I will argue that it is not possible to state Condition B as a purely formal constraint on syntactic structures. I present two arguments to this effect. The first exploits the fact that Condition B blocks “intersecting” reference as well as coreference. The second stems from a problem noted by Higginbotham (1983) and Partee and Bach (1984): there are many formally-distinct patterns of binding dependencies which encode the same interpretation. As Heim (1998) notes, there is independent evidence that quite a number of these distinct pattens must in general be permitted. The more distinct patterns can be encoded, the greater the number of formally distinct structures that Condition B must rule out. This gives rise to what I term the "grain problem": how can we ensure that the syntax is fine-grained enough to encode all of the attested patterns of binding relations while while maintaining Condition B as a simple and purely formal constraint? Fox’s (2000) Rule H is the most promising attempt to solve the grain problem. However, I will argue that Rule H and subsequent proposals, though they have led to many important insights, have not succeeded in this respect. This has a significant architectural implication: the "interpretativist" position is correct in a strong sense. Syntactic structures are not directly interpreted but are first mapped to some level of semantic representation. It is within this mapping procedure that Condition B and other such constraints apply. This suggests a return to earlier formulations of Condition B as direct constraints on possible interpretations (e.g. Chomsky’s (1973,241) “rule of interpretation”).
I end by addressing a residual empirical issue: just which patterns of binding dependencies are permitted, and how should we rule out those which are not attested? I propose a new constraint, Preserve Possible Disjointness, which accounts for Dahl’s paradigm (Dahl 1973, 1974), various related ellipsis phenomena, and Strong Crossover.
Chomsky, Noam. 1973. Conditions on transformations. In Fetschrift for Morris Halle, 232–286. New York: Holt, Reinhart and Winston. Dahl, Östen. 1973. On so-called sloppy identity. Synthese 26:81–112. Fox, Danny. 2000. Economy and semantic interpretation. Cambridge, Massachusetts: MIT Press. Higginbotham, James. 1983. Logical Form, binding and nominals. Linguistic Inquiry 14:395–420. Partee, Barbara, and Emmon Bach. 1984. Quantification, pronouns, and VP anaphora. In Truth, interpretation and information, ed. J. Groenendijk, T. Janssen, and M. Stokhof, 99–130. Dordrecht: Foris.