Previous studies of adult bilinguals have shown that cognates (translation equivalents similar in sound and spelling) are translated faster than non-cognates and different representations for the two categories in bilingual memory have been suggested Stewart 1994, van Hell andde Groot 1998). Assuming that bilingual children's representations are similar to those of adults, effects of form similarity between words should also be observed. This paper examines formsimilar nouns in the early lexical development of a bilingual German/English child aged 1;11±2;9 as well as effects of form similarity in picture naming and translation in two groups of German/English children aged 8±9. Form similarity here differs from the cognate status of a word in that it implies similarity of sound only. Considering the way hearing children acquire words, it seemed necessary to restrict the similarity of words to this modality. Similarly, the presentation of items in the translation tasks was auditory. The results show an effect of form similarity in early lexical development, whereby form-similar words occurred frequently in the beginning of the observation period in both languages and were more likely to have a translation equivalent in the child's English. In the translation task, form similarity resulted in lower latencies for both language directions. The results thereby con®rm that form similarity affects representations in both adult and child learners.

This study explores parallels and differences in the comprehension of wh-questions and relative clauses between early foreign-language (FL) learners and monolingual children. We test for (a) effects of syntactic first-language (L1) transfer, (b) the impact of input on syntactic development, and (c) the impact of individual differences on early FL syntactic development. We compare the results to findings in child second language (L2) naturalistic acquisition and adult FL acquisition. Following work on adult FL acquisition, we carried out a picture-based interpretation task with 243 child FL learners in fourth grade at different regular, partial, and high-immersion schools in Germany plus 68 monolingual English children aged 5 to 8 years as controls. The child FL learners display a strong subject-first preference but do not appear to use the L1 syntax in comprehension. Input differences across different schools affect overall accuracy, with students at high-immersion FL schools catching up to monolingual performance within 4 years of learning. Finally, phonological awareness is implicated in both early FL learning and naturalistic child L2 development. These findings suggest that early FL development resembles child L2 acquisition in speed and effects of individual factors, yet is different from adult FL acquisition due to the absence of L1 transfer effects.

This paper addresses the morphosyntactic dcvclopment of two bilingual children and the issues raised by the controversy behveen the single system and the separate development hypotheses. Set within a generative grammar frame\vork, evidence on GermanlEnglish and LatvianlEnglish is presented from the earliest stages of language development: for the GermanlEnglish child from 2;O to 2;6, for the LatvianlEnglish child from 1;3 to 1;ll. For the GermanlEnglish data, the results show early word order patterns which are in line with the language-specific orders ofboth languages. In the LatvianlEnglish data, there is correct inflectional marking on nouns and verbs in Latvian from the earliest stagcs,contrary to the paucity ofsuch marking in English. Hence the evidence from both children supports the view that the bilingual child separates the two languages from the beginning (Separate Development Hypothesis). In addition, the data show a developmental lead-lag pattern whereby functional categories emerge first in the more inflected language (German and Latvian) and later in English.

The present study investigates children's syntactic and pragmatic processing when specifying referents presented in short video clips. Within Relevance theory, the assumption of 'optimal relevance' implies that utterances are intended to involve the least processing effort on the part of the listener. In the present context, lexically specified NPs are assumed to be more in line with optimal relevance than pronouns. Subjects were 48 normally developing children aged 3;4-8;10 and 30 SLI children aged 5;1-8;9, divided into a low and a normal MLU group. Children's responses were coded according to levels of pragmatic processing and syntactic positions. Normally developing children's referent specifications were found to be increasingly relevant with increasing age. Differences between SLI and normal children were only found for the low MLU group with SLI who used fewer pronouns than the younger children, thereby showing that syntactic limitations alone cannot account for children's specification of referents.

No abstract