Slide Slam J7 Sandbox Series
Different time-course of activating tonal alternations in the production of Mandarin Tone 3 sandhi words: Evidence from reaction time survival analysis
Xiaocong Chen1, Caicai Zhang1, Yiya Chen2; 1The Hong Kong Polytechnic University, 2Leiden University Center for Linguistics and Leiden Institute for Brain and Cognition
Phonological entities (e.g., segments or lexical tones) may exhibit systematic changes conditioned by the phonological context. However, existing speech production models do not sufficiently specify how different phonological variants are activated during speech production We report results of a study aiming to shed light on the time-course of the activation of underlying and surface representations for phonological tonal variants. Our empirical base is Mandarin Tone 3 sandhi, a phonological alternation whereby a low-dipping tone (underlying form) changes to a rising tone, which resembles the lexical Tone 2, in the surface form when followed by another third tone. Previous research using the picture-word interference paradigm and stimulus onset asynchrony (SOA) manipulation suggests that both underlying and surface representations (Tone 3 and a Tone 2-like T3 sandhi variant) are activated but with different time course during sandhi word production (Nixon et al., 2015). This study thus aims to conceptually replicate the findings of Nixon et al. (2015), and to gain further insight into the time-course of activation via novel survival analysis (Reingold & Sheridan, 2014). Twenty-five native Standard Mandarin speakers were asked to produce disyllabic target Tone 3 sandhi words (e.g., 雨伞, yu3san3, ‘umbrella’) upon seeing a picture while their speech onset latencies were recorded. Each picture was preceded by a visual distractor, presented simultaneously with its spoken form auditorily. The distractor consisted of four types: a T3 identical-morpheme distractor (the same character as in the target word, e.g., 雨 yu3), a Tone 3 hetero-morpheme distractor (a different character sharing the underlying tone 3, e.g., 语 yu3), a Tone 2 distractor (a different character sharing the surface tone 2, e.g., 鱼 yu2), and a control distractor (a different character carrying either Tone 1 or Tone 4 with no tonal overlap, e.g., 预 yu4). Our results showed that both the Tone 3 and Tone 2 distractors yielded significantly shorter speech onset latencies than the control distractor, suggesting the activation of both tonal variants in production. Survival analysis on RT distributions revealed an earlier facilitatory effect of the Tone 3 distractors (321ms and 457 ms for the T3 identical- and hetero-morpheme distractor respectively) than the Tone 2 distractor (521 ms), confirming the different time course of the activation of the two tonal variants (Nixon et al., 2015). Furthermore, the latency differences between the T3 identical- and hetero-morpheme distractors revealed a significant effect of morphological identity. Further exploratory analysis suggests possible interactions of the effect of the distractors with the targets’ lexical familiarity and entropy of the first character. Our results offer important insights into the time course and factors that condition the encoding of phonological alternations during speech production. References: Nixon, J. S., Chen, Y., & Schiller, N. O. (2015). Multi-level processing of phonetic variants in speech production and visual word processing: evidence from Mandarin lexical tones. Lang. Cogn. Neurosci., 30(5), 491–505. Reingold, E. M., & Sheridan, H. (2014). Estimating the divergence point: A novel distributional analysis procedure for determining the onset of the influence of experimental variables. Front. Psychol., 5, 1432.