Catálogo de publicaciones - libros

Raising and control have been central concerns of generative syntax since the 1960s and continue to be an empirical focus of every comprehensive model that has come along since. The analysis of these constructions in each framework has typically relied crucially on the most fundamental assumptions underlying that framework. Thus, raising and control continue to provide an excellent window into generative models of syntax, and a useful tool for measuring the validity of their claims. In the 40 years since the publication of Rosenbaum (1967) and the 33 years since the publication of Postal (1974), attention to these constructions has persevered through each significant paradigm shift in generative syntax. Interest in these constructions has also broadened (from an initial focus on English and French) to include analyses of similar (or apparently similar) grammatical phenomena in a wide range of languages. Most recently, interest in raising and control has once again surged with the rise of the Minimalist Program. At the same time, some of the most recent analyses venture into relatively underexplored languages and/or grammatical phenomena. Concerned as we are with empirical results informing theoretical paradigms, we think that renewed attention to these two constructions, combined with an expanding empirical basis for analysis, makes this a particularly appropriate time to produce a book that gathers in one place some of the more interesting work being done on the topic at this time.

While it is clear that the study of language development contributes to linguistic theory, it is perhaps less widely recognized that by examining the time course of language development we can integrate the study of language into the broader study of biological development. The hope persists that this type of analysis will play a role in the genetic underpinnings of language, as it has already done in some areas of grammar.

The Minimalist Program advances explanatory adequacy to the extent that stipulative principles and filters of GB are deduced from the smallest number of simple, ‘natural’ axioms. Fundamental among these are: (i) Sound and meaning are ineliminable: there are lexical features and properties. (ii) There is a (recursive) structure building operation: Merge (A & B) produces C. (iii) The language faculty interacts with external systems: to be usable, the objects of the syntactic component must be legible to the interfaces.

Finiteness has been considered to be the most prominent syntactic property of clauses because it determines whether certain syntactic operations can apply across a clause boundary. More specifically, finiteness has been used as a cover term for the ability of a clause to check nominative case on its subject. In such a way, finiteness regulates the distribution of a nominative subject, either an overt DP-subject or a null subject pro . In a finite clause, the subject is able to check its nominative case and thus becomes inaccessible to further computation, whereas in a nonfinite clause the subject cannot check its nominative case, and can either be a PRO resulting in control structures or be targeted by a higher probe in raising constructions. According to standard approaches to finiteness within the Principles & Parameters and early Minimalist frameworks (Chomsky 1981, 1993, 1995), nominative case is checked by a finite T (= INFL) functional category; T is able to check ( nominative) case (i.e. it is finite) when it is fully specified for Tense and Agreement. In the most recent minimalist approaches (Chomsky 2000, 2001a, b), nominative case checking is considered to be the by- product of subject-agreement valuation in T. Such an approach implies that finiteness is a property of agreement.

This chapter investigates various licensing constraints imposed on shared subject arguments in Romanian obligatory control constructions and argues for an analysis of obligatory control (OC) in this language analogous to that of raising predicates. On the one hand, the discussion contributes to the current debate with respect to whether OC can and should be construed as raising or not (i.e. the Hornstein–Landau debate1), and on the other hand, the analysis provides an account of seemingly optional subject dislocation that is intimately tied to the Theme–Rheme sentence partitioning in Romanian and, consequently, independent of the control phenomenon per se. With respect to the first point, I propose that movement out of controlled clauses is a parametrized option made available by UG and governed by well-defined conditions. Specifically, it is available in languages where complements to control verbs lack phasal status, or can void phasehood, a proviso that guarantees an active subject goal available to both thematic and non-thematic checking operations with matrix probes. Regarding the second point, I show that dislocation of the subject DP, which may but need not occur, is not incumbent on morpho-syntactic featural requirements related to OC (such as Case or theta-role valuation), but determined by well-defined semantico-pragmatic constraints, such as topic and focus movement, construable as OCC features (Chomsky 2001b) on the various probing heads. This analysis has the merit of limiting the amount of movement required by reductionist approaches to OC, accounting for optionality in a systematic manner, and providing adequate empirical coverage of the phenomena under discussion.

Traditional classifications of control distinguish two types: obligatory control (OC) and non-obligatory control (NOC). Recently, it has been demonstrated that OC should also be divided into two types: exhaustive and partial control (cf. Landau 2000). This more refined view of OC is taken to be evidence against the movement theory of control (MTC – cf. Hornstein 2001), the argument being that partially controlled PRO does not behave like a trace. In this paper, I contribute to advancing our understanding of control by looking at agreement and flotation in control configurations and developing an analysis for partial control that results from movement plus stranding.

It is well known that some languages, despite permitting null subjects, do not display the full array of characteristics usually associated with the Null Subject Parameter as proposed by Rizzi (1982, 1986). In the 1980s, there was much debate about languages such as Chinese, which display no overt verbal agreement and still allow for the subject position to be null. More recently, languages like Brazilian Portuguese (BP) and (colloquial) Finnish have been discussed. In those languages, third person null referential subjects are not allowed in matrix contexts, which has been related to the fact that (at least in BP) verbal agreement is ‘poor’. However, third person null referential subjects are productive in embedded contexts in both languages, which is problematic for theories which try to explain the Null Subject Parameter.