Lexical stress and phonetic processing in word learning in20‐ to 24‐month‐old English‐learning children
ABSTRACT To investigate the interaction between segmental and supra-segmental stress-related information in early word learning, two experiments were conducted with 20- to 24-month-old English-learning children. In an adaptation of the object categorization study designed by Nazzi and Gopnik (2001), children were presented with pairs of novel objects whose labels differed by their initial consonant (Experiment 1) or their medial consonant (Experiment 2). Words were produced with a stress initial (trochaic) or a stress final (iambic) pattern. In both experiments successful word learning was established when the to-be-remembered contrast was embedded in a stressed syllable, but not when embedded in unstressed syllables. This was independent of the overall word pattern, trochaic or iambic, or the location of the phonemic contrast, word-initial or -medial. Results are discussed in light of the use of phonetic information in early lexical acquisition, highlighting the role of lexical stress and ambisyllabicity in early word processing.
Lexical stress and phonetic processing in word learning in
20- to 24-month-old English-learning children
Caroline Floccia,1Thierry Nazzi,2,3Keith Austin,4Fre ´de ´rique Arreckx5
and Jeremy Goslin1
1. School of Psychology, University of Plymouth, UK
2. Universit? Paris Descartes, France
3. Laboratoire Psychologie de la Perception, CNRS, Paris, France
4. School of Psychological Science, University of Manchester, UK
5. School of Early Years and Primary Education Studies, University of Plymouth, UK
To investigate the interaction between segmental and supra-segmental stress-related information in early word learning, two
experiments were conducted with 20- to 24-month-old English-learning children. In an adaptation of the object categorization
study designed by Nazzi and Gopnik (2001), children were presented with pairs of novel objects whose labels differed by their
initial consonant (Experiment 1) or their medial consonant (Experiment 2). Words were produced with a stress initial
(trochaic) or a stress final (iambic) pattern. In both experiments successful word learning was established when the to-
be-remembered contrast was embedded in a stressed syllable, but not when embedded in unstressed syllables. This was
independent of the overall word pattern, trochaic or iambic, or the location of the phonemic contrast, word-initial or -medial.
Results are discussed in light of the use of phonetic information in early lexical acquisition, highlighting the role of lexical stress
and ambisyllabicity in early word processing.
Decades of research have established young infants’
superiority over adults when discriminating phonetic
contrasts (Eilers, Gavin & Wilson, 1979; Lasky, Syrdal-
Lasky & Klein, 1975; Streeter, 1976; Trehub, 1976), with
infants exhibiting the ability to discriminate various
phonetic contrasts from their first year of life (e.g.
Eimas, Siqueland, Juscyzk & Vigorito, 1971; Trehub,
1976). However, this advantage does not transfer to the
early perceptual processing of words, especially when
infants are required to process both form and meaning,
with 14-month-old infants failing to learn minimal pairs
of non-words differing by a single consonant (Stager &
Werker, 1997; but see Rost & McMurray, 2009; Yoshida,
Fennell, Swingley & Werker, 2009).
This sharp contrast between phonemic discrimination
and the processing of words has generated many studies,
as it can offer important insight into the specificity of
lexical representations in early word learning or recog-
nition. The emerging picture reveals the immaturity of
the phonetic processing of early lexical representations,
highlighted in tasks involving the detection of consonant
or vowel mispronunciation for familiar or newly learnt
words (Bailey & Plunkett, 2002; Fennell & Werker, 2003;
Hall? & de Boysson-Bardies, 1996; Mani & Plunkett,
2007, 2008; Nazzi, 2005; Nazzi, Floccia, Moquet &
Butler, 2009; Swingley, 2003; Swingley & Aslin, 2000;
Vihman, Nakai, DePaolis & Hall?, 2004). Two sets of
explanations have so far been offered to account for this
limited level of phonetic processing of early lexical rep-
resentations. Proponents of the discontinuity hypothesis
have argued for the existence of holistic representations
at the onset of the second year of life (e.g. Charles-Luce
& Luce, 1990; Hall? & de Boysson-Bardies, 1996; Met-
sala & Walley, 1998), followed by a reorganization of the
lexicon as vocabulary increases in the second year of life
(Walley, 1993). According to the continuity hypothesis,
detailed phonetic lexical representations are maintained
throughout infancy and childhood, but access to these
representations is impinged on at around 1 year of age
because of the computational load of linking sounds to
meaning (Stager & Werker, 1997; see Saffran & Estes,
2006, for a review).
However, most research has focused on aspects of
phonetic specificity and lexical representations, while
very little attention has been paid to interactions between
supra-segmental and segmental information (but see
Address for correspondence: Caroline Floccia, School of Psychology, University of Plymouth, Drake Circus, Plymouth PL4 8AA, UK; e-mail:
? 2010 Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
Developmental Science 14:3 (2011), pp 602–613DOI: 10.1111/j.1467-7687.2010.01006.x
Vihman et al., 2004; Johnson, 2005). Prosody plays a
fundamental role in early language development, guiding
early acquisitions such as maternal voice recognition (see
Floccia, Nazzi & Bertoncini, 2000, for a review) and
language discrimination abilities (Nazzi, Jusczyk &
Johnson, 2000), but also providing some cues for word
segmentation in continuous speech (Jusczyk, Houston &
Newsome, 1999; Nazzi, Iakimova, Bertoncini, Fr?donie
& Alcantara, 2006) or for the acquisition of syntactic
parameters (e.g. Christophe, Nespor, Guasti & Van
Ooijen, 2003). Therefore it is perhaps unsurprising that
during the first year of life there is abundant evidence for
a prosodic bias in early speech processing, before the
integration of segmental information (see Hçhle, Bijel-
jac-Babic, Herold, Weissenborn & Nazzi, 2009, for a
Given its importance in infants’ perception of lan-
guage, one might ask how prosodic information interacts
with the construction and processing of stable phonetic
representations at the onset of lexical development. In
this article we aim to study this interaction by examining
how the location of stress in disyllabic words influences
the level of syllable-onset consonant detail in young
English-learning children. More specifically, we focus on
the interactions between stress position and phonetic
contrast in lexical representations by examining the
interaction between the stress pattern (trochaic versus
iambic) and position of the phonetic contrast (onset of
the initial syllable versus onset of the second syllable) in
disyllabicstimuli. Using these
20–24-month-old children in a novel word learning task.
This age range was selected to examine the interaction
between supra-segmental and phonetic information in
word learning without having to confront temporary
inabilities to access or to build phonetic representations.
In addition, around that age children are actively
engaged in lexical learning but are still expanding their
lexicon, as shown by average scores on the Oxford
Communicative Development Inventory approaching
70% of the 416 words in comprehension and 30% in
production (Hamilton, Plunkett & Schafer, 2000).
In English the acoustic realization of stress and its
underpinning phonological system provide an interesting
opportunity to look into the interactions between seg-
mental and supra-segmental representations in the
emerging lexicon, for it places the young English learner
in a situation of poor ‘acoustic and phonetic transpar-
ency’, whose level is defined as ‘the ease with which a
segment can be identified as distinct from another
competing candidate in a language’ (Sebasti?n-Gall?s,
Dupoux, Segui & Mehler, 1992, p. 19). According to
these authors, the level of transparency would be ham-
pered by different factors such as variable and contras-
tive stress, ambisyllabicity and vowel reduction, all
aspects that are present in English (Anderson & Jones,
1974; Brown, 1977; Gimson, 1989; Ladefoged, 1993;
Treiman & Danis, 1988; Vihman, DePaolis & Davis,
1998) and therefore contribute to a low degree of
acoustic-phonetic transparency in unstressed syllables.
Rules of stress assignment in English lead to disyllabic
words being mainly stressed on the first syllable, resulting
in trochaic patterns such as in the word EFfort (Cutler &
Carter, 1987). Iambic patterns, where stressed syllables
are preceded by an unstressed syllable such as in efFECT,
are much rarer in English.
Thus far only a few studies have specifically addressed
the interactions between stress assignment and the pho-
netic specificity of lexical representations. Vihman et al.
(2004) followed up on studies by Hall? and Boysson-
French-speaking infants were able to identify familiar
words that had their initial consonants changed, but not
their medial consonants changed. Hall? and Boysson-
Bardies suggested that this was because initial unac-
cented syllables are not fully specified at 11 months of
age, and Vihman et al. (2004) argued that this could be
due to the importance of stressed syllables in word rep-
resentations, as in French the stress (or rather fixed
accent) nearly always falls on the last syllable of words
(Di Cristo, 1998). Therefore, as the stress pattern in
English is largely trochaic, Vihman et al. (2004) predicted
that (at least in trochaic words) the recognition of
familiar words would be impaired if the initial consonant
was changed, but not if the change was made on a medial
consonant. This prediction was indeed validated, but the
study also revealed that recognition was completely
blocked when the initial consonant was changed in
English (e.g. vunny instead of bunny), whereas the medial
consonant change in the French study (balane instead of
banane) did not have such a marked effect. Vihman et al.
(2004) argued that this was an indication of the crucial
role that word-initial phonemes play in word recognition,
and that segmental information is better represented in
stressed than unstressed syllables1(also see Curtin, 2010,
for a recent study showing 14-month-olds’ abilities to
encode phonetic information in stressed syllables).
However, Johnson (2005) alleged that infants would
require access to the phonetic detail of unstressed sylla-
bles to extract iambic word forms from the speech
developed by Jusczyk and Aslin (1995), she familiarized
10.5-month-old infants with made-up iambic words (e.g.
giNOME). The infants were then tested using passages
containing exact representations of the familiarized
words, some novel words, or the familiarized words with
a different initial consonant (e.g. piNOME). While the
infants were found to listen longer to the familiarized
words than to the novel words, there was no such effect
1In further discussion we will refer only to the processing and
representations of content words, and deliberately ignore function
words. Shi, Werker and Cutler (2006) have shown that at 13 months,
English-learning infants possess a detailed phonetic representation of
unstressed functors, which suggests that the phonetic specificity of
words also depends on the frequency of the encountered items, but for
clarity purposes we will not discuss this issue here.
Lexical stress and phonetic processing603
? 2010 Blackwell Publishing Ltd.
when the initial consonant of the familiarized words was
changed. As the infants’ behaviour was modulated by the
change in the initial consonant, Johnson (2005) con-
cluded that the phonetic representation of unstressed
syllables is sufficiently detailed to encode a word-initial
Although there would appear to be a direct contrast
between the findings of Johnson (2005) and Vihman
et al. (2004), differences in the design of the two studies
make it difficult to draw any concrete conclusions on the
interaction between supra-segmental and segmental
representations. While Vihman et al. presented trochaic
forms, Johnson used iambic forms, and also Vihman
et al. used a word recognition task whereas Johnson used
a novel word forms segmentation task. We must also take
Johnson’s use of word-initial consonant contrasts into
consideration, as Vihman et al. suggest that this position
plays a key role in word processing. Therefore it is always
possible that the disparity between their findings could
be accounted for by the differences in experimental fac-
tors, tapping onto different levels of processing and⁄or
In this study we aim to contribute to this literature by
examining the link between phonetic and supra-seg-
mental processing within the context of new word
learning. Specifically, we wish to examine whether the
detection of phonetic contrasts is modulated by three
potential factors: the stress of the syllable in which they
are embedded, whether they are word-initial or -medial,
and whether they are found in the (more frequent) tro-
chaic or in the iambic stress pattern.
The lack of clarity regarding the first of these factors
was discussed above. For the second factor, the relative
importance of word onsets over offsets in adult word
processing has been emphasized in a number of studies
(e.g. Content, Kearns & Frauenfelder, 2001; Marslen-
Wilson, 1987). However, it is questionable whether chil-
dren exhibit this sensitivity at an early age, as part of an
in-built property of speech processing mechanisms.
Recent studies suggest otherwise, as equal sensitivity to
syllable onset or coda was reported in a mispronuncia-
tion detection task with English-learning 14–22-month-
olds (Swingley, 2009), and in a word learning task with
French-learning 20-month-olds (Nazzi & Bertoncini,
2009). These results suggest that 20- to 24-month-olds
could be equally sensitive to phonetic contrasts located
in either word-initial or -medial position.
Finally, regarding the third factor of interest to this
study, it must be noted that in English the privileged
status of the trochaic over the iambic stress pattern is
well established, with strong syllable onsets providing
preferential cues for word segmentation (Cutler & But-
terfield, 1992; Cutler & Norris, 1988; Echols, Crowhurst
& Childers, 1997), an efficient strategy given that most of
the English lexicon is made up of trochaic words.
Moreover, there is evidence for the emergence of an early
trochaic bias between 6 and 9 months in young English-
learning infants (Jusczyk, Cutler & Redanz, 1993; Jus-
czyk et al., 1999), and for an advantage of trochaic over
iambic forms in word segmentation tasks between 7.5
and 13.5 months (Jusczyk et al., 1999; Nazzi, Dilley,
Jusczyk, Stattuck-Hufnagel & Jusczyk, 2005). In addi-
tion, stress patterns also impact on children’s produc-
tions during their second year of life, as seen by a
tendency to omit initial unstressed syllables, saying for
example NAna instead of baNAna (Allen & Hawkins,
1980; Echols & Newport, 1992; Vihman et al., 1998). In
sum, these studies indicate a possible trochaic bias in
early word processing or representation developed from
repeated exposure to the stress pattern regularities of
English.2Accordingly, one could expect the phonetic
representations of 20- to 24-month-old children to be
more detailed for trochaic than iambic sequences.
In a joint design, two experiments were conducted to
examine how children access phonetic information in
newly learnt words, depending on the manipulation of
the three factors under scrutiny: syllable stress, word
position, and word stress pattern. In both experiments
children were presented with pairs of disyllabic pseudo-
words, half produced with a trochaic pattern, half with
an iambic pattern. In Experiment 1, pseudo-word pairs
differed by a consonant located at the onset of the first
syllable (word-initial), while in Experiment 2 the differing
consonant was at the onset of the second syllable (word-
Twenty- to 24-month-old children were engaged in a
variant of the word learning procedure developed by
Nazzi and Gopnik (2001; see also Havy & Nazzi, 2009),
in which they were presented with novel objects labelled
using pairs of pseudo-words that differed only on
were disyllabic, with half of the pairs produced with a
trochaic pattern, and the other half with an iambic
⁄⁄). For example, the child would first be presented
with an object labelled a ⁄
a different object labelled a ⁄
then introduces a third novel object also called a ⁄
and asks the child to give him or her ‘the other ⁄
These pseudo-word pairs
⁄, and then presentedwith
⁄. The experimenter
2Iambic words are also quite rare in children's early vocabulary, as they
are in the English lexicon in general. A count in the Oxford Commu-
nicative Developmental Inventory (Hamilton et al., 2000), which
includes 416 words, shows that the vast majority of the words intended
to describe most of the vocabulary of young children in perception and
production until 25 months is made up of trochaic words. Out of the
352 content words (excluding 12 onomatopoeias and 40 function
words), 225 are monosyllabic, 119 are disyllabic, and 20 have three or
more syllables. Out of the 119 disyllabic words, 112 are trochaic (the
exceptions being giraffe, balloon, outside, asleep, today, inside). Out of
the 20 words containing three or more syllables, 15 are stress initial
(the exceptions being tomorrow, policeman, pyjamas, banana and
604Caroline Floccia et al.
? 2010 Blackwell Publishing Ltd.
For children to succeed in this task they must be able to
learn the label⁄referent pairings and also distinguish
between the phonetic contrast that separates the two
original labels. This procedure provides ecological plau-
sibility as children are actively involved in manipulating
new objects and associating them with novel words (see
also Nazzi, 2005; Nazzi et al., 2009; Nazzi & Bertoncini,
Thirty children (including 13 girls) with an average age of
this experiment. All were raised in English-speaking
families, had no recorded auditory or developmental
problems, and were no more than 6 weeks premature.
(from 3.8% to 100%) and 72.05% in comprehension (from
all eight trials of the experiment, four completed seven
trials and two completed six trials. Incomplete trials were
due to the child either giving no answer, or picking up or
putting both of the labelled objects into the pot. An
additional 25 children were also tested, but were rejected
minimum of six trials (seven);distraction (one) or because
Three children were also excluded due to Oxford CDI
scoresbelow the10thpercentileof the published norms(in
production: 0.96% and 1.44% at 20 months, and 1.92% at
Three novel objects were used for each of the eight test
trials. Each of these 24 unique small objects was selected
such that they would be unfamiliar to the children, and so
would not already be associated with a name. All of the
sets of three objects were physically distinct, differing in
shape, colour, and texture in an effort to equalize their
wasused tolabelthe first twonovelobjects, witheach pair
differing by a single feature (place or voice) initial con-
words were constructed such that they could be easily
pronounced in either atrochaicor aniambic template(see
Table 1). Itmust be noted, however,that the phonologyof
English is such that the quality of vowels can be modified
depending on the location of stress (Hayes, 1995); there-
fore not all versions of the pairs, trochaic versus iambic,
are necessarily phonetically equivalent.
Prior to the eight test trials, two similar warm-up trials
were presented. In the first one the two initial objects
consisted of a toy cow and a toy car, labelled respectively
an animal and a car. The third of the objects was a toy
pig, also labelled an animal. In the second warm-up, one
of the objects of the pair was a toy dog, labelled a dog,
while the other was a novel object, labelled a ⁄
third of the objects was also novel and labelled a ⁄
⁄ versus ⁄⁄).Non-
Each child was tested individually in the BabyLab facility
of the School of Psychology, after active consent had
been obtained from the parents or caregivers. The session
was filmed for scoring purposes, using a Sony Handy
Cam DCR – HC35E. The procedure was an adaptation
of the categorization task used by Nazzi (2005), but in
this version the child was not asked to categorize objects,
but simply to associate the object to its newly learnt label
(see Havy & Nazzi, 2009). The warm-up trials were
identical to the test trials except that the presentation of
the objects and the identification question were repeated
if the child’s initial response was incorrect (although the
child was not told that the answer was incorrect). All 30
children responded correctly to their first or second
attempt in the warm-up trials.
Each of the trials was composed of a presentation
phase, followed by an identification question. In the
presentation phase two objects were given to the child,
one at a time, with the child encouraged to manipulate
the object for a few seconds before placing it on the table
in front of them. Within each trial, the objects were ar-
Bold items are those with a sonorant pivotal consonant (leading to more ambisyllabicity)
The percentage of correct identification of the target items in Experiment 1 with trochaic (TR) or iambic (IM) stress patterns.
3Re-analyses of data including children who always chose an object on
the same side, and those who performed poorly at the Oxford CDI, gave
exactly the same pattern of results, in Experiments 1 and 2.
Lexical stress and phonetic processing605
? 2010 Blackwell Publishing Ltd.
ranged on the table in a left-to-right sequence (child’s
perspective) in order to minimize memory load. The
experimenter (a British English native speaker, naive to
the aims and hypotheses of the experiment) spoke while
presenting each object, saying (for example): ‘Look! A
⁄. This is a ⁄⁄. Do you want to play with the
⁄? Yes, play with the ⁄
right, let’s put the ⁄
⁄ on the table. Here.’ Each
object was named between five and seven times in the
presentation phase. These two objects were named with
two phonemically distinct non-words, such as ⁄
⁄, sharing the same stress pattern (both
were produced with either the trochaic or iambic
After the presentation phase, the experimenter tested
lexical recognition by introducing a third object, nam-
ing it twice, and putting this object in a pot placed in
the middle of the table at equal distance from the first
two objects. For example, if the first object was named
⁄, and the second a ⁄
could be named a ⁄
⁄ (or a ⁄
menter asked the child to put ‘the other ⁄
pot. While waiting for the response, the experimenter
looked at either the child’s face or the object in the pot
in order to avoid influencing the child’s response. After
the child’s response, positive feedback was provided
regardless of the choice made. Successful performance
corresponded to the selection of the similarly labelled
The order of presentation of the trials and the labels
associated with each object in the triad, the position on
the table and order of presentation of the objects were
counterbalanced between participants. Each child was
presented with four trochaic and four iambic pairs of
object labels, with a pseudo-random order of presenta-
tion so that no more than three consecutive pairs could
have the same stress pattern. Children’s productive and
receptive vocabulary was assessed using the vocabulary
part of the British equivalent of the CDI Toddlers
(Hamilton et al., 2000).
⁄. See this ⁄⁄? All
⁄, the third object
⁄). The experi-
⁄’ in the
Results and discussion
The average rate of correct target item identification
(success rate) of the 30 participants was 57.5%, signifi-
cantly greater than the 50% chance rate (t(29) = 2.14,
p = .040). ANOVA analyses of success rates were con-
ducted with stress condition (trochaic versus iambic) and
consonant contrast type (voice versus place) as within-
participant factors. No main effect of consonant contrast
type was found (F(1, 29) < 1), nor did this factor sig-
nificantly interact with the factor of stress condition (F(1,
29) < 1), therefore contrast type will not be included in
further analyses. Stress pattern was shown to have a
significant effect over success rate (F(1, 29) = 5.53, p =
.026), with performance in the trochaic condition higher
than in the iambic condition, as can be seen on Figure 1
(see also the details of results by items in Table 1).
Planned comparisons showed that participants per-
formed above chance level in the trochaic condition
(65.6%, t(29) = 2.68, p = .012), but were at chance level
in the iambic condition (48.6%, t(29) < 1). Tests of cor-
relation between the success rates of stimuli with trochaic
and iambic stress patterns were not significant (r = ).05,
p = .80). There was no correlation between overall suc-
cess rate in the task and either CDI production (r = .34,
p = .067) or comprehension (r = .28, p = .12) scores.
Age correlated with the CDI scores in production
(r = .47, p = .008), but the correlation between age and
the CDI scores in comprehension only approached sig-
nificance (r = .32, p = .081), possibly because of a ceiling
Further analyses, comparing success rates for indi-
vidual pseudo-word pairs, shown in Table 1, highlight
considerable variability between stimuli pairs, ranging
from 33.3 to 81.3%. To account for this variability we
explored two factors that were not controlled in the
original design: the possibility of embedded familiar
words (as suggested to us by the Editor), and the degree
of ambisyllabicity in the pivotal intervocalic consonant
between first and second syllable nuclei.
When considering embedded words it is possible that
prior knowledge of familiar words embedded within our
non-word stimuli could have coloured the representation
of the to-be-discriminated items. For example, the word
show is embedded in the pair ⁄
extracted, leaves a lexically acceptable residue – accord-
ing to the Possible Word Constraint (PWC) formulated
by Norris, McQueen, Cutler and Butterfield (1997;
Johnson, Jusczyk, Cutler & Norris, 2003), which states
that a new word cannot be segmented if it leaves a single
consonant as a residue. With this in mind, an examina-
tion of the stimuli reveals that the word show is also
embedded – to a lesser extent – in ⁄
together with the words poo and two. Four similar cases
were also identified in our stimuli: mess and toe in
⁄ vs. ⁄⁄ and ⁄
and do in ⁄
⁄ vs. ⁄
⁄, key in ⁄⁄ vs. ⁄
⁄ vs. ⁄⁄, and if
⁄ vs. ⁄⁄,
⁄ vs. ⁄⁄, bee
⁄, pee and bee in ⁄
⁄. Finally, the word
% of correct choice
iambic items in Experiment 1 (contrast on the first consonant)
and Experiment 2 (contrast on the medial consonant).
Percentage of correct responses for trochaic and
606Caroline Floccia et al.
? 2010 Blackwell Publishing Ltd.
doughnut could sound like a mispronunciation of
⁄ vs. ⁄⁄. Although some of these words are
not listed in the Oxford CDI (mess, doughnut, poo, pee,
two and do), they are likely to have been heard by chil-
dren by 23 months. However, post-hoc analyses revealed
no significant differences in success rates for the eight
pairs of stimuli with familiar embedded words (60.7%)
and the remaining eight pairs of non-words without
embedded words (58.6%; paired t-test, t(29) < 1).
Discounting the possibility that embedded familiar
words might affect success rates, we now turn to ambisyl-
labicity. As defined by Lahiri (2001, p. 1351): ‘Ambisyl-
labicity results from the attraction of the first consonantal
onset of an unstressed syllable to form the coda of the
preceding syllable. This consonant then becomes ambi-
syllabic, since it belongs both to the onset and the coda of
the two syllables’. For example, the ⁄m⁄ in lemon belongs
to both syllables. Whether or not a consonant is ambi-
syllabic in English depends on many factors: ambisyl-
labicity is much more frequent in trochaic words than
an obstruent, and if the preceding vowel is short rather
definitions in mind it appears that our trochaic pairs are
more likely to be analysable as including an ambisyllabic
medial consonant than are the iambic pairs. In addition,
while all trochaic stimuli had long initial vowels, three
contained sonorant medial consonants (⁄
⁄, ⁄⁄ vs. ⁄⁄ and ⁄
which could lead to increased ambisyllabicityoverthe five
remaining pairs which had obstruent medial consonants.
To examine the effect of potential ambisyllabicity on
success rates, an ANOVA was conducted using within-
participant factors of stress pattern (trochaic versus
iambic) and ambisyllabicity (sonorant versus obstruent
pivotal consonant). Due to the relative paucity of stimuli
pairs with sonorant intervocalic consonants and the
random distribution of stimuli across participants, only
14 out of the 30 children presented with both types of
stimuli were included in this analysis. This showed a
marginal effect of ambisyllabicity (F(1, 13) = 4.33,
p = .058), with stimuli containing the less ambisyllabic
obstruent intervocalic consonants having a higher aver-
age success rate (61.9%) than those with the more
ambisyllabic sonorant consonants (46.3%). No interac-
tion was found between stress pattern and ambisyllab-
icity (F(1, 13) = 2.34, p = .15).
To summarize, in this experiment we examined 20- to
24-month-old children’s phonetic encoding of a word-
initial consonant contrast in trochaic and iambic carrier
words using aword learning task. The results showed that
children were able to distinguish the initial phonemic
contrast, but only when embedded within a trochaic car-
rier, and not when the initial syllable of the carrier was
properties of English continue to affect the accuracy of
children’s representations of word forms by the end of
⁄ vs. ⁄⁄),
their second year. Further implications of this finding will
be discussed when they can be combined with those of
at the start of the second syllable. This will allow a fuller
understanding of how word position, syllable stress, and
word stress pattern modulate phonetic encoding in early
In this experiment the procedure used in Experiment 1
was repeated but in this case the phonemic contrast
between non-words was positioned on the pivotal inter-
Thirty children (including 16 girls) with an average age
of 21.2 months (from 18.3 to 23.8) were successfully
tested in this task. These children were selected using the
same criteria as those of Experiment 1, and scored an
average of 40.2% in the production Oxford CDI (from
3.6% to 96.4%), and 69.0% in comprehension (from
35.1% to 100%). Out of these 30 children, 11 completed
the eight trials, 17 completed seven trials and two com-
pleted six trials. A further 14 children were tested but
were rejected because they showed a systematic bias to
pick up the object on one side (five), did not complete a
minimum of six trials (five), chose to ‘tidy up’ all the
objects (one), or systematically put both objects in the
pot (one). We also excluded two children who scored
below the 10th percentile of the published norms of the
Oxford CDI (in production: 1.9% at 21 months and
2.2% at 23 months).
Test stimuli were constructed in the same manner as in
Experiment 1, differing only in the location of the pho-
nemic contrast, which here is located on the pivotal
intervocalic consonant. Moreover, the initial syllables
carrying the consonant used for the phonemic contrast in
Experiment 1 were reused in this experiment, moved
from the first to the second syllable position of the non-
words. For example, the contrastive pair ⁄
⁄ in Experiment 1 became ⁄
Experiment 2. The only exceptions to this were the
⁄ vs. ⁄⁄ and ⁄
contrast) in Experiment 1 transforming into ⁄
⁄ and ⁄⁄ vs. ⁄
articulation contrast), due to experimenter error. Wher-
ever possible non-word pairs in this experiment were
constructed simply by reversing the syllables of stimuli
from Experiment 1, although this was possible only for
⁄ vs. ⁄
⁄ vs. ⁄⁄ pairs (voice
Lexical stress and phonetic processing607
? 2010 Blackwell Publishing Ltd.
⁄ inExperiment1became ⁄⁄ vs. ⁄⁄ in
The procedure was identical to that used in Experiment 1.
The average success rate across the 30 participants was
60.6%, significantly above the chance level of 50% (t(29)
= 3.96, p < .001). Further ANOVA analyses of success
rate were also conducted with within-participant factors
of stress pattern (trochaic versus iambic) and type of
consonant contrast (voice versus place). As in Experi-
ment 1, there was no significant main effect of consonant
contrast (F(1, 29) < 1), nor did it interact with stress
pattern (F(1, 29) < 1). Stress pattern was found to have a
significant effect over success rates in Experiment 2 (F(1,
29) = 5.09; p = .032), with participants performing better
in the iambic than in the trochaic condition (see Figure 1,
and Table 2 for the results by items). Further analyses
showed that children performed above chance only in the
iambic condition (69.2%, t(29) = 4.38, p < .001), not in
the trochaic condition (52.5%, t(29) < 1), the opposite
pattern to that found in Experiment 1. No significant
correlation was found between success rates in the tro-
chaic and iambic stress patterns (r = ).30, p = .10), nor
between overall success rate and CDI production (r =
).03) or comprehension (r = .18) scores. Children’s age
correlated with CDI scores in production (r = .56, p =
.001) and in comprehension (r = .56, p = .001).
As in Experiment 1, analyses of success rate variability
between non-word pairs were conducted, investigating
the possible effects of embedded words and ambisyllab-
icity. Five non-word pairs were found to contain
embedded words, show, poo and two in ⁄
⁄ and in ⁄⁄ vs. ⁄
⁄, key in ⁄⁄ vs. ⁄
⁄. A t-test comparing children’s success rates
between these five pairs (66.1%) and the 11 pairs without
(t(29) = 1.26, p = .22). Therefore, as in Experiment 1,
there is no evidence to suggest that the children’s
knowledge of embedded words had a significant effect
over their success rates.
⁄, bee and pee in ⁄
⁄, and bed in ⁄
An examination of potentially ambisyllabic stimuli in
Experiment 2 highlighted differences in initial vowel
length, with four trochaic pairs having initial short
⁄ vs. ⁄⁄, ⁄
⁄ and ⁄⁄ vs. ⁄⁄) and four initial long vowels.
It is worth noticing here that the ambisyllabicity of the
pivotal consonant was mainly related to the nature of the
pivotal consonant in Experiment 1, and to the length of
the preceding vowel in Experiment 2. In addition, as all
the medial consonants in the trochaic items were
obstruent in Experiment 2, it was not possible to analyse
the potential ambisyllabic effect of sonorant⁄obstruent
consonant types, as was conducted in Experiment 1.
As ambisyllabicity should be greater for stimuli with
initial short vowels than long vowels, a post-hoc ANOVA
analysis of individual success rates was conducted using
within-participant factors of word stress pattern (tro-
chaic versus iambic) and ambisyllabicity (short versus
long vowels). This analysis did not reveal any significant
effect of ambisyllabicity on success rates (F(1, 29) = 2.52,
p = .12), although items with the long vowels (leading to
less ambisyllabicity) tended to be more successful
(66.7%) than those with the short vowels (leading to
more ambisyllabicity; 56.7%). There was no interaction
between stress pattern and degree of ambisyllabicity (F(1,
29) = 2.73, p = .11; see Figure 2).
⁄ vs. ⁄⁄, ⁄⁄ vs.
Bold items are those with an ambisyllabic pivotal consonant
The percentage of correct identification of the target items in Experiment 2 with trochaic (TR) or iambic (IM) stress patterns.
Exp1 (change in
Exp2 (change in
Exp1 (change in
Exp2 (change in
% of correct choice
iambic items in Experiments 1 and 2 as a function of ambi-
syllabicity of the medial consonant.
Percentage of correct responses for trochaic and
608Caroline Floccia et al.
? 2010 Blackwell Publishing Ltd.
Comparison of the results of Experiments 1 and 2
By combining the results of Experiments 1 and 2 we made
a direct comparison of phonemic discrimination success
rates with a between-participants factor of word position
(phonemic contrast being word-initial versus word-med-
ial) and a within-participant factor of stress pattern
(trochaic versus iambic). The effect of the third variable,
namely the stress status of the syllable containing the
phonemic contrast, was represented by an interaction
between word position and stress pattern. This analysis
revealed no main effects; neither the effect of word po-
sition (F1(1, 58) < 1) nor stress pattern reached signifi-
cance (F1(1, 58) < 1). However, there was a significant
interaction between stress pattern and position in the
word (F1(1, 58) = 10.60, p = .002), showing that partic-
ipants were more successful with trochaic stimuli when
the phonemic contrast was word-initial, and with iambic
stimuli when the phonemic contrast was word-medial.
When considering the variability in success rates
between individual non-word pairs, there was a consis-
tent, yet marginal, effect of ambisyllabicity noted in both
experiments. Extending this method of analysis across the
results of the two experiments we conducted an ANOVA
on individual success rates with experiment (1 or 2) as a
between-participants factor and stress pattern (trochaic
or iambic) and ambisyllabicity (stimuli prevalent to
ambisyllabicity or not) as within-participant factors. This
revealed an interaction betweenambisyllabicityand stress
pattern (F(1, 41) = 4.47, p = .04), with ambisyllabicity
reducing success rates in trochaic stimuli (ambisyllabic:
68.2%, non-ambisyllabic: 44.9%), but not in iambic
stimuli (ambisyllabic: 58.5%, non-ambisyllabic: 59.8%).
Overall the combined results from both experiments
indicate that 20- to 24-month-old children are able to
distinguish phonetic contrasts only in stressed syllables,
and that even this ability is lost if the stressed syllable in
trochaic words is ambisyllabic.
This experiment revealed that when the phonemic con-
trast between non-words was located on a medial
consonant, children were better at distinguishing those
words when their stress pattern was iambic rather than
trochaic. By combining the results of Experiments 1 and
2 we learn that the consonantal phonemic contrast is
salient only when its syllabic carrier is stressed. This
would appear to be irrespective of the location of the
stressed syllable, or the stress pattern of the word.
Before discussing the general implications of these
results, it was important to consider the possibility that
the pattern of results obtained so far could be explained
by the acoustic characteristics of the stimuli. It is possible
that the stress patterns and⁄or of the to-be-discriminated
contrasts were realized such that only the contrasts
embedded in stressed syllables could be distinguished. To
test for this possibility we randomly selected audio
recordings of two testing sessions for each experiment,
with the constraint that these recordings would cover the
entirestimuli set for both experiments. Then, foreach pair
we extracted the sentences containing the first two
occurrences of both to-be-learned pseudo-words (for
example: ‘I have a ⁄
⁄!’; ‘Do you want to play with
⁄?’), plus the two occurrences of the target
pseudo-word (for example: ‘Now I have another
⁄!’;‘Canyouput theother ⁄
please?’). A total of 96 pairs of sentences were extracted,
and presented to a group of 15 naive English-speaking
adults (including eight females; mean age 24.2 years) in a
same⁄different decision task and a transcription task. In
the former task, participants were presented with the two
sentences describing one of the learned objects followed
by the two sentences describing the target object. The
participants were then asked to decide whether the two
object names were the same or different by pressing
appropriate keys. Of the 64 pairs of object names pre-
sented, half were similar and half different. The error rate
with these adultswas only 7.52%, with 2.32% (SD = 1.86)
errors on false alarms, and 5.20% (SD = 2.86) errors on
misses. Sensitivity (A¢) measure of signal detection was
computed for each participant and each condition (see
Table 3). Adults scored very high on A¢ values, ranging
from 0.93 to 1, showing discrimination close to ceiling
level. An ANOVA with phonemic contrast position (first
or medial consonant) and word stress pattern (iambic or
trochaic) as repeated measures was conducted on A¢. In
contrast to the experiments with children, these analyses
14) = 2.73, p = .12). However, word position had a main
significant effect (F(1, 14) = 19.35, p = .001), because of
lower values of A¢ when the contrast was word-initial
(.972) than word-medial (.987). Similarly, stress pattern
was found to have a main significant effect (F(1,
⁄ in thepotfor me,
the adult same⁄different decision task, together with mean A¢ values (and standard deviations). A percentage of 4.17% misses
corresponds to an average of 0.66 items missed out of the 16 items in this category (per participant)
Mean percentages (and standard deviations) of misses, false alarms and total errors (false alarms and misses) collected in
Stress pattern Contrast% of misses
% of false alarms
% of total errorsSD
Lexical stress and phonetic processing 609
? 2010 Blackwell Publishing Ltd.
14) = 5.53, p = .034), due to lower A¢ values for iambic
(.974) than for trochaic sequences (.985).
After the discrimination task, our adult participants
were asked to transcribe 96 pairs of sentences, each
consisting of two utterances describing a particular
object. Given the opacity of the English orthographic
system, vowel transcriptions were highly variable, with
the item ⁄
⁄ transcribed variously as begoo, begu,
bagu, bergue, bugoo, bagooe, begou, bugue. However, as
we were primarily interested in the transcription of the
consonantal contrast, transcriptions were reported as
correct if both syllables of the item were represented, and
the consonant of interest was correctly transcribed.
Using these criteria only 2.36% (SD = 2.79) of tran-
scriptions were erroneous, showing that all the contrasts
were easily perceived by our adult participants (see Ta-
ble 3). Moreover, these few errors were equally distrib-
uted across all of the factors under analysis.
The results of both the adult discrimination and
transcription experiments show that adults could dis-
tinguish our phonemic contrasts at near ceiling level, and
that they were no better at detecting contrasts in stressed
syllables than unstressed syllables. Therefore it is unlikely
that children were prevented from perceiving the pho-
nemic contrast in unstressed syllables due to the acoustic
characteristics of the stimuli.
In a final comparison we also tested a group of 12
adults (mean age 32.8 years, including eight females) in a
word learning task similar to that used with the children
using the recorded sentences extracted for the adult
same⁄different experiment. Participants first heard two
sentences containing the non-word label for ‘object 1’,
followed by two sentences containing the similar, but
phonemically contrasted, non-word label for ‘object 2’.
After this ‘learning phase’ the participants heard two
sentences describing a target item corresponding to the
label of one of the learned objects, and were asked to
make a speeded judgment as to whether the target was
the same as object 1 or 2. The presentation order of the
32 object pairs used in this experiment was randomized
across participants. Error rates were too low to allow any
informative analyses (mean 0.98%). The only significant
reaction time effect was an interaction between word
stress pattern and phonemic contrast position (F(1,
11) = 6.32, p = .029). This revealed that one particular
combination of conditions was significantly slower than
the other three, namely the initial consonant contrast in
iambic non-words. Therefore, although adults had
slightly greater difficulty in differentiating phonemic
contrasts in unstressed syllables, this was specific to
iambic words, and was not generalized across both
iambic and trochaic words, as found with children.
The aim of this study was to investigate the interaction
between supra-segmental information and phonetic
specificity of early lexical processing. To this end, we
tested 20- to 24-month-old English-learning children in a
word learning task where the to-be-learnt items differed
by a single consonant located either in the first (Exper-
iment 1) or second syllable onset (Experiment 2) in both
trochaic and iambic word forms. These experiments were
designed to investigate the interaction between three
factors. First, the position of the phonemic contrast
(word-initial or word-medial) allowed us to test the
possibility that word-initial position might not be as
important in children as it is in adults’ word recognition
(e.g. Swingley, 2009). For the stress status of the contrast-
bearing syllable (stressed versus unstressed) we predicted
that stressed syllables would lead to better acous-
tic⁄phonetic processingthan unstressed
Sebasti?n-Gall?s et al., 1992). Finally, for the stress
pattern of the word (trochaic versus iambic) we wished to
examine whether the trochaic bias found in both 7.5-
month-old English-learning infants (e.g. Jusczyk et al.,
1999) and English-speaking adults’ segmentation (e.g.
Cutler & Norris, 1988) might also affect phonetic pro-
cessing in 20- to 24-month-olds.
One of the findings of this study was that the absolute
position of thephonemic
word-initial or word-medial, had no effect upon the
children’s ability to discriminate between consonants.
This was similar to those of a mispronunciation detec-
tion task with 14- to 22-month-old English-learning
children (Swingley, 2009), and a word learning task with
20-month-old French-learning children (Nazzi & Ber-
toncini, 2009), which also failed to find processing
asymmetries between word-initial and non-initial con-
trasts. It is possible that the word learning task used by
ourselves and Nazzi and Bertoncini (2009) and Swing-
ley’s (2009) mispronunciation detection task were not
sufficiently sensitive to reveal the asymmetries shown in
some adult studies (e.g. Creel & Dahan, 2010). To
evaluate this possibility a computer-mediated version of
this task with children’s gaze tracked when presented
with visual representations of objects could potentially
reveal processing differences between non-initial and
If the absence was found to be robust across tasks then
it could be argued that the importance of word-initial
position in adult lexical recognition is an emerging
property after 20 to 24 months, as a consequence of
computing lexical neighbourhood in an expanding lexi-
con. That is, when encoding a new word, children would
gradually learn to devote more attention to word onsets
than to word offsets (segments or syllables), with the
initial consonant or syllable of a word being to serve as
an efficient sorting element in a Cohort-like model of
word recognition (e.g. Marslen-Wilson, 1987). Before
that stage the relatively sparse lexical neighbourhood
would allow the distance between lexical entries to be
measured by information at different positions of the
word. It is also possible that word onsets come to be
increasingly important over the course of lexical devel-
610Caroline Floccia et al.
? 2010 Blackwell Publishing Ltd.
opment due to increases in processing speed, and con-
sequently working memory abilities, which are typically
observed throughout childhood (see Fry & Hale, 1996;
Marchman & Fernald, 2008). Therefore, increases in
processing speed may lead to reductions in the temporal
window of analysis used during speech perception,
reducing focus to smaller and smaller chunks of speech
until recognition is reliant upon the fast and efficient
analysis of word onsets.
Moving to another of our experimental factors, it was
found that the stress pattern of our stimuli did not have
an absolute effect over the detection of phonemic con-
trasts. These contrasts were detected more accurately
within stressed syllables, no matter whether they were
located within trochaic or iambic words. Similarly, stress
pattern did not affect the low degree of accuracy for
contrasts located within unstressed syllables. Neither our
study nor Vihman et al.’s (2004) with 11-month-olds
found evidence to support a trochaic bias, such as that
found with studies of English-learning infants (Curtin,
Mintz & Christiansen, 2005; Jusczyk et al., 1999) and
adults (e.g. Cutler & Norris, 1988; Cutler & Butterfield,
1992; Echols et al., 1997). Therefore we could not lend
credence to the hypothesis that the predominant word
form, represented by the trochaic pattern, facilitates
word-level acquisition. This lack of facilitation is similar
to findings recently reported for another type of per-
ceptual bias, namely the labial-coronal bias. Labial-
coronal (LC) words (e.g. pat) are more frequent than
coronal-labial (CL) words (e.g. tap) in most languages
(Vall?e, Rousset & Bo?, 2001), more frequent in chil-
dren’s productions during their second year (MacNeilage
&Davis, 2000), and preferred
10 months of age (Gonzalez Gomez & Nazzi, submitted;
Nazzi, Bertoncini & Bijeljac-Babic, 2009). However, they
are not learned any more effectively than CL words in
word learning tasks similar to that employed in this study
(Nazzi & Bertoncini, 2009). To draw further conclusions
on this subject, younger children would have to be tested,
as the 20- to 24-month-old children tested in these
studies may have been able to overcome any such bias
Although neither the position of the phonemic con-
trast nor the stress pattern of the stimuli had absolute
effects on phonemic discrimination, when these factors
were combined it was clear that children were able to
access phonetic information more accurately in stressed
syllables than in unstressed ones. This suggests that in
English supra-segmental information modulates the
availability of segmental information in young children
(as had been found for known word recognition at
11 months; Vihman et al., 2004; but see Johnson, 2005).
same⁄different and transcription tasks, showed that
adults had no difficulty in perceiving the phonetic iden-
tity of either stressed or unstressed syllables. This indi-
cates that children’s difficulty with phonemic contrasts in
unstressed syllables is not likely to be due to poor
in perception by
phonetic realization. However, future research will be
required to establish whether these difficulties are due to
children’s inability to process phonetic information in
unstressed syllables or to represent that information
(which could be perceived, but not used, to build lexical
Finally, analyses of the variation in the phonetic dis-
crimination between individual stimulus pairs high-
lighted the fact that ambisyllabicity has an important
modulating influence upon children’s performance.
Although the study was not originally designed to
address this issue, children seemed to have more prob-
lems in discriminating phonetic differences if there was
prevalence to ambisyllabicity in the pivotal intervocalic
consonant of the stimuli. This effect was found only in
trochaic word forms, a pattern which itself increases the
tendency to be analysed as ambisyllabic segments over
iambic pattern. This effect was such that in Experiment 1,
consonants had the highest rates of discrimination
success, while performance for the trochaic items with
ambisyllabic consonants was at chance. That no such
effect was found in paired iambic items suggests that this
was not due to the identity of the pivotal consonant.
We propose two possible explanations to explain the
effect that ambisyllabicity has over the detection of
phonemic contrasts. First, children compute a full rep-
resentation of every strong syllable, and the specification
of this representation is more consistent and accurate if
there is no variability caused by ambisyllabic segments. A
more specific representation of stressed syllables would
help onset detection, when the contrast is at the onset of
either the first syllable (Experiment 1) or the second
syllable (Experiment 2). Another possibility is that the
ambisyllabic effect is due to attention, with the structure
of the task leading children to focus on both strong and
weak syllable onsets in the search for the discriminative
contrast. If one of the segments is ambisyllabic this
impedes this attentional mechanism. For either hypoth-
esis one would predict that the effect of ambisyllabicity in
trochaic items would be stronger when the to-be-at-
tended contrast is precisely located on the ambisyllabic
pivotal consonant. However, our results do not support
this prediction. Thus, more research is clearly required to
examine the links between stress and ambisyllabicity in
early lexical representations.
In summary, our study shows that in an object
manipulation word learning situation young English-
learning children spontaneously focus their attention
upon the stressed syllables in new sequences, but do not
process unstressed syllables in great detail. This finding
holds irrespective of the location of stressed syllables in
the sequences, that is in both trochaic and iambic words.
This indicates that at the end of a child’s second year the
processing of the phonetic detail of newly learnt words is
not yet fully mature, and, at least in English, is modu-
lated by stress-related properties of the to-be-analysed
syllables. More research will be necessary to understand
Lexical stress and phonetic processing 611
? 2010 Blackwell Publishing Ltd.
the developmental pathway that leads these English-
learning children from such an asymmetric level of pro-
cessing between stressed and unstressed syllables to fully
mature processing of phonetically specified representa-
tions of words in their native language.
We wish to thank two anonymous reviewers and Nuria
Sebasti?n-Gall?s for very helpful comments on earlier
versions of this paper.
Allen, G.D., & Hawkins, S. (1980). Phonological rhythm: def-
inition and development. In G.H. Yeni-Komshian, J.F.
Kavanagh, & C.A. Ferguson (Eds.), Child phonology: Vol. 1.
Production (pp. 227–256). New York Academic Press.
Anderson, J., & Jones, C. (1974). Three theses concerning
phonological representations. Journal of Linguistics, 10, 1–26.
Bailey, T.M., & Plunkett, K. (2002). Phonological specificity in
early words. Cognitive Development, 17 (2), 1265–1282.
Brown, G. (1977). Listening to spoken English. London:
Charles-Luce, J., & Luce, P.A. (1990). Similarity neighbour-
hoods of words in young children’s lexicons. Journal of Child
Language, 22 (3), 727–735.
Christophe, A., Nespor, M., Guasti, M.T., & Van Ooijen, B.
(2003). Prosodic structure and syntactic acquisition: the case
of the head-direction parameter. Developmental Science, 6
Content, A., Kearns, R.K., & Frauenfelder, U.H. (2001).
Boundaries versus onsets in syllabic segmentation. Journal of
Memory and Language, 45, 177–199.
Creel, S.C., & Dahan, D. (2010). The effect of the temporal
structure of spoken words on paired-associate learning.
Journal of Experimental Psychology: Learning, Memory, and
Cognition, 36 (1), 110–122.
Curtin, S. (2010). Young infants encode lexical stress in newly
encountered words. Journal of Experimental Child Psychol-
ogy, 105 (4), 376–385.
Curtin, S., Mintz, T.H., & Christiansen, M.H. (2005). Stress
changes the representational landscape: evidence from word
segmentation. Cognition, 96 (3), 233–262.
Cutler, A., & Butterfield, S. (1992). Rhythmic cues to speech
segmentation: evidence from juncture misperception. Journal
of Memory and Language, 31 (2), 218–236.
Cutler, A., & Carter, D.M. (1987). The predominance of strong
initial syllables in the English vocabulary. Computer Speech &
Language, 2 (3–4), 133–142.
Cutler, A., & Norris, D. (1988). The role of strong syllables in
segmentation for lexical access. Journal of Experimental Psy-
chology: Human Perception and Performance, 14, 113–121.
Di Cristo, A. (1998). Intonation in French. In D. Hirst & A.D.
Cristo (Eds.), Intonation systems: A surveyof twenty languages
(pp. 195–218). Cambridge: Cambridge University Press.
Echols, C., Crowhurst, M.J., & Childers, J.B. (1997). The
perception of rhythmic units in speech by infants and adults.
Journal of Memory and Language, 36, 202–225.
Echols, C., & Newport, E.L. (1992). The role of stress and
position in determining first words. Language Acquisition, 2,
Eilers, R.E., Gavin, W., & Wilson, W.R. (1979). Linguistic
experience and phonemic perception in infancy: a crosslin-
guistic study. Child Development, 50, 14–18.
Eimas, P.D., Siqueland, E.R., Jusczyk, P.W., & Vigorito, J.
(1971). Speech perception in infants. Science, 171, 303–306.
Fennell, C.T., & Werker, J.F. (2003). Early word learners’
ability to access phonetic detail in well-known words.
Language and Speech, 46, 245–264.
Floccia, C., Nazzi, T., & Bertoncini, J. (2000). Unfamiliar voice
discrimination for short stimuli in newborns. Developmental
Science, 3 (3), 333–343.
Fry, A.F., & Hale, F. (1996). Processing speed, working
memory and fluid intelligence. Psychological Science, 7 (4),
Gimson, A.C. (1989). An introduction to the pronunciation of
English. New York: Edward Arnold.
Gonzalez Gomez, N., & Nazzi, T. (submitted). Acquisition of
non-adjacent phonological regularities in the first year of life:
evidence from a perceptual equivalent of the labial-coronal
Hall?, P., & de Boysson-Bardies, B. (1994). Emergence of an
early lexicon: infants’ recognition of words. Infant Behavior
and Development, 17, 119–129.
Hall?, P., & de Boysson-Bardies, B. (1996). The format of
representation of recognized words in infants’ early receptive
lexicon. Infant Behavior and Development, 19 (4), 463–481.
Hamilton, A., Plunkett, K., & Schafer, G. (2000). Infant
vocabulary development assessed with a British Communi-
cative Development Inventory: lower scores in the UK than
the USA. Journal of Child Language, 27, 689–705.
Havy, M., & Nazzi, T. (2009). Better processing of consonantal
over vocalic information in word learning at 16 months of
age. Infancy, 14 (4), 439–456.
Hayes, B. (1995). Metrical stress theory: Principles and case
studies. Chicago, IL: University of Chicago Press.
Hçhle, B., Bijeljac-Babic, R., Herold, B., Weissenborn, J., &
Nazzi, T. (2009). Language specific prosodic preferences
during the first half year of life: evidence from German
and French infants. Infant Behavior and Development, 32,
Johnson, E.K. (2005). English-learning infants’ representations
of word forms with iambic stress. Infancy, 7 (1), 99–109.
Johnson, E.K., Jusczyk, P.W., Cutler, A., & Norris, D. (2003).
Lexical viability constraints on speech segmentation by
infants. Cognitive Psychology, 46 (1), 65–97.
Jusczyk, P.W., & Aslin, R.N. (1995). Infants’ detection of
sound patterns of words in fluent speech. Cognitive Psy-
chology, 29, 1–23.
Jusczyk, P.W., Cutler, A., & Redanz, N.J. (1993). Infants’
preference for the predominant stress patterns of English
words. Child Development, 64 (3), 675–687.
Jusczyk, P.W., Houston, D.M., & Newsome, M. (1999). The
beginnings of word segmentation in English-learning infants.
Cognitive Psychology, 39, 159–207.
Ladefoged, P. (1993). A course in phonetics. Orlando, FL:
Lahiri, A. (2001). Metrical patterns. In M. Haspelmath (Ed.),
Language typology and language universals (pp. 1347–1366).
Berlin: Walter de Gruyter.
612Caroline Floccia et al.
? 2010 Blackwell Publishing Ltd.
Lasky, R.E., Syrdal-Lasky, A., & Klein, R.E. (1975). VOT
discrimination by four to six and a half month old infants
from Spanish environments. Journal of Experimental Child
Psychology, 20, 215–225.
MacNeilage, P.F., & Davis, B.L. (2000). The motor core of
speech: a comparison of serial organization patterns in
infants and languages. Child Development, 71, 153–163.
Mani, N., & Plunkett, K. (2007). Phonological specificity of
vowels and consonants in early lexical representations.
Journal of Memory and Language, 57 (2), 252–272.
Mani, N., & Plunkett, K. (2008). Fourteen-month-olds pay
attention to vowels in novel words. Developmental Science,
Marchman, V.A., & Fernald, A. (2008). Speed of word recog-
nition and vocabulary knowledge in infancy predict cognitive
and language outcomes in later childhood. Developmental
Science, 11 (3), 9–16.
Marslen-Wilson, W.D. (1987). Functional parallelism in spo-
ken word-recognition. Cognition, 25 (1–2), 71–102.
Metsala, J.L., & Walley, A.C. (1998). Spoken vocabulary
growth and the segmental restructuring of lexical represen-
tations: precursors to phonemic awareness and early reading
ability. In J.L. Metsala & L.C. Ehri (Eds.), Word recognition
in beginning literacy (pp. 89–120). Mahwah, NJ: LEA.
Nazzi, T. (2005). Use of phonetic specificity during the acqui-
sition of new words: differences between consonants and
vowels. Cognition, 98, 13–30.
Nazzi, T., & Bertoncini, J. (2009). Phonetic specificity in early
lexical acquisition: new evidence from consonants in coda
positions. Language and Speech, 52 (4), 463–480.
Nazzi, T., Bertoncini, J., & Bijeljac-Babic, R. (2009). A per-
ceptual equivalent of the labial-coronal effect in the first year
of life. Journal of the Acoustical Society of America, 126 (3),
Nazzi, T., Dilley, L.C., Jusczyk, A.M., Stattuck-Hufnagel, S., &
Jusczyk, P.W. (2005). English-learning infants’ segmentation
of verbs in fluent speech. Language and Speech, 48, 279–298.
Nazzi, T., Floccia, C., Moquet, B., & Butler, J. (2009). Bias for
consonantal over vocalic information in 30-month-olds:
crosslinguistic evidence from French and English. Journal of
Experimental Child Psychology, 102 (4), 422–537.
Nazzi, T., & Gopnik, A. (2001). Linguistic and cognitive abil-
ities in infancy: when does language become a tool for cate-
gorization? Cognition, 80 (3), B11–B20.
Nazzi, T., Iakimova, G., Bertoncini, J., Fr?donie, S., &
Alcantara, C. (2006). Early segmentation of fluent speech by
infants aquiring French: emerging evidence for crosslinguistic
differences. Journal of Memory and Language, 54, 283–299.
Nazzi, T., Jusczyk, P.W., & Johnson, E.K. (2000). Language
discrimination by English-learning 5-month-olds: effects of
rhythm and familiarity. Journal of Memory and Language, 43,
Nazzi, T., & New, B. (2007). Beyond stop consonants: conso-
nantal specificity in early lexical acquisition. Cognitive
Development, 22 (2), 271–279.
Norris, D., McQueen, J.M., Cutler, A., & Butterfield, S.
(1997). The possible-word constraint in the segmentation of
continuous speech. Cognitive Psychology, 34 (3), 191–243.
Rost, G.C., & McMurray, B. (2009). Speaker variability aug-
ments phonological processing in early word learning.
Developmental Science, 12 (2), 339–349.
Saffran, J.R., & Estes, K.G. (2006). Mapping sound to mean-
ing: connections between learning about sounds and learning
about words. In R. Kail (Ed.), Advances in child development
and behavior (Vol. 34, pp. 1–38). New York: Elsevier.
Sebasti?n-Gall?s, N., Dupoux, E., Segui, J., & Mehler, J.
(1992). Contrasting syllabic effects in Catalan and Spanish.
Journal of Memory and Language, 31, 18–32.
Stager, C.L., & Werker, J.F. (1997). Infants listen for more
phonetic detail in speech perception than in word-learning
tasks. Nature, 388, 381–382.
Streeter, L. (1976). Language perception of two-month-old
infants shows effects of both innate mechanisms and expe-
rience. Nature, 259, 39–41.
Swingley, D. (2003). Phonetic detail in the developing lexicon.
Language and Speech, 46, 265–294.
Swingley, D. (2009). Onsets and codas in 1.5-year-olds’ word
recognition. Journal of Memory and Language, 60, 252–269.
Swingley, D., & Aslin, R.N. (2000). Spoken word recognition
and lexical representation in very young children. Cognition,
76 (2), 147–166.
Trammel, R.L. (1993). English ambisyllabic consonants and
half-closed syllables in language teaching. Language Learn-
ing, 43 (2), 311–356.
Trehub, S.E. (1976). The discrimination of foreign speech
contrasts by infants and adults. Child Development, 47 (1),
Treiman, R., Bowey, J.A., & Bourassa, D. (2002). Segmenta-
tion of spoken words into syllables by English-speaking
children as compared to adults. Journal of Experimental
Child Psychology, 83, 213–238.
Treiman, R., & Danis, C. (1988). Syllabification of intervocalic
consonants. Journal of Memory and Language, 27, 87–104.
Vall?e, N., Rousset, I., & Bo?, L.J. (2001). Des lexiques aux
syllabes des langues du monde. Typologies, tendances et
organisations structurelles. Linx, 45, 37–50.
Vihman, M.M., DePaolis, R.A., & Davis, B.L. (1998). Is there
a ‘trochaic bias’ in early word learning? Evidence from infant
production in English and French. Child Development, 69 (4),
Vihman, M.M., Nakai, S., DePaolis, R.A., & Hall?, P. (2004).
The role of accentual pattern in early lexical representation.
Journal of Memory and Language, 50, 336–353.
Walley, A.C. (1993). The role of vocabulary development in
children’s spoken word recognition and segmentation abili-
ties. Developmental Review, 13 (3), 286–350.
Yoshida, K.A., Fennell, C.T., Swingley, D., & Werker, J.F.
(2009). Fourteen-month-old infants learn similar-sounding
words. Developmental Science, 12 (3), 412–418.
Received: 3 November 2009
Accepted: 19 July 2010
Lexical stress and phonetic processing 613
? 2010 Blackwell Publishing Ltd.