Vowels in development: Speech sound perception in Dutch infants with and without a family risk of dyslexia

Abstract

This dissertation investigates the development of speech sound categories in infants with (FR) and without (NFR) a family risk of dyslexia. There were four research questions. The first was whether NFR infants would show evidence in favor of perceptual attunement. NFR infants aged 6-, 8- and 10 months were tested on their discrimination performance of a native (a:/ - /e:/) and non-native (English /ɛ/ - /æ) contrast. They showed evidence of discriminating the native contrast. For the non-native contrast, a decline in discrimination performance was found between the 6- and 8-month-old infants. These findings align with the theory of perceptual attunement. However, the 10-month-olds showed robust evidence of discriminating the non-native contrast. These findings show that the ability to discriminate a non-salient non-native contrast is not lost after the process of perceptual attunement has emerged. The results suggest that the performance of the infants depends on the applied experimental design, the age of the infants and the salience of the contrast. The second question focused on whether it was possible to classify individual NFR infants as (non-)discriminators. Infants’ discrimination performance of the native contrast was analyzed with a frequentist and Bayesian hierarchical approach. The latter approach was found to be a fruitful approach for individual classification. These findings call for assessment of individual discrimination in longitudinal studies and in studies that investigate the relation between early speech perception and later language or literacy skills to establish the true value of individual assessment. The third question investigated whether FR infants differed in their developmental discrimination trajectory from their NFR peers. Native (a:/ - /e:/) and non-native English (/ɛ/ - /æ/) speech perception was investigated in 6-, 8- and 10-month-old FR and NFR infants. Results were analyzed at the individual (Bayesian hierarchical modeling) and group level (frequentist modeling). Results of the NFR infants showed the same U-shaped pattern as described above: The 6- and 10-month-olds discriminated the non-native contrast. NFR and FR infants discriminated the native contrast. The FR infants did not show evidence of discriminating the non-native contrast at any of the ages. These findings suggest that the FR infants have a subtle delay in the development of their speech sound categories. Finally, the fourth question concerned distributional phonetic learning. This is a learning mechanism that has been proposed to account for perceptual attunement. It was investigated whether 8-month-old FR and NFR infants could discriminate the English /ɛ/ - /æ contrast after being exposed to either a unimodal or bimodal frequency distribution. There were no clear differences between no-FR and FR infants on this task. Furthermore, there was no difference in discrimination outcomes for infants exposed to the unimodal (which could lead to only one phoneme category) and bimodal condition (which could lead to two phoneme categories). These findings call for further research on distributional learning. Together, the findings point towards the dynamic nature of speech sound development and the differences that appear in this development between groups (NFR and FR) and individual infants.