Abstract
Abstract Based on argument indexation properties, this paper offers a model of four verb classes for Geshiza and Stau, two closely related and endangered Horpa lects spoken in Sichuan. Despite increased awareness and interest in Horpa, no exhaustive classification of verbs from the viewpoint of argument indexation has been offered. Even though argument indexation itself has been discussed in previous scholarship, the striking diversity of argument indexation patterns has not been addressed. It is shown that both Geshiza and Stau, the two lects chosen as the sample for this study, share the same verb classes, even though the class assignment of individual cognate verbs occasionally varies. The paper also describes how semantic and animacy features inherent in verbs, most prominently compatibility with human S and P arguments resulting in compatibility with speech act participants, play a crucial role in verb class assignment. Establishing a framework for Horpa verb classes helps in the ongoing descriptive work of these languages. In addition, it offers new material for typological studies of argument indexation.