The perception and production of Mandarin citation tones by prelingually deaf adults




Chen, Yu

Journal Title

Journal ISSN

Volume Title



While considerable attention has been paid to the study of speech perception and production by prelingually deaf children, little is known about prelingually deaf adults' performance in such tasks many years after rebuilding hearing and using spoken language. For prelingually deaf people who have Standard Chinese as their target language, because tonal information is hard to process by hearing devices due to technological limitations, it is especially important to investigate whether they can perceive and produce Mandarin tones correctly. This dissertation aimed to contribute to this knowledge by investigating Mandarin tone perception and production by three participant groups, namely the HA (hearing aid) group, the CI (cochlear implant) group, and the NH (normal hearing) group, through three experiments---synthesized tone perception, lexical tone perception, and lexical tone production. In the experiment of synthesized tone perception, we addressed whether prelingually deaf people could categorically perceive synthesized Mandarin tones, using identification and discrimination tasks. The results showed that while not all NH participants perceive all the synthesized tones categorically, the NH group surpassed the deaf groups both in tone identification and discrimination. Inside the deaf groups, the HA group performed better than the CI group: while a few HA participants could categorically perceive the synthesized tones, almost none of the participants in the CI group could do this. Thus, the results of this experiment might indicate that: first, Mandarin tones are not categorically perceived, or at most quasi-categorically perceived, even by NH people; second, it is hard to process fine-grained tone information even for prelingually deaf people who have abundant experience using spoken language; third, the CI devices do not convey the acoustic details needed to perceive tone as well as the HA devices do. In the experiment of lexical tone perception, we checked the deaf participants' performance in identifying Mandarin lexical tones by analyzing their mouse movements during the decision-making process. The results showed that both of the deaf groups could identify Mandarin lexical tone with quite high degrees of accuracy (around 70\%) under the distractions of competing tones and segmental distractors, indicating that the groups' performance in this experiment was much better than in the experiment of synthesized tone perception. In addition, the CI participants reached the same level as the HA participants in identifying Mandarin lexical tones, revealing that the CI participants were able to gather valuable tonal information indirectly from real human-produced speech sounds. However, the deaf groups still performed much worse than the NH group. The results showed that the deaf participants were vulnerable to the effect of target tones and of rhyme complexity in the syllables that these tones were embedded in: they performed much worse in identifying T2T3 and T2T4 than other tone pairs, and in identifying tones in nasal rhymes, compared to other rhyme types. In the experiment of lexical tone production, we investigated how deaf participants behaved in Mandarin tone production tasks using acoustic analysis and subjective assessment with multiple judges. The results confirmed that the deaf participants could produce Mandarin citation tones quite well, and the CI participants even performed a little better than the HA participants in tone production. Nevertheless, although the deaf groups had set up similar tone patterns as the NH group, the deaf participants still performed worse than their NH counterparts. Similar to the results of lexical tone perception, the deaf participants' performance was also impacted by target tone and rhyme complexity. In the current study, T2 and T3 (especially T3) were much harder to produce than T1 and T4 for the deaf participants. Tones were also harder to produce in syllables with nasal rhymes although the impact of rhyme complexity was not as obvious as in the lexical tone perception experiment. Overall, the current study indicated that, while all three groups performed much better in perceiving and producing Mandarin lexical tones than in perceiving synthesized tones, the NH group performed much better than the two deaf groups in all three experiments; inside the two deaf groups, although the CI group performed much worse than the HA group in the synthesized tone perception experiment, the two groups performed similarly both in the lexical tone perception and the production experiments. Compared with the performance of the NH group, the performance of the deaf groups revealed that the acoustic characteristics of tones themselves, the types of rhymes the tones are embedded in, and the different hearing devices were important factors that impact the prelingually deaf adults' Mandarin tone perception and production. Under the theoretical framework of the speech chain theory, these results demonstrated that the underlying mechanisms responsible for prelingually deaf Mandarin-speaking adults' challenges in tone perception and production were their deficits of information processing in the three levels (the acoustic level, the physiological level, and the linguistic level) over the speech chain. That is, because the deaf participants could not clearly hear the acoustic signals associated with the tones (the acoustic level), they showed weaker mental representations for these tones (the linguistic level), and thus experienced more difficulties in realizing them both in perception and production (the physiological level). From the results of the synthesized tone perception experiment, we can see that the deaf groups, especially the CI group had difficulty perceiving synthesized tones categorically because they could not directly access tone acoustic information. In contrast, both deaf groups performed much better in the lexical tone perception and in the production experiments, indicating their hearing devices provided access to indirect acoustic information on tones and therefore unexpected benefits to these deaf participants in terms of tone perception. In particular, considering the CI device has worse performance in processing tone information than the HA device, and the CI participants have worse original hearing loss than their HA counterparts, the current results indicate that: (a) the CI device is a useful tool for rebuilding hearing and developing spoken language even when it comes to aspects of speech that one would not expect them to help with, and (b) other cues to tones exist beyond those assumed to be the most important for speech processing (duration, amplitude, overtones of rhymes, etc.). Future studies should explore what cues to tone are available to CI users, and to Mandarin speakers more generally, that help them acquire tonal structures in the language.



Prelingually deaf adults, Mandarin, Citation tones, Tone perception, Tone production