Alice Davison

University of Iowa

Long-distance anaphors in Hindi/Urdu: Issues

In this paper I give an overview of the conditions on local andlong-distance binding of reflexives and reciprocals, including the domains in which binding may take place, and other coindexing relations such as verbal agreement. I consider these data in relation to the Raising to INFL expressed in various forms in Pica (1984), Huang and Tang (1989) (Cole et al.(1990), Cole and Sung (1994). I argue that it accounts well for the c-command condition, the subject condition, the expanded domains (with coindexation not restricted to co-arguments) and the contrast between simple and complex anaphors. I propose that INFL is actually TENSE/ASPECT, which has a strong D feature represnting the Extensed Projection Principle requirement for a subject, rather than AGR-S, which checks Nominative Case and morphological agreement on the verbal complex in this language.

Lexical choices for referentially dependent elements in Hindi/Urdu have sharply distinct properties:
(i) pronouns, which are marked for person and number, and refer to discourse antecedents as well as syntactic antecedents, and

(ii) anaphors, which are invariant for person, number and agreement features, and may not have 3p discourse antecedents.

The anaphors include morphologically simple and complex forms:

(iii) Xo anaphors: apnaa 'self-gen', apnee-P, khud, swayam 'self'

-locally and long-distance bound.

(iv) XP anaphors: reflexive: apnee aap (P) 'Self's self'

reciprocal: eek duusree (P) 'one-other'

Both apnee (Genitive or with postpositional) Case and apnee-aap-P may be locally subject-bound, with more emphasis conveyed by the complex form (1). The genitive form has only the simple form, so it is always available for long-distance coindexing. But this is not because it is an adjunct. Both genitive arguments (2) and adjuncts (3) have local and long-distance subject antecedents, like non-genitives (4), (5).

The Raising to INFL analysis involves cliticizing the X0 APNAA to TENSE (or ASPECT) at LF, for referential identification of the reflexive by coindexing with the specifier of TENSE/ASP, the subject. Head to head raising of TENSE-APNAA successively moves this complex head eventually to another TENSE in a matrix clause, allowing referential agreement with its specifier. XP raising to X0 is blocked by the Uniformity Principle on chains. Since movement to local INFL is a prerequisite for long distance movement, further movement to a higher INFL is blocked. This account predicts that complex anaohors which are only locally bound will not have to be subject oriented, a prediction which is true for reciprocals but not for reflexives (4), which must have subject antecedents.

This account of the local and long-distance binding of anaphors follows Cole and Sung (1994) except that the relevant functional head cannot be AGR, but rather has to be TENSE, which enforces the EPP subject requirement. This variation on the proposal predicts that local binding (especially of simplex anaphors) is possible only in syntactic domains with TENSE/ASPECT, accounting for the lack of local binding in some (event) nominalizations, and in causative complements, which lack TENSE/ASPECT as well as internal agreement. Local binding of both X and XP anaphors is possible in clausal domains without overt verbal agreement, but with ASPECT morphology on the verb; local binding is also possible in small clauses with predicate nominals (and no overt agreement). But some local binding is possible in NP if the nominal is a picture noun; the host functional projections for APNAA may include a Pred or Nom head (Bowers (1993)).

The verbal agreement system in Hindi/Urdu is independent of subject coindexing, but very clearly correlated to Nominative Case checking on both subjects and objects. Only subjects bind reflexives, but non-nominative subjects do not trigger agreement. If postpositional case blocks subject agreement, then verbal agreement can be with the object. Sentences with local reflexive binding (6) and long distance binding (7) have only subject antecedents for the reflexives, but object agreement with a Nominative object. This possibility is accounted for by assuming that APNAA cliticizes to TENSE, whose specifier is the subject. Thereferential identification of APNAA is by Spec/head agreement with the DP subject, which -if it has postpositional Case-does not project any phi-features for agreement. In fact we may assume AGR-S is not projected in sentences like (7) since it has no Nominative subject case to check in this instance. I am assuming that the nominative object in (7) has its Case and phi features locally checked by AGR-O. There is only one set of features, whether they are subject or object features, andthey are manifested specifically on the functional heads (TENSE ASPECT) above the VP complex. In sentences like (6) and (7), AGR-S could not identify a reflexive in terms of phi-features, since the phi features it shows are those of the object. AGR can identify its specifier only in terms of phi-features, not referential indices, since Case features on DP and AGR features on verbs are not interpretable and therefore erased before LF.

Examples
1) a. apnee-koo deekhoo! 'Look at yourself (you are a mess)'

      self-dat  see-imper (Montaut (1991))



 
 
b.    apnee-aap-koo deekhoo! 'Look out for yourself (don't worry about others)
      self self-dat see-imper (Ibid)



 
 
2) bacceei-nee [duusree bacceej-see] apnaa /us-kaa khilaunaa
   child-erg second child-from self's 3psg-of toy
  chiin liyaa
  snatch take-pf '[One childi] snatched from [another childj] [self'si/*j toy].'



 
 
3) Siilaai-nee vijayj-koo [PROj apniii/jus-kiii/*j praSamsaa
   Shila-erg   Vijay-dat        self's 3psg-poss   praise
   kar-nee-see   manaa      kiyaa apnii= Vijay/Shila
   see-inf -from prohibited do-pf us-kii=Shila/3p/*vijay
'Shilai forbade Vijayj [PROj to praise selfi/j]'



 
 
4) wooi bacceej-koo apnee-aapi/*j-see kaisee alag kar saktii hai?
3psg.f child-dat self's self-from how separate do-impf is
How can shei separate/remove the childj from selfi/*j?



 
 
5) maaNi-nee raamj-koo [PROj apnee(aap)i/j-koo gumnaam
mother-erg. Ram-dat. self-dat anonymous
patr likh-nee] kee liyee manaa liyaa
letter write-inf. of of sake forbid do-pf
'Motheri forbade Ramj [PROj to write selfi/j anonymous letters]'
apnee-koo = Ram(preferred)/mother apnee aap-koo=Ram/*mother



 
 
6) maaNi-nee baccooNj-koo [PROj apneei/j kamree-meeN kitaab
mother-erg children-dat self's room-in book-f.
paRh-nee dii
read-inf-obl. give-pf-f
'Mother allowed the children [PRO to read a book in her/their room]'.



 
 
7) baccooN-nee apnii kitaabeeN paRh-iiN
children-erg self's books read-pf
3ppl.m. f.pl. f.pl
'The children read self's books'
 
Link to Schedule