Constituent order in subordinate clauses: A comparison of Mudburra, Warlmanpa and Warriyangga/Thiinma

Slides for the presentation Constituent order in subordinate clauses: A comparison of Mudburra, Warlmanpa and Warriyangga/Thiinma, presented at the Australian Languages Workshop, North Stradbroke Island, Australia, on 9 July 2022. Abstract: In this preliminary study, we examine constituent order in subordinate clauses in datasets from three Australian languages. We then use a random forests analysis to identify variables that help predict SV vs. VS and OV vs. VO orders. Of the five potential predictor variables we analyse ('language,' 'text type [directly elicited/semi-elicited/unelicited],' 'subordinate clause type [relative/conditional/quoted speech],' 'matrix-subordinate order,' and 'finiteness of the subordinate clause'), only 'language' emerges as a significant predictor. This result highlights the fact that these three languages each have a unique profile for subordinate clause constituent order, despite being related. Additionally, it raises questions about what other factors (e.g., discourse-pragmatic information, main-clause word order, etc.) might, in a larger study, be found to shape subordinate clauses