Boonen Nathalie, Kloots Hanne, Gillis Steven
Computational Linguistics & Psycholinguistics Research Centre, University of Antwerp, Antwerp, Belgium.
Int J Pediatr Otorhinolaryngol. 2020 Jun;133:109956. doi: 10.1016/j.ijporl.2020.109956. Epub 2020 Feb 15.
Studies on speech intelligibility usually focus on either individual words, sentences or longer sequences of speech. Since these different kinds of speech samples can be judged using different methodologies, a difference in the reported intelligibility scores can either be due to the difference in the type of sample or methodology. The present study compares the speech intelligibility of seven-year-old children with a cochlear implant (CI) with that of their normally hearing (NH) peers. The first aim is to compare the intelligibility of short sentences and longer sequences of speech using the same methodology. Secondly, it has been suggested that i.a. advances in CI technology and changes in candidacy criteria may have had a positive influence on the intelligibility of children with CI. In order to assess this issue, the intelligibility of seven-year-olds implanted ten years apart will be compared.
The speech of two cohorts of early implanted children with CI (n = 16) and NH peers (n = 16), matched on several criteria but implanted in different years, was collected. More specifically, short and longer samples were selected from recordings of a wordless picture book retelling. Both types of samples were judged on a visual analogue scale by 105 inexperienced listeners.
The results showed that the intelligibility of children with CI was lower than that of their NH peers. Moreover, longer samples were significantly more intelligible than short samples for both groups. No significant effect was found between the two cohorts of children with CI. However, the intelligibility of three out of four children with CI was found to be on a par with that of their NH peers, indicating a large amount of variability between subjects.
Listeners ascribed higher intelligibility to longer samples than to short samples, despite the fact that both types were extracted from the same recordings and listeners followed the same judgement procedure. The amount of context thus facilitated speech decoding. No effect of the calendar year of implantation was found, suggesting that e.g., the evolution in CI technology did not have a significant impact on CI users' intelligibility after six years of device use.
关于言语可懂度的研究通常聚焦于单个单词、句子或更长的言语序列。由于这些不同类型的言语样本可使用不同方法进行评判，所报告的可懂度分数存在差异可能是由于样本类型或方法的不同。本研究比较了七岁人工耳蜗植入（CI）儿童与其听力正常（NH）同龄人之间的言语可懂度。首要目标是以相同方法比较短句和更长言语序列的可懂度。其次，有人提出，例如CI技术的进步和入选标准的变化可能对CI儿童的可懂度产生了积极影响。为评估此问题，将比较相隔十年植入的七岁儿童的可懂度。
收集了两组早期植入CI儿童（n = 16）及其NH同龄人（n = 16）的言语，两组在多个标准上匹配但植入年份不同。更具体地说，从一本无字图画书复述的录音中选取了短样本和长样本。两类样本均由105名无经验的听众在视觉模拟量表上进行评判。
结果显示，CI儿童的可懂度低于其NH同龄人。此外，对于两组而言，长样本的可懂度明显高于短样本。在两组CI儿童之间未发现显著影响。然而，发现四分之三的CI儿童的可懂度与他们的NH同龄人相当，这表明个体之间存在很大差异。
尽管两类样本均取自相同录音且听众遵循相同评判程序，但听众认为长样本的可懂度高于短样本。因此，语境量有助于语音解码。未发现植入年份的影响，这表明例如，CI技术的发展在设备使用六年之后对CI使用者的可懂度没有显著影响。
