Doctoral Dissertation Research: Disentangling neural indices of implicit vs. explicit second language processing

鉴于在全球化的世界中，第二语言（L2）熟练程度的相关性日益增强，以及多语言可能给个人的社会经济、认知和学术成果带来的好处，提高第二语言学习是一个日益重要的领域。这一努力的一个关键部分涉及提高我们对学习的相对贡献的理解，即外显（即涉及有意识的意识）与内隐（即不涉及有意识的意识）。为了研究显性和隐性第二语言处理的相对贡献和关系，本项目将人工语言学习实验与最近开发的脑电图（EEG）机器学习分析技术相结合，脑电图（EEG）是一种神经语言学方法，可以以非侵入性方式检测与语言处理相关的大脑活动。这将为语言学习模型之间的理论争论提供信息，这些模型假设隐式/显式处理被强接口、弱接口或无接口分开。这种对内隐/外显语法处理的更好理解可能会通过建议如何平衡内隐方法（例如，自然暴露）和外显方法（例如，以语法为导向的练习）来最大限度地提高学习效果，从而为第二语言教学提供信息。实验参与者将完成一项语言学习任务，涉及四个带有隐藏语法规则的新单词（两个单词通常与生物同时出现，两个单词与非生物同时出现）。之后，参与者将回答汇报问题，以确定他们是否/何时意识到这一规则。先前的研究表明，即使没有规则的参与者也表现出学习的迹象，在违反规则的试验中，与遵守规则的试验相比，具有明显的神经活动和较慢的反应。该项目使用基于机器学习的分析技术扩展了此类研究，该技术可以评估不同神经认知事件在大脑活动的时间和位置上的相似程度。第一个分析测试了隐式（无规则）语法处理的神经标记是否出现在显式（规则感知）处理期间。第二个分析测试了内隐处理和外显处理在语言预测的发生上是否不同（这里定义为当参与者单独阅读有生命/非有生命编码的人工语言单词时，与处理有生命的单词相关的神经模式的发生，而不是非有生命的单词）。第三个分析测试了语言任务的反应时间是否与内隐和外显加工的神经标记的时间表现出一致的关系，以确定参与者的外部反应是由有意识还是无意识加工驱动的。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。

Given the increasing relevance of second language (L2) proficiency in a globalizing world as well as the benefits that multilingualism might confer for individuals' socioeconomic, cognitive, and academic outcomes, improving L2 learning is an area of growing importance. A critical part of this endeavor involves advancing our understanding of the relative contributions of learning that is explicit (i.e., involves conscious awareness) vs. implicit (i.e., does not involve conscious awareness). To investigate the relative contributions of and relationship between explicit and implicit L2 processing, this project combines an artificial language learning experiment with recently-developed machine learning analysis techniques for electroencephalography (EEG), a neurolinguistic method that can detect brain activity related to language processing in a non-invasive way. This will inform theoretical debates between models of language learning that posit that implicit/explicit processing are separated by a strong interface, a weak interface, or no interface. Such an improved understanding of implicit/explicit grammar processing might inform second language pedagogy by suggesting how implicit methods (e.g., naturalistic exposure) and explicit methods (e.g., grammar-oriented drills) should be balanced to maximize learning.Experiment participants will perform a language learning task involving four new words with a hidden grammatical rule (two words usually co-occur with living things and two with non-living things). Afterwards, participants will answer debriefing questions to determine if/when they became aware of the rule. Prior research indicates that even rule-unaware participants show signs of learning, in that rule-violating trials feature distinct neural activity and slower responses when compared to rule-following trials. This project extends such research with machine learning-based analysis techniques which can assess the degree to which different neurocognitive events are similar in the timing and location of brain activity. The first analysis tests whether neural markers of implicit (rule-unaware) grammar processing occur during periods of explicit (rule-aware) processing. The second analysis tests whether implicit and explicit processing differ in the occurrence of linguistic prediction (defined here as the occurrence of neural patterns related to the processing of living words as opposed to non-living words, when participants read the living/non-living-encoding artificial language word in isolation). The third analysis tests whether language task reaction times show a consistent relationship with the timing of neural markers of implicit and explicit processing, to determine whether participants’ external responses are driven by conscious vs. unconscious processing.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.