Against Optional wh-Movement - CiteSeerX

8 downloads 0 Views 298KB Size Report
L. Mikkelsen and C. Potts, pp. 71-84. Somerville, MA: Cascadilla Press. .... I want Bill to paint [a picture of myself]. (25) a. Johni wants mej to paint [a ...
Deriving Reflexives Rose-Marie Déchaine and Martina Wiltschko University of British Columbia

1. The proposal: a typology of reflexives The morphology of reflexives correlates with their binding properties (Everaert 1986, Pica 1987, Reinhart & Reuland 1993) and reflexives are often distinguished according to whether they are X0 or XP anaphors (SE vs. SELF anaphors respectively). It has independently been argued that the categorical status of pronouns correlates with their binding properties (Wiltschko 1999). In particular, Déchaine and Wiltschko 2002 propose that there are (at least) three categorical types of proforms: pro-DP (Rexpression), pro-ϕP (variable), pro-NP (constant) with the structures in (1): (1)

a.

DP

ϕ

2

ϕ

c.

2

ϕP

D

ϕP

b.

2

NP | N

NP |

NP | N

N

We extend the analysis in (1) to reflexives, and argue that there are (at least) three types of reflexives: DP-, ϕP-, and NP-reflexives. We further show that the category of a reflexive determines the binding relation it enters into. DP-reflexives, as R-expressions, enter into a binding relation via assigned co-reference. φP-reflexives, as variables, enter into a binding relation via operator binding. NP-reflexives, as nominal constants, enter into a binding relation via theta-binding. A general consequence of reducing the binding properties of reflexives to their categorical identity is that there are no dedicated reflexives. It therefore follows that the binding effects associated with binding principle A (i.e. anaphors must be locally bound) ∗ Thanks to S.Burton, R.Sharp and audiences at UBC and WCCFL 21. This research was supported by the Austrian Academy of Science (APART 435) and SSHRC MCRI grant 412-97-0016.

© 2002 Rose-Marie Déchaine and Martina Wiltschko. WCCFL 21 Proceedings, ed. L. Mikkelsen and C. Potts, pp. 71-84. Somerville, MA: Cascadilla Press.

72

WCCFL 21

are epiphenomenal in that they derive from the interaction of independently motivated factors. The paper is organized as follows: §2 discusses DP-reflexives; §3 discusses NP-reflexives; §4 discusses φP-reflexives; §5 concludes. 2. DP-reflexives: English 1st/2nd person X-self We analyze English 1st and 2nd person reflexive pronouns (myself, yourself, ourselves, yourselves) as instances of DP-reflexives. We illustrate how they enter into binding relations (2.1) and then show how principle A effects arise (2.2). Finally, we show that there is no one-to-one correspondence between reflexive forms and reflexive construals (2.3), which is precisely the state of affairs predicted by the present analysis. 2.1. Binding of DP-reflexives We claim that DP-reflexives, as R-expressions, enter into a binding relation via assigned co-reference.1 By assigned coreference we mean that an identity relation holds between the arguments of a transitive relation R: (2) R [x,y] x=y. Applying this to English reflexives such as (3)a yields (3)b. (3) a. b.

I saw myself. see(x,y) where x = y

A question which arises under this approach is why assigned coreference is necessary with reflexives but impossible with personal pronouns as shown in (4). (4) a. b.

*I saw me. *see(x,y) where x = y

We take the complementarity of reflexives and personal pronouns to be an instance of blocking (Burzio 1989, Déchaine & Manfredi 1994, Williams 1997), as defined in (5). (5) The Blocking Principle: Select the most specified form. (x is more specified than y iff x has more features than y). 1. This is similar to Lidz’ (1995, 1996) notion of “near reflexives” and ter Meulen’s (2000) concept of “dependent coreference”.

Déchaine and Wiltschko

73

In a blocking-theoretic account local binding of me is impossible because a more specified form (myself) is available for assigned coreference. Although both reflexives and personal pronouns are DPs, the former have more structure and so count as more specified: (6)

a. DP 2 D ϕP my 2 ϕ NP | N self

b.

DP | D | me

Note that in (6)a the possessive pronoun is analyzed as an instance of D (Abney 1987). This captures the fact that English 1st and 2nd person reflexives have the same form as possessed nouns as illustrated in (7).2 (7) 1st/2nd person reflexive= pronominal possessor 1sg myself my book 2sg yourself your book 1pl ourselves our books 2pl yourselves your books We have shown how assigned coreference accounts for the binding of DP-reflexives, how blocking accounts for the complementarity between reflexives and pronouns, and how the DP-status of reflexives accounts for their possessor syntax, which in turn accounts for why they can be locally bound. The fact that they sometimes must be locally bound (i.e. seem to obey principle A) follows from the syntax of inalienable possession (Helke 1970, Pica 1987, Reinhart & Reuland 1993). The obligatory local binding of –self parallels the local binding of inalienably possessed nouns: (8) a. b.

I saw myself. *I said that Lucy saw myself.

(9) a. b.

I raised my hand. *I said that John raised my hand.

2. English 3rd person reflexives are analyzed as φP-reflexives (Déchaine & Wiltschko 2002), as reflected in the imperfect match between reflexives and possessive DPs: himself, *hisself, his book; themselves, *theirselves, their books.

74 (10) a. b.

WCCFL 21 I lost my way. *I said that John lost my way. (cf. Helke 1970)

2.2. DP-reflexives do not occur in all and only reflexive contexts The form of DP-reflexives as well as their binding properties follow from their DP-status combined with the syntax of inalienable possession. We now turn to the distribution of DP-reflexives, showing that they do not occur in all and only reflexive contexts. First, observe that there are reflexive contexts which do not require the presence of a DP-reflexive. For example, English inherently reflexive verbs can be, but need not be, marked with a DP-reflexive: (11) a. b.

I washed. I washed myself.

This provides evidence for the claim that reflexivity is independent of the appearance of an overt reflexive. Second, it is also the case that DPreflexives do not occur only in reflexive contexts. For example, they may be used as logophoric pronouns, in which case they are locally free: (12) a. b.

I believe that [Mary loves Paul more than myself]. I believe that [Mary loves Paul more than me]. (adapted from Zribi-Hertz 1995: 335, (6a/8a))

Under our analysis reflexivity is a by-product of assigned coreference. If there is no accessible local antecedent, assigned coreference fails, in which case DP-reflexives have the status of logophors (Kuno 1972, 1987, Zribi-Hertz 1989, Safir 2001). The logophoric use of DP-reflexives follows from blocking: in logophoric contexts, the more highly specified form (reflexive) is selected over the less specified form (personal pronoun). An advantage of this analysis is that it correctly predicts that assigned coreference is not restricted to co-arguments (pace Reinhart & Reuland 1993). This accounts for why DP-reflexives occur as nominal predicates: (13) a. b.

You are [not yourself] today. John considers you [not yourself] today.

In our proposal (13) is an instance of identificational predication. The claim that DP-reflexives may function as nominal predicates predicts that they may occur as adjunct nominal predicates. This prediction is borne out by the existence of so called emphatic DP-reflexives (“intensives”, C.L. Baker 1995, Zribi-Hertz 1995) which have the same distribution as adjunct depictive predicates as shown in (14) and (15):

Déchaine and Wiltschko (14) a. b.

I [myself] saw Mary. I saw Mary [myself].

(15) a. b.

John, tired, wrote the letter. John wrote the letter tired.

75

Emphatic reflexives, like depictive predicates, may be either subject- or object-oriented. However, it is not possible to have more than one occurrence of an emphatic reflexive, as illustrated in (16). (16) a. Fred [himself] is not usually as alert as Karen. b. Fred is not usually as alert as Karen [herself]. c. *Fred [himself] is not usually as alert as Karen [herself]. (C.L. Baker 1995, (46), cited in Zribi-Hertz 1995: 337, (12)) This uniqueness requirement on emphatic reflexives follows from their status as identificational predicates, which is independently claimed to impose a uniqueness restriction (Kiss 1998). We have shown that a DP-analysis of English self-reflexives accounts for the following properties. First, if there is a local antecedent available, then binding arises via assigned coreference. Second, the obligatoriness of binding follows from the syntax of inalienable possession. Third, if there is no local antecedent, then DP-reflexives may function as logophors. Finally, DP-reflexives are predictably found in contexts of nominal predication; this accounts for their use as emphatic reflexives. 3. NP-reflexives We now examine the properties of NP-reflexives, arguing that their binding properties, their syntactic distribution, and their locality conditions follow from their NP-status. As NPs, NP-reflexives are nominal constants. Whereas DP-reflexives enter binding via the identity relation (assigned coreference), this mechanism is not available for NP-reflexives by virtue of the fact that they are non-referential. For an NP-reflexive, the only way it can enter into coreference is via the predicate which introduces it. This is precisely the mechanism which Williams (1994) refers to as theta-binding.3 With a DPreflexive a transitive relation R is not, strictly speaking, reflexive. Rather, coreference is established by the identity relation as in (17)a. With NP reflexives however, the transitive relation R is truly reflexive as in (17)b: 3. While Williams (1994) takes theta-binding to be a general mechanism for all binding relations, we claim that theta-binding is restricted to NP-reflexives.

76

WCCFL 21

(17) a. b.

R [x,y] x=y R [x,x]

DP-reflexive NP-reflexive

We have discussed what makes (17)a unavailable for NP-reflexives, namely their inability to support reference (Déchaine & Wiltschko 2002). As for the unavailability of (17)b for DP-reflexives, this reflects general restrictions on the type of nominal argument that can remain VP-internal. In particular, we assume that only bare NPs must stay within the VP (Borer 1994). This accords with the fact that only bare nominals incorporate (Baker 1988). Applying this analysis to NP-reflexives we expect NPreflexives to have the distribution of bare NPs, which in turn predicts that they should have the same distribution as incorporated nouns. This prediction is borne out. For example, Plains Cree has productive noun incorporation (Wolfart 1973, Hirose 2000) as shown in (18). (18)

nôt-âposw-ê=w hunt-rabbit-intrans-3sg ‘s/he hunts rabbits’

The same position occupied by Plains Cree incorporated nouns may be filled by reflexive -iso, which occurs in verbal and nominal contexts: (19) a.

b.

nipah-isô-w kill-refl-3sg ‘he kills himself’ (cf. Wolfart 1973) nipah-isô-win kill-REFL-nom ‘self-killing/suicide’

English is often described as lacking noun-incorporation. Correlated with this is the absence of productive reflexive verb formation: (20) a. b.

*John rabbit-discovers. *John self-discovers.

We suggest that whatever is responsible for the lack of nounincorporation in English is also responsible for the lack of NP-reflexives in verbal contexts. However, it is not the case that there are never NPreflexives in English. They are possible in nominal contexts: (21)

self-discovery

Déchaine and Wiltschko

77

Although reflexive verb-formation is not productive in English, it seems to be possible whenever it involves back-formation from the corresponding reflexive nominalization. This is illustrated in (22): (22) a. b.

This tape will self-destruct in 5 minutes. (Mission Impossible) This kit will allow you to self-rescue. (R. Sharp, p.c.)

In these examples, the reflexive verb self-destruct is backformed from the nominalization self-destruction; likewise, the reflexive verb self-rescue is back-formed from the corresponding nominalization self-rescue. Another property of NP-reflexives is that they are strictly locally bound. This follows from the fact that NP-reflexives are theta-bound and so are restricted to co-argument binding, as illustrated in (23) for Plains Cree. (23)

John nitaweyihtam Mary ê-nipah-isô-t. John know Mary C-kill-self-3sg ‘Johni knows that Maryj selfi*i/i-killed’

In this proposal, Reinhart & Reuland’s (1993) notion of reflexivity reduces to pro-NP reflexives. This predicts that if a language has both proDP and pro-NP reflexives, only the former will permit non-local binding. This is borne out in English picture noun phrases: the DP-reflexive (myself) can be either locally or nonlocally bound (24), but the corresponding NPreflexive (incorporated self) can only be locally bound (25). (24) a. b.

John wants me to paint [a picture of myself]. I want Bill to paint [a picture of myself].

(25) a.

Johni wants mej to paint [a self*i//j-portrait]

b.

Johni said that this tapej will self*i/j-destruct in 5 minutes.

Our analysis of NP-reflexives claims that, as NPs, they are nominal constants and have the same distribution as incorporated nouns. In addition, NP-reflexives can only enter into a binding relation via theta-binding, which is restricted to co-argument binding. 4. φP-reflexives As φPs, φP-reflexives function as variables and coreference is a byproduct of operator binding.4 Assuming the principle of full interpretation 4.

See also Zribi-Hertz’s 1995 notion of “unspecified bindable expression”.

78

WCCFL 21

then, as a variable, a ϕP minimally requires the presence of an operator to be bound, as in (26). (26)

OPx […x…]

Recall that the category ϕP generalizes across nominal types, and so includes personal pronouns (Déchaine & Wiltschko 2002). This predicts that ϕP “reflexives” and “pronouns” can be non-distinct, in which case we expect an undifferentiated bound variable pronominal that can be both locally and non-locally bound. This prediction is borne out in Haitian, where the same form (li) is used for both local and non-local binding: (27)

Jean wè li Jean see 3sg ‘Jean sees him/her/himself’ (Déchaine & Manfredi 1994)

Indirect evidence that this pronominal is licensed by operator binding comes from the fact that Haitian li is used as a resumptive pronoun, which is standardly taken to involve bound variable anaphora: (28)

Chyeni [m te kase pat lii a] te mòde m. Dog 1sg Tns broke leg 3sg Def Tns bite 1sg ‘The dog whose leg I broke bit me’ (Koopman 1982:175, (14a))

This shows that a general purpose ϕP is attested, as predicted by our analysis. We also expect that some languages will have more specialized φPs. We here limit our attention to those φPs that have a reflexive function: long-distance φP-reflexives (4.1) and locally bound φP-reflexives (4.2). 4.1. Non-local binding of ϕP-reflexives: long-distance reflexives As variables, ϕP-reflexives can, but need not be, locally bound. This accounts for the distinction between local and long-distance reflexives. If there are two reflexive forms, then there is a contrast between simplex and complex reflexives. In Norwegian, the complex reflexive seg selv must be locally bound while the simplex reflexive seg may be non-locally bound (Pica 1984, Reinhart & Reuland 1993): (29) a.

b.

Jon foraktet seg selv Jon despised SEG self ‘Jon despised himself’ *Jon foraktet seg Jon despised SEG

Déchaine and Wiltschko (30) a. b.

oss forakte *Jon bad Jon asked us despise

79

seg selv SEG self

Jon bad oss forakte seg Jon asked us despise SEG ‘Jon asked us to despise him’

(Hellan 1988:104,109-13)

Under our proposal, the domain distinction between seg selv and seg follows from blocking. To see this consider the formulas in (31): (31) a. b.

λx [x despise x] λx [x ask [us despise x]

seg selv seg

The more specified form (seg selv) is only used for local binding whereas the less specified form (seg) is used elsewhere. If a more articulated structure correlates with a more specified form, it follows that seg selv is selected for local binding and seg is the elsewhere case: (32) a.

φP

b.

2

ϕ seg

NP |

φ | seg

N selv Thus, when a language has two reflexive forms they stand in a blocking relation to each other. In such a system, the simplex φP-reflexive is predictably the long-distance reflexive. 4.2. Local binding of ϕP-reflexives: Romance s-pronouns We now turn to systems where simplex φP-reflexives are locally bound, as is the case for Romance s-pronouns. We first show how the φP analysis accounts for the multi-functionality of s-pronouns and then extend the analysis of 3rd person s-pronouns to 1st and 2nd person. Across Romance, s-pronouns are used to mark local reflexive relations, as shown for French (33)a and Spanish (33)b. (33) a.

Jean se voit. J SE see.3sg ‘Jean sees himself’

80

WCCFL 21 b.

Juan se vio J SE see.3sg ‘Juan saw himself’

Romance s-pronouns are not restricted to marking the reflexive relation. French s-pronouns are also found in reciprocal contexts, middles and inchoatives, but fail to occur as impersonal subjects (Wehrli 1986): (34) a.

Jean et Marie se voient. J. and M. SE see.3pl ‘Jean and Marie see themselves’/‘Jean and Marie see each other’

b.

Ces livres se vendent bien. These books SE sell well ‘These books sell well’

c.

La porte s’ est ouverte. the door SE is open ‘The door opened’

d.

*Il se sait qu’ ils ont menti. It SE know that they have lied

Spanish s-pronouns are found in reciprocal contexts, middles and inchoatives as well as impersonal subjects (Sharp in prep.). (35) a.

Juan y Maria se vieron. J and M SE see.3pl ‘Juan and Maria saw themselves’/ ‘Juan and Maria saw each other’

b.

Estros libros se venden bien. These books SE sell well ‘These books sell well.’

c.

Se abrió la puerta. SE opened the door ‘The door opened’

d.

Se sabe que mentieron. SE know that lied.3pl ‘It is known that they lied’

Déchaine and Wiltschko

81

We propose that the distinction between French versus Spanish lies in the nature of the operator which binds the variable introduced by the spronoun. Either all the members of R are locally operator-bound (French), or at least one member of R is locally operator-bound (Spanish). Consequently, French s-pronouns are compatible with (36)a-c: here all arguments of the relation R are operator bound, the limiting case being (36)c, which corresponds to valency reduction.5 For Spanish, the requirement is weaker, in that at least one argument must be operatorbound: this permits the possibility of impersonal subjects as in (36)d.6 (36) a. b. c. d.

λx R[x,x] λx λy R[x,y] & R[y,x] x ≠ y λx R[x,x] & ∀y R[x,y] x = y ∃x R[x,y]

‘reflexive’ ‘reciprocal’ ‘object-oriented’ ‘impersonal subject’

To our knowledge, this account of Romance s-pronouns is the only one that derives their multi-functionality.7 Another advantage of the φP analysis is that it generalizes to Romance personal pronouns in the 1st and 2nd person (Déchaine & Wiltschko 2002). These are ambiguous between a pronominal, a reflexive and a reciprocal use as shown in (37). (37) a.

b.

Jean nous a vu. J. 1pl has seen ‘Jean saw us.’ Nous nous sommes vu 1pl 1pl be seen ‘We saw ourselves.’/’We saw each other.’

While 1st and 2nd pronouns have a pronominal use, this is not the case for s-pronouns (Kayne 2000). Instead an l-pronoun occurs in these environments, as shown in (38). (38) a.

Jean les a vu. J. 3pl has seen ‘Jean saw them.’

5. See ter Meulen 2000 for a similar view of inchoatives and middles. 6. The same holds of Italian (Manzini 1986). 7. S-pronouns only mark reflexivity for direct objects; a phrasal form is used for PP-arguments (Zribi-Hertz 1980, 1990, 1995). The fact that reflexivity is not marked in a uniform way is consistent with our claim that there are no dedicated reflexives.

82

WCCFL 21 b.

Ils se sont vu. 3pl SE are seen ‘They saw themselves.’/’The saw each other.’

As before, whenever there are two competing forms, the selection of one over the other is determined by the blocking principle. This implies that s-pronouns are more highly specified than l-pronouns: as such, s-pronouns block l-pronouns. This analysis predicts that within Romance there might exist a language in which an l-pronoun can be used as a reflexive.8 5. Conclusion We have proposed that there are (at least) three categorically distinct types of reflexives: DP-reflexives, φP-reflexives, and NP-reflexives. The category of a reflexive in turn determines the nature of the binding relation it enters into: assigned co-reference, operator binding or theta-binding: (39) a. b. c.

R [x,y] x=y OPx […x…] R [x,x]

DP-reflexive φP-reflexive NP-reflexive

As a general consequence, there are no dedicated reflexive proforms (Lidz 1996, 1996; Rooryck & van Wyngaerd 1997, 1998; ter Meulen 2000). This accounts for the morphosyntactic heterogeneity of reflexives, and for the fact that a reflexive need not mark all and only local reflexive relations. Our analysis also predicts that whenever a language has more than one reflexive, then blocking determines which one is selected. The table in (40) summarizes the results of this proposal: (40) Reflexive typology D = R-expression R[x,y] x=y local possessor syntax binding non-local binding predication

reflexive logophor emphatic reflexive

ϕ = variable OPx [x…(x)…] reflexive; reciprocal; valency reduction long distance reflexive emphatic reflexive

N = constant R[x,x] co-argument binding, N-syntax ---

8. Greek confirms the prediction that personal pronouns can be used as reflexives. See Anagnostopoulou & Everaert (1999) for relevant discussion.

Déchaine and Wiltschko

83

What emerges is that only NP-reflexives are subject to a strict locality condition and they only occur in argument-positions. In contrast, DP- and φP-reflexives may be locally bound, but need not be. With DP-reflexives, local binding correlates with possessor syntax, while non-local binding yields reflexive logophors. As for φP-reflexives, local binding correlates with multi-functionality (e.g. the same form is used for reflexives, reciprocals and valency reduction), while non-local binding yields longdistance reflexives. Finally, both DP- and φP-reflexives are not restricted to argument positions: as nominal predicates they function as emphatic reflexives. References Anagnostopoulou, Elena & Martin Everaert. 1999. Toward a more complex typology of anaphoric expressions. Linguistic Inquiry 30, 97-119. Baker, Mark. 1988. Incorporation. University of Chicago Press. Baker, C. Lee. 1995. Contrast, discourse prominence, and intensification, with special reference to locally-free reflexives in British English. Language 71, 63101. Borer, Hagit. 1994. On the projection of arguments. In: E.Benedicto and J.Runner (eds.) Functional projections. University of Massachusetts occasional papers in linguistics 17, 19-48. Burzio, Luigi. 1989. On the nonexistence of disjoint reference principles. Rivista di Grammatica Generativa 14, 3-27. Déchaine, Rose-Marie; Victor Manfredi. 1994. Binding Domains in Haitian. Natural Language and Linguistic Theory 12, 203-57. Déchaine, Rose-Marie, Martina Wiltschko. 2002. Decomposing pronouns. Linguistic Inquiry 33. Everaert, Martin. 1986. The Syntax of Reflexivization, Foris, Dordrecht. Helke, Michael. 1970/1979. The Grammar of English Reflexives. PhD dissertation, MIT, published by Garland, New York. Hellan, Lars. 1988. Anaphora in Norwegian and the Theory of Grammar. Dordrecht, Foris. Hirose, Tomio. 2000. Origins of predicates: evidence from Plains Cree. PhD dissertation, UBC Kayne, Richard. 2000. Parameters and Universals. Oxford University Press. Kiss, Katalin E. 1998. Identificational focus versus information focus. Language 74, 245-73. Koopman, Hilda. 1982. Les constructions relatives. In: C.Lefebvre, H.MagloireHolly and N.Piou (eds.) Syntaxe de l’haïtien. Ann Arbor, Michigan, Karoma, 167-203. Kuno, Susumo. 1972. Pronominalization, reflexivization, and direct discourse. Linguistic Inquiry 2, 161-195. Kuno, Susumo. 1987. Functional Syntax. University of Chicago Press. Lidz, Jeffrey. 1995. Morphological reflexive marking: evidence from Kannada. Linguistic Inquiry 26, 705-710.

84

WCCFL 21

Lidz, Jeffrey. 1996. Dimensions of reflexivity. PhD dissertation, UDelaware. ter Meulen, Alice. 2000. On the economy of interpretation: semantic constraints on SE-reflexives in Dutch. In: H.Bennis, M.Everaert and E.Reuland (eds.) Interface Strategies. Amsterdam, KNAW, 239-255. Manzini, Maria-Rita. 1986. On Italian si. In: H.Borer (ed.) The Syntax of Pronominal Clitics. New York, Academic Press, 241-262. Pica, Pierre. 1984. On the distinction between argumental and non-argumental anaphors. In: W.de Geest and Y.Putseys (eds.) Sentential Complementation. Dordrecht, Foris, 185-193. Pica, Pierre. 1987. On the nature of the reflexivization cycle. Proceedings of NELS 17, 483-499. Reinhart, Tanya and Eric Reuland. 1991. Anaphors and logophors: an argument structure perspective. In: J.Koster and E.Reuland (eds.) Long-distance Anaphora. Oxford University Press, 283-321. Reinhart, Tanya and Eric Reuland. 1993. Reflexivity. Linguistic Inquiry 24, 657720. Rooryck, Johann and Guido van den Wyngaerd. 1997. The self as other: a minimalist approach to zich and zichzelf in Dutch. Proceedings of NELS 28, 359-373. Rooryck, Johann and Guido van den Wyngaerd. 1998. Puzzles of identity: binding at the interface. Proceedings of NELS 29, 333-374. Safir, Ken. 2001. The syntax of anaphora. Ms., Rutgers University. Sharp, Randy. (in preparation). Reflexivity in multilingual machine translation. PhD dissertation, University of British Columbia. Wehrli, Eric. 1986. On some properties of French clitic se. In: H.Borer (ed.) Syntax & Semantics 19: The Syntax of Pronominal Clitics. Academic Press, 263-283. Williams, Edwin. 1994. Theta Theory in Syntax. Cambridge, Mass.: MIT Press. Williams, Edwin. 1997. Blocking and anaphora. Linguistic Inquiry 28, 577-628. Wiltschko, Martina. 2000. The categorical determination of pronominal binding properties. Proceedings of NELS 30 Wolfart, H. Christoph. 1973. Plains Cree: a Grammatical Study. Transactions of the American Philosophical Society, N.S. 63.5. American Philosophical Society, Philadelphia. Zribi-Hertz, Anne. 1980. Coréférences et pronoms réfléchis: notes sur le contraste lui/lui-même en français. Linguisticae Investigationes 4, 131-179. Zribi-Hertz, Anne. 1990. Lui-même argument et le concept de “pronom A”. Langages 97, 100-127. Zribi-Hertz, Anne. 1995. Emphatic or reflexive? On the endophoric character of French lui-même and similar complex pronouns. Journal of Linguistics 31, 333-374.