Syntactic processing across socioeconomic status: Linking input to comprehension

在不同的社会经济地位(SES)中，父母的投入在数量和质量上都不同，这影响了孩子早期的语言能力和后来的学习表现。虽然这些关系已经牢固地建立起来，但它们的机制还没有被很好地理解。对输入(例如，总单词)和结果(例如，词汇量)的综合测量跟踪了多年学习的结果，但这忽略了儿童如何通过与句子的反复接触来获得语言知识。为了直接观察这一过程，本项目结合了高密度输入录音、眼球跟踪方法和计算建模，以分离儿童用来解释句子的线索，以及策略如何随输入差异而变化。在科学层面上，这项研究为比较不同社会群体的系统差异提供了一个框架。虽然现有的习得模型假设便利性样本(即，SES较高的儿童)在所有儿童中提供了一个合适的学习代理，但本项目检验了另一种假设，即与SES相关的输入差异会影响儿童在听句子时注意到的信息。孤立儿童之间的理解差异将揭示学习者在理解新句子时必须掌握的策略。在社会层面上，学习和交流有赖于对句子的准确解读。描述儿童的理解策略可以更好地预测未来的结果，并提供了针对早期语言影响学习成绩的多种途径的方法。这一框架将使其他输入不同于主流同龄人的人群受益，例如双语学习者、人工耳蜗者。为了理解在父母输入和语言结果之间起中介作用的句子水平过程，本研究考察了语言体验如何影响来自不同SES背景的5岁和6岁儿童的理解策略。这个项目测试了这样一种假设，即输入属性决定了哪些线索对解释句子含义是有用的。对于SES较高的群体，更大的输入量支持对细粒度单词属性的学习，而单词多样性的增加增强了在解释句子时获取详细模式的需求。然而，对于SES较低的群体，数量和多样性的减少使粗粒度模式更有用，这既是因为它们需要更少的输入来学习，也是因为当单词出现在规范的句子框架中时，它们意味着相似的含义。这个项目使用高密度的话语录音来测量父母输入中的特定单词和句子框架模式，并通过眼球跟踪实验区分儿童的句子解释。计算模型将这些数据结合在一起，得出儿童在理解过程中对细粒度单词知识和粗粒度框架知识的依赖程度的估计。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Across socioeconomic status (SES), parental input varies in quantity and quality, and this impacts children's early language abilities and later academic performance. While these relationships are firmly established, their mechanisms are not well understood. Aggregated measures of input (e.g., total words) and outcomes (e.g., vocabulary size) track the consequences of multiple years of learning, but this ignores how children acquire language knowledge through iterative encounters with sentences. To observe this process directly, the present project combines high-density input recordings, eye-tracking methods, and computational modeling to isolate the cues that children use to interpret sentences and how strategies vary with input differences. At a scientific level, this research provides a framework for comparing systematic variation across social groups. While existing models of acquisition assume that the convenience sample (i.e., higher-SES children) offers a suitable proxy of learning in all children, this project tests an alternative hypothesis that SES-related differences in input affect what information children attend to when listening to sentences. Isolating between-child differences in comprehension will reveal strategies that learners must acquire to interpret new sentences. At a societal level, learning and communication depend on accurately interpreting sentences. Describing children's comprehension strategies enables better predictions of future outcomes and offers methods for targeting multiple pathways by which early language can affect academic performance. This framework will benefit other populations whose input differs from mainstream peers, e.g., dual language learners, cochlear implant users.To understand the sentence-level processes that mediate between parental input and language outcomes, the present project examines how language experience influences comprehension strategies in 5- and 6-year-olds from varying SES backgrounds. This project tests the hypothesis that input properties regulate what cues are informative for interpreting sentence meanings. For higher-SES groups, greater input quantity supports learning of fine-grained word properties, and increased word diversity enhances the need to access detailed patterns when interpreting sentences. For lower-SES groups, however, decreased quantity and diversity make coarse-grained patterns more useful, both because they require less input to learn and imply similar meanings when words occur in canonical sentence frames. This project measures word-specific and sentence-frame patterns in parental input using high-density recordings of utterances, and distinguishes children's sentence interpretations using eye-tracking experiments. Computational models combine these data to derive child-specific estimates of reliance on fine-grained word knowledge versus coarse-grained frame knowledge during comprehension.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.