Infants' abilities to segment word forms from spectrally degraded speech in the first year of life.

Infants begin to segment word forms from fluent speech-a crucial task in lexical processing-between 4 and 7 months of age. Prior work has established that infants rely on a variety of cues available in the speech signal (i.e., prosodic, statistical, acoustic-segmental, and lexical) to accomplish this task. In two experiments with French-learning 6- and 10-month-olds, we use a psychoacoustic approach to examine if and how degradation of the two fundamental acoustic components extracted from speech by the auditory system, namely, temporal (both frequency and amplitude modulation) and spectral information, impact word form segmentation. Infants were familiarized with passages containing target words, in which frequency modulation (FM) information was replaced with pure tones using a vocoder, while amplitude modulation (AM) was preserved in either 8 or 16 spectral bands. Infants were then tested on their recognition of the target versus novel control words. While the 6-month-olds were unable to segment in either condition, the 10-month-olds succeeded, although only in the 16 spectral band condition. These findings suggest that 6-month-olds need FM temporal cues for speech segmentation while 10-month-olds do not, although they need the AM cues to be presented in enough spectral bands (i.e., 16). This developmental change observed in infants' sensitivity to spectrotemporal cues likely results from an increase in the range of available segmentation procedures, and/or shift from a vowel to a consonant bias in lexical processing between the two ages, as vowels are more affected by our acoustic manipulations. RESEARCH HIGHLIGHTS: Although segmenting speech into word forms is crucial for lexical acquisition, the acoustic information that infants' auditory system extracts to process continuous speech remains unknown. We examined infants' sensitivity to spectrotemporal cues in speech segmentation using vocoded speech, and revealed a developmental change between 6 and 10 months of age. We showed that FM information, that is, the fast temporal modulations of speech, is necessary for 6- but not 10-month-old infants to segment word forms. Moreover, reducing the number of spectral bands impacts 10-month-olds' segmentation abilities, who succeed when 16 bands are preserved, but fail with 8 bands.

Six-month-old infants' perception of structural regularities in speech.

In order to acquire grammar, infants need to extract regularities from the linguistic input. From birth, infants can detect regularities in speech based on identity relations, and show strong neural activation to syllable sequences containing adjacent repetitions of identical syllables (e.g. ABB: mubaba). Meanwhile, newborns' neural responses to sequences of different syllables (e.g. ABC: mubage, i.e. diversity-based relations) do not differ from baseline. However, this latter ability needs to emerge during development, as most linguistic units, such as words, are composed of highly variable sequences. As infants begin to learn their first word forms at 6 months, we hypothesize that the ability to represent sequences of different syllables might become important for them at this age. Using near-infrared spectroscopy (NIRS), we measured 6-month-old infants' brain responses to repetition- and diversity-based sequences in the bilateral temporal, parietal and frontal areas. We found that 6-month-olds discriminated the repetition- and diversity-based structures in frontal and parietal regions, and exhibited equally strong activation to both grammars as compared to baseline. These results show that by 6 months of age, infants encode sequences with diversity-based structures. They thus provide the earliest evidence that prelexical infants represent difference in speech stimuli, which behavioral studies first attest at 11 months of age.

Consonant bias in adult lexical processing under acoustically degraded listening conditions.

Consonants facilitate lexical processing across many languages, including French. This study investigates whether acoustic degradation affects this phonological bias in an auditory lexical decision task. French words were processed using an eight-band vocoder, degrading their frequency modulations (FM) while preserving original amplitude modulations (AM). Adult French natives were presented with these French words, preceded by similarly processed pseudoword primes sharing their vowels, consonants, or neither. Results reveal a consonant bias in the listeners' accuracy and response times, despite the reduced spectral and FM information. These degraded conditions resemble current cochlear-implant processors, and attest to the robustness of this phonological bias.

Erratum: Reproducibility of infant fNIRS studies: a meta-analytic approach (Erratum).

[This corrects the article DOI: 10.1117/1.NPh.10.2.023518.].

Reproducibility of infant fNIRS studies: a meta-analytic approach.

Significance: Concerns about the reproducibility of experimental findings have recently emerged in many disciplines, from psychology to medicine and neuroscience. As NIRS is a relatively recent brain imaging technique, the question of reproducibility has not yet been systematically addressed. Aim: The current study seeks to test the replicability of effects observed in NIRS experiments assessing young infants' rule-learning ability. Approach: We conducted meta-analyses and mixed-effects modeling-based inferential statistics to determine whether effect sizes were replicable and comparable in a sample of 23 NIRS studies investigating infants' abilities to process repetition- and diversity-based regularities in linguistic and nonlinguistic auditory and visual sequences. Additionally, we tested whether effect sizes were modulated by different factors such as the age of participants or the laboratory. We obtained NIRS data from 12 published and 11 unpublished studies. The 23 studies involved a total of 487 infants, aged between 0 and 9 months, tested in four different countries (Canada, France, Italy, and USA). Results: Our most important finding is that study and laboratory were never significant moderators of variation in effect sizes, indicating that results replicated reliably across the different studies and labs included in the sample. We observed small-to-moderate effect sizes, similar to effect sizes found with other neuroimaging and behavioral techniques in the developmental literature. In line with existing findings, effect sizes were modulated by the participants' age and differed across the different regularities tested, with repetition-based regularities giving rise to the strongest effects; in particular, the overall magnitude of this effect in the left temporal region was 0.27 when analyzing the entire dataset. Conclusions: Meta-analysis is a useful tool for assessing replicability and cross-study variability. Here, we have shown that infant NIRS studies in the language domain replicate robustly across various NIRS machines, testing sites, and developmental populations.

Seven-month-old infants detect symmetrical structures in multi-featured abstract visual patterns.

The present study investigated 7-month-old infants' ability to perceive structural symmetry in mosaic-like abstract visual patterns. We examined infants' (n = 98) spontaneous looking behaviour to mosaic-like sequences with symmetrical and asymmetrical structures. Sequences were composed of square tiles from two categories that differed in their colour scheme and internal shape. We manipulated sequence length (3 or 5 tiles) and abstractness of the symmetry (token vs. category level). The 7-month-olds discriminated structurally symmetrical from asymmetrical mosaics in the first half of the test phase (first 8 trials). Sequence length, level of symmetry, or number of unique tiles per sequence did not significantly modulate infants' looking behaviour. These results suggest that very young infants detect differences in structural symmetry in multi-featured visual patterns.

Infants differentially extract rules from language.

Infants readily extract linguistic rules from speech. Here, we ask whether this advantage extends to linguistic stimuli that do not rely on the spoken modality. To address this question, we first examine whether infants can differentially learn rules from linguistic signs. We show that, despite having no previous experience with a sign language, six-month-old infants can extract the reduplicative rule (AA) from dynamic linguistic signs, and the neural response to reduplicative linguistic signs differs from reduplicative visual controls, matched for the dynamic spatiotemporal properties of signs. We next demonstrate that the brain response for reduplicative signs is similar to the response to reduplicative speech stimuli. Rule learning, then, apparently depends on the linguistic status of the stimulus, not its sensory modality. These results suggest that infants are language-ready. They possess a powerful rule system that is differentially engaged by all linguistic stimuli, speech or sign.

Learning word order: early beginnings.

We examine the beginning of the acquisition of the relative order of function and content words, a fundamental but cross-linguistically highly variable aspect of grammar. A review of the existing empirical literature shows that infants as young as 8 months of age can distinguish between functors and content words, and have a rudimentary knowledge of the order of these two universal lexical categories in their native language. Furthermore, human adults and non-human animals such as rodents process the same linguistic information differently from infants, emphasizing the developmental relevance of bootstrapping function/content word order from surface cues available in the input. We discuss the implications of these findings for a synergistic view of language acquisition, considering how grammar acquisition interacts with word learning.

Behavioral and Imaging Studies of Infant Artificial Grammar Learning.

Artificial grammar learning (AGL) paradigms have proven to be productive and useful to investigate how young infants break into the grammar of their native language(s). The question of when infants first show the ability to learn abstract grammatical rules has been central to theoretical debates about the innate vs. learned nature of grammar. The presence of this ability early in development, that is, before considerable experience with language, has been argued to provide evidence for a biologically endowed ability to acquire language. Artificial grammar learning tasks also allow infant populations to be readily compared with adults and non-human animals. Artificial grammar learning paradigms with infants have been used to investigate a number of linguistic phenomena and learning tasks, from word segmentation to phonotactics and morphosyntax. In this review, we focus on AGL studies testing infants' ability to learn grammatical/structural properties of language. Specifically, we discuss the results of AGL studies focusing on repetition-based regularities, the categorization of functors, adjacent and non-adjacent dependencies, and word order. We discuss the implications of the results for a general theory of language acquisition, and we outline some of the open questions and challenges.

Finding Phrases: The Interplay of Word Frequency, Phrasal Prosody and Co-speech Visual Information in Chunking Speech by Monolingual and Bilingual Adults.

The audiovisual speech signal contains multimodal information to phrase boundaries. In three artificial language learning studies with 12 groups of adult participants we investigated whether English monolinguals and bilingual speakers of English and a language with opposite basic word order (i.e., in which objects precede verbs) can use word frequency, phrasal prosody and co-speech (facial) visual information, namely head nods, to parse unknown languages into phrase-like units. We showed that monolinguals and bilinguals used the auditory and visual sources of information to chunk "phrases" from the input. These results suggest that speech segmentation is a bimodal process, though the influence of co-speech facial gestures is rather limited and linked to the presence of auditory prosody. Importantly, a pragmatic factor, namely the language of the context, seems to determine the bilinguals' segmentation, overriding the auditory and visual cues and revealing a factor that begs further exploration.

Finding phrases: On the role of co-verbal facial information in learning word order in infancy.

The input contains perceptually available cues, which might allow young infants to discover abstract properties of the target language. Thus, word frequency and prosodic prominence correlate systematically with basic word order in natural languages. Prelexical infants are sensitive to these frequency-based and prosodic cues, and use them to parse new input into phrases that follow the order characteristic of their native languages. Importantly, young infants readily integrate auditory and visual facial information while processing language. Here, we ask whether co-verbal visual information provided by talking faces also helps prelexical infants learn the word order of their native language in addition to word frequency and prosodic prominence. We created two structurally ambiguous artificial languages containing head nods produced by an animated avatar, aligned or misaligned with the frequency-based and prosodic information. During 4 minutes, two groups of 4- and 8-month-old infants were familiarized with the artificial language containing aligned auditory and visual cues, while two further groups were exposed to the misaligned language. Using a modified Headturn Preference Procedure, we tested infants' preference for test items exhibiting the word order of the native language, French, vs. the opposite word order. At 4 months, infants had no preference, suggesting that 4-month-olds were not able to integrate the three available cues, or had not yet built a representation of word order. By contrast, 8-month-olds showed no preference when auditory and visual cues were aligned and a preference for the native word order when visual cues were misaligned. These results imply that infants at this age start to integrate the co-verbal visual and auditory cues.

Subject relative clauses are not universally easier to process: Evidence from Basque.

Studies from many languages consistently report that subject relative clauses (SR) are easier to process than object relatives (OR). However, Hsiao and Gibson (2003) report an OR preference for Chinese, a finding that has been contested. Here we report faster OR versus SR processing in Basque, an ergative, head-final language with pre-nominal relative clauses. A self-paced reading task was used in Experiments 1 and 2, while ERPs were recorded in Experiment 3. We used relative clauses that were ambiguous between an object or subject-gap interpretation and disambiguated later in the sentence. The results of Experiments 1 and 2 showed that SR took longer to read than OR in the critical disambiguating region. In addition, Experiment 3 showed that SR produced larger amplitudes than OR in the P600 window immediately after reading the critical disambiguating word. Our results suggest that SR are not universally easier to process. They cast doubts on universal hypotheses and suggest that processing complexity may depend on language-specific aspects of grammar.