Max Planck Institute for Psycholinguistics, Nijmegen, the Netherlands; Leiden University Centre for Linguistics, Leiden University, Leiden, the Netherlands.
Max Planck Institute for Psycholinguistics, Nijmegen, the Netherlands; Donders Institute for Brain, Cognition and Behaviour, Radboud University, Nijmegen, the Netherlands.
Cognition. 2025 Jan;254:105999. doi: 10.1016/j.cognition.2024.105999. Epub 2024 Oct 30.
When producing a sentence, speakers must rapidly select appropriate words in the correct order. Models of lexical access often assume that this lexical selection process is competitive and that each word is chosen from a set of competing candidates. Therefore, an important theoretical issue is which factors constrain this choice. Speech error evidence suggests that word class plays a decisive role here and that lexical access is, at any point in time, restricted to words that fit the part of the grammatical structure of the sentence that is being constructed. Using a novel version of the picture-word interference paradigm, Momma, Buffinton, Slevc, and Phillips (2020, Cognition) showed experimentally that word class indeed constrains lexical selection. Specifically, in speakers of American English, action verbs (as in she's singing) competed with semantically related action verbs (as in she's whistling), but not with semantically related action nouns (as in her whistling). Similarly, action nouns only competed with semantically related action nouns, but not with action verbs. As this pattern has important implications for models of lexical access and sentence generation, we conducted a conceptual replication of the study in Dutch. In two experiments, we found a semantic interference effect, but, contrary to the original study, no evidence for a word class constraint. In accounting for these results, we propose that word class constraints on lexical selection are graded rather than categorical, and that, at least for verbs and action nouns, the marking for word class is clearer in English than in Dutch.
在生成句子时，说话者必须快速选择正确顺序的合适单词。词汇访问模型通常假设这个词汇选择过程是竞争的，每个单词都是从一组竞争候选词中选择的。因此，一个重要的理论问题是哪些因素限制了这种选择。言语错误证据表明，词类在这里起着决定性的作用，词汇访问在任何时候都限于适合正在构建的句子语法结构部分的单词。Momma、Buffinton、Slevc 和 Phillips（2020，认知）使用一种新颖的图片-单词干扰范式，实验表明词类确实限制了词汇选择。具体来说，在美国英语使用者中，动词（如 she's singing）与语义相关的动词（如 she's whistling）竞争，但与语义相关的动作名词（如 her whistling）不竞争。同样，动作名词只与语义相关的动作名词竞争，而不与动作动词竞争。由于这种模式对词汇访问和句子生成模型具有重要意义，我们在荷兰进行了该研究的概念复制。在两项实验中，我们发现了语义干扰效应，但与原始研究相反，没有证据表明存在词类约束。在解释这些结果时，我们提出词汇选择的词类约束是渐变的而不是分类的，并且至少对于动词和动作名词，英语中的词类标记比荷兰语更清晰。
