Learning morphophonological alternations across languages and populations

Young children are typically inconsistent in the use of grammatical morphemes - alternating between the correct forms and morpheme omissions/overgeneralisations. This period is protracted in children with language learning difficulties such as Specific Language Impairment (SLI), and in children acquiring languages with complex morphological systems (eg Russian). It has been variously claimed that children's early morpheme productions are constrained by their still developing phonological skills, limited vocabulary and understanding of syntax. Thus, a better understanding of the linguistic phenomena affecting grammatical morpheme acquisition is essential for modelling child language development and language assessment/intervention. This thesis focuses on the effects of morphophonological alternations on morpheme acquisition. In the first study I predicted that the existence of several formal representations of a morpheme (i.e., allomorphs) might delay its mastery, particularly in atypical populations. Using productions of English-speaking children with SLI, I demonstrated that segmental allomorphs (e.g., She runs) are mastered earlier than the syllabic ones requiring an additional syllable (e.g., She dresses). The second project followed up on these results using an experiment with nonce verbs. It showed that both typically developing (TD) 5-year-olds and those with SLI have similar difficulties in producing syllabic allomorphs of the past and present tense morphemes. The consistency of the results suggests a significant effect of allomorphy on morpheme production, i.e.,children’s difficulties are not rooted solely in their limited understanding of morphosyntax. The third project investigated similar phenomena in Russian, for which the acquisition of nominal declensions extends to school age. The awareness about vowel deletions and stress patterns was investigated in an experiment with real and nonce words in 4–7-year-olds. The results showed a growing sensitivity to morphophonological patterns with age, particularly apparent in real words. The latter suggests high lexicalisation of the vowel deletion pattern. However, stressed vowels were deleted/preserved with more accuracy, indicating that the various types of morphophonological patterns might have a joint effect during morpheme production. The consistency of the findings across languages and populations suggests that morphophonological alternations need to be taken into account during research and clinical intervention, particularly when studying atypical populations such as SLI.