On the Syntax of "Have To"

On the Syntax of "Have To"

  This paper, based on the current theory of phrase structure, aims to analyze the internal structure of "have to" constructions, especially in American English. Radford (1997, 2004a, 2004b), following Chomskyan framework of Transformational grammar (Chomsky 1981, 1995, 2001), argued that two potential positions exist for the infinitival particle "to"-either T or Aux. If Radford"s analysis is to be accepted, then the sequence of "have" followed by "to" with a modal-like meaning casts a problem in its phrase structure. This paper will argue that if "have" is a lexical verb as maintained in Huddlestone and Pullum (2002) and Quirk, et al (1972, 1985) and if "to" is either a T constituent or Aux constituent depending on the existence of a negative element, then the "have to" construction is bi-clausal rather than monoclausal. In other words, "have" is in the matrix TP, whereas "to" is in the embedded TP. Therefore, the surface subject of the "have to" construction is a raised subject from the embedded TP to the matrix TP, making "have" a raising predicate, just like "seem" and "appear." To support this argument, this paper will offer empirical support from the distribution of certain adverbials such as "yet," "only," and "still" as well as expletive subjects like "there" in the "have to" construction One consequence of this analysis is the ability to account for the contraction phenomenon, where "have to" is pronounced as [hæfta] because the A-trace of the subject-to-subject raising does not block the contraction between "have" and "to."