Icelandic plus English: language differentiation and functional categories in a successively bilingual child

Abstract

This thesis investigates the formal and functional properties of the linguistic knowledge of a young bilingual child 'Katla' who successively acquires Icelandic (L1, from birth) and English (L2, from age 1:3). I present new longitudinal natural speech data which I collected in both Icelandic and English from Katla at regular intervals. Audio-recordings were made roughly three times per month at age 1 ;0-4;7 and transcribed in adapted CHILDES/CHAT format. Using a generative framework, I analyse Katla's data qualitatively and quantitatively, focusing on her morphology and syntax during the period 1;6-3;6: determiners and word order in nominals, copula constructions, progressive constructions, imperatives, negation, verb placement, verb inflections, auxiliaries, and periphrastic do. Katla's development is compared with monolingual English-speaking and Icelandic-speaking children, and, where applicable, with other bilinguals. Particular attention is paid to early grammar differentiation and cross- language influence, and to the relationship between child language and input (construction types and frequencies). The empirical results are evaluated in the light of current theories of language acquisition and generative approaches to syntax. Katla's first multi-word combinations (1;6) show productive use of functional morphology (determiners, copulas). Early on, there is evidence of movement into the DP, IP and CP domains, indicating continuity of these functional categories. Moreover, translational equivalents, language-specific functional morphemes and language-specific word orders in Katla's Icelandic and English bear evidence of early language differentiation in successive child bilingualism. The longitudinal development of morpho-syntax largely progresses along separate lines for Katla's two languages; there is no cross- language influence as regards head parameter and movement parameter settings. Some construction transfer occurs where L1 and L2 linear orders are similar. Ensuing implications for transfer and (de)learnability are addressed.