Distributional learning of speech sound categories is gated by sensitive periods

Resilience and vulnerability of neural speech tracking after hearing restoration

The role of early auditory experience in the development of neural speech tracking remains an open question. To address this issue, we measured neural speech tracking in children with or without functional hearing during their first year of life after their hearing was restored with cochlear implants (CIs), as well as in hearing controls (HC). Neural tracking in children with CIs is unaffected by the absence of perinatal auditory experience. CI users and HC exhibit a similar neural tracking magnitude at short timescales of brain activity. However, neural tracking is delayed in CI users, and its timing depends on the age of hearing restoration. Conversely, at longer timescales, speech tracking is dampened in participants using CIs, thereby accounting for their speech comprehension deficits. These findings highlight the resilience of sensory processing in speech tracking while also demonstrating the vulnerability of higher-level processing to the lack of early auditory experience.

Early influence of language experience in non-native speech perception: Discrimination of three-way Thai stop contrasts by Korean and Japanese infants

Young infants' remarkable ability to discriminate non-native phoneme contrasts played a critical role in shaping the tenets of the perceptual narrowing hypothesis: early on, infants are sensitive to most phoneme categories, including those not used in their native language, but lose this sensitivity as they attune to their language. However, supporting evidence was derived from limited geographical regions and languages, particularly on early sensitivity, requiring further studies to specify the extent of early sensitivity and reassess the dominant developmental pattern. This study aimed to fill this gap by examining discrimination patterns for three-way Thai stop contrasts by two other Asian language learners (Korean and Japanese) at age 4-6 months. The three stop categories in Thai are distinct along the voice onset time (VOT) dimension, encompassing both negative and positive values. Thai pre-voiced and voiceless (i.e., short lag) stops are similar to stop categories used in languages such as French, Dutch, and Spanish. Thai voiceless and voiceless aspirated (i.e., long lag) stops are similar to those in English, Chinese, and German. Therefore, Thai stop categories provide an ideal test continuum for confirming early universal sensitivities to two supposedly language-general VOT boundaries (-30 ms, +30 ms). We presented two Thai phoneme pairs (pre-voiced vs. voiceless, voiceless vs. voiceless aspirated) to Korean and Japanese infants aged 4-6 months and observed their discrimination patterns using a visual habituation paradigm. The results showed divergent discrimination between the two language learners. Korean infants showed sensitivity to the pre-voiced-voiceless pair, whereas Japanese infants did not. By contrast, only Japanese infants showed some sensitivity to the voiceless-voiceless aspirated pair with some directionality effect, whereas Korean infants did not. These results demonstrate systematic cross-linguistic differences reflecting input influence in early perceptual sensitivity and suggest the ambient language environment may influence consonant perception much earlier than has been considered by the perceptual narrowing theory, calling for further refinement of the extent of initial perceptual state in the theory.

Lack of early sensitivity and gradual emergence of native phoneme categories: A pattern from underrepresented language learners

Perceptual narrowing of speech perception supposes that young infants can discriminate most speech sounds early in life. During the second half of the first year, infants' phonetic sensitivity is attuned to their native phonology. However, supporting evidence for this pattern comes primarily from learners from a limited number of regions and languages. Very little evidence has accumulated on infants learning languages spoken in Asia, which accounts for most of the world's population. The present study examined the developmental trajectory of Korean-learning infants' sensitivity to a native stop contrast during the first year of life. The Korean language utilizes unusual voiceless three-way stop categories, requiring target categories to be derived from tight phonetic space. Further, two of these categories-lenis and aspirated-have undergone a diachronic change in recent decades as the primary acoustic cue for distinction has shifted among modern speakers. Consequently, the input distributions of these categories are mixed across speakers and speech styles, requiring learners to build flexible representations of target categories along these variations. The results showed that among the three age groups-4-6 months, 7-9 months, and 10-12 months-we tested, only 10-12-month-olds showed weak sensitivity to the two categories, suggesting that robust discrimination is not in place by the end of the first year. The study adds scarcely represented data, lending additional support for the lack of early sensitivity and prolonged emergence of native phonology that are inconsistent with learners of predominant studies and calls for more diverse samples to verify the generality of the typical perceptual narrowing pattern. RESEARCH HIGHLIGHTS: We investigated Korean-learning infants' developmental trajectory of native phoneme categories and whether they show the typical perceptual narrowing pattern. Robust discrimination did not appear until 12 months, suggesting that Korean infants' native phonology is not stabilized by the end of the first year. The prolonged emergence of sensitivity could be due to restricted phonetic space and input variations but suggests the possibility of a different developmental trajectory. The current study contributes scarcely represented Korean-learning infants' phonetic discrimination data to the speech development field.

Learning phonetic categories in infancy: The role of word-context information

Identifying the type of mechanisms at the core of phonetic categorization remains a central subject of research in infant language learning. Amongst different theories, one is that infants compute distributional information of phonemes based on their surrounding sounds (i.e., word context) such that phonemes that appear in different word contexts are more likely to be discriminated and categorized separately than phonemes that appear in similar word contexts. Following the procedure of Feldman et al. (2013a), we investigated the role of contextual information in the acquisition of phonetic categories in 8-month-old infants, using a non-native vowel contrast (English /ɒ/-/ʌ/). In Experiment 1, we established lack of discrimination of the non-native contrast without prior exposure to it. In Experiment 2, we manipulated the type of exposure prior to testing: half of the infants were exposed to minimal pair carriers (words that differ by one phoneme only; e.g., lituh and litoh), and the other half of the infants were exposed to non-minimal pair carriers (words formed by different phonemes; e.g., lituh and nutoh). All infants were tested for discrimination of the contrast (tuh vs. toh) presented as alternating (e.g., tuh-toh-tuh-toh) and non-alternating trials (e.g., tuh-tuh-tuh), as in Experiment 1. Infants in both conditions looked on average longer at alternating rather than non-alternating trials, suggesting that they discriminated the /ɒ/-/ʌ/ contrast after a brief exposure to the vowels embedded into words. Crucially, discrimination occurred regardless of whether words were minimal pair carriers or non-minimal pair carriers. A cross-experiment comparison revealed that infants showed different patterns of looking times based on whether they were exposed to the contrast before testing (Experiment 2) or not (Experiment 1). Our study shows that any type of word context helps infants to re-establish discrimination of non-native contrasts once sensitivity has been lost. These findings aid to better understand how the speech input modulates learning mechanisms during the establishment of phonetic categories in the first year of postnatal life.

Phonetic perceptual reorganization across the first year of life: Looking back

This paper provides a selective overview of some of the research that has followed from the publication of Werker and Tees (1984a) "Cross-language speech perception: Evidence for Perceptual Reorganization During the First Year of Life." Specifically, I briefly present the original finding, our interpretation of its meaning, and some key replications and extensions. I then review some of the work that has followed, including work with different kinds of populations, different kinds of speech sound contrasts, as well as attunement (perceptual reorganization) to additional properties of language beyond phonetic contrasts. Included is the body of work that queries whether perceptual attunement is a critical period phenomenon. Potential learning mechanisms for how experience functions to guide phonetic perceptual development are also presented, as is work on the relation between speech perception and word learning.

Individual differences in auditory perception predict learning of non-adjacent tone sequences in 3-year-olds

Auditory processing of speech and non-speech stimuli oftentimes involves the analysis and acquisition of non-adjacent sound patterns. Previous studies using speech material have demonstrated (i) children's early emerging ability to extract non-adjacent dependencies (NADs) and (ii) a relation between basic auditory perception and this ability. Yet, it is currently unclear whether children show similar sensitivities and similar perceptual influences for NADs in the non-linguistic domain. We conducted an event-related potential study with 3-year-old children using a sine-tone-based oddball task, which simultaneously tested for NAD learning and auditory perception by means of varying sound intensity. Standard stimuli were A × B sine-tone sequences, in which specific A elements predicted specific B elements after variable × elements. NAD deviants violated the dependency between A and B and intensity deviants were reduced in amplitude. Both elicited similar frontally distributed positivities, suggesting successful deviant detection. Crucially, there was a predictive relationship between the amplitude of the sound intensity discrimination effect and the amplitude of the NAD learning effect. These results are taken as evidence that NAD learning in the non-linguistic domain is functional in 3-year-olds and that basic auditory processes are related to the learning of higher-order auditory regularities also outside the linguistic domain.

Infant Phonetic Learning as Perceptual Space Learning: A Crosslinguistic Evaluation of Computational Models

In the first year of life, infants' speech perception becomes attuned to the sounds of their native language. This process of early phonetic learning has traditionally been framed as phonetic category acquisition. However, recent studies have hypothesized that the attunement may instead reflect a perceptual space learning process that does not involve categories. In this article, we explore the idea of perceptual space learning by implementing five different perceptual space learning models and testing them on three phonetic contrasts that have been tested in the infant speech perception literature. We reproduce and extend previous results showing that a perceptual space learning model that uses only distributional information about the acoustics of short time slices of speech can account for at least some crosslinguistic differences in infant perception. Moreover, we find that a second perceptual space learning model, which benefits from word-level guidance. performs equally well in capturing crosslinguistic differences in infant speech perception. These results provide support for the general idea of perceptual space learning as a theory of early phonetic learning but suggest that more fine-grained data are needed to distinguish between different formal accounts. Finally, we provide testable empirical predictions of the two most promising models and show that these are not identical, making it possible to independently evaluate each model in experiments with infants in future research.

The role of L1-L2 dissimilarity in L2 segment learning - Implications from the acquisition of English post-alveolar fricatives by Mandarin and Mandarin/Wu speakers

This study examines how the concept of L1-L2 dissimilarity should be addressed from a two-way perspective in L2 segment learning, and how it relates to the learning outcomes. We achieved this by investigating the productions of the post-alveolar fricatives /ʃ, ʒ/ by Mandarin and Mandarin/Wu speakers, which were subsequently assessed by native English listeners. In the first experiment, we analyzed the spectral moments of /ʃ, ʒ/ produced by Mandarin monolingual and Mandarin/Wu bilingual speakers to find out how the two groups of speakers pronounced the target segments. In the second experiment, native English listeners were tasked with rating the accentedness of the Mandarin- and Mandarin/Wu-accented /ʃ, ʒ/. Results showed native English listeners scored Mandarin/Wu-accented /ʃ/ as having no accent and Mandarin-accented /ʒ/ as having a heavy accent, indicating that English natives perceived the 'native vs. nonnative' segment dissimilarity differently from Chinese learners of English, and that the L1-L2 dissimilarity perceived from both sides may work together in defining the L2 segment learning outcomes.

Naturalistic speech supports distributional learning across contexts

At birth, infants discriminate most of the sounds of the world's languages, but by age 1, infants become language-specific listeners. This has generally been taken as evidence that infants have learned which acoustic dimensions are contrastive, or useful for distinguishing among the sounds of their language(s), and have begun focusing primarily on those dimensions when perceiving speech. However, speech is highly variable, with different sounds overlapping substantially in their acoustics, and after decades of research, we still do not know what aspects of the speech signal allow infants to differentiate contrastive from noncontrastive dimensions. Here we show that infants could learn which acoustic dimensions of their language are contrastive, despite the high acoustic variability. Our account is based on the cross-linguistic fact that even sounds that overlap in their acoustics differ in the contexts they occur in. We predict that this should leave a signal that infants can pick up on and show that acoustic distributions indeed vary more by context along contrastive dimensions compared with noncontrastive dimensions. By establishing this difference, we provide a potential answer to how infants learn about sound contrasts, a question whose answer in natural learning environments has remained elusive.

How Tone, Intonation and Emotion Shape the Development of Infants’ Fundamental Frequency Perception

Fundamental frequency (ƒ₀), perceived as pitch, is the first and arguably most salient auditory component humans are exposed to since the beginning of life. It carries multiple linguistic (e.g., word meaning) and paralinguistic (e.g., speakers’ emotion) functions in speech and communication. The mappings between these functions and ƒ₀ features vary within a language and differ cross-linguistically. For instance, a rising pitch can be perceived as a question in English but a lexical tone in Mandarin. Such variations mean that infants must learn the specific mappings based on their respective linguistic and social environments. To date, canonical theoretical frameworks and most empirical studies do not view or consider the multi-functionality of ƒ₀, but typically focus on individual functions. More importantly, despite the eventual mastery of ƒ₀ in communication, it is unclear how infants learn to decompose and recognize these overlapping functions carried by ƒ₀. In this paper, we review the symbioses and synergies of the lexical, intonational, and emotional functions that can be carried by ƒ₀ and are being acquired throughout infancy. On the basis of our review, we put forward the Learnability Hypothesis that infants decompose and acquire multiple ƒ₀ functions through native/environmental experiences. Under this hypothesis, we propose representative cases such as the synergy scenario, where infants use visual cues to disambiguate and decompose the different ƒ₀ functions. Further, viable ways to test the scenarios derived from this hypothesis are suggested across auditory and visual modalities. Discovering how infants learn to master the diverse functions carried by ƒ₀ can increase our understanding of linguistic systems, auditory processing and communication functions.

Learning to Perceive Non-Native Tones via Distributional Training: Effects of Task and Acoustic Cue Weighting

As many distributional learning (DL) studies have shown, adult listeners can achieve discrimination of a difficult non-native contrast after a short repetitive exposure to tokens falling at the extremes of that contrast. Such studies have shown using behavioural methods that a short distributional training can induce perceptual learning of vowel and consonant contrasts. However, much less is known about the neurological correlates of DL, and few studies have examined non-native lexical tone contrasts. Here, Australian-English speakers underwent DL training on a Mandarin tone contrast using behavioural (discrimination, identification) and neural (oddball-EEG) tasks, with listeners hearing either a bimodal or a unimodal distribution. Behavioural results show that listeners learned to discriminate tones after both unimodal and bimodal training; while EEG responses revealed more learning for listeners exposed to the bimodal distribution. Thus, perceptual learning through exposure to brief sound distributions (a) extends to non-native tonal contrasts, and (b) is sensitive to task, phonetic distance, and acoustic cue-weighting. Our findings have implications for models of how auditory and phonetic constraints influence speech learning.

A practical guide to applying machine learning to infant EEG data

Electroencephalography (EEG) has been widely adopted by the developmental cognitive neuroscience community, but the application of machine learning (ML) in this domain lags behind adult EEG studies. Applying ML to infant data is particularly challenging due to the low number of trials, low signal-to-noise ratio, high inter-subject variability, and high inter-trial variability. Here, we provide a step-by-step tutorial on how to apply ML to classify cognitive states in infants. We describe the type of brain attributes that are widely used for EEG classification and also introduce a Riemannian geometry based approach for deriving connectivity estimates that account for inter-trial and inter-subject variability. We present pipelines for learning classifiers using trials from a single infant and from multiple infants, and demonstrate the application of these pipelines on a standard infant EEG dataset of forty 12-month-old infants collected under an auditory oddball paradigm. While we classify perceptual states induced by frequent versus rare stimuli, the presented pipelines can be easily adapted for other experimental designs and stimuli using the associated code that we have made publicly available.

Towards a model of language neurobiology in early development

Understanding language neurobiology in early childhood is essential for characterizing the developmental structural and functional changes that lead to the mature adult language network. In the last two decades, the field of language neurodevelopment has received increasing attention, particularly given the rapid advances in the implementation of neuroimaging techniques and analytic approaches that allow detailed investigations into the developing brain across a variety of cognitive domains. These methodological and analytical advances hold the promise of developing early markers of language outcomes that allow diagnosis and clinical interventions at the earliest stages of development. Here, we argue that findings in language neurobiology need to be integrated within an approach that captures the dynamic nature and inherent variability that characterizes the developing brain and the interplay between behavior and (structural and functional) neural patterns. Accordingly, we describe a framework for understanding language neurobiology in early development, which minimally requires an explicit characterization of the following core domains: i) computations underlying language learning mechanisms, ii) developmental patterns of change across neural and behavioral measures, iii) environmental variables that reinforce language learning (e.g., the social context), and iv) brain maturational constraints for optimal neural plasticity, which determine the infant's sensitivity to learning from the environment. We discuss each of these domains in the context of recent behavioral and neuroimaging findings and consider the need for quantitatively modeling two main sources of variation: individual differences or trait-like patterns of variation and within-subject differences or state-like patterns of variation. The goal is to enable models that allow prediction of language outcomes from neural measures that take into account these two types of variation. Finally, we examine how future methodological approaches would benefit from the inclusion of more ecologically valid paradigms that complement and allow generalization of traditional controlled laboratory methods.

Cross-Dialectal Novel Word Learning and Borrowing

The objective of this paper was to study the cognitive processes underlying cross-dialectal novel word borrowing and loanword establishment in a Standard-Chinese-to-Shanghainese (SC-SH) auditory lexical learning and borrowing experiment. To investigate these underlying cognitive processes, SC-SH bi-dialectals were compared with SC monolectals as well as bi-dialectals of SC and other Chinese dialects (OD) to investigate the influence of short-term and long-term linguistic experience. Both comprehension and production borrowings were tested. This study found that early and proficient bi-dialectism, even if it is not directly related to the recipient dialect of lexical borrowing, has a protective effect on the ability of cross-dialectal lexical borrowing in early adulthood. Bi-dialectals tend to add separate lexical representations for incidentally encountered dialectal variants, while monolectals tend to assimilate dialectal variants to standard forms. Bi-dialectals, but not monolectals, use etymologically related morphemes between the source and recipient dialects to create nonce-borrowing compounds. Dialectal variability facilitates lexical borrowing via enriching instead of increasing the short-term lexical experience of learners. The long-term bi-dialectal experience of individuals, as well as their short-term exposure to each specific loanword, may collectively shape the route of lexical evolution of co-evolving linguistic varieties.