The Role of Statistical Learning in Acquiring Syntactic Categories

描述（由申请人提供）：拟进行的实验将探讨统计学习在“名词”和“动词”等句法范畴习得中的作用。句法范畴知识在语言习得中的作用是普遍存在的；例如，范畴知识既有助于学习语法模式，也有助于早期的单词学习。虽然语言习得的许多方面依赖于句法范畴知识，但这些范畴的学习过程尚不清楚。本研究的目的是检验统计学习在句法范畴习得中起作用的假设。不同句法类别的词汇可以通过统计线索进行区分，例如它们的分布特征（即与其他词汇的共存）和语音特征（即它们的声音结构），第一个具体目的是评估婴儿在名词和动词类别习得过程中对这些线索的敏感性。如果在类别学习中使用了单词的统计属性，那么在类别习得的早期阶段，婴儿应该能够使用这些属性来识别新的类别成员。此外，婴儿确定新单词类别的能力应该反映出他们符合类别的主要统计特征的程度，对这些统计特征的训练应该有助于他们在更小的年龄使用。婴儿对新单词进行分类的能力将通过测试它们是否被映射到对象或动作指示物来评估。第二个具体目的是测试婴儿在类别学习中整合统计信息和语义信息的能力。句法范畴的习得包括对统计属性和语义属性的学习，这些线索在自然语言中是相互关联的。如果婴儿能够整合这些线索，它们之间的相关性应该有助于学习。婴儿将熟悉包含新类别的人工语言，并且统计和语义线索之间的相关性将被操纵。如果婴儿能够整合这些线索，他们应该能够更好地发现一个类别中单词意义的共性，并学习这些单词可能出现的语法结构，当线索相关时，而不是当线索不相关时。未来的研究将测试患有特殊语言障碍和其他发育性语言障碍的儿童（例如，晚说话者），他们表现出受损的单词学习能力和对识别句法类别的分布线索的敏感性降低。对儿童进行这些任务的测试将解决统计学习受损在多大程度上是这些障碍的一个促成因素。

DESCRIPTION (provided by applicant): The proposed experiments will investigate the role of statistical learning in the acquisition of syntactic categories such as "noun" and "verb". The role of syntactic category knowledge in language acquisition is pervasive; for example, category knowledge both facilitates learning grammatical patterns and facilitates early word learning. While many aspects of language acquisition rely on syntactic category knowledge, the process by which these categories are learned is unclear. The goal of the proposed research is to test the hypothesis that statistical learning plays a role in acquiring syntactic categories. Words from different syntactic categories are distinguishable by statistical cues, such as their distributional properties (i.e., cooccurrence with other words), and their phonological properties (i.e., their sound structure), the first specific aim is to assess infants' sensitivity to these cues in the acquisition of noun and verb categories. If words' statistical properties are used in category learning, infants should be able to use these properties to identify novel category members in the early stages of category acquisition. Moreover, infants' ability to determine the category of novel words should reflect the degree to which they conform to the predominant statistical characteristics of the category, and training on these statistical properties should facilitate their use at younger ages. Infants' ability to categorize novel words will be assessed by testing whether they are mapped to object or action referents. The second specific aim is to test infants' ability to integrate statistical and semantic information in category learning. Acquisition of syntactic categories encompasses learning about both statistical properties and semantic properties, and these cues are correlated in natural language. If infants can integrate these cues, correlations between them should facilitate learning. Infants will be familiarized with an artificial language containing novel categories, and the correlations between statistical and semantic cues will be manipulated. If infants can integrate these cues, they should be better able to detect commonalities in the meanings of words within a category, and to learn the grammatical structures in which these words can occur, when cues are correlated than when they are not. Future studies will test children with Specific Language Impairment and other developmental language disorders (e.g., late talkers), who exhibit impaired word learning abilities and reduced sensitivity to distributional cues identifying syntactic categories. Testing children on these tasks will address the extent to which impaired statistical learning is a contributing factor in these disorders.