Differential Subject Expressions in Lelepa: A Quantitative Investigation

Abstract

In Lelepa, an Oceanic language spoken in central Vanuatu, syntactic subjects are described as being expressed obligatorily through a subject proclitic at the beginning of the verb complex and optionally with a preverbal subject NP, headed by either a lexical noun or a free pronoun (Lacrampe, 2014). However, evidence has suggested that subject proclitics are sometimes deleted. Hence, Lelepa features four different subject expressions: a sole proclitic, a lexical NP, a free pronoun, or zero subject, whose contexts are nonetheless not well understood. This thesis draws on a corpus of Lelepa natural speech and investigates the variability in subject expressions quantitatively with logistic regression and recursive partitioning, accompanied by descriptive statistical data. Results have supported existing description that subject proclitics are the most prevalent subject expression, although they are also found deleted in clause-chains or in the presence of a coreferential lexical NP. In addition, Lelepa speakers’ realisation of subject references in the corpus is usually the optimal choice that is neither ambiguous nor redundant, hence lexical NPs and free pronouns are commonly present where ambiguous but disfavoured elsewhere. However, redundant free subject pronouns may still be used for pragmatic functions such as floor-shifting or attention-getting, especially for human, direct-speech, or deictic speech-act participant referents. This thesis complements existing description of Lelepa morphosyntax, and extends our typological understandings of referential choice in Oceanic languages. The data pre-processing and statistical modelling in this thesis also demonstrate practical approaches to performing quantitative analyses of small and/or unbalanced datasets. Show more