Som-satser med och utan som

Sammanfattning: This thesis deals with different types of subordinate clauses in Scandinavian and English, e.g. relative clauses and subordinate wh-questions, that all share the characteristic of having a visible or invisible constituent in a pre-complementizer position (spec-CP). The pre-complementizer constituent, the subject or a non-subject, is sometimes followed by a complementizer (in Cº). In English for example, that is used in that-relatives: obligatorily after an invisible subject, otherwise optionally.Swedish uses the complementizer att ('that') in that-clauses but another complementizer, som, in various subclauses with a pre-complementizer constituent. For this reason, the different kinds of subclauses that are focused on in this study are captured by the term som-clauses. The problem in focus is the question of why a complementizer like som or that is obligatory in some som-clauses, optional in others, and in still others (such as pronoun-headed relatives) would be ungrammatical.The analysis of the som-clauses is carried out within a principle- and parameter-based framework. Initially, a split CP is proposed, offering a specific projection for the fronted constituent in som-clauses and for the complementizer som. That on the other hand is placed in a lower projection, an organisation supported by the combined use of som and at ('that') in relatives in colloquial Danish.The different positions of som and that indicate a functional diversity as well. This in turn has provided a clue to the further analysis, which is also founded on two basic assumptions: 1) a fronted invisible constituent has to be formally identified and 2) every clause has to be typed, for instance as a declarative or an interrogative. Som is primarily associated with the former function, whereas that is a (declarative) clause type complementizer.Contextual identification of a fronted invisible constituent and of clause type may in some cases permit omission of som and that. Evidently, there is no general demand for an explicit subordinator in subclauses. However, the obligatory use of som in Mainland Scandinavian after (visible) subjects in embedded wh-questions points to a need of a subordinate feature in some cases.