The Weekly Newsletter of MIT Linguistics

Syntax Square 9/25 - Ljiljana Progovac (Wayne State University/MIT)

Speaker: Ljiljana Progovac (Wayne State University/MIT)
Title: The absolutive vP-less basis of se “middles” in Serbian: A split accusative language?
Date and time: Tuesday September 25, 1-2pm
Location: 32-D461

 I pursue a unified account of se in its various manifestations in Serbian, including, but not limited to, passive-like, middle-like, reflexive, reciprocal (all of them referred to as “middles”), given that coordination tests establish that these various readings are not syntactically distinct. The idea is that these middle se structures in Serbian (1) avail themselves of only one syntactic argument, a proto-participant in the event (2), often giving rise to massive vagueness, especially when not pragmatically constrained. The only argument in se middles is further proposed to be absolutive-like, in the sense that it does not grammatically discriminate subjects from objects, or agents from patients, the pattern also found with absolutives in syntactically ergative languages such as Tongan, as will be illustrated.

(1) Deca se tuku/grle/prskaju!
children SE hit/hug/sprinkle
‘The children are hitting/hugging/sprinkling each other.’ ‘?The children are hitting/ hugging/ sprinkling themselves.’ ‘The children are hitting/hugging/sprinkling somebody (else.)’ ‘The children are hitting/hugging/sprinkling me!’ ‘One spanks/hugs/?sprinkles children.’

(2) ∃e [H(e) ∧ Participant (Children,e)]

This analysis ties into my work on language evolution (e.g. Progovac 2015, 2016), where I have reconstructed a vP-less and TP-less (intransitive and tenseless) small clause stage in the evolution of language/syntax, with only one (absolutive-like) argument per verb/clause. Arguably, this absolutive-like proto-layer provides a common foundation/denominator for building transitivity, either by adding an ergative argument on the top, or an accusative argument on the bottom, not inconsistent with the postulates of Dependent Case Theory (e.g. Yip et al. 1987; Marantz 1991; McFadden 2004; Baker 2015).
Previous proposals have noted that se in Serbian can be analyzed as neither an argument, nor a reflexive pronoun, but is instead some kind of grammatical marker (e.g. Franks 1995; Marelj 2004; Progovac 2005). A unified account of se in Serbian requires seeing se as flagging a different, parallel type of grammar in a language whose dominant grammar is accsative, suggesting that Serbian may be a split accusative language, on analogy with split ergative languages. Finally, this approach makes some specific testable predictions regarding the processing of se middles, in contrast to their transitive (vP) counterparts, and the fMRI experiments we conducted have yielded some promising initial results (Progovac et al. 2018).