| Summary: | The process of second language acquisition is usually assumed to be affected by differences between the source language (L 1) and the target language (L2). Within the Minimalist approach (Chomsky 1995) crosslinguistic variation is accounted for in terms of differences in the values of features of functional categories instantiated in specific languages. Mature English differs from Hungarian in that its Tense category does not carry the [+f] feature characteristic of Hungarian focused sentences. Also, English lacks an additional functional projection dominating IP, namely F(ocus)P(hrase), which hosts focused, wh-, and negative operators in Spec,FP and attracts the verb or adjectival predicate into its head in order to satisfy spec-head agreement. It follows that English learners of Hungarian will have . to instantiate a new functional category FP and reset the values of the Tense category in their IL grammar. In this thesis we account for the difficulties faced by adult English learners of Hungarian by adopting the hypothesis that the two main classes of features have distinct learnability properties. It has been suggested that interpretable features (among them phi-features of nouns as well as [+wh] and [+f] features) are acquired easier than non-interpretable features (such as features responsible for V2 word order, resumptive pronouns, verbal inflection and nominal case morphology, as well as verb-movement associated with the Focus Projection in Hungarian). We demonstrate that this effect is also found -in our English-Hungarian interlanguage data. We show that even though L2 learners manage to prepose wh, focus and negative operators, they have continued difficulties with the accompanying verb-movement properties of Hungarian. This is reminiscent of the difficulties we find in child L 1 language acquisition of Hungarian. However, we argue that learnability factors have to be complemented by considerations about the nature of the target language input L2 learners receive. We propose that the nature of the TL input accounts for the differences between child and adult learners of Hungarian. It is well known that robust data (i.e. simple, salient and frequently occurring sentences) are required for the acquisition of correct feature-specifications of a target language. Infrequent data may cause a delay in the process of establishing L2 feature specifications and result in incomplete representations. Ambiguous data, on the other hand, are Iikely to ultimately result in divergent L2 representations at near native level. Testing these predictions in a study of acceptability judgements of adult English-speaking learners of Hungarian, we show that adult English speaking learners of Hungarian have difficulties in acquiring double wh- and double focus constructions as well as focused infinitives, long and partial operator movement in Hungarian. It is demonstrated that in the case of double wh- and double focus constructions native speakers' intuitions are indeterminate/optional, therefore the data L2 learners receive are not robust, leading to optionality in learners' interlanguage grammars. Although enjoying categorical judgements in native grammars, the nature of the input is similarly non-robust in the case of focused infinitives as well as long and partially extracted operator sentences. This is argued to lead to the difficulties L2 learners exhibit with respect to these structures. In the face of non-robust target language data learners are found to fall back on L 1 values and/or to resort to general learning strategies, such as overgeneralization and analogy.
|
|---|
