Beyond cognates and false friends: An exploration of multiple mapping between orthography and semantics using the Multilink model.

Abstract

In the bilingual mental lexicon, multiple mappings between form (orthography and phonology) and meaning (semantics) are frequent. Cognates, such as English TOMAAT and Dutch TOMAAT, which have the same or a similar orthography and the same meaning; false friends, such as room, have the same orthography but a different meaning in each language; while homonymous or polysemous words, such as Dutch BLIK, which has multiple (unrelated or related) meanings. Such multiple mappings influence processing on tasks such as lexical decision, word naming, and word translation in a variety of ways. Previous research has shown a facilitation effect of shared orthography and semantics, but an inhibition effect of shared phonology (Lemhöfer & Dijkstra, 2004; Schwartz et al., 2007; Peeters et al., 2013). These studies have also shown an effect of word frequency. Van Hell & de Groot (1998) and Tokowicz & Kroll (2007) showed that there are differential effects depending on semantic properties of the words in question. This thesis uses the Multilink computer model (Dijkstra et al., 2018a), which simulates visual word retrieval, for an exploratory study on the processing of multiple mappings. Simulations are performed using cognates and false friends, but also within-language multiple mappings (homonymy and polysemy), as well as combinations of within-language and between-language multiple mappings. It is found that the outcomes of Multilink’s simulations show moderate to high correlations with experimental outcomes in the case of simple multiple mappings and low to moderate correlations in the case of more complex multiple mappings. These findings are discussed in the context of theories of semantic representation.