以聲學及事件相關電位探討人工電子耳兒童之聲調習得

Abstract

聲調在中文當中具有辨義的作用。然而，因為電子耳對於低頻聲音訊號處理仍有所限制，導致配戴電子耳的兒童學習聲調面臨諸多挑戰。目前已知聽障兒童若能早期植入且配戴時間長，對於聲調習得有所助益；此外，使用雙模式聆聽的兒童（一耳植入電子耳，另一耳配戴助聽器）可能因助聽器提供了電子耳缺乏的低頻訊息，而在聲調聽取上具有優勢。然而，目前仍少有研究直接比較僅配戴單側電子耳和雙模式聆聽兒童的聲調產出表現，故實驗一於中國山東省耳鼻喉醫院招募了16名電子耳兒童，測試聆聽模式和電子耳使用時間對聲調產出表現的影響，並根據配戴模式（雙模式/單側電子耳）和電子耳使用時間（二年/五年）進行分組，另有16位聽力正常的兒童與配戴五年電子耳的兒童進行聽齡和生理年齡配對，作為對照組。研究者評量受試者的聲調產出正確率，以及針對四個聲調輪廓進行了聲學分析。實驗一結果顯示，配戴五年電子耳的兒童在聲調產出的表現顯著優於配戴二年電子耳的兒童。此外，配戴五年雙模式的電子耳兒童能清楚呈現二三聲之聲學特徵，而配戴五年單側電子耳和兩年電子耳的兒童仍易混淆二三聲之轉折點。這表示配戴電子耳時長在聲調習得上有顯著幫助，且雙模式的聽覺模式可能在相似的聲調產出（二聲和三聲）提供額外的優勢。 即使長期配戴電子耳的兒童具有與正常聽力同儕相當的聲調產出表現，但由於早期的聽覺剝奪導致聽覺中樞系統發育異常，我們假設電子耳兒童的神經表徵可能與正常兒童不同。為了驗證這一假設，我們透過事件相關電位檢測電子耳兒童在感知聲調時的神經反應。首先，我們先進行實驗二作為前測，確認了在台灣招募的配戴五年電子耳的兒童亦能表現出和聽常同儕相當的聲調產出。接著，實驗三招募8位配戴電子耳五年的兒童以及29名生理年齡和聽齡匹配的正常聽力兒童，使用特異刺激典範（oddball paradigm）量測不匹配神經反應（Mismatch responses, MMRs），刺激音包含一/四聲及二/三聲對比組合。結果顯示，電子耳兒童在一/四聲對比中顯示p-MMR和LDN，在二/三聲對比中僅出現p-MMR。而聽齡和生理年齡配對的聽力正常兒童在一/四聲和二/三聲對比中都顯示出LDN。此結果表示長期配戴電子耳兒童的中樞聽覺系統處理聲調訊息仍不如正常聽力同儕成熟。總結來說，本研究結果指出長期配戴電子耳兒童可以表現出與正常聽力同儕相當的聲調產出，但他們的聽覺中樞系統在處理聲調輪廓仍存在劣勢。需要進一步研究以探討次要聲學線索如何幫助電子耳兒童習得並處理聲調。

Lexical tones play an important role in Mandarin because they differentiate a given syllable into different meanings. However, it can be a challenge for children with cochlear implants (CIs) to learn because lexical tones are distinguished by fundamental frequency contours, and CI devices allow limited transmission in the low-frequency range in which fundamental frequency information resides. It has been suggested that children with CIs implanted at an early age and with a longer CI experience perform better in the acquisition of Mandarin tones. Moreover, children with bimodal stimulation (CI plus a hearing aid [HA]) might have an advantage over those with CIs in tone perception due to their having access to much richer information in the low-frequency range. As few studies directly compare tonal performance in children with unilateral CI (without using HAs for the unimplanted ear) and children with bimodal stimulation, Experiment 1 compared the tone production between these two groups of CI children to determine whether hearing modality and length of CI experience affect children’s performance. Sixteen children with CIs were recruited in the Shandong Provincial ENT Hospital, China and were grouped according to hearing modality (bimodal stimulation vs. unilateral CI) and length of CI experience (2 years vs. 5 years). Children’s production accuracy was calculated, and the pitch contours of the four lexical tones were acoustically analyzed. The results revealed that children with five years of CI experience had better performance in tone production than those with only two years of experience. Furthermore, children with five years of bimodal stimulation experience could produce distinct Tone 2 and Tone 3 contours similar to their normal-hearing counterparts, whereas those with five years of unilateral CI experience and two years of CI experience could not. This suggests that CI experience played an important role in tone acquisition and bimodal stimulation may have an advantage in producing similar tones (rising tone and dipping tone). We further predict that even though children with longer CI experience might have the potential to produce tones comparable to those of normal-hearing children, the neural representation in CI children might still be different from that in normal children due to abnormal development of the central auditory system caused by auditory deprivation. To verify this hypothesis, we investigated CI children’s neural responses when listening to Mandarin tones through event-related potentials (ERPs). First, we conducted Experiment 2 to confirm that the CI children with five years of CI experience recruited in Taiwan could perform comparable tone production to normal-hearing counterparts like the results of Experiment 1. Then we designed an event-related potential (ERP) experiment to investigate the mismatch responses (MMRs) patterns in CI children in Experiment 3. Eight CI children were recruited five years after implantation, along with 29 chronological-age- and hearing-age-matched children with normal hearing. The passive oddball paradigm was used, including a Tone 1/Tone 4 contrast and a Tone 2/Tone 3 contrast as stimuli. Results showed that CI children had p-MMR and LDN in the Tone 1/Tone 4 contrast, and only p-MMR in the Tone 2/Tone 3 contrast. On the other hand, both groups of matched children with normal-hearing displayed LDN in Tone 1/Tone 4 and Tone 2/Tone 3 contrasts. This suggests that lexical tonal processing in children with longer CI experience might be less mature than that in their normal-hearing counterparts. Our findings illustrate that children with longer CI experience can perform tone production similar to their normal-hearing counterparts, but they still have disadvantages in processing lexical tones differing only in F0 contours. Further studies are needed to investigate whether and how secondary auditory help children with CIs acquire and process lexical tones well.