Presentation

Ashlie Pankonin¹, Alyson Abel²; ¹San Diego State University and University of California, San Diego, ²San Diego State University

Learning a new word involves incrementally constructing a lexical representation for the word, eventually reaching a point where the word’s representation is robust enough to be easily accessed and retrieved. A lexical representation is thought to at least include information about the word’s phonological form, semantic meaning, and grammatical features (e.g., word class). While it would seem that lexical representations with both form and meaning information would be more robust than form alone, studies have reported mixed findings. Word class has also been shown to affect the development of lexical representations, with greater difficulty typically shown for verbs than for nouns, but why these differences occur is not well understood. One reason for this obscurity might be that the robustness of a word’s representation is often measured using solely behavioral methods, which requires the representation to be highly developed. Electroencephalography (EEG), however, can provide insight into the process of lexical representation development, revealing representations that are still developing and otherwise inaccessible. This study investigates how the word class of a new word and identification of the word’s meaning during an incidental semantic learning task (ISLT) affect the robustness of the word’s representation. Behavioral and EEG data were collected from 53 English-monolingual adults as they completed an experimental ISLT followed by a word recognition task (WRT). Participants completed either the Noun (n = 31) or Verb (n = 22) condition, in which the only major difference was the word class of the target word. In the ISLT, participants heard 100 sentence triplets. All sentences in the triplet ended with the same nonword that replaced a target word. Half of the sentence triplets were constructed to support identification of the target word’s meaning and half were constructed to not. After each triplet, the participant was asked to provide a meaning for the nonword if they believed one existed. Each response was classified offline for meaning identification, disregarding whether meaning identification was supported or not. For the WRT, completed directly after the ISLT, participants heard 200 nonwords, 100 of which they had heard previously in the ISLT and 100 of which were novel, and they indicated whether they recognized each nonword. Behavioral analyses of lexical representation robustness via recognition accuracy revealed the unexpected finding that identifying a meaning for the nonword did not facilitate explicit recognition of it later, regardless of word class, F(5, 62.93) = 1.18, p = .331. The ongoing ERP analysis focuses on the N400 ERP component as a measure of semantic processing. We predict ERP patterns to diverge from the behavioral findings, revealing more robust representations for words, especially nouns, for which meanings were identified. Such findings would demonstrate the incremental process of word learning, support the argument that implicit measures such as EEG can better tap into lexical representations that are early in development and consequently less robust, and suggest that meaning identification and word class have subtle effects that might only be detected at implicit levels in the early stages of the creation of a lexical representation.

Topic Areas: Meaning: Lexical Semantics, Language Production