ON THE CAUSATIVE VERB FORMS OF ARABIC: FORM I AND FORM II AND THEIR ASSOCIATIONS WITH (IN)DIRECTNESS OF CAUSATION

This dissertation sheds light on the semantic domain of causation in Arabic. The aim is to examine two Arabic causative verb forms, Form I and Form II, and their associations with (in)directness of causation. The central working hypothesis throughout this work is the verb-semantics hypothesis by Shibatani and Pardeshi (2002), which posits that autonomy of the causee, degree of directed causation, requirement for an external causer, and merger of two subevents into one conceptual event are factors that predict the morpho-syntactic complexity of a causative construction. Following the lead of Ambridge et al. (2020) on their operationalization of the verb-semantics hypothesis by Shibatani and Pardeshi (2002), two experiments were conducted. In both experiments, 60 animations for 60 verbs were used to depict various causative scenarios. The first experiment explored how Arabic speakers mentally perceived 60 events that depicted various degrees of causativity. This was achieved through collecting ratings from 20 Arabic speakers on four semantic variables: autonomy of the causee, degree of directed causation, degree of event-merge, and the requirement for an external causer. The second experiment obtained judgments of the relative acceptability of the less- and more-transparent causative forms of the same 60 verbs from 24 native-speaking Arabic adults. 

Three analyses were conducted on the results to better understand how causatives manifest in language use cross-linguistically, with a dedicated focus on the Arabic language.The first analysis addressed whether the four semantic variables of the verb-semantics hypothesis of Shibatani and Pardeshi (2002) account for the restrictions on the use of Arabic verb Form I and Form II. It was found that the variables autonomy of the causee, degree of directed causation, and the requirement for an external causer each showed strong positive correlations with Form I. The variable degree of event-merge showed a weak positive correlation with Form I. No correlations were noticed between any of the four variables and Form II. The second analysis addressed whether Arabic speakers perceive events in a similar way to speakers of other languages. Because this study followed the methodology Ambridge et al. (2020) used to examine causatives in English, Hebrew, Hindi, Japanese and Kʼicheʼs, the results could be directly compared. It was discovered that yes, Arabic speakers conceptualize the events tested similarly to speakers of the five other comparison languages. Twenty-three verbs (>38.3%) were rated the same by all speakers (120 speakers, 20 from each language) on all four semantic variables. From the remaining thirty-seven verbs, twenty-three verbs received the same ratings in three of the semantic variables, but not in event-merge. The remaining fourteen verbs were associated with numerous disagreements among the participants. The third and final analysis addressed the claim that all human languages use morphosyntax to mark the difference between direct versus indirect causative events by testing whether this holds true for Arabic. Across-linguistic computational model developed by Aryawibawa et al. (2021) was used to answer this question on the reasoning that if the principle is truly cross-linguistic, then the universal model should be able to utilize speaker semantic judgements to make accurate predictions about the grammatical acceptability of the different morphosyntactic forms. The model accurately predicted Arabic speaker judgments by a moderate correlation of 0.05, suggesting that Arabic conceptualizes directness of causation in a similar way to other languages, which supports the view that the underlying semantic distinction of more versus less direct causation maps onto and manifests as a morphosyntactic distinction.