Introduction

Abstract The term nonfinite, though potentially misleading, will encompass traditional infinitives, gerundials (gerunds, gerundives, supines), and participles. No theoretical significance is attached to any of these labels. They refer to slightly different clusters of features potentially constituting distinct formatives which may or may not have similar properties crosslinguistically. The non-discreteness of the categories and problems with (non)finiteness in the traditional sense have prompted some scholars to use the neutral term converb(see Haspelmath and Ko nig 1995; Auwera 1998). In English, infinitives are morphologically distinct from gerundials and the -ingparticiple, for which the traditional terms are convenient labels for the status of the formative as modifier/adjunct or argument. There are also aspectual and other differences. The fluidity of the feature content of these formatives in language change is an argument against the traditional labels, except that the changes are expositorily simpler to track with the standard terminology. The term nonfiniteis retained because prototypical infinitives, gerundials, and participles are in fact nonfinite in the traditional sense, that is, they have no subject person agreement. On the other hand, there is no simple feature [ ± finite] (in the traditional sense) that works crosslinguistically (cf. Haspelmath 1995: 4ff.; Ledgeway 1998: 58). Moreover, since nonfinite structures can have tense independent of the matrix tense (depending on the nature of the matrix predicate), and this infinitival tense can be manifested morphologically, a simple feature [ ± tense] also fails to distinguish finite and nonfinite structures.