Turkish-German heritage speakers' predictive use of case: webcam-based vs. in-lab eye-tracking.

Abstract

Recently, Özge et al. have argued that Turkish and German monolingual 4-year-old children can interpret case-marking predictively disregarding word order. Heritage speakers (HSs) acquire a heritage language at home and a majority societal language which usually becomes dominant after school enrollment. Our study directly compares two elicitation modes: in-lab and (remote) webcam-based eye-tracking data collection. We test the extent to which in-lab effects can be replicated in webcam-based eye-tracking using the exact same design. Previous research indicates that Turkish HSs vary more in the comprehension and production of case-marking compared to monolinguals. Data from 49 participants-22 Turkish monolinguals and 27 HSs-were analyzed using a binomial generalized linear mixed-effects regression model. In the Accusative condition, participants looked for the suitable Agent before it is appeared in speech. In the Nominative condition, participants looked for the suitable Patient before it is appeared in speech. HSs were able to use morphosyntactic cues on NP1 to predict the thematic role of NP2. This study supports views in which core grammatical features of languages, such as case, remain robust in HSs, in line with the Interface Hypothesis. We were able to replicate the effect of the predictive use of case in monolinguals using webcam-based eye-tracking, but the replication with heritage speakers was not successful due to variability in data collection contexts. A by-participant analysis of the results revealed individual variation in that there were some speakers who do not use case-marking predictively in the same way as most monolinguals and most HSs do. These findings suggest that the predictive use of case in heritage speakers is influenced by different factors, which may differ across individuals and affect their language abilities. We argue that HSs should be placed on a native-speaker continuum to explain variability in language outcomes.