The predicate-type asymmetry between Conditions A and B in (8) is motivated by data such as (9) (1993, p. 675):

(9) The queen_i invited both Max and herself_i/*her_i to our party.

The question arises why the personal pronoun is illicit here, since the conjunct position is A-exempt (cf. (3)), and the complementarity between SELF anaphors and personal pronouns normally breaks down in A-exempt positions (see [2]). R&R respond that "at the stage of translating syntactic predicates into semantic ones" (1993, p. 676) the personal pronoun comes to stand in reflexive semantic relation to its antecedent (given a distributive reading), even though the two are not syntactic coarguments (by (7a)). Example (9) thus violates Condition B understood as a constraint on semantic predicates (the semantic reflexivity goes unmarked).

As promising as R&R's account of binding is, there are at least two serious problems. First, R&R's binding conditions predict that monomorphemic (alias SE) anaphors should never be bound by a coargument (i.e., assuming the predicate is not itself intrinsically/lexically reflexive). This follows from Condition B, since the reflexive predicate in this case would not be reflexive-marked. Though this prediction appears to hold for the Scandinavian languages on which R&R focus, it is not clear that this pattern holds more broadly. The Korean, Russian, and Malayalam anaphors in (10-12), respectively, all have the independent characteristics of monomorphemic anaphors and are presumably SE in R&R's system, yet all can be bound by a coargument (examples are readily available from other languages as well) (from Cole & Sung, 1994, p. 358; Rappaport, 1986, p. 99; Mohanan, 1982, p. 169):

(10) Chelswu_i-nun Yenghi_j-ka casin_i/j-ul silheha-nun kes-ul molunta
     Chelswu-TOP Yenghi-NOM self-ACC hate-and fact-ACC not know
     `Chelswu didn't believe that Yenghi hates him/himself'

(11) Volodja ocen' ljubit sebja.
     Volodja-NOM very-much loves self-ACC
     `Volodja loves himself very much.'

(12) Moohan swayam aaraadhik'k'unnu.
     Mohan self worships
     `Mohan worships himself.'

A second problem for R&R's analysis is posed by the Icelandic polymorphemic anaphor sig sjalfur. Though both sig sjalfur and its more commonly discussed inverse counterpart sjalfur sig (see [3]) have emphatic usages not clearly subject to the binding conditions, each may be used nonemphatically in certain cases (and both exhibit local binding only). However, the distribution of their nonemphatic uses differs importantly. Unlike sjalfur sig (which in all relevant respects behaves as predicted by R&R's system), sig sjalfur may not be bound by a coargument (sig sjalfur is possible in (13) only on an emphatic reading) ([4]). This is so even though sig sjalfur and sjalfur _sig both contain the putative SELF morpheme sjalfur ([5]):

(13) Jon_i talar oft vi *sig sjalfan_i/sjalfan sig_i
     John talks often to himself

However, like sjalfur sig, sig sjalfur may be nonemphatically locally bound when it is the object of an adjunct about-PP (see below on the adjunct status of about-PPs):

(14) Maria_i talai vi Jon um sig sjalfa_i/sjalfa sig_i
     Mary talked to John about herself

Neither sig sjalfur nor sjalfur sig, however, may be (nonemphatically) locally bound when the object of some other adjunct PP:

(15) Jon_i sa snak nalaegt *ser sjalfum_i/*sjalfum ser.
    John saw snake near himself

Though I will return below to an account of these and other facts of Scandinavian binding, the point here is that sig sjalfur cannot be bound in the very case R&R's theory most clearly leads us to expect otherwise (13).

Pollard and Sag's Obliqueness-Command Account

An alternative argument-structure analysis of binding has been presented by P&S within their larger HPSG framework (1992, 1994). P&S characterize binding as structure sharing (token identity) of the values for the INDEX subfeatures PERSON, NUMBER, and GENDER between two referential nominal-objects (nom-obj). In their theory, nonpronominal (npro) objects (alias R-expressions) are broadly distinguished from pronominal (pron) objects, this latter category including both anaphors (ana) and personal pronouns (ppro). Note that in P&S's system ppro refers to what are commonly called `pronouns,' whereas pron refers to all referentially dependent NPs, both ppro and ana.

P&S's binding theory is based on the relations called obliqueness command (or o-command) and obliqueness binding (or o-binding), where relative obliqueness is indicated by left-to-right position on a common SUBCAT(EGORIZATION) list (P&S, 1994, p. 253):

(16) Let Y and Z be synsem objects ([6]) with distinct LOCAL values, Y referential. Then Y locally o-commands Z just in case Y is less oblique than Z.

(17) Y o-commands Z just in case Y locally o-commands X dominating Z.

(18) Y (locally) o-binds Z just in case Y and Z are coindexed and Y (locally) o-commands Z. If Z is not (locally) o-bound, then it is said to be (locally) o-free.

P&S's binding principles are as follows (1994:254).

(19) P&S's Binding Principles:

    Principle A: A locally o-commanded ana must be locally o-bound.

    Principle B: A ppro must be locally o-free.

    Principle C: A npro must be o-free.

One key paradigm for which these principles are meant to account is that in (20) (1992, pp. 266, 288).

(20) a. Mary_i talked to John_j about herself_i/himself_j

     b. Mary_i talked about John_j to herself_i/himself_*j

     c. Mary_i talked to himself_j about Bill_*j

In P&S's analysis both to and about are nonpredicative prepositions (Ps); that is, P (therefore PP) structure-shares its CONTENT value (including INDEX) with that of the prepositional object (P-object). Consequently, the P-objects in (20) are effectively subcategorized for by the verb talked. P&S also adopt the standard assumption that to-PPs are less oblique than about-PPs (e.g., Perlmutter & Postal, 1983). In the grammatical interpretations of (20a-b), then, the ana is bound by a less oblique argument, whereas in the ungrammatical instances the ana is illicitly bound by the object of the more oblique about-PP.

One example of A-Exemption for which P&S's analysis but not R&R's can account is shown in (21), in which the ana occurs in a clause with an expletive subject (Kuno 1987; cited by P&S, 1994, p. 258):

(21) They_i made sure that it was clear to themselves_i that this needed to be
      done.

Themselves in (21) is not locally o-commanded because it, being nonreferential, does not qualify as an o-commander, and the extraposed S[comp] clause [that this needed to be done] is more oblique than themselves due to the operation of P&S's Extraposition Lexical Rule, which "removes an S[comp] from a SUBCAT list, replacing it by NP_it, and appends the S[comp] to the end [I.e. the most oblique position] of the SUBCAT list, preserving role assignment" (1994, p. 150). Consequently, the ana in (21) will end up less oblique than the S[comp] and be exempt from Condition A.

There are problems with P&S's account, however. The first is that it recognizes one undifferentiated class ana, and thus offers no account for important asymmetries between monomorphemic and polymorphemic anaphors. For example, P&S's analysis cannot explain why monomorphemic anaphors are relatively free for logophoric binding regardless the position in which they occur; that is, they do not exhibit the same sensitivity to argumenthood as do polymorphemic anaphors in the determination of A-Exemption (R&R, 1991). This suggests that Condition A never applies to monomorphemic anaphors (just as it does not apply to monomorphemic personal pronouns; see R&R (1991, 1993) for related discussion and evidence to support this conclusion). I will return to this point below.

Another problem for P&S's account has been noted by R&R in regard to (20). R&R point out that about-PPs exhibit the characteristics of adjunct rather than argument (i.e., nonpredicative) PPs. Thus, a personal pronoun may be locally bound when it is the object of about (21) (see [2]), and a polymorphemic anaphor in this same position can be logophorically bound more readily than can the object of clear argument PPs (with in (23)) (R&R, 1993, p. 715).

(22) We talked with Lucie_i about her_i.

(23) a. *Can you talk with myself about Lucie?
      b. Can you talk with Lucie about myself?

If, however, about heads an adjunct PP, then one of the primary motivations for P&S's appeal to local o-command (viz., the paradigm in (20)) dissipates.

The most serious problem for P&S's account arises from revisions they themselves make to their analysis of so-called nonpredicative Ps (e.g., to) later in their 1994 volume (pp. 342-343). Recognizing that their original treatment of such Ps is problematic ([7]), P&S posit that a P such as to is in fact predicative, its CONTENT being not merely the CONTENT of its object but instead a parameterized state of affairs (psoa) that assigns a thematic role to the object (see P&S, 1994, p. 25 and elsewhere on psoas). I will return to the HPSG treatment of Ps below, but suffice it to say here that P&S's revised view of so-called nonpredicative Ps wreaks havoc on their appeal to local o-command (e.g., neither Mary nor John can now be said to locally o-command the ana in (20a)) and makes their binding theory entirely unworkable (a problem they recognize; see their comments on p. 343).

What is needed, then, is an approach to binding theory that can retain the advantages of other argument-structure accounts while avoiding the problems discussed above. I would like to sketch a preliminary version of such an approach within the framework of HPSG, but in a way that differs in many respects from P&S's HPSG analysis. I will then show how my approach predicts the precise domain of A-Exemption while avoiding the problems of R&R's and P&S's accounts ([8]) .

To begin, I retain P&S's concept o-command (simplified in (24)) so as to account for certain hierarchy effects (e.g., A-exemption in (21)).

(24) Let y and z be synsem objects, with distinct LOCAL values, y
     referential. Then y o-commands z iff y is less oblique than
     z or some x dominating z.

Unlike P&S, however, I claim that the central relation invoked by the binding conditions is not local o-command but rather reflexive o-command:

(25) A synsem object x reflexively o-commands a synsem object y iff

    a. x o-commands y, and
     b. x and y are lexically selected by a common head.

The lexical selection referred to in (25b) may be either of the two types of selection discussed by Chomsky (1986): c-selection or s-selection. C-selection (or categorial selection) refers to syntactic subcategorization, construed in HPSG terms as membership in a SUBCAT list. Reflexivity between elements c-selected by a common head may for mnemonic purposes be called c-reflexivity. This type of reflexivity is similar to the local o-command employed by P&S in that each involves SUBCAT relations; however, c-reflexivity entails no hierarchical relationship (I account for hierarchy effects not via reflexivity but instead via the notion of general o-command in (25a)). S-selection (or semantic selection) refers to theta-role assignment, construed in HPSG terms as membership of a thematic argument in some psoa (parameterized state of affairs). Reflexivity between elements s-selected by a common head (I.e., members of a common psoa) may be referred to as s-reflexivity. This type of reflexivity is part of what R&R have in mind in their definition of syntactic predicates.

The above definitions form the basis for the following:

(26) x o-binds y iff x and y are coindexed and x o-commands y.

(27) x reflexively o-binds y iff

     a. x o-binds y, and
     b. x and y are lexically selected by a common head.

(28) If x is not (reflexively) o-bound, then it is said to be (reflexively) o-free.

The binding conditions are presented in (29-31). Note that Conditions A and B make reference to reflexive o-binding whereas Condition C makes reference only to o-binding. I employ P&S's nomenclature for nominal-objects here (I.e., pron includes both ana and ppro).

(29) Condition A:

    A reflexively o-commanded pron Z must be reflexively o-bound if

      a. Z is of sort phrase, and
      b. Z is c- and s-selected.

(30) Condition B:

    A ppro is reflexively o-free.

(31) Condition C:

    An npro is o-free.

I will focus below primarily on Condition A and to a lesser extent Condition B. To see how (29) accounts for the precise domain of A-Exemption, consider the following paradigm in Table 1 illustrating which constructions allow a polymorphemic ana to be exempt from Condition A (a-e) and which do not (f-I) (see R&R, 1993, and P&S, 1994, for empirical data illustrating A-Exemption with all of these positions). For each position the ana may occupy, the table indicates whether the ana is c-selected and/or s-selected by some head (if so, the selecting head or heads are identified in parentheses; I.e., V(erb), N(oun), or P(rep)). Also shown for each position is whether or not the ana is c-reflexively o-commanded and/or s-reflexively o-commanded (if so, the reflexive head/predicate is identified, where (V) indicates matrix verb and (v) indicates embedded clause verb).

Notes: "c-ref/o" = c-reflexively o-commanded, "s-ref/o" = s-reflexively o-commanded, "pict-NP" = picture NP, "det" = contains a nonreferential determiner (no possessor), "conjct" = conjunct in coordinate construction, "dis" = distributive reading, "col" = collective reading, "pos" = contains an NP-internal (possessor) subject, "pred" = predicate.

Table 1 is largely self-explanatory, but a few cases need to be clarified. First, though both adjunct and argument Ps c-select and s-select their objects (in accordance with P&S's revised view of nonpredicative Ps), I propose that objects of argument Ps are s-selected by the verb as well (e.g., in constructions such as "talk to X"). This means that ana objects of argument but not adjunct Ps can satisfy Condition A via s-reflexivity (ana objects of picture-NPs lacking an NP-internal possessor parallel adjunct PPs in this respect).

Second, I assume that neither conjunct of a coordinate phrase is subcategorized (c-selected) by the other (In HPSG, all conjuncts are head-daughters), or by the matrix verb for that matter. Consequently, an ana conjunct will always be A-exempt (by (27b)), even when given a distributive reading. However, because Condition B differs from Condition A by not referring to c- or s-selection, datum (9) above follows. That is, I posit that the verb's psoa lexically specifies a thematic role for each conjunct when a distributive reading obtains (so that s-reflexivity is possible in principle, ruling out the ppro in (9)). In contrast, the entire coordinate structure (not each conjunct) is assigned a thematic value when a collective reading obtains. From this the contrast in (32) discussed by R&R (1993, p.677) falls out (where (32a) allows a collective reading but both in (32b) forces a distributive reading).

(32) a. Max_I and Lucie talked about him_I.

    b. *Both Max_I and Lucie talked about him_I.

Finally, polymorphemic ana in raising structures such as "John believed himself to be clever" and "Lucie strikes herself as clever" are subject to Condition A because c-reflexivity (but not s-reflexivity) can be established with respect to the matrix verb, while the requirements of (29b) are satisfied as well (s-selection by the embedded verb, and c-selection by both verbs; see the HPSG treatment of these structures in P&S, 1994, pp. 134-136).

Thus, the binding conditions in (29-31) account for the domain of A-Exemption without succumbing to a self-contradictory treatment of ana-objects of prepositions as do P&S (and without invoking any additional mechanisms such as R&R's Chain Condition--see R&R 1993 for details). Another major advantage of my analysis is that in it monomorphemic ana are subject to neither Condition A (because they are not phrasal) nor Condition B (because they are not ppro). On the one hand, this explains why monomorphemic ana (I.e., ana of sort word) pattern with ppro in that, unlike polymorphemic ana, monomorphemic ana allow nonlocal binding and exhibit no asymmetry for A-Exemption with respect to argument versus nonargument positions. These were facts that P&S could not readily explain because they conflated polymorphemic and monomorphemic ana into one class, making both types equally subject to Condition A. On the other hand, my analysis explains why monomorphemic ana unlike ppro are free to be reflexively o-bound (see below for discussion of the Scandinavian exceptions to this). This was a fact that R&R could not explain because they treated monomorphemic ana (alias SE anaphors) and ppro alike insofar as local coargument binding is concerned (I.e., R&R considered neither monomorphemic ana nor ppro to be reflexive markers, so neither type was predicted to be bindable by a coargument).

Another interesting consequence of my analysis is that it goes part of the way toward predicting the typological absence of pron phrases headed by ppro nouns. By P&S's Semantics Principle (1994, p. 48), the phrasal mother in a ppro-headed phrase would have CONTENT of sort ppro because its head-daughter is ppro (keep in mind that in HPSG, the noun rather than the determiner heads an NP (alias DP)). However, this would in non-A-exempt positions lead to a paradox in that Condition A would require the ppro phrase to be reflexively bound (due to its phrasal status) at the same time Condition B prohibited it from being so bound (due to the phrase's ppro status). This paradox does not arise with ana-headed phrases because only Condition A applies to the mother in those cases.

Up to this point I have carefully avoided dealing with certain facts of Scandinavian binding that motivated R&R's analysis. The two most important of these facts (not predicted by my analysis above) include (a) the prohibition against coargument binding of monomorphemic ana in Scandinavian, and (b) the effect that intrinsic/lexical reflexivity of a verb has on the binding possibilities of the verb's ana object (cf. Everaert, 1986, 1991). I suggest that these facts are best treated as reflecting a language-specific system of predicate marking in Scandinavian languages that admittedly mimics R&R's system in crucial respects and is superimposed on the universal binding theory described above.

Taking Icelandic as illustrative, the language-specific marking system can be summarized as follows. First, there are various ways in which an Icelandic predicate may be marked by an anaphor.

(33) An Icelandic predicate may be marked as:

    a. having an emphatic argument

    b. being directly reflexive

    c. being indirectly reflexive

Let me briefly digress to explain (33b) and (33c).

I build this distinction between direct and indirect reflexive marking on P&S's revised analysis of so-called nonpredicative Ps alluded to earlier (1994, pp. 342-343). Having employed fine-grained thematic roles for the psoas of Vs throughout their 1994 volume (e.g., the psoa of give contains roles like GIVER and GIFT), P&S posit that the thematic value assigned by a nonpredicative P is instead of a superordinate sort that is assigned only by P (not V) (e.g., to assigns GOAL rather than GIVEE to its object). If the psoas of V and P are of the same sort (e.g., transfer for both the V give and the P to; see P&S, 1994, p. 338 on psoa sorts), the two psoas may structure-share so that at a superordinate level the CONTENT of the entire predicate phrase (e.g., "give a book to Pat") can be considered a single supersort psoa including all the relevant thematic roles that had been specified (and underspecified) at the lexical level (e.g., GIVER, GIFT, and GIVEE).

I adopt this basic proposal by P&S, with the difference that I assume a fine-grained thematic role is lexically specified in the V's psoa (e.g., GIVEE) when an argument PP is involved, this in addition to the supersort role specified in the P's psoa (e.g., GOAL). An argument P-object is therefore directly (I.e., lexically) specified by the verb, and direct (s-)reflexivity may ensue if the P-object is ana. In contrast, a true adjunct P-object (e.g., of many locative and temporal PPs) has no role specified in V's psoa, and the psoas of V and P are of distinct sorts (therefore not structure-shared). This analysis leaves room for a middle ground of indirect reflexivity, in which case the ana P-object is not directly specified in V's psoa and yet the psoas of V and P are of the same sort (so that they structure-share and receive a single interpretation at the predicate phrase level). An about-PP in some instances is a good candidate for such a "partial" adjunct (see below). Though the universal binding theory outlined earlier is sensitive only to the core case, namely direct (s-)reflexivity, I suggest (for reasons that will become apparent momentarily) that the language-specific marking system in Icelandic is sensitive to both direct and indirect reflexivity.

With this background in place, we can now state the predicate marking requirement in Icelandic:

(34) A directly reflexive predicate must be (compatibly) marked.

Icelandic anaphors carry the following values:

(35) sjalfur sig = [refl] OR [emphatic]

    sig sjalfur = [indirectly refl] OR [emphatic]

    sig = (unmarked)

In addition, certain verbs may be specified as intrinsically (lexically) reflexive separate from the marking function of the ana (per R&R).

This marking system accounts for the Icelandic paradigm in Table 2; space prohibits me from presenting all the data supporting this paradigm (but see (13-15); also see Sigurjonsdottir and Hyams (1992) for some of the data on sig):

Notes: "dir-refl" = directly reflexive; "indir-refl" = indirectly reflexive; "intr-refl V" = intrinsically reflexive verb; "emph" = emphatic; "non" = nonemphatic.

Notice that unmarked sig is prohibited in the one case (dir-refl with standard V) where marking is required by (34) but is not independently supplied by the verb (per R&R). In contrast, emphatic sig sjalfur can occur in this position because its emphatic marking satisfies (34). As an indir-refl marker, however, nonemphatic sig sjalfur would be incompatible here (cf. (13)) (I assume that the only marking incompatible with a directly reflexive predicate is indir-refl marking). Nonemphatic sig sjalfur may not occur with an intrinsically reflexive V for the same reason. Nonemphatic sjalfur sig is avoided as well in this latter position because it would redundantly mark reflexivity (per R&R).

In the indir-refl case with an about-PP (I.e., Icelandic um), both sig sjalfur and sjalfur sig may occur, the former because it marks indir-refl and the latter because it marks reflexivity more generally (of which indir-refl is one variety). Unmarked sig may also occur here because (34) requires marking only in the core case of directly reflexive predicates. Finally, neither sig sjalfur nor sjalfur sig may occur nonemphatically in adjunct PPs since these anaphors' indir-refl and refl values, respectively, are incompatible with a nonreflexive predicate.

The marking system in (33-35), then, accounts for the Icelandic paradigm in Table 2, including the crucial facts that motivated R&R's (1991, 1993) analysis but were left unresolved by my universal binding theory. I leave it an open question how the above language-specific marking systems might be modified to account for related facts in Dutch and other Scandinavian languages.

In this paper I have argued for an argument-structure based binding theory in which the reflexivity required by Condition A and prohibited by Condition B is defined in terms of the distinction between c-selection and s-selection (terms introduced by Chomsky, 1986). I also exploit the HPSG distinction between signs of sort phrase and word, in this way accounting for asymmetries between polymorphemic and monomorphemic anaphors. I attribute certain aspects of Scandinavian binding originally motivating R&R's binding theory to a language-specific system of predicate marking rather than to the universal binding conditions.

[1]. P&S (1992, 1994) argue that A-Exemption obtains more broadly that this: An anaphor is A-exempt if it has no _less oblique_ referential coarguments. I accept P&S's assessment in my own proposal below.

[2]. One important companion observation to A-Exemption is that the complementarity between polymorphemic anaphors and personal pronouns breaks down in roughly the same exempt environments (R&R, 1993, p. 661):

(ii) a. Max saw a gun near himself/him.
     b. Lucie saw a picture of herself/her.

R&R take this as an indicator that Principle B, like A, must be sensitive to reflexivity (cf. (8)).

[3]. See, for example, Sigurjonsdottir and Hyams (1992).

[4]. I am greatly indebted to Veturlidi Oskarsson and Sigridur Sigurjonsdottir for these judgments and for stimulating discussion of Icelandic.

[5]. In R&R's system SELF morphemes combine with personal pronouns or SE morphemes to form a polymorphemic SELF construction. In this case the SELF morpheme sjalfur combines with sig, the latter being a SE morpheme in Icelandic.

[6]. The term synsem (syntax-semantics) refers to the complex of syntactic and semantic feature information borne by a sign within the HPSG system, as over against phon, which refers to a sign's phonological structure. Synsem objects may be either words or phrases.

[7]. Their original treatment left no room for assigning an interpretation to P itself (only to its object), despite the fact that so-called nonpredicative Ps are commonly role-restricted (e.g., none can take an object assigned an agentive role; 1994, p. 342).

[8]. The analysis I present here may be extended to account for structural effects on the nonlocal binding of monomorphemic ana (e.g., subject-orientation, the blocking effect) (see Hamilton, in preparation).

References

Battistella, E. (1989). Chinese reflexivization: A movement to Infl approach. Linguistics, 27, 987-1012.

Chomsky, N. (1986). Knowledge of language: Its nature, origin, and use. New York: Praeger.

Cole, P., Hermon, G., & Sung, L.-M. (1990). Principles and parameters of long-distance reflexives. Linguistic Inquiry, 21, 1-22.

Cole, P., & Sung, L.-M. (1994). Head movement and long-distance reflexives. Linguistic Inquiry, 25, 355-406.

Everaert, M. (1986). The syntax of reflexivization. Dordrecht: Foris.

Everaert, M. (1991). Contextual determination of the anaphor/pronominal distinction. In J. Koster & E. Reuland (Eds.), Long-distance anaphora (pp. 77-118). Cambridge: Cambridge University Press.

Hamilton, R. (in preparation). Logophoric and structural binding of reflexives by adult Japanese learners of English as a second language. Doctoral dissertation, University of South Carolina, Columbia.

Huang, Y. (1994). The syntax and pragmatics of anaphora: A study with special reference to Chinese. Cambridge: Cambridge University Press.

Mohanan, K. P. (1982). Grammatical relations and anaphora in Malayalam. MIT working papers in linguistics, 4, 163-190. Department of Linguistics and Philosophy, MIT, Cambridge, Mass.

Perlmutter, D., & Postal, P. (1983). Some proposed laws of basic clause structure. In D. Perlmutter (Ed.), Studies in relational grammar 1 (pp. 81-128). Chicago: University of Chicago Press.

Pica, P. (1987). On the nature of the reflexivization cycle. In Proceedings of NELS 17 (Vol. 2) (pp. 483-499). Amherst: GLSA, University of Massachussetts.

Pollard, C., & Sag, I. A. (1992). Anaphors in English and the scope of binding theory. Linguistic Inquiry, 23. 261-303.

Pollard, C., & Sag, I. A. (1994). Head-driven phrase structure grammar. Chicago: University of Chicago Press.

Progovac, L. (1992). Relativized SUBJECT: Long-distance reflexives without movement. Linguistic Inquiry, 23. 671-680.

Progovac, L. (1993). Long-distance reflexives: Movement-to-Infl vs. relativized subject. Linguistic Inquiry, 24, 755-772.

Rappaport, G. (1986). On anaphor binding in Russian. Natural Language and Linguistic Theory, 4, 97-120.

Reinhart, T., & Reuland, E. (1991). Anaphors and logophors: An argument structure perspective. In J. Koster & E. Reuland (Eds.) Long-distance anaphora (pp. 283-321). Cambridge: Cambridge University Press..

Reinhart, T., & Reuland, E. (1993). Reflexivity. Linguistic Inquiry, 24, 657-720.

Reuland, E., & Koster, J. (1991). Long-distance anaphora: An overview. In J. Koster & E. Reuland (Eds.), Long-distance anaphora (pp. 1-25). Cambridge: Cambridge University Press.

Reuland, E., & Reinhart, T. (1995). Pronouns, anaphors, and case. In H. Haider, S. Olsen, & S. Vikner (Eds.), Studies in comparative Germanic syntax (pp. 241-268). Dordrecht: Kluwer.

Sigurjónsdóttir, S., & Hyams, N. (1992). Reflexivization and logophoricity: Evidence from the acquisition of Icelandic. Language Acquisition, 2, 359-413.

Tang, C.-C. J. (1994). A note on relativized SUBJECT for reflexives in Chinese. In B. Lust, G. Hermon, & J. Kornfilt (Eds.), Syntactic theory and first language acquisition: Crosslinguistic perspectives. Volume 2. Binding, dependencies and learnability (pp. 71-82). Hillsdale, NJ: Erlbaum.

If you wish to make a public comment about this paper; please send your message to: linconf@tamvm1.tamu.edu

Discussion of this paper is available at the Discussion Web Site at: