Direct Versus Indirect Causation as a Semantic Linguistic Universal: Using a Computational Model of English, Hebrew, Hindi, Japanese, and K'iche’ Mayan to Predict Grammaticality Judgments in Balinese

The aim of this study was to test the claim that languages universally employ morphosyntactic marking differentiate events more- versus less-direct causation, preferring mark them with less- and overt marking, respectively (e.g., Somebody broke window vs. MADE break; *Somebody cried boy cry). To end, we investigated whether a recent computational model which learns predict speakers’ by-verb relative preference for two causatives in English, Hebrew, Hindi, Japanese, K'iche’ Mayan is able generalize sixth language on it has never been trained: Balinese. Judgments acceptability more-transparent causative forms 60 verbs were collected from 48 native-speaking Balinese adults. composite crosslinguistic these judgments, not only had seen, but also––in split-half validation test––to seen any language. A “random-semantics” showed relatively small decrement performance verbs, whose behavior can be learned verb-by-verb basis, achieved zero correlation human judgments when generalizing unseen verbs. Together, findings suggest conceptualizes directness causation similar way unrelated languages, therefore constitute support view distinction between less-distinct constitutes universal.