The traditionally held view that grammaticalisation should be a semantically-motivated process (as discussed, for example, in Hopper and Traugott 2003: 75, summarising Bybee 1985; Bybee and Dahl 1989; Heine, Claudi and Hünnemeyer 1991; Heine et al. 1993; and Heine and Kuteva 2002) has not been without its critics. One particular area of study, so far for the most part unchallenged, is Fischer’s (1994, 1997, 2007) treatment of the grammaticalising periphrastic modal of obligation, have to, in English. She provides a syntactically-led grammaticalisation account in which it is believed that the present-day, developing modal form had links with an earlier, Middle English expression in which the transitive object of the infinitive was located in pre-infinitival position, shared by both the infinitive and have. The syntactically-determined explanation for the grammaticalisation of this modal expression also takes account of the fact that many of the visible grammaticalisation effects are demonstrated to have taken place following the general shift in word order during the Middle English period, from an SOV to an SVO order. In the present study, the alternative viewpoint (first proposed by Brinton 1991) in which the syntactic word order shift is seen to be most frequently associated with transitive objects that referred to entities incapable of acting as possessors is expanded to suggest a context-induced path of grammaticalisation (Heine, Claudi and Hünnemeyer 1991; Heine 2002). In addition, the shift of the object to post-infinitival position is seen to be unavoidably linked to the prior development of obligation senses in the older construction, so necessitating a semantically-motivated explanation of the grammaticalisation route for have to.
