Parametric variation: Null subjects in minimalist theory (review)

Parametric variation is a case study in parametric approaches to crosslinguistic variation, as well as an argument in favor of such an approach. It frames the discussion of parameters around various types of null-subject languages (NSLs) and the null-subject parameter (NSP) as the parameter governing their crosslinguistic distribution. The choice of the parameter is very timely; the NSP is arguably one of the most studied parameters to date, but, as the authors acknowledge in the introduction, its status remains highly controversial. The questions that feature prominently throughout the volume are: (i) How do languages differ with respect to what kinds of empty subjects they allow? (ii) How is the extended projection principle satisfied in NSLs? (iii) What does null subjecthood correlate with? (iv) What is the status of parameters in the grammar? and (v) What is the organization of parameters in the grammar?

The book starts with an introduction by Ian Roberts and Anders Holmberg. The first part of the introduction gives an overview of the research on null subjects, starting with Apollonius Dyscolus’s observations about Ancient Greek, and sets the scene for the chapters in the volume. The second part turns to the debates surrounding the NSP, and the third part examines the nature of parameters themselves and the architecture of the grammar that contains them. It discusses and counters the arguments that have been levied in the literature against the concept of parameters, focusing on Newmeyer (2004, 2005), to whose criticism the authors respond in most detail, defending a parametric approach against Newmeyer’s rule-based approach or Hawkins’s (2004) performance-based approach.

Ch. 1, ‘A deletion analysis of null subjects’ by Roberts, poses the question of whether the subject position in NSLs is occupied by an empty element or remains empty, with the extended projection principle (EPP) feature being checked by rich inflection. Building on Holmberg 2005, Roberts argues against the latter approach, since on minimalist assumptions, null subjects cannot be licensed by features of rich inflectional heads, as these features are not specified in the lexicon and themselves need to be licensed (by features of the subject). He considers the possibility that null subjects undergo incorporation to T, which he treats as a type of Agree between a Probe and a defective Goal containing a proper subset of the features of the Probe, building on his analysis of Romance clitics (cf. Roberts 2010). Ultimately, however, he rejects such an account in favor of a deletion-based one.

Ch. 2, ‘Null subject parameters’ by Holmberg, starts with Rizzi’s (1982) formulation of the NSP, which distinguished different types of NSLs by positing different types of pronominal inflection, but Holmberg reinterprets it as a difference in whether T has an uninterpretable D feature or not, and uses this to capture the distinction between consistent NSLs, such as Italian, and partial NSLs, such as Brazilian Portuguese, Finnish, Icelandic, or Marathi. To derive the two types of null subjects, Holmberg resorts to two mechanisms: Roberts-style incorporation and deletion. Consistent NSLs for him involve incorporation of a defective pronoun lacking a D feature into T. The D feature of T is valued by the null Aboutness-shift topic of the kind proposed by Frascarelli (2007). Partial NSLs, by contrast, do not have a D feature in T, which captures the indefinite interpretation of null subjects in such languages. Holmberg extends the proposal to capture the differences between Icelandic (a partial NSL) and Mainland Scandinavian languages (non-NSLs). Since in both types of languages T lacks D features, further microparameters are necessary. These lie in what types of phi-features reside in T, whether the EPP is checked by the element that values T’s phi-features, and whether the specifier of TP has to be filled by an overt element. The resulting system involves many ‘null-subject microparameters’, captured as features of T heads.

Ch. 3, ‘Control into finite clauses in partial null-subject...