Page 231 - Encyclopedia of Philosophy of Language

To the extent that there appears to be a left-right asymmetry, it is because English is predominantly right branching, according to this view.
Third, Principles A and B are not limited to third person pronouns. Facts like the following are sub- sumed under the principles, even though first and second person pronouns are not really anaphoric, at least as anaphora are characterized here.
(a) *I, think that nobody listens to myself,. (9) (b) *You, amuse you,.
Fourth, BTsays nothing about the fact that ana- phoric elements and their antecedents usually must agree in person, number, and gender. A reasonable first approximation is to say that coindexed NPs must agree in these features. There are, however, cases like (10) where pronouns do not agree with their apparent antecedents, suggesting that such agreement is actu- ally a function of the semantics.
I bought a Veg-o-matic, after I saw them, (10) advertized on TV.
1.2 The Binding Theory and Empty Categories
Government and Binding theory (GB)posits several kinds of empty categories—that is, elements that are not pronounced, but play a role in the syntax and/or semantics of sentences. The application of BTto these elements plays a central role in accounting for a var- iety of syntactic facts within this theory (though not in others). In this section, a fewof the most important uses of BTin connection with empty categories in GB will be summarized.
The trace left behind by NP movement (in con- structions like the passive) is treated as an anaphor, and hence is subject to Principle A. This accounts for distinctions like the following (where t represents a trace):
In such sentences, the relative positions of the pro- noun and the wh-phrase provide no basis for pre- dicting whether an anaphoric relationship is possible, but the relative positions of the pronoun and the trace permit them to be differentiated. More generally, treat- ing wh-traces as R-expressions predicts that exam- ples involving wh-movement will behave with respect to anaphora as if the wh-phrases had not been moved. Sentences like the following, exemplifying what is known as 'weak crossover,' provide some support for this prediction: in general, the pattern of gram- maticality between a pronoun and an NP that the pronoun does not c-command is preserved if the NP is moved away, leaving a trace. The use o f ' %' indi- cates that only some speakers accept the examples. The difference in acceptability in the examples in (14) raises many issues of analysis, which will not be detailed here. For most speakers, anaphora to a kind- denoting term like bulldogs is easier than to the quanti- ficational certain breeds: nonrestrictive relative clauses typically allow the anaphora, restrictive relative clauses are harder, and interrogatives the hardest (see Wasow 1979).
(a) Only people who own them, could love (14) bulldogs,.
(b) He breeds bulldogs, which, only people who own them, could love t,.
(c) %People who own them, love certain breeds,.
(d) % Which breeds, do people who own them, love t,?
A third type of empty category in GB is PRO, which appears in the subject position of most infinitives. PRO is analyzed as a pronominal anaphor—that is, as an element subject to both Principle A and Principle B of BT.This entails, paradoxically, it would seem, that PRO must be both bound and free in its gov- erning category. The contradiction is only apparent, however, because not every element has a governing category. In particular, the subject position of some nonfinite clauses has no governing category. Evidently, the distribution of PRO is crucially depen- dent on the definition of governing category, and hence linked to claims about the binding of pronouns. This connection and its derivation from BT are referred to as 'the PRO theorem.'
Notice that the PRO theorem depends crucially on the domain of Principles A and B being the same, for this was the basis of the apparent contradiction. But the identity of the two domains has another conse- quence: it implies that anaphors and pronominals will be in complementary distribution wherever there is a governing category. For overt anaphors and pro- nominals, this consequence turns out to be a good first approximation, but not a reliable generalization. For example:
(a) Pat, was told t, that I saw Chris. (b) *Chris, was told Pat that I saw t,. (c) Pat,isbelievedt,tobedangerous. (d) *Pat,isbelievedt,isdangerous.
(11)
This pattern parallels the distribution of overt anaphors, as Principle A predicts:
(a) They, told each other, that I saw Chris. (12) (b) They, told Pat that I saw each other,.
(c) John, believes himself, to be dangerous.
(d) *John,believeshimself,isdangerous.
The traces of wh-movement are treated as R- expressions, and hence subject to Principle C. This accounts for a phenomenon known as 'strong cros- sover,' exemplified in example (13):
(a) Which linguist, did you say t, thought I had (13) insulted him,?
(b) *Which linguist, did you say he, thought I had insulted t,?
Anaphora
209