Investigating frequency of occurrence effects in L2 speakers: Talent matters

This paper investigates the impact of word frequency and pronunciation talent on proficient speakers’ second language (L2) production variability/similarity. The data analysed stem from a corpus of quasi-spontaneous conversations between native speakers of German and English. Within-speaker production similarity was established through a comparison of amplitude envelope signals for word tokens of L2 English word types. Production similarity scores, type frequency, and speaker talent were used as input for several linear (mixed effects) models. The models yielded a significant effect of type frequency, and a signficant interaction of type frequency and talent, for predicting similarity. The results are discussed in the context of exemplar-based category formation and indicate that less talented speakers are less capable of forming L2 categories.