¿Qu´e significa ’fortune’? How Multilink simulates cognate processing
Keywords
Loading...
Authors
Issue Date
2022-06-20
Language
en
Document type
Journal Title
Journal ISSN
Volume Title
Publisher
Title
ISSN
Volume
Issue
Startpage
Endpage
DOI
Abstract
This project investigates if Multilink, a computational model for bilingual
word recognition, can simulate word recognition by Spanish/English bilinguals
in a laboratory task like lexical decision. ‘Special items’ to be simulated
were identical and non-identical cognates. Identical cognates are
translation equivalents with the same orthographic form in both languages
(e.g., HOTEL); non-identical cognates are slightly different in orthography
across the languages (e.g., TOMATO in English – TOMAAT in Dutch).
To assess to what extent Multilink simulates these processes with Spanish/
English bilinguals correctly, correlations are correlated as an indication
of the ’goodness-of-fit’ between simulation times and reaction times.
Multilink can already handle the processing of special words with a
Dutch/English lexicon (Dijkstra, Miwa, Brummelhuis, Sappelli, & Baayen,
2010). In this project, we ran simulations with ’special items’ in a newly
built Spanish/English lexicon. Like Dutch, Spanish and English are both
alphabetic languages. The model should be able to yield fitting simulations
on ’special items’ in diverse tasks and wide stimulus lists also for Spanish/
English bilinguals.
First, we constructed a full Spanish/English lexicon and applied Multilink
to it. The cycle times obtained from simulations with the lexicon as
a whole were compared with the reaction times from the Spanish lexicon
project SPALEX by Aguasvivas et al. (2018). Multilink simulations yielded
a good fit to the empirical data with this new lexicon, with correlations
going up to 0.6397.
Next, Multilink was used to simulate items from a study by Arana,
Oliveira, Fernandes, Soares, and Comesa˜na (2022). It was indeed possible to
simulate the processing of ‘special items’ such as cognates in this individual
experiment. More importantly, an empirical interaction effect of cognate
status and stimulus list composition could be accounted for by varying the
degree of lexical competition (by means of the lateral inhibition parameter
in the model). In all, Multilink’s simulations were promising in terms of
their ability to simulate cognate retrieval in differently composed stimulus
lists.
Description
Citation
Faculty
Faculteit der Sociale Wetenschappen