Poster Session 2 (with coffee break)
Friday, June 21st, 2013 [16:00 - 17:30]
Statistical learning in non-Chinese speakers exposed to a sequence of words without spaces
Wang, T. , Liu, Y. , Chen, J. & Li, A.
National Taiwan Normal University, Taiwan
A printed Chinese sentence is composed of a string of characters with no extra spaces between words. We hypothesize that automatic word segmentation by Chinese readers is developed via the statistical learning mechanism that is universal to all language learners. We tested the hypothesis by exposing 20 non-Chinese speakers to a sequence of 3600 characters constructed from six disyllabic words with 6 different characters. The words were repeated 300 times and concatenated randomly except that the same word did not follow itself immediately. The transitional probabilities between any two characters in the sequence were .46 to 1 within words, and 0 to .29 between words. The sequence was presented one character every half a second, from left to right and from top to bottom over 36 screen pages. A character disappeared when the next one came up. Occasionally, with a probability of 0 to .03, the presentation rate doubled. The participants? task was to follow the characters as they appeared, and detect the instances of double presentation each time it occurred. At the end of the disguised vigilance task, the participants were given a surprise test which consisted of words and nonwords (reversals of the two characters in a word) from the sequence. A word and a nonword were shown each time in a random order, with one above the other. The participants pressed the up arrow key if they thought the top character string had appeared before, and the down arrow key if the bottom one had appeared. The averaged accuracy rate was .53, slightly but significantly greater than chance performance. This suggests that automatic word segmentation in reading Chinese texts can be achieved via the universal statistical learning mechanism that is available to non-Chinese speakers.
Verbs, body parts and argument structure frames in child language: A comparison between English and Hebrew
Uziel-Karl, S. 1 & Sethuraman, N. 2
1 Ono Academic College
2 University of Michigan - Dearborn
Research findings have shown that speakers tend to connect verbs to a core meaning (giving is about HAND) (Maouene, et al., 2008), and independently, that, verb meaning and argument structure are linked (Gleitman, 1990; Naigles, 1990). The present study examines whether there is a strong, consistent link between a verb, a body-part denoting the action, and a specific argument structure frame (e.g. give- HAND- V NP). This is, perhaps, the first study to examine the relevance of body parts for early syntactic aquisition cross-linguistically. To this end, we examined longitudinal naturalistic speech samples of two native Hebrew-speakers and two native English-speakers between the ages 1;6-3;0 (CHILDES, MacWhinney, 2000) for uses of ten highly frequent verbs that had strong body part associations in each language. A total of 2,954 verb-containing utterances (Hebrew, N = 1,223; English, N = 1,731) were analyzed. Each verb was coded for body part association and for its different argument structure frames. Analysis reveals highly similar cross-linguistic patterns of association between verbs, body parts and argument structures: 1. Verbs associated with LEG (go, come) most often occur in intransitive constructions with a locative (Hebrew: 43%, English: 52%); 2. Verbs associated with HEAD (see, look, know, read) and HAND (get, make, put) tend to occur in transitive constructions (V NP: Hebrew: HEAD: 21%; HAND: 32%; English: HEAD: 38%; HAND: 60%); 3. Verbs associated with HEAD and LEG occur as bare verbs more often than verbs associated with HAND (V: Hebrew: HEAD: 34%; LEG: 52% vs. HAND: 17%; English: HEAD: 33%; LEG: 30% vs. HAND: 6%). Prior research has focused on the importance of linguistic cues over real-world observational cues for learning early verbs and syntax. Here we suggest that, in addition to abstract linguistic cues, body parts may contribute to children's acquisition of verb argument structures.
The development of initial consonant and vowel perception in early infancy: An ERP study
Cheng, Y. 1 , Wu, H. 2, 3 , Tzeng, Y. 1 & Lee, C. 1, 4
1 Institute of Neuroscience, National Yang-Ming University, Taipei, Taiwan
2 Buddhist Tzu Chi General Hospital, Taipei Branch, Taipei, Taiwan
3 School of Medicine, Tzu Chi University, Hualien, Taiwan
4 The Institute of Linguistics, Academia Sinica, Taiwan
Mandarin Chinese is a tonal language with a relatively simple syllable structure, in which vowels and tones are compulsory units for a syllable whereas initial and final consonants are optional. According to the phonological saliency hypothesis, a vowel carries higher saliency than a consonant and thus vowels should be acquired earlier than initial consonants. Supporting evidence has shown that Mandarin-speaking children?s vowel production stabilized by 3-year-old but the initial consonant production did not until age 5 to 6. This study examined the phonological saliency hypothesis by measuring mismatch response (MMR) to Mandarin vowel and initial consonant changes in newborns and 6-month-olds with a multi-deviant oddball paradigm. For initial consonants, the small deviant ?da? (VOT=11 ms) and the large deviant ?ga? (VOT=23 ms) were compared with the standard ?ba? (VOT=10 ms). For vowels, the small deviant ?di? (front-high vowel) and the large deviant ?du? (back-high vowel) were compared with the standard ?da? (central-low vowel). Newborns revealed a broad positive mismatch response (P-MMR) to both initial consonants and vowels, but showed no difference between large and small deviants. For 6-month-olds, the polarity of MMR to vowels depends on the deviant size. The du/da contrast elicited an adult-like MMN in intervals of 172 to 234 ms after the stimulus onset but the di/du contrast only elicited a P-MMR in 268 to 440 ms. However, only P-MMRs were found for both large and small deviants of initial consonants in 6-month-olds. In summary, infants show an adult-like MMR to vowel change at as early as 6 months of age, but show no such a sign when responding to initial consonants yet. Our findings characterize the developmental trajectories of MMRs to vowels and initial consonants and support the phonological saliency hypothesis.
Word learning: Children's individual tendencies and their effects on the developing phonological system
Szreder, M.
University of York
Claims about individual learning tendencies (or ?strategies?) in children often appear to either be based on post-hoc analysis (Ferguson & Farwell, 1975; Farwell, 1976), or to limit individual variation to distinct ?learning types? (Bates, Bretherton & Snyder, 1988; Bates, Dale & Thal, 1995.) We propose that individual tendencies should be included in the study of language development, but that their effects on the path of acquisition are not predictable when they are considered in isolation. Instead, we argue that language develops in response to all factors acting upon it (cf. Lindblom, Studdert-Kennedy & MacNeilage, 1983; Ferguson, 1986), which include individual learning tendencies but also articulatory constraints. We collected longitudinal data from three children, from the 25 word point onwards for six months, and examined the following variables: 1) accuracy (PCC: Shriberg & Kwiatkowski, 1982), i.e. articulatory precision; 2) whole-word templates (T score: Vihman et al., in prep), i.e. early phonological systematicity; 3) boldness (% attempted fricatives, affricates and liquids), i.e. the individual tendency to attempt challenging targets. The results show that articulatory routines only evolve into wide-spread templates if the child exhibits high boldness coupled with limited articulatory skills. Therefore, templates are grounded in the individual inclination to expand vocabulary in children whose articulatory skills are not yet developed well enough to allow for accurate production of many diverse targets. Conversely, while templates aid lexical expansion, they further reduce accuracy and increase boldness, showing that the interaction between the factors is bi-directional. We conclude that individual differences should not be viewed in terms of ?types of learners?. Rather, early phonological systematicity is the outcome of interplay between articulatory constraints and individual learning tendencies, whose dynamic is constantly changing in development. Factors involved in word learning and production interact bi-directionally resulting in re-organisation of the dynamically developing phonological system.
Lexical specificity training enhances phonological awareness in second language learners
Janssen, C. 1 , Segers, E. 1 , McQueen, J. M. 1, 2, 3 & Verhoeven, L. 1
1 Radboud University Nijmegen, Behavioural Science Institute, Nijmegen, The Netherlands
2 Radboud University Nijmegen, Donders Institute for Brain, Cognition and Behaviour, Nijmegen, The Netherlands
3 Max Planck Institute for Psycholinguistics, Nijmegen, The Netherlands
Vocabulary knowledge is strongly connected to later success in school, however, second language learners often enter primary school with limited vocabulary in the target language. Speech decoding skill, phonological awareness and lexical specificity are phonological variables that have been shown to be related to children?s vocabulary. Improvement of these variables may enhance vocabulary development of both first and second language learners. In this study, the role of lexical specificity in word learning was explored. Lexical specificity is defined as knowledge on how words ought to sound and is highly related with phonological awareness. First language learners (4-year-old monolingual Dutch children) and second language learners (4-year-old Turkish-Dutch bilingual children) were compared in their improvement on phonological awareness after having received a lexical specificity training. Preliminary results show that lexical specificity training enhances phoneme blending, the ability to identify a word when hearing parts of the word in isolation (a component of phonological awareness), both in first and second language learners. Moreover, phonological overlap of the Dutch and the Turkish language in stimulus materials of the lexical specificity training contributed most to the gain in phoneme blending for the second language learners. A first step has been taken in finding a way to close the gap in word knowledge between young first and second language learners.
Developmental changes in Mandarin speech perception abilities in children reflected by MMN responses
Chen, Y. & Liu, H.
Department of Special Education, National Taiwan Normal University
The purpose of this study is using MMN to explore the dynamic changes of Mandarin speech perception abilities from early to middle childhood. Twenty preschoolers (Mean =3.40 yr), 18 school-age children (Mean =8.57) and 26 Mandarin-speaking adults participated in this study. Two sets of synthesized speech stimuli varying in Mandarin segmental (Consonant: alveolo-palatal affricate vs. fricative contrast) or suprasegmental features (Lexical tone: /i2/ with rising tone vs. /i3/ with contour tone) were used. The proportion of standard to deviant was 8:1, and there were 1080 trials in each condition. Six frontocentral sites (F3/Fz/F4, C3/Cz/C4) were selected for analysis. Mean amplitudes of the MMRs were calculated for seven successive 50-ms intervals, with starting point defined as 100 ms after the onset of acoustic difference between the standard and deviant in each condition. Results for consonant condition showed that the typical early MMN response only appeared in adults. Both preschool and school children showed more positive responses to the deviant during the interval of 230 to 330 ms after the stimulus onset time, while school children showed additional negative responses in the later time interval from 380 to 480 ms. For lexical tone, all children and adults showed more negative responses to the deviant stimuli, and there is a clear maturational trend from late negativity in children to early negativity in adults. The maturation of MMRs to both speech contrasts could also be observed in the change of topography, shifting from central to frontal sites and right to left in the hemisphere aspect. In sum, our results demonstrated the developmental changes in MMRs to Mandarin speech sounds. Furthermore, using contrasts representing different speech features and the specifically-defined time window based on the acoustic properties, it might provide information to clarify the possible functional roles for the different ERP components.
Abilities of sequence learning and rule learning: experimental evidence in newborn chicks (Gallus gallus)
Santolin, C. 1 , Sartori, D. 1 , RosaSalva, O. 2 , Regolin, L. 1 & Vallortigara, G. 2
1 Department of General Psychology, University of Padova, Italy
2 CIMeC, University of Trento, Italy
Sequence learning allows the encoding of the properties of event sequences, discriminating series composed of items presented in different order (e.g., ABA, AAB), while rule learning allows the recognition of a familiar structure underlying a sequence, even when composed of perceptually unfamiliar items. Human infants are capable of rapidly encoding serial order in event sequences, recognizing and generalizing abstract algebraic-like patterns, a process that might be crucial for language development. These abilities seem to be not specific to language because they emerge using both auditory stimuli (sequences of speech and non-speech sounds) and visual stimuli (sequences of dogs/cats pictures) and they have been uncovered also in non-human animals (including some recent and highly debated reports of sophisticate rule learning in avian species). This evidence is consistent with an adaptive role of sequence and rule learning in a variety of sophisticate non-linguistic cognitive operations. In the present study, we imprinted naïve newborn chicks on a video-sequence of arbitrary visual elements (e.g., geometric shapes ordered according to an ABA rule, creating sequences such as cross-circle-cross). Afterward, we tested chicks' choice between this familiar sequence and a structurally different one, composed of the same elements in altered order (e.g., AAB, cross-cross-circle). We also tested chicks' ability to generalize the structure of the sequence they had learnt to sequences composed of novel elements (unfamiliar geometric shapes). Chicks successfully generalized, showing preferential approach of the perceptually-novel sequence whose structure resembled that of the imprinting stimulus. In further experiments we extended our initial results, investigating whether this same generalization ability can be revealed also in a non-social context, employing a learning paradigm in which chicks are trained to find food hidden behind screens depicting one type of sequence.
Native and non-native phoneme perception in bilingual toddlers: A longitudinal study
Von Holzen, K. & Mani, N.
Georg-August-Universität Göttingen
During the first year of life, the language environment an infant experiences shapes their speech perception abilities. The resulting shift in sensitivity, known as perceptual reorganization, shows increased sensitivity for native language phoneme contrasts (Kuhl, Stevens, Hayashi, Deguchi, Kiritani, & Iverson, 2006) and decreased sensitivity for non-native contrasts (Werker & Tees, 1984). However, at this age, exposure to a second language (Kuhl, Tsao, & Liu, 2003; Conboy & Kuhl, 2011) or a brief familiarization period (Maye, Werker, & Gerken, 2002) can affect how infants perceive both native and non-native contrasts. Less is known about the impact of a changing language environment on speech perception later in development, after perceptual reorganization has taken place. To examine this, the current study tested German-English bilingual toddlers on their ability to discriminate between both a native (/ba/ - /pa/) and a non-native (Salish, / ` ki/-/`qi/) phoneme contrast, using the alternating/non-alternating procedure (Best & Jones, 1998). Any difference between alternating and non-alternating trials, regardless of direction, is indicative of discrimination. To observe how this ability is influenced by exposure to a second language (English) over time, toddlers were tested three times, spaced three months apart. As predicted, discrimination of non-native phonemes was not demonstrated at any time during the preschool year. Surprisingly, toddlers discriminated the native phoneme contrast at the beginning and end but not in the middle of the preschool year, suggesting a U-shaped pattern. These results provide evidence that hearing a second language (English) at preschool may have an influence, albeit temporary, on native phoneme discrimination. Monolingual German toddlers are currently being tested as controls to provide further support for this conclusion.
Flexible use of the mutual exclusivity assumption in early lexical acquisition: Evidence from eye tracking
Kalashnikova, M. 1 , Mattock, K. 2 & Monaghan, P. 1
1 Lancaster University
2 University of Western Sydney
Mutual Exclusivity (ME) refers to children?s tendency to establish one-to-one relations between basic-level labels and their referents. It is employed as a default word-learning strategy, but it remains unknown whether ME is present at the onset of lexical acquisition or is developed as infants? linguistic experience increases. The present study assessed early use of ME in tasks of exclusivity and lexical overlap in relation to infants? receptive vocabulary and through fixation pattern analyses in these tasks. Eighteen-month-old infants completed an eye-tracking preferential looking ME paradigm. In the Exclusivity condition, a speaker assigned a novel label to an unfamiliar object. This object appeared next to another unfamiliar object, and the speaker made a request using a different novel label, requiring infants to reason by exclusion. The Overlap condition was identical except that two speakers each assigned a novel label for the same unfamiliar object, and each used the label that they had introduced for the request stage, requiring infants to accept lexical overlap. Infants maintained ME in the Exclusivity condition, but their performance was related to receptive vocabulary scores (r=.546, p=.005). Their fixations followed a quadratic trend since initial fixations at target were interrupted by fixations at distracter followed by further fixations at target. Infants? performance was at chance in the Overlap condition showing fixations at target and distracter but no clear one-to-one or overlapping mappings. These findings indicate that ME does not operate at the onset of word learning but emerges as the infants? linguistic experience increases. Furthermore, our fixation pattern analyses suggest that in maintaining ME, infants do not show a default bias towards one-to-one mappings by directly assigning one novel label to one unfamiliar object. Instead, they actively evaluate all the information available in the word-learning situation to correctly map a word to its referent.
Making sense of familiarity and novelty responses in a single experiment
DePaolis, R. 1 & Keren-Portnoy, T. 2
1 James Madison Univesity
2 University of York
Typically, in a single auditory headturn preference paradigm (AHPP) study either a familiarity or a novelty response at the group level is taken to suggest that the infants have noticed some aspect of the auditory signal. In this presentation we argue that a mix of familiarity and novelty responses can be interpreted together in a single experiment as a coherent group response (see also Aslin, 2000; Dawson & Kidd, 2010). We used the AHPP to record looking times (LT) either to words likely to be familiar to the infants (e.g. baby or nappy) or to words unlikely to be known by any of the infants (e.g. berber or netter). This word form recognition task elicited both familiarity and novelty responses, as well as responses of no-preference, in 59 ten-month-old infants. Independent indices of lexical (CDI) and phonetic (repeated consonant production) advance were also collected. The infants who exhibited an extreme response, whether familiarity or novelty, were more advanced on these indices than the infants who showed no preference (p = .051 for lexical, p = .029 for production indices). Importantly, the groups do not differ on either of two measures that are proxies for general precocity (mean age; speed of processing, as measured by mean total LT). In a trial-by-trial analysis (with infants from the extreme group divided into extreme-novelty and extreme-familiarity), the groups exhibited different trajectories of looking preferences (see Figure 1). An ANOVA revealed an interaction of Time (the six trial pairs) and Group (extreme-novelty, moderate and extreme-familiarity): F[2,56] = 6.392, p < .001, ?2 = .275, supporting the assertion that the groups differ in their responses. We discuss these findings within the framework of the Hunter and Ames (1988) model, which suggests that a familiarity preference is succeded by a novelty preference over time.
Early lexical assessment: A comparison between British English and Italian children
Cattani, A. , Naftanaila, C. , Gunning, L. , Jenkins, H. & Floccia, C.
Plymouth University, UK
The knowledge of lexicon skills achieved by toddlers is generally acquired through parental questionnaires (e.g. CDI inventory, Hamilton et al., 2000) which require parents to fill-in the number of words understood and produced by their child. A new picture naming task, recently developed and validated for Italian children (Bello, Giannantoni, Pettenati, Stefanini & Caselli, 2012) which assesses four lexical subcomponents (comprehension and production of nouns and predicates), is now in the process of adaptation to an English version. The aim of this research is to assess the applicability of the newly adapted English version and to compare the developmental trend of the lexicon subcomponents of the two versions. The 120 picture cards were first shown to a group of adults to assess the name agreement and only 2 cards were replaced. Then, a cross-sectional sample of 92 children between the ages of 21, 24, 27, 30 and 36 months was assessed in the adapted English naming task together with the equivalent age CDI, SETK-2 word naming task and, according to the age, the British Scale Vocabulary (Dunn et al., 2009) or the Preschool Language Scale (PLS4, Zimmerman et al., 2009) to assess the test reliability. We analysed the percentages of correct responses and type of errors divided into semantically related, non-semantically related and non-responses of the picture naming task. While there is an indication of a slow start for the youngest English children, all older children reached the ceiling of words understood and produced like the Italian data. The analyses of the type of errors revealed that the higher percentages of non-responses in the errors found in English children might be an effect due to cultural differences of the children while performing the test.
Word recognition in Italian infants
Majorano, M. 1 & Vihman, M. 2
1 Department of Philosophy, Education and Psychology, University of Verona, Italy
2 Department of Language and Linguistic Science, University of York, UK
Experimental studies in three languages have shown that 11-month-old infants recognize word forms familiar from everyday life (for French, Hallé & Boysson-Bardies, 1994: for English, Vihman et al., 2004; and for Dutch, Swingley, 2005). The aim of the present study is to investigate word recognition in Italian infants using lists of untrained isolated words. Based on production studies of Finnish, where the onset consonant is often omitted, it has been suggested that medial geminates pull the child?s attention away from the initial consonant, even if the first syllable is accented, as is the case in both Finnish and Italian as well as in English. According to this hypothesis, a change to the first consonant should not block infant word recognition. In Experiment 1, 18 11-month-old infants were tested using the head-turn preference procedure. Two phonotactically similar lists (familiar and unfamiliar) of 12 words each were composed. Six words in each list had a geminate in medial position. The infants looked longer in response to the familiar than to the unfamiliar words [t(16) = 6.89; p <.001], replicating the base-line findings for the other languages. A second group of 18 infants participated in Experiment 2. In this case, all of the words in the two lists (familiar and unfamiliar) had a geminate in medial position. In both lists the first consonant of each word was changed. Preliminary results indicate that the infants respond with longer looking to the familiar words despite the change to the first consonant. Since onset-consonant change does not block word-form recognition in this case, the findings of Experiment 2 appear to support the hypothesis that medial geminates are sufficiently salient to draw infant attention away from the onset consonant, even in a language with a trochaic stress pattern.
Comprehension of prepositions in Norwegian
Aasheim Norås, S. & Vulchanova, M.
Norwegian University of Science & Technology, Trondheim, Norway
To comprehend and apply prepositions correctly in speech is one of many challenges children face during their acquisition process. Two of these prepositions found in Norwegian and English are discussed here: when do Norwegian children comprehend the use of 'on' and 'under' in Norwegian, compared to English children? This study investigated Norwegian children at ages 1;3, 1;6 and 2;0 on their comprehension of typical and atypical 'on' and 'under' situations by applying the preferential-looking paradigm and an eye-tracker. The English data which the Norwegian data will be compared with come from a British study by Kerstin Meints and colleagues; the current Norwegian study is a replication of the British study and results from both investigations will be compared and discussed. The findings indicate that the two prepositions are more or less understood in the same way during the same age stages across the two languages, but that Norwegian children seem to lag behind English children at 2;0; and that this might be due to the complexity of Norwegian 'on'; 'på'.
Receptive vocabulary and lexical access: A monolingual-bilingual comparison with the Computerized Comprehension Test
Poulin-Dubois, D. 1 , Legacy, J. 1 , Zesiger, P. 2 , Deák, G. 3 & Friend, M. 4
1 Concordia University
2 Université de Geneve
3 University of California, San Diego
4 San Diego State University
The Computerized Comprehension Task (CCT) has recently been proposed as an objective and direct measure of receptive vocabulary in the second year of life. However, while previous research suggests a strong convergent validity between the CCT and parental report measures such as the MacArthur-Bates Communicative Development Inventory: Words & Gestures (MCDI: WG), no research currently exists examining the validity of the CCT in assessing language development in bilingual children. The present study examines lexical access and receptive vocabulary development in 16-18 month-old monolingual and bilingual infants using the CCT. Accuracy and reaction time (delay to touch the target picture) on the CCT were collected in French-English bilingual infants (mean L2 exposure = 36.4%) and monolingual French-speaking children. Main findings indicate that there is no difference in accuracy (M =29.15% and M= 26.34%, respectively) or reaction time (M= 3194 ms and M = 3184 ms, respectively) when comparing bilinguals? performance on the CCT in L1 and L2. Bilingual infants' L1 accuracy however was lower than monolingual infants' accuracy (p = .001). They also exhibited response times that were slower than the monolingual infants (p = .01). However, when combining their total CCT L1 and L2 vocabulary, bilingual children showed a larger vocabulary than monolinguals. In terms of relations between vocabulary measures, there was a positive correlation between infants' total receptive vocabulary on the MCDI:WG and on the CCT, r = .325, p = .05. Current data replicate those of a recent study showing no differences in lexical access in 24-month-old bilinguals and monolinguals.These results suggest that the CCT and MCDI provide partially converging measures of receptive vocabulary in bilingual as well as monolingual infants.
Perceptual distance predicts success in early word learning across regional accents
Escudero, P. 1 , Best, C. T. 1 & Kitamura, C. 2
1 MARCS Institute, University of Western Sydney
2 School of Social Sciences and Psychology , University of Western Sydney
Children under 17 months have substantial difficulty in learning to associate picture referents to novel words that form phonologically minimal pairs, i.e., differing in a single consonant (e.g. 'bin' versus 'din'). However, they can associate picture referents to some minimal vowel pairs, namely 'deet' versus 'dit', suggesting early word learning is affected by the type of phoneme contrast involved. In previous studies the novel words were spoken in the infants? native regional accent (often Canadian or American English). We examine whether accent differences in the production of novel minimal vowel pairs affect early word learning success. An experience-based attunement perspective would predict higher levels of success for native than non-native accents, while an acoustic/articulatory perspective would predict that the accent with the largest phonetic differences would yield more success regardless of nativeness. Australian English 15-month-olds (N = 48) were presented with 'deet', 'dit' and 'doot' spoken by either an Australian English (native accent) or a Canadian English (non-native) female. During the learning trials, toddlers saw a moving object on the screen and heard 'deet'. After habituation, they were presented with three test trials, all of which showed the same moving object but played either the same word ('deet') or different words ('dit' and 'doot'). Test trial order was counterbalanced across infants. A repeated-measures ANOVA (Word) x Accent x Test Order revealed a main effect of Word, qualified by a Word x Accent interaction. Australian toddlers exposed to Canadian stimuli looked longer at the switch ('dit' and 'doot') than at the same trials ('deet'), while no word differences were found for toddlers exposed to Australian stimuli. Thus, accent differences in vowels indeed affect early word learning. Moreover, accent-specific acoustic/articulatory properties and how they compare to the learners? phonetic categories appear to be stronger predictors of word learning success than language experience.
Exploring the effects of linguistic experience on disambiguation
Lozano, I. & Bosch, L.
University of Barcelona
In word learning contexts disambiguation has been defined as a heuristic useful in linking novel labels to likely referents. Evidence of disambiguation has been shown in monolinguals at around 18 months of age by increased tendency to associate novel labels to novel objects rather than to familiar ones. Disambiguation is less often used by toddlers growing up bilingual or trilingual, suggesting that its emergence depends on variables such as linguistic experience (Byers-Heinlein & Werker, 2009). The present research aims at further exploring the effects of linguistic experience on the emergence of this heuristic by analyzing the effects of 1) the phonological proximity between the bilingual?s two languages (high amount of cognates), and 2) age of onset of exposure to the second language. Three groups of 18-month-old toddlers (N= 60 in total) were tested in a looking-while-listening paradigm: A Spanish monolingual group and two Spanish-Catalan bilingual groups differing in age of onset of bilingual exposure (simultaneous exposure in home environments, versus a later onset of L2 exposure through attendance to day-care centers). Bilinguals were expected to differ from monolinguals in the use of disambiguation. However, phonological similarity in bilingual?s word forms can affect the use of this heuristic, with higher presence of disambiguation in these groups, differing from previous results obtained in bilinguals exposed to languages with reduced number of cognates. If age of onset of bilingual exposure also modulates the use of this heuristic, differences might be found between the bilingual groups, with the late onset group approaching monolinguals in their use of disambiguation. Preliminary analyses have revealed a significant difference between bilinguals? and monolinguals? disambiguation patterns and no differences between both bilingual groups. Age of onset of L2 exposure and cognate status of words do not seem to modify the expected pattern of results in bilinguals.
Developmental sequence and overgeneralization patterns of -er and more-
Suh, J. , Lee, S. & Jun, J.
Hankuk University of Foreign Studies
This research investigates the comparative expressions to determine children's inflection development in English. Children acquire syntactic knowledge before they learn the concepts of morphological inflection (Moshe Anisfield, 1984). With various grammatical functions, 'more' is one of the most common words. The frequency and actual utterances of 'more + adjective' were categorized and contrasted with those of 'adjective + er,' which appear later and lower in frequency. Furthermore, it will be compared with the frequency of the mother and adjectives by itself. To examine the characteristics, the original data of CHILDES (http://childes.psy.cmu.edu/) have been modified to extract the language use. This study deals with 3712 (USA: 2,357 files - 21,444 types, 1,726,002 tokens; UK: 1,355 files - 9,829 types, 753,225 tokens) files from children ages 1 to 7, which were tagged with morpho-syntactic information. So far most of the studies have focused on the inflectional verb development (Brown, 1973; Gleason & Ratner, 1998), but this concentrates on the patterns of comparatives and the developmental progression. The relation of overgeneralization in comparison are compared to Marcus et al.(1992)'s inflections on past tense verbs and the possible factors are discussed. It is expected that input frequency influences the development of irregular forms of comparatives in that children often misuse 'little' as 'littler' compared to 'more' and make more mistakes on using superlative than comparative.
The acquisition of gender and plural morphology in Italian-learning 12- to 24-month-olds
Ferry, A. , Nespor, M. & Mehler, J.
Scuola Internazionale Superiore di Studi Avanzati
Research demonstrates that infants begin learning words during the first year of life (Bergelson & Swingley, 2012). Yet, learning a language requires more than simply building a vocabulary; infants must also learn the morphological rules and regularities of their language. One such regularity, the singular/plural distinction has been studied extensively in English-learning children. Those findings demonstrate that children do not acquire the singular-plural distinction until approximately 24 months, when they begin to produce and comprehend the distinction (Wood et al., 2009). Yet, in English, the singular/plural marking is not as robust as in other languages. Italian uses robust inflection morphology marking both gender and plurality of the noun in the same manner. The distinction is marked as a necessary component of the noun, in the article, and in adjectives associated with the noun (e.g., il ragazzo alto; i ragazzi alti; la ragazza alta; le ragazze alte). Here, we ask whether this robust marking highlights the singular/plural distinction earlier for Italian-learning infants than for English-learning infants. We compare this to the acquisition of gender morphology, which is marked the same way but infants have been shown discriminate natural gender categories early in development (Leinbach & Fagot, 1993). We tested 12- to 24-month-olds in a looking while listening task. During each trial infants viewed two images that differed either in gender or in plurality (e.g., one man versus two men). Sentences directed the infants to look at one of the images (e.g., ?Guarda il ragazzo!) while their looking time was recorded. Results indicate that gender morphology is acquired early in development. The acquisition of the singular/plural distinction emerges earlier in Italian infants than previously found in English-learning infants. The results are discussed in terms of the acquisition of morphology and the role of language in acquiring the singular/plural conceptual distinction.
Cross-modal matching of onomatopoetic words and visual textures by Japanese 1-year-old infants
Ikeda, A. 1, 2 , Kobayashi, T. 2 & Itakura, S. 1
1 Kyoto University
2 NTT Communication Science Labs
Recent preferential-looking studies have shown that 12-month-old infants can map particular nonsense words (e.g., 'kiki') to particular shapes (e.g., angular shapes) as well as adults, suggesting that infants' sensitivity to non-arbitrary sound-meaning relationships (i.e., sound symbolism) precedes language acquisition (e.g., Starr & Brannon, 2012). Is this sound-meaning correspondence similarly applicable to other properties? This study focused on visual-tactile properties to investigate whether 1-year-old infants can map words to particular visual textures using Japanese onomatopoetic words, which have rich visual-tactile expressions involving sound symbolism. Fifty-five Japanese-learning 16-month-olds (mean=16.0, range=12.1-20.9) were tested using a preferential-looking paradigm conducted in six trials (three pairs: tsuru-tsuru [smooth] vs. toge-toge [spiky], tsubu-tsubu [pebbly] vs. chiku-chiku [pricking], fuwa-fuwa [fluffy] vs. gotsu-gotsu [craggy]) with familiarization and test phases. In the familiarization phase, paired pictures were shown side by side on a black background with audio stimuli ('Look! What is [it] like?'). In the test phase, the same pair was shown with an onomatopoetic word that was congruent with one picture (e.g., 'Which is tsuru-tsuru?'). The test phase was repeated while reversing the left-right arrangement of the pictures. We calculated the proportion of looks to the congruent picture during the test phase. The results show that the infants looked significantly longer than chance at the congruent picture when presented with a pair of tsuru-tsuru and toge-toge (p<.01), but they failed when presented with the other pairs. Furthermore, a frame-by-frame analysis of time-shift looking at the pair of tsuru-tsuru and toge-toge revealed that they spent more time looking at the congruent picture only during the target sentences. That is, they shifted their attention in response to the particular audio stimuli. These results provide evidence for infants' sound-meaning mapping even before learning onomatopoetic words about visual-tactile properties and suggest that their reaction to sound-symbolic words may be instant and momentary.
The role of parental input and the frequency of use of verbs in the construction of the earliest multi-word expressions
Città, G.
Doctoral School in Cognitive Science - University of Messina
The aim of this work is to investigate the role of linguistic inputs and the frequency of use of verbs in the process of production of child first multi-word expressions (21-24 months of age). Through the computational analysis of fragments of linguistic corpora, extracted from the CHILDES database, we analyzed two key factors: (a) the set of individual linguistic inputs, the Child Directed Speech (CDS), that is the language environment within which the child lives and the crucial element affecting progressively the age of acquisition, the rhythm of growth, the size of the vocabulary and the formation of abstract categories of language; (b) the frequency with which verbs occur in the CDS. The analysis revealed that, in the flow of the CDS, the frequency of use of verbs, in specific contexts and in specific positions within the sentence, is an element of primary influence. It increases the production of simple and rudimentary composition of words that have verbs in salient position. Furthermore, the analysis that we conducted discriminates the use of the individual morphological variants of the verb in the construction of propositions and shows that the child's patterns of use are partially different from the patterns of use in the CDS. A comparison between the frequency of use of morphological variants of a sample verb (to look) at high frequency, both in the child speech and in the CDS, shows that, despite using more frequently the morphological variant most commonly used by adults in the CDS, children use it in a different way: within the sentence, they use it in different positions.
Statistical learning supports lexical development: Roles for predictability and lexical organization
Lany, J.
University of Notre Dame
Experience with statistical regularities plays a key role in word learning: Tracking sequential statistical structure (e.g., determiners typically precede nouns) promotes learning grammatical categories (Gerken et al., 2005), as well as learning word-referent mappings (Lany & Saffran, 2010). The current experiment tested the extent to which statistical learning supports vocabulary development by increasing the predictability of speech sequences and/or by promoting lexical organization (i.e., forming connections among similar words). Infants (22-month-olds, N=84) listened to an artificial language in which X- and Y-word categories were reliably marked by statistical cues: disyllabic X-words followed a-words, while monosyllabic Y-words followed b-words. Infants were then trained on pairings between words and pictures. In the Correlated-Semantics condition, X-words referred to pictures from one semantic category (e.g., animals) and Y-words referred to pictures from another category (e.g., vehicles). Thus, infants could benefit from predictability (experience with the aX bY pattern) and lexical organization (experience with statistical-semantic correlations). In the Uncorrelated-Semantics condition, X- and Y-words referred to pictures from both semantic categories, and thus infants could benefit from predictability alone. If infants are affected by lexical organization, then performance should differ across conditions. Moreover, tracking statistical-semantic correlations may be challenging, and infants' ability to do so may depend on their native-language proficiency. Performance across conditions did not differ overall. However, within the Correlated-Semantics condition more linguistically advanced infants shifted their gaze to the Target more rapidly and were more likely to maintain their gaze to the labeled picture than less-advanced infants. In the Uncorrelated-Semantics condition, infants' performance was unrelated to language proficiency. These findings suggest that statistical regularities support word learning by making speech predictable. As infants develop, they also benefit from statistical-semantic correlations, suggesting that statistical learning plays an increasingly important role in word learning across development.
Consonant onset, vowel and consonant coda mispronunciation detection in 20 month-old English-learning infants
Turner, J. , Delle Luche, C. , Gunning, L. , Jenkins, H. & Floccia, C.
Plymouth University
Adult studies show that consonants and vowels play different roles in language processing, with consonants being favourably involved in lexical processing, and vowels more important for grammar and prosody (Nespor et al, 2003). Intermodal Preferential Looking (IPL) studies with English-speaking infants as young as 14-month-olds show that infants are equally sensitive to consonant and vowel mispronunciations in familiar words (Mani & Plunkett, 2007; Swingley & Aslin, 2002). However, in word-learning tasks French-learning infants are only sensitive to consonant contrasts and not vowel contrasts (14 months in Zeziger & Johr, 2011). In a similar task, a recent study with English toddlers aged 16 to 23 months suggests that they might be particularly sensitive to coda consonants in CVC items (Floccia et al., under review). This could explain the discrepancy between the English and the French results, as mispronunciation detection studies in English toddlers comparing consonants and vowels have always focused on onset consonants rather than codas. Thus, the matter of a differential role of consonants and vowels in lexical development in English-learning infants is not so clear-cut. Using a classic IPL procedure and stimuli similar to Mani and Plunkett (2007), we presented 20 month old infants with two pictures of familiar objects (e.g. ball, cat) along with a correct or an incorrect pronunciation of the monosyllabic target word. Mispronunciations were created by manipulating the phonemic class (consonant/vowel) and the consonant location (onset/coda). Halfway through data collection, the results currently show that infants look longer at familiar objects when presented with the correct pronunciations compared with mispronunciations, with no asymmetry between onset consonant, vowel and coda consonant mispronunciation detection. If these results were to be confirmed, they would strengthen the findings that English-learning children do not show a consonant bias in word processing, irrespective of the location of the mismatching target phoneme.
Differentiation and refinement of articulatory coordination in the development of vowel production
Oohashi, H. 1, 2 , Watanabe, H. 1 & Taga, G. 1
1 Graduate School of Education, The University of Tokyo
2 The Japan Society for the Promotion of Science
Development of speech production is attributed to changes in anatomy of articulatory organs and motor coordination. Although previous studies have revealed changes in acoustical features of speech sounds, limited information is available as to articulatory states during speech production in children. Here, we applied an acoustic-to-articulatory inversion technique to estimate articulatory states from vowel sounds produced by Japanese children aged from 6 to 60 months. We assumed seven articulatory parameters, that is, jaw, tongue position, tongue shape, tongue tip, lip aperture, lip protrusion and larynx height, and two scaling factors determining a length of oral and pharyngeal cavities. Three lower spectral peaks of each vowel were used to calculate the parameters of articulatory states by using pseudo-inverse Jacobian matrix that maps an acoustical space to an articulatory space. The stepwise discriminant analysis with vowel categories as the predictor and the estimated articulatory parameters as the independent variables showed that the development of children's articulation went through three stages. At the first stage (6-9 months of age), coordination of the tongue position and the lip aperture formed three vowels, that is, front, back and central vowels. At the second state (10-17 months of age), recruitments of jaw and tongue tip enabled differentiation of three vowels into the five Japanese vowels. At the third stage (18- months of age), coordination among articulatory controls including tongue shape produced more distinct vowels specific to Japanese. Our result suggests that initial vowel articulations are founded on coordination between lip and tongue, and that the articulations are differentiated and refined into those adjusted to the native language along development. In particular, the changes in the recruitment of tongue parameters are critical in the development of vowel production.
Supporting early vocabulary development: What sort of responsiveness matters?
McGillion, M. 1 , Herbert, J. 1 , Pine, J. 2 , Keren-Portnoy, T. 3 , Vihman, M. 3 & Matthews, D. 1
1 University of Sheffield
2 University of Liverpool
3 University of York
Maternal responsiveness is a multi-dimensional construct encompassing ?prompt, contingent and appropriate reactions? to infant communicative acts that has been positively related to language outcomes (Bornstein & Tamis-LeMonda, 1989). A substantial body of research has explored different aspects of responsiveness, such as its timing, syntax and semantic content (Tamis-LeMonda, Bornstein & Baumwell, 2001). However, perhaps because of the many ways in which responsiveness can be operationalised, there is currently a lack of consensus around what type of responsiveness is most helpful for later language development (Masur, Flynn & Eichorst, 2005). The present study aimed to deconstruct the responsiveness construct by considering several dimensions on a single cohort to determine which measure best predicts early expressive vocabulary development. Drawing on an existing longitudinal dataset of naturalistic video-recorded dyadic interaction, maternal verbal responsiveness with their 9.5 month old infants was considered in relation to expressive vocabulary at 18months. Three dimensions of responsiveness were operationalised: Semantic responsiveness: maternal utterances that referred to the infant?s focus of attention; Temporal responsiveness: maternal utterances occurring within two seconds of an infant vocalisation; Semantic and temporal responsiveness: semantically appropriate utterances occurring within two seconds of an infant vocalisation. Mothers who responded more often to their infant?s vocalisations tended to do so in a semantically appropriate manner, however, only utterances that were both semantically appropriate and temporally linked to an infant vocalisation were related to later expressive vocabulary development. Regression analysis revealed that prompt and semantically appropriate maternal language at 9.5 months was a significant predicator of expressive vocabulary at 18months. This finding underlines the dyadic nature of the responsiveness construct and draws attention to the importance of early vocalisations as potential signals of infant attention. Further coding and analysis will consider the relationship between these responsiveness measures and other aspects of infant directed speech.
Segmenting speech: Comparing the time course of word recognition in monolingual and bilingual infants
Junge, C. 1 , Ebanks, N. 2 , Neophytou, E. 2 & Mills, D. 2
1 University of Amsterdam, Amsterdam, the Netherlands
2 Bangor University, Bangor, United Kingdom
The ability to detect words in continuous speech is vital for building a vocabulary, as it is an important early predictor of language development (Newman, Bernstein Ratner, Jusczyk, Jusczyk & Dow, 2006; Junge, Kooijman, Hagoort & Cutler, 2012). Multiple cues in the native language, such as stress patterns, can be used to segment words from speech (Jusczyk, Houston & Newsome, 1999). Segmentation abilities develop between six and 12 months, and have been tested both behaviorally and electrophysiologically. Behavioral research on bilingual infants showed that French-English 8-month-olds are not delayed in the emergence of word segmentation abilities (Polka & Sundara, 2003). However, little is known about whether bilinguals differ from monolinguals in their time course of word recognition in continuous speech. The present study therefore examined how learning simultaneously two rhythmically different languages - English and Welsh - affect 10-month-olds' speed of word recognition in running speech. Infants were first presented with 8 different utterances containing the same trochaic word (e.g., 'eagle') before recognition was tested, within novel utterances, by comparing event-related-potentials to novel tokens of the familiarized words with those of unfamiliarized control words (i.e. 'eagle' vs. 'lemming'; counter-balanced across infants). Preliminary data show that both monolingual (n=10) and bilingual infants (n=10) distinguished familiarized words from control words, as indicated in both groups by an increased negativity for familiarized words (N200-500; Kooijman, Hagoort & Cutler, 2005). Furthermore, the onset of the word recognition effect in both groups was comparable, from 220 ms onwards. This indicates that by 10 months, bilingual infants pick up novel words in running speech as quickly as monolingual infants. We are currently also testing seven-month-olds to see whether this pattern holds.
Twelve-month-olds' failure to associate multiple novel words with objects in a habituation switch task
Kobayashi, T. 1 & Murase, T. 2
1 NTT Communication Science Labs
2 Shimane University
Recent habituation studies have shown that English-learning 12-month-olds can rapidly associate one label with one object, suggesting that their word-learning mechanism works efficiently from the earliest stage of lexical development (MacKenzie et al., 2012). However, infants do not necessarily hear one label for one object in particular linguistic environments. For example, Japanese children often hear two different labels from infant- and adult-directed speech (e.g., wanwan and inu for a dog), and bilingual children often hear two labels from different languages for one object. An unresolved issue is thus how infants process two labels for one object. This study focuses on Japanese children with extensive multiple-labeling experience to examine whether 12-month-olds can rapidly associate two novel words with one object in a habituation paradigm with a switch design (Werker et al., 1998). Monolingual Japanese-learning 12-month-olds (N=48, mean age=12.4, range: 12.0-12.9) were habituated to two computer-animated scenes in which one novel object was paired with two novel labels embedded in sentence frames (e.g., Look, [it]'s a ronron. [It]'s a yamitsu). While the infants in one condition heard two novel words phonologically characteristic of an infant-directed speech (IDS) word (ronron, tenten) and an adult-directed speech (ADS) word (tawa, yamitsu), those in the other condition heard only two novel ADS words (tawa, yamitsu, byoku, reroni). After being habituated to the two scenes, they were given two 'same' trials and two 'switch' trials. In the test trials, only one label was presented to assess each word-object association. Results show that the 12-month-olds in both conditions looked equally long in the 'same' and 'switch' trials, indicating that 12-month-olds cannot yet learn labels. A follow-up experiment with 12-month-olds (N=24) found that they formed word-object associations when one object was paired with one ADS word. Together, these findings suggest that presenting 12-month-olds with multiple labels hinders word-object associations.
What are the units? Factoring morphological statistics into the learning of verbs
Willits, J. & Jones, M.
Indiana University
Statistical learning mechanisms seem to be important to many aspects of language acquisition. But what are the units over which statistical learning operates? Phonemes? Syllables? Morphemes? Words? Claims about the sufficiency of statistical learning mechanisms rest on which units are being used. Previous research has shown that infants don't recognize verbs in fluent speech until 10-13 months, compared with 7.5 months for nouns. This makes sense given their relative distributional statistics; noun statistics provide better segmentation cues. However, this statistical difference goes away in the presence of inflections like -ing. In accordance with these morphological-level statistics, Willits, Seidenberg, & Saffran (2009) found that 7.5-MO infants presented with verbs in -ing contexts do show evidence of recognizing verbs. In the present work we present an analysis of infant and toddler verb-learning based on statistics in child-directed speech, with the critical manipulation being how verb inflections are treated in these analyses. The analysis incorporates a number of statistics about verbs, including their ease of articulation, word frequency, repetition frequency, and contextual diversity. All of these factors are found to significantly contribute to predictions about verbs' difficulty of acquisition. Multiple regression analyses also showed that the importance of these statistics changes drastically from month zero to month thirty. Frequency, repetition, and discourse situation contextual diversity play important roles at younger ages, and lexical contextual diversity plays a critical role at older ages. Critically, all of these statistics provide a better account of verbs' age of acquisition if inflections are broken off of verbs and counted as separate units in the analysis. Together, these studies help answer questions about which factors contribute to vocabulary development. They also address the scope of statistical learning-based theories of acquisition, suggesting that tracking morpheme-level statistics plays a critical role in infants' acquisition of their native language.
The influence of tonal contexts in Mandarin speech production: A motor control perspective
Huang, T. 1 , Liu, Y. 1 & Chen, J. 2
1 Department of Athletic Performance, National Taiwan Normal University
2 Department of Chinese as a Second Language, National Taiwan Normal University
The tones in spoken Mandarin provide an extra element to distinguish words in addition to the syllables. Speech production involves a major part of controlling the motor system to perform the appropriate tone-syllable combinations. However, there has been little research on the Mandarin tone variations from the motor control perspective. The purpose of the study was to examine the kinematic characteristics of different tones in performing different bi-tonal combinations of Mandarin. Six adult male native speakers of Taiwanese Mandarin performed 16 different reiterant speech tasks (replacing the original syllable of the word with /ba/) where one of the 16 possible bi-tonal combinations of the four tones was used and repeated for 5 times to construct the task. Each participant performed 10 trials for each task. Kinematics data on the lips were collected using a high speed camera (200 fps) and a digitizing software. Amplitudes and durations of the lower lip movements were analyzed with repeated measure ANOVAs on the individual data to examine the influence of the tonal contexts to the motor characteristics of the reiterant speech. The results show that the first word of the disyllabic words has greater influence to the following word than vice versa for amplitude and duration measures in both open and close gestures. In terms of the acoustic characteristics of the tones, the falling tones (tone 3 and 4) are influenced by the preceding tones with a higher percentage than is the rising tone (tone 2). These results provide bases to the further understanding of the influence of the tonal contexts in Mandarin speech production. Future work will continue to examine the influence of the tonal contexts to Mandarin speech production on non-native speakers from a motor control perspective.
Cross-linguistic universality of word acquisition ages in comprehension and production
Minami, Y. , Kobayashi, T. & Sugiyama, H.
NTT
Gentner (1982) analyzed the contents of children?s productive vocabularies in six languages, providing the pioneering work in cross-linguistic child development. Since then, many studies of linguistic universality have been investigated in this field (Bornstein et al. 2004). However, no study has investigated cross-linguistic universality in word-comprehension and -production days calculated directly from a CDI database. Therefore, this study investigates cross-linguistic universality using English and Spanish Lex2005 CDI database (Dale et al., 1996) along with our Japanese CDI database (from 1,699 toddlers). We defined the word-comprehension and -production days as the days when 50% of the children comprehend and produce a word, respectively. We determined these days by approximating rate curves of the children who comprehend and produce the word with two logistic functions, setting these functions to 0.5 and solving them by the Newton method. The correlations of comprehension and production days between two languages were calculated category on category and word on word. Casselli?s criterion (1999) was used to set the categories. We determined word-on-word correspondences only for verbs common to both CDI databases, looking them up in ordinary dictionaries. For category-on-category comparison of the comprehension days, English-Japanese, English-Spanish, and Japanese-Spanish correlation coefficients were 0.77, 0.80 and 0.64, respectively (p<0.01). For category-on-category comparison of the production days, English-Japanese, English-Spanish, and Japanese-Spanish correlation coefficients were 0.73, 0.76 and 0.73, respectively (p < 0.01). For word-on-word comparison of the comprehension days, English-Japanese, English-Spanish, and Japanese-Spanish correlation coefficients were 0.73, 0.71 and 0.62, respectively (p<0.01). For word-on-word comparison of the production days, English-Japanese, English-Spanish, and Japanese-Spanish correlation coefficients were 0.57, 0.61 and 0.63 (p < 0.01). These results confirmed there are strong correlations in word-comprehension days and word-production days as seen through both category-on-category and word-on-word comparisons. This suggests that there is cross-linguistic universality of word acquisition ages in comprehension and production.
The role of language context and language dominance in the development of bilingual infant babbling
Molnar, M. & Bease-Berk, M.
Basque Center on Cognition, Brain and Language (BCBL). Donostia. Spain
Cross-linguistic studies have illustrated that babbling of monolingual infants varies as a function of native language background by the end of the first year (e.g., Boysson-Bardies et al., 1992). These findings have been interpreted within the babbling drift hypothesis, which theory proposes that monolingual infant babble progressively resembles the phonetic and prosodic characteristics of the linguistic environment. However, the development of babbling in bilingual infants, who have two native languages, is unclear. At the perceptual level, bilingual-to-be infants have been shown to differentiate two rhythmically similar native languages by 4-months of age (Spanish and Basque: Molnar et al., 2011; Spanish and Catalan: Bosch and Sebastian-Galles, 2001); still, it remains controversial whether preverbal bilingual infants develop differentiated language systems in their productions during the babbling period (Poulin-Dubois and Goodz, 2001). In order to investigate the developmental characteristics of bilingual babbling, in the current study, we recorded babbling samples from bilingual and monolingual infants of Basque and Spanish at 12-, and 16-months of age. Bilingual infants (Basque-dominant and Spanish-dominant) were recruited from one-parent-one-language families, and their babbling samples were recorded in two separate sessions corresponding to the two linguistic contexts (Basque and Spanish). Linguistic context of each session was induced by the language-appropriate caregiver interacting with the infant. Acoustic analysis of adult native productions of Spanish and Basque have indicated that the two languages differ in terms of the overall temporal distributions of vowels and consonants (Molnar et al., 2011). Based on these acoustic features, differences across monolingual and bilingual (by language context and language dominance) infant productions have been observed.
Eighteen-month-olds activate semantic information in the auditory modality
Delle Luche, C.
University of Plymouth
Children build up a lexicon progressively, and there is still debate whether early words emerge as semantically isolated lexical islands or are already linked together by semantic relationships. Evidence to date reported word-to-word priming in children as young as 21 months of age, but results with younger toddlers is not very conclusive. The present experiment evaluates whether the lexicon of 18-month-olds already contains semantic relations between spoken words in an all-auditory Head-Turning Paradigm. Children are presented with lists of words belonging to the same semantic category (e.g., consecutive names of animals), or to a mix of two categories (e.g., names of clothes and body parts mixed together). If words activate other semantically related words, children should display more attention, or familiarise slower, to single category lists over mixed categories lists. Children heard two blocks of 6 lists of words, starting for example with 6 Same Category lists (3 lists of animal names and 3 lists of food items) and ending with 6 Mixed Category lists (6 lists of body parts and clothes mixed together). Listening times towards each type of list was recorded. With 36 infants successfully tested, we found that children listened marginally longer to the Same Category lists overall (F(1,33) = 4.00, p = .053, ?2 = 0.10). When restricting the analysis to the 25 children who know at least 50% of the words in each category, this effect is significant (F(1,23) = 6.70, p = .016, ?2 = 0.22), with 12.64s of average listening time for the Same Category (SD = 0.81) compared to 10.60s in the Mixed Category (SD = 0.89). These findings point to the existence of semantic relations between words in the early lexicon, complementing earlier reports of semantic links at 21 months (e.g. Arias-Trejo & Plunkett, 2009).
Isolated words in input to infants: A critical wedge?
Keren-Portnoy, T. & Vihman, M.
University of York, UK
The importance for early lexical development of hearing words in isolation as compared with having to segment them from running speech is hotly debated (e.g., Brent & Suskind, 2001, Gerken & Aslin, 2005). Although several studies have shown that infants can segment words from continuous speech, this may not be the primary way that infants learn their first words. Brent and Siskind (2001) demonstrated that isolated-word frequency in input speech better predicts later word use than word frequency overall, while Junge et al. (2012) found more reliable 10-month-old learning based on isolated words than words heard in passages. We tested the comparative effects of input word use in isolation and in sentence-final position on 12-month-olds? word learning. A picture book with animals whose names were unlikely to be familiar to infants (e.g., condor, dassie, pudu) was read by parents to their infants over 3 weeks. The animal names appeared in the book either in isolation or sentence-finally. Each infant was then tested using the Head-turn Preference procedure on (i) trained words heard in isolation, (ii) trained words heard sentence-finally, and (iii) phonotactically-matched untrained words. Results show a tendency (F = 2.936, df = 1.4, p = .09) for group differences in mean looking times. When comparing each two word types in pairs, using relative percentage of looking to each type, we found greater attention to words heard in isolation than to those heard in sentences (p = .02), a tendency towards greater attention to words heard in isolation over untrained words (p = .07) and no significant difference between attention to words heard in sentences vs. untrained. We are now running a follow up experiment, which will allow more clearly interpretable pairwise comparisons. Our study provides further evidence that isolated words may afford a ?critical wedge? into the speech stream.