Amodal phonological abstraction in infants, vowel acoustic features in their mothers’ infant-directed speech, and their relationships to later vocabulary size

Abstract

Phonological abstraction in infants is a critical aspect of language development. A recent study found that Dutch-speaking adults adopted from Korea before 6 months learned to perceive a Korean consonant feature contrast not used in Dutch and generalised it to other places of articulation and to speech production, better than native Dutch speakers without Korean exposure (Choi, Broersma, et al., 2017; Choi, Cutler, et al., 2017). This finding implies that phonological abstraction occurs before 6 months, preceding language-specific attunement in speech perception by 6-10 months. This thesis investigates that inference directly in infants, with studies of their phonological abstraction abilities, vowel acoustic hyper-articulation and hyper-dispersion in their mothers’ infant-directed speech (IDS), and their combined impact on later vocabulary development. Three specific objectives were set for this thesis. The first objective was to determine at which age(s) phonological abstraction is evident in infancy and how it might develop over age. The second objective was to assess to what extent, if any, IDS vowels may facilitate or otherwise impact on the development of phonological abstraction. The third objective was to evaluate what impact phonological abstraction and IDS vowels might have on receptive vocabulary development. These were achieved by conducting three experiments. The first experiment used a cross-sectional design, where infants under 6 months participated in an auditory word-learning task involving two artificial languages in which the consonants had contrasting places of articulation (POA: labials vs. coronals). A subsequent visual (talking face) test assessed whether the infants had formed an amodal phonological abstraction of the POA difference, that is, generalised the learned mini-languages to the visual modality. Results suggest that at this early age, infants had acquired the amodal phonological abstraction of the articulatory differences between the artificial languages. Experiment 2 included a longitudinal investigation to examine developmental changes in amodal phonological abstraction, retesting a subset of participants from Experiment 1 at 7 months and 10 months. Additionally, we examined whether infants' performance at any age predicted receptive vocabulary size at 13 months. While infants exhibited amodal phonological abstraction at 4 months, this skill did not persist at the later ages, and no correlations were found between this skill at any age and later receptive vocabulary size. Experiment 3 assessed acoustic hyper-articulation and hyper-dispersion of the corner vowels /i u a/ in IDS produced by the mothers of Experiment 2 infants in structured laboratory play sessions at 4, 7, 10 and 13 months, and examined whether these IDS acoustic features predicted receptive vocabulary size at 13 months or phonological abstraction skill at any age. Acoustic hyper-articulation in IDS was significant at 7 months, and vowel dispersion was consistently greater in IDS than adultdirected speech for all vowels, except /u/ at 7 months. Surprisingly, neither acoustic measure correlated with 13-month receptive vocabulary. However, phonological abstraction skill at 4 months correlated significantly with acoustic hyper-articulation at 7 months. This research provides valuable insights into the complex interplay between early phonological abstraction, maternal speech characteristics, and subsequent vocabulary development in infants.

Date of Award	2024
Original language	English
Awarding Institution	Western Sydney University
Supervisor	Catherine Best (Supervisor)