Optimizing word retrieval in Multilink across stimulus types and tasks
Keywords
No Thumbnail Available
Authors
Issue Date
2020-02-14
Language
en
Document type
Journal Title
Journal ISSN
Volume Title
Publisher
Title
ISSN
Volume
Issue
Startpage
Endpage
DOI
Abstract
Multilink is a localist-connectionist network for word retrieval in mono- and bilinguals (Dijkstra et al., 2019). The latest version of this model (v.2.0.) and has an enriched lexicon and lateral inhibition as key updates. In the present study, we will explore the performance of Multilink v.2.0. by simulation three experimental studies involving different tasks and stimulus types. First, we simulated Dutch and English lexical decision using data of the Dutch Lexicon Project (Keuleers, Diependaele, Brysbaert, 2010) and the English Lexicon Project (Balota, Hutchinson et al., 2007). Second, word translation data (Pruijn, 2015) were simulated for Dutch-English bilinguals in two directions. Third, we simulated English and Dutch (not mixed) word naming data (de Groot, 2002). These studies involved words that differed in length (3-8 letter words), cognate status, and frequency status. The simulated model cycle times and the empirical reaction times were strongly correlated with Dutch decision, r(878) = .614, as well as for English decision, r(863) = .555. Even higher correlations were observed between the cycle times and the word translation data, r(128) = .606 (forward translation) and r(124) = .69 (backward translation). English word naming data showed also a high correlation with the cycle times, r(371) = .0474. All correlations were significant (p < .0001). The high correlations across tasks and stimulus types demonstrate that the new version of Multilink shows excellent performance on word recognition, lexical semantic processing, and word production. As this model used the same lexicon and the same parameter settings, our findings contributed to the existence of a shared processing mechanism.
Description
Citation
Supervisor
Faculty
Faculteit der Sociale Wetenschappen