Reviewed by: Possible and probable languages: A generative perspective on linguistic typology Marco Nicolis Frederick J. Newmeyer. 2005. Possible and probable languages: A generative perspective on linguistic typology. Oxford: Oxford University Press. Pp. x + 278. US $45.00 (softcover). The stated goal of the book is “reorienting generative grammarians in a direction that is better able to capture the distinction between the possible and the probable in language” (p. 27). To achieve this, the author does not present new research results of his own, but offers a detailed critique of the dominant functionalist and formalist approaches to typological generalizations and parametric correlations. In a nutshell, Newmeyer argues that both “formalist approaches”, in particular the Principles and Parameters model (Chapter 3) and “functionalist approaches” (Chapter 5), spectacularly fail when it comes to accounting for typological generalizations. Their nonabsolute, stochastic nature leads the author to propose that their explanation does not lie in the realm of grammatical theory (p. 104), but rather in a theory of performance, such as the one put forward in Hawkins (2004). Newmeyer’s Universal Grammar is thus far lighter that what is assumed in most generativist studies: “UG tells us what a possible human language is, but not what a probable language is” (p. 104). This catchy sentence essentially amounts to discarding parametric theory as part of the generative enterprise. This is precisely the goal of Chapter 3 of the present book, as we will see below. Chapter 2 offers a very interesting, highly accessible and well-documented historical overview of mainly the GB and MP models, as well as Kayne’s Antisymmetry. Several views of what constitutes a “parameter” are clearly presented (e.g., Baker’s macroparameters and Kayne’s microparameters) and examples of several parameters are given (pp. 42ff.), including Rizzi’s Subjacency Parameter, the Null Subject Parameter, and the Head Directionality Parameter. The author concludes: “the commitment of generativists to account for cross-linguistic variation was greatly facilitated by the introduction of the Government-Binding theory in the early 1980s and its conception that principles are parametrized” (p. 72). Despite that conclusion, Newmeyer, at the outset of Chapter 3, sketches out the components of his own conception of linguistic theory as follows (taken from p. 73): (1) Notice that (1b) refers to the kind of language specific rules characterizing the EST model. By Newmeyer’s own admission, the Principles and Parameters model had “greatly facilitated” the emergence of accounts of cross-linguistic variation, so it is hardly obvious why something along the lines of (1b) benefits a theory aiming at characterizing typological generalizations and crosslinguistic variation. Perhaps aware of the problem, the author defines these language-specific rules quite cryptically, as follows: “Essentially, they are parameter-settings ‘detached’ from the parameters themselves (which are hypothesized not to exist)” (p. 74). Furthermore, Newmeyer’s arguments against Parametric Theory appear questionable and have been harshly criticized within the generativist community (see Holmberg and Roberts (2005) for a detailed and thorough critique of Newmeyer’s arguments). Given the strict space limits of this review, I will limit myself to pointing out a couple of inadequacies in Newmeyer’s account of Parametric Theory. First, the author explicitly assumes that parameters have a “putatively rich deductive structure” (p. 45). This has never been the standard idea in the field (with the possible exception of Baker (2001)). Rizzi (2004) argues in fact that “Parameters have virtually no internal structure, certainly no inherent deductive structure: they are simple binary choice points as elementary and easy to fix as possible. The deductive structure has to do with the interaction of parameters with the rest of the structure of UG, the system of principles, not with their inner constitution” (p. 333). Second, due partly to this conception of parameters, Newmeyer considers Gilligan’s (1987) survey on null subject phenomena in one hundred languages as damning for theories of pro-drop such as Rizzi’s (1982), so much so that the conclusion of the discussion is that “the Null Subject Parameter is no more” (p. 92). Such conclusion is far too pessimistic. While the author correctly observes that “five language types are attested whose existence neither theory [i.e., Rizzi’s and Safir’s...
Read full abstract