Implicational hierarchies in clausal complementation

A common feature of languages from diverse families and types is the expression of thoughts, beliefs, utterances, claims, conjectures, wishes, and many other cognitive and mental concepts via clausal subordinationverbs corresponding to these concepts (believe, say, think, want) combine with a subordinate (complement) clause. Subordination involves construing a dependent state of affairs in relation to a main state of affairs, and theories of clausal complementation investigate how languages tie together the syntactic, morphological, and semantic properties of configurations conveying the different concepts. Although languages vary in the shape and make-up of complementation configurations, a striking observation is that there is a dependency between the meaning of a subordination configuration and its morphosyntaxchanging one often results in a change of the other. For instance, if I forgot to water the plant (to water is an infinitive), it is entailed that I did not water the plant. On the other hand, if I forgot that I watered the plant already (I watered is a finite construction), it is expressed that I did water the plant. Typological studies have shown that, despite variation in the morphosyntactic coding of subordination, the semantic classes of complement clauses are cross- linguistically very similar. The current project brings together previously established bodies of knowledge from functional-typological, as well as structural-grammatical frameworks and combines them with new methodologies and tests. Studies on subordination have brought out a range of interesting phenomena, however despite certain well-established tendencies, no property could be isolated that holds cross-linguistically. This project pursues the hypothesis that, although there are no absolute universals in the area of complementation, there are nevertheless implicational properties (relative universals) which can be defined and tested across languages. The specific hypothesis is that there is a possibly universal implicational subordination hierarchy, which is defined semantically and detectable through a diverse set of grammatical properties. Developing a typology of complementation is not only relevant for different areas of linguistics, but for a wide range of language and cognitive studies. Questions about the relation between language and other cognitive systems can only be answered if a comprehensive understanding of the properties of language is available. The typology of subordination thus provides a new testing ground for gaining further insights into the nature of complex mental concepts such as attempts, decisions, desires, beliefs, and as a result the workings of the human mind.

Much work in modern linguistics has been devoted to finding universals-properties that hold across all languages-since by investigating and comparing individual languages, we can draw conclusions about the abstract cognitive system taken to underlie human language. This project investigated a universal of the implicational type, where certain properties, though not universal themselves, universally entail other properties. The project examined configurations formed by combining verbs such as claim, decide, or manage with subordinate (=complement) clauses (e.g, that it is raining or to eat cookies) and concluded that there is a universal "implicational complementation hierarchy" (ICH), which is based on elements of general human cognition-the way humans build complex representations from simpler ones. The ICH states that there is predictable behavior in the interactions between the grammatical (morphological, syntactic) properties and the meaning of complementation configurations. The project developed tests to detect the effects of the ICH cross-linguistically and showed that complement clauses with simpler meanings also tend to be less complex morphologically and syntactically. For example, operations that are generally restricted to simple clauses (such as word order permutations) tend to be more easily established across complement clauses with simpler meanings (e.g., events without temporal specifications). The project identified a connection between parts of (simple) clauses and types of complement clauses and investigated different clausal domains using the methodology of containment-i.e., the general observation of generative grammar that larger structures (morphological, syntactic, as well as semantic) contain smaller structures. Such Matryoshka-doll-like incremental building then allows us to make predictions about structural complexity and the (im)possibility of different configurations and combinations. While there is no specific universal grammatical property that can be tied, in an absolute manner, to a particular part of a clause or complementation class, the universality lies in the existence of three broad classes of clausal domains, which correspond directly to the cross-linguistically observable classes of complement clauses. The subprojects studied the three different structural domains in parallel: their morphological coding (differences such as inflection, infinitive, gerund), syntactic organization (e.g., the richness of syntactic structure), and semantic interpretation (e.g., the presence and type of temporal information). The phenomena investigated in relation to the largest clausal domain included various dependencies spanning the matrix and subordinate clause (e.g., agreement, case) and the mapping between structure and the discourse context. The phenomena investigated in the middle domain included the structure and meaning of tense and aspect, as well as dependencies such as case and agreement. The phenomena investigated in the smallest domain included a typology of cross-clausal argument sharing (cases in which two predicates share an argument) and argument structure alternations such as active and passive.