Predictors of gesture viewpoint in L2: A generalized linear mixed models analysis

A number of factors (speech, gesture, and demographics) have been found to underlie gesture viewpoint alternation (character versus observer) in L1. However, while alternation between gesture viewpoints is attested in L2, we do not know to what extent this is patterned and whether the same explanatory variables found for L1 also predict L2 viewpoint. Narrative descriptions of motion were examined from a curated corpus of L1 Japanese speakers of intermediate-level L2 English residing in either Japan or the United States. New mixed effects binary logistic regression models tested if a variety of features of the accompanying speech and gesture or select characteristics of participants were predictive of gesture viewpoint. Results indicated that gestural expression of Manner of motion and information status of the agent in speech were significant predictors of character viewpoint, while an increase in the number of Path expressions significantly predicted observer viewpoint. Semantics of L2 speech and absence of a grammatical subject were not found to be significant predictors of viewpoint in this corpus. Female participants were significantly more likely to use character viewpoint than male participants, but country of residence was not found to significantly predict gesture viewpoint. The results suggest that in descriptions of motion, the linguistic options offered by languages predict the viewpoint with which L2 gestures are produced, even by intermediate-level L2 users. Furthermore, L2 data have the potential to contribute to our understanding of the nature of gesture viewpoint.