
to reduce prime visibility drastically through naming task (see Ferrand et al., 1994) and
shorter presentation durations and massive applied Mehler et al.’s (1981) design to study
forward and backward masking. To this end the possible role of the syllable in word and
we use the masked priming technique whose picture naming. In order to test the validity of
utility has been clearly demonstrated in prior the syllable as a unit of naming, we exploited
research (see Ferrand et al., 1994). the fact that, in French, words with the same
initial phoneme sequence may have different
THE PRESENT STUDY initial syllables. Indeed, although the words
In a seminal study of speech perception, BAL.CON and BA.LADE share the first three
Mehler, Dommergues, Frauenfelder, and phonemes /b/, /a/, and /l/, their syllabic strucSegui
(1981) demonstrated that French sub- ture differs, such that BAL is exactly the first
jects responded faster in a monitoring task syllable of BAL.CON, but more than the first
when the target corresponded to the first sylla- syllable of BA.LADE, whereas BA is exactly
ble of the stimulus word than when it corre- the first syllable of BA.LADE, but less than
sponded to a longer or a shorter segment than the first syllable of BAL.CON. Note that
the first syllable. Each target corresponded to French has a regular syllable structure and
a complete syllable in only one of the two clear syllable boundaries (for a specific referwords:
PA was exactly the first syllable of the ence to French phonology, see Kaye & Lowword
PA.LACE,1 but less than the first sylla- estamm, 1984), with minimal consonantal amble
of the word PAL.MIER whereas PAL was bisyllabicity, especially between the first and
exactly the first syllable of PAL.MIER, but second syllables of polysyllabic words. For
more than the first syllable of PA.LACE. Thus example, for the French word BALADE there
the target PA was detected faster in the word is a clear syllable boundary between BA- and -
PA.LACE than in the word PAL.MIER, LADE. If we hypothesize that syllables play
whereas the target PAL was detected faster in a central role over and above phoneme units in
PAL.MIER than in PA.LACE. These results naming, then we should observe a facilitation
therefore provide evidence for the syllable as when primes and targets share phonemes that
a fundamental unit of speech perception in constitute a syllable unit as opposed to when
French, with a prelexical level of stored sylla- primes and targets share phonemes that do not
bles mediating between the speech input and constitute a syllable unit. Using the same type
the mental lexicon. More recently, Wheeldon of stimuli, Mehler et al. (1981) found evidence
and Levelt (1995) have used a production vari- for the syllable’s role as a unit of speech perant
of the syllable monitoring task of Mehler ception.
et al. (1981) in which subjects were instructed Thus, the purpose of the present research
to silently generate the translation of a given was (1) to find out if the phonological codes
stimulus word and to monitor their production that are generated from a masked prime are
for a given syllable target. They replicated the structured syllabically, and (2) to identify the
standard syllabic effect in that responses to level(s) of processing (perceptual or articulatargets
were faster when they corresponded to tory) at which syllabic codes are involved in
the initial syllable of the generated word than naming (if they are involved at all). One promwhen
they did not. According to the authors, ising avenue to constrain a future unified
these results provide evidence that subjects model of orthographic and phonological processing
is a multilevel, multitask approach dis1
Throughout the article we denote syllable structure cussed in detail by Jacobs and Grainger (1994;
using a period. see also Grainger & Jacobs, in press; Jacobs,
712 FERRAND, SEGUI, AND GRAINGER
1994). This involves using several tasks that nemes (CVC) were selected such that these
are thought to cover different areas of the phonemes made up the first syllable for one
global word recognition and naming spectrum member of the pair and the first two (CV)
and to share a certain number of common pro- phonemes formed the first syllable of the seccesses.
Thus, for example, it is clear that word ond member of the pair. For instance, in the
and picture naming share an articulatory com- pair BA.LADE/BAL.CON the first three phoponent,
whereas lexical decision and word nemes (/b/ /a/ /l/) are identical, but the CVC
naming must share some early perceptual sequence corresponds to the first syllable of
components. On the basis of commonalities the word BAL.CON and the CV sequence corand
differences in the patterns of effects ob- responds to the first syllable of the word BA.-
served in the different tasks we can therefore LADE. The average frequency was 96 occurlimit
the range of possible underlying mecha- rences per million for the CVC word targets
nisms. (ranging from a minimum of 1 to a maximum
In the following experiments, the subject’s of 982) and 77 occurrences per million for the
task was to name words, nonwords or pictures CV word targets (ranging from a minimum of
as quickly as possible or to rapidly classify 1 to a maximum of 768; Tre´sor de la langue
letter strings as words or nonwords. The ex- franc¸aise, 1971). For each of the 18 pairs the
perimental conditions differed in terms of the initial consonant was either a voiced or a
type of relationship between the prime and voiceless stop and the second was a liquid
the target. In Experiment 1, bi- and trisyllabic (either /l/ or /r/). For each target word, two
target words were preceded by visual se- types of primes were selected: (1) primes
quences that corresponded (e.g., bal%%%– that correspond to the first syllable (e.g.,
BAL.CON, ba%%%%–BA.LADE) or did bal%%%–BAL.CON and ba%%%%–BA.-
not correspond (e.g., ba%%%%–BAL.CON, LADE), or (2) primes that do not correspond
bal%%%–BA.LADE) to the first syllable. Ex- to the first syllable (e.g., ba%%%%–BAL.-
periment 2 tests the syllable’s role using non- CON and bal%%%–BA.LADE). In Experiwords
as targets. Experiment 3 tests the sylla- ment 1B, 16 pairs of monomorphemic trisylble’s
role in the lexical decision task using labic French words (eight letters long) were
words and nonwords as targets. Finally, in Ex- selected by applying the same constraints as
periment 4, target pictures were preceded by those used for the bisyllabic words in Experiprimes
that corresponded or did not corre- ment 1A. For example, in the pair PA.RA.-
spond to the written transcript of the first sylla- SITE/PAR.TI.SAN the first three phonemes (/
ble of the picture’s name. p/ /a/ /r/) are identical but the CVC sequence
corresponds to the first syllable of the word
EXPERIMENT 1: BISYLLABIC AND PAR.TI.SAN, whereas the CV sequence is the
TRISYLLABIC WORDS first syllable of the word PA.RA.SITE. The
prime–target pairings were constructed in the
Method
same way as those in Experiment 1A. The
Subjects. Forty psychology students at Rene´ average frequency was 6 occurrences per mil-
Descartes University, Paris, France, served as lion for the CVC word targets (ranging from
subjects for course credit, 20 in Experiment a minimum of 0 to a maximum of 16) and 5
1A (bisyllabic words) and 20 in Experiment occurrences per million for the CV word tar1B
(trisyllabic words). All were native speak- gets (ranging from a minimum of 0 to a maxiers
of French, with normal or corrected-to- mum of 20; Tre´sor de la langue franc¸aise,
normal vision. 1971). For both Experiments 1A and 1B, each
Stimuli and design. In Experiment 1A, 18 target word was presented twice to a given
pairs of monomorphemic bisyllabic French subject, paired with one of the two alternative
words (six letters long) of similar printed fre- primes in the first half of the experiment and
quency sharing the same initial three pho- with the other possible prime in the second
SYLLABLE PRIMING 713
half. The two categories of word targets (CVC room as the subject in order to check and note
and CV words) represent the factor Type of the responses of the subject. The next se-
Target, whereas the two categories of prime quence followed after a 2-s delay. Stimulus
stimuli (CVC%%% and CV%%%% primes) presentation was randomized, with a different
represent the factor Type of Prime. The two order for each subject.
factors were crossed in a 2 1 2 factorial de-
sign. Two groups of subjects were used to Results
counterbalance the presentation order of a Mean naming latencies and percentage of
given word target and its two corresponding errors are given in Table 1 for both the bisylprimes.
Thus, two orders of presentation were labic and trisyllabic words. Latencies longer
used and each subject received all the priming than 1000 ms were excluded (less than 1.5
conditions and saw the target words twice in and 2% of the data for bisyllabic and trisyleither
of the two orders. labic words, respectively). Separate ANOVAs
Procedure. Stimuli were presented in isola- were run for Experiments 1A and 1B with
tion on the center of the display screen of a Type of Target (CV or CVC words) and Type
personal computer with a 70-Hz refresh rate. of Prime (CV or CVC primes) entered as main
The items appeared on the screen as white factors. F values are reported by subjects (F1)
characters on a dark background. Each charac- and by items (F2).
ter (in uppercase) covered approximately Experiment 1A: Bisyllabic words. The main
0.387 of visual angle from a viewing distance effect of Type of Target was not significant
of 60 cm, so the target words subtended about (F1(1,19) 3.16 and F2 › 1) and neither
2.287 of visual angle. The masked priming was the main effect of Type of Prime (both
procedure with the naming task used in the Fs › 1). On the other hand, the interaction
experiments of Ferrand et al. (1994) was between these two factors was significant
adopted here. Each trial consisted of the fol- (F1(1,19) 42.11, p › .001, and F2(1,17)
lowing sequence of four stimuli presented on 58.36, p › .001). The interaction reflects
the same screen location. Primes were always the fact that naming latencies were faster when
completely covered by the mask. First a for- CVC word targets were preceded by CVC
ward mask consisting of a row of six (or eight) primes than when they were preceded by CV
hashmarks (######) was presented for 500 primes and when CV word targets were prems.
This was immediately followed by presen- ceded by CV primes than when they were
tation of the prime for 29 ms, followed imme- preceded by CVC primes. An analysis of the
diately by a backward mask (######) for 14 error data showed no main or interaction efms,
which, finally, was immediately followed fects (all Fs › 1).
by presentation of the target word both (the Experiment 1B: Trisyllabic words. The
prime and the target) presented in the same main effects of Type of Target and Type of
screen location as the masks. The target re- Prime were not significant in either analysis
mained on the screen until the subjects re- (all Fs › 1). On the other hand, the interaction
sponded. Primes were always presented in between target and prime type was significant
lowercase and targets in uppercase. Subjects (F1(1,19) 19.66, p › .001, and F2(1,15)
were asked to fixate the middle of the forward 8.33, p › .02). As can be seen in Table 1,
mask. They were also instructed to name as this interaction illustrates the same pattern of
rapidly and as accurately as possible the target effects as obtained with the bisyllabic words.
word. The existence of a prime was not men- An analysis of the error data showed no main
tioned. The computer recorded the naming or interaction effects (all Fs › 1).
times, measured from target onset to the trig-
gering of the voice key by the subject’s re- Discussion
sponse (via a Sennheiser MD211N micro- The main purpose of Experiment 1 was to
phone). The experimenter sat in the same establish whether briefly presented strings of
714 FERRAND, SEGUI, AND GRAINGER
TABLE 1
MEAN NAMING LATENCIES (ms) AND PERCENTAGE OF ERRORS
IN EXPERIMENT 1 (BISYLLABIC AND TRISYLLABIC WORDS)
CV word targets CVC word targets
Bisyllabic word targets (e.g., ba.lade) (e.g., bal.con) Mean
CV primes (e.g., ba%%%%) 567 (0.4) 589 (0.7) 578
CVC primes (e.g., bal% % %) 599 (0.6) 558 (0.4) 578.5
Mean 583 573.5
CV word targets CVC word targets
Trisyllabic word targets (e.g., pa.ra.site) (e.g., par.ti.san) Mean
CV primes (e.g., pa%%%%%%) 608 (0.8) 626 (1.2) 617
CVC primes (e.g., par%%%%%) 632 (1.0) 604 (0.9) 618
Mean 620 615
letters activate syllable-sized units involved in faster word naming latencies as long as word
the process of word naming. It was observed naming is at least partly dependent on lexical
that naming latencies were faster when the retrieval. According to the access code hyprime
stimuli formed the first syllable of the pothesis, one should not observe syllabic
target word than when primes were one letter priming effects in naming tasks using nonmore
or less than the first syllable. These re- word stimuli. According to the output units
sults, obtained both for bisyllabic and for tri- hypothesis, on the other hand, syllabic effects
syllabic words, indeed suggest that syllable- should also be observed in the speeded namsized
units are rapidly generated from a ing of nonword stimuli. This is tested in Exprinted
string of letters. One possible locus of periment 2.
this syllabic facilitation effect is at the level
of phonological output units that store motor EXPERIMENT 2: NONWORD NAMING
control programs allowing the physical real-
Method
ization of speech sounds. The existence of syllable-
sized output units (cf. Levelt & Wheel- Subjects. Twenty psychology students at
don, 1994) within the theoretical framework Rene´ Descartes University, Paris, France,
for word and picture naming sketched by Fer- served as subjects for course credit. All were
rand et al. (1994) could explain the syllabic native speakers of French, with normal or corpriming
effects observed in Experiment 1. rected-to-normal vision and had not particiSublexical
orthographic units that are acti- pated in the previous experiment.
vated upon prime presentation will send acti- Stimuli and design. The design was the
vation directly to the syllabic output units thus same as that in Experiment 1A except for the
facilitating the pronunciation of any target use of nonword rather than word targets. Eighstimulus
that shares the same syllable units. teen pairs of bisyllabic nonwords (six letters
An alternative explanation of the results long) sharing the first three phonemes (CVC)
from Experiment 1 would be that initial sylla- were created such that these phonemes made
bles are access codes for visual word recogni- up the first syllable for one member of the
tion (e.g., Spoehr & Smith, 1973; Taft, 1979, pair and the first two (CV) phonemes formed
1987) and that presentation of a prime stimu- the first syllable of the second member of the
lus that corresponds to a word’s initial syllable pair. The nonwords were generated from the
facilitates the retrieval of the corresponding French words in Experiment 1A by replacing
lexical representation. This would lead to (after the first three phonemes) a consonant
SYLLABLE PRIMING 715
TABLE 2
MEAN NAMING LATENCIES (ms) AND PERCENTAGE OF ERRORS IN EXPERIMENT 2 (BISYLLABIC NONWORDS)
CV nonword targets CVC nonword targets
Bisyllabic nonword targets (e.g., ba.lode) (e.g., bal.don) Mean
CV primes (e.g., ba%%%%) 612 (2.0) 656 (2.2) 634
CVC primes (e.g., bal%%%) 646 (2.4) 616 (1.9) 631
Mean 629 636
with another consonant, or a vowel with an- interaction between target and prime type was
other vowel, in such a way that only legal significant (F1(1,19) 19.87, p › .001, and
syllables in French were used. The prime– F2(1,17) 6.61, p › .02). As in Experiment
target pairings were constructed in the same 1, the interaction reflects the faster naming
way as with the word stimuli in Experiment latencies obtained when CVC nonword targets
1A. For example the nonword target BA.- are preceded by CVC primes than when they
LODE was primed with ba%%% and are preceded by CV primes and when CV nonbal%%%%,
and the same was true for the word targets are preceded by CV primes than
target BAL.DON. As in Experiment 1, two when they are preceded by CVC primes. An
groups of subjects were used to counterbal- analysis of the error data showed no main or
ance the presentation order of a given non- interaction effects (all Fs › 1).
word target and its two corresponding primes.
Thus, two orders of presentation were used Discussion
and each subject received all the priming con- The fact that we observe a reliable syllable
ditions and saw the target words twice in one priming effect with nonword targets in Experiof
the possible orders. ment 2 suggests that the syllabic effect obProcedure.
Exactly the same procedure as tained with word stimuli in Experiment 1 is
that in Experiment 1, was used, except that not due to the prime stimuli facilitating access
subjects were instructed to read aloud non- to the target word’s lexical representation.
word stimuli. This would result if access codes for visual
word recognition were syllabically structured,
Results and if speeded word naming responses are at
Mean naming latencies and percentage of least partly determined by lexical retrieval.
errors2 are given in Table 2. Latencies longer The fact that nonword targets showed syllabic
than 1000 ms were excluded (less than 3% of effects that were very similar to those obtained
the data). An overall ANOVA was performed with word targets suggests that these effects
on the reaction time data with Type of Target arise in processes other than those required to
(CV and CVC nonwords) and Type of Prime isolate a specific lexical representation corre(
CV and CVC primes) entered as main factors. sponding to the target. One likely locus for
The main effect of Type of Target was not these syllabic effects would be the units that
significant (F(1,19) 1.67 and F2(1,17) are involved in computing an articulatory out- 1.95); neither was the main effect of Type of put in the naming task. As already noted in
Prime (both Fs › 1). On the other hand, the the introduction, there is accumulating evidence
that such codes for output phonology
are indeed structured syllabically (e.g., Lev- 2 All the nonword stimuli had straightforward, unam- elt & Wheeldon, 1994). biguous pronunciations following standard French spell-
ing-to-sound translation rules. Any alternative pronuncia- In order to provide a further test of the hytion
was considered an error. pothesis that the locus of the syllable priming
716 FERRAND, SEGUI, AND GRAINGER
effect is in output rather than access processes, Results
Experiment 3 uses a lexical decision task, Mean naming latencies and percentage erwhich
does not require that the target’s pho- rors are given in Table 3 for both word and
nology be output. If this hypothesis is correct, nonword targets. Separate ANOVAs were run
one should not observe the syllable priming for word and nonword targets with Type of
effect for either word or nonword targets in Target (CV or CVC targets) and Type of
such a task. On the other hand, if the syllable Prime (CV or CVC primes) entered as main
priming effect is affecting lexical access pro- factors.
cesses, it should be observable in the lexical In an analysis of variance conducted on the
decision task. correct lexical decision latencies and percent-
EXPERIMENT 3: LEXICAL DECISION age of errors to word targets, no main or interaction
effects were significant in either the by-
Method subjects or the by-items analyses (all Fs › Subjects. Twenty psychology students at 1). An analysis of variance on the nonword
Rene´ Descartes University, Paris, France, reaction times and error data also showed no
served as subjects. All were native speakers main or interaction effects (all Fs › 1).
of French, with normal or corrected-to-normal
vision, and had not participated in the previous Discussion
experiments. The same stimuli that produced a syllabic
Stimuli and design. The word targets were priming effect in the naming task in Experiexactly
the same as those used in Experiment ments 1 and 2 did not produce a similar effect
1A. Eighteen new nonword targets (different in the lexical decision task in Experiment 3.
from those in Experiment 2)3 were constructed This null result provides further support for
for the purposes of the lexical decision task. the hypothesis that the syllable priming effect
These nonwords were created by applying the lies in output rather than access processes.
same constraints as used for the nonwords in However, in order to provide a more convinc-
Experiment 2. For each nonword target, two ing positive result in favor of the output photypes
of primes were selected: (1) primes nology hypothesis, Experiment 4 uses pictures
that correspond to the first syllable (e.g., of common objects as targets in a naming task.
bar%%%–BAR.DOL and ba%%%%–BA.R- Since picture naming presumably involves the
ELE); and (2) primes that do not correspond to same output phonology as word naming we
the first syllable (e.g., ba%%%%–BAR.DOL should observe syllabic effects in this experiand
bar%%%–BA.RELE). ment.
Procedure. Exactly the same procedure as
that in Experiment 1 was used except that the EXPERIMENT 4: PICTURE NAMING
task was a lexical decision task. Subjects were Method
instructed to decide as rapidly and as accu-
rately as possible whether the letter string in Subjects. Eighteen psychology students at
uppercase that remained on the screen was a Rene´ Descartes University, Paris, France,
French word. Subjects responded ‘‘yes’’ by served as subjects. All were native speakers
pressing one of the two response buttons with of French, with normal or corrected-to-normal
the forefinger of the preferred hand and ‘‘no’’ vision, and had not participated in the previous
by pressing the other response button with the experiments.
forefinger of the nonpreferred hand. Stimuli and design. The design was identical
to that used in Experiment 1 except that
word targets were replaced by line-drawings 3 It was not possible to use the nonwords from Experi- of common objects. Ten pairs of monomor- ment 2 together with the words from Experiment 1A simply
because these nonwords were generated from the phemic bi- and trisyllabic French words correwords
in Experiment 1A. sponding to simple black-on-white drawings
SYLLABLE PRIMING 717
TABLE 3
MEAN LEXICAL DECISION LATENCIES (ms) AND PERCENTAGE OF ERRORS
IN EXPERIMENT 3 (BISYLLABIC WORD AND NONWORD TARGETS)
CV word targets CVC word targets
Bisyllabic word targets (e.g., ba.lade) (e.g., bal.con) Mean
CV primes (e.g., ba%%%%) 587 (5.5) 588 (4.9) 587.5
CVC primes (e.g. bal%%%) 591 (6.1) 590 (5.3) 590.5
Mean 589 589
CV nonword targets CVC nonword targets
Bisyllabic nonword targets (e.g., ba.rele) (e.g., bar.dol) Mean
CV primes (e.g., ba%%%%) 715 (8.1) 711 (7.5) 713
CVC primes (e.g., bar%%%) 719 (1.7) 718 (8.0) 718.5
Mean 717 714.5
of common objects served as target pictures Experiment 1 was used except that word tarwith
20 additional object fillers. The average gets were replaced by picture targets that subfrequency
was 31 occurrences per million for jects had to name as quickly and as accurately
the CVC picture names (ranging from a mini- as possible. Before starting the experiment
mum of 1 to a maximum of 130) and 35 occur- proper, the subjects were given a booklet inrences
per million for the CV picture names cluding all drawings (following Ferrand et al.,
(ranging from a minimum of 1 to a maximum 1994). Next to each object was printed the
of 288; Tre´sor de la langue franc¸aise, 1971). word. The subjects were asked to examine all
The pictures were selected from a French book the drawings, to study their names, and to use
of pictures (L’imagier du Pe`re Castor, 1991) only those names to refer to the pictures.
digitized (using a HP-Scan Jet IIcx), and ed-
ited with PaintBrush. At a viewing distance Results
of 60 cm, the mean angular size of the pictures Mean naming latencies and percentage of
was 2.57 horizontally and vertically. As in Ex- errors are given in Table 4. Latencies longer
periment 1, the pairs of target picture names than 1500 ms were excluded (less than 3% of
shared the same initial three phonemes (CVC) the data). An ANOVA was performed on the
and were selected such that these phonemes reaction time data with Type of Target (CV
made up the first syllable for one member of and CVC picture names) and Type of Prime
the pair and the first two (CV) phonemes (CV and CVC primes) entered as main factors.
formed the first syllable of the second member The main effect of Type of Target was not
of the pair. Thus, in the pair CA.ROTTE/ significant (F1 › 1 and F2 › 1), neither was
CAR.TABLE (meaning carrot and schoolbag the main effect of Type of Prime (both Fs ›
in English) the initial CVC sequence corre- 1). On the other hand, the interaction between
sponds to the first syllable of the word CAR.- target and prime type was significant
TABLE, whereas the CV sequence is the first (F1(1,17) 9.96, p › .01 and F2(1,9) syllable of the word CA.ROTTE. Each picture 23.02, p › .001). An analysis of the error data
target was primed by both a CVC and a CV showed no main or interaction effects (all Fs
prime with the order of presentation of these › 1). As in Experiments 1 and 2, the interacpriming
conditions being counterbalanced tion reflects the faster naming latencies to
across subjects, as in the previous experi- CVC picture targets when preceded by CVC
ments. primes compared to CV primes, and the faster
Procedure. The same procedure as that in RTs to CV picture targets when they are pre-
718 FERRAND, SEGUI, AND GRAINGER
TABLE 4
MEAN NAMING LATENCIES (ms) AND PERCENTAGE OF ERRORS IN EXPERIMENT 4 (PICTURES)
CV picture targets CVC picture targets
Picture targets (e.g., ca.rotte) (e.g., car.table) Mean
CV primes (e.g., ca%%%%%) 715 (11.5) 750 (13.0) 732.5
CVC primes (e.g., car%%%%) 755 (12.0) 706 (12.0) 730.5
Mean 735 728
ceded by CV primes compared to CVC Smith, 1973; Taft, 1979, 1987) fails to account
primes. for the absence of such effects in the lexical
decision task and cannot explain the presence
Discussion of these effects in nonword naming. In all
When a target picture is preceded by a cases in which a syllabic effect was observed,
prime that corresponds to the written tran- subjects had to produce an articulatory output
script of the picture name’s first syllable, nam- thus implying that it is the units involved in
ing latencies were faster when compared to generating such an articulatory output that are
the case in which the prime does not corre- structured syllabically.
spond to the first syllable of the picture’s Within the theoretical framework for word
name. This observation of syllabic priming and picture naming presented by Ferrand et
with picture targets provides strong evidence al. (1994), this rapid activation of output phothat
briefly presented letter string primes rap- nology arises via the direct connection beidly
activate phonological output units that are tween orthographic input units (i.e., letter or
structured syllabically. The fact that very simi- letter cluster codes) and output phonology.
lar syllabic priming effects are observed in When a prime stimulus corresponds to the inipicture,
word, and nonword naming but not tial syllable of the following target, then the
in a lexical decision task strongly suggests that appropriate initial syllable unit will be actithey
are indeed located at the level of output vated for articulatory output, thus allowing a
phonology (articulatory encoding) where pre- faster generation of the target’s articulatory
sumably the same processes are involved in output compared to when the prime does not
naming these different types of targets. correspond to the target’s initial syllable. The
fact that we did not observe syllabic priming
GENERAL DISCUSSION effects in the lexical decision task does not
The results of the present experiments imply, of course, that this task is insensitive
clearly demonstrate that briefly presented to phonological priming. With monosyllabic
masked primes corresponding to the initial word targets, Ferrand and Grainger (1992,
syllable of word, nonword, or picture targets 1993, 1994) have consistently observed phofacilitate
the naming of these targets in nological priming effects (primes were pseu-
French. On the other hand, these same prime dohomophones of the target) in the lexical destimuli
did not affect performance to word and cision task. However, while orthographic
nonword targets in a lexical decision task. All priming effects appeared with prime expothese
results converge to suggest that the syl- sures similar to those used in the present exlabic
priming effects observed in such extreme periments, phonological priming effects representation
conditions is located at the level quired longer prime exposures (see Ferrand &
of output phonology. An alternative explana- Grainger, 1993). Therefore, it is possible that
tion in terms of syllabically structured access syllabic priming effects would appear in the
codes in visual word recognition (Spoehr & lexical decision task with longer prime expo-
SYLLABLE PRIMING 719
sure durations, if such sublexical input pho- been proposed in models such as those of
nology is structured syllabically. The point, Crompton (1981), Dell (1988), and Levelt
however, in demonstrating that there is no syl- (1989, 1992). According to Crompton
labic effect in the lexical decision task when (1981, p. 663), ‘‘. . . the construction of
using exactly the same presentation conditions articulatory programs involves the accessing
that allowed the effects to emerge in word of a library of articulatory routines. These
naming, is to rule out an access code account routines correspond to syllables, while their
of the word naming results. addresses in the library are expressed in a
Indeed, the recent work of Carreiras, Alv- form analogous to classical phonemic reprearez,
and De Vega (1993) does suggest that sentation. The instructions employed in acsyllable-
sized units play a role in visual cessing the library involve the transfer of
word recognition. These authors found that phonological information from the lexicon.
Spanish words containing syllables that oc- This transfer takes place syllable-constitcur
frequently in Spanish were harder to rec- uent by syllable-constituent.’’ This model,
ognize than words containing low frequency therefore, clearly predicts faster reaction
syllables. Since this inhibitory syllable fre- times when one of the syllables of a target
quency effect was obtained in both the lexi- word (or nonword) is preactivated by the
cal decision and word naming tasks, it is prime. Levelt’s (1989) production model
likely to reflect lexical access processes that also predicts a priming effect for syllable
are sensitive to syllable-sized units. Ac- units. According to Levelt (1989, 1992,
cording to Carreiras et al., all words that 1993, 1994), each speaker possesses a syllahave
one syllable in common with the target bary (i.e., a library of syllable-sized motor
word will be partially activated upon target programs) that stores motor programs correpresentation
and subsequently interfere in sponding to all the syllables of the language.
the recognition process. Within the theoreti- Phonological representations consist of phocal
framework described by Ferrand et al. nemes, but these phonemes will be used to
(1994), the syllable-sized units responsible address the syllabary, and producing a word
for the inhibitory syllable frequency effect involves retrieving/assembling one or more
would be located at the level of sublexical of these articulatory syllables. Access to the
input phonology. Upon presentation of a syllabary is governed by the same principles
written word, these syllabic representations that govern access to the lexicon, that is, a
would receive activation from sublexical or- syllable would be all the more easily acthographic
units and send on activation to cessed the more frequently it is used.
whole-word representations, thus influenc- A count of the number of different syllaing
the process of visual word recognition. bles in French yielded 6000 from a sample
If sublexical input phonology is syllabically of over 200,000 word types (Cerf, Danon,
organized, then one is led to predict that Derouault, El Beze, & Merialdo, 1989). This
syllabic priming effects should appear in the number is very close to the number of syllalexical
decision and perceptual identifica- bles in English, about 6600 (Levelt, 1989).
tion tasks when using longer prime expo- Of course, not all possible syllables are
sures than in the present experiments. This stored in this dictionary of syllables. If this
is clearly an important point for further in- syllabary contained not only actual but also
vestigation. any potential syllable in a language such as
French, we would end up with about 33,600
Relation to Speech Production possible syllables (count based on the comThe
syllable priming effects reported in puterized French database BDLEX: Cerf et
the present study support the hypothesis that al., 1989; Dupoux, 1993). This number of
syllable units are functional units in the pro- all possible syllables in the French language
duction of French. Such units have already seems excessive in comparison with the size
720 FERRAND, SEGUI, AND GRAINGER
of the mental lexicon. In contrast, the num- a syllabified prosodic frame. Our results exber
of actual syllables is relatively small. tend this finding showing that the syllabic
However, speakers must be able to produce effect is also observed in a standard word
new but well-formed kinds of syllables: and picture naming task as a function of the
Levelt (1989) suggests that they can produce initial syllable match between a masked letthese
new syllables by analogy. This is in ter string prime and the to be named target.
fact presumably the way they acquired their
syllable repertoire to start with. Another ar- Relation to Spoken and Visual Word
gument in favor of a mental syllabary is that Recognition
most syllables are highly overused units of Research in languages with clear syllable
articulation: it would be wasteful to fully boundaries such as French, Spanish, Cataprogram
them time and again (Levelt, 1993). lan, or Portuguese shows positive evidence
Levelt and Wheeldon (1994) tested this for the listener’s sensitivity to the syllable
syllabary hypothesis on the basis of its pre- in speech perception, whereas the results for
diction of a syllable frequency effect. Sub- English are less clear (see Segui et al., 1990,
jects had to produce bisyllabic words in re- and Cutler, 1993, for reviews). These results
sponse to abstract visual patterns they had suggest that the syllable is probably not a
learned to associate with these words. The universal unit of processing. Instead, Cutler
authors used four types of words: low-fre- (1993) and Otake, Hatano, Cutler, and
quency words with low-frequency syllables, Mehler (1993) suggest that language rhythm
low-frequency words with high-frequency determines the segmentation units most natsyllables,
high-frequency words with low- ural to native listeners. Thus, French, Spanfrequency
syllables, and high-frequency ish, Portuguese, and Catalan have syllabic
words with high-frequency syllables. They rhythm, and French, Spanish, Portuguese,
observed both a word and a syllable fre- and Catalan listeners use the syllable in segquency
effect, the two effects being inde- mentation, while English has stress rhythm,
pendent, suggesting that the mental lexicon and segmentation by English listeners is
and the mental syllabary are indeed indepen- based on stress. In the case of Japanese
dent stores. which has a mora rhythm, Japanese listeners
More directly related to the present study use the mora in segmentation.
is the recent work of Wheeldon and Levelt The results obtained in the present experi-
(1995). As noted in the introduction, these ments mirror those obtained in speech perauthors
replicated the syllable monitoring ception (at least in French), suggesting a
effect first observed in speech perception by similarity between units involved in speech
Mehler et al. (1981). In their Experiment 2, production and speech perception in French.
Wheeldon and Levelt (1995) used a transla- Similarly, the mora (rather than the syllable)
tion task where Dutch subjects (who had a plays an important role in the recognition
good knowledge of English) had to generate (Otake et al., 1993) and production (Kubothe
Dutch translation of an auditorily pre- zono, 1989) of Japanese. It would therefore
sented English word and to monitor their appear that spoken word recognition and
production for a prespecified target string. production involve the same basic units.
The results of this experiment show that sub- Since research on speech perception has
jects were much faster in monitoring for a shown that ambisyllabicity undermines the
target string when it matched the first sylla- role of syllabic units in English, it will be
ble of the produced word than when it did interesting to examine whether this is also
not, regardless of the length of the target the case with speech production. We are curstring.
Wheeldon and Levelt concluded that rently examining this in a series of experisubjects
based their responses on the output ments conducted in English and in Spanish.
from the process that assigns phonemes to One other interesting question for future
SYLLABLE PRIMING 721
research concerns the extent to which the Ongoing research is currently examining
phonological codes generated from a printed these important questions.
word are similar to those involved in under-
standing a spoken word. Since syllabic units REFERENCES
appear to be involved in the recognition of BERENT, I., & PERFETTI, C. A. (1995). A Rose is a REEZ:
spoken French (e.g., Mehler et al., 1981), The two-cycles model of phonology assembly in reading English. Psychological Review, 102, 146–
then if visual word recognition involves the 184.
same sublexical phonology as spoken word CARREIRAS, M., ALVAREZ, C. J., & DE VEGA, M. (1993).
recognition, one should also observe such Syllable frequency and visual word recognition in
syllabic effects in the recognition of printed Spanish. Journal of Memory and Language, 32, 766–
780. French words. Some recent experiments CERF, H., DANON, T., DEROUAULT, A. M., EL BEZE, M., &
suggest that this is indeed the case. Cole´ MERIALDO, B. (1989). Speech recognition in French
(1993) ran a visual analog of the Mehler et with a very large dictionary. Proceedings of Eurosal.
study where subjects had to say whether a peech 89. Paris: La Vilette, France.
given stimulus word (e.g., PA.LACE) began CLEMENTS, G. N.,&KEYSER, S. J. (1983). CV phonology:
A generative theory of the syllable. Cambridge, MA:
with a target letter sequence (e.g., PAL) pre- MIT Press.
sented visually just before. She found a sig- COLE´ , P. (1993). The use of phonology in reading words
nificant syllable monitoring effect in good by 3rd grade readers. Poster presented at the Internabeginning
readers and also in adult readers. tional Conference on The Psychology of Language
and Communication, Glasgow, Scotland, September. Nevertheless, one could argue that subjects CROMPTON, A. (1981). Syllables and segments in speech
use some form of articulatory code to per- production. Linguistics, 17, 663–716.
form this syllable detection task and that CUTLER, A. (1993). Language-specific processing: Does
such results do not necessarily imply that the evidence converge? In G. T. M. Altman & R.
sublexical input phonology is organized syl- Shillcock (Eds.), Cognitive models of speech processing:
The second Sperlonga meeting. Hillsdale,
labically. Such a conclusion is supported by NJ: Erlbaum.
the present demonstration of an extremely CUTLER, A., MEHLER, J., NORRIS, D., & SEGUI, J. (1986).
rapid generation of syllabically structured The syllable’s differing role in the segmentation of
codes for output phonology in the masked French and English. Journal of Memory and Language,
25, 385–400. priming paradigm. In these extreme condi- CUTLER, A., MEHLER, J., NORRIS, D., & SEGUI, J. (1992).
tions, no syllabic priming was observed in The monolingual nature of speech segmentation by
the lexical decision task, thus reinforcing bilinguals. Cognitive Psychology, 24, 381–410.
our conclusions that results obtained with DELL, G. S. (1988). The retrieval of phonological forms in
production: Tests of predictions from a connectionist the naming task are indeed located in the model. Journal of Memory and Language, 27, 124–
coding required to generate an articulatory 142.
output. Thus, as noted above, it will be im- DUPOUX, E. (1993). The time course of prelexical proportant
to examine in future research cessing: The syllabic hypothesis revisited. In
whether syllabic priming effects are observ- G. T. M. Altmann & R. Shillcock (Eds.), Cognitive
models of speech processing: The second Sperlonga
able in visual word recognition tasks such meeting. Hillsdale, NJ: Erlbaum.
as lexical decision and perceptual identifi- EVETT, L. J., & HUMPHREYS, G. W. (1981). The use of
cation with appropriate prime exposures. If graphemic information in lexical access. Quarterly
sublexical input phonology is structured Journal of Experimental Psychology, 33, 325–350.
FERRAND, L. (1995). Repeated prime-target presentations syllabically then one should observe such do not eliminate repetition and phonological priming
effects. Furthermore, with respect to the hy- in naming digits. Acta Psychologica, 89, 217–227.
pothesis that visual and spoken word recog- FERRAND, L., & GRAINGER, J. (1992). Phonology and
nition share the same sublexical phonologi- orthography in visual word recognition: Evidence
from masked nonword priming. Quarterly Journal cal codes, syllabic priming effects should of Experimental Psychology, 45A, 353–372.
also occur in a cross-modal (masked visual FERRAND, L., & GRAINGER, J. (1993). The time course of
prime–auditory target) priming situation. orthographic and phonological code activation in the
722 FERRAND, SEGUI, AND GRAINGER
early phases of visual word recognition. Bulletin of LEVELT, W. J. M. (1989). Speaking: From intention to
the Psychonomic Society, 31, 119–122. articulation. Cambridge, MA: MIT Press.
FERRAND, L., & GRAINGER, J. (1994). Effects of orthogra- LEVELT, W. J. M. (1992). Accessing words in speech
phy are independent of phonology in masked form production: Stages, processes and representations.
priming. Quarterly Journal of Experimental Psychol- Cognition, 42, 1–22.
ogy, 47A, 365–382. LEVELT, W. J. M. (1993). Timing in speech production
FERRAND, L., GRAINGER, J., & SEGUI, J. (1994). A study with special reference to word form encoding. Annals
of masked form priming in picture and word naming. of the New York Academy of Science, 682, 283–295.
Memory & Cognition, 22, 431–441. LEVELT, W. J. M. (1994). The skill of speaking. In P.
FORSTER, K. I., & CHAMBERS, S. (1973). Lexical access Bertelson, P. Eelen, & G. d’Ydewalle, International
and naming time. Journal of Verbal Learning and perspectives on psychological science, Volume 1:
Verbal Behavior, 12, 627–635. Leading themes. Hillsdale, NJ: Erlbaum.
FORSTER, K. I., & DAVIS, C. (1984). Repetition priming LEVELT, W. J. M.,& WHEELDON, L. (1994). Do speakers
and frequency attenuation in lexical access. Journal have access to a mental syllabary? Cognition, 50,
of Experimental Psychology: Learning, Memory, and 239–269.
Cognition, 10, 680–698. LUKATELA, G., & TURVEY, M. T. (1994). Visual lexical
FREDERIKSEN, J. R., & KROLL, J. F. (1976). Spelling and access is initially phonological: 2. Evidence from
sound: Approaches to the internal lexicon. Journal phonological priming by homophones and pseudohoof
Experimental Psychology: Human Perception and mophones. Journal of Experimental Psychology:
Performance, 2, 361–379. General, 123, 331–353.
FUDGE, E. C. (1969). Syllables. Journal of Linguistics, 5, MEHLER, J. (1981). The role of syllables in speech pro-
253–286. cessing: Infant and adult data. Philosophical Trans-
GRAINGER, J., & FERRAND, L. (1994). Phonology and actions of the Royal Society, B295, 333–352.
orthography in visual word recognition: Effects of MEHLER, J., DOMMERGUES, J., FRAUENFELDER, U., &
masked homophone primes. Journal of Memory and SEGUI, J. (1981). The syllable’s role in speech seg-
Language, 33, 218–233. mentation. Journal of Verbal Learning and Verbal
GRAINGER, J., & FERRAND, L. (1996). Masked ortho- Behavior, 20, 298–305.
graphic and phonological priming in visual word rec- MEYER, A. S. (1990). The time course of phonological
ognition and naming: Cross-task comparisons. Jour- encoding in language production: The encoding of
nal of Memory and Language, 35, 623–647. successive syllables of a word. Journal of Memory
GRAINGER, J., & JACOBS, A. M. Orthographic processing and Language, 29, 524–545.
in visual word recognition: A multiple read-out MEYER, A. S., & SCHRIEFERS, H. (1991). Phonological
model. Psychological Review, in press. facilitation in picture-word interference experiments:
JACOBS, A. M. (1994). On computational theories and Effects of stimulus onset asynchrony and types of
multilevel, multitask models of cognition: The case interfering stimuli. Journal of Experimental Psycholof
word recognition. Behavioral and Brain Sciences, ogy: Learning, Memory, and Cognition, 17, 1146–
17, 670–672. 1160.
JACOBS, A. M., & GRAINGER, J. (1994). Models of visual OTAKE, T., HATANO, G., CUTLER, A., & MEHLER, J.
word recognition-Sampling the state of the art. Jour- (1993). Mora or syllable? Speech segmentation in
nal of Experimental Psychology: Human Perception Japanese. Journal of Memory and Language, 32,
and Performance, 20, 1311–1344. 358–378.
JARED, D., & SEIDENBERG, M. S. (1990). Naming multi- PERFETTI, C. A., & BELL, L. C. (1991). Phonemic activasyllabic
words. Journal of Experimental Psychology: tion during the first 40 ms of word identification:
Human Perception and Performance, 16, 92–105. Evidence from backward masking and masked prim-
KAYE, J. (1989). Phonology: A cognitive view. Hillsdale, ing. Journal of Memory and Language, 30, 473–
NJ: Erlbaum. 485.
KAYE, J. D., & LOWESTAMM, J. (1984). De la syllabicite´ SEGUI, J. (1984). The syllable: A basic perceptual unit in
(On syllabacity). In F. Dell, D. Hirst, & J. R. Verg- speech perception? In H. Bouma & D. G. Bouwhuis
naud (Eds.), Forme sonore du langage. Paris: Her- (Eds.), Attention and Performance X: Control of lanmann.
guage processes. Hillsdale, NJ: Erlbaum.
KLAPP, S. T., ANDERSON, W. G., & BERRIAN, R. W. SEGUI, J., DUPOUX, E., & MEHLER, J. (1990). The role of
(1973). Implicit speech in reading reconsidered. the syllable in speech segmentation, phoneme identi-
Journal of Experimental Psychology, 100, 368–374. fication, and lexical access. In G. T. M. Altmann
KUBOZONO, H. (1989). The mora and syllable structure in (Ed.), Cognitive models of speech processing: Psy-
Japanese: Evidence from speech errors. Language & cholinguistic and computational perspectives. CamSpeech,
32, 249–278. bridge, MA: MIT Press.
L’Imagier du Pe`re Castor (1991). Paris: Flammarion. SELKIRK, E. O. (1982). The syllable. In H. van der Hulst &
SYLLABLE PRIMING 723
N. Smith (Eds.), The structure of phonological repre- formance XII: The psychology of reading. Hillsdale,
sentation (Part 2). Dordrecht, The Netherlands: Foris. NJ: Erlbaum.
SPOEHR, K. T., & SMITH, E. E. (1973). The role of sylla- TOUSMAN, S., & INHOFF, A. (1992). Phonology in multibles
in perceptual processing. Cognitive Psychology, syllabic word recognition. Journal of Psycholinguis5,
71–89. tic Research, 21, 525–544.
TAFT,M. (1979). Lexical access via an orthographic code: Tre´sor de la langue franc¸aise (1971). Nancy: CNRS.
WHEELDON, L. R., & LEVELT, W. J. M. (1995). Monitor-
The basic orthographic syllabic structure (BOSS).
ing the time course of phonological encoding. Jour-
Journal of Verbal Learning and Verbal Behavior, nal of Memory and Language, 34, 311–334.
18, 21–39.
TAFT, M. (1987). Morphographic processing: The BOSS (Received August 8, 1994)
re-emerges. In M. Coltheart (Ed.), Attention and Per- (Revision received April 13, 1995)