Krishnamurti Bhadriraju. The Dravidian Languages

Подождите немного. Документ загружается.

6.3 Cases 221

‘mouth’ (acc–dat); a variant of -t-is-d-inkey ‘hand’: key-d-a

(inst–abl) (Burrow and

Bhattacharya 1970: §§63–7).

Central Dravidian Kolami has only four stems that take the oblique marker -t-, nal:

t-, si

d: si

t-, ul: u

t- (all meaning ‘day’), v¯ega

d ‘ﬁeld’: v¯ega

t- (Emeneau 1955b/1961: 61).

Naiki (Chanda) uses -t- in some non-masculine nouns in some of the cases (genitive,

locative, ablative) in the singular, k¯ı ‘hand’: k¯ı-t-un (loc), ¯ur ‘village’: ¯ur-t-a (gen).

In Parji, -t- occurs as an augment in the case of certain neuter nouns before ablative,

genitive and locative cases, mer ‘tree’: mer-t-o (gen), mer-t-i (loc), juve ‘well’ : juve-t-

are ‘from the well’.

Ollari has a trace of this sufﬁx in a few neuter words, e.g. polub-t-un ‘in the village’,

k¯ı-t-in ‘in the hand’; it has a variant in -

t- but we do not know the conditions, mar-

te-v¯er

‘root of the tree’.

Gadaba has -t- in some singular stems, t¯o

ta ‘garden’: t¯o

ta-t-in (loc), polub-t-un ‘in

the village’.

North Dravidian Only Brahui in North Dravidian has anything similar to PD

∗

-tt- in its

genitive, xar¯as-t¯a ‘of the bull’; in the other oblique cases there is an augment -t- before

the case sufﬁxes in the plural: dat–acc sg xar¯as-e ‘(to) the bull’, pl xar¯as-t-e ‘(to) the

bulls’, abl sg xar¯as-¯an ‘from the bull’, xar¯as-te-¯an ‘from the bulls’ (Bray 1909: 43).

Summary The foregoing distribution justiﬁes the reconstruction of

∗

-tt- as an oblique

marker in substantives of certain inanimate subclasses of stems. For South Dravidian I,

we can include in these two clear subclasses: -m/-n ending (mara-m ‘tree’) neuter nouns

and those that end in -

tu/-

tu preceded by a (C)

V or (C)VCV-. Kanna

da regularly and

da partially changed -tt- > -t- [-d-]. The classes of stems taking the augment -tt- are

retained intact in South Dravidian I. In South Dravidian II and Central Dravidian these

subclasses are expanded to include other stems but still the grammatical category of

gender is identical. In South Dravidian II, -tt- is combined with other augments to form

complex augments,

∗

tt +

∗

i (Telugu, Ko

da),

∗

tt +

∗

i +

∗

n (Kui). Pengo and Ma

da use

-ti as a genitive marker, -t- oblique, -i- genitive. Telugu, Ko

da, and Kui have adopted

the genitive base as the oblique stem, hence its double function. There are several less

pervasive oblique markers, discussed below.

6.3.1.2

∗

-an/

∗

-in;

∗

-nV

South Dravidian I In Old Tamil -a

n occurred with demonstrative pronouns, quantiﬁers

and numerals; -i

n after disyllabic and trisyllabic stems ending in a¯au¯u¯eaiin the

instrumental, dati

ve and occasionally sociative cases. The stem ending in -

in by itself

was genitive, which could be used syntactically as an adnominal.

222 Nominals

Ta. atu ‘that one’: at-a

n-ai (acc) ‘that’, at-an-¯al (instr) ‘by that’, ¯aru ‘six’: ¯ar-ar-ku

(dat) ‘to six’ (

n → r/

−−−

[+ stop, − voice]); e

zutt-ir-ku ‘to letter’, kanav-in-¯al ‘by dream’.

In classical texts there were usages without the augments also, e.g. kal ‘stone’: ka

r-ku

‘to stone’, ka

nu ‘eye’: ka

n-¯al ‘by the eye’.

Malay¯a

lam had -an as an augment of demonstratives in early inscriptions. Sometimes

-in was used, instead, e.g. ira

tu ‘two: ira

t-in-¯al ‘by twos’, itu ‘this one’: it-in-ukku

‘to this’. Otherwise, -in had the same distribution as -i

n of Old Tamil. Stems that take

-tt- add -in- also in dative and genitive, (dat) mara-tt-in-n

‘to the tree’, (gen) mara-tt-

in-

de ‘of the tree’ (Asher and Kumari 1997: 191–4).

In Iru

la -(a)n occurs as augment with animate nouns including the personal pronouns

before instrumental, e.g. n¯a ‘I’: nan-an-¯ale ‘by me’, n¯am ‘we’: nam-an-¯ale ‘by us’, pi

‘child’: pi

le-n-¯ale ‘b

y the child

’. I

nKo

dagu -¨ın/-n are

used as augments after neuter

demonstrative pronouns in accusative, dative and genitive cases, ad-¨ın-a (acc), ad-¨ın-gu

(dat), ad-¨ın-

du (gen). They also occur after many neuter nouns of one or two syllables in

accusative, instrumental and ablative cases, e.g. ¯ur ‘village’: ¯ur-n-, ba

te ‘road’: ba

te-n-.

Numerals take -¯an-, e.g. e

tu ‘eight’: e

t-¯an-a ‘eight’ (acc), e

t-¯a

da (gen).

Kota has -n after neuter demonstratives, ad-n-k ‘to that’ (dat), ad-n-l ‘in that’ (loc),

ed-n-l-tr (abl) ‘from what?’ Toda adds -n in adnominal use of some noun stems, p¯aw

‘river’: a

p¯aw-n

b ¨or

‘the name

of that

river

’.

In Kanna

da -a

r, a sandhi variant of -an (see Tamil above), became generalized as

the oblique marker of neuter demonstratives in the singular and plural and in numerals,

ad-ar-ke (r ←

r/ k; dat) ‘to that’, ad-ar-im (instr) ‘by that’, era

du ‘two’: era

d-ar-o

lage

(loc) ‘in two’. In Pampa Bh¯arata, this occurs with the plural demonstrative forms, av-a

r-,

iv-a

r- in dative and instrumental–ablative. The augment -in/-i occurs more widely after

consonant-ﬁnal stems in all cases except the accusative, maga

l-in-ge ‘to the daughter’,

d¯evar-i (n)-ge ‘to the gods’. Ba

daga has -n- after the third neuter demonstratives, ad-n-a

(acc, gen).

In Tu

lu -n occurs as an augment after human nouns and after stems ending in -e,

guru-n-a (gen) ‘of a Guru’, kudke-n-a ‘of a fox’, maga

l ‘daughter’: maga

l-n-a (also

kudk-a and maga

l-a, which establish -a as the genitive sufﬁx). Koraga has -n- for human

nouns in genitive, appe-n-a ‘of mother’.

South Dravidian II Telugu has -an-i (a sequence of two augments) as oblique augment

in demonstrative neuter forms, in the singular and plural, before metathesis took place,

e.g. adi ‘that’: d¯an-i-<

∗

ad-an-i, idi ‘this’: d¯ın -i- <

∗

id-an-i, ¯edi ‘which?’: d¯en-i- <

∗

ed-

an-i; so also the plurals, e.g. awi ‘they’ w¯an-i- <

∗

aw-an-i, etc.

Stems ending in neuter

Curiously, spoken Telugu has in the plural w¯a

t-i <

∗

aw-att-i with the augment

∗

-att which com-

pares better with Classical Tamil augment -a

rr- in the neuter plural demonstratives (see 6.3.1.3).

It appears that Old Telugu normalized the augment used in the singular, while spoken Telugu

6.3 Cases 223

singular -mu take -na- as oblique marker, gu

rramu ‘horse’: gurramu-na-ku/gurr¯a-na-

ku ‘to the horse’. Masculine singular nouns ending -

du in the nominative replace it by

-ni, maga

du ‘husband’: maga-ni-ki ‘to the husband’. These -na/-ni do not seem to be

historically related to

∗

-an/

∗

-in.

Gondi has -n as an augment after masculine nouns ending in a vowel, marri ‘son’:

marr¯ı-n-, muriyal ‘father’: muriya-n-, tammur ‘younger brother’: tammu-n-. This feature

is found in all dialects. The augment -n is also used after the plural sufﬁx -k or -ø of

some non-masculine as well as masculine nouns, e.g. ka

r-k ‘eyes’: ka

r-k-n-gag¯a ‘in the

eyes’, mark ‘sons’: mark-n-. Female human nouns that end in a vowel or the derivative

sufﬁx -al, or any of the consonants -r,-

r,-

d, replace the stem-ﬁnal consonant by the

augment -n, e.g. tange ‘elder sister’: tange-n-, but y¯ayal ‘mother’: y¯aya-n-, s¯el¯a

r ‘sister’:

s¯el¯a-n. Some other Gondi dialects add -t, e.g. s¯el¯a

r-t-, which apparently was the original

condition. The addition of -n instead in female human nouns must be a recent innovation

in the northern and central

dialects (see Rao 1987b: 146

–8).

da attests an underlying -an-i (two sufﬁxes -an + -i) in the neuter demonstrative

forms, which have undergone phonological changes, partly similar to those in Telugu,

adi ‘that’: da-ni, idi ‘this’: de-ni, avi ‘they’: va-ni, ivi ‘these’: ve-ni.

The masculine

plural ending -r takes -i as the oblique formative and those ending in common plural in

-k or -

take -a as the oblique marker.

Kui has masculine nouns ending in -nju in the singular and -ru in the plural, e.g. nega-

nju ‘good man’, nega-ru ‘good men’; the obliques are formed by adding -i to negan- and

negar-asnegan-i- and negar-i- to which the other case sufﬁxes are added (note that -ju

corresponds to PD

∗

-tu which occurs only in the nominative singular, see section 6.3).

In the case of nouns meaning female human, the oblique augment is clearly -n/-ni, e.g.

aja ‘mother’: aja-ni (gen), aja-n-gi (dat), aja-ni-i (acc); pl aja-ska ‘mothers’, aja-ska-ni

(gen), aja-ska-n-gi (dat). Even ‘neuter plurals that end in -nga are declined like masculine

nouns’ (Winﬁeld 1928: 25–30). Here also, we notice an extension of human inﬂectional

increment to non-human classes.

Kuvi has -n/-na as augment of nouns referring to humans, e.g. aya ‘woman’: aya-ni-ki

(dat), aya-n-i (gen); in plural the variant is -

n, aya-ska-

n-i (acc pl).

In Pengo -n occurs in the genitive plural of non-human nouns ending in -ku, e.g.

key-ku ‘hands’: key-ku-n-i (gen). Man

da also has -n- in the oblique-genitive, e.g. kiy-ni

neter

‘blood

of hand

’, mar-ni

¯ake

‘leaves

of the tree

’.

from the earliest times has preserved the inherited contrast in the augments. This is one class of

exception to the metathesis rules which involve mainly coronal consonants (section 4.5.7.3).

It is not clear if the metathesized forms are a shared innovation with Telugu or borrowings from

Early Telugu. The sequence

∗

id-an- should have become

∗

ed-an and

∗

d¯en- in Early Telugu. But

the vowel is ¯ı in d¯ı n-, perhaps to avoid homonymy with the interrogative d¯en- from ¯edi + an. But

da has the correct quality of the vowel in the obliques of idi and iwi, since the interrogative

forms are different, i.e. ayen ‘who?’, ayed ‘which?’

224 Nominals

Central Dravidian Kolami has no instances of an oblique with -n. Naiki (Chanda)

has -n in animate nouns in some of the cases, e.g. t¯olel ‘brother’: t¯ole-n-un (acc.), b¯ay

‘woman’: b¯ay-n-un (acc); it also occurs in genitive, pul-ne

tala

‘head

of tiger

’.

Parji has -n as oblique marker of some stems in ablative and genitive cases, e.g. kici-

n-a ‘of ﬁre’, kici-n-ar ‘from ﬁre’. Ollari has examples of -n- in genitive, kor-n-e

cen

‘cock’s

comb

’.

Gadaba has -n/-in/-un as a genitive marker, -n after a vowel and -in ∼-un after a

consonant, e.g. m¯are-n- ‘tree’, verg-in ‘cat’, polb-un ‘village’. Here the V before -n is

conditioned by the preceding vowel.

North Dravidian In Ku

rux -in/-i occurs after non-masculine

singular demonstrative

stems before all cases, ¯ad ‘it’: ¯ad-i-ge (dat), ¯ad-in-t¯ı (abl).

There is no evidence of an oblique marker involving -n in the other North Dravidian

languages.

Summary On the basis of comparative evidence,

∗

-an can be reconstructed for South

Dravidian I and South Dravidian II, particularly with respect to non-masculine demon-

stratives in the

singular,

∗

at-u,

∗

it-u,

∗

y¯at-u. Data from Tamil, Malay¯a

lam, Ko

dagu, Kota,

daga and Kanna

da (here, a

r- rather than -an) of South Dravidian I and Telugu and

da of South Dravidian II support this reconstruction. In some members the augment

has extended distribution: Iru

la has shifted -an to personal pronouns. The variant -in

occurs clearly in South Dravidian I, Tamil and Malay¯a

lam, with a different distribution

from that of -an, but -an and -in have become free variants in some of the languages,

e.g. Mala

¯a

lam, K

dagu and

Kanna

da.

In South Dravidian II, the augments na-/ni- occur in some of the members typically

with animate (generally human) nouns as distinguished from

∗

-tt which occurs pre-

dominantly with non-human/inanimate nouns. This occurs in Telugu and Gondi (with

masculine nouns), and Kui–Kuvi (with feminine nouns). Modern Kanna

da, Tu

lu and

Koraga of South Dravidian I and Naiki of Central Dravidian also seem to follow this

pattern with a morpheme of

the shape -

nV. The reﬂexes of

∗

-in also occur with other

classes of stems, not semantically so deﬁnable.

The other Central Dravidian languages do not show this trend, but they have the

-nV morph occurring in the genitive forms in Parji, Ollari and Gadaba. It is possible

to reconstruct

∗

-nV as another oblique marker. North Dravidian (other than Ku

rux) has

lost both

∗

-Vn and

∗

nV- .

6.3.1.3

∗

-an-tt-/

∗

-in-tt-

Tamil, Malay¯a

lam and Telugu give evidence of this complex augment. In Early T

amil

-a

rru occurs after plural neuter demonstrative roots av-, iv-; also after interrogative

and indeﬁnite pronouns, y¯a- ‘(an interrogative root)’, and pala- ‘many’, cila- ‘few’,

6.3 Cases 225

e.g. av-a

rr-u

l ‘in them’, iv-arr-o

tu ‘with these’, pala-v-arr-ai ‘many things’ (acc).

Malay¯a

lam has parallel uses of -a

rr-, e.g. av-arr-in-/avai(y)-irr-in/avai(y)-ir-in (dat).

Shanmugam (1971a: 241) considers that Old Kanna

da -a

r- is related to this morph rather

than to

∗

-an, by simpliﬁcation of the geminate to a single consonant.

Only Telugu has anything to compare with -a

tt- in the spoken forms, awi ‘those

(non-human)’: w¯a-

ti- (<

∗

aw-att-i), iwi ‘these (non-human)’: w¯ı

t-i-(<

∗

iw-att-i), ¯ewi

‘which ones? (non-human)’: w¯e-

ti-(<

∗

ew-att-i), respectively. Koya, a dialect of Gondi,

has v¯a-

ti- requiring a proto-form

∗

aw-antt-i. This looks like a borrowing from Early

(preliterary) Telugu because of the retention of the pregeminate nasal.

Telugu also has -i

ti-(<

∗

in-tt-i) as a complex oblique marker with number words,

deﬁnite and indeﬁnite, e.g. re

du ‘two’: re

d-i

ti- (obl/gen), m¯u

du ‘three’: m¯u

d-i

ti-

‘three’ (obl/gen), anni ‘that many (non-hum)’: ann(i)-i

ti-, etc. This is a combination

in + tt + i → in-

tt-i. From this it appears, on comparative grounds, that Old Tamil att-

is a combination of an +

tt- with the loss of n before a geminate (PP <

∗

NPP is a regular

rule in Tamil–Malay¯a

lam). For further details, see section 6.5 below.

6.3.1.4

∗

-a/

∗

-i

The vowels

∗

-a and

∗

-i occur as oblique markers in South Dravidian II and North Dra-

vidian. Therefore, they are reconstructible for Proto-Dravidian. In South Dravidian I -a

occurs as oblique marker with personal pronouns in dative, e.g. Ta. e

n-a-kku ‘to me’,

n-a-kku ‘to you’. Tu

lu has -e- as oblique maker after plural nouns, maro-ku

l-e (obl–

gen) ‘trees’, maro-ku

l-e-g¨ı ‘to trees’. Kanna

da also has -a in personal pronouns, nann-a

‘my’, nan-a-ge ‘to me’, etc.

Telugu adds -a to the common plural in -lu to form the oblique stem, which also

signals the genitive case at the syntactic level, bomma-lu ‘dolls’: bommal-a (obl–gen),

r¯aju ‘king’, r¯aju-lu ‘kings’: r¯ajul-a- (obl–gen), r¯aju-l-a-nu ‘kings’ (acc). The vowel -i

is added to human plural nouns ending in -ru and demonstrative pronouns, w

du ‘he’:

w¯an-i, w¯aru ‘they’ (hum): w¯ar-i (obl–gen): w¯ar-i-ki (dat), w¯ar-im¯a

talu ‘their words’

(gen). -i is also added as an oblique marker to a class of nouns of three morae each,

ending in a sonorant,

V] n, r, l, e.g. ¯uru ‘village: ¯ur-i, k¯alu ‘leg’: k¯al-i, c¯enu ‘ﬁeld’:

c¯en-i. It is not certain if this is derivable from

∗

-in with the loss of ﬁnal -n, but there is

no internal evidence for such a loss.

In Ko

da -a- and -i- have similar distribution, -a- after the common plural nouns

and -i after the masculine plural, -ru, e.g. ayli-k ‘girls’: aylik-a- (obl–gen), ayli-k-a-

‘girls’ (acc–dat), buba-r ‘parents’; bubar-i (obl–gen). -i occurs in the oblique–genitive

forms of van-i ‘that man’, ven-i ‘this man’, ayen-i ‘which man?’ Kui has -i as oblique

marker after masculine nouns in the singular and plural, e.g. neganju ‘a good man’:

negan-i (obl–gen), ¯abaru ‘fathers’: ¯abar-i (obl–gen). In most cases, the oblique stem

also functions as an adnominal or in the genitive case.

226 Nominals

rux adds -¯a- as an augment in the case of personal pronouns in the dative case,

e.g. n¯ım ‘you (pl)’: n¯ım-¯a-ge ‘to you’. In Brahui, monosyllabic nouns take the augment

a- before the sociative and genitive cases, e.g. m¯ar ‘son’: m¯ar-a-to ‘with son’, b¯a ‘mouth’:

b¯a-a-n¯a;-e- occurs as the oblique-genitive marker after all personal pronouns (except

the 2pl) before ablative, instrumental and locative cases, e.g. nan ‘we’: nan-e-¯an (abl),

nan-e-at (instr).

It is hard to interpret Tu

lu -e and Brahui -e comparatively. In any case, they are not

apparently related, since Tu

lu -e corresponds to PD

∗

-ay for which there is evidence.

∗

-i

can be reconstructed for South Dravidian II.

6.3.1.5 Summary

Looking at the permutations and combinations of the different augments, it appears

tempting to interpret these in terms of minimal constituents,

∗

tt. The inﬂectional

increments

(which in most cases are also mark

ers of the genitive case)

can be generated

by combining these in different sequences:

1. -a-: SD I (in personal pronouns), SD II: Telugu, Ko

da; ND: Ku

rux, Brahui

2. -n-: Ko

dagu in SD I; Gondi in SD II; Central Dravidian

3. -tt-: SD I (Kanna

da has a weakened variant -d-); Gondi, Pengo, Man

da;

Central Dravidian; Brahui

4. -i-: SD II

5. -a-n-: SD I, SD II: Telugu, Ko

6. -n-a-, -n-i-: Telugu

7. -tt-a-: Tu

lu (also geniti

ve), Kuvi

8. -tt-i-: SD II: Telugu, Ko

da, Kui, also Pengo and Man

da in genitive

9. -i-n-: SD I

10. -n-

tt-: Telugu, Ko

da, Pengo, Man

11. -a-n-

tt-: Tamil, Malay¯a

lam (with loss of nasal as

∗

att-inav-arr-), Telugu

∗

-a

t-i: w¯a

t-i obl of awi ‘they (non-hum)’

12. -n-

tt-i-: SD II in some restricted classes of stems, Te. illu: i

ti (<

∗

il +

tt + i), Ko

da ilu: inr-o (<

∗

il-n-do) (loc).

13. -i-n-

tt-i-: Telugu -i

ti- in numerals etc.

All possible sequences can be generated by the following schema that I proposed in

1980 in a paper (see Krishnamurti 1985: 221):

∗

n + tt → n-tt

∗

tt-V-n

∗

i-tt,

∗

tt-i

∗

i-n,

∗

n-i

∗

-i-n-tt,

∗

n-tt-i

∗

tt-i-n

∗

a-tt,

∗

tt-a

∗

a-n,

∗

n-a

∗

-a-n-tt,

∗

n-tt-a

∗

tt-a-n

It is not possible to recover the criteria underlying the use of different augments in Proto-

Dravidian. It can be done for some augments in certain subgroups, like -an in the case of

neuter demonstrative singular stems, shared by Tamil and Telugu (

∗

at-an-,

∗

it-an- etc.);

6.3 Cases 227

similarly -tt- occurs with non-personal nouns ending in -n,

∗

mara-n ‘tree’ :

∗

mara-tt-. In

any case the criteria are deﬁnitely not phonological. Ta. -a-

rr-(av-arr-ai ‘them’ neu)

and Te. -i-

t-i (¯a

r-i

ti-ki ‘to six’) can best be explained by positing loss of a nasal in

∗

-a-n-tt in Tamil and Malay¯a

lam and with retention of a nasal and with change of the

vowel (

∗

-i-n-tt-i) in Telugu.

6.3.2 Case markers

The case markers that can be clearly reconstructed for Proto-Dravidian are the accusative,

dative and genitive. The other cases do occur in all languages but the markers are

restricted to subgroups or individual languages. The distribution of case markers is

sometimes determined by the gender of the noun [± animate], [± human], etc. and

sometimes by phonological criteria, such as consonant-ending/vowel-ending. For data

I have generally followed Shanmugam (1971a) but I have independently checked his

data and analysis for individual languages and added material where necessary, from the

source publications. Paradigms

from selected languages of

each subgroup are given in

a following section to facilitate comparison. Members of a paradigm with case markers

which are not reconstructible are given in parentheses.

The syntax of cases will be dealt with in chapter 9.

6.3.2.1 Accusative case

(i)

∗

-ay

South Dravidian I This morph is found clearly in a subgroup of South Dravidian I,

namely Tamil, Malay¯a

lam, Ko

dagu, Iru

la, Ku

rumba. Traces of its retention are found

in Kui–Kuvi (somewhat uncertain) and in Brahui. It is therefore reconstructible for

Proto-Dravidian.

Old Tamil: -ai is the accusative marker, obligatorily used with [+ animate] nouns

(Shanmugam 1971a: 256), e.g. v¯e

zatt-ai ka

t¯ay ‘you see the elephant’. It is optionally

dropped

after personal pronouns in Old Tamil:

t-icin-¯or-¯e

‘those who saw

you

’. In Modern Tamil -ai becomes -e and it is obligatory with animate nouns, e.g.

enn-e ‘me’; with inanimate nouns the use of an accusative is a sign of deﬁniteness that

the speaker intends to convey, maram ‘tree’: mara-tt-e ‘the tree’ (acc).

Malay¯a

lam: -ai occurred in early inscriptions. It was replaced by -e in records from

the tenth century, e.g. ˜n¯an

avan-e

ticcu

‘I

beat

him

’. With non-human nouns its

use was optional.

Iru

la accusative case markers are -e/-ne,-ne after stems ending in -e,-e elsewhere,

e.g. v¯ele-ne ‘work’, p¨ede-ne ‘daughter’, avan-e ‘him’, gi

da-tt-e ‘plant’ (Perialwar 1978b:

9–10).

dagu: -a (< -ai) occurs after stems in a ﬁnal liquid, ava

l-a ‘her’ (acc); -na occurs

elsewhere, mara-t-n-a ‘tree’ (acc), m¯eji-n-a ‘table’ (acc).

228 Nominals

Alu Ku

rumba, the accusative case marker is -na, uli-na ‘tiger’ (acc) following

vowel-ending stems. The marker is optional with inanimate nouns.

South Dravidian II Kui: the accusative sufﬁx is -i used in the case of human nouns,

e.g. mr¯ıeni-i ‘son’ (acc): its use is optional in the case of non-personal nouns, mrahnu-

tin-i ‘tree’ (acc).

Kuvi: -i in m

reha-

-i ‘man’ (acc), ka

da- t-i ‘river’(acc); optional in [−human] nouns,

e.g. hiccu

dupdu

‘put out

the ﬁre

’.

North Dravidian Brahui: -e marks the accusative case, which appears to have been

derived from

∗

ay, e.g. d¯a shar-e ‘this village’ (acc), musi huchch-e ‘three camels’ (acc).

This is also used in a dative sense.

(ii)

∗

-Vn

This morph is attested in all major subgroups.

South Dravidian I Toda: -n is the marker for accusative; it is represented as -

n after

stems in ﬁnal retroﬂexes, e.g. ¯o

n ‘man’ (acc), k¯ux-npa

t ‘catch the girl’; n¨ım-n ‘you’

(acc).

Kota: -n occurs as the accusative

marker, obligatorily with animate

nouns and option-

ally

with non-animate ones,

l-n ‘thief’, pujg¯u

l-n ‘tigers’; ¯un mar-t-n/marm-nerc

¯o

‘he cut down the tree’. In non-initial syllables the V of -Vn is lost in Toda and Kota.

Kanna

da: -aM,-an,-¯an,-¯a occurred in the inscriptions. The long-vowel forms oc-

curred in earlier records, e.g. bi

zid¯on-¯a ‘the one who has fallen’ (acc), k¯o

te-y-an ‘fort’

(acc), kayy-an ‘hand’ (acc). Its use is optional after non-human nouns,

e.g. kar

naN-

aM → kar

nan-am ‘Kar

na’ (acc), kajjaM-aM → kajjam-am ‘work’ (acc) (Ramachandra

Rao: 58–9). In Modern Kanna

da -annu is the accusative marker, obligatorily used in

human nouns or as a marker of deﬁniteness in the case of non-human and inanimate

nouns (Sridhar 1990: 160–1).

daga: the ﬁnal nasal is lost (?) and the accusative morph looks deceptively like

a reﬂex of

∗

-ay, but it is not so, since Ba

daga is closer to Kanna

da than to Pre-Tamil,

e.g. adu-n-a ‘that’, mara-v-a ‘tree’.

lu: -nu/-n¨ı and -anu mark the accusative, e.g. ka ˜nji-n¨ıko

nola ‘take away the calf’,

en-anu ‘me’, in-anu ‘you’.

The marker is seldom lost after human nouns, but it is optional after non-human nouns. After a

noun phrase involving an attribute to the noun head the accusative is always expressed (Emeneau

1984: 76–8).

For the phonemic interpretation of the morphophonemes N, M, see fn. 20.

6.3 Cases 229

South Dravidian II Telugu: -nu/-ni, alternating with -n, are complementary morphs

marking accusative, obligatorily with animate nouns, but optionally with non-animate

ones, p¯a

ra-nu ‘Brahmin’ (acc; inscriptional), ¯ali-n(i) ‘wife’, bh¯arya-n(u) ‘wife’ (acc);

w¯ari-n(i) ‘them’. The ﬁnal vowel was optional in poetic language, but not so in Modern

Telugu. The ﬁnal vowel of the case morph is harmonized to the vowel in the preceding

syllable. In Old Telugu, sometimes the marker is dropped in poetic language in the case

of personal pronouns, nin- ‘you’ (acc), alternatively ninn-u(n), nin-un, etc.

Gondi: -n after long vo

wels, -

¯un after consonants,

e.g.

bayy¯e-n ‘mother’, k¯ond¯a-t-¯un

‘ox’ (acc), tar¯as-¯un ‘snake’ (acc). In the K

oya dialect it is -

ini/-in (Tyler 1969: 53),

e.g.

t¯a-t-ini ‘mountain’ (acc).

da: in Ko

da, Pengo and Man

da, the accusative and dative cases are represented

by the same marker, -

i in Ko

da obligatorily after human nouns, e.g. aya guruye-

osinar ‘they are bringing the Guru’, na-

i ‘me’; r¯eto

rista

‘I will release

the crab

’.

Pengo: -a

after plural stems and -i

elsewhere, e.g. key-di-

‘hand’, keyku-k-a

[?]

‘hands’ (acc), neku

r-ti-

‘dog’ (acc).

Central Dravidian The whole subgroup shares two features in the marking of ac-

cusative case: (i) -n occurs after vowel-ending stems and -Vn after consonant-ending

stems; (ii) the use of the accusative marker is optional in the case of inanimate nouns.

Kolami: -n∼-un,-n after any stem in a ﬁnal vowel, liquid or semivowel, and -un

elesewhere, e.g. ella-n ‘house’ (acc), k¯olavan-un ‘a Kolami man’ (acc).

Naik

ri (wrongl

y claimed to be a dialect of K

olami) has the same marker for

the

accusative and dative, e.g. -

/-¯u

, e.g. pul-¯u

‘tiger’ (acc), ell¯a-

‘to the house’, an-¯u

‘me, to me’ (Thomasiah 1986: 95)

Naiki (Chanda): -n ∼ -un/-on, e.g. k¯onda-n ‘bull’ (acc), pul-un ‘tiger’ (acc); optional

use in neuter nouns, e.g. kokke

¯ı v

ilupti

‘you

tore

the cloth

’.

Parji: -n∼-in (freely varying with -i before some stems) (Burrow and Bhattacharya

1953: 21), e.g. ¯enu-n

ten

‘I tied

an elephant

’, pall-in

petten

‘I picked

a fruit

’,

rol-in

ciur

‘give

the bride

’. Optional use in inanimate nouns, n¯ır

ender

‘bring

water

’.

Ollari:

n ∼ -in,

e.g.

m¯al-in ‘daughter’,

dur

-n ‘panther’ (acc).

Bhattacharya (1957:

21) says that these morphs freely vary with -

/-i

which are accusative–dative. Inanimate

nouns take the marker optionally, ¯an

kis

ton

‘I

put out

the ﬁre

’.

North Dravidian Ku

rux: -an after non-human nouns ending in a consonant, -n after

such nouns ending in a vowel, and -in elsewhere. The personal pronouns also have the

same distribution, e.g xess-an

ax¯alagd -as

‘he is sowing

rice

’, ¯a

d¯o-n

¯er¯a

‘see

that

’. The accusative is sometimes used for dative, perhaps under the inﬂuence of

surrounding Indo-Aryan (Shanmugam 1971a: 355).

230 Nominals

Malto: the case sufﬁx is -n/-in in most cases; after stems ending in -du, the ﬁnal syllable

is replaced by -a/-an, e.g. male-n ‘to the man’, maler-in ‘to the men’;

t¯u

d-a/

t¯u

d-an ‘to

the tiger’.

Summary Of the two case sufﬁxes which can be reconstructed for Proto-Dravidian,

∗

-ay and

∗

-Vn, we notice that the languages derived from Pre-Tamil (Tamil, Malay¯a

lam,

dagu, Iru

la, Ku

rumba) have a shared innovation in the loss of

∗

-Vn, while the other

South Dravidian I languages have lost

∗

-ay. This innovation provides us evidence to

show that Toda–Kota must have split off from Pre-Tamil prior to the other Pre-Tamil

descendants. The loss of

∗

-Vn in Pre-Tamil can be attributed to a period after Toda–Kota

branched off and before the palatalization rule set in Pre-Tamil, i.e. ca. third century

BCE. By that time Ko

dagu–Iru

la–Ku

rumba must also have separated from Pre-Tamil

as another sub-branch. Zvelebil (1972c) suggests a revised stemma for this subgroup

based on the retention of

∗

-ay as accusative. It is doubtful if its retention can be invoked

in setting up isoglosses for subgrouping.

The allomorph

∗

-ay is independently preserved in Brahui. Its seeming retention in

the other languages like Parji, Kui–Kuvi may be misleading, indicating the possibility

of loss of ﬁnal -n in all allomorphs.

Most languages show that the accusative is optional in the case of noun stems carrying

the gender features [−animate] or [−human] which probably deﬁned the original distri-

bution

of the two allomorphs in Proto-Dra

vidian. The representation

of the accusative

and dative by the same marker is typically found in languages which are surrounded by

New Indo-Aryan, in which there is a single marker for both these cases (Masica 1991:

§10.4).

6.3.2.2 Dative case

A geminate velar consonant

∗

-kk- is the core of the dative sufﬁx. A preceding nasal is

needed to account for the developments in some of the languages, so

∗

-nkk. The geminate

may be weakened to -k- [g] in some languages. It is used for a wide range of meanings

as goal, indirect object (listener, recipient), purpose, comparison, cause, location in time

and place, etc. It also occurs in a genitive sense

to denote adnominal relationship (in

kinship), possession (with stative predicates), direction, etc.

South Dravidian I Tamil: -kku occurs in classical poetry, e.g. iravalar-kku

¯ıyum

(Pat. 81.23) ‘will give

to the poor

’, p¯avai-kku

p¯u

kkoytu

(PN 11.4) ‘having

plucked

the ﬂowers

for the doll

’. Adnominal (genitive) uses: nampikku

makan

‘son

of Nampi

’, ena-kku

v¯ı

‘my

house

’, ¯ur-kku

nakku

‘east of

the village

’

(Shanmugam 1971a:

264

–7). Stative use

in Modern Tamil, e.g.

avar-ukku tami

z teriyum-

aa? ‘Does he know Tamil?’