The development of verb-initial structures cross-linguistically: insight from Austroasiatic

Sponsored by the SNSF 2018-2021


Research summary

In linguistics, the verb is traditionally considered to be the core element of the clause, and its core status is emphasized in psycho- and neurolinguistic research as well as being the basis of many theories on the syntax/semantic interface (see Levin & Rappaport-Hovav 2005 for an overview). At the same time, basic verb-initial (V1) word order reportedly occurs in only ca. 10-19% of the world's languages (c.f. Dryer 2013, Clemens & Polinsky to appear). Explanations for this may come from recent experiments and large-scale comparisons, which show that V1 sequences in natural languages violate apparent cognitive or processing biases on at least three levels: information structure (“ɢɪᴠᴇɴ before ɴᴇᴡ principle”; Chafe 1994, Lambrecht 1994, Junge et al. 2015), dependency (“dependency length minimization”; Hawkins 2004, Newmeyer 2005, Futrell et al. 2015), and parsing (“Actor-first”; cf. Choudhary 2010).

This research suggests that V1 structures are dispreferred cross-linguistically since a V1 language will necessarily violate at least two of these biases. However, 10% (or more) of the world’s languages translates to quite a large number of languages that have V1 structures, and such languages can have quite stable word order. Further, the extremely robust Actor-first bias does not seem to apply in V1 sentences (Schlesewsky & Bornkessel 2006, Bickel et al. 2015), suggesting the possibility of a (not yet identified) driving force that counteracts the Actor-first preference. The most likely way to identify such a force is to trace the evolution of V1 structures in the history of language families and to compare the findings to specific effects of horizontal transfer (areal skewings), universal drifts, and chance.

Word order is highly prone to change, from both internal (i.e. reanalysis, pragmatics) and external (i.e. areal and contact) influence (Harris & Campbell 1995, Aikhenvald & Dixon 2007), so all possible factors need to be considered. Accordingly, this project takes a broad, corpus-based approach to study the evolution and distribution of V1 structures in Austroasiatic (AA) languages, which are of particular interest for the study of V1 configurations (Jenny et al. 2014). We then compare the findings with V1 structures in other language families to investigate V1 development (as e.g. in Insular Celtic, Modern Welsh), V1 maintenance (e.g. AA, Austronesian, Modern Irish), and V1 loss (e.g. Afroasiatic, western Austronesian, Modern Breton, Middle Welsh). This project focuses on AA languages from the Khasian, Palaungic, Nicobarese, Aslian, and Katuic groups, namely because: 1) they exhibit V1 patterns to different degrees and in different structures, 2) the V1 patterns cannot be explained as result of language contact, and 3) they are spoken on the periphery of the AA area as local rather than state languages.

This last point (3) is particularly salient, since peripheral (residual) languages are expected to be more conservative (and thus more likely to contribute to reconstruction) than central (or “spread”) languages (Nichols 1992, Dixon 1997; but cf. Celtic and Tocharian). These peripheral languages are understudied, so we are more likely to gain new insight into the effects and development of V1 structures. The inability to explain these structures through language contact (point 2) means we must consider a diachronic source. The range of word orders in these languages (point 1) means that we can more likely account for V1 features diachronically, particularly since they are from the same tree. This in turn will give us indications of likely developmental pathways of word order for other language phyla and families.

The basic question we are asking is: “what V1 structures do we find where and when, what motivates their existence, and how are they maintained, developed, or lost?” Secondary questions are: “how does verb-initiality interact with or motivate subsystems of grammar?”, and “to what degree do languages differ with regard to V1 structures?” These are difficult questions, particularly due to the complexity of the research, what is known of the languages, and the history of the areas under study.