Modeling the Stages of Pronoun Acquisition

该奖项是NSF社会、行为和经济科学博士后研究奖学金（SPRF）项目的一部分，由SBE的学习科学项目提供支持。SPRF计划的目标是为学术界、工业界或私营部门和政府的科学事业准备有前途的早期职业博士级科学家。SPRF奖励包括在知名科学家的赞助下进行为期两年的培训，并鼓励博士后进行独立研究。美国国家科学基金会寻求促进科学界各阶层的科学家，包括那些未被充分代表的群体的科学家，参与其研究项目和活动；博士后阶段被认为是实现这一目标的一个重要的专业发展阶段。每个博士后必须解决各自学科领域的重要科学问题。在加州大学欧文分校的丽莎·珀尔博士的赞助下，这个博士后奖学金奖支持一位研究儿童对代词理解的早期职业科学家。代词高度依赖语法信息和语境信息，是研究儿童多信息整合能力发展的理想测试案例。关于代词的形态、代词的形式和语篇连接词是如何单独决定代词指的是什么，语言学家已经有了相当完善的理论，但对于这些信息是如何在听者的头脑中组合起来的，我们知之甚少，而对于听者如何决定优先处理哪些信息，我们就更不了解了。由于这些线索中的一些是概率性的，这意味着它们倾向于但不需要特定的代词解释，而且它们在输入中的丰度在同一种语言的不同方言中是不同的，所以这个问题进一步复杂化了。了解儿童的线索整合机制是如何工作的，以及它对统计变化的反应，可以为语言变化和认知发展的理论提供信息。它还可用于诊断非典型发育，并为通过干预儿童投入的数量和质量来改善入学准备的努力提供信息。该项目包括一系列协调的实验、语料库和计算建模研究。第一阶段使用实验和语料库方法来确定儿童第一次对三种特定线索敏感的年龄。(i)词性一致形态，（ii）句内语义关系或语篇关系，以及（iii）代词形式（无效代词与显性代词），都被发现影响西班牙语代词的解释。第二阶段通过使用实验方法跟踪儿童在3岁、4岁和5岁时如何对这些线索进行相对排序来解决线索整合的问题。第三阶段解决的问题是儿童如何从他们的输入中提取这些排名，以及输入中的方言间差异是否会导致不同的排名结果。语料库工作将首先确定儿童接收的输入中每个线索的分布。接下来，将使用一个计算模型来估计分配给每个线索的权重，这是基于它们在输入中的分布和儿童在第二阶段观察到的排名行为。这个模型将提供一个精确的，可量化的建议，关于儿童如何衡量每个线索，以及这些权重在发展过程中如何变化。这种方法的新颖之处在于，它将来自不同理论领域（词法学、语篇语用学、社会语言学）的既有见解汇集在一起，展示它们如何在学习者的思想中共存，而且相互作用。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This award was provided as part of NSF's Social, Behavioral and Economic Sciences Postdoctoral Research Fellowships (SPRF) program and supported by SBE's Science of Learning program. The goal of the SPRF program is to prepare promising, early career doctoral-level scientists for scientific careers in academia, industry or private sector, and government. SPRF awards involve two years of training under the sponsorship of established scientists and encourage Postdoctoral Fellows to perform independent research. NSF seeks to promote the participation of scientists from all segments of the scientific community, including those from underrepresented groups, in its research programs and activities; the postdoctoral period is considered to be an important level of professional development in attaining this goal. Each Postdoctoral Fellow must address important scientific questions that advance their respective disciplinary fields. Under the sponsorship of Dr. Lisa Pearl at the University of California, Irvine, this postdoctoral fellowship award supports an early career scientist investigating children's understanding of pronouns. Pronouns are highly dependent on both grammatical information and contextual information, making them an ideal test case in which to study children's developing ability to integrate multiple pieces of information. Linguists have fairly well-developed theories about how pronominal morphology, pronominal form, and discourse connectives contribute individually to deciding what a pronoun refers to, but far less is understood about how these pieces of information are combined in the mind of the listener, and even less is known about how the child listener decides which pieces of information to prioritize. The question is further complicated by the fact that some of these cues are probabilistic, meaning that they favor but do not require certain pronoun interpretations, and by the fact that their abundance in the input varies across different dialects of the same language. Understanding how children's cue integration mechanism works and how it reacts to statistical variation can inform theories of language change and cognitive development. It can also be useful for diagnosing atypical development and inform efforts at improving school readiness through interventions in the quantity and quality of children's input. This project encompasses a series of coordinated experimental, corpus, and computational modeling studies. Phase 1 uses experimental and corpus methods to pinpoint the age at which children first become sensitive to three specific cues. (i) Verbal agreement morphology, (ii) intra-clausal semantic relations or discourse relations, and (iii) pronominal form (null vs. overt pronouns), have all been found to influence pronoun interpretation in Spanish. Phase 2 addresses the question of cue integration by tracking how children rank these cues relative to one another across ages 3, 4, and 5, using experimental methods. Phase 3 addresses the question of how children can potentially extract these rankings from their input, and whether inter-dialectal differences in the input lead to different ranking outcomes. Corpus work will first determine the distribution of each cue in the input that children receive. Next, a computational model will be used to estimate the weight assigned to each of these cues, based on their distribution in the input and children's observed ranking behavior from Phase 2. This model will provide a precise, quantifiable proposal for how children weight each cue and how these weights change over the course of development. The novelty of this approach lies in bringing together established insights from different theoretical domains (morphosyntax, discourse pragmatics, sociolinguistics) to show how they can not only coexist but also interact in the mind of the learner.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.