Cooper W E
Ann N Y Acad Sci. 1983;405:48-63. doi: 10.1111/j.1749-6632.1983.tb31617.x.
The perception of fluent speech involves both acoustic-driven and knowledge-driven sources of information. Acoustic sources that directly convey information about higher-order linguistic characteristics include the suprasegmental aspects of speech, associated with the acoustic cues of fundamental frequency, duration, and intensity. From the standpoint of cochlear prostheses, interest in listeners' perception of these suprasegmentals is heightened by the fact that these properties are typically among the best preserved sources of speech information in the sensorineural hearing-impaired. Studies with normal listeners indicate that suprasegmentals provide cues to such linguistic factors as word stress, syntactic structure, and semantic interpretation. In addition, they supply information about many nonlinguistic characteristics of the speaker. Research on knowledge-driven processes indicates that normal listeners utilize knowledge of lexical, syntactic, and semantic constraints of the language to guide the segmentation of speech into word-like units and to select among candidate word choices. These findings are related to research with hearing-impaired listeners and prosthesis testing.
流畅言语的感知涉及声学驱动和知识驱动的信息源。直接传达有关高阶语言特征信息的声学源包括言语的超音段方面，与基频、时长和强度的声学线索相关。从人工耳蜗的角度来看，由于这些属性通常是感音神经性听力受损者中保存最好的言语信息源，因此听众对这些超音段的感知备受关注。对正常听众的研究表明，超音段为诸如单词重音、句法结构和语义解释等语言因素提供线索。此外，它们还提供有关说话者许多非语言特征的信息。对知识驱动过程的研究表明，正常听众利用语言的词汇、句法和语义限制知识来指导将言语分割成类似单词的单元，并在候选单词选择中进行筛选。这些发现与对听力受损听众的研究以及假体测试有关。
