Explaining diverging evidence

The syntactic status of clause-initial complement-taking predicates has been controversially discussed in the literature with analyses ranging from main clause to parenthetical. This chapter sheds light on the question by providing a usage-based account of 200 occurrences of initial I think in a corpus of spoken English. It investigates to what extent the two formal cues (i) presence or absence of the that-complementizer and (ii) prosodic prominence provide evidence for the relative prominence of I think. The data present diverging evidence, which can be reconciled however by (i) adopting a dynamic model of grammar and (ii) reassessing the function of the that-complementizer in spoken language, viz. as a filler used for rhythmic purposes or to give weight to the initial clause.