Clause Type Language Type: Ergative Accusative Transitive Erg-Abs Nom-Acc Intransitive Abs Nom

The study of case has been an integral part of the study of clausal structure for cen-
turies. In the ancient Greek tradition, for example, the actants of a clause were defined
primarily in terms of semantic case notions like accusative ‘accused/affected’, dative
‘giving’, or vocative ‘called/named’. In the Pān.inian tradition (ca. 6th century BCE),
the relationship between the arguments of a clause and their case realization was me-
diated via a level of semantic roles such as actor and undergoer. While the ancient no-
tions have remained with us as part of the modern view of syntax, languages also came
to be classified according to the kind of subject vs. object marking they displayed. In
particular, a distinction is drawn between ergative and accusative types of languages.
This distinction goes back to Fillmore (1968) and has been taken up in various forms
in the literature (see Manning 1996 for an overview). Most of the discussions sur-
rounding the proposed distinction assume an opposition between ergative-absolutive
and nominative-accusative types of clauses. This is illustrated in Table 1.
Table 1. Clause Types
Clause Type Language Type

Ergative Accusative
Transitive Erg-Abs Nom-Acc
Intransitive Abs Nom
Thus, languages are classified by the case pattern exhibited by transitive and intransi-
tive clauses, whereby the deeper distinction between ergative and accusative languages
is taken to manifest itself in the grouping of the grammatical relations. In ergative lan-
guages, objects and intransitive subjects group together in terms of case marking; in
accusative languages, subjects of intransitives and transitives are treated alike and are
differentiated from objects in terms of case marking.
Within this typology, Urdu/Hindi has been situated as a morphologically split-erga-
tive language in which the ergative marker is sensitive to perfect morphology (e.g.,
Anderson 1977, Mahajan 1990, Dixon 1994). This typological classification requires
1
2 M. B UTT AND T. H. K ING
the ergative to alternate with an unmarked case (nominative/absolutive) on subjects of

transitive sentences. This alternation is found in Urdu/Hindi, and so the entailment is
confirmed. Another entailment is that the ergative not appear on intransitive subjects.
However, this is not supported by the facts of the language, as Urdu/Hindi intransitives
also show an ergative/nominative alternation on subjects (section 2.1.).
The classification in Table 1 represents an oversimplification of case patterns as no
other case marking alternations are assumed to exist. Even if they are assumed to ex-
ist, they generally are not taken to be of typological relevance in characterizing the
clause structure of a language. In this paper, we argue that case alternations involving
datives, accusatives, and instrumentals must also be taken into account for an insightful
understanding of the role of case in the clause structure of Urdu/Hindi.
With respect to case alternations on objects, the literature to date has produced sev-
eral well known examples. These include the alternation of overtly marked accusative
objects in Turkish with unmarked (nominative) objects, which correlates with specific
vs. nonspecific interpretations (Enç 1991), the distribution of the Finnish partitive (de
Hoop 1992, Vainikka 1993, Kiparsky 1998, 2001), which is associated with (non)telicity
and (un)boundedness, and the alternation between Scottish Gaelic genitive and un-
marked objects (Ramchand 1997), which correlates with aspectual affectedness. None
of these languages are ergative (or split-ergative); so it would seem at first that these
case alternations are not relevant for the present discussion. However, an alternation
similar to the nominative-accusative split in Turkish is found in Urdu/Hindi (e.g., Allen
1951, McGregor 1972:49–50, Butt 1993, Masica 1991, Mohanan 1994, Singh 1994),
and the accusative case involves a notion of affectedness (Saksena 1982, Butt 1998).
Similarly, although it is well known that South Asian languages allow non-nomi-
native subjects besides the ergative (e.g., Verma and K.P. Mohanan 1990, Mohanan
1994), we are not aware of any ergativity discussion which integrates the fact that split-
ergative languages such as Urdu may also show patterns such as instrumental-unmarked,
dative-unmarked, dative-accusative, as well as the canonical ergative-unmarked (where
the unmarked option has been called “absolutive” or “nominative”; see section 2.2.).
Consider the miminal pair in (1), found in some dialects of Urdu/Hindi, in which an
ergative subject alternates with a dative one. This alternation is restricted to infinitive
plus ‘be’ contexts, and the sentences only differ with respect to the case marker.
(1) a. nadya=ne zu ja-na h

Nadya.F.Sg=Erg zoo.M.Sg.Obl go-Inf.M.Sg be.Pres.3.Sg
‘Nadya wants to go to the zoo.’
b. nadya=ko zu ja-na h
Nadya.F.Sg=Dat zoo.M.Sg.Obl go-Inf.M.Sg be.Pres.3.Sg
‘Nadya has to go to the zoo.’
In addition, as mentioned above, the object case may alternate in accordance with spe-
cific vs. non-specific interpretations, as in (2), giving rise to ergative-unmarked and
ergative-accusative patterns not predicted by the view in Table 1.
T HE S TATUS OF C ASE 3
(2) a. nadya=ne jiraf dekh -na h

Nadya.F.Sg=Erg giraffe.M.Sg.Nom see-Inf.M.Sg be.Pres.3.Sg
‘Nadya wants to see a giraffe/giraffes.’
b. nadya=ne jiraf=ko dekh -na h
Nadya.F.Sg=Erg giraffe.M.Sg=Acc see-Inf.M.Sg be.Pres.3.Sg
‘Nadya wants to see the giraffe.’
This variety of patterns in case marking, combined with the correlation between
case morphology and semantic effects, renders the generally accepted ergative vs. ac-
cusative language division as too simplistic. It also poses a challenge for Case Theory,
in which the occurrence of case has generally been understood in the (by now clas-
sic) division between structural and inherent Case (Chomsky 1981). Under this view,
structural Case identifies the core grammatical relations subject and object configura-
tionally and interacts with agreement, while inherent case takes care of all other types
of (morphological) case markings, both regular (such as the dative on indirect objects
or the ergative on subjects) and irregular or quirky case marking.
With respect to Urdu/Hindi, this division cannot be upheld in a strict form, as Ma-
hajan’s (1989, 1990) treatment of the interaction between word order, agreement, and
case marking in Hindi shows. His analysis includes the idea that a given NP (e.g., an
ergative NP) can be both structurally and inherently case marked: structurally because
the NP has to be identified as a given grammatical relation (e.g., subject for ergative
NP s) and undergo the appropriate movements; inherently because the overt case mark-
ing (e.g., ergative) is assumed to come from a stipulation in the verb’s lexical entry.
In addition, a semantically meaningful use of case as in (2) contrasts sharply with
truly idiosyncratic or quirky case over which no coherent semantic or syntactic gen-
eralizations can be formulated. In particular, a recognition within formal semantics of
the fact (often noted in descriptive and typological studies, see Wierzbicka 1981 for
an overview) that a difference in overt case marking affects the compositional seman-
tics of a clause (de Hoop 1992, Krifka 1992, Verkuyl 1993, Ramchand 1997) questions
the notion that inherent (morphological) case is a semantically uninteresting (uninter-
pretable), lexically stipulated requirement on overt form (cf. Chomsky 1995).
Woolford’s (1997) four-way case system for Nez Perce addresses some of these
problems. She posits crosslinguistic availability of the patterns ergative-nominative,
ergative-dative, and ergative-objective. She furthermore associates ergative, dative,
and accusative with agents, goals/experiencers, and themes, respectively, thus acknowl-
edging that case may be semantically sensitive.
The view articulated in this paper is close to Woolford’s. However, she employs
the standard distinction between structural and lexical/inherent Case which assumes
that structural Case is checked by either functional or lexical heads, while lexical Case
is stipulated in the verbal entries. We instead propose a three-way distinction between
structural, semantic, and quirky case. This three-way distinction arises out of the Urdu/-
Hindi case system as a whole, using Lexical-Functional Grammar (LFG) as a gram-
matical theory which does not force a (phrase structural) predisposition towards the
treatment of case marking.
As Urdu/Hindi is a language with very flexible word order, we differ from Wool-
ford (1997) in that we have no notion of phrase structural Case in which case markers
are checked by lexical or functional positions. Rather, following LFG’s linking theory,
we capture the notion of structural Case via a theory of linking which relates thematic
roles to grammatical relations. We also differ with respect to the notion of lexical case;
we differentiate the case markers over which one can state regular semantic generaliza-
tions (semantic case) from those whose appearance is truly idiosyncratic (though there
is usually a historical explanation).
The paper is organized as follows. Section 2 contains an overview of the current
usage of case in Urdu/Hindi. Section 3 provides some information as to the diachronic
picture. Sections 4 and 5 present our approach to case and discuss the synchronic status
of the case markers as clitics. Finally, section 6 presents detailed analyses of case mark-
ing alternations on subjects and focuses on the alternation between ergative and dative
subjects in the infinitive constructions presented in (1). The conclusions we draw are
summarized in section 7.
In this section, we show that Urdu/Hindi exhibits a structural system of grammati-

cal functions which is overlaid with the use of semantic morphological case. That is,
core grammatical relations can be marked with a number of different case alternations
which have an effect on the semantic interpretation of the clause. This position is not
new: statements to the effect that Hindi should be analyzed as a fundamentally “ac-
cusative” system can be found in Kachru (1987), Dixon (1994:175), Mohanan (1994),
and Manning (1996). The connection of case marking to semantics also appears re-
peatedly in the papers cited.
The basic Urdu/Hindi case markers are given in Table 2.
Table 2. The Case Markers
Clitic Case Gram. Func. Morph. Effect

nominative subj/obj none
ne ergative subj oblique marking on NP
ko accusative obj oblique marking on NP
dative subj/ind. obj oblique marking on NP
se instrumental subj/obl/adjunct oblique marking on NP
k- genitive subj (infinitives) agrees with head noun
specifier none
mẽ/p r/t k/ locative obl/adjunct oblique marking on NP
There are seven cases which are morphologically realized by seven markers: the nom-
inative is indicated by the lack of a case marker, while the accusative and dative share
the marker ko. The locative employs one of three markers or a null marker depend-
ing on the meaning. The instrumental se spans a range of functions, some of which
are discussed in section 2.5. The only case marker that inflects is the genitive (section
2.6.): this is because it arose out of an inflecting participial construction. All of the
case markers mark the core grammatical functions subject, object, or indirect object
(Mohanan 1994:64–66). Each case marker appears in a variety of constructions.
In the early days of generative syntax, the occurrence of ergative case was often dealt
with by recourse to more familiar structures or concepts. One strategy was to view
an ergative construction as a type of passive. However, this analysis was found to be
wanting as the syntax of ergative languages became better understood. Another strat-
egy is to view the appearance of ergative case as an instance of “quirky” or “lexical”
case. In this scenario, individual verbs or verb forms lexically stipulate the appearance
of the ergative case. However, this approach is not optimal because the ergative has a
more systematic distribution than would be expected from lexical stipulation.
Bittner and Hale (1996) analyze the Hindi ergative as a marked Structural Case as-
signed by I . While this is an improvement over viewing the ergative as quirky case,
they do not take into account semantic factors which interact with the structural nature
of the ergative. Bittner and Hale account for ergative subjects of intransitives by ana-
lyzing unergative intransitives as underlyingly transitive (cf. Hale and Keyser 1993).
However, it remains a mystery why the ergative is obligatory with overtly transitive
perfect clauses, but is optional with perfect unergative intransitives. For example, as
shown in (3), the ergative alternates with the nominative on some intransitive verbs
(e.g., Kachru 1978) where it is correlated with an expression of volitionality (Butt and
King 1991, Tuite, Agha and Graczyk 1985, Mohanan 1994, Davison 1999).
(3) a. ram kh ãs-a
Ram.M.Sg.Nom cough-Perf.M.Sg
‘Ram coughed.’ (Tuite, Agha and Graczyk 1985:264)
b. ram=ne kh ãs-a
Ram.M.Sg=Erg cough-Perf.M.Sg
‘Ram coughed (purposefully).’ (Tuite, Agha and Graczyk 1985:264)
Urdu/Hindi ergative case has resisted syntactic analysis for several reasons. One is
that the full range of data with regard to the scope and distribution of ergative marking
in Urdu/Hindi is often not taken into account, e.g., the ergative-dative subject alter-
nation in (1). Another reason is that the restrictive view of case in terms of ergative
vs. accusative systems leads the researcher to ignore ergative-accusative case patterns
like the ones in (2). Woolford (1997:221), for example, points out that Bok-Bennema’s
(1991) general approach to ergativity does not allow for an ergative-accusative pattern.
The Urdu/Hindi ergative is sensitive to both syntactic and semantic factors. The
mixed nature of the Urdu/Hindi ergative is acknowledged by researchers who work
primarily on Hindi. As already discussed, Mahajan (1990), for example, proposes that
argument noun phrases in Hindi may have both structural and inherent Case. Struc-
tural Case is assigned in Spec AGRP, Spec IP, or in the complement to V position. The
inherent Case of an argument is specified in the lexical entry of a particular verb form.
The overt case morphology is treated as an instance of inherent Case. Inherent Case
marked noun phrases may also be assigned structural Case if they are in the appro-
priate position. This dual system of Case assignment applies to direct objects marked
with ko and ergative subjects because these nouns function as direct arguments which
are overtly marked with case clitics. A short-coming of Mahajan’s analysis is that the
ability to assign Structural Case is linked to the particular verb form. Psych-predicates
(dative subjects) and verbs with perfect morphology are taken to lack the ability to as-
sign Structural Case, which is why the object moves to a functional agreement posi-
tion, leaving the subject inside the VP and vulnerable to the assignment of Inherent
Case. Furthermore, because Mahajan 1990 formulates a purely structural approach to
ergativity, he cannot account for the ergative-nominative and ergative-dative case al-
ternations.
Davison (1999) provides the most complete account of the pattern of Hindi ergative
marking to date. Her research is based on a careful survey of case marking across sev-
eral verb classes. She treats the ergative as a structural Case which interacts with the
specifications of the lexical entry of the verb and proposes licensing conditions which
take finiteness and lexical and aspectual specfications into account. Davison’s proposal
is close to the one presented here: information coming from the verbs’ lexical entries
interacts with information provided by the ergative. However, under her account, the
semantic factors involved in case-alternations are associated with the lexical entry of
a verb. We instead believe that while some inherent case marking information must
be relegated to the lexicon, a more general story can be told with respect to the case
alternations observed in Urdu/Hindi.
De Hoop (1999) sketches an account of case marking which relies on the semantic
notions of weak and strong Case presented in previous work (de Hoop 1992). If devel-
oped in more detail, her account should be compatible with ours and supplement our
approach with more detailed semantic machinery. Such sophisticated semantic ma-
chinery is needed to account for the distribution of the ergative in Urdu. As shown in
(1) and repeated here, the ergative can appear in nonfinite constructions in Lahori (and
Delhi) dialects of Urdu/Hindi (Butt and King 1991, Bashir 1999). In particular, the
ergative alternates with the dative on subjects in this construction to provide different
modalities: if the dative is used, the subject generally must perform the action; if the
ergative is used, the subject is interpreted as wanting to perform the action.

Butt and King (1991) and Mohanan (1994) therefore argue that the ergative is as-
sociated with volitionality or the feature [ conscious choice]. Bashir (1999), based
on an examination of current usage of the ergative in modern day Urdu TV dramas,
concludes that the picture is not so simple. She observes the pattern in Table 3 and
proposes an explanation in terms of markedness.
Table 3. The Ergative and Semantic Entailments
Tense/Aspect Valency Ergative Semantic Entailment

a. Finite, Perfect Intransitive Unergative No No entailment
b. Yes [ conscious choice]
c. Finite, Perfect Transitive No Exceptional,
No entailment
d. Yes No entailment
e. Infinitive Any No [ source specified]
f. Yes [ source specified]
When the ergative is marked, as in Table 3b, it denotes conscious choice. When it is
structurally required, as in Table 3d, it carries no entailments. With respect to the infini-
tive construction, Bashir speculates that the ergative may be encroaching on the domain
of the dative. In this construction, the ergative is marked and entails a subject who has
control over the action. The dative, however, is unmarked and may be regarded as the
Elsewhere Case: the subject may or may not have control over the action, the precise
interpretation depends on the context.
We incorporate these results into our analysis in section 6 by assuming a disjunctive
entry for the ergative. When the ergative is structurally required (transitive verbs in the
perfect), no extra semantic information is contributed to the clause. When the ergative
is not structurally required, it contributes the information that the action is within the
internal control of the subject.
The nominative is phonologically null and in some discussions is referred to as an abso-

lutive. However, as seen in section 2.1., the ergative does not conform to the standard
ergative-absolutive pattern. Furthermore, the distribution of unmarked NPs exceeds
that of the pattern associated with absolutives. While nominatives do appear as ob-
jects of transitives and subjects of intransitives, as is the case for absolutives, they also
alternate with the accusative ko on objects, as in (5). The alternation is governed by an-
imacy and specificity effects similar to those found in Turkish (Enç 1991). Absolutives
are not typically involved with specificity effects.
(5) a. nadya=ne gar.i c la-yi h
Nadya.F.Sg=Erg car.F.Sg.Nom drive-Perf.F.Sg be.Pres.3.Sg
‘Nadya has driven a car.’
b. nadya=ne gar.i=ko c la-ya h
Nadya.F.Sg=Erg car.F.Sg=Acc drive-Perf.M.Sg be.Pres.3.Sg
‘Nadya has driven the car.’
Furthermore, as was seen in (3), nominatives alternate with ergatives with intransitive
unergative verbs. This is again not a property associated with absolutives. Finally, a
clause may contain more than one unmarked argument, as in (6). This is not a property
typically associated with either absolutives or nominatives.
(6) nadya gar.i c la-ti h
Nadya.F.Sg.Nom car.F.Sg.Nom drive-Impf.F.Sg be.Pres.3.Sg
‘Nadya drives a car.’
These uses of the unmarked case in Urdu/Hindi provide further evidence that a dif-
ferent view of the interaction between case and clausal structure is needed.
The agreement pattern in Urdu/Hindi is sensitive to case marking and grammatical
functions: the verb only agrees with subject or object nominatives. If the subject is
nominative, the verb agrees with it ((6)). If the subject is non-nominative and the ob-
ject is nominative, then the verb agrees with the object ((5a)). If both the subject and
the object are non-nominative, then the verb shows “default” masculine singular agree-
ment ((5b)). As such, agreement does not uniquely identify a grammatical function in
Urdu/Hindi (i.a., Mahajan (1989, 1990, 1991), Davison (1988, 1991a,b), Butt (1993,
1995), Mohanan (1994, 1995)). The full range of agreement patterns still awaits a de-
tailed analysis: in Urdu/Hindi the type of agreement varies from auxiliary to auxiliary
and across different tenses. We briefly return to the issue of agreement as it interacts
with case in section 4.3..
The accusative is form-identical with the dative ko (section 2.4.). Many approaches
therefore assume that Urdu/Hindi lacks an accusative and that the ko is an inherent
dative case (e.g., Mahajan 1990, Davison 1998). However, there are two distinct dis-
tributional patterns with regard to ko. If a single label for ko were assumed, a distinc-
tion would still have to be made at some level of analysis. In this paper, as in earlier
work (Butt 1995), we reify these distinct distributional patterns in terms of two differ-
ent names: accusative vs. dative.
As shown in (5), the accusative marker ko alternates with the phonologically null
nominative on direct objects. The appearance of accusative ko is connected with a sen-
sitivity to animacy (animate objects almost always require ko) and definite/specific in-
terpretations. This property of ko has been discussed extensively (e.g., Allen 1951,
McGregor 1972:49–50, Butt 1993, Masica 1991, Mohanan 1994, Singh 1994).
Furthermore, ko is associated with a notion of affectedness (Saksena 1982). Rele-
vant examples come from causativization patterns that generalize across verb classes.
Some verb classes (i.e., the so-called ingestives ‘eat’, ‘drink’, but also ‘write’) require
the causee to be marked with ko, as in (7a). Most verb classes require the causee to be
marked with instrumental se, as in (7b). Finally, a handful of verbs, as in (8), allow an
alternation between se and ko that has been analyzed in terms of an affected agent by
Saksena (1982).
(7) a. nj m=ne s dd f=ko/*se kh ana kh ıl-a-ya
Anjum.F.Sg=Erg Saddaf.F.Sg=Acc/Inst food.M.Sg.Nom eat-Caus-Perf.M.Sg
‘Anjum made Saddaf eat food (gave Saddaf food to eat).’
b. nj m=ne s dd f=se/*ko paoda k t.-a-ya

Anjum.F.Sg=Erg Saddaf.F.Sg=Inst/Acc plant.M.Sg.Nom cut-Caus-Perf.M.Sg
‘Anjum had Saddaf cut a/the plant.’
(8) a. nj m=ne s dd f=ko m sala c kh -va-ya

Anjum.F.Sg=Erg Saddaf.F.Sg=Acc spice.M.Sg.Nom taste-Caus-Perf.M.Sg
‘Anjum had Saddaf taste the seasoning.’
b. nj m=ne s dd f=se m sala c kh -va-ya
Anjum.F.Sg=Erg Saddaf.F.Sg=Inst spice.M.Sg.Nom taste-Caus-Perf.M.Sg
‘Anjum had the seasoning tasted by Saddaf.’
The accusative is thus another example of a case marker that is structural in the sense
that it can only appear on certain grammatical relations (it appears only on direct ob-
jects in most dialects of Urdu/Hindi), but is also semantic in that it participates in an
specificity alternation with the nominative and an affectedness alternation with the in-
strumental in causatives.
The dative is identical in form to the accusative. It differs from the accusative in that
it marks indirect objects, as in (9), and subjects, as in (10), and never alternates with
nominative objects. Nor does the dative indirect object in (9) ever become subject un-
der passivization, unlike the accusative.
(9) nj m=ne s dd f=ko cıt.t.h i d-i
Anjum.F.Sg=Erg Saddaf.F.Sg=Dat letter.F.Sg.Nom give-Perf.F.Sg
‘Anjum gave Saddaf a letter.’
The dative is associated with the -role goal/experiencer (cf. Verma and K.P. Mo-
hanan (1990) on experiencer subjects, and Mohanan 1994 and references therein). Ex-
periencer subjects encompass modal contexts such as in (10a), psych predicates as in
(10b), the alternation with the ergative as in (1), and subjects of N-V complex predi-
cates, as in (10c).
(10) a. nadya=ko skul ja-na p r.-a
Nadya.F.Sg=Dat school.F.Sg.Obl go-Inf.M.Sg fall-Perf.M.Sg
‘Nadya had to go to school.’
b. nadya=ko d. r l g-a
Nadya.F.Sg=Dat fear.M.Sg.Nom be attached-Perf.M.Sg
‘Nadya was afraid.’
c. nadya=ko k hani yad a-yi
Nadya.F.Sg=Dat story.F.Sg.Nom memory come-Perf.F.Sg
‘Nadya remembered the story.’
The dative is thus also a combination of structural and semantic case. It is restricted
to indirect objects and subjects, but cannot be relegated to the status of an unpredictable
inherent case since there are coherent syntactic and semantic generalizations as to its
distribution and use.
The instrumental se is extremely versatile. It may be used for instrumental adjuncts as

in (11a), for source expressions, both locative, as in (11b), and material, as in (11c), as
well as for comitatives, as shown in (11d), and for causees, as in (8).
(11) a. nadya=ne d rvaza cabi=se kh ol-a
Nadya.F.Sg=Erg door.M.Sg.Nom key.F.Sg=Inst open-Perf.M.Sg
‘Nadya opened the door with a key.’
b. nadya=ne aj lahor=se fon ki-ya

Nadya.F.Sg=Erg today Lahore=Inst phone do-Perf.M.Sg
‘Nadya called from Lahore today.’
c. s nar=ne sone=se har b na-ya
goldsmith.M.Sg=Erg gold.M.Sg.Obl=Inst necklace.M.Sg.Nom make-Perf.M.Sg
‘The goldsmith made a necklace out of the gold.’
d. nadya s dd f=se bat k r r h-i h
Nadya.F.Sg.Nom Saddaf.F.Sg=Inst talk.F.Sg.Nom do stay-Perf.F.Sg be.Pres.3.Sg
‘Nadya is talking to Saddaf.’
The instrumental is also used on adjuncts that express the demoted agent (logical
subject), as in the standard passive in (12). Standard passives are formed with the verb
ja ‘go’ (in all tenses) in combination with perfect morphology on the main verb.
(12) cor (p lıs=se) p kr.-a g -ya/ja-ta
thief.M.Sg.Nom police=Inst catch-Perf.M.Sg go-Perf.M.Sg/go-Impf.M.Sg
‘The thief was caught by the police.’ (adapted from Mohanan (1994:183))
The instrumental also occurs in a construction described as a passive of disability
in some grammars of Urdu/Hindi (e.g., Glassman 1976, Van Olphen 1980), as in (13).
(13) a. nadya=se y rdu=k-i cıt.t.h i p r.h -i n hı̃
Nadya.F.Sg=Inst this Urdu=Gen-F.Sg letter.F.Sg.Nom read-Impf.F.Sg not
ja-ti
go-Impf.F.Sg
‘Nadya does not have the ability to read this Urdu letter.’
b. s=se c l-a n hı̃ ja-e-g-a
Pron=Inst walk-Perf.M.Sg not go-3-Fut-M.Sg
‘She/he can’t possibly walk.’ (in the context of a broken leg) (Glassman 1976:275)
This (dis)ability “passive” differs syntactically from the standard passive. It is possi-
ble with intransitives, as in (13b), unlike the standard passive. The instrumental NP
is obligatory and exhibits subject properties with regard to control and anaphora (see
Mohanan 1994 for a list of diagnostics for grammatical subjecthood in Hindi).
Thus, se can appear in a wider number of contexts than the other case markers: it can
serve to mark core grammatical relations as well as adjuncts. In the adjunct contexts
se has a range of meanings. In the passive and the (dis)ability construction, on the
other hand, the appearance of se is predictable, depending on the verb ja ‘go’ and the
lexical semantics of the verb (if the verb does not allow passivization or does not allow
a dispositional reading, then the construction is illformed).
The genitive may be roughly characterized as marking subjects of nonfinite clauses, as

in (14a), subjects of finite copula constructions, as in (14b), and specifiers of nominals,
as in (14c). Like other case marked nominals in Urdu/Hindi, genitives may be scram-
bled. As such, they are functionally, but not phrase structurally determined. Genitives
are also not semantically motivated: Mohanan (1994:177) considers and discards an
analysis in terms of a semantic notion such as possession.
(14) a. ram=ke b t.h -ne=p r mã=ne s=ko

Ram.M.Sg=Gen.M.Sg.Obl sit-Inf.Obl=on mother.F.Sg=Erg Pron=Dat
kh ana di-ya
food.M.Sg.Nom give-Perf.M.Sg
‘On Ram’s sitting down, the mother gave him food.’ (Adapted from Mohanan
1994:78)
b. ram=ka ek bet.a h
Ram.M.Sg=Gen.M.Sg one son.M.Sg.Nom be.Pres.3.Sg
‘Ram has one/a son.’ (Adapted from Mohanan 1994:177)
c. rani=ka bh ai
Rani.F.Sg=Gen.M.Sg brother.M.Sg.Nom
‘Rani’s brother’
In general, the genitive can be analyzed as marking specifiers. Were it not for the
rather complicated agreement pattern associated with it, this case would be crosslin-
guistically unremarkable. As can be seen from the examples in (14), the genitive in-
flects to agree with the head noun in terms of gender, number, and obliqueness. Payne
(1995) discusses the Hindi genitive and views the agreement pattern as an instance of
Suffixaufnahme.
In this section we briefly consider the historical origins of the modern Urdu/Hindi case
system. Beames (1872) and Kellogg (1893) both provide detailed discussions on the
possible origins of the modern case markings in Urdu/Hindi. Despite their very careful
discussions, much more diachronic work remains to be done to establish the historical
development of the modern case markers at the same level of detail as is available for
the Romance languages, for example. Unfortunately, the amount of research needed
falls outside the scope of this paper, but see Butt (2001a) for a more detailed look at
the ergative case in Urdu/Hindi.
The Sanskrit case system consisted of a complex set of declensions which varied ac-
cording to three numbers (singular, dual, plural) and approximately thirteen different
noun classes. The system distinguished nominative, accusative, instrumental, dative,
ablative, genitive, locative, and vocative. Unlike the Urdu/Hindi system of today, case
was marked by inflectional affixes, as in its sister language Latin. This system was
simplified over time. A rough indication of the developmental stages as described in
Sen (1973) is given in (15).
(15) A. Old Indo-Aryan
1200 BCE — 600 BCE (Vedic)
600 BCE — 200 BCE (Epic and Classical Sanskrit)
B. Middle Indo-Aryan (Aśokan inscriptions, Pāli, Prākrits, Apabhram
. śa—Avahat.t.ha)
200 BCE — 1100 CE
C. New Indo-Aryan (Bengali, Hindi/Urdu, Marathi and other modern North
Indian languages)
1100 CE — Present
Masica (1991:231) gives Table 4 for Middle Indo-Aryan. The endings are listed in
their surface forms as they occur distributed over several nominal paradigms. These
forms were either lost or were collapsed into the modern oblique marking (Beames
1872:209). In particular, the genitive and locative forms seem to have fallen together in
Apabhram . sa, and in old (or archaic) Hindi the ablative, dative, and accusative singular
were then also collapsed (Kellogg 1893:126).
Table 4. Case Forms in Middle Indo-Aryan
Singular Plural
Nominative -u, a, a -a, aı̃
Accusative [same as Nominative]
Instrumental -e , i , he, hi -e(h)ı̃, ehi, ahı̃
Ablative -hu, ahu, aho -hũ, ahũ
Genitive/Dative -ho, aho, ha, su, ssu -na, hã
Locative -i, hi, hi -hı̃
In modern Urdu/Hindi, the oblique is mandatory for all non-nominative (non-direct)

forms. This includes all nouns marked by overt case, as illustrated in (16), as well as
bare locatives, as shown in (17). The oblique manifests itself as the affix -e on mascu-
line singular nouns ending in -a, and as -õ in the plural on all nouns. However, on all
other forms (e.g., feminine singular nouns, masculine nouns not ending in -a such as
yassin in (16)), the oblique marking is not overt.
(16) a. yasin=ne k tt-a dekh -a
Yassin.M.Sg=Erg dog-M.Sg.Nom see-Perf.M.Sg
‘Yassin saw a/some dog.’
b. yasin=ne k tt-e=ko dekh -a
Yassin.M.Sg=Erg dog-M.Sg.Obl=Acc see-Perf.M.Sg
‘Yassin saw the dog.’
(17) adnan d.akxane g -ya h

Adnan.M.Sg.Nom post office.M.Sg.Obl go-Perf.M.Sg be.Pres.3.Sg
‘Adnan has gone to the post office.’
Because the oblique is a prerequisite for the ergative, dative, accusative, instrumen-
tal, genitive, and locative marking, as well as for postpositions (section 4.2.), Masica
(1991:231–236) analyzes the oblique as Layer I of three possible layers of case mark-
ing. We do not follow this analysis, but see this remnant of the Sanskrit system as en-
suring synchronic morphological wellformedness: if the noun is in the oblique form,
then modifiying adjectives must also be in the oblique form. The one use of the oblique
marking in the modern system in which the original case function has been preserved
is the use of the oblique to mark locatives as in (17). The meaning of this locative use is
restricted: it can only mean ‘to’ or ‘from’, not ‘in’, ‘towards’, etc. We thus distinguish
between two forms of the oblique morphemes (section 5): one as agreement inflection
on the complement of a K head, and one as a locative case marker.
None of the modern case forms are descended from the forms in Table 4, despite the
wide-spread belief in the literature that the modern ergative ne is descended directly
from the Sanskrit instrumental -ina. As shown in section 4, the modern ergative is a
clitic and not an affix, as would be expected if it were a continuation of the Sanskrit in-
strumental inflectional affix. Furthermore, a survey of Old Hindi writers in the middle
ages shows that while they used an ergative pattern, there was no sign of ne (Beames
1872:267–271). Instead, the oblique marked the subjects of eligible transitive verbs.
In the equivalent to (18), for example, modern Urdu/Hindi would require that ‘king’
and ‘brahmin’ be marked with the ergative ne due to the perfect morphology on the
transitive verb. However, this is not the case in (18). There is no overt manifestation
of oblique marking presumably because these are not masculine nouns ending in a.
(18) daha bār bujh-yeu rāj
ten time ask-Perf.M.Sg king
duj di-yeu na utar kāj
brahmin give-Perf.M.Sg not answer matter
‘Ten times the King asked, The Brahmin gave no answer in the matter.’
[Old Hindi] (Chand, Prithiraja-Rasau i.49; Beames 1872:267)
In (19) the oblique marking on the subject is overt: the form of the wh-word ‘who’ is in
the oblique. In (20), the subject ‘Kabir’ again shows no overt oblique marking, but here
the verbs all agree with the object, just as would be the case in the modern language for
a transitive verb with a non-nominative subject (section 6). This is particularly clear
for the verb ‘take’: it agrees with the feminine noun ‘pen’.
(19) jihi rac-e suraga bhu satta pātāla
who.Obl create-Perf.M.Pl heaven.M.Nom earth.Nom seven.Nom hell.M.Nom
‘Who made heaven, earth, the seven hells.’ [Old Hindi]
[He who created heaven, earth and the seven hells.]
(Chand, Prithiraja-Rasau i.11; Beames 1872:267)
(20) masi kāgad chū-yo nahı̄ kalam gahı̄ nahi hāth
ink.Nom paper.M.Nom touch-Perf.M.Sg not pen.F.Sg take.Perf.F.Sg not hand
jāro juga māhātma jehi kabir jan-ā-yo nāth

four.Pl age.Pl glory.Nom who.Sg.Acc Kabir.Obl know-Caus-Perf.M.Sg lord.Nom
‘Kabir touched not ink nor paper, he took not pen in hand; He made known the lord to
whom is glory in the four ages.’ [Old Hindi] (Kabir, Sakhi 183; Beames 1872:269)
The origin of the ergative ne is not known. Beames and Kellogg propose the par-
ticipial form lage of the verb lag ‘stick to’ as a possible ancestor (Beames 1872:264)
and speculate that the Urdu/Hindi ergative may be derived from a subjective dative
that was in use in another dialect of Hindi around the time of the Moghul Emperor
Shah Jehan’s reign (1627–1658). We are exploring the idea that the Sanskrit locative
janyē ‘for the sake of, because of, caused by’ may have been the ancestor of ne as well
as some dative forms, such as Assamese no and Gujarati nẽ, in other South Asian lan-
guages (Butt 2001a). This form also gave rise to the Bengali postposition jonno ‘for’
(Chatterji 1926:769). As other case markers in Urdu/Hindi appear to have their origin
in former nouns, janyē is a reasonable candidate as an ancestral form of the modern
ergative case marker.
The instrumental se may either be connected with Sanskrit sam ‘with’ or with the
locative singular noun sañge ‘in attachment to’ (Kellogg 1893:132). The genitive arose
out of a participial form of ‘do’ (fn. 11), and the dative/accusative ko is generally traced
to the Sanskrit locative noun kákshe ‘armpit, side’ (Kellogg 1893:130). The locatives
are derived from various nouns.
The basic paradigm of pronouns in Urdu/Hindi is given in Table 5 (based on Kellogg

1893, Glassman 1976, Beg 1988). As can be seen, the pronoun system is quite com-
plicated. While there are no gender distinctions and the Sanskrit dual has been lost,
the second person includes some honorific distinctions and the third person makes a
difference between proximal and distal. The instrumental and locative appear to pat-
tern together. However, the nominative, ergative, and accusative/dative stand alone, as
does the genitive, which is the only case which inflects according to gender and number
(as with the nouns).
Dasgupta, Ford, and Singh (2000:121–146) examine data with respect to the inter-
action of case and pronouns in Hindi and conclude that the forms we have been treat-
ing on a par as case markers should be subdivided into two classes: true adpositions
and case expressing light adpositions. True adpositions include the instrumental se
and the locatives mẽ ‘in’ and p r ‘on’. Light adpositions are dative/accusative ko,
ergative ne, and the genitive k-. Their classification is based primarily on a seman-
tic argument: the dative/accusative, ergative, and genitive are structural cases because
they are not interpretable. The locatives and the instrumental are semantically inter-
pretable and hence not structural. We argue that this type of a division is unwarranted
for Urdu/Hindi due to the many semantically motivated alternations found with struc-
tural case markers.
Another argument for the classifcation advanced by Dasgupta et al. (2000) is that
the dative/accusative, ergative, and genitive appear to receive special attention in the
pronimal paradigm in terms of their morphological forms. However, they do not con-
sider the complete pronoun paradigm. As can be seen in Table 5, the dative/accusa-
tive, ergative and genitive do not form a natural class.
Table 5. Pronoun Paradigm
NOM ERG ACC / DAT INST LOC GEN

1.Sg m˜ m˜=ne m jh =ko m jh =se m jh =p r mer-a/i/e
m jh e
1.Pl h m h m=ne h m=ko h m=se h m=p r h mar-a/i/e
h me
2.Disresp. tu tu=ne t jh =ko t jh =se t jh =p r ter-a/i/e
t jh e
2.Familiar t m t m=ne t m=ko t m=se t m=p r t mhar-a/i/e
t mhe
2.Resp. ap ap=ne ap=ko ap=se ap=p r ap=k-a/i/e
3.Prox.Sg ye ıs=ne ıs=ko ıs=se ıs=p r ıs=k-a/i/e
ıse
3.Prox.Pl ye ın=ne ın=ko ın=se ın=p r ın=k-a/i/e
ınhõ=ne ınhõ=ko ınhõ=se ınhõ=p r ınhõ=k-a/i/e
ıne
3.Dist.Sg vo s=ne s=ko s=se s=p r s=k-a/i/e
se
3.Dist.Pl vo n=ne n=ko n=se n=p r n=k-a/i/e
nhõ=ne nhõ=ko nhõ=se nhõ=p r nhõ=k-a/i/e
nhẽ
In fact, the pronoun forms reflect their historical origin. The third person forms are
demonstratives which have been coopted by the pronoun paradigm (Sanskrit also em-
ployed demonstratives as third person pronouns, but not the same forms). The ap, the
respectful second person, is apparently derived from the Sanskrit noun ātman ‘soul,
self’ (Kellogg 1893:181). These pronouns therefore pattern like nouns with respect to
the genitive. Furthermore, the demonstrative origin common to the third person pro-
nouns, which is not shared by any of the other pronouns, accounts for the particular
forms of the oblique marking in the third person. The other pronouns represent con-
tinuations of the original Sanskrit pronoun system.
We take the variation in pronoun forms to be an indication of continuing histori-
cal change (several pronoun forms which were attested in Old and Middle Urdu are
no longer in use today). While the pronoun paradigm and its interaction with case
is interesting and worthy of further exploration, especially in comparison with the di-
achronic developments and synchronic patterns of related Indo-Aryan languages, we
now return to the discussion of our theory of case as applied to Urdu/Hindi.
In sum, the original Sanskrit case morphology has been almost completely lost. The
oblique marker is a last vestige of the original morphemes. It still fulfills a locative
function in restricted contexts, but now primarily serves as a morphological wellformed-
ness checker similar to the weak/strong agreement pattern of German adjectives and
determiners (see Butt, King, Niño, and Segond (1999:103–108) for a description and
LFG analysis).
Masica (1991), on the other hand, treats the oblique marking as Layer I in a series
of case layers. Layer II includes the items in Table 2 which we consider to be case
markers. Layer III encompasses the nominal part of postpositions (section 4.2.). Ma-
sica’s three layer system of case marking essentially mirrors the historical development
of case and postpositional marking in South Asian languages. Layer I encompasses
the vestiges of the old system. In some South Asian languages, these vestiges play a
greater role in the case marking system than what can be observed in Urdu/Hindi. For
those languages, an inclusion of the oblique marking as part of the case system seems
reasonable. Layer II encompasses those items which now function as case markers,
while Layer III denotes nouns such as up r as in mez ke up r ‘on top of the table’,
where the ‘top’ is linked to the main noun via the genitive (as in the English transla-
tion). In this paper, we essentially make the same distinctions as Masica, but formalize
them differently. Layer II is case proper for us, while Layer III encompasses the class
of postpositions. In what follows, we build on Masica’s basic insight and integrate it
into our analysis. As part of this analysis, we establish that the case markers under dis-
cussion are clitics and not inflectional affixes (section 4.1.) or postpositions (section
4.2.). This has been argued for by Mohanan (1994). In the next section, we repeat
some of her arguments, as well as adding our own.
Given that the case markers always appear after an NP, one possible assumption is that
they are affixes. In this section, we review evidence from stress, coordination, and an
intervening focus clitic (hi ‘only’) to argue that the case endings are clitics (i.e., phrasal
affixes).
One classic test for clitic status is the interaction with coordinate structures. Inflec-
tional affixes do not scope over a coordinate structure; instead, they can only be at-
tributed to the stem to which they attach. For example, the oblique marking -e on nouns
is an inflectional affix. In (21a) the case clitic ko requires this oblique marking on its
complement. Even when the complement is coordinated, the oblique marking cannot
take scope over the coordinated nouns, whether the first one is just the stem k tt, as in
(21a), or the noun with the non-oblique ending k tta, as in (21b). Instead, the oblique
ending must appear on both nouns in the coordinate structure, as in (22a).
(21) a. *[[k tt or gh or.]-e]=ko b. *[[k tt-a or gh or.]-e]=ko

dog and horse-M.Sg.Obl=Acc dog-M.Sg and horse.M.Sg-Obl=Acc
In contrast, the case endings in Urdu/Hindi can scope over coordinated noun phrases,
as in (22). For similar examples and argumentation see Mohanan (1994) and Payne
(1995) on the genitive.
(22) a. yasin=ne [k tt-e or gh or.-e]=ko dekh -a h

Yassin.M.Sg=Erg dog-M.Sg.Obl and horse-M.Sg.Obl=Acc see-Perf.M.Sg be.Pres.3.Sg
‘Yassin saw the dog and the horse.’
b. nadya [lahor or karach i]=se h
Nadya.F.Sg.Nom Lahore and Karachi=Inst be.Pres.3.Sg
‘Nadya is from Lahore and Karachi.’
The focus clitic may be placed between the case marker and the nominal, as seen in
(23), where hi ‘only’ appears between various nouns or pronouns and the case markers.
(23) a. s=hi=ne kam ki-ya

Pron.3.Sg=Foc=Erg work.M.Sg.Nom do-Perf.M.Sg
‘That one himself/only did (the) work.’
b. t jh =hi=ko di-ya
you.Obl=Foc=Dat give-Perf.M.Sg
‘I gave it to you (and not to anyone else).’ (Platts 1967:300)
c. m˜ vahã saik l=hi=se p h˜ch s k-ti hũ
I.Nom there bicycle=Foc=Inst reach able-Impf.F.Sg be.Pres.1.Sg
‘I can get there with just a bike.’ (Sharma 1999)
d. Gauri: to s=se m tl b?
so Pron.3.Sg.Obl=Inst meaning.M.Sg.Nom
Lakh a: h m=hi=se sara m tl b
Pron.3.Pl=Foc=Inst all.M.Sg.Nom meaning.M.Sg.Nom
Gauri: ‘So, what meaning of that [what’s the meaning of that]?’
Laka: ‘All meaning is from us/me alone [the meaning is all of my knowing].’
(L gan ‘Tax’, Hindi Movie)
The examples in (23) have been drawn from a variety of sources because judgements
on these sentences vary. Sharma (1999) documents speaker variability with respect
to clitic placement, but shows that clitics can be inserted between a noun and its case
marker. This supports our view that case markers are clitics.
With true affixes, the focus clitics behave differently: the clitic cannot split an affix
from its stem. This is seen in (24a) in which a clitic appears between the oblique ending
-e and the noun stem k tt. A similar example is shown for verbal affixes in (24b).
(24) a. *k tt-hi-e b. *kh el-hi-a

dog-Foc-Obl play-Foc-Perf.M.Sg
Furthermore, stress facts indicate that the case endings are clitics and not affixes.
Case markers do not affect the placement of stress, while affixes do. For example, in
trisyllabic words with three heavy syllables, the stress falls on the penult as in talA i
‘search’. If the case markers affected the stress system, one would expect A a + ko ‘to
Asha’ to become a Ako, but the stress remains on the first syllable.
The case endings in Urdu/Hindi are therefore not inflectional affixes. Since they are
not syntactically independent, we propose that they are clitics (see Masica 1991:234 for
a similar conclusion). In particular, they are clitics which syntactically attach to NPs.
Due to the above properties, and due to the fact that the case markers attach postnomi-
nally, case endings have been described as postpositions in many accounts of Urdu/Hin-
di. Again following Mohanan (1994), we wish to make clear that postpositions differ
from the case markers in terms of form and distribution. Consider the typical postpo-
sitions in (25) and example in (26).
(25) ke pich e ‘behind’ ke p hle ‘before’

ke nich e ‘under’ ke pas ‘next to’
ke up r ‘over’ ke sath ‘with’
ke nd r ‘inside’ ke liye ‘for’
ke samne ‘in front of’ ke t r f ‘in the direction of’
ke age ‘in front of (further along)’ ke bad ‘after’
(26) bılli bıstar [ke nich e] so rah-i h

cat.F.Sg.Nom bed.M.Sg Gen.Obl under sleep stay-Perf.F.Sg be.Pres.3.Sg
‘The cat is sleeping under the bed.’
All the items in (25) correspond to a relational preposition in English. In contrast to

English, they appear after the NP. The ke in each of these postpositions is the oblique
form of the genitive, which is now invariant for most postpositions. For the native
speaker, this invariant ke is not associated with the genitive, but is taken to be part of the
postposition. Similarly, the final -e on some of the postpositions in (25) is presumably
a reflex of the oblique marking, which indicated a locative.
As shown in (27), some postpositions allow inflection and establish the link to an
original genitive construction (also see Masica 1991:234) in which the postpositions
of today were nouns linked to another noun via the genitive (cf. English because of ).
(27) m˜=ne bılli=ka pich a ki-ya

I=Erg cat.F.Sg=Gen.M.Sg behind.M.Sg.Nom do-Perf.M.Sg
‘I went after the cat.’
In addition, the contentful part of the postposition may appear by itself, as in (28). This
is not the case for any of the case clitics, including the locatives mẽ, par, and tak listed
in Table 2.
(28) up r ao
up come.Impf
‘Come up!’
There is thus a clear difference in distribution and form between the case markers
and the ke postpositions. However, the locative case markers (mẽ ‘in’, p r ‘on’, tak ‘to-
wards’) do pattern with the postpositions in one respect. Although Urdu/Hindi does not
normally exhibit case stacking, se ‘from’ (and mẽ ‘in’) may stack on top of locatives,
be they case markers, as in (29a), or postpositions, as in (29b).
(29) a. n lõgõ=mẽ=se tin
that.Pl.Obl people.Obl.Pl=in=from three
‘three from among those people’
b. lmari [ke pich e]=se
cupboard Gen.Obl behind=from
‘from behind the cupboard’
However, the relevant generalization is over locatives, not over a particular syntactic
class (case clitics vs. ke postpositions). It would therefore be a mistake to base the iden-
tification of case clitics with postpositions on this one argument, especially as the case
clitics can all appear on subject noun phrases, while noun phrases with ke postpositions
do not mark subjects (Mohanan 1994).
As such, postpositions must be distinguished from case clitics. In the following sec-
tion, we propose that case clitics are heads of a KP, whereas postpositions are Ps which
head a PP.
Crosslinguistically it has long been noticed that only certain types of words become
clitics (see Sadock 1991 for discussion). In order to capture this generalization, it has
been proposed that functional heads can be clitics, while lexical categories such a nouns
cannot, unless they undergo historical development and change category along with the
change in prosodic status (see Franks 1999 and references therein).
The Urdu/Hindi data support this idea, and we propose that the case endings are
functional heads of a KP (KaseP). This gives rise to the basic structure in (30b). We
assume that the oblique marking on nouns (singular -e in (30)) is synchronically the
result of the complement-head relationship between the K and the NP. This marking is
obligatory when there is an overt K head. However, not all nominals allow the oblique
marking to surface, as discussed in section 3.1..
(30) a. l r.ke=ne b. KP
boy.M.Sg.Obl=Erg
NP K
ne
N
lar.ke
This use of KP differs from that proposed by Lamontagne and Travis (1986) and em-
ployed by Löbel (1994) and Bayer, Bader, and Meng (2001). Lamontagne and Travis
(1986) proposed the KP in analogy with IP and CP. Under this view (as under our view),
the K is a functional element. However, in our approach, the type of functional infor-
mation it can contribute is not restricted to a simple case value. Instead, K can con-
tribute a complex of features associated with case, including grammatical function in-
formation and semantically relevant material such as volitionality. This is not as radical
a proposal as it might at first seem: before the split Infl hypothesis, I included informa-
tion on tense, aspect, mood, and agreement; we extend this idea in that the functional
head can contribute information, in addition to checking it against features.
The encoding of a complex of features in functional categories such as K is also
employed by Löbel (1994). However, in her approach, as in that of Bayer, Bader, and
Meng (2001), no distinction is made between Ks which are bound pieces of morphol-
ogy and Ks which are independent syntactic words (e.g., prepositions). In Bayer,
Bader, and Meng’s (2001) analysis of German case marking, which is inflectional and
most reliably found on determiners and adjectives, the category KP is used for both da-
tive case (inflectional) and governed prepositional phrases. In Löbel’s (1994) approach
to case alternations on objects, K makes no distinction between sublexically bound and
syntactically independent words. For example, German determiners such as der ‘the’
are formed via Head Movement of K to D. Such approaches cannot be incorporated
into our analysis because we subscribe to the Lexical Integrity Principle (Bresnan and
Mchombo 1995). This principle holds that words are built out of different structural
elements and are composed by different principles than syntactic phrases. Within LFG,
this principle is encoded through c(onstituent)-structure representations such as (30b)
which do not extend into the morphology. That is, bound morphemes cannot appear
independently in the phrase structure, and thus not as a K. Since case markers in Urdu
are clitics, i.e., are prosodically dependent on another element but have different prop-
erties from bound morphemes, they may appear as the head of a K phrase.
Our proposal is structurally quite close to that of Neeleman and Weerman (1999),
who employ a CaseP which carries structural features and which may remain empty
with nominatives. We also posit a fundamental distinction between nominative and
non-nominative cases (cf. Neeleman and Weerman 1999:63), but we include semantic
information in our case markers and generally allow them to play a more active role.
We analyze genitives as occuring in the Spec NP position. For a similar analysis

of genitives see Davison (1998).
The oblique marking on the NP complement of K was discussed in section 4.3.1.
The agreement between the genitive k- and the head noun is due to the fact that this
case marker originated from a participial construction. This agreement can be viewed
synchronically as the result of NP internal agreement.
(31) a. asım=ke b. KP
Asim.M.Sg=Gen.Obl
K
l r.ke=ne
boy.M.Sg.Obl=Erg NP K
ne
KP N
asım=ke
N
l r.ke
Bare nominatives, i.e., the nominals which have no overt case ending and no oblique
inflection, distribute syntactically like KPs with overt case marking. We therefore as-
sume that these project a KP, albeit one without an overt K head, as in (32b). Since LFG
does not posit empty categories, the K head of the KP is not projected in the structure in
(31b) (see King 1995 and Bresnan 2001 for constraints on LFG phrase structure). Un-
der the assumption that the oblique endings are the result of the overt K head’s require-
ments on its complement, no oblique ending occurs in the nominative. The nominative
case comes from default rules (section 5.2.) which state that subject and object KPs in
Urdu/Hindi require case and that if there is no other case, the nominative is assigned.
(32) a. l r.ka b. KP
boy.M.Sg.Nom
NP
N
l r.ka
Recall that another type of bare nominal exists in Urdu/Hindi: locatives as in (33).
Since these distribute like the overtly case marked nominals, we again assume a KP in
which the K head is not projected. Again, the locative case and the feature structure
associated with it is associated with the KP via default rules (section 5.2.).
(33) adnan d.akxane/zu g -ya h

Adnan.M.Sg.Nom post office.M.Sg.Obl/zoo.M.Sg.Obl go-Perf.M.Sg be.Pres.3.Sg
‘Adnan has gone to the post office/zoo.’
(34) a. zu b. KP c. d.akxane d. KP
zoo.M.Sg.Obl post office.M.Sg.Obl
NP NP
N N
zu d.akxane
We now need to account for the presence of the oblique inflection. In this instance,
we analyze the oblique e as a case marker that is a bound morpheme whose surface re-
alization is governed by the morphophonological properties of the nominal (e.g., mas-
culine nouns ending in a overtly realize this morpheme, as in (34c)).
Recall that Urdu/Hindi verbs agree with either nominative subjects or objects (in
that order), or default to masculine agreement if there is no nominative argument. One
way of capturing this pattern would have been to distinguish nominatives and non-
nominatives in terms of NP vs. KP and then agree only with NPs. However, for this
analysis to be convincing, there would have to be more distributional properties that
differentiate nominatives from non-nominatives than those that unite them. This is not
the case. In addition, the bare locative would be assumed to project to a KP while the
bare nominative would not. There is no good reason to assume this. We therefore do
not link agreement with the structure of case and for the moment leave aside a treat-
ment of agreement in Urdu/Hindi.
The point that agreement and case are not as intimately related as often assumed is
made particularly well in Subbarao’s (1999) typological examination of agreement in
South Asian languages. Subbarao demonstrates that agreement is sensitive to a variety
of factors, and that no coherent generalization can be formulated based on the interac-
tion between agreement and case as necessitated by an analysis of movement for case
or -features. In South Asian languages, case marking is just one factor among many
and is not always relevant for agreement.
In this section we provide an overview of the LFG architecture as relevant to case as-
signment. Section 6 provides analyses of particular cases, focusing on the ergative and
dative.
In LFG, information from different components combines to produce a consistent
and coherent analysis. The different modules of grammar (e.g., grammatical functions,
semantics, and phonological information) are encoded in terms of projections from lex-
ical entries and phrase structure rules, which in turn encode syntactic and morpholog-
ical constituency. This is informally illustrated in (36) for (35). A sentence like (35)
has two syntactic structures associated with it. The first is a phrase structure tree, re-
ferred to as the c(onstituent)-structure. LFG avoids the use of traces. The c-structure
therefore closely reflects the actual string and contains a faithful representation of lin-
ear order and constituency information. The grammatical functions are encoded in the
f(unctional)-structure as an attribute value matrix (AVM).
(35) ram kh ãs-a

‘Ram coughed.’
(36) a. Constituent-structure: b. Functional-structure:

S
PRED cough SUBJ
KP V
TNS - ASP TENSE PAST
kh ãs-a
ram PRED Ram
CASE NOM
SUBJ
PERS 3
NUM SG
In LFG morphological and syntactic elements are given distinct c-structure repre-
sentations. In contrast, the f-structure information can be contributed by bound mor-
phemes as well as by independent lexical items. As such, the f-structure may be iden-
tical whether the case markers are clitics, affixes, or null in c-structure. However, the
source of this information will differ: from the clitic itself for clitics; from the affixed
noun for affixes; from default rules for null markers (section 5.2.).
A priori the role of case marking can be handled in various ways in LFG. For exam-
ple, the architecture does not presuppose an intimate connection between case, agree-
ment and structural position, although this would be one possibility. In the follow-
ing, we present our view of case marking in terms of three distinct uses of case: struc-
tural, semantic, and quirky. We assume these types of case marking to be universally
available, but do not assume that every language will make use of each possibility.
Urdu/Hindi uses an interaction of all three types case assignment, while English spe-
cializes in structural case by position.
The association of grammatical functions with thematic roles is handled via a flexible,
yet constrained theory of Mapping (see Bresnan and Zaenen 1990 for an overview).
Arguments of a predicate are specified in the lexicon with the features [ r(estricted)]
and/or [ o(objective)]. Roughly, patient-like roles are [ r], secondary patient-like
roles are [+o], and other roles are [ o]. For example, the a(rgument)-structure of the
English verb pound would look as in (37). These specifications constrain the way ar-
guments are associated with grammatical functions, which are also classified by means
of these features as shown in (38).
(37) a-structure pound ag pt

[ o] [ r]
(38) Gram. Functions Features Gram. Functions Features

SUBJ [ r, o] OBL [+r, o]
OBJ [ r, +o] OBJ [+r, +o]
The intrinsic role classifications of the argument structure are related to the fully spec-
ified grammatical functions by mapping principles (not discussed here), as in (39).
(39) a-structure pound ag pt

[ o] [ r]
f-structure SUBJ OBJ
For the purposes of this paper, the Mapping Theory can be viewed as analagous to
Structural Case in Chomskyan approaches. However, as the identification of grammat-
ical functions is not necessarily tied to position, Mapping Theory can easily deal with
free word order languages like Urdu/Hindi.
The role of case marking with respect to Mapping Theory can vary from language
to language. In some languages, case marking and grammatical functions may be one-
to-one and onto (English). In languages like Urdu/Hindi, case marking interacts with
Mapping Theory in that it provides information that filters out possible linkings. This
is shown in section 6. However, we first discuss the three types of case marking that
we take to be universally available.
Structural case involves case assigned on the basis of syntactic information. It is usu-
ally correlated with grammatical function. It may also be associated with phrase struc-
ture position.
Structural case is often an instance of default case and hence functions as the Else-
where Case (cf. Zaenen, Maling, and Thraı́nsson’s (1985) notion of default vs. lexi-
cally stipulated case). For languages which require that all NPs have case, this can be
stated as in (40a), analagous to the Case Filter (Rouveret and Vergnaud 1980, Chomsky
1980).
In Urdu/Hindi the nominative is a default case. As such, there are principles which
assign nominative case to subjects and objects, as in (40b,c). In languages in which
all subjects have nominative case, (40b) would be obligatory; in languages such as
Urdu/Hindi, in which there are non-nominative subjects, the default principles are op-
tional and only apply if nothing else assigns case to the subject.
(40) a. Wellformedness principle: KP : ( CASE )
b. Default: (( SUBJ CASE )= NOM)
c. Default: (( OBJ CASE )= NOM)
While the identification of grammatical functions is not necessarily tied to posi-

tional information within Mapping Theory, some languages may restrict a position to
a particular case-marked grammatical function. Thus, non-thematic grammatical re-
lations and their corresponding case marking may be licensed by structural position
(King 1995). We have not found an example of positional case in Urdu/Hindi.
We take semantic case to be the most general type of case marking in Urdu/Hindi. The
defining characteristics of semantic case are: (i) predictability via the formulation of
generalizations across predicates and constructions; (ii) a subjection to syntactic re-
strictions (such as only appearing on certain grammatical functions). The association
of case morphology with grammatical functions can be restricted by the case markers
themselves. This is in line with Nordlinger (1998), who proposes the notion of con-
structive case for Australian languages whereby the case morphology provides infor-
mation as to grammatical relations.
For example, the entry for the Urdu/Hindi ergative ne would specify that it can only
appear on subjects and that when it appears with transitive perfects, it has no semantic
entailments. On the other hand, with intransitive perfects or infinitives, some form of
conscious control over the action is entailed. This syntactic and semantic information
is part of the lexical entry of the case marker (see section 6.1. for a concrete example)
and must be consonant with other information, such as the argument structure (Map-
ping Theory) and the verbal morphology. This differs from the Chomskyan treatment
of morphological case, in which the case marker is an overt spell-out of features deter-
mined independently by the syntax or lexical properties of the verb.
Finally, quirky case is used only when there is no regularity to be captured: the case
assignment is truly exceptional to the system. For example, consider the Urdu/Hindi
transitive verb la ‘bring’ in (41). The subject should be ergative since this is a perfect
transitive verb. However, it is nominative; this requirement must be stipulated in the
lexical entry, as in (42).
(41) nadya kıtab la-yi
Nadya.F.Sg.Nom book.F.Sg.Nom bring-Perf.F.Sg
‘Nadya brought a book.’
(42) la ‘bring’ ( PRED )= ag[ o] th[ r]

( SUBJ CASE ) = NOM
Our notion of quirky case is extremely restricted. Quirky case only occurs when no
generalizations can be made about the choice of case with the predicate in question.
Our approach thus contrasts with Zaenen, Maling, and Thrainsson (1985) and Wool-
ford (1997), where semantic case is treated as lexically inherent or quirky case. Our
approach also contrasts with that of Wunderlich and Lakämper (1999), who employ a
notion of semantic case, but who take it to encompass the case markers and preposi-
tions that mark oblique (non-core) arguments and adjuncts. Structural case for them
identifies the core grammatical relations subject, object, and indirect object, while se-
mantic case supplements the structurally case marked arguments. In our approach, in
contrast, core grammatical relations are identified via Mapping Theory and semantic
case interacts with these core grammatical relations.
In this section we focus on the ergative/nominative and ergative/dative case alterna-

tions on subjects to show how our view of case interacts with the analysis of a clause.
A crucial part of our account is that the case markers contribute to the syntactic and
semantic analysis of the clause. That is, the ergative in Urdu/Hindi cannot be analyzed
as a case that classifies a language in terms of how it treats it grammatical functions: it
is a case marker that is confined to a certain grammatical function (subject), like other
case markers, and it has semantic import when viewed in alternation with other cases,
as is also the case for datives, accusatives, and instrumentals in Urdu/Hindi.
The entries for the Urdu/Hindi ergative ne and the dative/accusative ko can be encoded
as in (43).
(43) a. ne b. ko
( CASE ) ERG [ ( CASE ) ACC
(SUBJ ) (OBJ )
[ ( SEM - PROP CONTROL ) = INT ( SEM - PROP SPECIFIC)
((SUBJ ) OBJ) ( CASE ) DAT

((SUBJ ) VFORM) PERF ] (OBJ ) (SUBJ )
( SEM - PROP CONTROL ) ]
The entry for the ergative states that a subject is required (( SUBJ )). This subject can
either have the semantic property of internal control (( SEM - PROP CONTROL) = INT),
or if there is an object in the clause ((( SUBJ ) OBJ)) and the form of the verb is perfect
((( SUBJ ) VFORM) PERF) (i.e., in the case of a transitive perfect), the semantic en-
tailments are left open. The abstract feature CONTROL INT stands for the connection
of the ergative with volitionality and takes into account the more complex picture of
ergative usage as described by Bashir (1999) (section 2.1.). The skeletal f-structures
resulting from the entry for ne are shown in (44).
(44) a. b.
SUBJ CASE ERG CASE ERG
SUBJ
SEM - PROP CONTROL INT
OBJ
VFORM PERF
Thus, the option in which the ergative is structurally required in transitive verbs with
perfect morphology is semantically unmarked. When it is not structurally required, as
with the intransitives in (3) or the infinitive construction in (1), the ergative entails that
the subject has control over the action. The equations ensuring only a compatibility
with intransitive perfects and infinitives have been left out in the interests of readability.
This split in usage is modeled by the disjunction ( ) in the entry of the ergative case.
Similarly, ko encodes a disjunction between its use as an accusative object marker

(( OBJ )), where it marks specificity (( SEM - PROP SPECIFIC) ), and its use as a
dative, where it marks goals and experiencer subjects (( OBJ ) (SUBJ )) or
modality with respect to the infinitive constructions in (1). The corresponding skele-
tal f-structures are shown in (45).
(45) a.
CASE ACC
OBJ
SEM - PROP SPECIFIC +
b.
CASE DAT
OBJ
SEM - PROP [ CONTROL ]
c.
CASE DAT
SUBJ
SEM - PROP [ CONTROL ]
This use of features to indicate the semantic contribution of the case marker is inten-
tionally rough. The semantics of the KP is not determined by the case marker alone, but
must be evaluated within the context of the utterance and other semantic information in
the clause, such as the lexical semantics of the verb, the temporal and aspectual dimen-
sions, etc. This type of precise semantic evaluation is not the domain of the syntax or
the morphology. However, the syntax and the morphology must provide relevant infor-
mation to the semantics. As such, the abstract features encoded in the f-structures can
be viewed as instructions to the semantic component. For example, the feature value
CONTROL EXTERNAL is a signal that the KP may be subject to external control: the
precise modality of this is determined by the semantics. In the remaining sections
we demonstrate our analysis on various case alternations in Urdu/Hindi.
Consider (46) which shows an alternation with respect to volitionality.
(46) a. ram kh ãs-a

‘Ram coughed.’
b. ram=ne kh ãs-a
Ram.M.Sg=Erg cough-Perf.M.Sg
‘Ram coughed (purposefully).’
Simplified lexical entries for the subject and verb are shown in (47) (e.g., -features
are omitted). Mapping Theory associates the [ o] agent with a SUBJ.
(47) ram N kh ãs V -a

( PRED)=‘Ram’ ( PRED)=‘cough ag[ o] ’ ( TNS - ASP TENSE ) = PAST
( VFORM) = PERF
The entry for the ergative ne given in (43a) allows for the two possibilities in (44).
However, only the possibility in (44b) will be well formed for (46b), as (44a) requires
the presence of an object. The only well formed f-structure resulting from the combi-
nation of the noun ram with the ergative ne is as shown in (48). It has the desired effect
that ergative subjects of intransitives entail volitionality.
(48)
PRED Ram
SUBJ CASE ERG
Thus, the case marker itself assigns ergative case to its head noun. In addition, it
forces its head noun to be a subject in the f-structure and to be compatible with the
given semantic interpretation. If it is incompatible with the other requirements listed
in the lexical entry, the result is an ill-formed structure.
The entry for kh ãs-a ‘cough’ does not specify the ergative in its lexical entry. It is
therefore free to occur with a nominative subject as well, as in (46a). This nominative
is assigned by default ((40b)).
Now consider the alternation ergative/nominative in transitive sentences, exempli-

fied by (49). The ergative is required when the verbal morphology is perfect.
(49) a. ram gar.i c la-ta (h )

Ram.M.Sg.Nom car.M.Sg.Nom drive-Impf.M.Sg be.Pres.3.Sg
‘Ram drives a car.’
b. ram=ne gar.i c la-yi (h )
Ram.M.Sg=Erg car.M.Sg.Nom drive-Perf.M.F.Sg be.Pres.3.Sg
‘Ram has driven a/the car.’
With perfect transitive verbs, only the second disjunct of the ergative entry in (43)
comes into play, i.e., in the f-structure in (44a). This disjunct does not include any
information about internal or external control, so the semantics of (49b) are left under-
specified. The ergative cannot appear in (49a) because the verbal form is not of the
right kind for either of the disjuncts to apply (the first disjunct is understood to be con-
strained to apply only to perfects and infinitives).
Next consider the ergative/dative alternation in the infinitival construction in (50).


We analyze this construction as an instance of functional control: the verb ‘be’ acts as
a modal which predicates an event of the subject. The subject of the modal controls the
PRO subject of the embedded clause. The entry for this form of the verb ‘be’ is given
in (51). Again, the verb makes no direct specifications as to case. It does, however,
require that some notion of CONTROL be involved. Thus, dative and ergative subjects
are compatible with this entry, but nothing else.
(51) h ( PRED )= be ag/go[ o],Ev[ r]

( SUBJ SEM - PROP CONTROL )
( SUBJ)=( XCOMP SUBJ)
( TNS - ASP TENSE ) = PRES
( XCOMP VFORM ) = INF
With respect to the disjunction in the entry of the ergative marker in (43), only one
of the possibilities is viable: because the VFORM of the embedded XCOMP is INF, and
because the lexical entry of the verb requires the feature SEM - PROP CONTROL, only the
first disjunct applies (i.e, the f-structure in (44b)). The resulting f-structure analysis
is shown in (52).
(52)
PRED be ,
TNS - ASP TENSE PRES
PRED Nadya
SUBJ CASE ERG
PRED go
XCOMP SUBJ
VFORM INF
Now consider the dative version in (50b). The entry for the dative case in (43b) al-
lows for the three cases in (45). The disjunctions for the marking of object (( OBJ )) and
indirect object (( OBJ )) will not result in a well formed analysis for (50) as the argu-
ment structure of h can only be linked to a subject and a clausal complement by the
mapping theory. The only viable option is a dative subject (( SUBJ )) with the feature
SEM - PROP CONTROL . This feature in the f-structure in (53) tells the semantic compo-
nent that there is something to be said about the control the subject had over the action.
Context-dependently this can give rise to a modal interpretation such as ‘must’ (ex-
ternal control), as in (49b), but it need not.
(53)
PRED be ,
TNS - ASP TENSE PRES
PRED Nadya
CASE DAT
SUBJ
SEM - PROP CONTROL
PRED go
XCOMP SUBJ
VFORM INF
Recall from Table 3 and section 2.1. that Bashir (1999) found that the use of ko in
this construction is less marked than the use of the ergative, in the sense that subjects
marked with ko could also be used when the subject has control over the action, while
the ergative is restricted to subjects which have control over the action. We therefore
make a difference between the amount of information postulated. In the ergative use,
the value of the feature SEM - PROP CONTROL is supplied, in the dative use, only the ex-
istence of the feature is asserted: the value is left open. Again, the dative case marker
itself ensures that it appears on a subject and that it contributes the information as to the
existence of the CONTROL feature. For analyses of other semantic alternations within
this approach, see Butt and King (2001).
In conclusion, we have argued that case marking has both syntactic and semantic re-
flexes whose diversity reflects distinctions across verbal classes and constructions. We
have presented a view of case which allows case morphology to play an active role
in contributing to the grammatical analysis of a clause via a notion of semantic case.
In addition, we posit structural and quirky case as further case marking mechanisms
which are universally available, but do not require that every language make use of
them. In our analysis of the Urdu/Hindi case system, all three types of case assignment
work together to characterize the many invidual case usages and case alternations.
With regard to the morphosyntactic properties of the Urdu/Hindi case markers, we
established that they are clitics: a consideration of both synchronic and diachronic facts
points to a distinction between the oblique affix, the case clitics, and the postpositions.
These distinctions have been analyzed as Layer I, II, and III, respectively by Masica
(1991). We incorporate the three-way distinction into our formal analysis in terms of
oblique marking, case clitics, and postpositions.
A careful examination of the synchronic uses of each of the case markers also showed
that the perception of Urdu/Hindi as a tense/aspect split-ergative language is an over-
simplification. Instead, all the Urdu/Hindi case markers, including the ergative, func-
tion on a par as semantic cases. The ergative, for example, is restricted to subjects and
signals that the subject has control over a given action in contexts where it is not struc-
turally required, e.g., infinitives and perfect intransitives. The dative is restricted to
subjects and indirect objects and is associated with goals/experiencers. This associa-
tion gives rise to a modal interpretation in constructions like the infinitive + ‘be’ one.
The ergative/dative alternation in these contexts indicates that the dative and ergative
are in a closer structural relationship than previously acknowledged, though historical
discussions have postulated a slippery slope between ergative and dative case. In our
analysis the close relationship between the dative and ergative is rendered by the idea
that both the ergative and the dative contribute the feature CONTROL to the semantic
component, but that they differ in the realization of this feature.
Miriam Butt University of Manchester Institute for Science and Technology (UMIST)
Tracy Holloway King Palo Alto Research Center (PARC)
1. We would like to thank the audience of the Utrecht Workshop on Case, the Graduiertenkollegskol-
loquium in Stuttgart, and the members of the seminar on Case (Konstanz, Fall of 1998), where parts of this
paper were presented. We are grateful to Alice Davison and several anonymous reviewers who commented
on earlier versions of the paper. We would also like to thank Ashwini Deo and Helen de Hoop for interesting
comments and discussion. Miriam Butt’s contribution to this paper was made possible by financial support
from the DFG (the German Science Foundation) via the SFB 471 at the University of Konstanz.
2. Due to typological work undertaken as part of colonial exploration in the 19th century (e.g., Ray
and Haddon 1893), it was recognized that a number of languages used a separate case for the marking of
“agentive nominatives”. Well known examples included Greenlandic and Basque (e.g., Pott 1873). Due to
a series of misunderstandings, this agentive nominative eventually came to be known as the “ergative” (cf.
Manaster Ramer 1994).
3. The South Asian languages Urdu and Hindi are closely related. Both are among the official languages
of India and are spoken primarily in the north of India. Urdu is the national language of Pakistan. The data
presented in this paper are drawn primarily from the dialect of Urdu spoken in Lahore, Pakistan.
4. For a more complete typology of case patterns see Plank (1995). However, even this more complete
typology does not do justice to the case alternations found in Urdu/Hindi as discussed in this paper.
5. We do not include a discussion of the locative: for its synchronic uses see Mohanan (1994); for its
origin see Kellogg (1893:132–133) and Beames (1872:292–296). In terms of morphology, the oblique mark-
ing appears on singular nouns ending in -a (-e) and on all plural nouns (Glassman 1976, Masica 1991:240).
See section 3.1. and 4 for more discussion.
6. This also applies to Bobaljik (1993).
7. This has the effect that the noun phrases which only have structural Case are exactly the nominative
arguments, whereby nominative in Urdu is phonologically null (Table 2).
8. See Butt (1998) for an analysis of the causative patterns in terms of aspectual affectedness.
9. See Butt (1997) for an analysis of this construction as a complex predicate with dispositional seman-
tics whose subject is an instrumental NP . Bhatt (1998) proposes an alternative analysis in terms of negative
polarity. Davison (1990) examines this construction within a larger discussion on “peculiar passives”.
10. The fact that the genitive inflects can be traced directly to its historical origin. After a fierce debate
in the last century, the view espoused by Hoernle won out and was taken over by Beames (1872:285) and
Kellogg (1893:129). Under this view, the genitive is analyzed as having arisen from kr.ita ‘done by’, the
Sanskrit past participle of kr.i ‘do’ as follows. Sanskrit kr.ita Prakrit kerita keriai modern Urdu/Hindi
k-. The original participal inflected for agreement and the genitive case marker has not lost this property.
11. We thank Ashwini Deo for help with the glossing and transcription of these examples.
12. This case marker is also similar to the dative suffix in Dravidian languages: Telugu and Tamil -ku,
Malayalam -kku, Old Kannada -ke or -kke, Tulu -ku or -gu (Sahoo 2001:38). As such, further investigation
is called for.
13. One of the reviewers asserts that a discussion of the status of case is incomplete without address-
ing the issue of pronouns. We had not included a discussion of pronouns in an earlier version of this paper
because the Urdu/Hindi pronoun system is the result of several differing historical developments, as is gen-
erally the case crosslinguistically.
14. The locative p r ‘on’ represents the class of locative case markers here.
15. No mention is made of t k ‘towards’.
16. The -hõ affix only appears on examples with the ergative in Dasgupta et al. (2000:128). A reviewer
concurs with these judgements. However, Beg (1988:157–159) and Kellogg (1983:174–175) report that this
form is used with all oblique non-nominative forms. This is in accordance with Lahori Urdu. A dialectal
difference thus seems to characterize the use of pronoun forms in Urdu/Hindi.
17. For example, the first person singular is a continuation of the Sanskrit instrumental singular first per-
son pronoun mayā. This form has been reanalyzed as a direct form in Urdu/Hindi, but not in Punjabi,, where
the m˜ is treated as an oblique, as seen by the object agreement and the fact that third person nouns must be
marked with the ergative.
(i) mẽ/ dnan=ne kıtab vekh -i
I.Obl/Adnan.M=Erg book.F.Sg.Nom see-Perf.F.Sg.
‘I/Adnan saw the book.’ [Punjabi]
The t jh and m jh forms are continuations of old dative forms. The oblique -e inflection on these forms
signals the dative, but we do not believe that this -e is the same as the oblique -e discussed above. There is
some indication that the t jh and t jh e forms evolved independently (Beg 1988:151,156).
18. A reviewer points out that Naim’s (1975) Urdu grammar also argues for this position.
19. The focus clitic has the distribution of an adverb. As such, it has a wider syntactic distribution than
the case clitics, although it is always postposed and may not be the last element of a clause. In (i), the focus
clitic hi is shown attached to a verb.
(i) nadya=ne ta kh el-a=hi h
Nadya.F.Sg=Erg cards play-Perf.M.Sg=also be.Pres.3.Sg
‘Nadya has only played cards.’
20. Dasgupta, Ford, and Singh (2000:130) list four clitic+case marker combinations as ungrammati-
cal: mẽ=hi=ne ‘I=Foc=Erg’, tu=hi=ne ‘you=Foc=Erg’, ınhõ=hi=ne ‘they.Obl=Foc=Erg’, and nhõ=hi=ne
‘they.Obl=Foc=Erg’. This leads them to conclude that ne must be treated as an inflectional affix. However,
Kellogg (1893:176) lists nhĩ=hi=ne ‘they.Obl=Foc=Erg’ as a possible form. This evidence, taken with the
examples above, indicates that ne is not an affix. Another argument from Dasgupta, Ford, and Singh for
the affix status of ne with respect to pronouns comes from (Hindi) orthography, where the ne is sometimes
written as one word together with the pronoun. However, it is well known that orthography provides at best
a weak argument for the morphosyntactic status of an item. Consider, for example, the German spelling
reform, by which radfahren ‘bicycle-drive’ is now spelled Rad fahren, though the status of Rad ‘wheel’ it-
self has not undergone any morphosyntactic or prosodic change since the spelling reform. Another example
comes from Lahiri’s (2000) work on the Bengali progressive and perfect. Both of these are formed with a
version of the auxiliary ‘be’, and in both the progressive and the perfect the main verb in combination with
the ‘be’ is spelled as one word in the Bengali script. Howerver, Lahiri shows that that the ‘be’ in the perfect
has the status of a clitic, while the progressive ‘be’ has been reanalyzed as an affix. We therefore conclude
that the ne cannot be analyzed as an affix. However, work remains to be done on the compatibility of pro-
nouns with clitics, as Platts (1967:301) mentions certain combinations which are not felicitous: mẽ=hi=ne
‘I=Foc=Erg’, mera=hi ‘I.Gen=Foc’, tera=hi ‘you.Gen=Foc’. One of the reviewers concurs with Dasgupta
et al.’s judgements, again indicating dialectal variation.
21. Thanks to Rami Nair (p.c.) for discussion on this point. A reviewer notes that evidence from stress
is problematic as the Hindi stress system itself is highly problematic. This criticism is warranted. Hayes
(1990:162–167) represents the most useful and lucid summary of the known facts. We present the stress
evidence here as suggestive when taken together with the evidence from coordination and clitic placement.
22. A reviewer points to the contrast between m kan ke pas ‘house.M.Sg Gen.Obl side’ vs. m kan ki or
‘house.M.Sg Gen.F.Sg side’, which is discussed by Dasgupta, Ford, and Singh (2000:125). Here ke pas is
invariant, but the genitive in ki or shows feminine agreement with or. The contrast is taken as evidence that
postpositions are not always invariant. However, we analyze ke pas as a postposition, but m kan ki or as a
standard Noun-Genitive Noun construction in which agreement with the head noun is expected.
23. Due to lack of space, we do not put forth a detailed analysis of the internal structure of Urdu/Hindi
nominals. For a detailed description of Hindi data, see Verma (1971). Of particular interest to the data dis-
cussed here is the position of Adjectival Phrases in the structure; as in many languages, AP s show agreement
patterns with their head nouns in Urdu/Hindi.
24. This also applies to Dasgupta, Ford, and Singh’s (2000) idea of a little p, which may encode inflec-
tional or periphrastic case markers. Little p is used to check oblique case features, much as K does in other
approaches. The terminology of little p is meant to be indicative of the slippery slope between postpositions
and case markers. We sympathize with this concern, but see no gain in adopting the idea of little p.
25. We would like to thank a reviewer for suggesting this structure which allows for coordination data
such as that in (i).
(i) asım=ke pıtta ji or amir=ke dada ji ne
Asim.M.Sg=Gen.Obl father Resp and Amir.M.Sg=Gen.Obl grandfather Resp Erg
‘Asim’s father and Amir’s grandfather’
26. The basic architecture we assume is diagrammed in (i) (based on Butt and King 1998 and Butt, Dal-
rymple, and Frank 1997). The projections are defined in terms of mathematical functions for which inverse
functions are also available; this means that any level of representation can be related to any other level. The
lines represent the most direct connections.
(i) c(onstituent)-structure
argument-str discourse-str prosodic-str
f(unctional)-str phonology
semantic-str
semantics
27. LFG allows for functional projections and X syntax; we use S here for expository purposes. See
Bresnan 2001 on constraints on LFG phrase structure rules.
28. This a-structure can be conceived of as an attribute-value matrix (Butt 1998).
29. This avoids the problems encountered by structural definitions of argument structure (i.e., equating
deep structure with argument structure) (Rappaport 1983).
30. A canonical example is the assignment of adnominal genitive in English. (See footnote 31 on the
notation.)
(i) English Adnominal Genitives (simplified structure)
NP
NP N
( ADJUNCT )= =
( CASE )=GEN hat
Boris’s
31. We do not discuss the details of the LFG formalism here; these can be found in Bresnan (2001) and
references therein. Basically, the up arrows ( ) encode mappings between nodes of the constituent-structure
tree and the functional-structure. The ‘ ’ refers to the particular atribute-valume matrix (AVM) that the phrase
structure node in question corresponds to. So, in the examples in this paper, the ‘ ’ usually refers to the
functional-structure of the noun phrase containing the case marker. For example, in (47) the ‘ ’ refers to the
AVM with PRED ‘Ram’ in it in (48); thus, the first line of (43) states that this part of the functional-structure
contains the pair CASE ERG, as is seen in (48), while the second states that this part of the functional-structure
is contained within the SUBJ of the next bigger AVM, as is also seen in (48). The notation (SUBJ ) is an
instance of inside-out functional uncertainty.
32. The features INT vs. EXTERNAL are proposed in analogy to the distinction made between internal
vs. external causation by Levin and Rappaport Hovav (1995).
33. SUBJ is an abbreviation used for expository purposes. The SUBJ information will be realized as
an f-structure attribute, while the exp information is relevant to the argument-structure which is not shown
here (section 5.1.).
34. Our representations would not be crucially different if instead encoded the accusative-dative split
by having two separate entries for ko.
35. Similarly, the SPECIFIC + in the entry of the accusative is an indication of the semantic factors that
must be considered. The precise semantic interpretation of accusative ko, especially as contrasted with the
use of nominative objects, remains to be explored. See Dayal (1999) for discussion.
36. The locative oblique ‘zoo’, not shown in the f-structure, functions like the ergative and dative case
markers in that it is associated with information specifying that it must be an OBL , etc.
37. Event arguments link to an XCOMP or COMP ; see Butt (1995) for discussion.
38. There are some further interesting cases in which a nominative appears with an infinitive construc-
tion: t fan ane th e ‘storms were going to come’. As the translation indicates, these examples involve a tem-
poral interpretation. We would posit a different analysis for these constructions and embed that analysis in a
larger picture of case marked infinitives and the use of vala ‘one’ on infinitives to express temporal relations.
39. For example, Beames (1872) and Kellogg (1893) speculate that the origin of the ergative may be
traced to a related dative subject construction and Bashir (1999) proposes that the ergative is slowly en-
croaching on the domain of the dative.
Allen, W.S. 1951. A Study in the Analysis of Hindi Sentence-Structure. Acta Linguistica Hafniensia 6:68–
86.
Anderson, Stephen R. 1977. On Mechanisms by which Languages become Ergative. In Mechanisms of Lan-
guage Change, ed. Charles Li. 317–363. Austin, Texas: University of Texas Press.
Bashir, Elena. 1999. The Urdu and Hindi Ergative Postposition ne: Its Changing Role in the Grammar. In
The Yearbook of South Asian Languages and Linguistics, ed. Rajendra Singh. 11–36. New Delhi: Sage
Publications.
Bayer, Josef, Markus Bader, and Michael Meng. 2001. Morphological Underspecification Meets Oblique
Case: Syntactic And Processing Effects in German. Lingua 111:465–514.
Beames, John. 1872–79. A Comparative Grammar of the Modern Aryan Languages of India. Delhi: Mun-
shiram Manoharlal. Republished 1966.
Beg, Mirza Khalil A. 1988. Urdu Grammar: History and Structure. New Delhi: Bahri Publications.
Bhatt, Rajesh. 1998. Complex Predicates as Polarity Items. Talk presented at the Complex Predicates Work-
shop at the Annual Meeting of the DGfS, Halle, March.
Bittner, Maria, and Ken Hale. 1996. The Structural Determination of Case and Agreement. Linguistic Inquiry
27(1):1–68.
Bobaljik, Jonathan D. 1993. On Ergativity and Ergative Unergatives. Papers in Case and Agreement II, MIT
Working Papers in Linguistics Volume 19:45–88.
Bok-Bennema, Reineke. 1991. Case and Agreement in Inuit. Foris: Dordrecht.
Bresnan, Joan. 2001. Lexical-Functional Syntax. Cambridge, Massachusetts: Blackwell.
Bresnan, Joan, and Sam Mchombo. 1995. The Lexical Integrity Principle: Evidence from Bantu. Natural
Language and Linguistic Theory 13(2):181–254.
Bresnan, Joan, and Annie Zaenen. 1990. Deep Unaccusativity in LFG. In Grammatical Relations: A Cross-
Theoretical Perspective, ed. Katarzyna Dziwirek, Patrick Farrel, and Errapel Mejias-Bekandi. 45–57.
Stanford, California: CSLI Publications.
Butt, Miriam. 1993. Object Specificity and Agreement in Hindi/Urdu. In Papers from the 29th Regional
Meeting of the Chicgo Linguistic Society, 80–103.
Butt, Miriam. 1995. The Structure of Complex Predicates in Urdu. Stanford, California: CSLI Publications.
Butt, Miriam. 1997. Aspectual Complex Predicates, Passives and Disposition/Ability. Talk held at the Lin-
guistics Association of Great Britain, Spring Meeting, April. Extensive handout available at
http://ling.uni-konstanz.de/pages/home/butt.
Butt, Miriam. 1998. Constraining Argument Merger through Aspect. In Syntax and Semantics: Complex
Predicates in Nonderivational Syntax, Volume 30, ed. Erhard Hinrichs, Andreas Kathol, and Tsuneko
Nakazawa. 73–113. New York, New York: Academic Press.
Butt, Miriam. 2001a. A Reexamination of the Accusative to Ergative Shift in Indo-Aryan. In Time over Mat-
ter: Diachronic Perspectives on Morphosyntax, ed. Miriam Butt and Tracy Holloway King. 105–141.
Butt, Miriam. 2001b. Case, Agreement, Pronoun Incorporation and Pro-Drop in South Asian Languages.
Talk given at the Workshop The Role of Agreement in Argument Structure, Utrecht. http://ling.uni-
konstanz.de/pages/home/butt.
Butt, Miriam, Mary Dalrymple, and Anette Frank. 1997. An Architecture for Linking Theory in LFG. In
On-line Proceedings of the LFG97 Conference, San Diego, ed. Miriam Butt and Tracy Holloway King.
http://csli-publications.stanford.edu/LFG2/lfg97.html.
Butt, Miriam, and Tracy Holloway King. 1991. Semantic Case in Urdu. In Papers from the 27th Regional
Meeting of the Chicago Linguistic Society, ed. Lisa Dobrin, Lynn Nichols, and Rosa M. Rodriguez, 31–
45.
Butt, Miriam, and Tracy Holloway King. 1998. Interfacing Phonology with LFG. In Proceedings of the
LFG98 Conference, ed. Miriam Butt and Tracy Holloway King. Stanford, California. CSLI Publications.
http://csli-publications.stanford.edu/LFG3/lfg98.html.
Butt, Miriam, and Tracy Holloway King. 2001. Case Systems: Beyond Structural Distinctions. In New Per-
spectives on Case Theory, ed. Ellen Brandner and Heike Zinsmeister. Stanford, California: CSLI Publi-
cations. To Appear.
Butt, Miriam, Tracy Holloway King, Marı́a Eugenia Niño, and Frédérique Segond. 1999. A Grammar Writer’s
Cookbook. Stanford, California: CSLI Publications.
Chatterji, Suniti Kumar. 1926. The Origin and Development of the Bengali Literature, Volume II. Calcutta:
D. Mehra, Rupa & Co. 1975 edition.
Chomsky, Noam. 1980. On Binding. Linguistic Inquiry 11:1–46.
Chomsky, Noam. 1981. Lectures on Government and Binding. Foris: Dordrecht.
Chomsky, Noam. 1995. The Minimalist Program. Cambridge, Massachusetts: The MIT Press.
Dasgupta, Probal, Alan Ford, and Rajendra Singh. 2000. After Etymology: Towards a Substantivist Linguis-
tics. München: LINCOM EUROPA.
Davison, Alice. 1988. Constituent Structure and the Realization of Agreement Features. In Papers from the
24th Annual Regional Meeting of the Chicago Linguistic Society, Part Two: Parasession on Agreement
in Grammatical Theory, ed. Diane Brentari, Gary Larson, and Lynn MacLeod, 41–55.
Davison, Alice. 1990. Peculiar Passives. Language 56(1):42–66.
Davison, Alice. 1991a. Feature Percolation and Agreement in Hindi-Urdu. Presented at the South Asian Con-
ference, University of Wisconsin, November. Unpublished ms., University of Iowa.
Davison, Alice. 1991b. Finiteness and Case in Hindi-Urdu Complements. Paper presented at the Eleventh
South Asian Language Analysis Roundtable, University of Illinois, May. Unpublished ms., University
of Iowa.
Davison, Alice. 1998. Structural Case, Lexical Case and Complex Predicates. South Asian Language Review
8.
Davison, Alice. 1999. Ergativity: Functional and Formal Issues. In Functionalism and Formalism in Linguis-
tics, Volume I: General Papers, ed. Michael Darnell, Edith Moravcsik, Frederick Newmeyer, Michael
Noonan, and Kathleen Wheatley. 177–208. Amsterdam: John Benjamins.
Dayal, Veneeta. 1999. Bare NPs, Reference to Kinds, and Incorporation. In Proceedings of SALT IX, ed.
Tanya Matthews and Devon Strolovitch. 34–51. Ithaca, New York: Cornell University.
de Hoop, Helen. 1992. Case Configuration and Noun Phrase Interpretation. Doctoral dissertation, Rijksuni-
versiteit Groningen.
de Hoop, Helen. 1999. Optimal Case Assignment. In Linguistics in the Netherlands, ed. Renée van Bezooijen
and René Kager. 97–109. Amsterdam: John Benjamins Publishing Company. AVT Publications 16.
Dixon, R.M.W. 1994. Ergativity. Cambridge: Cambridge University Press.
Enç, Mürvet. 1991. The Semantics of Specificity. Linguistic Inquiry 22(1):1–25.
Fillmore, Charles J. 1968. The Case for Case. In Universals of Linguistic Theory, ed. Emmon Bach and R.T.
Harms. 1–88. New York, New York: Holt, Rinehart and Winston.
Franks, Steven. 1999. Clitics at the interface. In Clitic Phenomena in European Languages, ed. Marcel den
Dikken and F. Benkema. Amsterdam: John Benjamins. To Appear.
Glassman, Eugene H. 1976. Spoken Urdu. Lahore: Nirali Kitaben.
Hale, Kenneth, and Jay Keyser. 1993. On Argument Structure and the Lexical Expression of Syntactic Re-
lations. In The View from Building 20, ed. Kenneth Hale and Jay Keyser. 53–109. Cambridge, Mas-
sachusetts: The MIT Press.
Hayes, Bruce. 1990. Metrical Stress Theory: principles and case studies. Chicago: The University of Chicago
Press.
Kachru, Yamuna. 1978. On Ergativity in Selected South Asian Languages. Studies in the Linguistic Sciences
8(1):111–127.
Kachru, Yamuna. 1987. Ergativity, Subjecthood and Topicality in Hindi-Urdu. Lingua 71:223–238.
Kellogg, S. H. 1893. Grammar of the Hindi Language. Delhi: Munshiram Manoharlal Publishers Pvt. Ltd.
Second Edition, reprinted 1990.
King, Tracy Holloway. 1995. Configuring Topic and Focus in Russian. Stanford, California: CSLI Publica-
tions.
Kiparsky, Paul. 1998. Partitive Case and Aspect. In The Projection of Arguments: Lexical and Compositional
Factors, ed. Miriam Butt and Wilhelm Geuder. 265–308. Stanford, California: CSLI Publications.
Kiparsky, Paul. 2001. Structural Case in Finnish. Lingua 111:315–376.
Krifka, Manfred. 1992. Thematic relations as links between nominal reference and temporal constitution. In
Lexical Matters, ed. Ivan Sag and Anna Szabolcsi. 29–53. Stanford, California: CSLI Publications.
Lahiri, Aditi. 2000. Hierarchical restructuring in the creation of verbal morphology in Bengali and Germanic:
Evidence from Phonology. In Analogy, Levelling, Markedness: Principles of Change in Phonology and
Morphology, ed. Aditi Lahiri. Berlin: Mouton de Gruyter.
Lamontagne, Greg, and Lise de Mena Travis. 1986. The Case Filter and the ECP. McGill Working Papers
in Linguistics 3(2):51–75.
Levin, Beth, and Malka Rappaport Hovav. 1995. Unaccusativity. Cambridge, Massachusetts: The MIT Press.
Löbel, Elisabeth. 1994. KP/DP-Syntax: Interactions of Case-Marking with Referential and Nominal Fea-
tures. Theoretical Linguistics 20(1):37–70.
Mahajan, Anoop. 1989. Agreement and Agreement Phrases. MIT Working Papers in Linguistics 10:217–
252.
Mahajan, Anoop. 1990. The A/A-Bar Distinction and Movement Theory. Doctoral dissertation, Massachusetts
Institute of Technology.
Mahajan, Anoop. 1991. Clitic Doubling, Object Agreement and Specificity. In North Eastern Linguistic So-
ciety 21, ed. Tim Sherer, 263–277.
Manaster Ramer, Alexis. 1994. The origin of the term ‘ergative’. Sprachtypologische Universalien Forschung
(STUF) 47(3):211–214.
Manning, Christopher D. 1996. Ergativity: Argument Structure and Grammatical Relations. Stanford, Cal-
ifornia: CSLI Publications.
Masica, Colin. 1991. The Indo-Aryan Languages. Cambridge: Cambridge University Press.
McGregor, R.S. 1972. Outline of Hindi Grammar: With Exercises. Oxford: Clarendon Press.
Mohanan, Tara. 1994. Argument Structure in Hindi. Stanford, California: CSLI Publications.
Mohanan, Tara. 1995. Wordhood and Lexicality: Noun Incorporation in Hindi. Natural Language and Lin-
guistic Theory 13(1):75–134.
Naim, C.M. 1975. Introductory Urdu, Volume II. Chicago: Committee on Southern Asian Studies, University
of Chicago.
Neeleman, Ad, and Fred Weerman. 1999. Flexible Syntax: A Theory of Case and Arguments. Dordrecht:
Kluwer.
Nordlinger, Rachel. 1998. Constructive Case: Evidence from Australian Languages. Stanford, California:
CSLI Publications.
Payne, John. 1995. Inflecting Prepositions in Indic and Kashmiri. In Double Case, ed. Frans Plank. 283–298.
London: Academic Press.
Plank, Frans. 1995. Research into Syntactic Change III: Ergativity. In Syntax: An International Handbook
of Contemporary Research, ed. Joachim Jacobs, Arnim von Stechow, Wolfgang Sternefeld, and Theo
Vennenmann. 1184–1199. Berlin: Walter de Gruyter.
Platts, John T. 1967. A Grammar of the Hindustānı̄ or Urdu Language. Delhi: Munshiram Manoharlal.
Pott, A.F. 1873. Unterschied eines transitiven und intransitiven nominativs. Beiträge zur vergleichenden
Sprachforschung auf dem Gebiete der arischen, celtischen und slawischen Sprachen 7:71–94.
Ramchand, Gillian. 1997. Aspect and Predication. Oxford: Oxford University Press.
Rappaport, Malka. 1983. On the Nature of Derived Nominals. In Papers in Lexical-Functional Grammar, ed.
Lori Levin, Malka Rappaport, and Annie Zaenen. Bloomington, Indiana: Indiana University Linguistics
Club.
Ray, Sidney H., and Alfred C. Haddon. 1893. A Study of the Languages of Torres Straits with vocabularies
and grammatical notes, Part I. Proceedings of the Royal Irish Academy, Third Series II:463–616.
Rouveret, Alain, and Jean-Roger Vergnaud. 1980. Specifiying Reference to the Subject. Linguistic Inquiry
11:97–202.
Sadock, Jerrold. 1991. Autolexical Syntax. Chicago: University of Chicago Press.
Sahoo, Kalyanamalini. 2001. Oriya Verb Morphology and Complex Verb Constructions. Doctoral disserta-
tion, Norwegian University of Science and Technology.
Saksena, Anuradha. 1982. Case Marking Semantics. Lingua 56:335–343.

Sen, Subhadra Kumar. 1973. Proto-New Indo-Aryan. Calcutta: Eastern Publishers.
Sharma, Devyani. 1999. Nominal Clitics and Constructive Morphology in Hindi. In On-line Proceedings of
the LFG99 Conference, ed. Miriam Butt and Tracy Holloway King. Stanford, California: CSLI Publica-
tions. http://csli-publications.stanford.edu/LFG/4/lfg99.html.
Singh, Mona. 1994. Perfectivity, Definiteness and Specifity: A Classification of Verbal Predicates in Hindi.
Doctoral dissertation, The University of Texas at Austin.
Subbarao, K. V. 1999. Agreement in South Asian Languages. To appear in the Proceedings of the South
Asian Language Analysis Roundtable, University of Illinois at Urbana-Champaign.
Tuite, Kevin, Asif Agha, and Randolph Graczyk. 1985. Agentivity, Transitivity, and the Question of Active
Typology. In Papers from the Parasession on Causatives and Agentivity at the 21st Regional Meeting of
the Chicago Linguistic Society, ed. William Eilfort, Paul Kroeber, and Karen Peterson, 252–270.
Vainikka, Anne. 1993. The Three Structural Cases in Finnish. In Case and Other Functional Categories in
Finnish Syntax, ed. Anders Holmberg and Urpo Nikanne. 129–159. Berlin: Mouton de Gruyter.
Van Olphen, Herman. 1980. First-Year Hindi Course. Austin, Texas: Department of Oriental and African
Languages and Literatures, University of Texas, Austin.
Verkuyl, Henk. 1993. A Theory of Aspectuality. Cambridge: Cambridge University Press.
Verma, M. K., and K. P. Mohanan (ed.). 1990. Experiencer Subjects in South Asian Languages. Stanford,
California: CSLI Publications.
Verma, M.K. 1971. The Structure of the Noun Phrase in English and Hindi. Delhi: Motilal Barnasidass.
Wierzbicka, Anna. 1981. Case Marking and Human Nature. Australian Journal of Linguistics 1:43–80.
Woolford, Ellen. 1997. Four-Way Case Systems: Ergative, Nominative, Objective and Accusative. Natural
Language and Linguistic Theory 15:181–227.
Wunderlich, Dieter. 1997. Cause and the Structure of Verbs. Linguistic Inquiry 28(1):27–68.
Wunderlich, Dieter, and Renate Lakämper. 2000. On the Interaction of Structural and Semantic Case. Lin-
gua. Special issue, to appear.
Zaenen, Annie, Joan Maling, and H. Thráinsson. 1985. Case and Gramamtical Functions: The Icelandic Pas-
sive. Natural Language and Linguistic Theory 3:441–483.
ability passive, 10 Middle Indo-Aryan, 12
actor, 1 morphological, 4, 25
adposition, 14 nominative vs. absolutive, 2, 7
affectedness, 8 quirky, 3, 5, 23, 25
agentive nominative, 31 Sanskrit, 11
agreement semantic, 3, 9, 23, 25
genitive, 11 specificity, 2
Urdu/Hindi pattern, 8, 22 split-ergative, 28
agreement and case, 22 stacking, 19
argument structure, 23 structural, 3, 9, 23, 24
Australian languages, 25 typological classification, 1
Urdu accusative, 8
Basque, 31 Urdu dative, 9
Bengali Urdu genitive, 11
progressive and perfect, 33 Urdu instrumental, 10
Urdu nominative, 7
Case Urdu/Hindi paradigm, 4
inherent, 3 Case Filter, 24
strong, 6 case stacking
structural, 3 locative, 19
weak, 6 Case Theory, 3
case causative, 8
accusative system, 1 clause
alternation, 2, 26 intransitive, 1
bare locative, 21 transitive, 1
bare nominative, 21 clitic, 19
classification, 1 case, 17
clitic, 17 placement, 17
constructive, 25 test, 16
dative lexical entry, 26 comitative, 10
default, 24 complex predicate
default rules, 24 N-V, 9
ergative, 5, 13, 31 conscious choice, 6
ergative lexical entry, 25, 26 constructive case, 25
ergative origin, 14 control, 27, 29
ergative system, 1
ergative-dative alternation, 2, 5, 26, 28
ergative-nominative alternation, 26, 27 Elsewhere Case, 24
historical development, 11 experiencer subject, 9
historical origin, 14
lexical, 5 Finnish, 2
locative, 18, 19 focus clitic, 16
39
40
Greenlandic, 31 pronoun
case system, 14
KP, 19 historical origin, 15
Urdu/Hindi paradigm, 14
LFG, 3 psych predicate, 9
architecture, 22 Punjabi, 32
linking, 23
quirky case, 25
Mapping Theory, 23
Middle Indo-Aryan Sanskrit
case system, 12 case system, 11
modal, 9, 29 Scottish Gaelic, 2
semantic case, 3, 25
Nez Perce, 3 specificity, 7
split-ergative, 1, 13
oblique marking, 12, 13, 16 stress, 17
Old Hindi, 13 structural case, 24
Suffixaufnahme, 11
Pān.ini, 1
passive
Turkish, 2
ability, 10
perfect
morphology, 13 undergoer, 1
phrasal affix, 16
postposition, 18 volitionality, 6

Clause Type Language Type: Ergative Accusative Transitive Erg-Abs Nom-Acc Intransitive Abs Nom

Uploaded by

Copyright:

Available Formats

Clause Type Language Type: Ergative Accusative Transitive Erg-Abs Nom-Acc Intransitive Abs Nom

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Clause Type Language Type: Ergative Accusative Transitive Erg-Abs Nom-Acc Intransitive Abs Nom

Uploaded by

Copyright:

Available Formats

The study of case has been an integral part of the study of clausal structure for cen-

Table 1. Clause Types

Clause Type Language Type

the ergative to alternate with an unmarked case (nominative/absolutive) on subjects of

(1) a. nadya=ne zu ja-na h

(2) a. nadya=ne jiraf dekh -na h

In this section, we show that Urdu/Hindi exhibits a structural system of grammati-

Table 2. The Case Markers

Clitic Case Gram. Func. Morph. Effect

(4) a. nadya=ne zu ja-na h

Tense/Aspect Valency Ergative Semantic Entailment

The nominative is phonologically null and in some discussions is referred to as an abso-

b. nj m=ne s dd f=se/*ko paoda k t.-a-ya

(8) a. nj m=ne s dd f=ko m sala c kh -va-ya

The instrumental se is extremely versatile. It may be used for instrumental adjuncts as

b. nadya=ne aj lahor=se fon ki-ya

The genitive may be roughly characterized as marking subjects of nonfinite clauses, as

(14) a. ram=ke b t.h -ne=p r mã=ne s=ko

In modern Urdu/Hindi, the oblique is mandatory for all non-nominative (non-direct)

(17) adnan d.akxane g -ya h

jāro juga māhātma jehi kabir jan-ā-yo nāth

The basic paradigm of pronouns in Urdu/Hindi is given in Table 5 (based on Kellogg

Table 5. Pronoun Paradigm

NOM ERG ACC / DAT INST LOC GEN

(21) a. *[[k tt or gh or.]-e]=ko b. *[[k tt-a or gh or.]-e]=ko

(22) a. yasin=ne [k tt-e or gh or.-e]=ko dekh -a h

(23) a. s=hi=ne kam ki-ya

(24) a. *k tt-hi-e b. *kh el-hi-a

(25) ke pich e ‘behind’ ke p hle ‘before’

(26) bılli bıstar [ke nich e] so rah-i h

All the items in (25) correspond to a relational preposition in English. In contrast to

(27) m˜=ne bılli=ka pich a ki-ya

We analyze genitives as occuring in the Spec NP position. For a similar analysis

(33) adnan d.akxane/zu g -ya h

(35) ram kh ãs-a

(36) a. Constituent-structure: b. Functional-structure:

(37) a-structure pound ag pt

(38) Gram. Functions Features Gram. Functions Features

(39) a-structure pound ag pt

f-structure SUBJ OBJ

(40) a. Wellformedness principle: KP : ( CASE )

b. Default: (( SUBJ CASE )= NOM)

c. Default: (( OBJ CASE )= NOM)

While the identification of grammatical functions is not necessarily tied to posi-

(42) la ‘bring’ ( PRED )= ag[ o] th[ r]

In this section we focus on the ergative/nominative and ergative/dative case alterna-

((SUBJ ) OBJ) ( CASE ) DAT

Similarly, ko encodes a disjunction between its use as an accusative object marker

Consider (46) which shows an alternation with respect to volitionality.

(46) a. ram kh ãs-a

(47) ram N kh ãs V -a

SEM - PROP CONTROL INT

Now consider the alternation ergative/nominative in transitive sentences, exempli-

(49) a. ram gar.i c la-ta (h )

Next consider the ergative/dative alternation in the infinitival construction in (50).

(50) a. nadya=ne zu ja-na h

(51) h ( PRED )= be ag/go[ o],Ev[ r]

SEM - PROP CONTROL INT

(21) a. [[k tt or gh or.]-e]=ko b. [[k tt-a or gh or.]-e]=ko

(24) a. k tt-hi-e b. kh el-hi-a