<<

46

■ Article ■ The Cases in the Oblique Case Constructions in Oriya*

Junji Yamabe

1. Preliminaries This article is concerned with the oblique case subject con- structions in Oriyal), exemplified in (1) and (2).2)

(1) a.mo-ra girlfrend naah-~.3) me-GEN girlfriend not. be (E)-3s 'Idon' t have a girlfriend .' b. bas-re mo-ra parisraa hei gal-aa. bus-Loc me-GEN urine happened-3s 'Iurinatedonabus .' (2) a. mo-te nabina bhala laag-e. me-ow Nabin like-3s 'I like N abin. ' lit. 'Nabin strikes me as good.' b. mo-te e barsa bi caakiri miLil-aa ni. me-ow this year also job received-3s not 'I did not get a job this year either.' These sentences have a logical subject4) in non-nominative, i.e. oblique, case, and hence, the name of oblique case subject con- struction.

山 部 順 治Junji Yamabe, Graduate School, University of Tokyo, Linguistics. Main Publication: "Word Order within Gerund Clauses in Bengali" , Tokyo University Linguistic Papers 10, pp. 227-256, Department of Linguistics, University of Tokyo, 1991 . The Cases in the Oblique Case Subject Constructions in Oriya 47

Oblique case subject constructions ("dative subject construc- tions" or "experiencer subject constructions", as more commonly called), attested in many languages of the world,5) are cross-linguis- tically characterizable in terms of semantics. Their logical subjects are "non-volitional" [Kachru 1981 for Hindi; Klaiman 1980,1981 for Bengali; Sridhar 1976,1979 for Kannada] or "non-agentive" [MacAlpin 1976:193 for Malayalam], that is, do not exert con- scious control over the realization or non-realization of the de- scribed situation.6) This characteristic distinguishes them from passive constructions. The logical subjects of the latter are voli- tional (= agentive), obligatorily in many languages including Oriya, and canonically in others such as English. While widespread cross-linguistically, is in South Asian lan- guages where oblique case subject constructions present great vari- ety and appear in text frequently. thus constitute of the geographical features of these languages, among which is Oriya, an Indo-Aryan language.7) Those in Oriya choose for the cases on the logical subjects be- tween genitive (henceforth abbreviated as GEN, related to mark- ing with GEN) as in (1) and dative (DAT, related to marking with OBJ)as in(2),and for those on the logicalobjects(henceforth, themes afterits semantic role)between nominative(NoM)and accusative(Acc).8)This articleis to describe the uses and non- uses of the four cases and go on to explain aspects of the observa. tions.9) The subsequent part of this paper is organized as follows.Sec- tion 2 explores the cases on logicalsubjects,and section 3 those on themes.Section 4 looks into the cases in nominalized clauses.Sec- tion 5 detailsthe combinations of the cases on the logicalsubjects and themes inclauses.Section 6 is a conclusion.

2.The Dative and Genitive Cases on Logical Subjects Like Oriya,Hindi distinguishestwo cases for the oblique case subjects(=the so-called"dative subjects"). 48 Journal of the Japanese Association for South Asian Studies, 7, 1995

(3) merii ek bahan hai. me.GEN one sister is 'I ha ve a sister.' (4) mujh-e vah laRkii acchii lagtii hai. me-OBj that girl like 'I like the girl .'

The Oriya (la) and the Hindi (3) involve the same semantic rela- tions and both have logical subjects in genitive. Further, the Oriya (2a) and the Hindi (4) involve the same semantic relations and both have dative logical subjects.10) In contrast, Bengali has only genitive subjects, and Dravidian languages such as Malayalam only dative subjects. Thus, Hindi and Oriya are similar in dividing the oblique case subjects to the two cases. consider that the dative and genitive cases have abstract meanings and are used (or not used) in accordance with them. First, dative means the prototypical role of goal, a locus toward which something moves: this is most clear when it is on the recipi- ent noun phrase in a ditransitive clause (= the one to whom some- one {gives/tell/shows} something) [Mohanan and Mohanan 1990, Verma and Mohanan 1990:9]. The datives on subjects are to ex- press their semantic status as goals in terms of localistic metaphor, as schematized in (5b) for the Malayalam example in (5a) [Mohanan and Mohanan 1990].

(5) a.baalane{dukkham/santooSam}wannu. boy.DAT grief.NoM/happiness.NoM come.PAST 'The boy became{sad/happy} .' [Mohanan and Mohanan 1990:47]

b. sadness/ happiness •¨ o boy [Mohanan and Mohanan 1990 : 49] Second, genitive means the prototypical role of possessor [Verma and Mohanan 1990 : 9], as is obvious when it is on a modifier of a noun as in nabina-ra baapaa Nabins father'. And in localistic The Cases in the Oblique Case Subject Constructions in Oriya 49 metaphor a possessor can be conceived of as a locus {at/by/in} which a possessum is found. On the basis of the above, we can characterize the semantic features distinguishing between dative and genitive cases as involvement and non-involvement of such notional directional movement as an arrow indicates in (5b).11) Cases, including the two of our concern, extend their ranges of uses from more to less typical circumstances. They , however, di- verge in the distances and courses of such extensions. As a conse- quence, corresponding ("the same") cases in different languages appear at widely differing, though substantially overlapping , ranges. Oblique case subjects overall are non-typical positions for dative and genitive cases to appear on, and accordingly form do- mains where a great deal of divergence is observed across lan- guages.12)In what follows we look into the distributions of the two cases in Oriya (2.1-2.4) and account for the observations in light of the semantic considerations above (2.5, 2.6).

2. 1 The Verb ach- 'Be(E)' The existential 'be' verb, ach-, forms a clause denoting a state of possession. The noun referring to the possessed entity (= possessum) appears in nominative and agrees with the verb. Where the possessum is a concrete entity, namely a thing or a person, the case on the possessor is genitive and never dative. Genitive is in use whether the possession is a permanent/integral/ inalienable one as in (6), or a temporary/alienable one as in (7) .

(6) {mo-ra / *mo-te} {bhaai / girlfriend I ghara / bahuta me-GEN/ me-ow brother / girlfriend / house / many braNa / daaRhi / lambaa baaLa} ach-i / baapaa pimple / beard I long hair be(E)-3s / father ach-anti.13) be(E)-3p 'Ihave{a brother/ a girlfriend/ahouse/rnany pimples/ a beard/long hair/a father}.' 50 Journal of the Japanese Association for South Asian Studies, 7, 1995

(7) {tuma-ra / *tuma-ku} {eka Tankaa / cigarette} -GENI you-OBJ one rupee I cigarette ach-i ki? be(E)-3s Q 'Doyouhave{one rupee/cigarettes}(with you)?'

For temporary possession, the possessor can alternatively be fol- lowed by the abstract location noun paakha-re `side-LOC', as in (8).

(8) tuma-paakha-re {eka Tankaa / cigarette} ach-i ki?'= (7)' Abstract possessed entities can be classified as follows: (i) emo- tions, ex.(9); (ii) physiological phenomena, (10); (iii) characters, (11); (iv) social relations, (12); (v) results and plans of volitional activities, (13); (vi) time, (14). Genitive on the possessor is in use with all these. Dative is marginal in some instances of (i), (ii) and (vi), and outright unacceptable elsewhere. (9) {mo-ra / ?mo-te} {bhaya / sandeha / dukkha / icchaa / me-GEN/ me-OBJ fear / doubt / sorrow / will / aagraha / taaThi prema / taaThi gruNaa / aasaa / aapatti} desire at.him love / at.him hatred / hope / objection ach-i. be(E)-3s (10) a. {tuma-ra / ?tuma-ku} {bhoka / jvara} ach-i ki? you-GEN/ you-OBJ hunger / fever be(E)-3s Q b. {taa-ra / *taa-ku} {cancer I TB I sardi} ach-i. -GEN/ he-OBJ cancer / TB / cold be(E)-3s (11) {taa-ra / *taa-ku} {computer-ra gyaana / homosexual he-GEN/ he-ow computer-GENknowledge / homosexual prakruti / dharja / bhala guNa / kharaapa abhyaasa} nature / patience / good virtue / bad habit ach-i. be(E)- 3s (12) {tuma-ra / *tuma-ku} {samparka / daaittva / dosa / you-GEN/ you-ow relation / responsibility / guilt I The Cases in the Oblique Case Subject Constructionsin Oriya 51

eThi adhikaara} naah-i•`.

here right not.be(E)-3s (13) {mo-ra / *mo-te} {jojanaa / kaama / chuTi / income} me-GENi me-OBJ plan / work / holiday / income ach-i. be(E)-3s (14) {aama-ra / ?aama-ku} {samaya / Deri} ach-i. we-GEN / we -OBJ time / delay be(E)-3s

Note that, according to our classification, some nouns have dual membership and some are not typical members of any class: aasaa 'hope'can be(i)an emotion or(v)aresult of activity that is mental; ruci 'taste, interest' is under an ordinary interpretation (v) a char- acter (as distinguished from aagraha (i) `desire'); bhala dhaaraNaa 'good idea' and semitiuddesya 'such intention'areperhaps(v)re- sults of activities that are mental; asubidhaa 'inconvenience' is left unclassifed (Regarding cases, this patterns with (i)). Inanimate nouns can be possessors in the sense that they can participate in the same construction as typical possessors, as in (15). (15) ei gacha {-ra / *-ku} bisa ach-i. this plant -GEN I -OBJ poison be(E)-3s 'This plant has poison .'

We assimilate such inanimates to the instances as in (6) where the possessum is concrete, or to those as in (11), (iii) characters, where the possessum is abstract.

2. 2 The Verb he- 'Become, Happen' The verb he- 'become, happen' takes nominative nouns to yield one-place predicates denoting events. For emotions, the logical subject can be either in genitive or dative. (16) {mo-ra / mo-te} {sandeha / ghruNaa / bhaya / dukkha} me-GEN/ me-OBJ doubt / hatred / fear / sorrow 52 Journal of the Japanese Association for South Asian Studies, 7, 1995

heuch-i. be.happening-3s

' I am getting {doubt/ hatred/ fear/ sorrow} .'

An exception to this generalization is raaga he- 'anger happen' = 'get angry', which takes a genitive but not a dative subject. (I do not have an explanation for this.) For physiological phenomena involving body-interior (accord- ingly intangible) aspects, the logical subject can be either in genitive or dative.") (17) {mo-ra / mo-te} {kaachu / chickenpox / appendicitis} me-GEN/ me-oBJ scabies / chicken pox / appendicitis heich-i. have.happened-3s 'I have got {scabies / chicken pox / appendicitis} .' Examples of nouns of this class are in (18). The resulting predi- cates can take either genitive or dative. (18) bhoka 'hunger', jvara 'fever', nida 'sleep', names of dis- eases (e.g. (17), cancer), kaasa 'cough', cinka 'sneeze', haakuTi 'hiccup', baanti 'vomit' For physiological phenomena that do not involve body-interior working, the logical subject can only be in genitive as in (19). (19) {mo-ra / *mo-te} {ghaa / bahuta braNa / me-GEN/ me-ow festering.wound / many pimple / ghimiri / botha} heich-i. heat rash / boil have.happened-3s 'I have got {a fester / many pimples / a heat rash/ a boil} .' The predicates parisraa he- 'urine happen', (1 b), and (bahuta) jhaaRaa he- ' (much) feces happen' are additional examples. It should be added that with these and those in (19), dative is out under the ordinary interpretations of them: it becomes good (or bet- ter) if we contrive to construe the sentences as referring to body- internal ailments, assimilating to them to instances like (17). The Cases in the Oblique Case Subject Constructions in Oriya 53

For events external to a body, the logical subject is in genitive. The nominative nouns can be abstract as in (20a-c), or concrete as in (20d). (20) a. {mo-ra / *mo-te} {raastaa-re durghaTanaa / me-GEN/ me-OBJ way-LOCaccident / operation / bhul / baaRei I badaLi} hel-aa. operation / mistake impact / shift happened-3s '1{got an accident on the way/got an operation/ made a mistake / hit on something / transferred}.' b. {aama-ra / *aama-ku} phaLa hel-aa ni. we-GEN/ we-OBJ result happened-3s not 'We did not get results .' c. 1960-re {taa-ra / *taa-ku} jamma heithil-aa. 1960-Loc he-GEN / he-oBj birth happened-3s 'He was born in 1960 .' d. {semaana-nkara / *semaana-nku} gata maasa-re they-GEN / they-OBJ last month-Loc jhia hel-aa. girl happened-3s 'Agirl was born to them last month .' With a noun referring to time, genitive is always fine and dative is sometimes good. (21) a. {taa-ra / taa-ku} kaali koRie barsa he-GEN/ he-oBJ yesterday 20 year hel-aa. happened-3s 'He became 20 years old yesterday .' b. {aama-ra / ?aama-ku} bahuta deRi hei gal-aa. we-GEN I we-OBJ much delay happened-3s 'We got delayed much .'

2. 3 The Verb aas- 'Come' The verb aas- 'come' takes nominative nouns denoting emotions 54 Journal of the Japanese Association for South Asian Studies, 7, 1995 and physiology, to yield predicates denoting eventive phases of those phenomena. They denote actualization (`do, happen') of the event denoted by the incorporated noun, as in (22a), or a stage before actualization (`feel, feel like'), as in (22b). (The factors con- ditioning this variation are not clear to me.) Most of them can take either a dative or a genitive subject as illustrated in (22).

(22) a. {mo-ra / mo-te} {haakuTi / haai / jvara / raaga} me-GEN/ me-oBj hiccup / yawn / fever / anger aasil-aa. came-3s '1 {hiccuped / yawned / got fever / got an gry}.' b. {mo-ra / mo-te} {vaanti / nida} aasuch-i. me-GEN/ me-ow vomit / sleep be.coming-3s 'Iam feeling{likevomiting/sleepy} .'

In the sense of understanding, aas- can take only dative but never genitive.

(23) {*mo-ra / mo-te} kichi aas-e ni. me-GEN / me-OBJ anything come-3s not 'Icannot understand a nything.'

For pasanda aas- 'liking come' = 'like', see (75).

2. 4 Verbs Taking Dative Subjects There are many predicates that take a dative subject but never a genitive subject. Two-place verbs of this class are exemplified in (24).15)

(24) bhala laag- 'like', subh- 'think', suNaa j- 'hear' (< 'be heard'), dis- 'see', dekhaa j- 'see' (< 'be seen'), bujhaa j- 'can understand'(<'be understood'), aas- `understand, know' (< 'come': ex. (23)), miL- 'receive, meet' (33b, 34b), laag- 'need, take the responsibility of (a child)' (< 'attach(i)') The Cases in the Oblique Case Subject Constructions in Oriya 55 The verb laag- 'feel' (< `attach(i)') takes nouns denoting emo- tions and physiological phenomena in nominative. (25) {*mo-ra / mo-te} {khusi / raaga / bhoka / parisraa / me-GEN/ me-OBJ happiness / anger / hunger / urine / baanti} laaguch-i. vomit be.attaching-3s 'I am feeling{ happy/angry/hungry/like urinating/ like vomiting} .'

The verb dekhaa- 'show' collocates with certain nouns, and means 'feel like'. This usage is highly ideomatic: I have found no more than three nouns of this kind.

(26) {*mo-ra/ mote} {baanti / parisraa I jhaaDaa} me-GEN/ me-oBj vomiting / urine / feces dekhaauch-i. be .showing- 3s 'Iam feelinglike { vomiting / urinating/ evacuating}.' lit. 'It is showing me vomiting etc.'

he reasons for dative marking here are clear. The verb laag- literally denotes concrete movement leading to a goal. In (25) it metaphorically extends this notion. The verb dekhaa-, (26), is ba- sically a and accordingly takes a metaphorical goal. The verbs in (24), too, can be analyzed as involving direc- tional movement: the idea, the visual impression, the auditory im- pression and the possessorship of the themes referents move up to (the senses of) the logical subjects referents.

2. 5 The Dative •\ Genitive Alternation In the sentences in (27) the described situations are, objectively viewed, identical. However, a change in the verb gives a change in case choice.

(27) a. {mo-ra / ?mo-te} Tike bhaya ach-i. (cf.(9)) me-GEN / me-OBJ a.bit fear be-3s 56 Journal of the Japanese Association for South Asian Studies, 7, 1995 b. {mo-ra / mo-te} Tike bhaya heuch-i. (cf.(16)) me-GEN/ me-ow a.bit fear be.happening-3s c. {(?)mo-ra / mo-te} Tike bhaya aasuch-i. (cf.(20)) me-GEN/ me-ow a.bit fear be.coming-3s d. {*mo-ra / mo-te} Tike bhaya laaguch-i. (cf.(25)) me-GEN me-oBJ a.bit fear be.attaching-3s 'I {(a) have / (b)(c) am getting / (d) am feeling} fear a little.' In (c), some speakers deem the genitive less than perfect. The nouns sandeha 'doubt', cintaa 'anxiety' and dukkha 'sorrow' follow more or less the same pattern. asubidhaa 'inconvenience', though not denoting an emotion, also conforms to this. It should be noted. that judgments, if measured in absolute terms, are variable accord- ing to speakers, and to occasions of interviews. Similar observations can be made for physiological phenomena, (28)-(29). (28) a. {tuma-ra / ?tuma-ku} bhoka ach-i ki? (cf. (10)) you-GEN you-OBJ hunger be-3s Q b. {??tuma-ra / tuma-ku} bhoka you-GEN you-OBJ hunger heuch-i ki? (cf. (17,18)) be.happening-3s Q c. {*tuma-ra / tuma-ku} bhoka you-GEN you-oBJ hunger laaguch-i ki? (cf.(25)) be. attaching-3s Q '{(a) Are you / (b) Are you getting / (c) Are you feel- ing} hungry?' (29) contrasts between ach- 'be' and he- 'happen'. (29) a. ftaa-ra I *taa-kuf {cancer I sardi / TB / kaachu} he-GEN/ he-obj cancer / cold / TB I scabies ach-i. (= (10b)) be-3s The Cases in the Oblique Case Subject Constructions in Oriya 57

b. {taa-ra / taa-ku} {cancer I sardi / TB I kaachu} he-GEN / he-oBj cancer / cold / TB / scabies heich-i. (= (17)) have.happened-3s 'He {(a) has / (b) has got} {a cancer / a cold / TB / scabies} .' dhaaraNaa 'idea' and anka 'maths', results of volitional activities (see also 2.1), present comparatively clear splits between he- and aas-.

(30) a. {pilaamaana-nkara / *pilaamana-nku} bhul children-GEN / children-ow wrong dhaaraNaa ach-i. idea be-3s b. {pilaamaana-nkara / ?pilaamana-nku} bhul children-GEN / children-ow wrong dhaaraNaa heuch-i. idea be.happening-3s c. {?pilaamaana-nkara / pilaamana-nku} bhul children-GEN / children-ow wrong dhaaraNaa aasuch-i. idea be.coming-3s 'Children {(a) have / (b)(c) are getting} misconcep- tions.' (31) a. {mo-ra / *mo-te} anka hu-e ni. (cf. (20)) me-GEN / me-OBJ maths happen-3s not b. {*mo-ra / mo-te} anka aas-e ni. (cf. (23)) me-GEN / me-OBJ maths come-3s not 'I cannot do maths.'

The observations above can be covered by the orders as in (32).

(32) GEN: more likely ach- •¬ he-•¬.. aas- •¬ laag- less likely DAT: less likely ach- •¬ he- •¬ aas- •¬ laag- more likely 58 Journal of the Japanese Association for South Asian Studies, 7, 1995 This arrangement is partly explainable in terms of the semantics of the verbs. At the beginning of section 2 we have proposed dis- tinguishing the dative and genitive cases via involvement and non- involvement (respectively) of notional directional movement. This concept can be decomposed into components: (i) change and (ii) spatial directionality. Of the four verbs at issue, laag- and aas- contain notions (i) and (ii), and he-, only (i), and ach, neither: they differ in the amount of contribution to the conception of direc- tional movement. We are thus led to expect that, other parts being equal, dative is more likely to appear with laag- and aas- than with he-, and more likely with he- than with ach-; and that the reverse holds for . (Note that the observed divergence be- tween laag- and aas- is left unaccounted-for.) Under this view, the instances where both cases are possible as (26b) can be regarded as representing a domain where the exten- sions of the two cases overlap. And the variance in judgments among speakers can be taken as due to their differing inclinations toward either one of the cases. The verbs paR- 'fall' and miL- 'receive' also participate in con- trasts in case choices. In (33)-(34), the verb he- 'happen, become' goes with genitive, (33a) (see also (20)); paR- 'get' (< `fall'), with either dative or genitive, (34a); miL- 'receive' with dative, (33a, 34b) (see also fn.15). (33) a. {mo-ra / *mo-te} e barsa bi caakiri me-GEN/ me-ow this year also job hel-aa ni. (cf. (20)) happened-3s not b. {*mo-ra / mo-te} e barsa bi caakiri me-GEN/ me-ow this year also job miLil-aa ni. (= (2b)) received-3s not 'I did not get a job this year either.' (34) a. {aama-ra / aama-ku} beLebeLe kaama paR-e. we-GENI we-OBJ sometimes work fall-3s The Cases in the Oblique Case Subject Constructions in Oriya 59 b. {*aama-ra / aama-ku} beLebeLe kaama miL-e. we-GEN/ we-oBj sometimes work receive-3s 'We sometimes t(a) have / (b) get} work.' Another example for paR-, as it takes either dative or genitive, follows:

(35) {mo-te / mo-ra} sehi jami paRil-aa. me-oBJ / me-GEN that land fell-3s 'I was allotted the land.' = 'The land fell into my hand.'

Admitting the ground is weak, we can assume for (33)-(35) also that varied choices of cases reflect varied degrees of involvement of directional movement in the conceived images of an objectively identical phenomenon. Regarding the differentiation between paR- 'fall' and he- 'become', we may speculate that the former contains a more solid conception of it than the latter since the change the former denotes is concrete (i.e. motion) while the one the latter denotes is abstract.

2.6 More on Directional Movement In this section, I present a set of facts that can be covered by the following generalization: If the referent of the theme noun phrase is anchored to a particular location within the described event, by expression or implication, a use of dative on the logical subject becomes difficult or impossible. This is a consequence of the se- mantics of the as a goal of movement. If a body part noun is expressed as the theme and if this serves to denote the location of the event, a logical subject cannot be in dative, as in (36) and (37).

(36) {mo-ra / *mo-te} {deha / mana} bhala laagu- me-GEN *me-oBj body / mind good attaching ni. not.be.3s 'I am not feeling well in my {body / mind}: 60 Journal of the Japanese Association for South Asian Studies, 7, 1995

(37) a. {mo-ra/ *mo-te} muNDa {dharuch-i / me-GEN / me-OBJ head be.catching-3s / bulouch-i} . be.turning(t).about-3s 'I have a headache.' lit. 'It is {catching / turning} my head.' b. {mo-ra /*mo-te} peTa kaaTuch-i. me-GEN/ me-ow stomach be.cutting(t/i)-3s 'I have a stomachache.' lit. 'It is cutting my stomach.'

Without body-part nouns, the same verbs go with a dative subject , as in (38) and (39). (38) {mo-te / *mo-ra} bhala laagu- ni. me-oBJ / me-GEN good attaching not .be.3s 'I am not feelina well.' (39) injection ne-laa beLe {mo-te / *mo-ra} injection take-GER time me-ow / me-GEN kaaTil-aa. cut(i/t)(Dast)-3s

'I got pain when I took an injection.' lit. 'It cut me . . . '

Where a body part is expressed as a locative noun , the logical subject can be in genitive but hardly in dative, as in (40c) and (41b). (40) a. {taa-ra / taa-ku} cancer heich-i. (= (17)) he-GEN/ he-oBJ cancer have.happened-3s b. {taa-ra / (?)taa-ku} lung-cancer heich-i. he-GEN/ he-OBJ lung cancer have.happened-3s c. {taa-ra / ??taa-ku} phusphus-re cancer he-GEN/ he-oBJ lung-LOC cancer heich-i. have.happened-3s 'He has got {(a) cancer / (b) lung cancer / (c) cancer at the lung} .' The Cases in the Oblique Case Subject Constructions in Oriya 61

(41) a. {taa-ra / taa-ku} kaachu hel-aa. (cf. (17)) he-GEN/ he-ow scabies happened-3s b. {taa-ra / ??taa-ku} deha-re kaachu hel-aa. he-GEN/ he-ow body-LOC scabies happened-3s 'He got {(a) scabies / (b) scabies in his body}.' The genitive case on taa-ra in (40c) seems to be ambiguous as to whether its assignment is due to the noun phusphus 'lung' or to the predicate cancer he-. Similarly for the genitive in (41b). In (42), dative is not good, with or without the body-part specification eThi 'here'.

(42) {mo-ra / ??mo-te} (eThi) kuNDhei heuch-i. mo-GEN / me-ow here itching be. happening-3s 'I am feeling itchy (at this part).' We analogize this instance to the ones like (40b) and (41c) on the ground that itching is a spatially limited process by nature. This account also applies to the division of body-interior vs. body-sur- face physiological phenomena, seen with examples (17, 18) vs. (19): the latter are phenomena only with local effect. An entity also may be circumscribed due to occurring in a tan- gible/spatial domain of the world. Hence the impossibility of da- tive for the body-external phenomena as in (20), in contradistinc- tion to the instances as in (17, 18). The restriction of concern extends beyond oblique case subject constructions proper. This can be seen in the fact that Oriya lacks instances of the so-called "dative of (in-)convenience", which is familiar from Continental European languages such as German, Spanish and Polish. Consider (43)-(44).

(43) a. {mo-ra / *mo-te} kaNTaa-re shirt ciri gal-aa. me-GEN/ me-OBJ nail-Loc shirt got.torn-3s 'My shirt got torn with a nail (on me).' b. {mo-ra / *mo-te} pua mari gal-aa. me-GEN/ me-oBj son died-3s 62 Journal of the Japanese Association for South Asian Studies, 7, 1995

'My son died (on me).' c. {mo-ra / *mo-te} se box-Taa sahaja-re kholaa me-GENi me-OBj that box-QL ease-LOC opened hel-aa. became-3s '(My) box opened easily.' = 'I could open my box eas- ily.' (44) {*mo-ra / *mo-te} barsaa hel-aa. me-GEN / me-ow rain happened-3s 'It rained on me.' = 'I got rained on.'

Here, the referents of the themes (the shirt , the son, the box) do not move in the depicted events. Indeed, the events denoted by the clauses or their effects can be construed as moving toward the intended experiencers, but this does not support dative marking . In (43), the genitive signifies the possessors (in the literal sense) of the themes. In (44), where a relation of possession does not obtain , genitive is not allowed.

3. The Nominative and Accusative Cases on Themes 3.1 Predicates Taking Accusative Themes First consider a basic transitive clause, (45). The is some- times marked with objective (ow), (45a), and sometimes with nominative (Nom), (45b).

(45) a. taa-ku / nabina-ku / se pua-ku / seiTaa-ku / MU~ b. *se / *nabina / *se pua / seiTaa / me.NOM him / Nabin / that boy / that one / *kichi jinisa-ku kichi dekhi paaril-i ni. jinisa any thing (-ow) see could-ls not 'I could not see {him / Nabin / that boy / that / any- thing} .'

To account for this case marking alternation , we assume, first, The Cases in the Oblique Case Subject Constructions in Oriya 63 that a transitive verb assigns its object feature (Acc), and, second, that this is related to marking according to (46) below. (See fn.8 for the overall framework the relevant assump- tions are embedded in.)

(46) A noun phrase with accusative case feature (Acc), if it is human and/or definite, is marked with the objective case morpheme (OBJ), and otherwise, is marked with the morpheme (Nom).16) Turning to oblique case subject clauses, (47) and (48), we can observe alternation in case marking on the themes that is similar to, but has more options than, the one seen in (45). Here, human and/or definite noun phrases are marked either objectively (OBJ)or nominatively (Nom), and others are only marked nominatively (Nom).17) (47) a. taa-ku / nabina-ku / ??seiTaa-ku / mo-te b. se / nabina / seiTaa / me-ow him / Nabin / that one / *kichi jinisa-ku disil-aa ni. kichi jinisa any thing(-OBJ) could.see-3s not 'I could not see {him / Nabin / that / anything}.' (48) a. tuma-ku / nabina-ku / eiTaa-ku mo-te b. tume / nabina / eiTaa / me-ow you / Nabin / this one / * paisaa-ku darakaara. paisaa money-(ow) need 'I need {you / Nabin / this / money} .' To account for this pattern, we assume that the predicates here assign to their themes either accusative (Acc) or nominative (Nom) case feature.18) These case features are related to marking according to (46) and (ii) in fn. 16, respectively. 64 Journal of the Japanese Association for South Asian Studies, 7, 1995

With some predicates, the theme, even if it is human, can only be marked nominatively.

(49) a. ??taa-ku / ??nabina-ku mo-te seThi miLil-aa ni. b. se / nabina me-OBJ there him / Nabin-(oBj) received-3s not 'I could not meet {him / Nabin} there.'

We assume that such predicates assign to the theme only nomina- tive but not accusative case feature. (50) classifies oblique case subject predicates according to whether they can assign accusative case feature to the theme. Note that membership varies across speakers, is not always clear-cut for each speaker.

(50) a. Nom OK, Acc OK: bhala laag- 'like', mane paR- 're- member', mane ach- 'remember', darakaara 'need' b. Nom OK, Acc %: dis- 'can see', dekhaa j- 'can see', suNaa j- 'can hear' c. Nom OK, Acc ?? / *: miL- 'receive, meet', laag- 'take responsibility of (a child)'

bujhaa j- 'can understand' seems to belong to the (50b) class. aas- 'understand' seems to belong to the (50c) class (This does not take a human theme).

3.2 The Double Object Case Constraint Consider the oblique case subject clauses in (51), where both of the two noun phrases are marked with objective (ow).

(51) a. mo-te niru-ku bhala laag-e. me-oBj Niru-oBJ like-3s 'I like Niru.' b. mo-te tuma-ku mane paRuthil-aa. me-oBj you-ow was.remembering-3s 'I was remembering you.' The Cases in the Oblique Case Subject Constructions in Oriya 65

c. mo-te tuma-ku disil-aa ni. me-OBj you-OBj could.see-3s not. 'I could not see you.'

A clause containing two objective-marked noun phrases are in fact ambiguous, allowing for two reversal readings in thematic rela- tions. That noun phrase in it that linearly precedes the other, or that is mo-te 'me-ow' tends to be taken as the logical subject. For example, (51a) can mean either 'I like Niru' or `Niru likes me', the former reading being preferred. Replacement of mo-te 'me-ow' in (51) with some other noun phrase remarkably reduces acceptability, as in (52).

(52) a. * {tuma-ku / taa-ku / nabina-ku} niru-ku me-OBj / you-OBJ / he-OBJ / Nabin-OBJ Niru-OBj bhala laag-e. like-3s '{You / He / Nabin} like(s) Niru.' b. * aama-ku tuma-ku mane paRuthil-aa. we-OBj you-OBj was.remembering-3s 'We were remembering you.' c. * aama-ku tuma-ku disil-aa ni. we-OBj you-ow could.see-3s not 'We could not see you.'

With the theme in nominative, the experiencer can be any noun.")

(53) a. {mo-te / tuma-ku / taa-ku / nabina-ku} niru me-ow / you-OBj / he-ow / Nabin-OBj Niru bhala laag-e. like-3s '{1 / You / He / Nabin} like(s) Niru.' b. {mo-te / aama-ku} tume mane paRuthil-a. me-ow / we-OBj you. NOM was.remembering-2 '{I was / We were} remembering you.' 66 Journal of the Japanese Association for South Asian Studies, 7, 1995 c. {mo-te / aama-ku} tume disil-a ni. me-ow / we-oBj you.NOM could.see-2 not '{1/We} could not see you .'

The observations in (51-53) show that what is blocked in (52) is

repeated occurrence of the morph -ku. We can capture this

through the rule as in (54), the Oriya version of the double object

case constraint.20)

(54) The Double Object Case Constraint (Oriya)

a. *X •c X, where Xs are due to assignment by the same

predicate.

b. X = -ku

Part (54a) says that the constraint is to label as ungrammatical

repetition in the same clause of some element(s), X. Part (54b)

nominates the morph -ku for X. Thus, of the instances of repeti-

tion of the objective case morpheme (OBJ), (54) only rules out the

sequence -ku •c -ku, and passes -te •c -ku, as in (51), -ku •c -te

and -te •c -te, as in (55) below. Notice that the allomorph -te

appears only with the two mo-te `me-ow' and to-te 'you(intimate) -OBJ', and the allomorph -ku with the other pro- nouns (e.g. taa-ku 'he-OBj', aama-ku 'we-OBj') and all nouns (e.g.

nabina-ku `Nabin-OBj', ghara-ku 'house-OBj').

The combination -te •c -te is acceptable at least for some speak-

ers.21)

(55) mo-te to-te {bhala laag-e / mane paRuthil-aa / me-OBJ you-OBJ like-3s / was. remembering- 3s / disil-aa ni} . could.see-3s not 'I {like/was remembering/ could not see} you .'

In ditransitive clauses, constraint (54) has little effect. For ex- ample, most consulted speakers accept all the choices in (56), al- though some feel slightly uncomfortable with the \-prefixed choices which lead to repetition of -ku.22) The Cases in the Oblique Case Subject Constructions in Oriya 67

(56) se aama-ku \ pua-ku / \seiTaa-ku dekhaail-aa ni. pua / sei-Taa he.NOM we-OBj son / that one(-OBJ) showed-3s not 'He did not show{the son/that one}to us.'

Further, a human proper noun and a human pronoun must be marked with objective, as in (57). Repeating -ku is the only accept- able choice here.

(57) se aama-ku taa-ku / rabi-ku dekhaail-aa ni. *se / *rabi he.NOM we-oBJ him / Rabi(-OBJ) showed-3s not 'He did not show{him/Rabi}to us.'

The Hindi counterpart to the Oriya constraint (54) differs from it in one respect. Saksena [1983] observes that in Hindi, repetition of the objective case morpheme is unacceptable even if the two occurrences get realized in distinct allomorphs, the one in -e and the other in -ko. (58) illustrates this for a ditransitive clause; (59), for a clause with an oblique case subject. Compare (58) with (56), and (59) with (51).

(58) mai~-ne us-e {phuul / *phuul-ko} phe-kaa me-ERG he-OBJ flower / flower-OBJ threw 'I threw the flower to him .'[Saksena 1983:45] (59) mujh-e {maa / *maa-ko} yaad aaii. me-OBJ mother / mother-OBJ remember 'I remembered mother .'[Saksena 1983:47]

Thus, the Hindi version refers to the morpheme OBJ, being blind to allomorphy, and can be expressed as in (60).

(60) The Double Object Case Constraint (Hindi)

a. *X •c X, where Xs are due to assignment by the same

predicate.

b. X = OBJ

The Oriya (54) and the Hindi (60) differ at the (b) parts, in other 68 Journal of the Japanese Association for South Asian Studies, 7, 1995 words, in the grammatical levels on which they apply. Though slight in itself, the difference bears on the fact that the dative subject constructions in Oriya allow objective case marking on the themes, as in (50), whereas those in Hindi do not, as in (59).

4. Nominalization and Cases When a basic-type clause is nominalized, the verb turns into the gerund form and the subject can either remain in nominative or get into genitive [Nayak 1987: 48-49] .23)In the examples to follow, (61)-(63) and (66)-(70), the (b) clauses are the nominalized ver- sions of the tensed (a) clauses.

(61) a. {tume / *tuma-ra} niru-ku bhala paa-a. you.NOM/ you-GEN Niru-OBJ like-2 'You like Nir u.' b. {tume / tuma-ra} niru-ku bhala paa-ibaa-Taa you.NOM / you-GEN Niru-OBJ like-GER-QL 'your liking Niru'

Consider the contrast between (b)'s of (62) and (63). Whereas an animate noun can be marked genitively, as in (62b), an inanimate noun cannot, as in (63b).24)

(62) a. {nabina / *nabina-ra} aasithil-aa. Nabin / Nabin-GEN came-3s 'Nabin came.' b. {nabina / nabina-ra} aas-ibaa-Taa Nabin / Nabin-GEN come-GER-QL 'Nabin's coming' (63) a. {ciThi / *ciThi-ra} aasithil-aa. letter / letter-GEN came-3s 'A letter came .' b. {ciThi / *ciThi-ra} aas-ibaa-Taa letter / letter-GEN come-GER-QL 'a letter coming' The Cases in the Oblique Case Subject Constructions in Oriya 69

This points to the following constraint: Only a noun phrase with the semantics of sentient role can go into genitive through nominalization. 25) An inanimate entity cannot be sentient, and ac- cordingly cannot appear in genitive in (63b). An object cannot get into genitive, whether it is inanimate as in (64) or human as in (65). (64) {hindi bhaasaa / *hindi bhaasaa-ra} paRh-ibaa-Taa Hindi language / Hindi language-GEN learn-GER-QL 'learning(of) the Hindi language'

(65) {hari-ku / *hari-ral dhar-ibaa-Taa Hari-ow / Hari-GEN catch-GER-QL '(someone's) arresting(of) Hari'

This is because an object cannot be construed as having a sentient role. (66) illustrates the situation where an (underlying) object has turned into a subject through passivization. The judgments on the genitive in nominalization, (66b), are elusive, varying both accord- ing to speakers and to interview sessions.

(66) a. {hari / *hari-ra} (pulis-dvaaraa) dharaa gal-aa. Hari / Hari-GEN police-by caught went-3s 'Hari was arrested (by the police).' b. {hari /* / (?)hari-ra} (pulis-dvaaraa) dharaa j-ibaa-Taa Hari / Hari-GEN police-by caught go-GER-QL 'Hari' being arrested (by the police)'

Note further that passive predicates overall collocate best with an inanimate rather than human subject, and that those going with a human subject such as (66) connote thing-like status for it. The above observations indicate that in clauses like (66b) genitive is unacceptable to the degree that one feels the subjects as thing-like, and accordingly non-sentient. Let us turn to oblique case subject constructions. For many (but not all) speakers, a dative subject, having a sentient role, can come 70 Journal of the Japanese Association for South Asian Studies, 7, 1995 into genitive in a nominalization context.

(67) a. {*tuma-ku / *tuma-ra} {taa-ku / nabina-ku / you-OBJ you-GEN he-oBJ / Nabin-ow / baapaa-nku} bhala laag-e. father-ow like-3s a'. {tuma-ku / *tuma-ra} chocolate bhala laag-e. you-ow / you-GEN chocolate like-3s 'You like {him / Nabin / (your) father / chocolate}.' b. {*tuma-ku / %tuma-ra} {taa-ku / nabina-ku / you-ow / you-GEN he-oBJ / Nabin-ow / baapaa-nku} bhala laag-ibaa-Taa father-oBJ like-GER-QL b'. {tuma-ku / %tuma-ra} chocolate bhala laag-ibaa-Taa you-OBJ/ you-GEN chocolate like-GER-QL

' your liking {him / Nabin / (your) father / chocolate} ' (68) a. {tuma-ku / *tuma-ra} bhoka laaguch-i. you-oBJ I you-GEN hunger be.feeling- 3s 'You are feeling hungry.' b. {tuma-ku / %tuma-ra} bhoka laaguth-ibaa-Taa you-OBJ I you-GEN hunger be.feeling-GER-QL

' your feeling hungry'

In (67a) = (51a) and (67b), dative on the experiencer is not good because it leads to a violation of the double object case constraint, (54). As can be seen above, a logical subject can come into genitive, when going with an objectively marked theme as in (67b) or going with an inanimate theme such as 'chocolate' in (67b'). In contrast, as in (69b) below, the same is not possible with a nominatively marked theme that is human and definite. (A speaker expressed the feeling that with the ? ?-marked choices in (69b), the referent of the theme noun phrase would be construed as a choice for employ- ment, i.e. something like a thing.) The Cases in the Oblique Case Subject Constructions in Oriya 71

(69) a. {tuma-ku / *tuma-ra} {se / niru} bhala laag-e. you-oBJ / you-GEN he.NOM/ Niru like-3s (cf. (52a)) a'. {tuma-ku / *tuma-ra} baapaa bhala laag-anti. you-oBj / you-GEN father like-3p 'You like {him / Niru / (your)father} ! b. {tuma-ku / ??tuma-ra} {se / niru / baapaa} you-OBJ/ you-ow he.NOM/ Niru / father bhala laag-ibaa-Taa like-GER-QL

' your liking {him / Niru / (your) father}' These observations indicate that for many speakers, genitive on the logical subject is allowed only where the theme is assigned accusative case feature, (67b), but not where this is assigned nomi- native case feature, (69b): schematically, % GEN- Acc, ?? GEN NOM, where GEN is due to nominalization. According to (46), a noun phrase assigned accusative (Acc) gets marked nominatively (Nom) if it is inanimate and/or indefinite, and otherwise marked objectively (om). The theme noun phrase in an oblique case subject construction cannot get into genitive. (70) a. {nabina / *nabina-ra} tuma-ku bhala laag-e. Nabin.Nom / *Nabin-GEN you-oBj like-3s 'You like Nabin.' = `Nabin is liked by you.' b. {nabina / *nabina-ra} tuma-ku Nabin.NOM/ *Nabin-GEN you-OBJ bhala laag-ibaa-Taa like-GER-QL

' you liking Nabin' = `Nabins being liked by you' This is because the theme in an oblique case subject construction does not play a sentient role.

5. Case Configurations in Clauses This section looks into how the cases on logical subjects and 72 Journal of the Japanese Association for South Asian Studies, 7, 1995 themes, examined in sections 2 and 3 respectively, go together in clauses.

5. 1 Multiple Choices darakaara 'need' has at least five options for the case configura- tion.26)

(71) a. mo-te tume darakaara (ach-a). (DAT - NOM) me-OBJ you.NOM need be-2 (cf. (48b)) b. mo-ra tume darakaara (ach-a). (GEN - Nom) me-GEN you.NOM need be-2 c. mo-te tuma-ku darakaara (ach-i). (DAT - Acc) me-ow you-OBJ need be-3s (cf.(48a)) d. mo-te tuma-ra darakaara (ach-i). (DAT - GEN) me-oBJ you-GEN need be-3s e. mo-ra tuma-ku darakaara (ach-i). (GEN - Acc) Me-GEN you-oBj need be-3s 'I need you.'

Several predicates have four options. pasanda (ach-) 'like' is an example, (72). mane paR- 'remember' and mane ach- 'remember' follow the same pattern. For some speakers, darakaara 'need' also follows this pattern (rather than the one as in (71)).

(72) a. mo-te se pasanda (ach-i). (DAT - Nom) me-ow he.NOM liking be-3s b. mo-ra se pasanda (ach-i). (GEN - Nom) me-oBj he.Nom liking be-3s c. mo-te taa-ku pasanda (ach-i). (DAT - Acc) me-oBJ he-oBJ liking be-3s d. mo-ra taa-ku pasanda (ach-i). (GEN - Acc) Me-GEN he-oBJ liking be-3s e. * mo-te taa-ra pasanda (ach-i). *(DAT - GEN) me-oBJ he-GEN liking be-3s 'I like him.' The Cases in the Oblique Case Subject Constructions in Oriya 73

A variant of this pattern is found with jaNaa (ach-) 'know', in which a genitive subject is marginal: (73a, c) are fully acceptable whereas (73b, d) are less than perfect.

(73) a. mo-te se jaNaa (ach-i). (DAT - Nom) b. ?mo-ra se jaNaa (ach-i). ?(GEN - Nom) c. mo-te taa-ku jaNaa (ach-i). (DAT - Acc) d. ?mo-ra taa-ku jaNaa (ach-i). ?(GEN - Acc) e. *mo-te taa-ra jaNaa (ach-i). *(DAT - GEN) 'I know him.'

Some predicates have two options. bhala laag- 'like' can assign either nominative or genitive to the theme, giving DAT - NOM, DAT - Acc, as in (74). For some speakers, dis- 'can see' (ex.(47)), dekhaa j- 'can see' and suNaa j- 'can hear' pattern the same (cf.(50b)). (For others, only the (74a) configuration is possible.) To gloss (74), bhala laag-a 'like-2', bhala laag-e 'like-3'.

(74) a. mo-te tume bhala laag-a. (DAT - NOM) b. *mo-ra tume bhala laag-a. *(GEN - NOM) c. mo-te tuma-ku bhala laag-e. (DAT - Acc) d. *mo-ra tuma-ku bhala laa-e. *(GEN - Acc) e. *mo-te tuma-ra bhala laag-e. *(DAT - GEN) 'I like you.'

pasanda aas- 'like' presents a variant of this, as in (75). Besides the (a) and (c) configurations, (b) is quite acceptable as well. Gloss: pasanda aas-e 'like-3s', ni 'not'.

(75) a. mo-te se pasanda aas-e ni. (DAT - Nom) b. ?/OKmo-ra se pasanda aas-e ni. ?/OK(GEN - Nom) c. mo-te taa-ku pasanda ass-e ni. (DAT - Acc) d. *mo-ra taa-ku pasanda aas-e ni. *(GEN - ACC) e. *mo-te taa-ra pasanda aas-e ni. *(DAT - GEN) 'I do not like him.' 74 Journal of the Japanese Association for South Asian Studies, 7, 1995

Predicates allowing the two case configurations DAT - NOM

and GEN - Nom follow: the verb he- combined with nouns denot-

ing emotions and physiological phenomena (e.g. bhaya he- 'fear

happen' (exx. (16, 26b)), kaachu he- 'scabies happen' (ex. (17))),

and paR- 'get' (•ƒ 'fall') (exx. (34b, 35)).

Predicates having only one case option follow. DAT - NOM

only: miL- 'get, meet' (exx.(31b,32c), laag- 'take the responsibility

of (a child)', manaa 'prohibited' and several adjectives, and pre-

sumably also aas- 'understand' (•ƒ 'come') (ex. (23b)). GEN - NOM

only: most instances of possession (exx. (6-15)), and most in-

stances of body-external events (exx. (20, 33)).

5.2 Negation and Cases

For clauses with a verbal predicate, negation does not affect case possibilities. The non-verbal predicate darakaara can be negated either by the existential naah~- not.be(E)' or by the predicational nuh~- 'not .be(P)'. The former leaves the variety illustrated in (71) intact, as in (76). The latter reduces the number of choices, as in (77): the theme must be in nominative. Gloss for (76), naah-a'not.be(E)- 2', naah-i'not.be(E)-3s'; for (77), nuh-a~ 'not.be(P)-2', nuh-e~ 'not .be(P)-3s'.

(76) a. mo-te tume darakaara naah-a~. (DAT - Nom) b. mo-ra tume darakaara naah-a~. (GEN - Nom) c. mo-te tuma-ku darakaara naah-i~. (DAT - Acc) d. mo-ra tuma-ku darakaara naah-i~. (GEN - Acc) e. mo-te tuma-ra darakaara naah-i~. (DAT - GEN) 'I do not need you.' (77) a. mo-te tume darakaara nuh-a~. (DAT - Nom) b. mo-ra tume darakaara nuh-a~. (GEN - Nom) c. *mo-te tuma-ku darakaara nuh-e~. *(DAT - Acc) d. *mo-ra tuma-ku darakaara nuh-e~. *(GEN - Acc) e. *mo-te tuma-ra darakaara nuh-e~. *(DAT - GEN) The Cases in the Oblique Case Subject Constructions in Oriya 75

‘ I do not need you.'

Likewise, pasanda can be negated with the existential naah•` - or the predicational nuh•` -, and when negated with the predicational nuh•` - it looses the options DAT - Acc and GEN - Acc, where the theme is in non-nominative. jaNaa can be negated only by the existential naah•` - but not by the predicational nuh•` -

6. Concluding Remarks

Of the logical subjects in the oblique case subject constructions in Oriya, those which can be conceived of as possessors appear in genitive as in (1), and those which can be conceived of as goals (e.g. experiencers) in dative as in (2). Some can appear in either case, exx. (16, 17, 18, 21, 22, 27b, c, 29b, 34a, 35). We can interpret these last instances as representing an overlap domain of the exten- sions from typical instances. On the theme noun phrases (= logical objects), two cases, nominative and accusative, are available. Some predicates allow nominative only, and some, either case. Crossing these options on the two noun phrases can yield maximally four options for case configurations of clauses, i.e. genitive-nominative, dative-nominative, genitive-accusative, dative-accusative (in the order of the logical subject - the theme), and the full set of these are materialized with some predicates, exx. (71, 72).

The two features, namely, the division of labor among more than one case and the variety of instances of multiple case options, characterize the oblique case subject constructions in Oriya in con- trast with those in other South Asian languages. To look at neigh- boring languages, those in Telugu (Dravidian) invariably take the case configuration dative-nominative (Only the (a)-type configura- tion of (71-75) is possible) [Nozomi Kodama, personal communi- cation], and those in Bengali (Indo-Aryan) invariably genitive-ac- cusative (Only the (d)-type configuration is possible).

To the extent the present paper is successful, it as a whole points to new research space. First, there is much to be clarified on the 76 Journal of the Japanese Association for South Asian Studies, 7, 1995 uses and non-uses of expressions which presumably is accessible by help of plural native speakers judgments (though there will certainly be limitations to this method). This work, if seriously pursued, can amount to more than restating textbook knowledge, or filling up a conventional questionnaire. Second, the observed facts do not merely make a mere list of idiosyncrasies but exhibit regularities that hold. Such regularities call for description and explanation. I hope that the present paper can motivate work along this line on grammatical aspects of the South Asian languages , which is prerequisite to comparative study on their sociology , ge- ography and history.27)

Notes * I would like to thank all the people taught me Oriya and afforded me hospi-

talities during my stays in Orissa. Special thanks are due to Mr . and Mrs. M.M. Guin for their cooperation in various aspects of data collection . I am grateful to Nozomi Kodama for his moral support, to Shingo Einoo for his detailed comments

on the earlier manuscript which contributed much to the content and exposition , and to an anonymous reviewer for checking my English in the penultimate version . 1) The judgments on sentences reported in the present article most closely reflect

those of Modan Mohan Guin (born in Jajpur , 1960, brought up in Sambalpur) and Sujata Guin (born in Cuttack, 1964, brought up in Rourkela). The series of inter-

views with them were carried out in Tokyo during 1994-1995 . The data have been cross-checked with those of other speakers from various places of Orissa , obtained in Rourkela and Nowrangpur, August-September 1994 , and in Tokyo, 1995. 2) For reason of space limitation, we omit discussion of the adjectives and of the

oblique case subject predicates as they embed a subordinate clause/verb . They also take an oblique case subject

3) Pronunciation (Oriya): a = [o], as = [a], c = [tf] , j = [d3], consonants + h = aspirates, capitals = retroflexes,•` = vowel nasalization. English words are cited in italics . Ab- breviations: Acc = accusative, DAT = dative, (E) = existential 'be' , GEN/GEN = genitive, GER = gerund, (i) = intransitive, LOC = locative, NOM/NOM = nominative, OBJ = objective, p = plural, (P) = predicational 'be' , Q = question particle, QL = qualifier, s = singular, (t) = transitive, 1/2/3/ = first/second/third person. *-, ?- and %-marked examples are unacceptable, less than perfect and acceptable only for a

subset of speakers, respectively.

4) A logical subject is defined as that of a predicate that is in the highest

position in the hierarchy: agent •„ beneficiary •„ recipient / experiencer •„ instru- ment •„ theme / patient •„ location [Bresnan and Kanerva 1989:23 and references The Cases in the Oblique Case Subject Constructions in Oriya 77

cited there]. In an oblique case subject construction, it is an experiencer (e.g. one who {feels / knows / sees / gets afflicted with} something) or a recipient (e.g. one who receives something). We classify a possessor as a recipient (i.e. one in the resultant state of receiving something) in the above-cited hierarchy. "Location" refers to a physical / concrete one, not a metaphorical / abstract one such as a pos- sessor [Bresnan and Kanerva 1992: 116-117]. We skip the question of which noun phrase in a clause is the syntactic subject (as distinguished from the logical subject). Accordingly, our uses of the terms " oblique case subject construction", "dative subject" and "genitive subject" should be understood to be for convenience of reference. 5) For example, Japanese ore-ni kore-ga {wakar-u / deki-ru} (koto) `(that) I (DAT) {can understand I can do} this (NOW). See Masica [1976] and fn. 18 for names of lan- guages having oblique case subjects. 6) For earlier treatments of the semantics of "dative subjects" in the general-linguistic literature, see Anderson [1971: 100-104] and Fillmore [1968: 24ff.]. 7) Verma and Mohanan (eds.) [1990] is a recent collection of articles on various as- pects of the oblique case subject constructions in South Asian languages. There seems to be no linguistic work focused on those in Oriya. School for native speakers do not recognize them as a distinct class of constructions. 8) The conceptual and terminological framework we presuppose for Oriya in the discussion of this paper is as follows. A predicate assigns each of its complements one of the case features: nominative, accusative, dative and genitive, abbre- viated as Nom, Acc, DAT and GEN, respectively. A case feature is related to mark- ing by one of the case morphemes: nominative, objective and genitive, ab- breviated as NOM,OBJ, GEN, respectively. A case morpheme is realized in one of the morphs: zero, /-e, -ku, -te, -ra/. Thus, we distinguish (at least) three gram- matical levels. The abovementioned items are mutually associated as in (i).

(i) (case features)

(case morphemes)

(case morphs)

This reads, for example, as follows. All instances of dative case feature (DAT) and some instances of accusative case feature (Acc) correspond to marking by the ob- jective case morpheme (ow). The morpheme OBJin turn realizes itself in either of the allomorphs /-ku/ and /-te/ depending on the lexeme it attaches to. Note that no established frameworks are currently available for describing case alternations in Indo-Aryan languages. Mohanan [1993] is a proposal for Hindi that is distinct from, and in some respects conflicts with, ours. 9) For the sake of simplicity, this paper concentrates on the items appearing in (i) of fn. 8. Treating the and the postpositions is outside its scope. 10) Dative in (3), i.e. mujh-e (OBJ) ek bahan hai, is dialectal. Hindi translates (1b) with favor of either case within the domain of oblique case subjects. 13) A noun denoting a respectable person is grammatically plural, as with baapaa ‘father'in(6).Aninanimatenoun,even referring to many instances, is grammati- cally singular, as with 'many pimples' in (6) and 'cigarettes' in (7). 14) Some speakers find genitive unacceptable where he- 'happen' combines with nouns denoting physiological phenomena that are exclusively body-interior such as hun- ger, fever and names of some diseases (e.g. cold, diabetes, cancer, AIDS, malaria). (Other speakers find genitive as well as dative fully acceptable, whose judgments are illustrated in the text.) 15) Predicates that take a dative subject can sometimes collocate with a genitive noun phrase looking like a logical subject, as in (i). (i) a. {mo-te / mo-ra} aaji daramaa miLil-aa. me-OBj / me-GEN today salary received-3s

‘ I received my salary today.' b. {tuma-ku / tuma-ra} bahuta paisaa laagib-a. you-OBj / you-GEN much money will.need-3s ‘You will need much money.'

The genitive here is due to the noun phrases being modifiers of nouns. The predi- cates are not its assigners. To see this, consider the sentences in (ii), where the genitive does not allow the intended readings. (ii) a. {mo-te/ #mo-ra} has miLil-aa ni. me-OBJ / me-GEN bus received-3s not

` I couldnt catch a bus today.' (GEN: 'My bus was not available.') b. {aama-ku / #aama-ra} aaji eka lakhya Tankaa miLil-aa. we-OBj / #we-GEN today one lakh rupee received-3s

‘We received one hundred thousand rupees today.' (GEN:'We received the one thousand rupees of ours today.') 16) On the counterparts to (46) in Indo-Aryan languages generally, see Masica [1982, 1986, 1991: 364-369]. The rules for dative and nominative case features are stated below (See also (i) in fn.8). The Cases in the Oblique Case Subject Constructions in Oriya 79

with roughly the same meanings. For example, it declines in (47) faster than in (45) as we proceed from higher to lower positions on the animacy/definiteness scale. Note also that some speakers do not simply allow objective marking on the theme noun with dis-. See also (50b). 18) It is not typologically rare that the themes in so-called "dative subject construc- tions" receive the same morphosyntactic treatment as the objects in basic transitive clauses, which is reflected in "accusative" or "objective" case marking. It is re- ported in South Asia for Bengali [Klaiman 1980: fn. 4, 1981: 16], Kashmiri [Altaha 1985: 196-197], Malayalam [Asher 1968: 99, Masica 1986: 163, MacAlpin 1976: 188-189], Tamil [Agesthialingom 1972: 8 (quoted in Sridhar 1979: 113), Lindholm 1976: 162-167, Arul Raj 1981: 313] and Sinhala [Gair 1990: 17, 18, 1992: 191], and outside for Choctaw [Davies 1986: 100], Early [Allen 1986: 398], Imbabura Quechua [Jake 1983: 250] and Udi [Harris 1984: 248]. 19) The verb, if present, obligatorily agrees with the theme if the latter is in nominative as in (53), and in the default third person singular if it is in non-nominative as in (51). 20) The Hindi version of this constraint is in (60). Versions of it, blocking two occur- rences of "objective" or "accusative" case, are found outside South Asia as well, for example, in Georgian [Allen 1984] and Japanese [Shibatani 1978: 262, 361]. 21) It seems that some others find (55) unacceptable (or less acceptable than (51)). This judgment, if it is real, can be captured by modifying the (54b) part as: X = -ku and -te . 22) For the latter set of speakers, replacement of aama-ku in (57) by mo-te as in (i) removes this tiny problem.

pua-ku / seiTaa-ku (i) se mo-te dekhaail-aa ni. pua / sei-Taa he.Nom me-OBj son / that one(-OBJ) showed-3s not ‘He did not show {the son/ that one} to us.'

23) We use the names of the case "genitive" ambiguously for the case feature (Gen) and the case morpheme (GEN)since the distinction seems to have no empirical bearing. Nominalized clauses with a genitive subject, as in (61b), are not included in the class of oblique case subject constructions. Such genitive subjects are more often than not semantically agentive/volitional. We leave open the question of whether a unified semantic account is possible of the genitive in oblique case subject con- structions and that in nominalization. 24) A fact parallel to this is reported for Bengali in Klaiman [1981: 60]. A somewhat similar observation is made for English, too: whereas a subject noun phrase denot- ing a person can be marked with the genitive s in a gerund clause (Bill's coming here), one denoting a thing cannot (the ice(*'s) melting [Grimshaw 1992: 122]). 25) The term sentient stands for conscious involvement in the action or state [Rozwadowska 1988: 158]. Note also that volitionality / agentivity entails sentient- ness, but not vice versa. For its relevance to the , see of the verb existential 'be' after it, (ib), a comparable observation being made for the Bengali cognate darkaar 'need' by Machida and Niwa [1991: 101].

(i) a. aama-ra ebe bi {seiTaa/ seiTaa-ra} darakaara ach-i. we-GEN still that(-GEN) need be(E)-3s b. aama-ra ebe bi {seiTaa / *seiTaa-ra} darakaara. (a) and (b) 'We still need it.'

The predicate darakaara 'need' provides a clear piece of evidence that the objec- tive case marker (OBj) on a logical subject reflects dative case feature (DAT) rather than accusative case feature (Acc). As in (ii), it can take an inanimate logical sub- ject. Such a noun phrase must be marked with objective.

(ii) {gaaRi-ku / *gaaRi} petrol darakaara. car-OBJ/ car petrol need ' A car needs petrol.'

27) Geographical consideration of the cases in the oblique case subject constructions in South Asian languages is carried out in the earlier longer version of the present article.

References Allen, Cynthia L. 1986 "Reconsidering the History of Like", journal of Linguistics, 22, pp. 75-409. Allen, W. Sidney 1984 "On Certain Case Constraint and Their Interpretation", Lin- gua, 63, pp. 1-15. Altaha, Fayez Mohammed 1985 Non-nominative Subjects in Kashmiri. doctoral disser- tation, State University of New York. Anderson, John M. 1971 The Grammar of Case. Cambridge, Cambridge University Press. Arul Raj, V. S. 1981 "The Syntactic Treatment of Subject in Tamil", in S. Agesthialingom and N. Rajasekharan Nair (eds.) Dravidian Syntax. pp. 307-324, Annamalainagar, Annamalai University. Asher, R. E. 1968 "Existential, , Locative and Copulative Sentences in The Cases in the Oblique Case Subject Constructions in Oriya 81

Inversion in UG: A Reply to Schachter's Comments" , in Tim Stowell and Eric Wehrli (eds.) Syntax and Semantics, 26, pp. 111-125, San Diego, Academic Press. Davies, William D. 1986 Choctaw Verb Agreement and Universal Grammar , Dordrecht, Reidel. Fillmore, Charles 1968 "Case for Case", in Emmon Bach and Robert T. Harms (eds.) Universals in Linguistic Theory. pp. 1-88, New York, Holt, Rinehart and Winston. Gair, James W. 1990 "Subjects, Cases, Infl in Sinhala", in Verma and Mohanan (eds .) [1990], pp. 13-41. Gair, James W. 1992 "AGR, INFL, Case and Sinhala Diglossia or: Can Linguistic Theory Find a Home in Variety", in Edward C. Dimock, Jr., Braj B. Kachru and Bh. Krishnamurti (eds.) Dimensions of Sociolinguistics in South Asia. pp. 179-209, Delhi , Oxford & IBH Publishing. Grimshaw, Jane 1991 Argument Structure. Cambridge, Mass., MIT Press. Harris, Alice 1984 "Case Marking, Verb Agreement, and Inversion in Udi", in Carol Rosen and David Perlmutter (eds.) Studies in Relational Grammar, 2, pp. 243-258 , Chicago, University of Chicago Press. Hook, Edwin Peter 1990 "Experiencers in South Asian Languages: A Gallery", in Verma and Mohanan (eds.) [1990], pp. 319-334. Jake, Janice L. 1983 Grammatical Relations in Imbabura Quechua. doctoral disseration, University of Illinois at Urbana-Champaign. [Published by Garland, New York, 1985.] Kachru, Yamuna 1981 "Transitivity and Volitionality in Hindi-Urdu", Studies in the Linguistic Sciences, 11, pp. 181-193. Klaiman, N. H. 1980 "Bengali Dative Subjects", Lingua, 51, pp. 275-295. Klaiman, M. H. 1981 Volitionality and Subject in Bengali: A Study of Semantic Param- eters in Grammatical Processes. Bloomington, Indiana University Linguistics Club. Lindholm, James M. 1976 "Nested Case Relations and the Subject in Tamil", in Verma (ed.) [1976], pp. 152-182. MacAlpin, David W. 1976 "Dative Subject in Malayalam", in Verma (ed.) [1976], pp. 183-194. Machida, Kazuhiko and Kyoko Niwa 1990 Bengarugo (The Bengali Language). To- kyo, Hakusuisha. Masica, Colin P. 1976 Defining a Linguistic Area. Chicago, University of Chicago Press. Masica, Colin P. 1982 "Identified Object Marking in Hindi and Other Languages" , in Omkar N. Koul (ed.) Topics in Hindi Linguistics, Vol. 2, pp. 16-50, New Delhi, Bahri. Masica, Colin P. 1986 "Definiteness-Marking in South Asian Languages", in Bh . Krishnamurti, Colin P. Masica and Anjani K. Sinha (eds.) South Asian Languages. pp. 123-146, Delhi, Motilal Banarsidass. Masica, Colin P. 1991 The Indo-Aryan Languages. Cambridge, Cambridge University Press. Mohanan, K. P. and Tara Mohanan 1990 "Dative Subjects in Malayalam", in Verma and Mohanan (eds.) [1990], pp. 43-57. 82 Journal of the Japanese Association for South Asian Studies, 7, 1995

Mohanan, Tara 1993 "Case Alternation on Objects in Hindi", South Asian Language Review, 3(1), pp. 1-30. Nayak, Rath 1987 Non-finite Clauses in Oriya and English. doctoral dissertation, Cen- tral Institute of English and Foreign Languages, Hyderabad. Rozwadowska, Bozena 1988 "Thematic Restrictions on Derived Nominals", in Wendy Wilkins (ed.) Syntax and Semantics, 21, pp. 147-165, San Diego, Academic Press. Saksena, Anuradha 1983 "A Case Marking Constraint", Lingua, 60, pp. 41-52. Shibatani, Masayoshi 1978 Nihongo no Bunseki (A Grammatical Analysis of Japanese). Tokyo, Taishukan. Sridhar, S. N. 1976 "Dative Subjects", Studies in Linguistic Sciences, 6, pp. 582-593. Sridhar, S. N. 1979 "Dative Subject and the notion of Subject", Lingua, 49, pp. 99- 125. Verma, Manindra K. (ed.) 1976 The Notion of Subject in South Asian Languages. Uni- versity of Wisconsin. Verma, Manindra K. and K. P. Mohanan 1990 "Introduction to the Experiencer Sub- ject Construction", in Verma and Mohanan (eds.) [1990], pp. 1-11. Verma, Manindra K. and K.P. Mohanan (eds.) 1990 Experiencer Subjects in South Asian Languages. Stanford, Center for the Study of Language and Information.