Phonological Cascade in Speech Production

描述（由申请人提供）：本研究探讨语音产生过程中不同语音过程相互作用的程度。语音产生理论通常假设语音加工的两个不同阶段。首先，词汇音韵学过程从长期记忆中检索单词形式的粗粒度表征。然后语音过程产生更细粒度的音系结构。尽管理论一致认为存在两个这样的加工阶段，但它们在相互作用的程度上存在分歧。这里提出的实验验证了这些阶段相互作用的一个具体建议：语音级联假说。这一假设提出，虽然进程之间的相互作用是存在的，但它是有限的。级联激活通过允许源自词汇语音处理的影响（例如，词频等词级变量）来支持相互作用。由于语音到词汇音位过程的反馈受阻，交互作用受到严重限制。这种对相互作用的限制阻止了源自语音加工的影响（例如，发音相似性等细粒度结构的影响）直接影响词汇语音加工。两个系列的实验验证了这一假设。第一项研究考察了在词汇、语音或语音加工方面有后天缺陷的个体的表现。第二项研究在神经完整的参与者中实验诱发的语言错误。语音级联假说预测，在词汇语音过程中产生的语音错误（无论是通过实验还是通过神经损伤）应该对单词水平变量（如词频）敏感，但由于缺乏反馈，不受语音结构的细粒度方面（如发音相似性）的影响。相反，由于级联，语音处理过程中产生的语音错误应该对这两种变量都很敏感。

DESCRIPTION (provided by applicant): This research examines the degree to which different phonological processes in speech production interact with one another. Theories of speech production typically assume two distinct stages of phonological processing. First, lexical phonological processes retrieve coarse-grained representations of word form from long term memory. Phonetic processes then generate more fine-grained aspects of phonological structures. Although theories agree that two such processing stages are present, they disagree on the degree to which they interact with one another. The experiments proposed here test one specific proposal for the interaction of these stages: the phonological cascade hypothesis. This hypothesis proposes that although interaction between processes is present, it is restricted. Cascading activation supports interaction by allowing effects that originate within lexical phonological processing (e.g., word-level variables such as word frequency) to influence phonetic processing. Interaction is crucially limited by blocking feedback from phonetic to lexical phonological processes. This restriction on interaction prevents effects originating within phonetic processing (e.g., effects of fine-grained structure such as articulatory similarity) from directly influencing lexical phonological processing. Two series of experiments test this hypothesis. The first examines the performance of individuals with acquired deficits to lexical phonological or phonetic processing. The second examines experimentally induced speech errors in neurologically intact participants. The phonological cascade hypothesis predicts that speech errors generated within lexical phonological processes (either experimentally or via neurological damage) should show sensitivity to word-level variables (e.g., word frequency) but, due to the lack of feedback, be uninfluenced by fine-grained aspects of phonological structure (e.g., articulatory similarity). In contrast, due to cascade, speech errors generated during phonetic processing should be sensitive to both types of variables. Project Narrative: By addressing a relatively unexplored issue in how the sound structure of words is processed to support the production of speech, the experiments proposed here will provide the basis for richer, more detailed theories of language production. Furthermore, increasing our understanding of how the adult language processing system breaks down in aphasia may assist the development of more refined therapies for acquired speech production impairments.