Читать книгу A Companion to Chomsky - Группа авторов - Страница 49

1 1 Our intended readership is broad, extending beyond generative linguists. For a more comprehensive list of the enduring discoveries of generative syntax that is aimed primarily at generative linguists, see D'Alessandro (2019).

2 2 Generativism has never denied that acceptability judgments are influenced by many factors demonstrably unrelated to grammaticality, such as processing costs and cultural norms. The name “acceptability judgment” (as opposed to “grammaticality judgment”), which originates with Chomsky (1965), explicitly acknowledges this fact. Generative linguists aim to control for these extraneous factors in their elicitation procedures, or alternatively take these confounds into consideration when developing their analyses.

3 3 For most of the generativist period, acceptability judgments have been collected using an informal methodology that does not meet the experimental standards expected in neighboring fields such as psychology (see Schütze and Sprouse 2013, 30 for details). Although it has attracted criticism (Edelman and Christiansen 2003, Ferreira 2005, Wasow and Arnold 2005, Featherston 2007, Gibson and Fedorenko 2010, 2013; see Hill 1961 for an early critique), recent research has demonstrated that, thanks to large effect‐sizes, this method of collecting judgments is as reliable as “formal” collection methods, with convergence between the two ranging between 86% to 99% (Sprouse and Almeida 2012, 2013, 2017a, 2017b; Sprouse, Schütze and Almeida 2013, Schütze 2020).

4 4 Viewing transformations as descriptions of nonlocal dependencies came later and indirectly, via Chomsky's adoption from Hockett (1958) of deep versus surface structure in Chomsky 1964 and via his technical innovation of traces in Chomsky 1973. See Chomsky 1977 for a partial subsumption of traces under silent pronouns (or, more precisely, PRO) and Koster 1978 for an early argument for abandoning the notion that any nonlocal dependencies are established via the application of transformational rules.

5 v The asterisk symbol is used to denote that an example sentence is judged by native speakers as unacceptable. In the absence of any observable confounding factors, this judgment is viewed as evidence that the sentence in question is also ungrammatical.

6 6 For the sake of simplicity, the phrase marker in (8) avoids representing the X′‐levels characteristic of X′‐theory.

7 7 This is a crude description, based on a generalization that provoked Chomsky's (1977) theory of Subjacency (see the concluding paragraphs of Section 4.3.2). For a more contemporary perspective, in which many more “hidden” dependencies are postulated than are represented in the examples in (14) (and (18)), see the discussion following example (28) and the references provided there.

8 8 The idea that question phrases in wh‐in‐situ languages are tails in a top‐gap A′‐dependency was first advanced by Aoun, Hornstein, and Sportiche (1981). At first glance, it appears that the top‐gap and tail‐gap A'‐dependencies that are utilized to form constituent questions across languages differ in their locality constraints, as the former appear to be establishable across island boundaries, unlike the latter (compare (i) below with (14) in the main text). Subsequent research has shown that this difference is only apparent. In island‐obviating cases, the entire island (Nishigauchi 1986, Choe 1987, Pesetsky 1987) or a Q‐morpheme associated with it (Cheng 1991, Hagstrom 1999, Cable 2010) is the tail of the dependency, not the question phrase contained within it.Ali [ISLANDkim gel‐di diye] gülümse‐di? [Turkish]Ali who arrive‐PST COMP smile‐PST‘Who is it such that Ali smiled because s/he arrived?’

9 9 These constraints form the Binding Theory. See Chomsky (1981) for an influential formulation.

10 10 Precisely how to define a binding domain has been a long‐standing topic of debate. For a definition based solely on hierarchical syntactic structure, see Chomsky (1981). For a definition which considers hierarchical structure and linear order (in particular, precedence) see Langacker (1969), Lasnik (1976), and Bruening (2014). For definitions that define binding over functional relations such as obliqueness, which are only partly reflected in hierarchical structure, see Pollard and Sag (1994) and Bresnan (2001).

11 11 This assumption is not universally upheld. For instance, offshoots of mainstream generativism such as Head Driven Phrase Structure Grammar (Sag and Pollard 1991), Lexical Functional Grammar (Bresnan 1982), and Simpler Syntax (Culicover and Jackendoff 2005) maintain that the difference in argument‐structure properties between seem‐ and promise‐type predicates is not reflected in syntactic structure. In the generative literature, this idea originates with Bach (1977).