UC Berkeley

This dissertation argues for the existence of a new A'-feature, discourse connected (DC), which grammatically encodes a constraint on the relation between the constituent it attaches to and discourse relations to previous sentences. Connectives like That's because and For example encode the rhetorical relations of explanation and elaboration. DC encodes these relations as well, but by A'-movement of a phrase to the left edge of a clause or noun phrase, specifically the movement of a phrase that is previously mentioned in the sentence that the current one is explaining or elaborating upon. I argue that there must be a DC-feature in the lexicon on par with other A'-features, such as wh. Given that DC encodes a discourse structural constraint, there must be a syntax--discourse structure interface.

One consequence of this dissertation is descriptive: a range of phenomena in the Eastern Cham language (Austronesian: Vietnam) are found to be instances of DC-marking. These phenomena include what appear on the surface to be topicalization, optional wh-movement, partitives, and inventory forms (e.g. bread, three loaves). I argue that these phenomena in Eastern Cham must be analyzed in terms of DC, not information structure or other previously proposed analyses for comparable constructions in other languages. As a result, multi-sentence discourses that control for discourse structure should be used as diagnostics for constructions that might be licensed by DC or something similar.

This dissertation adds a new A'-feature, DC, to the typology of A'-features. DC-movement shares a variety of characteristics with A'-movement more broadly. It exhibits sensitivity to syntactic islands, weak crossover, and locality effects. As with other A'-features in some cases, there is a parallelism between CP and DP: a phrase can be DC-moved to the left edge of either a clause or a noun phrase. Similarly, wh-phrases can undergo secondary movement to the left edge of the noun phrase in some languages. Additionally, this dissertation examines the position of DC among other A'-features, following recent work on the hierarchy of A'-features (Aravind 2017, 2018). Despite its apparent optionality, DC-movement is argued to be obligatory, like wh-movement. DC is also found to be independent from other features such as wh; the same phrase can be marked as both DC and wh. Unlike wh, DC can only be checked once on a respective phrase in a derivation, and the movement of multiple DC-phrases to the edge of the same clause exhibits Path Containment Effects (Pesetsky 1982), not the tucking-in pattern observed in multiple wh-movement (Richards 1997).

DC is argued to provide evidence for the need for a dynamic event semantics that allows the events introduced throughout the discourse to be tracked. This dissertation proposes that the DC-feature is introduced by a DC-particle on analogy with focus particles and the Q-particle on wh-phrases (Cable 2010). The DC-particle is shown to introduce a presupposition that checks the participants of two events in a discourse: the current event and an event in a prior sentence inferred by a subordinating discourse relation (i.e. explanation or elaboration).

The dissertation proceeds as follows. Chapter 1 introduces DC, along with the ongoing debate on the existence of pragmatic features in syntax. Additionally, the concept of hierarchical discourse constraints (HDCs) is introduced as a heuristic to understand that position of DC in comparison with linear information states (LISs), exemplified by information structural notions like old information topic. Chapter 2 examines the basic case of DC-movement, topicalization, and demonstrates that DC must be an \={A}-feature. In Chapter 3, wh-phrases are shown to be able to be DC-marked in Eastern Cham. An analysis is proposed in which DC-particles and Q-particles can be present in the same DP. Chapter 4 turns to DC-marking inside a noun phrase and finds a CP/DP parallelism: phrases can be DC-moved to the left edge of a noun phrase. Despite these cases involving movement inside a noun phrase, the event semantic interpretation of DC is affirmed: DC is only computed between the larger events of which the noun phrase is a part. Chapter 5 concludes. Additionally, it analyzes clitic right-dislocation in Catalan as DC-movement with minor syntactic and semantic differences from DC-movement in Eastern Cham (cf. Lopez 2009). Contrastive topic is also examined and proposed to be a second hierarchical discourse constraint, different from DC.