The ultimate attainment of L2 learners varies considerably; some learners achieve completely native-like performance -and arguably native-like competence-while others 'fossilize', their endstate competence and performance differing, often considerably, from native speakers of the L2. In this paper, we contrast two positions which argue that failure to achieve native-like competence reflects effects of L1 representations on the interlanguage grammar (ILG). These accounts differ as to whether non-native attainment reflects L1 syntactic or phonological properties. According to the Representational Deficit Hypothesis (is no (syntactic) parameter resetting in adult L2; consequently, speakers can never acquire functional categories or features that are required by the L2 but absent in the L1. In other words, there are permanent effects of L1 syntactic representations on the ILG. In contrast, we argue that it is transfer of L1 prosodic constraints that affects IL representations, with consequences for the production of inflectional morphology and function words, particularly during the course of development but also in the endstate. In this paper, we investigate the L2 acquisition of English tense morphology by Mandarin speakers; we show that Mandarin speakers have few problems in interpreting English tense appropriately and that their production of the morphology shows effects of stimulus type which is unexpected on syntactic accounts. We further argue that, while there are circumstances in which L2 speakers are able to adapt L1 prosodic structures in order to represent L2 functional morphology, there are certain situations where this will never be possible.
II Representational Deficit HypothesisAccording to the RDH, the only uninterpretable formal features that can be realized in the ILG are those that are represented in the L1. In the case of the feature [±past], the lack of overt tense morphology in Mandarin is taken to indicate a lack of the corresponding formal feature in that language. The prediction, therefore, is that Mandarin-speaking learners of English are unable to acquire [±past], leading to difficulties in realizing and interpreting tense morphology. The IL representation, then, is permanently defective in the sense that it lacks features required to appropriately represent the L2.Investigating this hypothesis, Hawkins and Liszka (2003) report that, in oral production tasks, two Chinese-speaking subjects had greater difficulty realizing English t/d in regular past tense contexts (63% suppliance) than in the case of past participles (100% suppliance). Contra Lardiere (1998Lardiere ( , 2003, they argue that the problem with tense cannot be attributed to a difficulty with final consonant clusters: although many inflected forms ends in clusters, the same subjects * We would like to thank two anonymous reviewers for comments and the audiences at BUCLD 29, GALANA 1 and the University of Hamburg for questions related to this work. We would also like to thank the following research assistants for their participat...