| Research has shown that cognates between Japanese and English have the potential to be a valuable learning tool (Daulton, 2008). Yet little is known on how Japanese learners of English produce cognates in context. Recently, studies have argued that cognates can cause a surprisingly high number of syntactic errors in sentence writing activities with Japanese learners (Rogers, Webb, & Nakata, 2014; Masson, 2013). In the present study, I investigated how Japanese learners of English understood and used true cognates (words that have equivalent meanings in both languages) and non-true cognates (words where the Japanese meaning differs in various ways from their English source words). Via quasi-replication, I analyzed participants' sentences to determine the interaction of true and non-true cognates on semantics and syntax. In an experimental study, twenty Japanese exchange students filled out a word knowledge scale of thirty target words (half true cognates and half non-true cognates) and wrote sentences for the words they indicated they knew. These sentences were analyzed quantitatively and qualitatively for both semantic and syntactic errors. Sentences with true cognates were semantically accurate 86% of the time, while those with non-true cognates were accurate only 62.3% of the time, which was a statistically significant difference. When the sentences were analyzed for syntax, there was no statistically significant difference in the number of errors between true and non-true cognates, which contrasts with previous research. Qualitative analysis revealed that the most problematic syntactic issue across both cognate types was using collocations correctly. Among those collocational issues, there were clear differences in the types of errors between true and non-true cognates. True cognate target words were more likely to lead to problems with prepositional collocations, while non-true cognate target words were more likely to lead to problems with verb collocations. These results suggest that for intermediate Japanese learners of English, semantics of non-true cognates should be prioritized in learning, followed by syntax of true and non-true cognates, which should be taught according to the most problematic error types per cognate status. |
