This paper addresses the question of how assertion and presupposition are reflected in the grammar. Since Kiparsky &Kiparsky 1970 andHooper &Thompson 1973, it's often been suggested that these notions provide the semanticpragmatic underpinning for a range of complementation patterns, including the licensing of so-called Main Clause Phenomena [MCP]. This paper presents a new large-scale experimental study, investigating the lexical and semantic-pragmatic licensing conditions of four types of MCP (Verb Second [V2], topicalization, speech act adverbs, and scene setting adverbs) in English, Swedish, and German. The central contribution of this paper is demonstrating what precise dimensions of assertion and presupposition are reflected in the grammar: for embedded V2, what matters is the discourse status of the embedded proposition as new vs. Given (in the sense of Schwarzschild 1999); a dimension which cross-cuts both factive and non-factive verbs. The other MCP investigated show no sensitivity to either of the lexical or pragmatic factors investigated. We further show that Givenness is not reflected in a (null) DP-layer, contrary to previous claims.
The Syntax of Assertion and PresuppositionThe main finding of the experiment is that V2 tracks the availability of p as discourse new information (as contrasted with the status of p as Given, in the sense of Schwarzschild 1999) -a dimension which cross-cuts both factive and non-factive verbs. The interpretation of p as new vs. Given is found to depend closely on the type of predicate; however, interactions with the polarity of the matrix clause speak against a selection-based account. The other MCP investigated, on the other hand, are found to be sensitive to neither type of lexical or pragmatic factor investigated; showing us that MCP encompass a less homogeneous set of constructions than generally assumed. We moreover show that Givenness does not imply that the embedded clause is underlyingly a DP, as per previous proposals (e.g. Haegeman 2014; Kastner 2015; following Kiparsky & Kiparsky 1970).
Overview of the paperSection 1.2 looks at previous approaches to MCP, finding a highly complex theoretical and empirical landscape, with substantial disagreement not just on the theoretical side, but also in terms of what the empirical facts actually are. The main open questions are discussed in Section 1.3. This discussion leaves us with the conclusion that without comparative data from a larger number of speakers and a wider range of verbs, MCP, and languages, further theoretical progress is difficult. Section 2 presents the current experiment, addressing these open questions. Section 3 discusses the analytical consequences of the present findings, showing further that Givenness is not semantically encoded in a null D-layer. We end on a note on the implications of the current findings for the theory of factivity.