In this ambitious book, D. Gary Miller’s objective is ‘to investigate a core of nonfinite structures, together with potential analyses, to serve as an empirical basis for further work on [their] synchrony and diachrony’ (xv). It is thus a critical review of descriptive and theoretical literature, with some novel proposals. What sets this book apart is the range of phenomena covered: M gives full attention to gerunds, infinitives, participials, and small clauses, and strives to unify their analyses, incorporating both historical change and synchronic syntax. The major diachronic domain is the history of English, with synchronic investigations picking among (mostly Indo-European) languages that raise special challenges with their nonfinite constructions. Minimalism (with its precursors) is the framework of most of the theoretical discussion. The book consists of one introductory and twelve numbered chapters, followed by a discussion and reference list of the Old and Middle English text sources, extensive general references (44 pp.), an admirably detailed language index with subentries by construction, and a subject index; unfortunately, authors are not indexed.
M’s survey of data and theories and thorough documentation of historical progressions that need to be accounted for make this volume of greatest interest to generative syntacticians. For historical linguists the appeal may be somewhat less; a fair amount of the data is familiar (albeit usefully assembled), though some widely held analyses thereof are contested. The degree of technical sophistication in the nondescriptive sections is high: it is a demanding read even for graduate students specializing in syntax. Still, anyone working on nonfinite structures is certain to benefit from the discussion.
In Chs. 0–2, M defines key terms and lays out important theoretical claims. His analyses build on proposals in Schütze 1997, whose central goal was to argue for a complete separation of structural licensing of DPs (‘abstract Case’) from the morphological case-marking and agreement system. Whether or not a clause licenses its subject position, Schütze claims, is independent of the clause’s case- and agreement-checking properties (e.g. nonfinite clauses may check nominative case on their subjects without licensing them), and PRO has the same licensing requirements as overt DPs but is blocked in some circumstances. Consequently, clauses with overt and PRO subjects contrast with those whose subjects must raise out.
M adapts Schütze’s proposals, most notably by suggesting that individual predicates can select for whether Accord, Schütze’s combined case- and agreement-checking operation, applies in its complement clause. M ties the distribution of PRO to the presence/absence of agreement, thus no longer maintaining a complete separation between the licensing and inflectional systems. Further, Schütze assumes that, while some languages do not morphologically realize agreement, languages that do must spell it out in all instances; M instead allows for optional spell-out.
M defines nonfinite as ‘lacking subject person agreement’, explicitly allowing for semantically independent and morphologically marked ‘tense’ on a nonfinite verb form. Nonfiniteness is a property of a Mood head between Comp and Tense, and is compatible with agreement as a marked option. In attempting to establish what it means for a clause to ‘have tense’ or be specified [ + Tense], M allows for infinitives that inflect for tense (e.g. the Ancient Greek future infinitive) and counts aspectual auxiliaries as constituting Tense in infinitives (e.g. to have gone).
Both the book’s presentation of the facts and its discussion of theoretical proposals are extremely detailed, reflecting M’s goal of showing that his ideas can be technically worked out across a wide range of data. Since this material does not lend itself to concise summary, we restrict our comments to an overview of the phenomena analyzed and general remarks on the approach and presentation.
In Ch. 3, M develops ideas about control from Landau 1999. He suggests that Agr and Tense features of the infinitive sometimes block inheritance of phi-features by PRO from its controller, explaining the exhaustive versus partial control distinction. He also discusses nonobligatory [End Page 443...