Is there really a syntactic category of subordination?

While different criteria are used in the literature in order to define subordinate clauses, these clauses are generally assumed to form a syntactic category, in the sense of a syntactically defined class that is relevant to speakers of individual languages. Many phenomena that are usually regarded as distinctive for subordination, however, do not actually provide evidence for such a category. This is illustrated in the paper with regard to syntactic embedding. Different criteria provide conflicting evidence as to the embedded vs. nonembedded status of particular clauses, the same criteria give different results for the same clause types in different contexts, and individual criteria do not always make it possible to identify distinct clause classes. This is because the various phenomena that are usually regarded as evidence for embedding are not actually motivated in terms of the same principles, nor are they syntactically motivated. Rather, these phenomena reflect a variety of semantic and pragmatic principles, and in some cases originate from diachronic processes independent of the syntactic status of the relevant clauses