The present paper aims to provide an improved understanding of the grammaticalization of parenthetical complement-taking predicates (CTPs). We point out problems in existing accounts (focusing on Boye & Harder 2007 and Brinton 1996, 2008) and propose a synthesis which incorporates the main insights of those accounts, while avoiding the problems. Based on the theory of grammatical vs. lexical status in Boye & Harder (2012), our proposal invokes as a key theoretical innovation a distinction between constructional slots for discursively secondary material, and CTP clauses as fillers of such slots. We argue that this distinction allows us to be precise not only about the status of parentheticals with respect to the grammatical vs. lexical distinction, but also about the status of sentence adverbs.

Research on parentheticals and CTP clauses has been focused on constructions where the CTP clauses co-occur with clauses that are sentence-like and propositional. We broaden the scope to include constructions with clauses that are reduced and constructions that designate illocutions or states-of-affairs rather than propositions. We argue that while all CTP are structurally qualified for undergoing grammaticalization, both the syntactic and semantic type of the clauses that co-occur with them restrict grammaticalization in certain ways.

While focusing on English, we regard our proposals and hypotheses as potentially applicable also to other languages. Accordingly, we include data also from other languages without however pretending that this is sufficient to actually confirm our hypotheses crosslinguistically.