Generating syntactically well-formed sentences requires not only a system of syntactic rules, operations or constraints, but also a lexical component of the grammar that captures the relevant features of the predicates the syntax works with. The influence of the lexicon on the syntax reveals itself most clearly in the case of clause-embedding predicates such as claim, ask, promise, refuse, seem, know, believe, be glad, etc. For example, differences seem to exist between predicates with respect to selectional restrictions (believe that vs. *believe whether), and possible control readings (x promises y to come → x will come; x persuades y to come → y should come).
The leading question of PB3 is to what extent the syntax of embedded clauses is determined by the lexical specifications of clause-embedding predicates, and to what extent the lexical properties of such predicates are influenced by syntactic properties of subordination structures. It is often hard to decide which of these factors is more prominent.
Our research group is developing and utilizing a database of clause-embedding predicates in which predicates of different languages and from diachronic stages are stored. The database includes a series of corpus examples for each predicate, which are annotated for a series of properties in order to document that predicate's embedding behavior. This makes it possible to run queries in the database on the basis of predicate classes and the other coded properties of predicates and example sentences. Currently, the largest part of the database covers German, but we are expanding coverage in order to investigate cross-linguistic variation in clause-embedding predicates. The contemporary German portion of the database is now available as a public beta at http://www.owid.de/plus/zasembed2017/. This public version, and the custom-made search interface, were made possible by a collaboration with the Institut für Deutsche Sprache, who are hosting the database on their OWIDplus platform.
The area Status and realization of the sentential argument is concerned with two aspects of sentential embedding that have been previously neglected. First, by which means (Case/Adpositions) are sentential arguments licensed? Second, how are additional (sentential) arguments licensed? (cf. Frank hämmert Maria ein, dass sie sparen soll - Frank keeps dinging it into Maria that she should save money - vs. *Frank hämmert, dass Maria sparen soll - Frank hammers that Maria should save money).
In the area Non-canonical argument realization we are carrying out a comparative investigatio of predicate classes, whose propositional argument positions are not canonically specified. For example, the clausal argument of certain predicates can be realized by an adverbial: He regrets when she is sick. In other constructions, the clausal argument can be supplied by a whether-clause, even though the matrix predicate doesn't normally select for interrogatives: We can neither confirm nor deny whether the course took place.
In the area Implicit Causality, we are currently developing a semantic theory of implicit causality (IC) verbs, which include - among others - clause-embedding psych verbs. The analysis encompasses both the tendency of IC verbs to trigger explanations in subsequent discourse als well as their strong preference for anaphoric reference to only one of the two arguments in those explanations. In addition to our theoretical work, we are conducting psycholinguistic experiments to test our theory, using both online (e.g. eyetracking during reading) and offline (discourse continuation experiments) methods.
The area Finiteness and Dependency in clausal embedding examines the concept of finiteness. We are attempting to understand the different degrees of finiteness that can be discerned in different embedded clause types in terms of the extent of their dependence on elements of the matrix clause. This sort of dependency operates on at least two dimensions -- in the reference of the subject and in the temporal interpretation. We are investigating to what extent these degrees of dependency are realized in the syntactic structure, and how this interacts with the semantic properties of clause-embedding predicates.
The area Clause-embedding predicates in creole languages focuses on the possible changes in clause-embedding predicates due to language contact. Owing to the extreme circumstances under which they arose, creoles are the result of intimate language contact between a superstrate language (typically a European language) and several substrate languages (e.g. West African languages). One of the leading questions in this area will be to what extent the syntactic and semantic properties of clause-embedding predicates in creoles result from the superstrate or substrate languages, or whether these predicates developed their properties independently from general principles of syntactic computation and the syntax-semantics interface.