Yacovone Anthony, Waite Briony, Levari Tatyana, Snedeker Jesse
Department of Psychology, Harvard University.
J Exp Psychol Gen. 2025 Mar;154(3):711-738. doi: 10.1037/xge0001677. Epub 2024 Dec 16.
It is well-established that people make predictions during language comprehension--the nature and specificity of these predictions, however, remain unclear. For example, do comprehenders routinely make predictions about which words (and phonological forms) might come next in a conversation, or do they simply make broad predictions about the gist of the unfolding context? Prior EEG studies using tightly controlled experimental designs have shown that form-based prediction can occur during comprehension, as N400s to unexpected words are reduced when they resemble the form of a predicted word (e.g., when expecting cake). One limitation, however, is that these studies often create environments that are optimal for eliciting form-based prediction (e.g., highly constraining sentences, slower-than-natural rates of presentation). Thus, questions remain about whether form-based prediction can occur in settings that more closely resemble everyday comprehension. To address this, the present study explores form-based prediction during naturalistic spoken language comprehension. English-speaking adults listened to a story in which some of the words had been altered. Specifically, we experimentally manipulated whether participants heard the original word from the story (), a form-similar nonword (), or a less-similar nonword (). Half of the target words were predictable given their context, and the other half were unpredictable. Consistent with the prior work, we found reduced N400s for form-similar nonwords () relative to less-similar nonwords ()-but only in predictable contexts. This study demonstrates that form-based prediction can emerge in naturalistic contexts, and therefore, it is likely to be a common aspect of language comprehension in the wild. (PsycInfo Database Record (c) 2025 APA, all rights reserved).
人们在语言理解过程中会进行预测,这一点已得到充分证实。然而,这些预测的性质和特异性仍不清楚。例如,理解者在对话中是否会常规性地预测接下来可能出现哪些单词(以及语音形式),或者他们只是对正在展开的语境的主旨进行宽泛的预测?先前使用严格控制的实验设计的脑电图研究表明,在理解过程中可以发生基于形式的预测,因为当意外单词与预测单词的形式相似时(例如,当预期是“蛋糕”时),对这些单词的N400反应会减少。然而,一个局限性是,这些研究通常创造出最有利于引发基于形式的预测的环境(例如,高度受限的句子、比自然语速慢的呈现速度)。因此,对于基于形式的预测是否能在更接近日常理解的环境中发生,仍存在疑问。为了解决这个问题,本研究探索了自然主义口语理解过程中的基于形式的预测。以英语为母语的成年人听了一个其中一些单词被改变的故事。具体来说,我们通过实验操纵参与者听到的是故事中的原始单词()、形式相似的非单词()还是不太相似的非单词()。一半的目标单词根据其上下文是可预测的,另一半是不可预测的。与先前的研究一致,我们发现相对于不太相似的非单词(),形式相似的非单词()的N400反应减少了——但仅在可预测的语境中。这项研究表明,基于形式的预测可以在自然主义语境中出现,因此,它很可能是日常语言理解中的一个常见方面。(PsycInfo数据库记录(c)2025美国心理学会,保留所有权利)