Department of Psychology, University of Toronto, Toronto, Ontario, Canada.
Ear Hear. 2010 Aug;31(4):555-66. doi: 10.1097/AUD.0b013e3181daae5a.
The available research indicates that cochlear implant (CI) users have difficulty in differentiating talkers, especially those of the same gender. The goal of this study was to determine whether child CI users could differentiate talkers under favorable stimulus and task conditions. We predicted that the use of a highly familiar voice, full sentences, and a game-like task with feedback would lead to higher performance levels than those achieved in previous studies of talker identification in CI users.
In experiment 1, 21 CI users aged 4.8 to 14.3 yrs and 16 normal-hearing (NH) 5-yr-old children were required to differentiate their mother's scripted utterances from those of an unfamiliar man, woman, and girl in a four-alternative forced-choice task with feedback. In one condition, the utterances incorporated natural prosodic variations. In another condition, nonmaternal talkers imitated the prosody of each maternal utterance. In experiment 2, 19 of the child CI users and 11 of the NH children from experiment 1 returned on a subsequent occasion to participate in a task that required them to differentiate their mother's utterances from those of unfamiliar women in a two-alternative forced-choice task with feedback. Again, one condition had natural prosodic variations and another had maternal imitations.
Child CI users in experiment 1 succeeded in differentiating their mother's utterances from those of a man, woman, and girl. Their performance was poorer than the performance of younger NH children, which was at ceiling. Child CI users' performance was better in the context of natural prosodic variations than in the context of imitations of maternal prosody. Child CI users in experiment 2 differentiated their mother's utterances from that of other women, and they also performed better on naturally varying samples than on imitations.
We attribute child CI users' success on talker differentiation, even on same-gender differentiation, to their use of two types of temporal cues: variations in consonant and vowel articulation and variations in speaking rate. Moreover, we contend that child CI users' differentiation of speakers was facilitated by long-term familiarity with their mother's voice.
现有研究表明，人工耳蜗（CI）使用者在区分说话者方面存在困难，尤其是在区分同性说话者时。本研究旨在确定儿童 CI 用户在有利的刺激和任务条件下是否能够区分说话者。我们预测，使用高度熟悉的声音、完整的句子以及带有反馈的游戏式任务将比以前在 CI 用户的说话者识别研究中取得的更高的性能水平。
在实验 1 中，21 名年龄在 4.8 至 14.3 岁的 CI 用户和 16 名正常听力（NH）5 岁儿童需要在包含反馈的四选一强制选择任务中区分他们母亲的脚本语音与陌生男性、女性和女孩的语音。在一种情况下，发音包含自然韵律变化。在另一种情况下，非母亲说话者模仿每个母亲语音的韵律。在实验 2 中，实验 1 中的 19 名儿童 CI 用户和 11 名 NH 儿童在随后的一次任务中返回，需要在包含反馈的二选一强制选择任务中区分他们母亲的语音与陌生女性的语音。同样，一种情况有自然韵律变化，另一种情况有母亲模仿。
实验 1 中的儿童 CI 用户成功区分了他们母亲的语音与男性、女性和女孩的语音。他们的表现比年幼的 NH 儿童差，后者表现接近上限。在自然韵律变化的情况下，儿童 CI 用户的表现优于模仿母亲韵律的情况下。实验 2 中的儿童 CI 用户区分了他们母亲的语音与其他女性的语音，并且他们在自然变化的样本上表现优于模仿样本。
我们将儿童 CI 用户在区分说话者方面的成功，甚至在区分同性说话者方面的成功归因于他们使用了两种类型的时间线索：辅音和元音发音的变化以及说话速度的变化。此外，我们认为，儿童 CI 用户对说话者的区分是由于他们长期熟悉母亲的声音。
