Essays in Syntactic Theory

Samuel David Epstein

• Publisher: Routledge
• Year: 2013
• ISBN:
  9781134651801
• Edition: 1
• Topic: Lingue e linguistica
• Subtopic: Linguistica

The essays in this important collection explore wide-ranging aspects of the syntax and semantics of human languages. Key topics covered include movement phenomena and the syntax of logical form, methods in generative linguistics and the role of rules vs. principles in syntactic theory. This volume makes a vital contribution to substantive and methodological debates in linguistic theory.

1

A Note on Functional Determination and Strong Crossover

In this letter the analysis of Strong Crossover (SCO) phenomena provided in Koopman and Sportiche (1983) (K and S) is examined. It will be shown that certain cases of SCO are beyond the scope of that analysis (an analysis also discussed in Chomsky (1982)). Alternatives will be presented which handle the entire range of cases.

The SCO configuration is illustrated by K and S as follows:

In Section 1, 1 will briefly summarize the K and S analysis of SCO. The ill-fonmedness of such configurations is claimed to be derivable without appealing to Principle C of the Binding Theory, the properties of such configurations following from four independently motivated principles, namely:

• (2)

• a. The K and S definition of “variable”
• b. The Functional Determination algorithm (Chomsky ( 1982))
• c. Principle A of the Binding Theory (Chomsky (1982))
• d. Principle B of the Binding Theory (Chomsky ( 1982))

In Section 2, 1 will examine a structure of Standard English exhibiting the SCO configuration ( 1 ). We shall see that the K and S analysis of SCO incorrectly fails to rule out such structures.

In light of this problem, I will then supplement the K and S system (2a-d) with the following principles:

• (2)

• e. Principle C of the Binding Theory (Chomsky (1982))
• f. Control Theory
• g. Case Theory
• h. The Theta Criterion (in particular the notion “theta-chain” (Chomsky (1981)))

• and

• i. The (or A) Resumptive Pronoun Parameter (Chomsky (1982))

We shall see that the derived system (2a-i), also fails to rule out certain SCO configurations of Standard English. I shall suggest that the inability to rule out such instances of SCO is a consequence of the incorporation of both principles (2a) and (2b).

In Section 3, 1 provide two alternative analyses of SCO in Standard English. In the first analysis, principle (2a), the K and S definition of “variable”, is abandoned. Under this analysis, Functional Determination and Binding Theory are shown to be superfluous with respect to ruling out SCO configurations. Under the second analysis, principle (2b), Functional Determination, is altogether abandoned.

The correctness of this analysis will indicate that Functional Determination plays, at most, a superfluous role in ruling out instances of SCO in Standard English.

1 The K and S Analysis of SCO

K and S propose the following (universal) definition of “variable” (a definition presumed to apply at all syntactic levels), under which variables need not be empty categories:

They note that the Bijection Principle (BP):

requires the incorporation of (3). That is to say that Weak Crossover constructions such as:

(5) [_S' who_i [_S does his_i mother love e_i]]

can be ruled out by the BP only if both the (overt) pronoun “his_i”, as well as “e_i” are defined as variables.

K and S argue that (3), the definition of “variable”, is independently motivated in that SCO configurations can be ruled out under this definition by Principles A and B of the Binding Theory, i.e. without appealing to Principle C of the Binding Theory.

Thus, for example, consider a structure exhibiting the SCO configuration:

K and S account for the ill-formedness of (6) as follows. First, they note that under definition (3), “... it is the pronoun “he” which is interpreted as a variable, and no longer the trace e, of the wh -phrase “who”.” Crucially then, the structure is not ruled out by some principle prohibiting the local

-binding of the pronominal “he”. In fact, notice that “he” is not a pronominal; it is by definition (3), a variable. Rather, such structures are ruled out by Functional Determination, as applied to ej, and Principles A and B of the Binding Theory. In particular, (6) is ruled out because:

[e_i is]... locally A-bound to “he”, ignoring traces of successive cyclic movement in the intermediate COMPs which appear to play no particular role. “He” has an independent θ-role. so e_i is an empty pronominal, i.e. a PRO.
But principles A and B of the Binding Theory... require PRO to be ungovemed and e_i... is ungovemed: hence [(6) is]... ruled out by these principles. The SCO violations are thus explained by [(3)] and principles A and B of the Binding Theory. (Koopman and Sportiche (1983, 148)

The reader will notice that the above analysis also correctly rules out SCO configurations containing empty objects such as

2 Generable SCO Configurations

Under the K and S analysis, S-structures of the following type (noted independently in Epstein (1983) and in Sportiche (1983, 35)) are incorrectly generable:

(8) [_S' Who_i [_S did he_i try [_S' e_i [_S e_i to go]]]]

In (8), as in (6), “he” is locally

-bound, hence a variable (under (2a)). The subject ej (again, ignoring traces in COMP) is locally A-bound by “he”, which has an independent theta-role. (Notice that if Subjacency is a constraint on movement (see e.g. Lasnik and Saito (1984)) the trace in COMP need not be present. Even if this trace is present it does not count as an

-binder under Functional Determination (see Chomsky (1982)). Consequently the ej subject is PRO (under (2b)). Under Principles A and B of the Binding Theory (2c and 2d) (8) is generable since e_i, (i.e. PRO), is ungovemed. Thus, under the K and S analysis (2a-2d) such SCO configurations are generable. (Sportiche (1983, 35) claims that the un-grammaticality of examples such as (8) is “... due to the accidental property of English of not allowing resumptive pronouns in subject position...” However, this account fails to specify the formal principles and/or parameters governing the distribution of resumptive pronouns (see also Sportiche (1983, 149-150))).

Notice that incorporating Principle C of the Binding Theory (2e) is without effect here. Functional Determination identifies the e_i subject in (8) as PRO. Consequently, no Binding violation results. Furthermore, notice that the Theory of Control (2f) is satisfied in (8); PRO is properly controlled. In addition, we can not rule out (8) under Case-Theory (2g). Specifically, (8) can not be ruled out under the assumption that wh-trace (or variables) require Case. Such an assumption is orthogonal here because the ej subject in (8) is, by Functional definition, PRO, not wh-trace (nor a variable).

Concerning the Case-status of the (lexical) NP “who”, in (8), notice first that this operator (a non-argument in an

-position) does not require a theta-role (see Chomsky (1981, 179-180)). Consequently, under the reduction of the Case-Filter to the Theta Criterion (see e.g. Chomsky (1981, 336)), this operator need not be Case-marked, i.e. visible for theta-role assignment. (See also McNulty (in preparation) for further discussion of these issues.)

Turning now to (2h), notice that if we were to replace Functional Determination with the following principle:

we would still be unable to rule out (8). Free assignment of features certainly allows the assignment of the features [+ anaphor, + pronominal] to the ej subject, in which case no principle of grammar is violated. In particular notice that the Theta Criterion (Chomsky ( 1981 )) apparently provides us with no means by which to rule out (8). Identifying (8) as a Theta Criterion violation would seem to require that “theta-chain” be defined in such a way that the constituents [“whoj”, “hej”, “ej”] obligatorily constitute a single theta-chain. Under such a definition, this three-membered chain in (8) would be assigned two theta-roles. The structure would then be ruled out as a violation of the Theta Criterion. However, such a definition of “theta-chain” seems untenable, since it would presumably entail that in, for example,

(10) [_S' [_S hej tried [_S' [_S e_i to go]]]]

there also exists a single theta-chain, namely [“hej”, “ej”], which is illicitly assigned two theta-roles. This, of course, is an unwanted result. The theory of theta-chains would thus seem to require the standard assumption that any occurrence of PRO heads a theta-chain. Thus, we see that the...

Table of contents

• Front Cover
• Essays in Syntactic Theory
• Routledge Leading Linguists
• Title
• Copyright
• Dedication
• Contents
• Acknowlegments
• Introduction
• 1 A Note on Functional Determination and Strong Crossover
• 2 Quantifier-pro and the LF Representation of PROarb
• 3 The Local Binding Condition and LF Chains
• 4 Adjunction and Pronominal Variable Binding
• 5 Quantification in Null Operator Constructions
• 6 Differentiation and Reduction in Syntactic Theory: A Case Study
• 7 Derivational Constraints on A-Chain Formation
• 8 Overt Scope Marking and Covert Verb-Second
• 9 “UN-Principled” Syntax and the Derivation of Syntactic Relations
• Index