Speaker: Christopher Legerme (MIT)
Title: Complementizers and Verb Fronting with Doubling in Haitian Creole
Time: Tuesday, September 10th, 1pm – 2pm
Location: 32-D461
Abstract: For this Syntax Square, we will investigate the syntax of Haitian Creole and the interaction of the ki/ke complementizer alternation (1) with Verb Fronting with Doubling constructions (VFD, Glaude and Zribi-Hertz, 2012) (2). HC shows the following subject-object asymmetry where the form of the complementizer varies according to whether the subject or object undergoes wh-movement (Koopman, 1982; Law, 1995).
1.a. Kiyès ki te wè Mari?
who COMP PST see Mari
Who saw Mari?
b. *Kiyès ki Mari te wè?
who COMP Mari PST see
Who did Mari see?
c. *Kiyès ( ke ) te wè Mari?
who COMP PST see Mari
Who saw Mari?
d. Kiyès ( ke ) Mari te wè?
who COMP Mari PST see
Who did Mari see?
Takahashi and Gra?anin-Yuksek (2008) (henceforth TGY) argue that the realization of ki depends on whether C can agree with a single nominal (DP) goal for both its unvalued features. Let us call this the SPLIT-AGREE analysis. TGYs approach has been popularly adopted in much recent work (Baptista and Obata, 2015; Obata et al., 2015; Sugimoto and Pires, 2022). Therefore, in VFD constructions the prediction is that ke should surface because C must agree with two different syntactic objects, namely, with the verb for wh-features and the subject for phi-features. Indeed, VFD constructions generally obligatorily lead to the object-extraction form of the complementizer ke (which is optionally null) (2).
2.a (Se) kontan ?ki / ?ke / ?Ø Bouki kontan
FOC happy COMP Bouki happy
It is that Bouki is HAPPY (e.g., not SAD).
b. (Se) kouri ?ki / ?ke / ?Ø Bouki kouri
FOC run COMP Bouki run
It is that Bouki RUNS (e.g., not WALKS).
However, I will show that, for certain unaccusative predicate constructions that are compatible with an otherwise precluded EPP violation in the language (i.e., verbs where the subject may be null or absent), the appearance of the complementizer ki can be triggered with the extraction of non-nominal constituents, contrary to the expected pattern of complementizer distribution in the language. For example, this happens in cases where verbs like rete to remain or manke to lack occur in the VFD construction (3) (the example below is simplified here for space).
3.a Se rete/manke ki rete/manke Malis san goud nan kont la
FOC remain COMP remain Malis 100 gourde in account the
It is that Malis has 100 gourdes REMAINING/LACKING in his account. (literally, it is that there is remaining Malis 100 gourdes)
(3) is incompatible with the SPLIT-AGREE analysis of TGY because of our expectations from (2). Clearly, argument structure is important for the form of the complementizer chosen by the grammar and it is important that our theory of HC be flexible enough to derive the data in (3). I argue in favor of an empirically superior alternative analysis of the complementizer alternation facts based on the antilocality approach that bans movement in the narrow syntax when it results in the dependency of two positions that are too close to each other (Grohmann, 2003; Erlewine, 2020, 2018). Challenges for this alternative proposal are discussed.