Will Nediger presented on "Focus and Condition C" at the 32nd West Coast Conference on Formal Linguistics at USC on March 7, 2014.

Abstract: Binding Condition C, as originally formulated by Chomsky (1981), states that R-expressions must not be bound. Condition C rules out all of the ungrammatical sentences in (1); in each case, there is an R-expression, John, which is bound by a co-referent nominal, either a pronoun or another R-expression.

(1) a. *He likes John.
b. *John likes John.
c. *He/*John thinks that Sally knows that you like John.

However, a number of phenomena have been observed to be prima facie exceptions to Condition C. First, it has been observed that epithets can be bound in certain circumstances, as in (2), violating Condition C under the assumption that epithets are R-expressions (e.g. Lasnik 1976).

(2) John ran over a man who was trying to give the idiot directions.

Second, focused R-expressions seem to be exceptions to Condition C, both as binders and bindees, as in (3) and (4).

(3) a. Only JOHN likes John.
b. (Nobody thinks John is polite.) JOHN thinks John is polite.

(4) a. John only likes JOHN.
b. (John doesn’t like anybody.) John likes JOHN.

Third, as observed by Schlenker (2005), bound R-expressions are allowed when they serve a disambiguating function, as in (5).

(5) A linguist working on binding theory was so devoid of any moral sense that he forced a physicist working on particles to hire the linguist’s girlfriend in his lab. (Schlenker 2005)

I deal with the above by revising aspects of the binding-theoretic approach of Safir (2004). Safir adopts a competition-based approach to binding: essentially, he argues that R-expressions cannot be bound because bound R-expressions can always be replaced by a less dependent nominal without affecting the interpretation, along a formal dependency scale for reference. In (1a-b), for example, John can be replaced with himself, while in (1c), John can be replaced by him. However, Safir’s approach, as is, does not account for the data in (2)-(5). I argue that the seemingly problematic data in (2)-(5) is actually consistent with Safir’s approach, provided some revisions are made in that approach.

I argue, following Dubinsky and Hamilton (1998), that epithets are not R-expressions, but pronouns, and thus not subject to Condition C. Crucially, in cases where bound epithets are disallowed across clauses, as in (6a), the ungrammaticality is due not to Condition C, but to an antilogophoricity condition on epithets, which prohibits them from occurring in the complement of an attitude verb with a co-referent subject, hence the contrast between (6a) and (6b).

(6) a. *John told us of a man who was trying to give the idiot directions.
b. John ran over a man who was trying to give the idiot directions.

Second, the data in (3) also follows straightforwardly from another extension of Safir’s approach, if we take the interpretation to include both the ordinary semantic value and the focus semantic value, in the terms of Rooth (1992). Rooth defines the focus semantic value as the set of propositions with which the proposition containing the focus is being contrasted. The focus semantic value of (3a) is the set of all propositions of the form “x likes John.” If the bound R-expression is replaced by himself, the focus semantic value of the resulting sentence is the set of all propositions of the form “x likes himself” – a different set. Since the bound R-expression cannot be replaced without altering the focus semantic value, (3a) is grammatical, i.e. the R-expression is licensed because there is no less contentful co-referent nominal (anaphor or pronoun) that is more viable regarding its interpretation.

Schlenker develops a pragmatic approach to Condition C which can deal with the data in (3)-(5). Roughly, Schlenker posits a principle called Minimize Restrictors!, which says that a more restricted
definite description should never be used when a less restricted one can be used, unless the more restricted one serves some pragmatic purpose. He takes R-expressions to be more restricted than pronouns and anaphors. However, since (as I argue) the data are consistent with Safir’s approach, and Safir gives a highly articulated syntactic account of all three binding conditions, while Schlenker only deals with Condition C, I argue that my modified version of Safir’s approach should be adopted. Crucially, I incorporate important aspects of Schlenker’s pragmatic account into my modified approach. My approach to the disambiguation cases and some of the focus cases, such as (4) and (5), incorporates aspects of Schlenker’s general approach.

I argue that the sentences in (4) are grammatical not because replacing the bound R-expression with a pronoun or anaphor would change the interpretation, but because it would not retain the structural parallelism present in (4a-b). It is well-known that focus constructions are sensitive to parallelism (roughly, for a focus construction to be felicitous, the preceding discourse should contain a member of the focus semantic value). I argue that the sentences in (4) are felicitous because they satisfy a stricter form of
parallelism: it is possible for a member of the focus semantic value of “John likes JOHN” to contain a bound R-expression (e.g. Mary in “John likes Mary”), but it is not possible for a member of the focus semantic value of “John likes HIMSELF” to contain a bound anaphor, thus allowing only the bound R-expression in (4). This analysis is similar to Schlenker’s, in that Condition C is overruled for pragmatic reasons.

Furthermore, I argue that in disambiguation cases such as (5), Safir’s theory allows for a pragmatic override of Condition C in situations where the use of a bound R-expression serves the additional pragmatic function of disambiguation (as in Schlenker’s account). Specifically, I argue that a bound R-expression may be used if and only if disambiguation could not be achieved without violating Condition C.