Eivind Nessa Torgersen
University of Reading
This presentation reports on an ongoing study of the effects of sentence context and local acoustic structure on the online processing of the voicing contrast in L2 with language learners. The purpose of this study is to examine if online speech processing in L2 is different from processing in L1. The study is within the cross modal semantic priming paradigm and examined how subjects categorised a target word which was embedded in various sentential conditions: biased and neutral semantic contexts and a control condition where the target word in isolation was embedded in pink noise. Using the same experimental technique, previous research has found immediate effects of sentence context in identification tasks with L1 speakers (Borsky et al., 1998), but not in lexical decision tasks (Borsky et al., 2000). The subjects are Norwegian learners of English at two different levels of knowledge of the L2 language and the visual stimuli were presented in two different temporal conditions, direct offset and slightly delayed in relation to the prime. The target stimuli were from a 6-step GOAT to COAT continuum differing only in the temporal cue for VOT. Subjects were told to indicate if the target word matched the visual probe or not by pressing one of two buttons on a response box while paying attention to both sound and the meaning of the unfolding sentence. A computer program recorded responses and response times. The sentences were recorded on a CD where one channel went to the subject (via headphones) and a tone on the second channel (inaudible to the subjects) went to the computer and triggered the visual display and start of the clock. Preliminary analysis of the data shows an effect of both semantic context and VOT on phoneme categorisation. Biased sentential contexts have faster response times if the VOT values and corresponding target words are congruent with the sentence context. Response times for the neutral sentential context and pink noise condition fall throughout the continuum from low to high VOT values, which reflects the acoustic structure of the target word. There is a shift in the categorisation towards the sentence bias for low and high VOT values depending on the sentence bias. Neutral sentential condition and pink noise have identifications in an area between the biased contexts, which clearly indicates the effect of context. Categorisations of extreme VOT values are more congruent in the delayed probe position. No other effects have been found for this condition. The target words embedded in pink noise have faster responses compared to the words embedded in sentential contexts.
References
Borsky, S., Tuller, B., & Shapiro, L. (1998). "How to milk a coat:" The effects of semantic and acoustic information on phoneme categorization. Journal of the Acoustical Society of America, 103 (5), 2670-2676.
Borsky, S., Shapiro, L., & Tuller, B. (2000). The temporal unfolding of local acoustic information and sentence context. Journal of Psycholinguistic Research, 29, 155-168.