I think and other complement-taking mental predicates: A case of and for constructional grammaticalization

Abstract

This article provides a critical assessment of previous claims that complement-taking mental predicates (CTMPs) like I think, I suppose , etc. are instances of grammaticalization. In so doing, it calls attention to the main problems one encounters when applying commonly agreed-upon grammaticalization criteria to CTMPs. It is demonstrated that the syntactic mobility of CTMPs is crucial to their decategorialization while being at odds with the parameter of positional fixation. In addition, CTMPs' ability to occur both in adverb-like, parenthetical positions, and in verb-like, clause-initial position, suggests that their decategorialization is incomplete. The possibility to reactivate productive verbal properties in expressions that display a high degree of formulaicity is explained in terms of grammatical persistence. Another challenge facing the grammaticalization of CTMPs is the existence of variation in terms of tense, aspect and modality. The aforementioned obstacles are documented by present-day spoken British English corpus data. It is argued that, rather than regarding them as pragmaticalized or lexicalized as has alternatively been suggested, CTMPs should be approached from the usage-based perspective of constructional grammaticalization, which is concerned with the grammaticalization of schematic constructions that are part of a wider taxonomy rather than being isolated sequences.