Third person effects on binding *.

Moskovsky, Christo

Abstract

The article provides evidence from a number of languages that sentences which are structurally identical, and differ only in morphological number, display different binding options." 1st and 2nd person sentences allow instances of pronominal binding (e.g. "I am not thinking of me") while structurally identical sentences in the 3rd person do not (e.g. "*[He.sub.i] is not thinking of [him.sub.i]"). Such evidence presents a problem for existing versions of the binding theory. It is argued that the data under discussion can be accounted for in terms of a discourse "avoid ambiguity" factor operating on 3rd person, but not on 1st and 2nd person, pronouns. The article contends that the binding options of 3rd person pronouns are determined by both structural (syntactic) and nonstructural (discourse) factors, while the binding options of 1st and 2nd person pronouns are determined by structural factors alone, and in this sense only the latter represent a pure case of syntactic binding. It therefore follows that attempts at formulating structural (syntactic) constraints on binding should avoid 3rd person pronouns as the picture there is additionally complicated by the operation of a discourse factor (avoid ambiguity) and should deal with binding of 1st and 2nd person pronouns, as only they reflect structural conditions on binding. In reality, however, studies investigating anaphoric binding deal almost exclusively with instances of binding in 3rd person sentences, which may have negatively affected current formulations of locality constraints on binding.

**********

Nowadays it is more or less unanimously accepted that anaphora involves "the interaction of all of the standard components of a theory of language: pragmatics, semantics, syntax, and phonology" (Wasow 1986: 108). The classic binding theory (BT) comprises three binding conditions as in (1):

(1) BINDING THEORY

(A) An anaphor (= reflexive or reciprocal) is bound in a local domain;

(B) A pronominal is free in a local domain;

(C) An r-expression (1) is free (in the domain of the head of its chain) (Chomsky 1986b: 166)

and is designed to account for sentence-level referential relations as in (2):

(2) a. [John.sub.i] despises himselfi/*[him.sub.i].

b. [John.sub.i] thinks that Mary despises [him.sub.i]/*[himself.sub.i].

c. *[She.sub.i] thinks that [Mary.sub.i] despises him. (2)

It should be noted that the BT relies entirely on structural concepts which are rooted in the X-bar theory (see Chomsky 1981, 1986a; Chomsky and Lasnik 1992). The key configurational condition for binding is "c-command," and locality constraints on binding are defined in terms of "governing category" (Chomsky 1981: 211; 1986b: 169).

While the BT in (1) neatly accounts for the referential relations in sentences like (2), there are numerous instances in which it fails to do so:

(3) The picture of [himself.sub.i] in Newsweek shattered the peace of mind that [John.sub.i] had spent the last six months trying to restore.

Such use of the reflexive is usually associated with a particular point of view or perspective, and is analyzed under the term "logophoricity" (Sells 1987; Reinhart and Reuland 1991). Referring to sentences like (3), Pollard and Sag (1992: 278) point out that it "is difficult to imagine any principle involving a configurationally determined notion of binding domain, however formulated, that would account for such facts." Such examples abound in Zribi-Hertz (1989) and reflect the fact there can be a variety of factors determining the referential options for a pronoun in a particular sentence, and that sometimes nonstructural factors can interfere with, and even override, syntactic conditions on binding. The pertinent question in relation to this, then, is whether it is reasonable to extend a syntactic binding theory to capture the full range of anaphoric relations, including those falling outside of sentence-level grammar, and the obvious conclusion is that a syntactic theory of binding should confine itself to the structural aspects of anaphora, and that logophorocity (among analogous phenomena) "requires a separate treatment" (Reinhart and Reuland 1991:317). Another obvious conclusion is that studies investigating syntactic (as opposed to pragmatic) aspects of binding need to identify and deal with empirical data that represent structural conditions on binding alone and are not contaminated by the operation of nonsyntactic (e.g. phonological, semantic, pragmatic) factors. Clearly, analyzing sentences like (3) cannot be expected to yield any ground-breaking insights into the nature of the structural conditions regulating binding. The problem is that while (3) displays unequivocal logophoric effects, the operation of nonsyntactic factors may not always be as obvious as in (3). In what follows, I will consider a discourse factor, operating to eliminate ambiguity in 3rd person sentences, which appears to have largely been ignored in the literature dealing with binding.

Early research on reflexive and pronominal binding considered the possibility that binding is subject to pragmatic, rather than syntactic, factors. Dowty (1980) proposes a "neo-Gricean conversational" principle as in (4), arguing that constraints on noncoindexing of pronouns (in other words, condition (B) of the binding theory) are not part of grammar, but are determined by discourse factors alone: (3)

(4) NEO-GRICEAN CONVERSATIONAL PRINCIPLE

If a language has two (equally simple) types of syntactic structures A and B, such that A is ambiguous between meanings X and Y while B has only meaning X, speakers of the language should reserve structure A for communicating meaning Y (since B would have been available for communicating X unambiguously and would have been chosen if X is what was intended) (Dowty 1980: 32).

Applying this principle (which, following Reinhart [1983: 166], I will informally call "avoid ambiguity") to (5) and (6) will yield the correct results:

(5) * [She.sub.i] saw her/.

(6) [She.sub.i] saw herself/. (Dowty 1980: 32)

In the terms of the conversational principle formulated in (4), the reflexive herself is the unambiguous syntactic structure B: herself can refer only to the subject she and, therefore, is always chosen when such is the intended reference. On the other hand, the pronominal her can have a number of referents including she; her is, then, the ambiguous syntactic structure A which, in line with (4), is reserved for the cases when the intended reference is outside of the sentence.

Taking Dowty's argument a little further, it will be logical to assume that, if (4) is really a discourse principle (rather than a structurally determined principle of grammar) operating for the elimination of ambiguity, then such a principle will be irrelevant in those cases in which there is no ambiguity. In other words, if the two syntactic structures A and B are equally unambiguous, we can expect that no "avoid ambiguity" principle will come into play, that is, (4) will not be operable. It can be argued that, while 3rd person pronouns can be a source for ambiguity, this should not be so with 1st and 2nd person pronouns. Third person pronouns are ambiguous in the sense that they can (potentially) have more than one referent: in the sentence John thinks that Mary hates him the pronoun him can refer to John, but it can also refer to a number of other referents. Conversely, in a sentence like I think that Mary hates me, the pronoun me can only refer to I and to nothing else. In this sense, 3rd person pronouns are inherently ambiguous in a way that 1st and 2nd person pronouns are not. If such reasoning is correct, sentences like (5) and (6) in the 1st and 2nd person would not be a source of ambiguity, therefore the avoid ambiguity principle (4) will be irrelevant. As a result, the speaker should be able to choose freely between the available structures A and B, that is, we can expect that a pronominal can appear in the place of the reflexive with grammatical results. It must be admitted that there is not much evidence in English to support such a hypothesis. Dowty himself does not discount the possibility that his neo-Gricean coversational principle may have been grammaticized even if it originated in pragmatics. In fact, subsequent research (see, e.g., Chomsky 1981, 1986b; Chomsky and Lasnik 1992; Lasnik 1989; Reinhart 1983; among many others) has demonstrated that binding relations in sentences like (5) and (6) are regulated by structural conditions. This being the case, then the distinction between 1st and 2nd person on the one hand, and 3rd person on the other, should not affect the binding options, and sentences like (7) will be ruled out as well:

(7) * I saw me.

Nevertheless, the operation of the avoid ambiguity principle in the 3rd person cannot completely be discounted. As Dowty (1980: 39, fn. 4) points out, "it is easier to violate non-coindexing restrictions in English with first person than with third person pronouns," and in this sense (7) is not as bad as (5). The classic BT does not account for such differences which, it is assumed here, reflect the fact that sentences like (5) involve two violations: a violation of the BT and a violation of the avoid ambiguity principle, while sentences like (7) involve only a BT violation.

Another piece of evidence for the operation of a discourse avoid ambiguity principle in the 3rd person comes from the analysis of sentences like (8) and (9), which illustrate how a phonological factor, such as contrastive stress, can sometimes override structural conditions on binding and allow for a pronominal to be bound in an obvious violation of condition B of the binding theory:

(8) Personally, I suspect me. (4)

(9) You are not afraid of him--you are afraid of you. (5)

These sentences were tested for acceptability with a number of speakers. Most of them found (8) and (9) somewhat awkward out of context, but nonetheless acceptable. All informants, however, judged structurally equivalent sentences in the 3rd person completely unacceptable irrespective of the contrastive stress:

(10) He is not afraid of you--* [he.sub.i] is afraid of [him.sub.i].

(10) is structurally identical to (9), so the contrast in acceptability cannot be accounted for in terms of syntactic structure. One possible way to account for the contrast is to assume that avoid ambiguity, which only operates in the 3rd person, cannot be overridden by phonological factors.

Also, it has long been recognized that English reflexives in the 1st and 2nd, but not in the 3rd person sometimes can appear without a structural antecedent: (6)

(11) Physicists like myself were never too happy about the parity principle.

(12) * Physicists like himself don't often make mistakes. (Ross 1970: 229-230)

One possible explanation for the contrast in acceptability between (11) and (12) is that with 1st and 2nd person reflexives no ambiguity can arise, which makes it possible to use them without a structural antecedent.

Such evidence from English in support of the existence of a discourse avoid ambiguity principle operable only in the 3rd person may seem inconclusive. However, there is compelling evidence from other languages. According to Toman (1991), Czech reflexive possessives can freely be interchanged with nonreflexive possessives in the 1st and the 2nd person:

(13) [Vy.sub.i] jste otravil [vasi.sub.i][svou.sub.i] kocku. you have poisoned your/self's cat 'You have poisoned your cat.' (Toman 1991: 153)

but not in the 3rd person where only the reflexive is acceptable:

(14) Karel/otrfivil *[jeho.sub.i/] [svou.sub.i] kocku. Karl poisoned his self's cat 'Karl poisoned his (own) cat.' (Toman 1991: 153)

Toman offers no account for the contrast between (13) and (14), but it is obvious that the difference between the two sentences is not structural. According to Reinders-Machowska (1991), the situation with Polish possessives is practically the same.

Bulgarian possessives behave in the same manner. In the 3rd person, the use of a nonreflexive possessive in (15) is unacceptable:

(15) [Ivan.sub.i] e otrovil * [negovata.sub.i]/[svojata.sub.i] kotka. Ivan is poisoned his self's cat 'Ivan has poisoned his (own) cat.'

In contrast, structurally identical sentences in the 1st and 2nd person allow a seemingly free interchange between reflexive and nonreflexive possessives:

(16) Az sam otrovil mojata/svojata kotka. I am poisoned my self's cat 'I have poisoned my (own) cat.'

The same holds for nonclitic, nonpossessive reflexives as well. Once again we observe the same pattern: in the 1st and 2nd person the reflexive can be replaced with a pronominal with a grammatical result, which is not possible in the 3rd person:

(17) Az ne mislja za mene/sebe si. I not think about me self 'I am not thinking of me/myself.'

(18) [Ivan.sub.i] ne misli za * negoi/sebe [si.sub.i]. Ivan not think about him self 'Ivan is not thinking of himself.'

Sentences (17) and (18) are identical in structure and the only difference between them is person. Therefore, it is not reasonable to assume that the restrictions on the use of the pronoun in (18) are structural in nature. If they were structural, then they would be expected to affect all persons, not only the 3rd person. An important consequence of the contrast between (17) and (18) is that different locality constraints apparently operate in the two sentences. The position taken here is that in (18) the pronoun nego 'him' is barred as a result of the operation of an avoid ambiguity principle of the type formulated in (4), rather than as a condition B violation: if it were a condition B violation, the pronoun mene 'me' in (17) would also be barred. It appears then that sentences like (18) (i.e. 3rd person sentences) do not constitute appropriate data for the analysis of locality constraints on binding, because they involve additional nonsyntactic effects. Sentences like (17), on the other hand, do not involve such effects and, therefore, the range of the relevant binding domain should be calculated based only on the analysis of sentences like (17), and not those like (18).

While (as suggested earlier) the operation of "avoid ambiguity" in English may not be as obvious as in other languages, (7) there is evidence that it does operate. The contrast between (19) and (20) is a particularly clear example of its operation:

(19) I am not thinking of me.

(20) * [John.sub.i] is not thinking of [him.sub.i].

Obviously, the relevant binding domain for the pronominal me in (19) is not its governing category or its complete functional complex (which in this case are equal to the whole clause), but something smaller, and if this is the case with me in (19) it should also be the case with him in (20). The fact that the proposed reading of (20) is not acceptable should be attributed to the operation of the avoid ambiguity principle.

The 1st and 2nd person data presented here show a close-to-complete breakdown in the complementarity relation between reflexives and pronominals (encoded in conditions A and B of the BT), and in light of that, one might be tempted to conclude that 1st and 2nd person data are not representative of core binding relations, and that 3rd person data alone embody genuine reflexivity. (8) Such a position would be reinforced by the fact that some languages (e.g., German and French) have no 1st and 2nd person reflexives, and a distinction between reflexive and nonreflexive pronouns only holds in the 3rd person. However, in languages which have distinct 1st and 2nd person reflexive forms, alongside the numerous instances of what appears to be free variation between reflexives and pronominals (as, e.g., in [13], [16] and [17] above), there are also examples of sentences manifesting complete compliance with the binding conditions, irrespective of morphological number. Bulgarian verbal clitics are a good case in point:

(21) a. [Ivan.sub.i] [se.sub.i] / * [go.sub.i] pogledna (v ogledalo-to). Ivan self-CL him-CL looked in mirror-the 'Ivan looked at himself/him (in the mirror).'

b. Az go poglednah (v ogledalo-to). I him-CL looked in mirror-the 'I looked at him (in the mirror).'

c. Az se / * me poglednah v ogledalo-to. I self-CL me-CL looked in mirror-the 'I looked at myself/me in the mirror.'

d. Ivan me pogledna (v ogledalo-to). Ivan me-CL looked in mirror-the 'Ivan looked at me (in the mirror)."

In fact, more unambiguous examples of the operation of the BT, with 100% complementarity irrespective of morphological number, may be hard to come by. In English, the situation with direct object pronouns is very much the same:

(22) a. She saw herself/*her (in the mirror).

b. I saw myself/*me (in the mirror).

The position taken here is that, as a purely syntactic (= structural) component, the BT cannot be sensitive to a distinction in morphological number. Therefore, sentences like (21) and (22) are taken to represent core binding relations, with the view that instances of breakdown in complementarity (e.g. [13], [16] and [17]) can be accounted for in terms of locality constraints and LF movement of the reflexive. This point is taken up again towards the end of this article.

Another question that needs to be addressed is whether with 3rd person pronouns, which have been shown to be subject to the operation of a discourse avoid ambiguity principle of the type formulated in (4), the operation of the binding conditions is, in fact, taken over by avoid ambiguity. The view taken here is that such an assumption is not warranted: while all examples presented so far involve sentences in which the effects of avoid ambiguity and the binding conditions are mixed, there are cases in which avoid ambiguity operates in the absence of binding, as in (23), and also cases in which binding operates in the absence of avoid ambiguity, as in (24):

(23) * The news about John?? astounded him??.

(24) * With no living creature around, [John.sub.i] pampered [him.sub.i] for hours.

In (23) John does not c-command him: the pronoun him is not bound by the NP John, and the unacceptability of (23) in the proposed reading cannot be attributed to the BT. Conversely, in sentences like (24) condition B operates in the absence of avoid ambiguity: the context eliminates any possible ambiguity, therefore, Dowty's discourse principle will be irrelevant; (24) is ruled out as a condition B violation. This is supported by the fact that an equivalent 1st person sentence (which would generally be insensitive to avoid ambiguity) would be also be ungrammatical:

(25) * With no living creature around, I pampered me for hours.

It is not among the goals of this article to provide insights into the psychological mechanisms underlying avoid ambiguity. It is important to note, though, that as a discourse principle, avoid ambiguity should not be expected to operate absolutely (the same way a syntactic principle would be expected to operate), but only so far as there is a perceived ambiguity on the part of the speaker/listener. It therefore seems logical to assume that the perceived level of ambiguity with regard to specific sentences will differ among individual speakers (and over time, possibly, even with the same speaker). Indeed, acceptability judgement data indicate that while speakers unanimously and consistently reject BT violations such as (26):

(26) [Mary.sub.i] found [her.sub.i] in the library. they are rather ambivalent with regard to sentences involving avoid ambiguity, such as (23) above and (27):

(27) a. The man behind [him.sub.i] attacked [John.sub.i].

b. The news about [John.sub.i] reached [him.sub.i] in the early hours of the morning.

c. The article about [him.sub.i] in The Observer convinced [John.sub.i] that the media attacks against him were very carefully planned and organized.

d. The continuous rumors about [him.sub.i] were a huge source of anxiety for [John.sub.i].(10))

There is a great deal of variation in acceptability judgements on most of these sentences ranging from "completely acceptable," through "partially acceptable," to "completely unacceptable," which is nothing less than what we can expect with the operation of a nonsyntactic (discourse) principle.

Since this article does not attempt to propose a pragmatic theory of referential dependencies, it will have little else to say about avoid ambiguity: its existence has been discussed with the sole purpose of drawing attention to the fact that 3rd person sentences must be excluded from the analysis of the operation of the binding conditions, because quite often, referential dependencies in such sentences involve--in addition to structural conditions on binding--nonsyntactic binding effects which may be hard, or even impossible, to tease apart from the strictly structural conditions on binding. However, it will not be an overstatement to say that most, if not all, studies of binding have analyzed instances of binding in the 3rd person, which may have brought about the wrong results. Chomsky (1986b: 167) discusses sentences like (28) in some detail:

(28) a. * They/told [NP stories about [them.sub.i]].

b. [They.sub.i] heard [NP stories about [them.sub.i]]. suggesting that the contrast between the two sentences can be accounted for in terms of an optional implicit argument with the properties of PRO in the SPEC position of the object NP. It is not among the goals of this article to consider the merits of his analysis. The point that is relevant to the current discussion is that in the 1st person, this contrast disappears:

(29) a. [We.sub.i] told [NP stories about [us.sub.i]].

b. [We.sub.i] heard [NP stories about [us.sub.i]].

which indicates that the factor responsible for the contrast in (28) must be nonstructural, otherwise we would expect the same contrast to occur in (29) as well. Reinhart and Reuland (1993: 661) present (30):

(30) [Lucie.sub.i]'s joke about [herself.sub.i]/*[her.sub.i].

to illustrate cases in which complementarity between reflexives and pronominals holds. However, in structurally identical 1 st or 2nd person sentences, this complementarity disappears:

(31) a. My joke about myself/me.

b. Your joke about yourself/you.

Likewise, Burzio (1991) uses a Russian example (from Timberlake 1979):

(32) [On.sub.i] uze rasskazal mne o [svoei.sub.i]/*[ego.sub.i] zizni.

He already told me about self's his life

'He has already told me about his own life.'

to support his claim that complementarity breakdown is not "of sufficient crosslinguistic generality" (Burzio 1991: 98). However, in identical 1st or 2nd person sentences, both pronoun and reflexive would be grammatical, and complementarity disappears:

(33) [Ja.sub.i] uze rasskazal emu o [svoei.sub.i]/[moei.sub.i] zizni.

I already told him about self's my life

'I have already told him about my life.'

What all these examples have in common is that they all manifest avoid ambiguity effects which disappear in corresponding 1st or 2nd person sentences. To the extent that such examples involve nonsyntactic binding effects in addition to structural conditions on binding, it is not impossible that theoretical assumptions about binding made based on the analysis of such data are flawed. This seems to be particularly valid for formulations of locality constraints on pronominal binding. Sentences such as (19) and (20), repeated here as (34) and (35):

(34) I am not thinking of me.

(35) * [John.sub.i] is not thinking of [him.sub.i].

suggest that the ungrammaticality of (35) cannot be attributed to a locality constraint violation, but rather to the operation of the discourse avoid ambiguity factor. If this is true, then (35) does not represent the actual locality constraint operating on pronominal binding. It would appear, therefore, that attempts to formulate locality constraints on binding should target data which are insensitive to avoid ambiguity, such as (7), repeated here as (36):

(36) * I saw me.

and should propose a locality constraint reflecting the fact that clause-internal binding of me is bad in (36), but is alright in (34).

It could be argued that such an approach would present problems for condition A of the binding theory, because then in sentences like (17), (19), (31), and (33), which apparently allow free variation between reflexives and pronominals, it will be the reflexive that will be bound in a locality constraint violation. This need not necessarily be the case if LF movement of the reflexive (or part of it) is assumed. In long-distance anaphora, LF movement of reflexives is now a standard assumption (see, e.g., Cole and Sung 1994, and the references cited there), but even with regard to local reflexives, such as the English himself, there have been proposals(11) that the reflexive (or the self part) undergoes movement at LF and ends up in a position which is much "closer" to its antecedent where condition A would be respected.

In conclusion, the article has provided and discussed crosslinguistic evidence indicating that the distribution of 3rd person pronominals is not only subject to the operation of condition B of the binding theory, but also subject to the operation of a discourse avoid ambiguity factor. While avoid ambiguity can operate in the absence of binding and vice versa, in the 3rd person the two quite often coincide. For that reason, the operation of avoid ambiguity is not obvious, and it only becomes obvious when 3rd person sentences involving a bound pronominal are compared with structurally identical 1st or 2nd person sentences. The implications for the BT are clear: it seems that the only secure way to eliminate the effects of avoid ambiguity would be to deal with 1 st and 2nd person sentences only.

Received 8 January 2002 Revised version received 4 March 2003

University of Newcastle Callaghan, Australia

Appendix. Sentences used in the acceptability judgement task.

The new book about [her.sub.i] will earn [Mary.sub.i] a lot of fame. This piece of news about [him.sub.i] astounded [John.sub.i].

The article about [him.sub.i] in The Observer convinced [John.sub.i] that the media attacks against him were very carefully planned and organized.

The huge success of the book about [her.sub.i] was not unexpected for [Mary.sub.i]

The news about [John.sub.i] reached [him.sub.i] in the early hours of the morning.

Mary's unconcealed desire to divorce [John.sub.i] terribly upset [him.sub.i]. The investigator's report about [Mary.sub.i] gave John a reason to leave [her.sub.i].

The news about [me.sub.i] reached [me.sub.i] in the early hours of the morning.

The new book about [Mary.sub.i] will earn [her.sub.i] a lot of fame.

The man in the bed above [John.sub.i] asked [him.sub.i] a question.

The man behind [him.sub.i] attacked [John.sub.i].

[Mary.sub.i] found [her.sub.i] in the library.

The continuous rumors about [him.sub.i] were a huge source of anxiety for [John.sub.i].

This piece of news about [John.sub.i] astounded [him.sub.i].

The huge success of the book about [Mary.sub.i] was not unexpected for [her.sub.i].

Most men at [her.sub.i] work find [Mary.sub.i] sexually attractive.

The article about [John.sub.i] in The Observer convinced [him.sub.i] that the media attacks against him were very carefully planned and organized.

The news about [him.sub.i] reached [John.sub.i] in the early hours of the morning.

The announcement about [Mary.sub.i]'s engagement to John came as a shock to [her.sub.i].

This piece of news about [me.sub.i] astounded [me.sub.i].

The new book about [you.sub.i] will earn [you.sub.i] a lot of fame.

The voicemail message for [Mary.sub.i] removed [her.sub.i] from the list of suspects.

The women in [his.sub.i] life passionately love [John.sub.i].

The stack of Playboy magazines under [John.sub.i]'s bed belongs to [him.sub.i].

The management's agreement to offer [Mary.sub.i] a position at their Vienna branch found [her.sub.i] unprepared.

The announcement about [her.sub.i] engagement to John came as a shock to [Mary.sub.i].

The huge success of the book about [me.sub.i] was not unexpected for [me.sub.i].

Mary's unconcealed desire to divorce [him.sub.i] terribly upset [John.sub.i].

Most men at [Mary.sub.i]'s work find [her.sub.i] sexually attractive.

The investigator's report about [Mary.sub.i] gave [her.sub.i] a reason to leave John.

The man behind [John.sub.i] attacked [him.sub.i].

The continuous rumors about [John.sub.i] were a huge source of anxiety for [him.sub.i].

The voicemail message for [her.sub.i] removed [Mary.sub.i] from the list of suspects.

The women in [John.sub.i]'s life passionately love [him.sub.i].

The man in the bed above [him.sub.i] asked [John.sub.i] a question.

[John.sub.i] suspects that Mary doesn't like [him.sub.i].

The stack of Playboy magazines under [his.sub.i] bed belongs to [John.sub.i].

Notes

* Peter Peterson was very helpful at an early stage of the preparation of this article, and Alan Libert provided a number of useful comments and suggestions on the final version: I am grateful to them both. I must also acknowledge the constructive comments provided by an anonymous reviewer. Correspondence address: School of Language and Media, University of Newcastle, Callaghan, NSW 2308, Australia. E-mail: Christo.Moskovsky@newcastle.edu.au.

1. Informally, the term "r-expression" refers to lexical names; see Chomsky (1981: 102), and Chomsky (1986b: 79).

2. Condition (C) in (1) will not be considered here, and it may well be the case, as proposed by some authors (Reinhart 1983, 1986; Koster 1987; Grodzinsky and Reinhart 1993), that this condition is not a part of grammar, though the issue will not be addressed in this article.

3. For a much more refined proposal, see Levinson (1991).

4. From The Return of Perry Mason, 28 January 1994, Prime Television.

5. From The Presidio, 24 February 1994, Prime Television.

6. However, they will certainly have some kind of a discourse antecedent.

7. One possible reason is the fact that there are no reflexive possessives in English.

8. A point raised by an anonymous reviewer.

9. Based on distance between binder and bindee (rather than governing category); for a detailed proposal along these lines, see Moskovsky (2002).

10. See Appendix for the complete list of sentences tested for acceptability.

11. Chomsky (1986b: 175) suggests that the English reflexive undergoes LF movement analogous to clitic climbing in the Romance languages: an idea which, for reasons unclear to me, has largely been ignored.

References

Burzio, Luigi (1991). The morphological basis of anaphora. Journal of Linguistics 27, 81107.

Chomsky, Noam (1981). Lectures on Government and Binding. Dordrecht: Foris.

--(1986a). Barriers. Cambridge, MA.: MIT Press.

--(1986b). Knowledge of Language. Its Nature, Origin and Use. New York: Praeger.

--; and Lasnik, Harold (1992). The principles and parameters theory. In Syntax: ein internationales Handbuch zeitgendssischer Forschung, J. Jacobs, A. yon Stechow, W. Sternefeld, and T. Vennemann (eds.), 506-569. Berlin: Walter de Gruyter.

Cole, Peter; and Sung, Li-May (1994). Head movement and long-distance reflexives. Linguistic Inquiry 25, 355-406.

Dowty, David R. (1980). Comments on the paper by Bach and Partee. In Papers from the Parasession on Pronouns and Anaphora, K. J. Kreiman and A. E. Oteda (eds.), 29-40. Chicago: Chicago Linguistic Society.

Grozdinsky, Yosef; and Reinhart, Tania (1993). The innateness of binding and coreference. Linguistic Inquiry 24, 69-101.

Koster, Jan (1987). Domains and Dynasties: The Radical Autonomy of Syntax. Studies in Generative Grammar 30. Dordrecht: Foris.

Lasnik, H. (ed.) (1989). Essays on Anaphora. Dordrecht: Kluwer.

Levinson, Stephen C. (1991). Pragmatic reduction of the binding conditions revisited. Journal of Linguistics 27, 107-161.

Moskovsky, Christo (2002). Aspects of Binding in Bulgarian. European University Studies. Series XXI: Linguistics, vol. 242. Frankfurt am Main: Peter Lang.

Pollard, Carl; and Sag, Ivan (1992). Anaphors in English and the scope of the binding theory. Linguistic Inquiry 23, 261-303.

Reinders-Machowska, Ewa (1991). Binding in Polish. In Long-Distance Anaphora, Jan Koster and Eric Reuland (eds.), 137-150. Cambridge: Cambridge University Press.

Reinhart, Tania (1983). Anaphora and Semantic Interpretation. London: Croom Helm.

--(1986). Centre and periphery in the grammar of anaphora. In Studies in the Acquisition of Anaphora, B. Lust (ed.), 123-150. Dordrecht: Reidel.

--; and Reuland, Eric J. (1991). Anaphors and logophors: an argument structure perspective. In Long-Distance Anaphora, Jan Koster and Eric Reuland (eds.), 283-321. Cambridge: Cambridge University Press.

--; and Reuland, Eric J. (1993). Reflexivity. Linguistic Inquiry 24, 657-720.

Ross, J. R. (1970). On declarative sentences. In Readings in English Transformational Grammar, R. Jacobs and P. Rosenbaum (eds.), 222-272. Waltham, MA: Ginn.

Sells, Peter (1987). Aspects of logophoricity. Linguistic Inquiry 18,445-481. Timberlake, A. (1979). Reflexivization and the cycle in Russian. Linguistic Inquiry 10, 109141.

Toman, J. (1991). Anaphors in binary trees: an analysis of Czech reflexives. In Long Distance

Anaphora, Jan Koster and Eric Reuland (eds.), 151-170. Cambridge: Cambridge University Press.

Wasow, Thomas (1986). Reflections on anaphora. In Studies in the Acquisition of Anaphora, Barbara Lust (ed.), 107-122. Dordrecht: Reidel.

Zribi-Herz, Anne (1989). Anaphor binding and narrative point of view: English reflexive pronouns in sentence and discourse. Language 65, 695-727.