How does adult learners' L1 interact with word frequency in the error rates and patterns of L2 classifier use: a cross-linguistic comparison

Previous literature has reported mixed findings on the effect of L1 (classifier languages vs non-classifier languages) on the acquisition of L2 classifiers. This study aims to investigate whether any effect of L1 may be modulated by the word frequency of the target classifiers. Chinese classifiers were categorized into pre-established frequency bands A to C in descending order of word frequency, and learner data was extracted from the HSK Dynamic Composition Corpus containing sentences with classifiers written by L1-Korean and L1- English learners, with a total of 5248 sentences analyzed. Two native Mandarin speakers reported the error rates, and errors were also further categorized into four different types (i.e. misuse, omission, overuse, misorder) to investigate any differences in error patterns across different L1 groups. Results showed a significant interaction between L1 and classifier word frequency, with comparably low error rates for both L1 groups on high-frequency classifiers, but a significantly higher error rate for L1-Korean learners on low-frequency classifiers. Aligned with exemplar theory, more frequently encountered classifiers may have a relatively robust representation in the mental lexicon for both L1 groups, as opposed to the less stable representation for less-frequently encountered classifiers which are more prone to transfer effects from L1 to L2.