Learning to Accommodate Variation in Speech Input

学习适应语音输入的变化

• 批准号: 0921362
• 负责人: Lori Holt
• 金额: $ 50万
• 依托单位: Carnegie-Mellon University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2009
• 资助国家: 美国
• 起止时间: 2009-10-01 至 2013-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0921362&HistoricalAwards=false
• 关键词: Learning; Accommodate; Variation; Speech; Input

To become fluent native speakers and listeners, children ultimately must perceive and produce speech that is consistent with the local dialect. For example, English spoken in Calgary, Canada sounds slightly different than English spoken in Pittsburgh, Pennsylvania. The variation in these two English environments is patterned and, in learning language, children must acquire these regularities. Seemingly at odds with this, children must also maintain some flexibility in mapping speech sounds to meaning. For example, infants sometimes encounter talkers whose speech deviates quite drastically from their community's dialectal norms, as when addressed in English by a non-native speaker from Italy. Thus, while learning the normative patterns of their local dialect, children also must maintain enough cognitive flexibility to accommodate talkers who do not fall within these norms. How do infants and toddlers accomplish these seemingly competing tasks in language learning? Answering this question will reveal general characteristics of children's learning at multiple time scales: from spontaneous interpretation of fleeting speech sounds, to short-term adaptation to deviations from expectation, to longer-term learning of the regularities of the local dialect. This research maps the range of acoustic variability to which infants are exposed in their native speech environment in different social interactions with adults. It will create a detailed acoustic speech production corpus with which to understand the challenges confronting infant language learners. The research will also study how infants' and toddlers' speech perception is shaped by regularities of the local dialect, revealing the mechanisms of long-term sensitivity to dialect norms and short-term adaptation to deviations from dialect norms. It will also investigate how toddlers' growing vocabularies influence adaptation to variability in spoken language.The research takes an innovative cross-disciplinary approach to these questions. Understanding early development of the knowledge of the sound structure of English and its interaction with developing vocabularies will contribute to our understanding of typical language development. It will also further our understanding of accent development in multilingual learning situations and early communicative deficits. An important outcome of the project will be the dissemination of a speech corpus, providing a rich acoustic sample of speech for other researchers to use in studying English dialects.

为了成为流利的母语使用者和倾听者，儿童最终必须感知并产生与当地方言一致的语音。例如，在加拿大卡尔加里说的英语听起来与在宾夕法尼亚州匹兹堡说的英语略有不同。这两种英语环境的差异是有模式的，在学习语言时，儿童必须获得这些模式。似乎与此不一致的是，儿童在将语音映射到意义时也必须保持一定的灵活性。例如，婴儿有时会遇到说话的人，他们的语言与他们社区的方言规范大相径庭，比如来自意大利的非母语人士用英语说话。因此，在学习当地方言的规范模式的同时，孩子们还必须保持足够的认知灵活性，以适应不符合这些规范的说话者。婴幼儿是如何完成这些看似竞争性的语言学习任务的呢？对这个问题的探讨将揭示儿童在多个时间尺度上学习的一般特征：从对短暂语音的自发解释，到对偏离预期的短期适应，再到对当地方言的长期学习。本研究绘制了婴儿在与成人的不同社会互动中暴露于其母语环境中的声学变化范围。它将创建一个详细的声学语音生产语料库，以了解婴儿语言学习者面临的挑战。本研究还将研究当地方言的重复如何塑造婴幼儿的言语感知，揭示对方言规范的长期敏感性和对偏离方言规范的短期适应机制。本研究还将探讨幼儿词汇量的增长如何影响对口语变化的适应。了解英语语音结构知识的早期发展及其与词汇发展的相互作用将有助于我们理解典型的语言发展。这也将进一步加深我们对多语言学习情况下口音发展和早期交际缺陷的理解。该项目的一个重要成果将是传播一个语音语料库，为其他研究人员研究英语方言提供丰富的语音声学样本。