Contrastive tongue shapes of the three sibilant fricatives in Taiwan Mandarin read speech

Abstract

This study investigates the production of three sibilant fricatives in Taiwan Mandarin, dental [s], retroflex [?], and alveolopalatal [?], using ultrasound recording. Previous studies have pointed out that the contrast between dental [s] and retroflex [?] tends to be lost in connected speech in Taiwan Mandarin [1, 2]. We ask whether the merger could happen in read speech as well. Two Taiwan Mandarin speakers (1 male and 1 female) read a list of nonse words, which contained one of the three fricatives, followed by one of three vowels [a], [?], and [o]. Tongue shape was traced at the mid-point of each fricative and SSANOVA was used to compare the three fricatives. The analysis revealed two major points. (1) In read speech, both speakers made clearly distinguishable tongue shapes for all of the three fricatives, suggesting that speaking style plays a role in determining the likelihood of the merger; however, more data are required for further verification. (2) The amount of variation in tongue shape in different vowel contexts changed between speakers and fricatives. Speaker 1 showed a large amount of variation in the shape of the tongue body for dental [s], but not for retroflex [?] and alveolopalatal [?], while speaker 2 showed no such variation for any of the three fricatives. The pattern of variation made by speaker 1 conformed to previous observations that dental fricative requires an active engagement of the tongue tip but not the other part of the tongue, which remains flexible to coarticulate with the following vowels. Post-alveolar fricatives, however, require the active engagement of the whole tongue, which limits the possibility of coarticulation [3]. For both speakers, across vowel contexts, bunched tongue shape was observed for retroflex [?] and an advancement of the tongue root was observed for alveolopalatal [?].