Embedded verb second in German: Experiments at the syntax-pragmatics interface

The interpretation and licensing of clauses with Embedded Verb Second (EV2) in German have been widely debated over the past decades. The goal of this thesis is to experimentally testempirical claims in more recent work on EV2 in German, in particular those in Djärv 2019a and Caplan & Djärv 2019. Djärv (2019a) and Caplan & Djärv (2019) argue that (i) EV2 clauses must denote discourse-new information, and that (ii) EV2 clauses are therefore unacceptable under embedding predicates that presuppose that the complement proposition is discourse-old. These claims were tested using a judgment task to elicit the naturalness of EV2 clauses when they constitute discourse-old information. The results provide two key findings. First, they show that the EV2 clauses are judged natural when they constitute discourse-old information. Second, the results show that canonical verb-final configurations are preferred over EV2 complements if the embedded proposition denotes discourse-old information. The results partially bear out previous claims and I argue that it is the lexical semantics of embedding predicates alone that determines whether EV2 is available, not the discourse status of any particular instance of EV2. Moreover, this thesis provides additional evidence against the claim that EV2 gives rise to speaker-commitment interpretations of the embedded proposition, replicating the results in Djärv 2019a and Caplan & Djärv 2019.