NO SUPPORT FOR HIERARCHICALLY-RANKED LANGUAGE-TRANSFER CUES: EXPERIMENTAL EVIDENCE FROM ARTIFICIAL LANGUAGE LEARNING

Abstract

A variety of theories attempt to model cross-linguistic influence in multilingual language learning. The current study uses an artificial miniature language paradigm to test the predictions of one of these models: the Typological Primacy Model (Rothman, 2011, 2013, 2014). This model holds that cross-linguistic influence is determined by learners’ perceived similarity between their known languages (L1, L2, etc.) and the novel language, and further hypothesizes that similarity between languages is determined using a hierarchy of input cues in which some cues are more important than others in determining similarity between languages and thus cross-linguistic influence. Specifically, the highest level of the hierarchy is lexical cues, such as cognates, meaning that, if present, they should always be the most important factor in determining cross-linguistic influence. We created two artificial languages and taught them to Korean-English bilinguals. One language had English cognates and Korean-like word order within phrases, while the other had Korean cognates and English-like word order within phrases. Both languages had variable sentence order, allowing participants to choose between subject-verb-object and subject-object-verb. The proposed Typological Primacy Model predicts that participants who learned the language with the Korean cognates would prefer to use Korean-like subject-object-verb word order despite the presence of English-like phrases, while participants who learned the language with the English cognates would prefer to use English-like subject-verb-object word order despite the presence of Korean-like phrases. We found that there was no difference between the two groups’ word order choices, leading us to conclude there is no evidence to support the Typological Primacy Model’s predictions that cognates determine the source of cross-linguistic influence.