Bilingual Speech Segmentation

双语语音切分

• 批准号: 6969761
• 负责人: DANIEL J WEISS
• 金额: $ 6.77万
• 依托单位: PENNSYLVANIA STATE UNIVERSITY, THE
• 依托单位国家: 美国
• 财政年份: 2005
• 资助国家: 美国
• 起止时间: 2005-06-01 至 2007-05-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6969761
• 关键词: age difference; behavioral /social science research tag; clinical research; cues; discrimination learning; human subject; infant human (0-1 year); language development; multilingualism; speech recognition; university student

DESCRIPTION (provided by applicant): The goal of this research is to gain insight into how language learners are able to form multiple representations when exposed to more than one language. Previous research has conflicted with respect to the age at which this ability is present (see Sebastian-Galles & Kroll, 2003). The proposed investigation will span infant and adult learners in order to understand whether the ability to form multiple representations is available to infants at 10 months of age and whether there are changes in processing over the course of development. Previous perceptual research investigating the development of bilingual language acquisition has typically focused on infants' abilities to discriminate between two different languages. The research proposed here employs a speech stream segmentation task that more directly assesses whether listeners are capable of forming multiple representations. Specifically, the experiments are designed such that participants must form separate representations for 2 artificial "languages" in order to correctly parse the languages and correctly identify the "words". Thus, this research also addresses a critical problem for the statistical learning literature. Namely, how can language learners correctly segment the input from multiple languages? If the statistics are combined across languages, incorrect segmentation of both languages could occur. A second goal of this research is to investigate the acoustic cues that may facilitate the formation of multiple representations. Language learners may face a difficult task of trying to determine when they are in the presence of multiple languages. Therefore, the task of the learner is to identify relevant features that denote a language change. In the proposed experiments, a series of cues will be manipulated in an effort to determine which features may be useful for cuing language learners to the presence of multiple languages. In sum, the proposed research will fill gaps in current knowledge regarding both bilingual language acquisition as well as statistical learning.

描述（由申请人提供）：本研究的目标是深入了解语言学习者如何能够形成多种表征时，暴露于一种以上的语言。先前的研究在这种能力出现的年龄方面存在冲突（见Eschertian-Galles & Kroll，2003）。拟议的调查将跨越婴儿和成人学习者，以了解是否有能力形成多个表征是提供给婴儿在10个月大，是否有变化的过程中的发展。以前的知觉研究调查双语语言习得的发展通常集中在婴儿区分两种不同语言的能力。这里提出的研究采用了语音流分割任务，更直接地评估听众是否能够形成多个表示。具体地，实验被设计为使得参与者必须形成2种人工“语言”的单独表示，以便正确地解析语言并正确地识别“单词”。因此，这项研究也解决了统计学习文献的一个关键问题。也就是说，语言学习者如何正确地从多种语言中分割输入？如果跨语言组合统计数据，则可能会发生两种语言的不正确分割。本研究的第二个目标是调查可能有助于形成多种表征的声学线索。语言学习者可能面临着一个困难的任务，试图确定他们何时在多种语言的存在。因此，学习者的任务是识别表示语言变化的相关特征。在拟议的实验中，一系列的线索将被操纵，以努力确定哪些功能可能是有用的提示语言学习者的存在多种语言。总之，拟议中的研究将填补空白，目前的知识双语语言习得以及统计学习。