Assessing the role of orthographic versus uniquely auditory input acquiring new L2 segments. 7èmes Rencontres Internationales du Réseau Français de Phonologie, University of Aix-en-Provence

While there is no doubt as to the positive contribution of written input in helping second language (L2) learners parse and acquire syntactic and morphological structure, the extent to which orthographic representation may facilitate the acquisition of new segments is less clear. For example, in languages such as French and English which lack an exceptionless one-to-one grapheme-phoneme correspondence, spelling-based pronunciation errors are often attested (e.g. Altenberg & Vago 1983, Young-Scholten 1995). One recent study which bears on this question is Steele (2002). In this experiment, 13 beginner Mandarin-speaking learners of French were tested on their production of stop+rhotic clusters as well as singleton rhotic onsets via a vocabulary-learning task. The words to be learned were presented aurally accompanied by a corresponding image; no orthographic input was given. As concerns the clusters, the learners ' realizations were generally non-target-like. In voiced contexts (e.g. brebis, baudrier), the rhotic was typically realized as one of the voiceless fricatives [x], [X] or [h]. In voiceless clusters (e.g. traîneau, macramé), all of the learners deleted the rhotic to varying degrees. Steele argues that deletion in this context is related to the learners ' misanalysis of /“ / in the absence of orthographic input. Specifically, the L2 learners parse the devoiced rhotic of a stop+rhotic cluster as aspiration (e.g. Target [tXEno]; learner analysis [tHEno]). Such an analysis is particularly logical for native speakers of Mandarin, as the voicing contrast in this language is one between voiceless plain an