Semantics and Statistics in the Formation of Causatives in Hijazi Arabic: Data From Semantic Ratings, Grammatical Acceptability Judgments and Computational Modeling Article Swipe
YOU?
·
· 2025
· Open Access
·
· DOI: https://doi.org/10.1525/collabra.146110
· OA: W4416290461
The aim of the present study was to investigate the possibility that Hijazi Arabic – like other languages previously studied – uses morphosyntactic marking to differentiate events of more- versus less-direct causation, preferring to mark them with less-transparent and more-transparent marking respectively (e.g., Somebody broke the window vs Somebody MADE the window break; *Somebody cried the boy vs Somebody MADE the boy cry. First, we obtained from 20 adult native speakers, directness-of-causation ratings (event-merge, physical/directive causation, requiring an external causer and autonomy of the causee) for each of 60 actions, corresponding to 60 verb roots. Next, we showed that these semantic ratings – alongside the statistical learning predictors of preemption and entrenchment – predicted the relative acceptability of less-transparent, Form I causatives (e.g., *kataba) versus more-transparent, Form II causatives (e.g., kattaba), as judged by 24 new participants from the same population. Finally, we showed that a computational model that learns mappings between bundles of lexical and semantic features and the two causative forms can simulate (with correlations of around r=0.9) verb-by-verb patterns of human acceptability judgments. These findings add to a growing body of literature which suggests that language learners may be aided by a typological universal in the form of a negative relationship between directness of causation and transparency of causative marking.
