In lieu of an abstract, here is a brief excerpt of the content:

Reviewed by:
  • Parametric variation: Null subject in Minimalist Theory
  • Gréte Dalmi
Theresa Biberauer, Anders Holmberg, Ian Roberts, and Michelle Sheehan. 2010. Parametric variation: Null subject in Minimalist Theory. Cambridge: Cambridge University Press. Pp. vi + 368. US $110.00 (hardcover).

The Null Subject Parameter is an important component of recent generative grammatical theorizing. This book sets as its main goal the defense of the parametric approach to cross-linguistic variation within the Minimalist framework. It consists of eight chapters and a substantial introduction. The term “null subject” is used in this book in a broad sense. It includes referential null subjects, non-referential expletive null subjects, and generic null subjects. Languages that allow null subjects in this broader sense outnumber those that do not (Haspelmath et al. 2005). Since Rizzi’s (1986) seminal work on null subjects it has become clear that null subject languages (NSLs) come in various types. [End Page 470]

In the introduction, Roberts and Holmberg give the following classification of NSLs:

  • Type 1 Expletive null subject languages (e.g., German, Dutch) allow expletive null subjects in impersonal passives.

  • Type 2 Partial null subject languages (e.g., Finnish, Brazilian Portuguese) allow referential null subjects under special circumstances: while 1st and 2nd person referential subjects can be freely omitted, 3rd person referential subjects are dropped only if they have a clearly identifiable antecedent in a higher clause. In Type 2 languages 3SG generic subjects are always null.

  • Type 3 Consistent null subject languages (e.g., Italian, Greek) are characterized by a number of diagnostic features: (i) stylistic inversion of the subject; (ii) absence of the complementizer-trace effect; (iii) rich agreement morphology.

  • Type 4 Radical null subject languages (e.g., Chinese, Japanese) allow null subjects as well as other null arguments in active finite clauses.

In chapter 1, Roberts starts out with Holmberg’s observation that 3SG referential null subjects never co-occur with lexical expletive subjects. He derives this incompatibility from the fact that 3SG referential null subjects behave like weak pronouns: they are never found in post-verbal position, they cannot be left-dislocated but they can license floating quantifiers. This leads Roberts to conclude that the canonical [Spec, TP] subject position in stylistic inversion and complex inversion is occupied by a referential null subject. In morphologically rich consistent NSLs the T head can attract defective goals because it has a [+D] feature. Defective 3SG referential subjects move to the designated subject position to satisfy the Extended Projection Principle (EPP), after which they must delete due to their defective nature.

In chapter 2, Holmberg discusses (i) expletive null subjects, (ii) referential null subjects, and (iii) generic null subjects in Type 2 and Type 3 NSLs. While consistent NSLs allow referential 3SG null subjects in certain well-defined contexts, they never allow generic 3SG null subjects. Partial NSLs, on the other hand, allow all three types of null subject listed in (i)–(iii). Holmberg (2005) derives this distribution of 3SG null subjects from the presence of the [+D] feature on the finite T head in Type 3 Consistent NSLs and from the absence of the same feature in Type 2 Partial NSLs. The [+D] feature reflects the pronominal nature of verbal inflections appearing under T0. Given that Type 3 Consistent NSLs do not allow generic null subjects at all, 3SG generic inclusive null subjects are not found in them either. Some Type 2 Partial NSLs allow both the generic inclusive and the generic exclusive interpretations of 3SG null subjects (e.g., Brazilian Portuguese), while others allow only the generic inclusive interpretation of 3SG null subjects (e.g., Finnish) in active finite clauses.

In chapter 3, Holmberg and Sheehan investigate syntactic environments that can license 3SG referential null subjects in active finite subordinate clauses in three languages: Finnish, Brazilian Portuguese, and Marathi. They convincingly show that the lexical antecedent licensing such referential null subjects is not a controller in [End Page 471] the sense of Landau (2004); null subjects alternate with pronominal subjects in these environments. Thus, the relationship that holds between the lexical antecedent and the null subject resembles pronominal binding rather than canonical control. The only peculiarity of 3SG referential null subjects is that they...

pdf

Share