English Infants Research Articles

Early phonetic learning without phonetic categories: Insights from large-scale simulations on realistic input

Before they even speak, infants become attuned to the sounds of the language(s) they hear, processing native phonetic contrasts more easily than nonnative ones. For example, between 6 to 8 mo and 10 to 12 mo, infants learning American English get better at distinguishing English and [l], as in "rock" vs. "lock," relative to infants learning Japanese. Influential accounts of this early phonetic learning phenomenon initially proposed that infants group sounds into native vowel- and consonant-like phonetic categories-like and [l] in English-through a statistical clustering mechanism dubbed "distributional learning." The feasibility of this mechanism for learning phonetic categories has been challenged, however. Here, we demonstrate that a distributional learning algorithm operating on naturalistic speech can predict early phonetic learning, as observed in Japanese and American English infants, suggesting that infants might learn through distributional learning after all. We further show, however, that, contrary to the original distributional learning proposal, our model learns units too brief and too fine-grained acoustically to correspond to phonetic categories. This challenges the influential idea that what infants learn are phonetic categories. More broadly, our work introduces a mechanism-driven approach to the study of early phonetic learning, together with a quantitative modeling framework that can handle realistic input. This allows accounts of early phonetic learning to be linked to concrete, systematic predictions regarding infants' attunement.

Quantifying child directed speech cross-culturally across development

Child-directed speech (CDS) influences language development (e.g., Golinkoff et al., 2015), but varies across cultural and demographic groups (Hoff, 2006). Recent work examining speech heard by North American English (NAE) infants found an increased proportion of CDS with age (Bergelson et al., 2018). Quantity of CDS remained relatively constant across age, while quantity of adult-directed speech (ADS) decreased. We replicate these findings using a different methodology, and expand them to include other language communities. Our data come from daylong audio recordings of 58 children ages 2–36 months from the ACLEW dataset (Bergelson et al., 2017; 30 children acquiring NAE, 10 UK English, 8 Argentinian Spanish, and 10 Tseltal/Mayan). Ten randomly selected 2-min segments (Tseltal: nine 5-min segments) from each child were annotated for speaker gender, age (child or adult), and addressee for each utterance. We calculated the minutes per hour of CDS, ADS, and all speech. Preliminary analyses find high variability in overall language input across individuals, age, and culture, and partially replicate the Bergelson et al. (2018) pattern of results. Ongoing annotation will permit finer-grained analyses of sub-group differences. Further analyses will examine the influence of factors such as speaker gender, number of speakers, and maternal education.

An order effect in English infants’ discrimination of an Urdu affricate contrast

An order effect was found in English infants’ discrimination of an Urdu contrast. In Experiment 1 7- and 11-month-old English infants were tested on the Urdu contrast between the affricates /tʃʰ/ and /tʃ/. The order of presentation was counterbalanced: At each age half the infants were habituated to the aspirated and tested on the unaspirated affricate, the other half habituated to the unaspirated and tested on the aspirated. As expected, younger infants discriminated the contrast whereas older infants did not, showing the expected decline in discrimination. Order of presentation seemed to affect the older infants’ response. Experiment 2 tested the order effect directly. The results showed no asymmetry in the performance of 7-month olds but clear asymmetry in that of 11-month-olds, who discriminated the contrast only when the non-English-like aspirated affricate was presented first. Experiment 3 tested adult native-speakers of both Urdu and English. Although the English listeners showed a reduced sensitivity to the contrast, there was no effect due to order of presentation of the stimuli in either adult group. The finding of an asymmetry in the infants suggests that infants’ perceptual narrowing for speech sounds may be a more complex phenomenon than has generally been assumed.

"Mummy, keep it steady": phonetic variation shapes word learning at 15 and 17 months.

Fifteen-month-olds have difficulty detecting differences between novel words differing in a single vowel. Previous work showed that Australian English (AusE) infants habituated to the word-object pair DEET detected an auditory switch to DIT and DOOT in Canadian English (CanE) but not in their native AusE (Escudero etal., ). The authors speculated that this may be because the vowel inherent spectral change variation (VISC) in AusE DEET is larger than in CanE DEET. We investigated whether VISC leads to difficulty in encoding phonetic detail during early word learning, and whether this difficulty dissipates with age. In Experiment 1, we familiarized AusE-learning 15-month-olds to AusE DIT, which contains smaller VISC than AusE DEET. Unlike infants familiarized with AusE DEET (Escudero etal., ), infants detected a switch to DEET and DOOT. In Experiment 2, we familiarized AusE-learning 17-month-olds to AusE DEET. This time, infants detected a switch to DOOT, and marginally detected a switch to DIT. Our acoustic analysis showed that AusE DEET and DOOT are differentiated by the second vowel formant, while DEET and DIT can only be distinguished by their changing dynamic properties throughout the vowel trajectory. Thus, by 17 months, AusE infants can encode highly dynamic acoustic properties, enabling them to learn the novel vowel minimal pairs that are difficult at 15 months. These findings suggest that the development of word learning is shaped by the phonetic properties of the specific word minimal pair.

Bilingual Infants Demonstrate Advantages in Learning Words in a Third Language.

Prior research suggests that bilingualism may endow infants with greater phonological flexibility. This study investigated whether this flexibility facilitates word learning in additional languages (n=96). Experiment 1 compared 18- to 20-month-old monolingual (English) and bilingual (English/Mandarin) infants on their ability to learn words distinguished by click consonants from a Southern African language, Ndebele. English-Mandarin bilingual infants were sensitive to Ndebele click contrasts, but monolingual English infants were not. In Experiments 2a and 2b, we investigated whether enhanced bilingual sensitivity extended to analogous nonlinguistic labels: hand claps and finger snaps. Although discriminated by infants, neither group distinguished words labeled by hand claps and finger snaps. Results suggest that bilingual infants' sustained openness to non native contrast may facilitate the uptake of words in distant languages.

Listen up! Developmental differences in the impact of IDS on speech segmentation

While American English infants typically segment words from fluent speech by 7.5-months, studies of infants from other language backgrounds have difficulty replicating this finding. One possible explanation for this cross-linguistic difference is that the input infants from different language backgrounds receive is not as infant-directed as American English infant-directed speech (Floccia et al., 2016). Against this background, the current study investigates whether German 7.5- and 9-month-old infants segment words from fluent speech when the input is prosodically similar to American English IDS. While 9-month-olds showed successful segmentation of words from exaggerated IDS, 7.5-month-olds did not. These findings highlight (a) the beneficial impact of exaggerated IDS on infant speech segmentation, (b) cross-linguistic differences in word segmentation that are based not just on the kind of input available to children and suggest (c) developmental differences in the role of IDS as an attentional spotlight in speech segmentation.

British English infants segment words only with exaggerated infant-directed speech stimuli

The word segmentation paradigm originally designed by Jusczyk and Aslin (1995) has been widely used to examine how infants from the age of 7.5months can extract novel words from continuous speech. Here we report a series of 13 studies conducted independently in two British laboratories, showing that British English-learning infants aged 8–10.5months fail to show evidence of word segmentation when tested in this paradigm. In only one study did we find evidence of word segmentation at 10.5months, when we used an exaggerated infant-directed speech style. We discuss the impact of variations in infant-directed style within and across languages in the course of language acquisition.

Detailing vowel development in infancy using cortical auditory evoked potentials and multidimensional scaling

The present study used an efficient measure of perceptual sensitivity to map perception across the British English vowel space for 80 monolingual English infants (4–5, 8–9, and 10–11 months old). Auditory evoked potentials were measured for spectral changes between concatenated vowels, which, for infants, typically evokes a positivity about 150–200 ms after each spectral change. These were measured for 28 pairs of seven monophthongal vowels (/i/, /ɪ/, /ɛ/, /a/, /ɑ/, /ɔ/, /u/) that were presented in a random concatenated sequence with changes every 300–400 ms. ERPs were averaged across epochs following each spectral change, with the magnitude of the response for each vowel pair used as similarity measure for multidimensional scaling. The 4–5 month old infants had two-dimensional perceptual maps that closely matched the F1 and F2 acoustic differences between vowels. In contrast, the older infant response was less related to acoustic differences and they had selectively larger responses for neighbors around the vowel quadrilateral (e.g., /i/-/ɪ/); suggesting a shift to a more phonologically driven processing. These results provide a more detailed picture of phonetic development than has been shown before, and demonstrate an efficient procedure to map speech processing in infancy.

Exploring the Role of Stress in Bayesian Word Segmentation using Adaptor Grammars

Stress has long been established as a major cue in word segmentation for English infants. We show that enabling a current state-of-the-art Bayesian word segmentation model to take advantage of stress cues noticeably improves its performance. We find that the improvements range from 10 to 4%, depending on both the use of phonotactic cues and, to a lesser extent, the amount of evidence available to the learner. We also find that in particular early on, stress cues are much more useful for our model than phonotactic cues by themselves, consistent with the finding that children do seem to use stress cues before they use phonotactic cues. Finally, we study how the model’s knowledge about stress patterns evolves over time. We not only find that our model correctly acquires the most frequent patterns relatively quickly but also that the Unique Stress Constraint that is at the heart of a previously proposed model does not need to be built in but can be acquired jointly with word segmentation.

Utterance‐Final Lengthening Is Predictive of Infants’ Discrimination of English Accents

Infants in their first year manifest selective patterns of discrimination between languages and between accents of the same language. Prosodic differences are held to be important in whether languages can be discriminated, together with the infant's familiarity with one or both of the accents heard. However, the nature of the prosodic cues that actually facilitate infant discrimination has not been directly examined. We analyzed the accent discrimination of 5‐ and 7‐month‐old British English infants, looking for durational features that could predict a range of discrimination results using the Headturn Preference Procedure. We previously found that both 5‐ and 7‐month‐olds based in Plymouth could discriminate Plymouth‐accented and Welsh‐accented English, while 5‐month‐olds could not discriminate Welsh‐accented and Scottish‐accented English. Most surprisingly, 7‐month‐olds failed to discriminate Plymouth‐accented and French‐accented English. From half of all utterances used in these four experiments, we calculated a range of durational metrics, both globally across utterances and locally at utterance‐final edges. Utterance‐edge metrics were the relative duration of the final consonant+vowel interval (nfinalCV) and the final vowel alone (nfinalV). Separately for 5‐ and 7‐month‐olds, we determined the difference in scores for all durational metrics between: (a) familiar training and novel test utterances and (b) familiar test and novel test utterances. Regression analyses showed unique predictors of discrimination. For 5‐month‐olds, the predictor was the difference in nfinalCV between familiar training and novel test utterances. For 7‐month‐olds, the predictor was the difference in finalV between familiar test and novel test utterances. This demonstrates an early sensitivity to precisely those localized timing features that are useful across languages for segmentation. We therefore predict that similar sensitivity should be shown in infants of other language backgrounds.

Evaluating the potential risks and benefits of infant rotavirus vaccination in England

Rotarix®, a vaccine for the prevention of gastroenteritis in young children, was introduced in England in July 2013. At around this time, an elevated risk of intussusception (a cause of bowel obstruction) was reported among infants vaccinated in Australia and the USA. A risk-benefit analysis compared potential vaccine-related risks (additional intussusception admissions and deaths) with estimated vaccine benefits (prevented rotavirus general practitioner visits, emergency visits, admissions and deaths) in the 2012 birth cohort. Detailed data from England included the incidence of intussusception events aged <2 years by week of age, the coverage of vaccination aged <2 years by week of age, and the incidence of rotavirus gastroenteritis (RVGE) events aged <5 years by week of age. Recent estimates of vaccine-related risk from Australia were applied during the 1–21 day period after the first and second dose of vaccination. Rotarix® is estimated to cause one additional intussusception admission in every 18,551 vaccinated English infants (5th and 95th percentiles, 6728–93,952), equivalent to 35 (7-98) additional intussusception admissions each year. The vaccine is estimated to prevent three rotavirus deaths, 13,000 rotavirus admissions, 27,000 rotavirus emergency visits and 74,000 rotavirus GP consultations in children aged <5 years, and lead to annual savings of over £11 million, each year. We estimate 375 (136-1900) fewer RVGE admissions for every additional intussusception admission, and 88 (18-852) fewer RVGE deaths for every additional intussusception death. The estimated benefits of Rotarix® vaccination would greatly exceed the potential risk in England.

Visual and sensori-motor influences on speech perception in infancy

Speech perception is multisensory and is comprised not only of auditory information, but also of visual (Burnham and Dodd, 2004; Kuhl and Meltzoff, 1984; Patterson and Werker, 2003) and proprioceptive motor information (Yeung and Werker, 2013). Building on previous studies examining perceptual attunement in young infants (Werker and Tees, 1984, inter alia), this set of experiments examines the role that vision and motor proprioception play during the perception of auditory speech sounds. In the first experiment, the developmental trajectory of perceptual attunement in English-learning infants is again explored using the dental-retroflex contrast of Hindi. We replicate the finding that English-learning 6-month-olds are able to discriminate auditory-only dental and retroflex stimuli, while English-learning 10-month-olds are not. In the second experiment, looking time measurements are used to explore the possibility that the addition of dynamic visual information acts as a perceptual anchor, now permitting discrimination of this same non-native contrast by 10-month-old infants. Finally, in the third experiment, we investigate the role of a temporary motor manipulation, designed to prevent relevant movement of the tongue in 6- and 10-month-old English infants during perception of the non-native Hindi contrast, to determine the effect of proprioceptive, sensori-motor mechanisms in auditory speech perception across development.

The influence of babbling patterns on the processing of speech

This study compared the preference of 27 British English- and 26 Welsh-learning infants for nonwords featuring consonants that occur with equal frequency in the input but that are produced either with equal frequency (Welsh) or with differing frequency (British English) in infant vocalizations. For the English infants a significant difference in looking times was related to the extent of production of the nonword consonants. The Welsh infants, who showed no production preference for either consonant, exhibited no such influence of production patterns on their response to the nonwords. The results are consistent with a previous study that suggested that pre-linguistic babbling helps shape the processing of input speech, serving as an articulatory filter that selectively makes production patterns more salient in the input.

A Language Stimulation Analysis of Cantonese-English Bilingual Infants in Hong Kong

This study measured naturalistic language stimulation in Hong Kong among eight bilingual (Cantonese, English) infants. Two findings were that (a) infants received more indirect (adult-to-adult) than direct (adult-to-infant) language stimulation and (b) infants received 6.8 hr of language stimulation per day with more fillers, vocatives, and baby talk than declaratives, interrogatives, and imperatives.

Phonological Specificity of Tones in Early Lexical Representation at 16 Months of Age

By about 14 months of age, English infants show sensitivity to mispronunciations of consonants and vowels of familiar and newly learnt words when asked to identify a referent. They are also capable of learning two similar-sounding words. These ?ndings suggest that infants possess phonologically detailed representations of words. However,very little is known about the role that tones play in early lexical representations. Word meaning is also distinguished by tones, which consist primarily of pitch level or contour variations in tonal language, like Chinese. While in English, a typically non-tonal language, pitch change only plays a role in the phrasal level as intonation. This study aims to answer the question: Do 16-month-old Chinese and English infants treat tones as phonological information in their lexical representations? Using the intermodal preferential looking paradigm with the mispronunciation task, study 1 examined whether Chinese infants at 16 months were sensitive to mispronunciations of tones in monosyllabic, familiar words. 15 infants were separated into two groups. For group 1, the familiar words were pronounced correctly in block1, while mispronounced with the falling tone in Mandarin (T4) in block 2. The block order was reversed for the infants in group 2. Using a similar task to study 1, study 2 investigated whether English infants at 16 months were sensitive to tonal change in familiar English words when asked to identify a referent. In contrast with study 1, all the 12 infants received one kind of block order. In block 1, the words were pronounced with T4 as the familiar tone. A falling pitch contour is common in English and quite similar to T4 whereas it is less common to pronounce words with a fall-rise contour in English. Therefore, the dipping tone in Mandarin (T3) was used as the novel tone in block 2 and were considered as tone mispronunciations in study 2. The results of study 1 showed that Chinese infants in group 1 looked longer at a target object when its label was correctly pronounced than when the label was mispronounced, in other words, a mispronunciation effect was found. In group 2, there were no systematic difference in infants' preferences for the target between the pre- and the post-naming phase, neither when the target label was mispronounced or pronounced correctly. The results of study 2 showed that English infants looked longer at a target object when its label was pronounced with T4 than when the label was pronounced with T3. A positive correlation was found between English infants' receptive vocabulary size and their performance in the mispronunciation condition. In conclusion, the results indicate that 1) tones interact with phonological information for both Chinese and English infants at 16 months of age; 2) 16-month-old English infants may be on the way of learning to eliminate the phonological specificity of tone.

Syntactic trees and small-world networks: syntactic development as a dynamical process

Language has been argued to exhibit a complex system behavior. In our approach, the syntactic relations of dependency between words have been represented as networks. In a previous study, two English infants’ corpora of utterances were analyzed longitudinally, offering a view of the ontogeny of syntax. Abrupt changes were detected in the growth pattern of the Giant Connected Component – the largest connected set of nodes in a graph. In the present study, we have further analyzed and compared three more infants, from the CHILDES database, learning three different languages: Dutch, German and Spanish. Our results show, along with previous work with English-speaking infants, that all three infants’ syntactic networks change their topology in a similar way, from tree-like networks to small-world networks. This change happens at a similar period in all three infants (between ~700 and ~800 days), regardless of the language they acquire. Our study also shows that the hubs – the most connected nodes in these small-world networks – are always the so-called functional words, which, according to linguistic theory, just contribute to the syntactic structure of human language. The emergence of these hubs happens abruptly, following a logarithmic growth pattern. This developmental pattern challenges usage-based theories of language acquisition and suggests that syntactic development is driven by the growth of the lexicon.

Contextual influences on the excursion size of fundamental frequency in infant babbling: A comparative study of English and Mandarin.

Davis and MacNeilage (1995) reported contextual influences on infants’ segmental production in which vowel production preference and error were strongly influenced by word level contexts. However, few studies have addressed the issue of contextual influences on supra‐segmental production in pre‐linguistic periods of acquisition. The present study investigated the influence of utterance length and utterance position on the size of f0 excursion in the babbling of English and Mandarin infants. Spontaneous babbling speech of four monolingual English infants and four monolingual Mandarin infants at the age of 8 and 14 months were collected. Canonical babbling syllables with either a falling or rising f0 contour were acoustically analyzed. The magnitude of f0 excursion size was calculated as the difference between maximum f0 and minimum f0 within a syllable. Results showed that F0 excursion size was significantly larger in monosyllables than in individual syllables of multisyllabic utterances. It is also larger in utterance final syllables than in non‐final syllables. The influence of age and ambient language properties of f0 will also be discussed. [Work supported by NSF 0844080.]

The development of perceptual grouping biases in infancy: A Japanese-English cross-linguistic study

Perceptual grouping has traditionally been thought to be governed by innate, universal principles. However, recent work has found differences in Japanese and English speakers’ non-linguistic perceptual grouping, implicating language in non-linguistic perceptual processes ( Iversen, Patel, & Ohgushi, 2008). Two experiments test Japanese- and English-learning infants of 5–6 and 7–8 months of age to explore the development of grouping preferences. At 5–6 months, neither the Japanese nor the English infants revealed any systematic perceptual biases. However, by 7–8 months, the same age as when linguistic phrasal grouping develops, infants developed non-linguistic grouping preferences consistent with their language’s structure (and the grouping biases found in adulthood). These results reveal an early difference in non-linguistic perception between infants growing up in different language environments. The possibility that infants’ linguistic phrasal grouping is bootstrapped by abstract perceptual principles is discussed.

Cross‐linguistic evidence for the influence of native language prosody in infant speech segmentation.

Research using artificial languages with English‐learning infants has shown both prosodic and distributional cues are used for speech segmentation by 7 months. When these cues conflict, infants younger than 7 months rely on distributional cues while older infants rely on prosodic cues). In the present study we assessed the role of prosodic information in segmentation when it is not favorably aligned with distributional cues, to determine whether language‐specific rhythmic biases guide segmentation as suggested by studies using natural speech. Two continuous streams of naturally produced syllables (English and French) were constructed using nines syllables that are permissible in both languages. Within each stream statistical cues were manipulated independently of language‐appropriate stress cues. English‐ and French‐learning 8‐month‐olds were familiarized with their native language stream and then present test probe strings to determine what syllable sequences were extracted from the stream. Probes were selected to assess the role of stress cues in segmentation. Findings show that English infants make use of a trochaic template; Canadian‐French infants show a weaker and less focused reliance on stress cues to segment words from connected speech. These cross‐linguistic differences reflect processing biases that may be set by language experience and/or elicited by speech input properties.

Perception of vowel length by Japanese- and English-learning infants.

This study investigated vowel length discrimination in infants from 2 language backgrounds, Japanese and English, in which vowel length is either phonemic or nonphonemic. Experiment 1 revealed that English 18-month-olds discriminate short and long vowels although vowel length is not phonemically contrastive in English. Experiments 2 and 3 revealed that Japanese 18-month-olds also discriminate the pairs but in an asymmetric manner: They detected only the change from long to short vowel, but not the change in the opposite direction, although English infants in Experiment 1 detected the change in both directions. Experiment 4 tested Japanese 10-month-olds and revealed a symmetric pattern of discrimination similar to that of English 18-month-olds. Experiment 5 revealed that native adult Japanese speakers, unlike Japanese 18-month-old infants who are presumably still developing phonological perception, ultimately acquire a symmetrical discrimination pattern for the vowel contrasts. Taken together, our findings suggest that English 18-month-olds and Japanese 10-month-olds perceive vowel length using simple acoustic?phonetic cues, whereas Japanese 18-month-olds perceive it under the influence of the emerging native phonology, which leads to a transient asymmetric pattern in perception.