Krishnamurti Bhadriraju. The Dravidian Languages

Подождите немного. Документ загружается.

7.7 Non-ﬁnite

verbs: past-stem

based

331

n-ru ‘having eaten’; p¯ar- ‘to see’: p¯ar-ttu ‘having seen’, k¯e

l- ‘to hear’: k¯e-

tu (<

∗

k¯e

ttu) ‘having heard’, kal- ‘to learn’: ka

rru (<

∗

kat-tu <

∗

kal-tt-) ‘having learnt’; aka

z- ‘to

dig’: aka

z-ntu ‘having dug’, kol- ‘to destroy’: ko

n-ru (<

∗

kol-nt-) ‘having destroyed’,

veku

l- ‘to be enraged’: veku-

tu (< veku

l-nt-) ‘having been enraged/roused’. Another set

of markers in ancient Tamil is -i which derives from PD

∗

-i, and several others, which are

peculiar to Classical Tamil, namely -¯ıi,-¯aa,-¯uu, and -pu. We are not sure of the phonetics

of the extra long vowels or vowel sequences not allowed by Dravidian phonotactics. They

are considered simple vowels, lengthened rhetorically, and therefore not phonemic, like

‘truly’ in English. Meenakshisundaran says that OTa. -iin- can be taken to derive from

-ici

n- with the loss of -c-(-isin- > -ihin- > -iin-) (see section 7.3.1.5.1). In Gondi–Ko

the past participial

suf

ﬁx is

∗

-ci- but in

Kui

–Kuvi it

∗

-ca-. It is

not certain if Old

Tamil vowel sequences are to be traced to

∗

-aha-,

∗

-uhu- and

∗

-ihi- with Ø < -h- < -c-

and with identical vowels before and after. In any case, such an ancient phenomenon as

double nuclei in Old Tamil writings requires a satisfactory historical and comparative

explanation and solution. The examples cited from Ramaswami Aiyar (1938: 759–60)

are to

z-¯aa, beside to

zu-tu (PN) ‘having worshipped’, i

t-¯uu, beside i

tu ‘having placed’

(PN), tar-¯ıi, beside ta-ntu ‘having given’ (KT).

2. Malay¯a

lam: the perfective participle is formed by adding -u to the past stem, van-nu

∗

va-ndu) ‘having come’, cey-tu ‘having done’, ka

tu ‘having seen’. This contrasts

with the ﬁnite verb only in the ﬁnal vowel, vann

‘(subject)

came

’.

4. Kota: the past stem takes -

t in

forming the past participle. This

appears to be a

generalization of the -

t- allomorph (which

occurs after -

n ﬁnal stems) of the

past stem,

va-d-

t ‘having come’, id-t ‘having said’, tin-kc-t ‘having fed’.

5. Toda: here also the past participle is not identical with the past stem but is formed

by adding a sibilant to the past stem, e.g. pi

c ‘having gone’ (past stem p¯ı-), poz ‘having

come’ (past stem pod-).

6. Kodagu: the past participial marker is -t¨ı /-it¨ı added to the past stem, e.g. bar-‘to

come’: ba-nd-it¨ı ‘having come’, m¯a

du- ‘to make’: m¯a

d-i-t¨ı.

7. Kanna

da: like Tamil the past stem ending in -tu or -i is also the perfective participle,

e.g. k¯e

l- ‘to hear’: k¯e

l-du ‘having heard’, bar- ‘to come’: ba-ndu ‘having come’, ¯a

du-

‘to dance’: ¯a

d-i ‘having danced’ etc.

8. Tu

lu: the perfective participle is formed by adding -t¨ı/-d¨ı, sometimes both, to the

verb stem, kal- ‘to learn’: kal-t¨ı /kal-t¨ıd¨ı ‘having learned’, t¯u- ‘to see’: t¯u-d¨ı ‘having

seen’, bar- ‘to come’: bat-t¨ı /bat-t¨ıd¨ı ‘having come’.

7.7.1.2 South Dravidian II

In this subgroup the perfective participle is not identical to the past-tense stem as in

South Dravidian I, but to a sequence of Proto-Dravidian past

∗

-cc and

∗

-i, another past

morph (section 7.4.1.5). The vowel is -a in three languages. It is likely that

∗

-cc-i was

332 The verb

the original marker and some of the languages replaced the vowel by -a owing to vowel

harmony with the stem vowel.

9. Telugu: the perfective participle is formed by adding -i, e.g. pe

tu- ‘to put’: pe

t-i

‘having put’, winu- ‘to hear’: win-i, etc. Historically, some verbs originally had -ci/-si

as the past marker leading to -c/-w or -s/-y alternation in paradigms between the past

and non-past, pilucu- ‘to call’: pili-ci ‘having called’, pila-w-aN inf ‘to call’, c¯eyu ‘to

do’: c¯e-si ‘having done’: c¯ey-aN inf ‘to do’. But these stems are reanalysed with -i as

past and the alternations were attributed to the stems before past and non-past sufﬁxes

(see TVB: 162–4).

10. Gondi: in Gondi -ci/-ji/-si are phonologically conditioned allomorphs used as

markers of the perfective participle, e.g. toh- ‘to show’: toh-ci ‘having shown’, son-

‘to go’: son-ji ‘having gone’, s¯u

r- ‘to see’: s¯u

r-si ‘having seen’. The southern dialects,

bordering on Telugu, also have the -i variant, e.g. ¯u

d- ‘to see’: ¯u

d-i ‘having seen’

(southeastern dialect).

11. Ko

da: the perfective participle is marked by -zi/-si/-i which occur in comple-

mentary environments, e.g. toR- ‘to tie’: toR-si- ‘having tied’, pi

d- ‘to carry’: pi

d-zi

‘having carried’, as- ‘to hold’: as-i ‘having held’.

12. Kui: the perfective markers are -sa/-ja/-a again in complementation (partially

phonological and partially morphological), ¯a- ‘to become’: ¯a-ja ‘having become’, tin-

‘to eat’: tin-ja ‘having eaten’. -i occurs alternatively in a limited number of verbs instead

of -a, ¯aj-a/¯aj-i ‘having cooled’.

13. Kuvi has similar morphs with similar distribution, i.e. -sa/-ha/-a,-ca/-ja (dial),

e.g. h¯ı- ‘to give’: h¯ı-ha ‘having given’, t¯os- ‘to show’: t¯os-sa ‘having shown’, ve-‘to

beat’: ve-ca ‘having beaten’, ven- ‘to hear’: ven-ja ‘having heard’; p¯ay- ‘to beat’: p¯ay-a

‘having beaten’.

14. Pengo: the gerund markers are -si/-zi/-hi/-i; Burrow and Bhattacharya also give

-ci/-ji as variants, but it is

not certain if

c/s and j/z contrast in the same dialect,

e.g.

as-

‘to seize’: as-si ‘having seized’, k¯uk- ‘to call’: k¯uk-ci/-si/-hi ‘having called’ (apparently

in different dialects), k¯a - ‘to watch’: k¯a-hi ‘having watched’, v¯a- ‘to come’: v¯a-zi ‘having

come’, koy- ‘to cut’: koy-ji ‘having cut’. A variant with the -a vowel also occurs rarely,

e.g. ta- ‘to bring’, ta-ha v¯a- ‘get it and come, bring it’.

7.7.1.3 Central Dravidian

15. Kolami: the gerund marker is -¯ut in the Adilabad dialect and -t in Wardha; an

alternative mode is by adding -na to the past stem in -t, s¯ı-t -na ‘having given’; Adilabad

siyy-¯ut ‘having given’.

16. Naiki (Chanda): the perfective participle is formed by adding -un to the past stem

in -t, pak-t-un ‘having beaten’, tin-d-un ‘having eaten’.

7.7 Non-ﬁnite

verbs: past-stem

based

333

17. Parji: like the languages of South Dravidian II and unlike Kolami–Naiki, the

perfective participle is formed by adding -ci/-i;-ci occurs after verb stems with ﬁnal

alternants -t/- p in past and non-past conjugations, elsewhere -i is used, e.g. ver-i ‘having

come’, cen-i ‘having gone’, nilp-/nilt- ‘to stand’: nil-ci, k¯ap-/k¯at- ‘to wait’: k¯a-ci ‘having

waited’.

18. Ollari: just like in Parji, the perfective markers are -si/-zi/-i, e.g. s¯u

r-i ‘having

seen’, kar-si ‘having crossed’, u

n-zi ‘having drunk’.

19. Gadaba: there are three alternants -ji/-ci/-i,-ji after n-ﬁnal roots (Class 5, exception

pun-), -ci after roots with ﬁnal - p/-t alternation, e.g. goyalp- ‘to rinse’: goyal-ci, and -i

elsewhere, v¯ar- ‘to come’: v¯ar-i ‘having come’. Bhaskararao (1980: 47–8) calls this ‘the

incompletive sufﬁx’.

7.7.1.4 North Dravidian

There is no perfective marker in these languages, which can be traced to Proto-

Dravidian sources. The perfective participle is formed by adding -¯ar in Ku

rux, which

Subrahmanyam (1971: 222) thinks to be a borrowing of Hindi -kar ( j¯a- ‘to go’: j¯akar

‘having gone’), e.g. ci- ‘to give’: ci-¯ar ‘having given’. In Malto one of the two markers

is deﬁnitely native, -i and -le, on

d-i or on

d-le ‘having drunk’. Brahui has no non-ﬁnite

perfective participle.

7.7.1.5 Summary

Since

∗

-ci/-cci has reﬂexes in South Dravidian II and three languages of Central

Dravidian, it can be reconstructed for Proto-Dravidian. It appears to be a sequence of two

past morphs

∗

cc and

∗

-in with the ﬁnal consonant lost in the proto-stage itself, since the

gerund is a free form occurring before pause syntactically. Alternatively,

∗

-cci can be

taken as an innovation in South Dravidian II, since it is present in all the languages, with

the isogloss enclosing areally (through structural diffusion) some languages of Cen-

tral Dravidian, namely, P

arji

–Ollari–Gadaba, but not Kolami–Naiki (see feature 26 in

table 11.1c). The whole sequence

∗

-cc-in corresponds with the relic sequence -icin used

in Early Tamil as an extended form of the past ﬁnite verbs (see section 7.3.1.5.1).

7.7.2 Past adjective or the past relative participle

In most languages the past adjective is reconstructible. It is normally formed by adding

an adjectival sufﬁx to the past stem, e.g. Ta. va-nt-a,Te.wacc-in-a ‘the one who/which

came’. This participle is the main instrument of the relative clause (in the absence

of correlative pronouns). For instance, constructions like ‘Sita, who sang a song’ or

‘the song, which Sita sang’ are rendered like ‘the-song-sung-Sita’, ‘Sita-sung-song’ by

334 The verb

converting the ﬁinite verb into a relative participle which is placed before the noun head

(subject, object, instrument etc.).

7.7.2.1 Distribution in different subgroups

South Dravidian I In Old Tamil the adjectival sufﬁx -a is added to the past stem,

e.g. i

tu- ‘to place’, past stem i

t-: i

t-a ‘that placed’, n¯okku ‘to see’: n¯okk-i

n-a/-iy-a

‘that seen’, or-¯ıy-a ‘that disengaged’. Malay¯a

lam follows a similar process, e.g. cey-

‘to do’: cey-t-a ‘that done’, e

zutu- ‘to write’: e

zut-iy-a ‘that written’. Kota and Toda

have innovated independent formatives in relative participles. Ko

dagu adds -¨e to the

past stem, e.g. m¯a

du- ‘to make’: m¯a

d-in-¨e/-iy-¨e ‘that made’, bar- ‘to come’: ba-nd-¨e

‘the one who/which came’. Kanna

da forms the relative participle by the addition of -a

to the past stem, e.g. bar- ‘to come’: ba-nd-a ‘the one who/which came’, k¯e

l- ‘to hear’:

k¯e

l-d-a ‘that heard’, m¯a

du- ‘to make’: m¯a

d-id-a ‘that made’. Tu

lu forms the past relative

participle by adding -i/-¯ı or -ina/-na in different social and regional dialects, e.g. bar-

‘to come’: ba-tt-i/ba-tt-¯ı/bat-na (South and North Brahmin and South Common), batt-

in-a (North Common), ¯a- ‘to become’ ¯ay-i/¯a-n-a (South Brahmin and Common), ¯a-t-¯ı

(North Brahmin and Common) (Kekunnaya 1994: 111–12). Tu

lu makes a distinction

between the distant past (inherited past) and the immediate past (perfect tense). There

are adjectives formed on both past and perfect stems. The perfective is innovated in Tu

lu,

e.g. tin- ‘to eat’: tin-d-i ‘the one who ate/that which is/was eaten’, tin-t-i ‘that had (been)

eaten’, p¯o- ‘to go’: p¯o-y-i ‘that went’, p¯o-t-i ‘that had gone’, kal- ‘to learn’: kal-t-i ‘that

learnt’, kal-t¨ıd-i ‘that had learnt’.

South Dravidian II Telugu forms the past relative participle by the addition of -in-a,

which consists of the past -in- and the adjectival -a, ceppu ‘to say’: cepp-in-a ‘the one

who told/that told’. In some disyllabic verbs, the ﬁnal consonant is doubled in the

past adjective as a free variant of the -ina form, ce

du- ‘to perish’: ce

d-ina/ce

d-a ‘that

perished’, tagu- ‘to ﬁt’: tag-ina/tagg-a ‘that which ﬁtted’. In Gondi, the past adjective is

formed by the addition of -¯a/-a to the past stem, e.g. ki- ‘to do’: ki-t-¯a ‘done’, u

d-t-a ‘that

ploughed’ (Koya dialect), k¯as- ‘to boil’: k¯as-t-a¯er ‘boiled water’. Ko

da–Kui–Kuvi–

Pengo–Man

da add -i to the past stem, e.g. Ko

da s¯a- ‘to die’: s¯a-t-i ‘thedead...’,g¯ur-

‘to sleep’: g¯ur-it-i ‘the one that slept’; Kui p¯u- ‘to blossom’: p¯u-t-i ‘the blossomed’, tin-

‘to eat’, past stem tis-: tis-i ‘the one that ate/was eaten’; Kuvi p¯ay- ‘to beat’: p¯ay-it-i

‘that beat/was beaten’; Pe. ni

- ‘to rise’: ni

-t-i ‘that rose’, ¯end- ‘to dance’: ¯end-t-i ‘that

danced’; Man

da hen-ti

¯ak-i

‘leaves

that changed colour

’.

Central Dravidian Kolami and Naiki add -a to the past stem in forming the relative

participle, e.g. Kol. ¯ar- ‘to dry’: ¯ar-t-a ‘that dried’, tin- ‘to eat’: tin-d-a ‘the one that

7.7 Non-ﬁnite

verbs: past-stem

based

335

ate/was eaten’; Naiki arup- ‘to cut’: arup-t-a ‘the one cut’. In Parji, Ollari and Gadaba,

there do not seem to be relative participles formed on the past stem. There are, however,

forms based on non-past stems. In Gadaba the non-past relative participle is formed

by adding -dan to the verb stem, e.g. ¯u

d- ‘to plough’: ¯u

d-dan ‘that which is ploughed’.

There is another relative participle sufﬁx, -te for human and -o for non-human. It appears

that in one allomorph -t is a past marker, e.g. a

dg- ‘to dig’: a

dig-t-en

d ‘the man who

dug’, a

dg-o-n

d-i ‘the woman who dug’.

North Dravidian In Ku

rux, the past adjective is formed by adding -¯a to the the past

stem, kundr- ‘to be born’: kundr-k-¯a ‘that born’, un- ‘to drink’: unkh-k-¯a ‘that drank/

drunk’. There is nothing comparable in Brahui. Malto forms the participle by adding

-pe of obscure origin, e.g. dary- ‘to draw’: dary-pe ‘that drawn’.

Summary The sufﬁx -a is found in South Dravidian I, Telugu and Gondi of South

Dravidian II, Kolami and Naiki of Central Dravidian and Ku

rux of North Dravidian. The

sufﬁx -i is found in Tu

lu of South Dravidian I and in a subgroup of South Dravidian II, i.e.

da–Kui–Kuvi–Pengo–Man

da. It appears that both

∗

a and

∗

i can be reconstructed for

Proto-Dravidian as adjectiv

al markers, also found more extensi

vely as genitive suf

ﬁxes

in nominal inﬂection.

7.7.3 Conditional and concessive forms

The conditional forms meaning ‘if (subject) does/did’ and the concessive forms meaning

‘even if (subject) does/did’ are rendered as non-ﬁnite verbs in most Dravidian languages.

Such constructions are based on the past stem in some languages; in others they are based

on non-past stems or formed with sufﬁxes added to the basic stem. The meaning does not

necessarily denote the time of action. In any case, these constructions are not traceable

to Proto-Dravidian.

South Dravidian I Old Tamil adds -i

n to the bare stem in the case of weak verbs and

to the non-past stem in -kk/-pp in the case of strong verbs; the concessive is formed by

adding the conjunctive particle -um to the conditional form, e.g. k¯a

n-i

n ‘if one sees’,

n¯ı-kk-i

n/n¯ı-pp-in ‘if one removes’, iru-pp-in ‘if one stays’; concessive t¯onru- ‘to see’:

t¯o

nr-in-um ‘evenifonesees...’InLater Tamil the conditional form is based on the past

stem by the addition of -¯al/-¯el, e.g. k¯a

n- ‘to see’: ka

t-¯al ‘if one saw’, u

n- ‘to eat’: u

t-¯el

‘if one ate’. The concessive is formed by adding -um to the conditional, e.g. ka

t-¯al-

um ‘even if one saw’. In Old Malay¯a

lam -il/-kil are added in forming the conditional;

in Modern Malay¯a

lam just as in Tamil the conditional is formed by adding -¯al to the

past stem, e.g. OMa. var-il/varu-kil ‘if one comes’, Mdn Ma. cey-t-¯al ‘if one did’. The

336 The verb

concessive is formed by adding -um to the conditional form. Kota adds -m¯el to the past

stem in forming the conditional va-d-m¯el ‘if one came’. In Ko

dagu, ¯engi is added to the

past stem to form the conditional, e.g. ba-nd-¯engi ‘if one came’. Subrahmanyam (1971:

131) considers this morpheme as a complex consisting of the verb en- ‘to say’ plus

the conditional morph -kil found also in Malay¯a

lam. Ko

dagu loses the ﬁnal -l, but the

length of the vowel is unexplained. Kanna

da has a totally different conditional marker

-are, concessive marker ar-¯u. Old Kanna

da has -o

de instead, added to the past stem and

-are in Modern Kanna

da, e.g. bare- ‘to write’: bare-d-are ‘if one wrote’, bare-d-ar¯u

‘even if one wrote’, p¯e

l- ‘to speak’: p¯e

l-d-o

de ‘if one spoke’. Tu

lu adds -

da to ﬁnite

verbs in forming the conditional, e.g. kal- ‘to learn’: kal-t-e ‘he learnt’, kal-t-e-

da ‘if he

learnt’.

South Dravidian II In Old (Literary) Telugu the conditional was formed by adding

-in-an, which was homophonous with the concessive. It appears that -in was the original

conditional sufﬁx, which compares well with South Dravidian I -in; the conjunctive

-an was added to it to derive a concessive meaning originally, but it also came to be

interpreted as a conditional with bleaching of the meaning of the sufﬁx -an. In Modern

Telugu the concessive is -in¯a, which can be derived from older -in-an by the loss of -n and

lengthening of the preceding vowel. In Middle Telugu inscriptions, we have in-¯anu as

an intermediate stage of the concessive form, before the loss of ﬁnal -nu, which reﬂects

a conﬂict between pronunciation and writing. In Early Modern Telugu the conditional

is represented by -it¯e/-t¯e/-

t¯e which is totally unrelated to -in or -inan. It is said that this

has resulted from a wrong analysis of verbs in the second singular followed by -¯eni ‘if’

a conditional particle, e.g. c¯esiti(wi) + ¯eni ‘if you had done’ → c¯esit¯eni → c¯esit¯e(n).

With n-ﬁnal roots the sufﬁx is -

t¯e, e.g. an- ‘to say’: a

t¯e(ni) ‘if you said’. The meaning

got generalized to all subjects. In Gondi, the conditional sufﬁx is -¯eke/-eke added to the

past or non-past stem, e.g. v¯a- ‘to come’: v¯a-t-¯eke ‘if one came’. The past conditional is

used mainly in forming the subjunctive mood (contrafactual condition, e.g. ‘if you came,

I would have given you’ etc.). The concessive is expressed by the addition of -t¯er-/gir,

v¯a-t-¯ek(e)t¯er ‘if one came’. Ko

da adds -i

a to the past stem to form the conditional,

e.g. v¯a- ‘to come’: v¯a-t-i

a ‘if one came’; the concessive is formed by the addition of a

conjunctive sufﬁx -ba, v¯a-t-i

aba‘even if one came’. The sufﬁx -i

a probably has an

underlying -in followed by -ga, although we cannot attribute any meaning to the latter

element (section 7.6.2 (15)). Kui adds -eka to the past stem to form the conditional, e.g.

t¯ak- ‘to walk’: t¯ak-it-eka ‘if (one) walked/walks’; the emphatic particle -ve is added to

the conditional to form the concessive, e.g. sah- ‘to beat’: sah-t-eka-ve ‘even if (one)

beat/beats’. In Kuvi, the conditional marker is -ihi added to the past stem, e.g. t¯os- ‘to

show’: t¯os-t-ihe ‘If (I) beat (him)’. In Pengo the conditional marker is -is added to the

7.7 Non-ﬁnite

verbs: past-stem

based

337

past stem, e.g. v¯a- ‘to come’: v¯a-t-is ‘if (one) comes’. Note that Pengo -is and Kuvi -ihi

are cognate.

Central Dravidian In Kolami (Wardha dialect) the conditional marker is -te added to

the future ﬁnite form. This is clearly a borrowing from Telugu, but used differently. In

the Adilabad dialect the marker is -t¯e-, again reminiscent of borrowing from Modern

Telugu (-t¯e- see above), e.g. kak- ‘to do’: kak-t¯e ‘if (one) does/did’. Naiki adds -te to the

past or future ﬁnite verb, e.g. si- ‘to give’: si-t-an ‘I gave’, si-t -an-te ‘if/when I gave’.

This is also a clear case of borrowing from Telugu, but it is interesting that a bound

form is borrowed and added innovatively to a free form. Parji has a different process. It

adds (a) -o

d or -o

d-el, e.g. t¯ond- ‘to appear’: ili

t¯ond-o

‘if

a bear

appears

’; or (b)

-ek, e.g. var- ‘to come’: v¯ani

var-ek

‘if

rain

comes

’; or (c) -em, e.g. men- ‘to be’:

netta

p¯ap

men-em

‘when

the dog

a baby

. . . ’ Ollari adds -ko

r-en/-go

r-en to the

base, e.g. s¯ı- ‘to give’: s¯ıko

r-en ‘if one gives’, un- ‘to drink’: un-go

r-en ‘if one drinks’.

I wonder if the conditional morph -en is added to the reﬂexive -ko

r here. Gadaba has

two allomorphs -o

t/-go

t which are added to the basic stem to form the conditional, e.g.

anuyp-/anuyk- ‘to kill’: anuyk-o

t ‘if one kills’, tin- ‘to eat’: tin-go

t ‘if one eats’.

In North Dravidian there are no non-ﬁnite verbs used in conditional and concessive

meanings.

Summary It is not possible to reconstruct a marker for conditional in Proto-Dravidian.

The distribution of relevant markers is as follows:

(1) -i

n: Old Tamil (added to basic or the non-past stem)

(2) -il: Old Tamil, Malay¯a

lam (basic stem or non-past)

(3) -¯al: Middle and Modern Tamil, Malay¯a

lam (added to the past stem)

(4) -¯el: Middle and Modern Tamil (added to the past stem)

(5) -¯en-kil:Ko

dagu (added to the past stem)

(6)

de:

Old Kanna

(added to the past stem)

(7) -are: Modern Kanna

da (added to the past stem)

(8) -

da:Tu

lu (added to the past stem)

(9) -¯en(i): Old Telugu (added to the past stem or any free form including a ﬁnite

verb); -t¯e: Modern Telugu by metanalysis of the second singular ﬁnite verb

ending in -ti (wi) + ¯e(ni) → t¯e(n);-¯eni(n) ‘if’ is a clitic which can be added

to any free form in a sentence. But -t¯e ∼ -

t¯e/-it¯e is a grammaticalized bound

form which is added to the stem variant in the past tense, c¯uc- ‘to see’: c¯us-

t¯e ‘if one sees’; Kolami–Naik

ri have borrowed the sufﬁx -t¯e from Middle

Telugu. Note the construction in the other Central Dravidian languages is

different.

338 The verb

(10) -is/-ih: Pengo–Kuvi

(11) -¯ek/-ek: Gondi, Kui, also Parji

(12) -i

a:Ko

(13) -em: Parji

(14) -ko

r-en: Ollari

(15) -o

t/-go

t: Gadaba

It may be possible to reconstruct -in/-il for Pre-Tamil on the basis of (1), (2); (12),

(13) and the second

morph of (14) seem to be distantl

y derived from or related

∗

-in

although the vowel -e in (13) and (14) and the velar element in (12) are bothersome.

Again (3)–(5) and (9) reconstruct to

∗

-¯el/-¯en;-em/-en of (13) and (14) are perhaps also

related to this

set. OTe. -

¯en- appears to be

the conditional form of an archaic

stem -

¯en

∗

ay-m- ‘to be’) with the the sufﬁx

∗

-in. In any case the variety and diversity of forms

indicate its origin after the split of different branches of Proto-Dravidian.

7.8 Non-ﬁnite verbs: non-past-stem

based

The non-past stem is the base of several non-ﬁnite verbs like the non-past participle (du-

rative, present, present–future, future–habitual, etc. contrasting with the past participle),

non-past relative participle (verbal adjective) and the inﬁnitive. The imperative verb also

has a non-past stem incorporation because all imperatives intrinsically refer to non-past

time.

7.8.1 South Dravidian I

The literary dialect of Tamil has a non-ﬁnite verb with -pu as a sufﬁx, e.g. t¯a

z- ‘to fall’:

t¯a

z-pu ‘on falling’ (when something falls); i

tu- ‘to place’: i

tu-pu ‘on placing something’.

Note that Ko

da of South Dravidian II has exactly the same construction and, therefore,

∗

-pu is reconstructable for Proto-South Dravidian. The consonant - p must be a reﬂex of

the non-past

∗

-pp. The present relative participle is formed by adding -a to the present

tense stem in -ki

r-/-kinr- ‘the one who/which is doing’/‘ ...is done’, e.g. p¯ar- ‘to see’:

p¯ar-kki

r-a/p¯ar-kkinr-a ‘that which sees/is seen’. Another non-past relative participle

is formed by adding the aorist marker -um/-kk-um, e.g. e

zu- ‘to rise’: e

z-um ‘the one

who/which rises/will rise’, p¯ar- ‘to see’: p¯ar-kk-um ‘that which sees/will see’.

In Malay¯a

lam the future adjective is formed by adding -um, e.g. a

ri- ‘to know’:

ri-y-um ‘one who/which knows’.

In Kota, the present participle is formed by adding -r to the past stem in -t (S

; see

Emeneau 1994: 87–92), n¯o- ‘to see’: n¯o-

c-r ‘seeing’, org- ‘to sleep’: org-y-r ‘sleeping’.

(-r- seems to correspond to -ki

r- of Tamil.) The non-past adjective is also unique, formed

by adding -vd/-bd, e.g. tin- ‘to eat’: tin-bd ‘that is/will be eaten’. In Toda, the non-past

adjective is formed by adding -θ to the verb stem, e.g. na

r- ‘to walk’: na

r-θ ‘that which

walks’.

7.8 Non-ﬁnite

verbs: non-past-stem

based

339

In Ko

dagu, the non-past adjective is formed by adding -v-¨e/-pp-¨e to the basic verb

stem. Apparently, the consonantal element is the non-past-tense marker followed by an

adjectival sufﬁx, m¯a

du- ‘to make’: m¯a

du-v-¨e ‘that which does’, ba- ‘to come’: ba-pp-¨e

‘that which comes’.

The Kanna

da present participle is formed by adding the present-tense marker, i.e.

-ut(t)um, e.g. a

lu- ‘to weep’: a

l-utum ‘weeping’, bar- ‘to come’: bar-utt

um ‘coming’.

The non-past adjective is formed by adding the adjectival sufﬁx -a to the non-past stem,

e.g. ¯a

l- ‘to rule’: ¯a

l-v-a ‘that rules/will rule’, en- ‘to say’: en-ba ‘that which says/will

say’.

In Tu

lu, the present participle marker is -ontu/-ondu (different social dialects) added

to the past verb stem of two classes of verbs and to the bare stem in two other classes,

e.g. kal- ‘to learn’: kal-t-ontu/-ondu ‘learning’, t¯u- ‘to see’: t¯uv-ontu/-ondu ‘seeing’.

The non-past adjective is for

med by adding an adjective

marker -

i to the present–future

stem of the verb, kal-p-i/kal-pub-i ‘that which learns/will learn’. Subrahmanyam (1971:

261) thinks that -ontu is the perfective participle of the reﬂexive verb, which is added to

the past stem of a basic verb to form the ‘present adverb’. This process is found also in

Tamil and Malay¯a

lam.

7.8.2 South Dravidian II

The Old Telugu durative participle is formed by adding -cun, which has no corresponding

form in any other language. There is a phonological possibility that it could be related to

-cin (-c + -in) found in Classical Tamil; alternatively OTe. -cu could be the source of Ka.

-t(t)u, the present participial marker. One of the variants of the non-past adjectives is

formed by the marker -u(n), e.g. win- ‘to hear’: win-u(n) ata

du ‘the man who listens’.

This is reconstructible to PD

∗

-um/-un. There is another marker -u

du(n)/-wu

du(n) in

Old Telugu which meant ‘after, on...Ving etc.’, e.g. an- ‘to say’: an-u

du(n) ‘as soon

as one says/has said’. The other markers of relative adjectives in non-past are -e

di/-e

(> Mdn Te. -¯e), e.g. c¯eyu-/c¯es- ‘to do’: c¯es-edi/-e

du ‘the one who does’: Mdn Te.

c¯es-¯e id.

The Gondi durative participle is formed by adding -c¯er/-j¯er/-s¯er which occur in com-

plementation, e.g. toh- ‘to show’: toh-c¯er ‘showing’, un- ‘to drink’: un-j¯er ‘drinking’,

v¯ar- ‘to sing’: v¯ar-s¯er ‘singing’. The non-past adjective is formed by adding -v¯al to

the verb stem, veh- ‘to tell’: veh-v¯al m¯ayn¯al ‘the person that tells’. This sufﬁx is con-

sidered a nominal formative. The non-past conditional is formed on the non-past stem

by adding the conditional marker -¯ek¯e, e.g. a

t- ‘to cook’: a

t-n-¯ek¯e ‘if one cooks/while

cooking’.

In Ko

da -pu marks the simultaneative, e.g. koRku

ker-pu

‘when/as

the cocks

crow

’; it has an alternant -bu after stems ending in a nasal, e.g. man- ‘to be’: man-bu

‘while staying’. The durative aspect in the ﬁnite verb is formed by adding the non-past -n

340 The verb

to the perfective marker -si-/-zi-, e.g. ki- ‘to do’: ki-zin-a ‘I am/was doing’. The durative

conditional is formed by adding -i

to the durative stem in -zin-/-sin-, e.g. ki-zin-i

‘as

one is/was doing’. The non-past adjective marker is -i added to the non-past stem in

-n, e.g. g¯ur- ‘to sleep’: g¯ur-n-i ‘that which sleeps’; the non-past marker is dropped after

stems ending in -n or -

n, u

n-Ø-i ‘that which eats/will eat’.

In Kui the present (durative) participle is formed by adding -ki/-pi ∼ -bi/-ji occur-

ring in complementation. These are related to Proto-Dravidian non-past

∗

-kk/

∗

-pp and

the perfective marker

∗

-(c)ci, e.g. aj- ‘to fear’: as-ki ‘fearing’, ¯ar- ‘to call’: ¯ar-pi ‘call-

ing’, tin- ‘to eat’: tin-ji ‘eating’, in- ‘to say’: in-ji ‘saying’. The non-past adjective

is formed by adding the adjectival sufﬁx -i to the non-past base in -in/-n, e.g. ¯end-‘to

dance’: ¯end-in-i ‘that which dances/will

dance

’, k¯o- ‘to reap’: k¯o-ni ‘that which reaps/will

reap’.

Kuvi present (durative) participle is formed by the addition of -ci/-ji/-si/-hi corre-

sponding to the perfective participle of Gondi and Ko

da, e.g. h¯ı- ‘to give’: h¯ı-hi ‘giving’,

ve- ‘to beat’: vec-ci ‘beating’, ven- ‘to listen’: ven-ji ‘listening’. The non-past adjective

is formed by adding -i to the non-past stem in -in/-n, e.g. p¯ay- ‘to beat’: p¯ay-in-i ‘that

which beats/will beat’. The non-past adjective in P

engo is formed by adding the adjec-

tival sufﬁx -i to the non-past stem in -n, e.g. ko

r- ‘to buy’: ko

r-n-i ‘one who buys/will

buy’. This is entirely similar to the form in Ko

da.

7.8.3 Central Dravidian

In Kolami the present participle is formed by adding -san, e.g. ¯ol ‘to see’: ¯ol-san ‘eating’;

the future participle is formed by adding -ak, ¯ol-ak ‘about to see’; the durative adjective

is formed by adding -a, e.g. tin- ‘to eat’: tin-a ‘the one(s) who is/are eating’; the future

adjective is formed by adding -eka, var- ‘to come’: var-eka ‘that which comes/will

come’. Predicative nouns can be formed by adding -r- followed by personal sufﬁxes,

e.g. kalk-eka-r-an ‘I am the man who does’ etc. Naiki (Chanda) forms the non-past

participle by adding -eka, e.g. ser- ‘to go’: ser-eka ‘going’.

In Parji the non-past participle is formed by adding -o

d/-ek/-em with slightly different

meanings, e.g. men- ‘to live’: men-o

d ‘while living/if (one) lives’, ver- ‘to come’: ver-ek

‘on coming home’, men-em ‘when (one) is living’. The habitual adjective is formed by

adding the adjectival sufﬁx -an to the future stem in -r-/-d-/-t-/-iy-, e.g. ven- ‘to hear’:

ven-d-an ‘one that hears’, cok- ‘to climb’: cok-r-an ‘the one that climbs’, v¯ıt- ‘to sow’: v¯ıt-

iy-an ‘the one that sows’. Predicative nouns are formed by adding the third-personal suf-

ﬁxes (not the ﬁrst and second) to these, e.g. ci- ‘to give’: ci-r-an-ed ‘the man who gives’.

Ollari non-past participle is formed by adding -i

/-u

, e.g. un- ‘to drink’: un-u

‘while

drinking’, val- ‘to ﬂy’: val-i

‘while ﬂying’. The habitual adjective is formed by adding

the adjectival

suf

ﬁx -an to a non-past (habitual)

stem, e.g.

s¯ı- ‘to give’: s¯ı-d-an ‘one who

gives/will give’, pun- ‘to know’: pu-y-an ‘one who knows/will know’.