Türker (2016) found that English learners of Korean performed better at comprehending figurative language when the expressions shared lexical and conceptual structure across the two languages than when they diverged. (2016) examined Spanish learners of Basque, and found stronger electrophysiological responses to syntactic violations in structures that are common between Spanish and Basque compared to violations in structures unique to Basque. (2017), who observed a similar pattern with Spanish learners of French. This observation was later supported by Carrasco-Ortíz et al. Foucart and Frenck-Mestre (2011) found tentative evidence German learners of French show more electrophysiological sensitivity to gender errors when the French nouns have the same gender as German than when they have a different gender. For example, Lowie and Verspoor (2004) observed that Dutch learners of English acquire prepositions that are similar in form and meaning across the two languages (e.g., by/bij) more easily than ones that are dissimilar (e.g., among/tussen). The first is to take one group of L2 learners, with the same L1, and compare their acquisition of different structures in the L2, such that one structure is similar to L1 and the other is different. The goal of this paper is to propose a new method for evaluating the effect of similarity on the learnability of L2 structures, using deep learning.Ĭurrent approaches to determining similarity effects on L2 acquisition typically take an experimental angle, usually proceeding in one of two ways. Although individual structures (e.g., relative clauses, cognate inventory) can be compared fairly straightforwardly across languages, it is much harder to combine these structures appropriately to determine a global similarity metric across languages, which limits our ability to predict how difficult an arbitrary L2 will be to acquire for different L1 speakers. Similarity, in particular, is a difficult variable to examine, because it is so hard to pin down. Pedagogical context ( Tagarelli et al., 2016), cognitive processing differences across learners ( Ellis, 1996 Yalçın and Spada, 2016), L2 structural complexity ( Pallotti, 2015 Yalçın and Spada, 2016 Housen et al., 2019), or similarity between the target L2 and the learner's first language (L1) can all conspire to affect the speed and success of L2 acquisition ( Hyltenstam, 1977 Lowie and Verspoor, 2004 Foucart and Frenck-Mestre, 2011 Málek, 2013 Schepens et al., 2013 Türker, 2016 Carrasco-Ortíz et al., 2017). Learning a second language (L2) can be difficult for a variety of reasons. These findings serve as a proof of concept for a generalizable approach that can be applied to natural languages. The results showed that this approach can not only recover the facilitative effect of similarity on L2 acquisition, but can also offer new insights into the differential effects across different domains of similarity. We then compared the change for each L1/L2 bilingual model to the underlying similarity across each language pair. By observing the change in activity of the cells between the L1-speaker model and the L2-learner model, we estimated how much change was needed for the model to learn the new language. These models thus represented L1 speakers learning L2s. We next built a series of neural network models for each language, and sequentially trained them on pairs of languages. We built a set of five artificial languages whose underlying grammars and vocabulary were manipulated to ensure a known degree of similarity between each pair of languages. In this study, we present a different approach, employing artificial languages, and artificial learners. Further, the combinatorial explosion of possible L1 and L2 language pairs, combined with the difficulty of controlling for idiosyncratic differences across language pairs and language learners, limits the generalizability of the experimental approach. Yet global similarity between languages is difficult to quantify, obscuring its precise effect on learnability. Learning a second language (L2) usually progresses faster if a learner's L2 is similar to their first language (L1). 2School of Computing Science, University of Glasgow, Glasgow, United Kingdom.1English Language & Linguistics, University of Glasgow, Glasgow, United Kingdom.
0 Comments
Leave a Reply. |