Nanzan University August 1-4, 2007Some Milestones in the Development of Binding TheoryHoward LasnikUniversity of [email protected] course is dedicated to the memory of Tanya Reinhart-1-The original theories of anaphoric relations were transformational. Anaphoric pronounswere derived from full NPs; reflexives were derived from pronouns, as were missing subjects ofcertain infinitivals (EQUI). (1) "... there are actually good reasons for introducing [pronouns] transformationally assubstituends for NP's of the proper kind... [a] transformation ... replaces an animate NPby "he."" Chomsky (1955, p.502)(2) "In §109.6 we found that such sentences as "John wanted him tocome" (= 311) are introduced by transformation. If we investigatethese sentences in more detail, we discover that there are certain restrictionson the occurrence of pronouns. Alongside of 311 we have 325 butnot 326:325 (a) I wanted him to try(b) I wanted you to try(c) I wanted to try326 I wanted me to tryThe only way to avoid a special restricting statement on the level Pis to add a mapping MxP that carries 326 into 325c:327 MxP carries I-want-I-to{try into I-want-to{try" LSLT p.519(3) *You wanted you to tryThird person is slightly trickier:(4) 328 (a) he wanted him to try(b) he wanted to tryThe simplest way to handle this situation appears to be to set uptwo distinct elements he and he* corresponding to the element hew ofW, he being an element just like I and you (which accounts for 328b),and he* being an ordinary proper noun (which accounts for 328a, justas we have "he wanted John to try"). The establishment of this pair ofhomonyms on the level P is further supported by its usefulness for otherpurposes, as we will see below. We thus replace 327 by the more generalcharacterization 329MxP carries -want- -to{try into -want - to{try Thus, bound anaphoric and freely referring ones are different lexical items. For instance,(5) is ambiguous, rather than vague.-2-}{{}{}(5) John thinks he is clever(6) The setting up of two elements he and he* alsohas support in the referential use of these words. Whereas I and youhave an unambiguous reference in sentences like "John said that Iwould come," "John said that you would come," in "John said that hewould come" the reference is ambiguous. In our terms, if "he" in thissentence is derived from the syntactic element he (the pronoun), thereference is to John, as it is in 330b, where he becomes Ø; if it is derivedfrom the syntactic element he* (the proper noun), the reference is toa second person, as in 330a. While the results of our grammaticalanalysis naturally tell us nothing about reference, this syntactic discussiondoes provide the means for an adequate description of reference in somesemantic description of English. p.525(7) We also have330 (a) John wanted him to try(b) John wanted to try(c) John wanted me to tryThe simplest way to account for this is to revise MxP, replacing329 by I I 331 MxP carries you -want- you -to{try into NPy he I you -want - to{try (NPy … I, you) NPyMxP applied to "John wanted him* to try" gives 330a; i.e., it doesnot apply. Applied to "John wanted him to try," MxP gives 330b. pp.519-520What about reflexives?(8) Tself as we have stated it is only a special case of a quite generaltransformation. Note that "him," "you," etc. in 351 are each the objectof the complex verb "persuade-to try" (an instance of Verb-Complement).But it is true in general that the object of a verb undergoes the transformation Tself, even in kernel sentences. Thus we have355 he saw himself, etc.The transformation Tself must be reformulated, then, to hold moregenerally of NP-VT-NP, and it must be given as a mapping, since it isobligatory even for kernel sentences. We replace 352 by 356, definingthe component M0 P of MP-3-{({})}{}356MxP is determined by (Qself , *self), where I I Qself = you VPA{ VT, you (NP … I, you) NP he*self : (U, U, U, self) p.530Locality of the 'clause-mate' type was also first hinted at in LSLT:(9) The investigation of "self" and its distribution will, I think, turn out tobe of some importance for the study of transformational structure. Theoccurrence in a sentence of X-self, for some pronoun X, indicates aspecial relation between this element and some noun or some otherpronoun in the sentence. To give a general rule concerning all suchcases directly for all sentences may be quite difficult. But we may beable to show that whenever this relation exists between two positionsin a complex sentence Z, the elements filling these places have somefixed relation in the kernel structures from which Z is derived (e.g.,they may be subject and object). If so, the distribution of self may bestatable simply in terms of this kernel sentence relation. The simplificationthus introduced can be an important support for transformationalanalysis, in particular cases. p.534Chomsky (1965) has interesting discussion of reflexivization, referencing Lees and Klima(1963). In this discussion, referential indices are introduced as part of the theory, to furtherexplicate the notion of 'identity' relevant to deletion operations (of which pronominalization andreflexivization are two):(10) "As an additional illustration, consider the reflexivizationoperation (see Lees and Klima, 1963, for a detailed discussion). Ithas frequently been observed that in a sentence such as "Johnhurt John" or "the boy hurt the boy," the two phoneticallyidentical Noun Phrases are necessarily interpreted as differing inreference; sameness of reference requires reflexivization of thesecond Noun Phrase (this is also true of pronominalization).Various attempts have been made to build an account of this intothe syntactic component, but none has been very convincing.The availability of lexical features suggests a new approach thatmight be explored. Suppose that certain lexical items aredesignated as "referential" and that by a general convention,each occurrence of a referential item is assigned a marker, say, aninteger, as a feature. The reflexivization rule can be formulatedas an erasure operation that uses one Noun Phrase to delete-4-another. As in the case of relativization ... the erasureleaves a residue, in particular,