Krishnamurti Bhadriraju. The Dravidian Languages

Подождите немного. Документ загружается.

7.4 T

ense

291

The loss of paired intransitive–transitive stems may be taken as a shared innovation of

the languages of the Central Dravidian group, although traces are embedded redundantly

in such pairs as Pa. ¯u

g- v.i. ‘to swing’: v.t. ¯uk-ip/-it- (see section 7.3.5). They are

preserved only in a few archaic forms in Kanna

da and Tu

lu of South Dravidian I, and

rux–Malto of North Dravidian, but totally lost in Telugu of South Dravidian II.

7.4 Tense

There are two tenses reconstructible for Proto-Dravidian, i.e. past and non-past. Non-

past includes

the habitual (present/aorist/inde

ﬁnite/generic)

and future. Some languages

have also developed a separate present tense. Perfective, durative, inﬁnitive, concessive,

conditional etc. represent aspects associated with the past or non-past tense markers.

Some of these are traditionally considered moods, but there is always a tense marker

historically traceab

le in their formation, usuall

y the past or non-past without necessaril

referring to time. A number of modal auxiliaries are added to the non-ﬁnite form of

a verb to express modes such as ability–inability, probability, prohibition, permission,

obligation, etc. The modes are normally expressed by a limited number of auxiliary

verbs, added to the ﬁnite or non-ﬁnite form of the primary verb. The nucleus of the

verb (the main meaning carrier) also has some auxiliaries (called vectors/operators/

co-verbs, etc.) added to the primary root, covering such semantic categories as transitive,

causative, reﬂexive–reciprocal and benefactive.

Complex stems are a part of the

lexical

representation of extended forms of the main verb. Almost any verb (simple, complex or

compound) can take any modal auxiliary (hence, grammatical), but only some operators

modify the meaning of the primary verb. Finite and non-ﬁnite verbs have both afﬁrmative

and negative inﬂection. A ﬁnite verb is the head of the main clause, and a non-ﬁnite

verb, the head of a subordinate clause. Morphologically the ﬁnite verb, in most cases,

carries an agreement marker with the subject phrase in gender–number–person.

7.4.1 Past markers

The past-tense markers are -t -/-tt-, -nt- (v.i.) ∼ -nt-t- (v.t.), -i- ∼ -in-, -cc-, -kk- which

could have occurred in complementary environments in Proto-Dravidian, but these are

not fully recoverable, owing to some subgroups analogically regularizing one or more of

the original allomorphs. There is complementation with some overlap between -t- and

-tt- in different subgroups. In South Dravidian I and South Dravidian II, the past stem

is the same in ﬁnite verb inﬂection (past tense) and in the formation of the past verbal

adjective (relative participle) and the gerund (perfective participle). Some languages of

Central Dravidian and South Dravidian II have the gerund formed by the addition of

∗

-(c)ci not used in the ﬁnite forms. But the palatal affricate does occur as a past marker

in Old Tamil, Toda–Kota of South Dravidian I, Telugu of South Dravidian II and in

North Dravidian; it is, therefore, reconstructed as a past allomorph of Proto-Dravidian.

Therefore,

∗

-cc-i can be taken as a sequence of two past allomorphs

∗

-cc and

∗

-i.

292 The verb

7.4.1.1

∗

-t-

In South Dravidian I and Central Dravidian and

∗

-tt- in South Dravidian II and North

Dravidian. This occurs clearly after monosyllabic roots ending in an alveolar or retroﬂex

nasal, e.g.

(1) PD

∗

tiHn-(tin-t-/tin-tt-) ‘to eat’ [3263].

(See section 7.2.1 (3).)

SD I past

∗

tin-t-: Ta. tin-(tin-¯r-), Ma. tin-(tin-n-), Ko. tin-(tid-), To. t¨ın-

(t¨ı

d-), Ko

d. tinn-(tin-d-), Ka. tin(nu)-(tin-d-), Tu. tin-(tin-d-/tin-t-);

SD II past

∗

tin-tt-: OTe. tin-(tin-

t- < tin-

t- <

∗

tin-tt-);

∗

ti-tt- with the

loss of the root-ﬁnal nasal: Go. tin-(ti-tt-), Ko

da tin-(tiR-), Kui tin-(ti-s-),

Kuvi tin-(ti-c-), Pe. tin-(ti-c-/ci-c-), Man

da tin-(ti-c-);

CD past

∗

tin-t-: Kol.

Nk. (Ch.), Pa.

tin-(tin-d-), Oll. tin-(tin-

d-), Gad.

tin-(tiy-).

We note that South Dravidian I and Central Dravidian require a past-tense marker -t-

and South Dravidian II requires -tt-. This is true of the other (C)Vn- roots,

∗∗

aHn- >

∗

¯an-/

∗

a(y)n- >

∗

e(y)n- >

∗

i(y)n- [868] ‘to say’: SD I

∗

en-t-/an-t-/in-t-; CD: en-t-; SD II

∗

an-tt-/

∗

in-tt-; ND: ¯an- (see Krishnamurti 1997b/2001a: 333).

(2) PD

∗

n-(u

t-/u

t-) ‘to eat, drink’ [600]. (See section 7.3.1 (2).)

SD I: Ta. Ma. u

n-(u

t-), Ko. To. u

n-(u

d-), Ko

d. u

n-(u

d-), Ka. u

d-), Tu. u

n (u

d- past; u

t- perfect);

SD II: Go.–Ko

da–Kui–Pe. u

n-(u

t-); Man

da un-(uc-; probably on the

analogy of tic past stem of tin-: tic-);

CD: Kol. un-(un-d-), Nk.–Pa.–Oll.–Gad. un-(un-

d-);

ND: Ku

r. ¯on-(o

d-), Malt. ¯on-(on-

d-) ‘to drink’.

Derived nouns and the causative stem in South Dravidian I, South Dravidian II require

the reconstruction of

∗

¯u

n-; this and the long vowel in North

Dravidian are explained as

arising from a Proto-Dravidian laryngeal

∗

H within the root, i.e.

∗

n-. In Malto -t-,

-y-, -Ø are the allomorphs before different personal morphs in the past tense. Under the

aspectual system the perfective (past) aspect is expressed by -t-/-c- (Mahapatra 1979:

163–9). In Ku

rux, the past-tense marker -k- is added to on

d-(on

d-k-an ‘I drank’) and

not to the root ¯on-; here, the stem alternant seems to have incorporated a historically

older past morph.

7.4.1.2

∗

-tt-

Actually

∗

t ∼

∗

tt can be taken as morpheme alternants without clearly deﬁned distri-

bution. In Tamil there are bases consisting of two short syllables, of which the second

∗

tiHn- shows a long vowel, i.e. ¯ı from iH in the case of the derived noun and in the formation

of thecausative; the laryngealis lost in non-causative inﬂection (Krishnamurti 1997b; 2001a: 333).

7.4 T

ense

293

is -

tu/-

ru, which form the past stem by geminating the ﬁnal stop, e.g. ke

tu- ‘to perish’:

t-, pe

ru ‘to get’: perr-. In traditional descriptions these are shown to have -t- past and

not -tt-, but comparatively, both South Dravidian II and Central Dravidian independently

support the past marker as

∗

-tt- in these cases. The voiceless geminate in all languages

is explained better by this reconstruction.

(3) PD

∗

t-u (i

t-) ‘to put, place’ [442]

SD I: Ta. Ma. i

tu (i

t-), Ko

d. ¨ı

d-(¨ı

t-), Ko. i

d-(i

t-), To. ¨ı

d-(¨ı

t-), Ka. i

t-) (Tu

lu past is different);

SD II: Ko

da i

d-(i

d-t-), i

t-ki- ‘to put on, wear’, i

r-(iR-t-) ‘to serve

food’, Kui i

t-a (i

ti-), Pe. i

t-(i

t-t -), iz-(is-t-) ‘to put in, as salt’, Man

da i

t-.

Note that Proto-Dravidian

∗

t- has merged with

∗

-t- in South Dravidian II

(section 4.5.5.3, see ety. (78)); Te. i

du and Go. irr- do not have past stems

like the others.

CD: Kol. ¯ı

d-(i

t-), Nk. (Ch.) i

r-(i

t-), Pa. i

d-(i

t-), Gad. ir-(i

t-).

The languages of Nor

th Dravidian have no cognates

for this verb.

(4) PD

∗

caH-(ca-tt-) ‘to die’.

∗

caH > c¯a- ∼ cay-/cey- >

∗

key- (ND) [2426].

(See Krishnamurti 1997b/2001a: 331.)

SD I: Ta. c¯a-(cett-), Ma. c¯aka (ca-tt-), Ko

d. c¯a-(ca-tt-), To. soy-(so-t-),

Ka. s¯ay (sa-tt), Tu. saipini ‘to die’;

SD II: Ko

da s¯a-(s¯a-t-), Kui s¯ava (s¯a-t-), Kuvi–Pengo–Man

da h¯a-

(h¯a-t-);

CD: Pa. cay-(ca-˜n-), Oll. say-(sa-d-);

ND: Ku

r. kh¯e‘- (ke-cc-), Malt. key-(ke-cc-), Br. kah-(kask-).

lu of South Dravidian I , Telugu of South Dravidian II, the remaining Central Dravidian

languages and the languages of North Dravidian follow different routes in past formation.

Only Gondi has replaced this root by a borrowed word.

(5) PD

∗

k¯a-(k¯a-tt-) ‘to protect, guard’ [1416].

SD I: Ta. k¯a-(k¯a-tt-), Ma. k¯akkuka (k¯a-tt-), Ko

d. k¯a-(k¯a-t-), Ko. k¯a-v-

(k¯a-t-), To. k¯o-f-(k¯o-t-), Ka. k¯a-(k¯a-d-); Tu. k¯a-puni ‘to watch’;

SD II: Te. k¯a-cu v.t., k¯a-pu n., Ko

da k¯a-pu n., Kui k¯a-p-a (k¯a-t-), ‘to

wait for’, Kuvi–Pe.–Man

da k¯a-(k¯a-t-) ‘to watch, wait’;

CD: Kol.–Nk. kay-(kay-t-) ‘to herd cattle’, Pa. Oll. k¯a-p-(k¯a-t-) ‘to

wait’;

ND: x¯a-p- ‘to protect, guard’, Malt. q¯a-p- ‘to watch, wait for’, Br. xw¯a-f-

‘to graze’.

The past -tt- is supported by South Dravidian I, South Dravidian II and Central

Dravidian languages. Again, Kanna

da–Tu

lu of South Dravidian I, Telugu–Gondi of

294 The verb

South Dravidian II and North Dravidian have adopted a different way of forming the

past. We can clearly see that the inﬂected non-past stem has become the base in many

languages (see Kota–Tu

lu, Kui, Parji–Ollari and all North Dravidian languages).

It has been pointed out that there was a class of verbs which combined tense and

voice. Such verbs formed past transitive from

∗

-nt-t- which lost the nasal in several

South Dravidian languages and became identical with the additive past -tt-:

(6) PSD

∗

zi (-nt-) v.i. ‘time to pass, become spent, ruined’, (-nt-t-) v.t. ‘to

spend/waste time, void excreta, reject’ [1356].

SD I: Ta. ka

zi (-nt-/-tt-) v.i/v.t., Middle Ta. ka

zi-cc-(< ka

zi-tt-) past,

zi-cc-al n. ‘diarrhoea’, Ma. ka

ziyuka v.i., ka

zikka v.t., To. ko

dy-(ko

d-s-)

‘time to pass’, ko

dc-(ko

d-ˇc)

‘chase out of sight’, Ko

d. kayy-(kay-˜nj-)

v.i. ‘time passes’, kay-(kay-c-) v.t. ‘to pass time’. In this subgroup of South

Dravidian I the past sufﬁx -tt- got palatalized to -cc-. Kanna

da and Tu

have cognates but the past markers are different.

SD II: Ko

da ka

rs-(ka

rs-t-) ‘to defecate’, Kui k

rahpa-(k

rah-t-), Kuvi

rah-(g

rah-t-), Pe. k

rac-(-c-), Man

da g

rah-.

The uninﬂected stem in South Dravidian II looks like the past stem of South Dravidian I

incorporating the palatalized past sufﬁx.

7.4.1.3

∗

-nt-

In South Dravidian I, -nt- is the past-tense marker of mostly monosyllabic or disyllabic

roots ending in /u i y r

z/. It happens that most of them are intransitive. The corresponding

transitive forms have

∗

-nt-t-(> -tt- in SD I). It also occurs in Pre-Tamil descendants

(Malay¯a

lam–Ko

dagu–Kota–Toda) in a special class of stems (Arden’s seventh conjuga-

tion, Tamil Lexicon’s twelfth conjugation), in which the past is marked by -nt- and the

non-past by -pp-. Most of these bases (some seventy) are disyllabic ending in -a. This

class of verbs contrasts with the former class in which -nt- and -v- stand for past and

non-past in the intransitive and -tt- and -pp- for the transitive, respectively. In Middle

Tamil, Malay¯a

lam, Ko

dagu, Toda and Kota, following a front high vowel i or y, the past

markers -nt- and -tt- were palatalized to -˜nc-(Ta.˜nj, Ma. ˜n˜n,Ko

d. ˜nj,Ko.c,To.s) and

-cc-, respectively.

Kanna

da preserves -nd- in six verbs following the (C)V- allomorph of the root, n¯o-‘to

suffer’: no-nd-, bar- ‘to come’: ba-nd-, tar- ‘to bring’: ta-nd-, m¯ı- ‘to bathe’: mi-nd/m¯ı-d,

b¯e- ‘to be hot’: be-nd-; after other monosyllabic bases, the past is -d-(DVM: 198ff.).

In Toda Stem I corresponds to the intransitive stem in Tamil and Stem 2 corresponds to the past

stem of the transitive. Thus Ta. ka

zi =To.ko

dy-, Ta. ka

zi-cc-=To.ko

dc-. The sandhi rules account

for the past stems of Toda, t + y → s, c + i → ˇc.

7.4 T

ense

295

Parallels are noticed in Central Dravidian (mainly Parji and Gadaba), e.g. Pa. ver-(ve-

˜n- <

∗

we-nj-), Oll. var-(va-n-). Even where South Dravidian I languages do not have

-nt-, Parji–Ollari–Gadaba had a past -nd-, Pa. koy-(ko-˜n-) ‘to cut, reap’. These roots in

Old Tamil had -t- as past marker, cey-(cey-t-) ‘to do’, but Modern spoken Tamil has

sey-(se-˜nj) presupposing a past form with -nd-(sey-nd-).

In actual conjugation, the past-tense morph -nt- is mainly conﬁned to South Dravidian

I; there is, however, evidence of the incorporation of -nd- as a stem formative in several

languages (see section 5.4.5). It can, therefore, be reconstructed for Proto-Dravidian.

(7) PSD

∗

iru (iru-nt-) ‘to exist, live, belong to (with dative subject)’ [480].

SD I: Ta. iru (iru-nt-) v.i., iru-ttu ‘to cause to sit’, Ma. iri (iru-nn-) ‘to

remain, to sit’, iru-ttuka ‘to seat’, To. ¨ır-(¨ı θ -), Ko

d. ir-(i-˜nj- <

∗

ir-nd-),

Ka. ir-(ird-/-idd ), Tu. ippuni (itt-);

SD II: Kui r¯ında (r¯ınd-i) ‘to be stable’, r¯ıspa (r¯ıs-t-) ‘to set in position’.

Notice that Kui has incorporated a Proto-South Dravidian past morph as a stem formative

and also has a corresponding transitive which falls into the class of NP (v.i.), NPP (v.t.)

with loss of tense meaning.

(8) PD

∗

cal-(can-t- <

∗

cal-nt-) ‘to go,

pass, occur

’ [2781].

SD I: Ta. cel (ce

n-r-), cel-uttu ‘to cause to go’, Ma. cel-ka (cenn-) v.i.,

cel-uttuka v.t., Ko. cal-(cad-). To. sal-(sad-), Ka. sal-(sa-nd-), Tu. sall-uni

‘to be valid, ﬁt’, sand-uni ‘to pass as time’;

SD II: Te. canu (ca

t-), Go. son-, sond-, hon-(hott-), han-(hat-),

da son-(soR- <

∗

sott- <

∗

son-tt-); Kui sal-(sas-), Kuvi hal-(hacc-),

Pe.–Man

da hal-(hac-);

CD: Nk. ca-(ca

d-), Pa. cen-(cen-d-), Oll. sen-(sey-, sen

d-).

South Dravidian I clearly has -nt- sufﬁx. Tu. sand-, Go. sond- point to the restructuring of

past stem as the normal base. South Dravidian II has -tt- as past and Central Dravidian -d-;

Ollari–Gadaba -yy- forms point to an analogical extension of the key- ‘to do’ paradigm

to other monosyllabic verbs. The rounding of the root vowel (a > o) is an idiosyncratic

change in Gondi–Ko

da for which we do not know any reason at present.

(9) PD

∗

waH-r (imp. w¯a-, past wa-nt- <

∗

war-nt-) ‘to come, to happen’ [5270].

SD I: Ta. var-u (v¯a imp sg; past va-nt-, inf var-a-) v.i., var-u-ttu v.t. ‘to

cause to come’, Ma. var-uka (imp v¯a-, past va-nn-) v.i., var-u-ttuka v.t., Ko.

v¯ar-(v¯a-, va-d-), To. p¯or-(po-d-), Ko

d. bar-(b¯a-, ba-nd-), Ka. bar-(b¯a-,

ba-nd-), Tu. barpini ‘to come’;

SD II: Te. wacc- (imp r¯a- <

∗

wr¯a- <

∗

war-¯a), past wa-cci-; earlier past

marker -cci got incorporated into the basic stem through reanalysis, see

296 The verb

TVB:162–3); Go. w¯ay- (imp war-¯a, past w¯a-t-), Ko

da w¯a- ( imp sg r

a-,

past w¯a-t-), Kui–Kuvi–Pe.–Man

da v¯a-(v¯a-t-);

CD: Kol. var- (imp v¯a-, past va-tt-), Nk. (Ch.) var-/va-(va-t-), Pa. ver-

(ve-˜n- < ve-˜nj- < ver-nd- < var-nd-; radical a becomes e before an alveolar

in Pre-Parji, see section 4.4.5), Oll. var-(vad-, van-), Gad. v¯ar- (imp v¯a-,

var-, past va-dd-/va-nn-);

ND: Ku

r.-Malt. bar- ‘to arrive’, Br. banning (imp. ba, past bass-).

South Dravidian I and Parji of Central Dravidian agree in having -nt- as a past marker.

The other Central Dravidian languages have analogically restructured their past forms

with either -d-or-tt- as the marker. South Dravidian II (except Telugu) has regularized

-t-(<

∗

-tt-) as the past marker. North Dravidian has regularized -cc- as the past-tense

marker.

(10) PD

∗

k¯ay (-nt-) v.i. ‘to grow hot’, (-nt-t-) v.t. ‘to boil, heat’ [1458].

SD I: Ta. k¯ay (-nt-/-tt- ∼ -cc-) v.i./v.t.; also k¯ay-ttu/k¯ay-ccu v.t. ‘to boil,

ignite’, also k¯a-ntu v.i. (with an erstwhile past marker -nt- incorporated

into the stem, following

the loss of root-

ﬁnal -y, as expected); Ma. k¯ay-uka

v.i., k¯ay-kka/k¯ay-cc-uka v.t.; k¯a-nt-uka ‘to be hot’, Ko. k¯ay (k¯ac-) v.i., k¯ac-

(k¯ac-) v.t., To. k¯oy-(k¯os-) v.i., k¯oc-(k¯oˇc-) v.t., Ko

d. k¯ay-(k¯a-˜nj-) v.i., k¯a-c-

(k¯a-c-i) v.t., Ka. k¯ay-(k¯ay-d-/k¯a-d-), Tu. k¯ay-uni v.i.;

SD II: Te. k¯ayu (k¯as-i) ‘to shine’, Go. k¯as- v.i., k¯as-is/-ih- v.t., Ko

da k¯ay-

(k¯ay-it-) ‘to warm oneself at ﬁreplace’; (restructured stems occur: k¯a-mb-

v.i., k¯a-p- v.t.), Kui k¯ag-, Kuvi k¯ay- v.i.; (restructured bases: k¯a-nd- v.i.,

k¯a-t- v.t.), Pe. Man

da k¯ay- v.i., Man

da also k¯a-nd- ‘(sun) to be hot’;

CD: Nk. (Ch.) k¯ay ‘to be hot’, Pa. k¯a-p-ip- (past - p-it) v.t., Oll. k¯ay-p-

(k¯ay-t-) v.t., Gad k¯ay-kil ‘fever’;

ND: Ku

r. x¯ay- ‘to dry up’, xai-d- v.t., Malt. q¯ay-/qeyr- ‘to be boiled’.

Although only South Dravidian I has evidence

of -

nt- as past marker, the fact that

South Dravidian I

and South Dravidian II have a restructured verb

k¯a-ntu (see Tamil,

Malay¯a

lam, Pengo, Man

da) shows that its use was much more extensive at an earlier

stage of Proto-Dravidian.

7.4.1.4

∗

-i- ∼

∗

-in-

These sufﬁxes occur generally after disyllabic and trisyllabic stems of three or more

morae, ending in an obstruent, i.e. -P, -PP, NP followed by the non-morphemic -u. The

Tolk¯appiyam says that -i occurs as the past marker after verb stems ending in /u i ai

nrl

l/, citing

verbal adjectives ending in -iya/-i

na, e.g. peyar-iya, coll-iya, aru

l-iya. After the ﬁnal short vowel

-u,-i

na occurs, ¯o

t-ina (Israel 1973: 33–5).

7.4 T

ense

297

reﬂexes of these are lost in Gondi and Pengo of South Dravidian II, Kolami–Naiki of

Central Dravidian and Brahui of North Dravidian (DVM: §2.26, pp. 214ff.).

Old Tamil has -i

n- as past marker in ﬁnite verbs of the above canonical structure, e.g.

v¯a˙nk-i

n-¯en ‘I bought’, but v¯a˙nk-i ‘having bought’ (perfective participle). This is true of

all verbs ending in two consonants (-PP, -NP, -

lu,-

nu) following a short vowel, or of

)

/(C

)VC

V where the ﬁnal consonant, C

, is a stop or -v-. Kota has -y-

corresponding to Pre-Tamil -i- as a perfective marker, e.g. namb- ‘to believe’: namb

-y-, amk- ‘to press’: amk-y-. Toda also has -y- in the corresponding position, nob- ‘to

believe’: nob-y-; Ko

dagu has -i just like Tamil–Malay¯a

lam. In Kanna

da -i is combined

with a dental sufﬁx in denoting the past tense in ﬁnite verbs, m¯ad-id-anu ‘he did’, but

-i- occurs as perfective marker, e.g. m¯a

d-i ‘having done’. Tu

lu has -y as the past marker

of immediate past of some classes of verbs, e.g. pari- ‘to drink’: pari-y-. In the distant

past -t- is the tense marker, e.g. pari-t-.

In South Dravidian II, Old Telugu has -(i)ti- in past ﬁnite verbs except in the third

person; e.g. Te. 1sg cepp-iti-ni ‘I said’, 1pl cepp-iti-mi ‘we said’, 2sg cepp-iti-wi ‘you

said’, 2pl cepp-iti-ri ‘you said (pl)’, 3sg and inanimate pl -e-, cepp-e-nu ‘he/she/it/they

(neu) said’, 3h pl cepp-i-ri. It seems that -ti was the original past (cf. win-

ti-ni ‘I heard’)

and the -i- preceding -ti could be epenthetic. Pure -i-/-iy- occurs in the 3sg -e-n(< iy-

an;-an 3m sg) in which the gender–number morph got obscured because of sandhi

and hence was generalized as the 3sg and inanimate plural. The perfective participle is

formed by adding -i, e.g. cepp-i ‘having said’. A number of stems ending in -c and -s in

Telugu, like pilucu ‘to call’, have ﬁnite and non-ﬁnite forms like pili-ci ‘having called’,

pili-c-en ‘he/she/it called’, c¯eyu ‘to do’: c¯e-si ‘having done’, etc. In these, historically

ﬁnal -ci and -si were the original past markers. The wider use of -i as past participle

marker has led to a reanalysis of these constructions by treating -c/-s as part of the stem

(TVB: §2.85). Gondi and Ko

da also have

∗

-ci as the perfective marker in synchronic

grammars, e.g. Go. son-j¯ı ‘having gone’, k¯ı-s¯ı ‘having done’, Ko

da son-si, ki-zi id.

da has -it as past marker (i + t) in ﬁnite verbs, broadly after stems ending in an

obstruent, t¯o

r- ‘to appear’: t¯or-it-an ‘he appeared’. In Pengo the participle is formed by

adding -si/-zi/-hi, u

n-zi ‘having drunk’, k¯uk-si ‘having called’, ta-hi ‘having brought’,

nil-ci ‘having stood’ (cf. Ko

da ta-si, nil-si). Kui also has -it- as past marker in ﬁnite

verbs after stems ending in an obstruent; the past participle sufﬁx is -sa/-ja (after nasal

or vowel-ending stems, -a elsewhere). Kuvi has -it ∼ -t as past markers, e.g. ajj- ‘to

fear’: ajj-it-. The perfective participle is formed by adding -ca/-ja/-sa/-ha, e.g. ven-ja

‘having heard’.

Parji has -i- ∼ -ci as a marker of the gerund, cen-i ‘having gone’, ver-i ‘having

come’, but nil-ci ‘having stood’. Ollari has -i alternating with -si/-zi in forming the past

participle, e.g. s¯u

r-i ‘having seen’, un-zi ‘having drunk’. In Gadaba -i occurs in the

perfective particple as an alternant of -ji ∼ -ci, e.g. in-ji ‘having said’, but v¯ar-i ‘having

298 The verb

come’. The ﬁnite verbs in the past do not use the vowel -i, but -n occurs in Class 6, e.g.

v¯ar ‘to come’: past stem va-n-o-, k¯e- ‘to do’: ke-no-. This -n can be related to the morph-

complex -in of South Dravidian I. Kolami and Naiki do not have anything comparable

involving -i.

Malto has -y- as a past marker in the third person; in other environments -t- occurs in

the past, e.g. amb- ‘to leave’: amb-y-ah (3m sg), amb-y-ad (3f sg), amb-y-ar (3hum pl),

but amb-t-an (1sg). Malto also uses -y- as a completive marker, e.g. in the past it oc-

curs with the past tense morph -t-/-c- and in non-past with -d-/-n-, amb-iy-t-an ‘I left

off’ (Mahapatra 1979: 163–74). Droese uses -i as perfective marker, on

d-i ‘having

drunk’.

South Dravidian I and South Dravidian II use -in-/-i- as markers of the past tense/

aspect in ﬁnite and non-ﬁnite verbs; -i- occurs in combination with past -tt- in several of

the South Dravidian II languages. In Central Dravidian and North Dravidian it is found

only in non-ﬁnite verbs denoting past/perfective aspect. It appears that the dental sufﬁx

has spread at the expense of this form.

7.4.1.5

∗

-cc-

In early literary Tamil -ici

n- occurs as a redundant marker of past in a special class

of ﬁnite verbs, e.g. e

nr-icin ‘I said’, vant-icin ‘I came’, perr-icin ‘she obtained’. There

is no (g)np marker, but the inﬂected forms occurred in 1sg, 3sg, or 2 imperative sg.

Pronominalized nouns based on these forms occurred in the third person in all genders

and numbers, e.g. pirint-ici

n-¯o

l ‘she who separated’, ka

t-icin-¯or ‘those that saw’. This

form is interpreted as a sequence of two past allomorphs

∗

-cc and -in (DVM: §2.27).

Emeneau (1957b: §§39–56) surveys the past-tense markers in the whole family and

proposes that Toda and Kota, along with the relic usage of Tamil, have also inherited the

palatal/sibilant as a past marker, To. 1sg p¯ı-ˇs-pen ‘I went’, Ko. vad-c-¯e ‘I came’.

Telugu has a -cu as the ﬁnal syllable of bases of three morae, e.g. pilucu ‘to call’,

racu ‘to open’, ¯e

dcu ‘to cry’, ¯ı

dcu ‘to drag’ and the past forms are pilic-e-nu ‘he called’,

pilic-i ‘having called’, etc. Cognates from the other languages in South Dravidian lack

this

stem-

ﬁnal

syllable, Ta.

li, te

ra, a

zu, i

zu, respectively. I interpreted -cu in these

forms as a consequence of the incorporation of an original past marker

∗

-c(c)i which

occurs synchronically in South Dravidian II and Central Dravidian (TVB: §§2.84–90).

The alternation of -y/-s in tri-moraic bases that end in -yu, e.g. orayu ‘to rub’: oras-i

‘having rubbed’, c¯eyu ‘to do’: c¯es-i ‘having done’, etc. has also been explained as

resulting from incorporating the past allomorph -si (following the loss of -y, ora-si <

oray-ci-), retained intact in the other South Dravidian II languages (TVB: §§2.39–50).

In some cases Kanna

da, Tu

lu and even Tamil–Malay¯a

lam have cognates with -Vcu as

a formative, where Telugu has -Vyu, e.g. Te. enayu v.i./v.t. ‘to mix, mingle’: Ta. i

n-ai

‘to join’, Ma. in-ayuka, Ka. e

n-asu ‘to add together’; Te. al-ayu ‘to be tired’: Ta. al-acu,

7.4 T

ense

299

Ma. al-asuka, Ka. al-asu,Tu.al-asuni, Malto al-esi ‘sweat, heat’ (see Emeneau 1957b:

§48). Following the general trend in Dravidian of reanalysing former tense markers as

formatives, adequately illustrated earlier, it could be that there was a widely used palatal

sufﬁx as the past allomorph in the whole family and that it got incorporated into the

stem as a formative, later weakened to -y- in South Dravidian I (for -c- [-s-] > -y/-Ø,

see section 4.5.1.3). I have also cited more examples of South Dravidian II and North

Dravidian languages incorporating

∗

-cc as a formative, e.g. Te. par-acu ‘to ﬂee’: Malt.

par-c-, Te. ka

d-acu ‘to cross, time to pass’: Kui g

r¯as-(TVB: §2.89–90). While discussing

the distribution of -i , I gave examples for the occurrence of -ccV as a gerund in the other

languages of South Dravidian II.

Even in Central Dravidian several verb stems end in -c/-s, not found in the other

subgroups, Kol. -Nk. ars-, Pa. narc-, Oll. nars-, Gad. narc- ‘to fear’: Nk. ari, Pa. Oll. Gad.

nar n. ‘fear’ [3605] (see DVM: 223–4). These cases are covered by the same argument

as above.

rux–Malto–Brahui synchronicall

y employ -

cc as a past mark

er extensively, some-

times with another past marker -k, or other times all by itself, e.g. Type VI verbs take

-c- past in the 3sg rah- ‘to be’: rah-c-as ‘he was’; in the 1sg/pl and 2sg/pl the stem is

rah-ck-. Verbs representing ‘the historical past’ take -c-/- j- as past marker, e.g. Ku

khe- ‘to die’: ke-cc-, ci

- ‘to give’: ci-cc-, man- ‘to become’: man- j-, bar- ‘to come’:

bar-c-; Malt. key- ‘to die’: ke-c-, ci- ‘to give’: ci-c-, bar- ‘to come’: bar-c-; Brahui has

-ss- correspondingly, man- ‘to become’: mass-, bann- ‘to come’: bass-.

7.4.1.5.1 Possible relationship between

∗

-in and

∗

-cin Meenakshisundaran (1965:

84–5) says that Tolk¯appiyam takes -ci

n as the basic form of the sufﬁx, treated as an

‘expletive’ of the second person, although ‘he also states that this expletive can be used

in the other persons’ (cites Tolk. 759–60). Meenakshisundaran suggests that -ici

n with

the loss of -c- in Pre-Tamil would give rise to -iin. ‘This will give us -in the past tense

sign, as found in Tamil’ (1965: 85). He also traces the long -¯ıi occurring as past participle

marker in Cankam Tamil to -iin derived from -ici

n. In any case it is quite tempting to

ﬁnd some relationship between the past -in- and the relic usage of

∗

-cin- with the loss

of c-.

The past markers -

ci(n) ∼ -i(n)

(in South Dravidian II and Central Dravidian)

could have lost the ﬁnal -n which could be quite normal. We cannot presently resolve

the question whether -cc- and -in are independent past markers, and whether -cc-in is a

combination of both. There is more evidence to consider it as a morphological complex,

because Ku

rux and Malto, which do not lose

∗

c-, have a reﬂex of past

∗

-i.

7.4.1.6

∗

-kk-

Tamil has four disyllabic

roots ending in -

ku, puku ‘to enter’ [4238], naku ‘to laugh’

[3569], taku ‘to ﬁt’ [3005], toku ‘to aggregate, assemble’ [3476], whose past stems are

300 The verb

pukk-, nakk-, takk- and tokk-. Ramaswami Aiyar (1938: 750) puts these with stems of the

type in (C)V

tu/-

tu and says that their past stems are formed by gemination of the root-ﬁnal

stop. He suggests that the gemination may have been the result of ‘sandhi action of the past

sufﬁx’. Subrahmanyam (1971: 103), following this suggestion, derives gemination in

naku, etc. also by sandhi of k + t, which is neither phonologically motivated nor possible

in Dravidian. In South Dravidian, only Koraga, a minor tribal language adjacent to Tu

lu,

forms the past by the addition of -k. The Onti dialect uses -k-/-kk-/-g- as past markers,

e.g. ku

tu ‘to beat’: 1sg ku

tu-g-e, 1pl ku

tu-g-a, 2sg ku

tu-g-a, 2pl ku

tu-g-er¨ı,3msg

tu-g-i, 3f/neu sg ku

tu-g-u, 3m/f pl ku

tu-g-er¨ı ; in Tappu also the past sufﬁx is -g-;

the 3f/neu sg sufﬁx is -¨ıd¨ı instead of -u; in the Mudu dialect, again, the difference is

only in the personal sufﬁxes, 1sg/pl -e/-u, 2sg/pl -a/-r¨ı, 3mf sg -i,fn-¨ıd¨ı,mfpl-r¨ı. The

past sufﬁx is -k- after monosyllabic bases, e.g. ¯ı ‘to give’: ¯ı-k-e ‘I gave’; it is -kk- after

stems in ﬁnal i or e, e.g. negi-kk-i ‘he said’, but j¯ek¨ı-g-e ‘I washed’ (Bhat 1971). In my

opinion this is an innovation in Koraga of giving a past meaning to a non-past sufﬁx just

like Kota, which uses - p-(<

∗

-pp), a non-past marker, as a past marker. Note that the

non-past marker in Koraga is -n- which derives from

∗

-n.

ND: Ku

rux uses -k-/-

- as the past marker in ﬁnite verbs in the ﬁrst and second persons,

es- ‘to break’: 1sg es-k-an, 1pl es-k-am, 1pl (inc) es-k-at, 2sg es-k-ai, 2pl es-k-ar,3m

sg es

as ‘he broke’, 3f sg es

¯a, 3m/f pl es

ar (Hahn 1911: 45). Some verbs have -ck-

(combination of two past markers c + k) as the past marker, ra

-ck-an ‘I remained’,

etc. (Hahn 1911: 50). Brahui also has -k- as the past marker beside -s-/-ss- in different

classes of stems (Emeneau 1962d: 23).

7.4.1.7 Summary

Four sets of allomorphs surface as markers of the past tense in Dravidian: (1) -t- ∼ -tt- ∼

-nt-, (2) -i- ∼ -in-, -cc-in, (3) -cc-, (4) -kk-. The last one, -kk-, occurs only in North

Dravidian

and is geographically complementary to groups (1) and (2). In Ku

rux–Malto

some conjugations show either (3) or (4), but some complex morphs involve both (3)

and (4). One of the isoglosses binding North Dravidian from others is the loss of the

dental group (1) as past marker in most items while generalizing (3) and (4).

Following broadly the distribution of these in Old Tamil texts, we can say that in South

Dravidian I -t- and -tt- occur in complementary environments, -t- after monosyllabic

roots with a short radical vowel ending in /n

z y/; since /r

z/ do not occur root-ﬁnally,

the

y have to be followed by the vowel -

e.g.

tin- ‘to

eat

’, u

n-‘to

drink

’, ce

- ‘to

’,

Ramaswami Aiyar (1938) says that the gemination of the ﬁnal stop in four disyllabic roots (see

section 7.4.1.6), puku-(puk-k-) ‘to enter’, etc. derives gemination as a result of sandhi of ﬁnal

-k with the past -t (

∗

-kt- > -kk-), but phonologically this is an impossible type in Dravidian. DVM:

103 endorses this view. I have explained that the use of a velar stop for past in Koraga is an

innovation which has nothing to do with the ND -kk- (section 7.4.1.6).