Prediction in bilingual sentence processing: Is it linked to production?

Natalia Meir, Olga Parshina, Irina Sekerina

Abstract:

The Unified Competition Model (MacWhinney, 2012) accounts for crosslinguistic
differences in thematic role mapping. We investigated production
and predictive use of accusative case morphology in Russian-Hebrew
bilingual children. We also investigated the role of production in predictive
processing testing the Prediction-by-Production Account (Pickering &
Garrod, 2018) vs. the Missing Surface Inflection Hypothesis (Prévost &
White, 2000). Three groups of children aged 4–8 participated: Russian-
Hebrew-speaking bilinguals, Russian-speaking and Hebrew-speaking
monolingual controls. All children participated in the accusative case
production and Visual-World eye-tracking comprehension experiments.
Bilinguals were tested in both of their languages. The results of the study
confirmed the predictions of the Unified Competition Model showing
typological differences in the strength of the case-marking cue and its
predictive use in sentence processing in Russian- and Hebrew-speaking
controls. While Russian-speaking monolinguals relied on case marking to
predict the upcoming agent/patient, the performance of Hebrew-speaking
monolingual children varied. The findings for bilinguals showed that
despite their lower production accuracy in both languages, they were either
indistinguishable from monolinguals or showed an advantage in the
predictive use of case morphology. The findings support the Missing Surface
Inflection Hypothesis, which predicts a dissociation between production and
comprehension.