Signoret Carine, Johnsrude Ingrid, Classon Elisabet, Rudner Mary
Linnaeus Centre HEAD, Swedish Institute for Disability Research, Department of Behavioural Sciences and Learning, Linköping University.
School of Communication Sciences and Disorders and Department of Psychology, University of Western Ontario.
J Exp Psychol Hum Percept Perform. 2018 Feb;44(2):277-285. doi: 10.1037/xhp0000442. Epub 2017 May 29.
The perceptual clarity of speech is influenced by more than just the acoustic quality of the sound; it also depends on contextual support. For example, a degraded sentence is perceived to be clearer when the content of the speech signal is provided with matching text (i.e., form-based predictability) before hearing the degraded sentence. Here, we investigate whether sentence-level semantic coherence (i.e., meaning-based predictability), enhances perceptual clarity of degraded sentences, and if so, whether the mechanism is the same as that underlying enhancement by matching text. We also ask whether form- and meaning-based predictability are related to individual differences in cognitive abilities. Twenty participants listened to spoken sentences that were either clear or degraded by noise vocoding and rated the clarity of each item. The sentences had either high or low semantic coherence. Each spoken word was preceded by the homologous printed word (matching text), or by a meaningless letter string (nonmatching text). Cognitive abilities were measured with a working memory test. Results showed that perceptual clarity was significantly enhanced both by matching text and by semantic coherence. Importantly, high coherence enhanced the perceptual clarity of the degraded sentences even when they were preceded by matching text, suggesting that the effects of form- and meaning-based predictions on perceptual clarity are independent and additive. However, when working memory capacity indexed by the Size-Comparison Span Test was controlled for, only form-based predictions enhanced perceptual clarity, and then only at some sound quality levels, suggesting that prediction effects are to a certain extent dependent on cognitive abilities. (PsycINFO Database Record
语音的感知清晰度不仅受声音的声学质量影响；它还取决于上下文支持。例如，当在听到退化句子之前为语音信号的内容提供匹配文本（即基于形式的可预测性）时，退化的句子会被感知为更清晰。在此，我们研究句子层面的语义连贯（即基于意义的可预测性）是否会提高退化句子的感知清晰度，如果是，其机制是否与通过匹配文本进行增强的机制相同。我们还会探讨基于形式和意义的可预测性是否与认知能力的个体差异有关。20名参与者听取了通过噪声声码器处理后清晰或退化的口语句子，并对每个句子的清晰度进行评分。这些句子的语义连贯程度有高有低。每个口语单词之前要么是同源的印刷单词（匹配文本），要么是无意义的字母串（不匹配文本）。通过工作记忆测试来测量认知能力。结果表明，匹配文本和语义连贯都显著提高了感知清晰度。重要的是，即使在退化句子之前有匹配文本，高连贯性也会提高其感知清晰度，这表明基于形式和意义的预测对感知清晰度的影响是独立且累加的。然而，当控制由大小比较跨度测试所索引的工作记忆容量时，只有基于形式的预测提高了感知清晰度，而且仅在某些声音质量水平上如此，这表明预测效果在一定程度上依赖于认知能力。（PsycINFO数据库记录
