Encoding of contextual tonal alternations in word production in Teochew

Context-dependent tonal alternations occur abundantly in many Chinese dialects, but how they are encoded in speech production is still unclear. Previous research on Mandarin third-tone sandhi production has suggested that the encoding of sandhi tonal variant may involve an online computation process in the phonological encoding before articulation, but it is unknown whether this also applies to the tone sandhi in other Chinese dialects. In the current study, we employed an odd-one-out implicit priming task to investigate the encoding process of a similar tonal alternation phenomenon in Teochew (a Min Chinese dialect), whereby Tone 2 (T2) and Tone 3 (T3) exhibit different sandhi tonal variants in the non-phrasal-final positions depending on the following tones. Our two experiments showed that compared to the tonal unrelated condition, the word production latencies were significantly facilitated when the underlying tonal category was shared, but the only mismatch of surface sandhi tonal variant did not disrupt the production latencies. These results indicate that different contextual-dependent sandhi tonal variants in Teochew are probably represented as the same underlying representation in the lexicon, but their realization is more likely to occur in the course of articulation in anticipation of the subsequent tonal context. This stands in contrast to the findings from Mandarin third-tone sandhi, suggesting different processing mechanisms could be engaged for different types of tonal alternations.