Functional morphemes and mechanisms of early language processing

早期语言处理的功能语素和机制

• 批准号: RGPIN-2014-05212
• 负责人: Shi, Rushen
• 金额: $ 2.48万
• 依托单位: Université du Québec à Montréal
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2016
• 资助国家: 加拿大
• 起止时间: 2016-01-01 至 2017-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=611502
• 关键词: Functional; morphemes; mechanisms; early; language

The distinction between lexical and functional categories is the most fundamental one in natural languages. Lexical categories include nouns, verbs, adjectives and adverbs. Functional categories include determiners, auxiliaries, tense endings, etc. Lexical categories are also called content words, as they carry the main weight of meaning in utterances. In contrast, the primary role of functional elements is to indicate grammatical relations across words. Children start talking around age one. They produce one-word speech for months and then word combinations by 18-20 months of age. Speech below age two contains content words, typically missing functional elements. This characteristic may suggest that young children do not know functional categories. However, new research from several labs including mine using perceptual measures shows that babies start processing functional items early, long before they can produce them. In my early studies I found that 0-6-month-old babies distinguish function words versus content words. I later showed that infants before age one perceive and represent specific functional items in their native language, and use them to segment adjacent words. Shortly after age one, infants understand the grammatical relations across different functional elements and use functors to categorize content words. These findings reveal that infants have more sophisticated grammatical knowledge than previously assumed. In the next cycle of my NSERC grant, I will address 2 issues in early morpho-syntactic acquisition: 1) the mechanisms underlying infants’ acquisition of function words and their grammatical relation with content words; 2) the initial learning of bound functional elements (e.g., endings such as –ing in English and –er in French) and their morphological relations with the stem (e.g., walk_walk-ing_walk-ed). For 1), we will test the hypothesis that the combined properties of weak word forms and high frequency for function words (relative to content words) enable infants to perceive them as functional elements and to use them to discover grammatical properties of content words (e.g., using determiners to categorize the adjacent word as a noun, “a/the dog”, “a/the cat”). To precisely study these properties, we will create artificial languages containing novel words, with careful manipulation of word forms and frequencies. Perceptual experiments using these materials will be conducted with one-year-olds. For 2), novel words with a familiar French ending –er and with an artificial ending will be used to test how preverbal and early verbal infants learn such endings and the morphological variation of related words. Across experiments, varying training conditions with novel word stems and endings will be used to examine input factors. We will test the hypothesis that the output of distribution-based morphological learning guides infants’ subsequent learning of word meaning. That is, infants first learn that certain forms are morphologically related (e.g., walk-walking), which then enables them to expect all forms of this variation to have the same core meaning. This hypothesis differs from the standard view that children first learn the meaning of individual words before interpreting (at a much later age) their internal morphological structure. Both 1) and 2) will test more than the learning of specific items and their relations. They will test infant’ ability to generalize structures at the abstract level. The studies will yield a better understanding of the mechanisms underlying early language development, in particular, the initial acquisition of abstract morpho-syntactic structures. They will contribute important data to the larger debate on the nature and origin of syntactic knowledge in infants.

词汇范畴和功能范畴的区别是自然语言中最基本的区别。词汇类别包括名词、动词、形容词和副词。功能范畴包括限定词、助词、时态结尾等。词汇范畴也被称为实词，因为它们在话语中承载着主要的意义。相比之下，功能成分的主要作用是表明词与词之间的语法关系。 孩子们在一岁左右开始说话。它们会在几个月内发出一个单词的语音，然后在18-20个月大的时候进行单词组合。两岁以下的讲话包含实词，通常缺少功能成分。这一特征可能表明幼儿不知道功能范畴。然而，包括我在内的几个实验室使用知觉测量进行的新研究表明，婴儿很早就开始处理功能物品，远远早于他们产生它们。在我早期的研究中，我发现0-6个月大的婴儿区分虚词和实词。后来，我展示了一岁以前的婴儿能够感知并用母语表达特定的功能项目，并用它们来切分相邻的单词。一岁后不久，婴儿就能理解不同功能成分之间的语法关系，并使用函数词对实词进行分类。这些发现表明，婴儿拥有比之前假设的更复杂的语法知识。 在我的NSERC赠款的下一个周期，我将解决早期形态句法习得中的两个问题：1)婴儿习得虚词的机制及其与实词的语法关系；2)绑定功能元素的初始学习(例如，英语中的-ing和法语中的-er)及其与词干的形态关系(例如walk_walk-ing_walk-ed)。对于1)，我们将检验这样一种假设，即虚词(相对于实词)的弱词形式和高频属性的组合使婴儿能够将它们感知为功能元素，并使用它们来发现实词的语法属性(例如，使用限定词将相邻的词归类为名词、“a/狗”、“a/猫”)。为了准确地研究这些特性，我们将创造包含新词的人工语言，并仔细操纵词形和频率。使用这些材料的感知实验将在一岁的孩子身上进行。2)使用熟悉的法语结尾的新词和人工结尾的新词来测试语言前和早期语言的婴儿如何学习这种结尾以及相关单词的形态变化。通过实验，不同训练条件下的新词干和词尾将被用来检验输入因素。我们将检验这一假设，即基于分布的形态学习的输出指导婴儿随后的词义学习。也就是说，婴儿首先了解到某些形式在形态上是相关的(例如，走路)，这使他们能够期待这种变异的所有形式都具有相同的核心含义。这一假设不同于标准的观点，即儿童首先学习单个单词的含义，然后(在较晚的年龄)解释其内部形态结构。1)和2)将测试的不仅仅是对特定项目及其关系的学习。他们将测试婴儿在抽象水平上概括结构的能力。 这些研究将更好地理解早期语言发展的机制，特别是抽象形态句法结构的初始习得。他们将为关于婴儿句法知识的性质和来源的更广泛的辩论贡献重要的数据。