[SynSem] Diana Kakashvili (TSU) & Stavros Skopeteas (U. Göttingen)

Diana Kakashvili (Tbilisi State University) & Stavros Skopeteas (U. Göttingen)

Objects and goals in Tush

OV languages differ in the conditions under which they permit postverbal constituents. Some OV languages, such as written Turkish and Tamil, only allow for postverbal constituents that are outside the focus domain. In contrast, other OV languages, including Sinhala, Georgian, and Eastern Armenian, allow postverbal foci (Herring and Paolillo 1995; Öztürk 2013; Borise 2019; Pregla 2024). Moreover, variation exists not only between languages but also within languages: different arguments vary in their likelihood of appearing in the postverbal domain. For instance, direct objects are less likely to occur postverbally than obliques, and among obliques, goal arguments of motion verbs are even less likely to do so (Haig et al. 2024). This disparity cannot be attributed solely to information structure, as goal arguments of motion verbs do not represent background information more often than direct objects.
The question is whether this difference is rooted to the syntax of different types of complements or it reflects linearization preferences that are independent from syntax. Tush (a Nakh language spoken in Georgia) is a good candidate to examine exactly this question: corpus facts (narratives texts by 16 native speakers) reveal that direct objects are only postverbal if they represent given information, while goal arguments of motion verbs can be postverbal independent of information structure. In order to identify the syntactic determinants of the different types of complements, we compared their behavior with respect to diagnostics of the syntactic properties of the preverbal and postverbal domain. Our results show that direct objects and goal arguments behave alike with respect to properties that are determined by the core syntax (such as negative concord) but differ in properties that are sensitive to information structure (such as the placement of remnants in split topicalization).
 

This talk is part of CauLaGeNet talk series organised by SFL SynSem group and LLING, (CNRS & U. Nantes).