The word ‘semi-modals’ is not a precise term. It refers to a loose constellation of verb constructions which, according to many commentators (for example, Hopper and Traugott 22003, Bybee et al. 1994, Krug 2000), have been moving along the path of grammaticalization (see sections 1.2, 11.3) in recent centuries. The semi-modals are probably the most cited cases of grammaticalization in the ongoing history of English. Among these, in turn, the prototypical, most indubitable cases of semi-modal status are be going to and have to, which are long-standing representatives of this evolving class, traceable back to the late ME or EModE period – see Krug (2000), Biber et al. (1999: 487), Danchev and Kytö (2002); also Mair (1997) on be going to; Fischer (1992) on have to. It is well known that in these two constructions, the lexically independent verbs have and go have, over the centuries, gradually acquired an auxiliary-like function in construction with the infinitive to.

This chapter is naturally to be seen as complementary to the previous chapter on the modals; and just as we did not refrain from referring to the semi-modals where relevant in Chapter 4, we will where relevant return to the modals in Chapter 5. To some degree, the boundary between these two chapters is artificial.

Auxiliary–lexical verb gradience

Krug (2000: 3–5, passim) argues for a class of ‘emergent modals’, a small prototype category on the cline between auxiliary and main verb, including not only be going to and have to, but also have got to and want to, as well as one or two more marginal candidates, such as need (to) and ought to.