Embedded V2, Factivity and Main Point of Utterance

Since the seminal work of Hooper & Thompson 1973, many researchers have pursued the insight that V2, as a classic Main Clause Phenomenon [MCP] is licensed in formally subordinate clauses to the extent that such clauses are asserted. H&T categorised embedding predicates into 5 classes largely according to whether their complement clauses could be interpreted as asserted, a status which they took to be the converse of presupposed. The class of verbs of communication such as say occupied one pole – allowing MCP freely in their complements – while factives such as be happy that occupied the other. In an important update of this tradition, Simons 2007 has considerably sharpened H&T’s concept of assertion, proposing that the crucial distinction is whether the subordinate clause contributes a proposition that makes the utterance relevant; as a diagnostic, in a question/response sequence, “whatever proposition communicated by the response constitutes an answer (complete or partial) to the question is the main point of the response.” Simons demonstrates that given this definition/diagnostic, even factive clauses may constitute the Main Point of Utterance MPU; hence, in such contexts, they should also allow V2. In this talkwe present the results of three experiments (one on Swedish and two on English) that aimed to test empirically the claim that the possibility of V2 in an embedded clause (EV2) follows fromwhether or not the embedded clause constitutes theMPU (cf. Julien 2007, Jensen & Christensen 2013). In the first experiment, 104 L1 speakers of Swedish were asked to judge the acceptability of question-response pairs where, following Simons 2007, the question was manipulated to vary the location of the MPU in the response: in the main or the embedded clause. There were two other independent variables: the classification of the embedding verb in the response, and whether or not the embedded clause in the response exhibited V2. We show that, on the one hand, the results support the claim that Swedish EV2 is possible under semi-factive (discover/realize) and non-factive (think/claim) clause-embedding predicates, but not under purely factive ones (be happy/be surprised) (Wiklund et al. 2007). Strikingly, the judgments also mirror the frequency difference between EV2 in the complements to epistemic vs. communicative non-factives (e.g. suppose vs. say) reported for Danish corpus data in Jensen & Christensen 2013. However, the results show no interaction between the effect of embedded V2 and embeddedMPU: that is, our data suggest, contra Julien 2007, Jensen & Christensen 2013, that the low acceptability/frequency of V2 under factives cannot be explained by the twin hypotheses that MPU licenses EV2 and that factives cannot embed MPU. An alternative interpretation, preserving the idea that MPU licenses EV2, would be that participants may have essentially ignored the MPU-licensing questions when evaluating the acceptability of the responses. Under such an account the low acceptability of EV2 under