Krishnamurti Bhadriraju. The Dravidian Languages

Подождите немного. Документ загружается.

4.4 Historical

phonolo

gy: vowels

101

(19) PD

∗

t¯e

l ‘scorpion’. SD I: Ta. Ma. t¯e

l,Ko

d. t

l¨ı, Ka. c¯e

z(u), t¯e

z (

z instead

l could be a hyper-standard form after

z became

l), Tu. t¯e

l¨ı, c¯e

l¨ı, t¯el¨ı;

SCD (SD II): Te. t¯elu. ND: Malt. t¯ele,Br.t¯elh

. (No cognates from most of

SD II and CD) [3470].

(20) PD on-

tu (-tu neuter sg sufﬁx) ‘one’. SD I: Ta. onru (> Mdn Ta. o

nu), Ma.

onnu,Ko.o

d,Ko

d. ond¨ı, Ka. ondu,Tu.o ˜nji; SD II: OTe. o

du, Mdn Te.

ti ‘single, alone’, Go. und¯ı, un

d¯ı ,Ko

da un

ri; ND: Ku

r. ¯on, ¯ond, Malt.

-ond,Br.asi [990 d].

(21) PD

∗

k¯ol ‘stick’. SD I: Ta. Ma. Ko. k¯ol,Ko

d. k¯ol¨ı, Ka. k¯ol, k¯olu,Tu.k¯ol¨ı,

k¯olu; SD II: Te. k¯ola, Go. k¯ola,Ko

da k¯ol,Kuik¯o

du, Kuvi k¯olu,Pe.k¯ol

‘pestle’, Man

da k¯ul; CD: Kol. k¯ola, Nk. Pa. k¯ol ‘pestle’ [2237].

4.4.2 Alternations i/e and u/o in Proto-South Dravidian

(SD I and SD II)

It has already been stated that -V

is a part of the formative syllable and only three vowels

can occur in this position,

namely

iau, and not all the ﬁve vowels of Proto-Dra

vidian

which occur in V

position. The following sound changes are signiﬁcant in that Rule 4a

represents the Proto-South Dravidian stage which includes South Dravidian I and South

Dravidian II, while 4b occurred only in a subgroup of South Dravidian I, i.e. Proto-Tamil,

and 4c in Early Kanna

da. The sound changes

in question have been discussed

thoroughly

by Burrow (1940)

and Krishnamurti (1958a, 1961, 1980, 2001a: ch. 2 Postscript) and

now know their proﬁle from the Proto-South Dravidian stage through Early Tamil (which

included Malay¯a

lam also). For subsequent adoption of the principles of reconstruction

of this change, see Emeneau (1970a), Zvelebil (1970b) and Subrahmanyam (1983).

Rule 4. South Dravidian umlaut

(a) PD

∗

u >

∗

o /#(C

)

−−−

-a (Proto-South Dravidian)

(b) PSD

∗

o > iu/#(C

)

−−−

-a (Proto-Tamil).

Rule 4a merges high vowels with mid vowels when followed by -a in the next syllable.

This

means that PSD

∗

e represents PD

∗

i and

∗

e and PSD

∗

o represents PD

∗

u and

∗

o, when followed by -a in the next syllable. At a much later stage, by Rule

4b these

instances of

∗

o became i, u, respectively, in Early Tamil. As a consequence of these two

changes, Tamil and

Malay

¯a

lam have i, u corresponding to

Telugu and Kanna

da e, o in the

environment [C-a. By looking only at etymologies in which Tamil and Malay¯a

lam have

i, uC-a and Telugu and Kanna

da e, oC-a, there is no way one can reconstruct the Proto-

Dravidian vowel qualities. When high vowels or a Ø formative occur as V

, all Proto-

Dravidian vowels remain in all the languages concerned. Similarly, if any of the languages

preserves a long vowel in cognates, its quality can be taken to represent the quality of

102 Phonolo

gy: historical

and comparative

the Proto-Dravidian vowel (see ety. (13), (15), (17), (19), (21)). Since this is a change

that covered mainly South Dravidian I and South Dravidian II languages, evidence

from the other subgroups like Central Dravidian and North Dravidian would also help

in reconstructing the Proto-Dravidian vowel qualities in doubtful cases, i.e. if none

of the South Dravidian languages has cognates with high vowels or zero beginning

the formative syllable or with a long radical vowel. The following examples attest the

application of Rules 4a and b and also identify the diagnostic environments for the

reconstruction of the Proto-Dravidian vowel qualities:

(22) PD

∗

it-V- ‘meat’. PSD

∗

et-a-V-; SD I: Ta. iracci ‘meat’, Ma. iracci,Ko

eraci; SD II: OTe. e

raci,Pe.jey ‘meat’ (<

∗

rey- <

∗

ir- ay) [529].

The Proto-Dravidian quality is preserved in Ta. i

rri ‘ﬂesh’ where

∗

i occurs in a closed

syllable.

(23) PD

∗

k¯e

l /

∗

l-V- n. ‘family’, v. ‘to ramify’. PSD

∗

k¯e

l /

∗

l-a-; SD I: Ta.

k¯e

l ‘kindred’, k¯e

l-mai ‘friendship’, ki

l-ai v.i. ‘to multiply’, n. ‘kindred,

relations’, Ko. k¯e

l,To.ke



l ‘partnership’, Ka. ke

l-e, ge

n-e ‘companion-

ship’, k¯e

l-a‘companion’, Tu.ge

n-e ‘coupling’; SD II: Kui kl¯amba ‘family,

lineage, kin’ (Kui ¯a <

∗

¯e) [2018].

The Proto-Dravidian vowel quality is based on the long vowel stems in Tamil and

Kanna

da.

(24) PD

∗

col ‘ﬁreplace’. PSD col-V-; SD I

∗

ol-V- > Ta. ul-ai, Ma. ul-a,Ko.

elk¯al ‘ﬁreplace between two stones’, To. was

∗

ol-), Ko

d. ol-e, Ka. ol-e,

Tu. ul-e (loss of

∗

c- in SD I); SD II

∗

col > Ko

da solu,Kuiso

du,Pe.hol,

Man

da huli; CD: Pa. col

gel (-kel ‘stone’), Gad. soygel [2857].

The Proto-Dravidian vowel quality can be established on the basis of cognates from

South Dravidian II and Central Dravidian.

(25) PD

∗

n¯u

z /

∗

z-V- ‘to squeeze through’. PSD

∗

n¯u

∗

z-a > Ta. nu

z-ai v.i.

‘to creep in’, v.t. ‘to insert’, n. ‘a narrow passage’, Ma. nu

z-ayuka ‘to creep

in, squeeze through’, Ka. no

z-e, nurgu, nuggu (<

∗

z-ung-) ‘to squeeze

through’, Tu. nurguni ‘pass through’; ND:

∗

z- > Ku

r. nu

r-(nu

d-) ‘to

hide’, Malt. nu

d-. Also cf. Ta. nu

z-u-ntu, nu

z-u-tu ‘to insert’, n¯u

z-ai ‘hole’

[3714].

Long root-vowel, high vowel as V

and evidence from ND attest to Proto-Dravidian

vowel quality and quantity.

There are a few residual forms for which it may not be possible to reconstruct the

Proto-Dravidian vowel quality for want of diagnostic environments as deﬁned above, e.g.

4.4 Historical

phonolo

gy: vowels

103

(26) PSD

∗

el-a- > Ta. il- ai ‘leaf, petal’, Ma. il-a,Ko.el,To.es,Ko

d. elaka

da,

Ka. el-e, el-a; ?Go. (Mu.) ko

rk-ila ‘new leaf’ [497].

It is not certain what the original quality of vowel was before the merger in Proto-South

Dravidian. For other such cases, see Subrahmanyam (1983: 209–10).

4.4.2.1 Reconstruction of Proto-Dravidian vowel qualities after merger

It was established beyond doubt that PD

∗

u merged with

∗

o in Proto-South Dra-

vidian and not vice versa (Krishnamurti 1958a: 464–65). It was shown as part of Proto-

Dravidian morphophonemics that (C

-V

- (where C

∗

k) produce

)

- by syllable-contraction rule (Rule 2). When V

is a high vowel /i u/, the re-

sultant long vowel has

the same quality as V

; but when V

is /a/, the resultant long

vowel is

∗

¯e for both

∗

i and

∗

e as V

and

∗

¯o for both

∗

u and

∗

o as V

. From this it is clear

that when the formative vowel was -a, PSD

∗

e replaced PD

∗

e and

∗

o replaced

∗

respectively, before the contraction rule operated (see Krishnamurti 1958a: 464–5), e.g.

PSD

∗

t¯e- < tiy-a- ‘sweet’ (see ety. 7), PSD

∗

k¯e-mp- <

∗

kiy-a-mp (ety. 8).

(27) PD

∗

t¯e- <

∗

tew-i ‘to be full, satiated; to belch’. PSD

∗

tew-V- > Ta. tev-i

‘to be full’, tev-i

tu, tek-u

tu ‘to be cloyed, sated’, tev-i

tu ‘loathing of food

from satiation’, tevvu ‘to ﬁll’, tik-ai v.i. ‘to be complete’; Ma. tika

t-uka

‘to become full to the throat, to belch’, t¯e

t-uka ‘to belch, ruminate’ [3405,

3453]. Cf. Ta. t¯e-nku (or ?t¯e

n-ku) ‘to become full’, t¯e-kku ‘to be sated,

to belch’, Ma. t¯e˙n˙n-uka ‘to feel nausea’, t¯ekk-uka ‘to belch’, Ko. t¯ekl ‘a

belch’, To. t

ok- ‘to belch’, Ko

d. t

ek¨ıl¨ı ‘a belch’, Ka. t¯egu, t¯eku ‘a belch’

Tu. t¯ekε ‘brimful’; SD II: Te. t¯encu ‘to belch’, Ko

da d¯ek- ‘to belch’, Kui

t¯ek-; ND: Ku

r. t¯e

kh t¯e

khrn¯a ‘to feel heavy after a dinner’ [3451, 3453].

DEDR puts the relevant cognates in three entries, but items under 3453 illustrate the

contraction rule given at the head of the entry.

(28) PD

∗

t¯o-l <

∗

tok-al ‘skin, hide’. SD I: Ta. Ma. Ko. t¯ol,To.tw

¨ıs,Ko

d. t¯ol¨ı,

Ka. t¯o-l;SDII:Te.t¯olu, Go. t¯ol,Ko

da t¯olu,Kuit¯o

du, Kuvi t¯olu,Pe.t¯ol;

CD: Kol. Nk. Pa. Oll. Gad. t¯ol. Compare these with Ta. tuk-al, tokku ‘skin

of fruit’, Ka. togal, toval ‘leather, skin of fruit’, Te. tokka ‘skin of fruit’,

Oll. tokka, Gad. tokk¯a [3559].

(29) PD

∗

p¯o<

∗

puk-a- ‘to go’. SD I: Ta. puk-u (pukk-) ‘to enter, go’, Ma. puku-

ka,To.pux-(puk-), Ka. puku, pogu (pokk-), Tu. pogg- uni ‘to enter’ [4238].

This item is related to 4572 PD

∗

p¯o ‘to go’: SD I >Ta. Ma. p¯o ‘to go’,

p¯okku ‘to cause to go’, To. p

¨ıx- v.i., Ka. p¯o, p¯ogu, h¯ogu,Tu.p¯o-pini;

SD II > Te. p¯o(wu), Ko

da, Pe. p¯ok- ‘to send’, Man

da p¯uk- [4572].

104 Phonolo

gy: historical

and comparative

DEDR separates these two sets on the basis of the differences in their surface

phonology.

Contrasting with the above are cases of lengthening of the root vowel without change

after contraction, if the underlying formative vowel (V

)is

∗

i or

∗

u, e.g. OTe. n¯ı -ru,

niw-u

ru ‘ashes’: Ta. Ma. Ka. n¯ı -ru, Go. n¯ı -r,Ko

da n¯ı -ru,Pe.n¯ı z/n¯ıs [3693].

(30) PD

∗

niw-i/u- > n¯ı - ‘to be elongated, to stretch out’. PSD

∗

niw-i/u-,

∗

n¯ı -;

SD I > Ta. nim-ir ‘to be straight’, Ma. nivir-uka, Ka. nimir, nigur ‘to be

stretched out’, nig-u

l ‘to stand erect’, nig-ur ‘to lengthen out’, Tu. nigacuni;

SD II: Te. nig-u

du ‘to stretch out’, Ko

da nig-

ri- ‘become erect’; CD: Pa.

nikip-(nikit-) ‘to stretch out’ [2922]: SD I> Ta. Ma. n¯ı-

l ‘to grow long’,

n¯ı -

tu id., Ko. To. n¯ı

r v.i., n¯ı

t v.t., Ko

d. n

¨ı

d-, v.i., n

¨ı

t- v.t., Tu. n¯ı

t-uni, n¯ı

d-uni,

Ka. n¯ı -

l ;SDII> Te . n¯ı -lugu,Kuin¯ı l -(n¯ı

t-) ‘to stand up’,

dr¯ınj-(<

∗

d¯ı n j-

∗

d-inj-) [3692].

∗

-w- changes to -m-or-g- in different languages (see section 4.5.7.2.1).

4.4.2.2 Exceptions and extensions to the umlaut rule

1. Tamil: in Old Tamil there are some thirteen residual forms which appear not to have

undergone the dissimilatory rule (Rule 4b). Almost all these forms are attested in early

literary texts (see Krishnamurti 1958a: 465–8), e.g. Ta. cey-al ‘action’: cey ‘to do’, cel-

avu ‘going’: cel ‘to go’, ce

r-al ‘anger’: cer-u ‘to be angry’, ko

l-ai ‘hold, determination’:

l ‘to hold’, kol-ai ‘killing, murder’: kol ‘to kill’, to

t-a˙nku ‘to begin’: to

t-u ‘to begin’.

Do these represent

retention of the Proto-Dra

vidian stage or the Proto-South

Dravidian

stage? At least one etymology is diagnostic in pointing to these as retentions of the

Proto-South Dravidian stage:

(31) PD

∗

pic-ar ‘name’. PSD

∗

pec-ar [pes-ar] >

∗

pey-ar;SDI:Ta.pey-ar, piy-

ar, p¯e-r ‘name’, Ma. pey-ar, piy-ar, p¯e-r,Ko.p¯er,To.p

or n., p

oˇsf v.t.

‘to name’, Ko

d. peda (r-loss), Ka. pesar (u), hesaru,Tu.pudar¨ı, Koraga

podari, hudari; SD II: Te. p¯eru, Go. par¯ol, padur (< pedur), Ko

da p¯er(u),

Kui p¯aru; CD: Kol. Nk. p¯er (lw < Te.), Pa. Oll. Gad. pidir; ND: Ku

r. Malt.

pinj- ‘to name’, Br. pin ‘name’ [4410].

The Central Dravidian and North Dravidian forms are diagnostic in that they show

∗

as the Proto-Dravidian vowel. The ultimate connection could be

∗

pin-cc- >

∗

picc-ar >

∗

pic-ar whence Ka. pes-ar. The Tu

lu, Ko

dagu and Central Dravidian forms point to -d-

(< -t- < -c-). In that case Tamil–Malay¯a

lam pey-ar represents the Proto-South Dravidian

stage and not the Proto-Dravidian stage (see Krishnamurti 1958a: 466). The few residual

forms have to be taken as exceptions to the dissimilatory rule. That these are nouns

formed synchronically from verb roots within Tamil does not qualify them as exceptions

(as proposed by Subrahmanyam 1983: 214–16), because forms like Ta. ve

l ‘white’:

4.4 Historical

phonolo

gy: vowels

105

lanku ‘to shine’, vi

l-ar ‘to become pale’ [5496a], Ta. pe

r-u ‘to beget, bear’: pir-a ‘to

be born’, pi

r-avi ‘birth’ [4422] are similarly related, except that Rule 4a applies to them.

Kota, Toda, Iru

la and Ku

rumba dialects (except P¯alu Kurumba) and Ba

daga conform

to the rule by showing only e and o before C-a. Kapp (1978) says that P¯alu Ku

rumba

does not show the implementation of i, u [C-a > e, o [C-a, e.g. nila, nilamu ‘ground,

soil’, mu

lamu ‘cubit’, ile ‘leaf’. The fact seems to be that this dialect shares with Early

Tamil the dissimilation rule (Rule 4b), i.e. e, o [C-a > i , u [C-a, because of the presence

of such forms as ule ‘ﬁreplace’: Ta. ulai (<

∗

ol-ay < PD

∗

col-ay; see ety. (24)), and puge

‘smoke’: Ta. pukai (< PD

∗

pokai).

2. Tu

lu: Rules 4a and b require that all other languages of South Dravidian I and II

have only e, oC-a while Tamil and Malay¯a

lam have i, uC-a after the operation of the

ule. All Nilgiri dialects which

branched from Pre-Tamil at dif

ferent points in time show

the inherited qualities as expected. So do Ko

dagu and Kanna

da. Tu

lu has two regional

dialects, North (N) and South (S), and two caste dialects, Brahmin (B) and Common

(C). A recent study of Tu

lu dialects demonstrates that the NB, SB and SC attest the

inherited qualities (e, oC-a), while the NC has changed these to high vowels (i, uC-a).

This must be a recent dialectal change, which is not related to what happened in Early

Tamil (Rule 4b); e.g. Tu. es-a

lu ‘petal’, mo

lampu ‘knee’ (SB, SC, NB): is-a

lu, mur-ampu

(NC) (Kekunnaya 1994: 42).

3. Kui–Kuvi: it has already been demonstrated (Krishnamurti 1958a: 465) that the

umlaut rule (Rule 4a) preceded the syllable contraction rule (Rule 2) in all South

Dravidian languages; therefore, it is a shared innovation of South Dravidian I and South

Dravidian II. It also preceded the metathesis and vowel-contraction rules of South-

Central Dravidian (South Dravidian II), perhaps going back to over a millennium BCE.

These require us to reconstruct

∗

¯e and

∗

¯o in South Dravidian II also for older

∗

i /

∗

eC-a,

∗

oC-a. The metathesis and vowel-contraction rule (see section 4.4.3, Rule 6 be-

low) is still an ongoing sound change in Kui–Kuvi–Pengo and Man

da (Krishnamurti

1978a). The long mid vowels which resulted from Rule 2 or Rule 6 (metathesis and

vowel contraction) are retained in the other South-Central Dravidian (South Dravidian

II) languages but lowered to -¯a- in Kui–Kuvi (Krishnamurti 1980), e.g.

Rule 4c. Lowering of long mid vowels in Kui–Kuvi

¯e, ¯o > ¯a /#(C

)(C

)

−−−

(Pre-Kui–Kuvi)

PSD

∗

cow-ar (<

∗

cup + ar) ‘salt’. SD I

∗

ow-ar: Ta. Ma. uvar, Ka. ogar,Tu.ubar¨ı, ogar¨ı;

SD II:

∗

cow-ar > (Te. ogaru ‘astringent’, lw from Ka.), Go. sovar, hovar, ovar,Ko

s¯oru,Pe.h¯or,Kuis¯aru, Kuvi h¯aru, Man

da j¯ar [2674a,b; see ety. (10)].

(32) PD

∗

z-V n. ‘pit, hollow’, v.i. ‘to form pit’. PSD ku

z-i/-u, ko

z-a-; SD I:

Ta. Ma. ku

z-al ‘ﬂute’, Ko. korl ‘tube’, kuy ‘pit’, To. kw¯e



l ‘clarinet’, Ko

106 Phonolo

gy: historical

and comparative

l-a, Ka. ko

z-al, ko

z-avi ‘ﬂute, tube’, Tu. ko

l-avε; SD II: OTe. kr¯o-lu,

kr¯o-wi,Pe.k

roy ‘pit’, Kui k

r¯au (k

r¯anga, pl) ‘pit, hole’, Kuvi (F) graiy¯u

(gr¯anga, pl), g

r¯ayu ‘hole’ [1818].

Since this sound change occurs only in the case of long vowels resulting from contraction

eC-a, oC-a, and not in original

∗

¯e,

∗

¯o, it was possible that the qualities of these vowels

could have been phonetically opener and lower than mid vowels, somewhat like ¯ε and

before they merged with ¯a (Krishnamurti 1980: 502–3). There are seven cases shared

by Kui and Kuvi, eight in Kui alone and three in Kuvi alone.

4. Modern literary languages: Rules 4a and 4b are not operative in Modern Tamil,

Kanna

da and Telugu. Tamil and Malay¯a

lam have changed older i , uC-a to modern e,

oC-a, e.g. OTa. il-ai > Mdn Ta. elε ‘leaf’, i

t-am > e

d-am ‘space’, puk-ai ‘smoke’ >

poh-ε. These forms violate 4b, which ceased to operate by the Middle Tamil period.

Therefore

they appear to go back to

the old pre-assimilatory stage.

Mdn Kanna

has

tig-a

tu ‘rind’, hu

l-a ‘wo

’, nil-avu ‘standing’; s

o also Te.

cil-aka < cil-uka ‘par

rot

’,

d-ata < u

d-uta ‘squirrel’. These violate Rule 4a, which is no longer operational in

these languages. Notice in all such cases in Telugu and Kanna

da it is the formative

vowel which has changed from high to low (Krishnamurti 1958a: 468).

4.4.2.3 Kanna

da umlaut

Around the eighth century CE, Kanna

da independently introduced a sound change:

radical mid vowels became high when followed by a high vowel, before written literature

emerged in the language (Krishnamurti 1958a: 467; earlier Sreekanthaiya 1935, Burrow

1940: 296–7 and Gai 1946: 5–6 mentioned this change).

Rule 5. Mid-to-high vowel harmony

e, o > i, u /#(C

)

−−−

-V

=[+high])

id-ir ‘opposite’ (<

∗

ed-ir): Ta. et-ir,Te.ediri ‘opponent’, ed-uru ‘opposite’, sur-i ‘to

pour down’ (<

∗

cor-i): Ta. cor-i id., Te. tor-˜agu ‘to spill’, pur-i ‘tofry’(< por-i): Ta.

por-i,Te.por-˜a

tu id. Here, the umlauted e, o are not involved in the change.

4.4.3 Long vowels through metathesis and vowel-contraction

in South-Central Dravidian (SD II)

After high vowels had merged with mid vowels in South Dravidian II (Telugu, Gondi,

da, Kui, Kuvi, Pengo and Man

da), a sound change came about as an innovation in this

subgroup. One of the consequences of this sound change was to shift non-nasal apical

consonants /

∗

z/ which occurred as

∗

in Proto-Dravidian stems (C

- to the position before V

, allowing V

and V

to contract into long

(Krishnamurti

1955, 1961: 51–68, 1978a: 18–19).

4.4 Historical

phonolo

gy: vowels

107

Rule 6. Metathesis and vowel contraction

PD PSD

-V

- > (C

ii > ¯ı

uu > ¯u

aa > ¯a

i/e a > ¯e

u/o a > ¯o

If the proto-form had no initial consonant, then C

would become C

, creating words

with alveolar and retroﬂex consonants in this subgroup. Where the proto-form had an

initial consonant, the resulting

form would have a consonant

cluster with an apical

resonant as the second member. Since this change took place in Proto-South Dravidian

after the umlaut rule (Rule 4), PSD

∗

e and

∗

o represent PD i /e and u/o respectively; after

contraction we have ¯e and ¯o from

∗

e-a and

∗

o-a, respectively. This isa major historical and

typological change leading to innovative phonotactics in South Dravidian II. Examples:

(33) PD

∗

t-ank- ‘to be compressed, to hide’. PSCD

∗

d-ang-/

∗

d¯a-ng:OTe.

gu ‘to lie hidden’,

a-cu v.t. ‘to hide’ (beside a

d-˜agu, a

d-˜acu), Mdn Te.

d¯a-gu, d¯a-cu (beside a

nugu, a

nuc-), Ko

d¯a

- v.i.,

d¯ap-, v.t., Kui

d¯apa

(

d¯at-) ‘to lie in wait for’ (Kol.

d¯a

g-,

d¯ap- and Nk.

dh¯a

g-,

dh¯ap- were

apparently borrowed with

d- from Early Telugu); SD I: Ta. a

tanku, a

takku,

Ma. a

ta˙n˙nu, a

takku,Ko

d. a

dak- v.t. Ka. a

dangu, a

daku; ND: Malt. a

rg-

‘to press down’, Ku

r. Malt. a

rk- ‘to press down’ [63].

(34) PD

∗

l-V- ‘young, tender’. PSD

∗

l-a-/

∗

l¯e-:Te.l¯e- ‘tender’, also el-a,

Go. leyor, leyon

d ,

diyy¯or ‘young man’, leya,

diyy¯a ‘young woman’, Kui

l¯avenju ‘young man’, l¯aa, r¯a

a ‘young woman’ (l¯a- < l¯e-), Kuvi r¯a

a ‘young woman, virgin’; SD I: Ta. i

l-a, i

l-ai ‘young’, Ma. i

l-a, Ko. To.

l ‘young’, Ko

d. ¨e

le¨e ‘youth’, Ka. e

l-a, e

l-e ‘tenderness, youth’, Tu. e

l-e

[513].

(35) PD

∗

mar-am/n ‘tree’. PSD

∗

mar-an/

∗

mr¯a-n:OTe.mr¯anu ‘tree’, Mdn Te.

m¯anu,Ko

da maran, mr¯anu,Kuimr¯anu, Kuvi mr¯anu, m

r¯anu;SDI:Ta.

Ma. maram,Ko

d. Ka. mara; CD: Pa. meri, Oll. marin, Gad. m¯aren; ND:

r. mann, Malt. man [4711a].

(36) PSCD

∗

por-a-/

∗

pr¯o- ‘to sell’: Ko

da por-(port-), Kui, Kuvi pr¯a-(pr¯at-),

Pe. pro-(prot-), Man

da (BB) pre-(pret-)[4536].

Further examples can be seen in DEDR 1278, 4973, 3949, 3174, 3340, 4283, etc. Krish-

namurti (1978a, 1983) discusses this sound change and shows that it has spread through

the mechanism of ‘lexical diffusion’ and has not completed its course in some of the

108 Phonolo

gy: historical

and comparative

languages, particularly, Kui–Kuvi–Pengo–Manda. It has ceased to be an ongoing change

in Telugu, Gondi and Ko

da (1978a: 9). Even South Dravidian I languages show loss

of word-initial vowels before alveolar and retroﬂex consonants, thereby promoting the

apical consonants to word-initial position; Iru

la and Tu

lu have over two dozen items

beginning with non-nasal apicals (Krishnamurti 1978a: 18, fn.14). This phenomenon is

discussed in greater detail under consonants.

4.4.4 Vowel changes in individual languages: South Dravidian I

4.4.4.1 Toda

Emeneau (1957b, 1970a) and Subrahmanyam (1976b, 1983) have attempted to identify

the conditions of the most complex changes in Toda vowels. There are still a number of

exceptions to the rules proposed. Subrahmanyam (1983: 53–60) describes the conditions

for retention and says that ‘change’ occurs elsewhere. Normally we notice conditions

(in the environment of a given segment) for change and not necessarily for retention.

In the following sections I have tried to recast the rules of Subrahmanyam into those

specifying the conditions of change and not of retention.

Rule 7a. Low vowel rounding in Pre-Toda

∗

¯a > o, ¯o /# (C

)

−−−

)-V

is [ −low]; in the case of

∗

a,C

= an alveolar

∗

r [+ apical, + anterior]; or C

is not a single retroﬂex

∗

z [ +apical, −anterior] of Pre-Toda followed by a vowel

which is [+low] as (V

); elsewhere,

∗

a and

∗

¯a will remain unchanged.)

Examples: PD

∗

n ‘eye’ > To. ko

n (C

is a retroﬂex but not followed by -a/-ay)

[1159], but

∗

pal ‘tooth’ > To. pas (alveolar l is an anterior, i.e. a non-back consonant

retains a) [3986], PSD

∗

nakk- ‘to lick’ > To. nok [4353]; here, the consonant following

the root vowel is back and high, hence the change;

∗

p¯at-u ‘to ﬂy’ > To. p¯or- (not followed

by -a/-ay) [4020]. There are a number of exceptions to the rules stated above: (1) To.

n ‘loan’: Ta. ka

t-a

n [1113], To. ko

n ‘threshing ﬂoor’: Ta. ka

l-an [1376], To. o

dg-

‘be quiet’, o

rk- ‘to subdue’: Ta. a

tanku v.i., a

takku v.t. [63] all have a single retroﬂex

followed by

∗

-a but still they do not retain the original

∗

a root vowel; (2) deictic aθ - ‘he,

she, it’, pl aθ -¯am, at ‘that much’, atfok ‘then’, etc. do not change a to o [1], and tan-,

tam- reﬂexive pronoun obliques have retained the original vowel – perhaps a grammatical

constraint on the operation of the sound change; (3) through contraction with the sufﬁx

vowel, the root vowel is lengthened in a few cases, To. k¯al (k¯a

d-) ‘fall from height’:

Ta. ka

zal ‘slip’ [3582], To. p¯aw, p¯a, p¯afn ‘old’ : Ta. pa

za [3999], To. p¯ag ‘to use’: Ta.

z-aku ‘be used’ [4000] (Subrahman

yam 1983: 51

–60, 63–70).

∗

i split into ¨ı and i;¨ı before and after consonants other than c, ˇc and s, and i

elsewhere.

4.4 Historical

phonolo

gy: vowels

109

Rule 7b. Splits of front and back vowels (Emeneau 1970a: 36–46; Subrahmanyam

1983: chs. 4–13, pp. 51–157)

(i)

∗

i> ¨ı/#(C

)

−−−

and C

are not palatal or sibilant in pre-Toda)

PSD

∗

ir-u ‘to be’ > Ta. iru-(iru-nt-): To. ¨ır-(¨ıθ-) ‘to sit, live’ [480], PD

∗

tin (tin-t) ‘to

eat’ > Ta. ti

n-(tin-r-): To. t¨ın (t¨ıd-) [3263]; PD

∗

kil- ‘to sound’ > Ta. cilai ‘to sound’:

To. kis

-(kis-θ -) ‘to crow’ [1574]. The conditions of split are not clear although ¨ı appears

mostly before pre-Toda alveolar consonants. In contrast i remains in PD

∗

cil- ‘to be not’

>Ta. il ‘non-existence’, illai ‘no, not’: To. il- ‘not to exist’ [2559], Ta. ki

lu ‘to pinch’:

To. ki



l [1589], To. kic, kiˇc ‘ﬁre’ (<

∗

kic-) [1514], To. siry ‘joy’ (< ciri ‘laughter’)

[1562], etc.

(ii)

∗

e > ¨o/ #(C

)

−−−

= [–alveolar];

Pre-Toda) [784]

Ta. e

tu ‘eight’ : To. ¨o

t,butkerf ‘shoe’ (<

∗

kerpu [DED 1963]). There are 41 examples

given: 29 follow the rule and 12 violate it. Subrahmanyam invariably calls all exceptions

loanwords without any evidence from external history, like pe

n ‘butter’ (cf. Ta. ve

ney).

(iii)

∗

e > i/#(C

)

−−−

-V

= [+high] or C

is [+palatal])

E.g. To. nis

‘emblic myrobalan’ : Ta. nelli [3755],To. ir-(irθ-): Ta. eri ‘to throw a

weapon’ [859] Subrahmanyam gives 16 examples that conform to the rule and 6 that

violate it.

(iv)

∗

e > ¨ı is said to occur in the remaining environments. But the speciﬁc

conditions are not clear (Subrahmanyam 1983:125). In most cases C

an alveolar, retroﬂex, velar or the glide /y/, i.e. [+apical] or [+high], e.g.

To. ¨ır ‘female buffalo’: Ta. erumai [816], k¨ıy (k

is-) ‘to do’: Ta. cey [1957].

In a few cases

∗

e > ¨u, e.g. To. t¨u

lc-(t¨u

lˇc-) ‘to become clear’: Ta. te

l-i, Ka.

l-i [3433].

(v) a. ¯e >

o / #(C

)

−−−

-V

= y); elsewhere, ¯e remains.

In the 26 etymologies provided, roots end in -y only in three cases, derivative -y

(< -i) occurs in 6; the rest are shown to retain ¯e either because they are loanwords from

Tamil, Kanna

da and Ba

daga, or cannot be explained at present (Subrahmanyam 1983:

131–6). Out of the 13 examples given for the sound change, 11 have an apical consonant.

It appears that the rule can be reformulated as

(v) b. ¯e >

o /#(C

)

−−−

-V

=[+apical])

To. p¯ey ‘to thatch’: Ta. v¯ey [5532], To.

r ‘to rise’: Ta. ¯eru [916], To. s

or (s

od-) ‘to

arrive’: Ta. c¯er (c¯er-nt-) [2814].

(vi)

∗

u > ¨u /#(C

)

−−−

-V

∗

cc;or,V

∗

PSD

∗

pul-i ‘tiger’ > Ta. pul-i,To.pus-y [4307], PD

∗

muc- ‘to cover’ > Ta. muccu,

To. m¨uc [4915]. Eleven cases have been shown of which three have long vowels through

compensatory lengthening, e.g. Ta. Ma. ukir,To.

ur ‘ﬁnger nail’ [561].

(vii)

∗

u > ¨ı /#(C

)

−−−

-V

or (C

= [+labial], V

= [+low])

110 Phonolo

gy: historical

and comparative

The segment(s) preceding V

may be a single consonant or a consonant cluster (PP

or NP), e.g. To. p¨ı

l ‘hole in a wall’: Ta. pu

zai [4317], To. k¨ıp ‘rubbish’: Ta. kuppai

‘heap’ [1731a]. All the 12 examples ﬁt the description including one with compensatory

lengthening, To. p

¨ı

l ‘to tumble over’: Ta. pura

l, pira

l [4285]. There are two exceptions

to the above rule (see Subrahmanyam 1983: 6, fn2).

(viii) (

∗

u >)

∗

o > wa/ #(C

)

−−−

-V

= alveolar [+apical, −back], V

= [+low]: one of the environments is enough

for the change)

(ix)

∗

o> w¨ı/#(C

)

−−−

-V

= [−low])

Here, we can ignore the ﬁrst stage. Toda has inherited the Proto-South Dravidian

change of i, u > e, o [C-a (see below). Rules (viii) and (ix) are complementary. They

reﬂect

the developments of PD

∗

o with which

∗

u had merged in Proto-South Dravidian

before [C-a. PSD

∗

kot-a ‘to become short’ > Ta. kurai (-nt-), To. kwar-(kwarθ-) [1851],

PSD

∗

t-ay ‘umbrella’ > Ta. ku

tai,To.kwa

r [1663]; PSD

∗

t-i ‘house’ > Ta. ku

ti,

To. kw¨ıs

‘shed for calves’ [1655].

(x)

∗

u > u /#(C

)

−−−

-u/C

u/C

-Ø

∗

u remains unchanged when no vowel occurs, or -u occurs in V

position. PD

∗

n ‘to

drink’ > Ta. u

n,To.u

n (u

d-) [600], PD pu

z-u ‘worm’ > Ta. pu

zu,To.puf [4312].

(xi) a.

∗

¯o > wa, w¯a/# (C

)

−−−

-V

= alveolar or retroﬂex, i.e.

[+apical], V

= [+low])

Examples: PSD

∗

t-ay v.i. ‘to

break

’ > Ta

t-ai (-nt-),

To.

r-(wa

r-θ-)

[946], PSD

∗

kot-ay ‘to pierce’ > Ka. kor-e ‘to pierce as cold’, To. kwar-(kwar-θ-) ‘to feel cold’

[2168, 1859], SD

∗

k¯o

t-ay ‘west wind’ > Ta. k¯o

tay,To.kw¯a

r [2203].

(xi) b.

∗

¯o > w¨ı, w

¨ı/#(C

)

−−−

-V

(Co = one or more than one consonant;

= [+high] or zero)

PSD

∗

t-unku ‘to be restrained’ > Ta. o

t-unku v.i, o

t-ukku v.t., To. w¨ı

rg-(w¨ı

rgy-) ‘be

crushed’, w¨ı

rk- ‘to crush’ [954], PD

∗

ontu ‘one’ > Ta. onru,To.w¨ıd [990d], PD

∗

k¯o

‘horn’ > Ta. k¯o

tu,To.kw

¨ır [2200].

(xi) c.

∗

w > Ø/# p

−−−

The sequence pw- created by the Rules (xi) and (xii) is simpliﬁed to p, e.g. Ta. porai

‘load’: To. pa

r [4042], Ta. pukai ‘smoke’: To. pax ‘smoke, tobacco’ [4240]; PSD pon

‘gold’ > Ta. pon,To.p¨ın [4570], Ta. ponku ‘to boil’: To. p¨ıg ‘to bubble up’, PD

∗

p¯or n.

‘ﬁght’ > Ta. p¯or,To. p

¨ır ‘quarrel’ [4540], PSD

∗

p¯o ‘to go’ > Ta. p¯o (p¯oy-i), To. p

¨ıx p

¨ı

[4572].

(xii)

∗

o > ¨u /#(C

)

−−−

= y)

PSD

∗

moyal > Ta. muyal ‘hare’: To. m¨u



s [4968], Ta. toy ‘to breathe hard as in asthma’,

Ka. suy ‘to breathe’: To. t¨us

‘to breathe heavily’ [2680, 3512]; there are exceptions and

the conditions of change are not clear.