T
SANSKRIT, PRAKRIT, AND APABHRANCA
1. The Origin of Sanskrit
OMETIME in the course of the second millennium B.C.
Indo-European tribes occupied, in varying degrees of com-
pleteness, vast areas in Iran, Asia Minor, and north-west India.!
The problems of their movements and affiliations are still far
from solution, but on linguistic grounds we postulate a group
conveniently styled Aryan, whose speech can be regarded as
the ancestor of the speeches of India and Iran. Of these
Indian speeches* our oldest evidence is the Rgveda, and the
language of this great collection of hymns is obviously a hieratic
and conventional one. It testifies to the cultivation of sacred
poetry by rival families of priests among many distinct tribes
during a considerable period of time, and in various localities.
Some of the hymns were doubtless composed in the Punjab,
others in the region which in the Brahmanas is recognized as the
home of the Kurus and Paficalas, tribes representing the con-
solidation of units familiar to us in the Rgveda. It is even
claimed that Book vi is the poetry of the period before the tribes
entered India proper, though the contention is still implausible.
That, under these circumstances, the speech of the Rgveda
should show dialectic mixture is only to be expected, and, despite
the great difficulties involving the attempt to discriminate, some
progress is possible towards determining the characteristics of
the dialect which lies at the basis of the Rgveda. It was marked
by the open pronunciation of intervocalic dh, bh, d, and dh as h,
4, and /4; by the change of Z into r; and by the intrusion of
the pronominal instrumental plural termination ebAis into the
1 Cf. Keith, Religion and Philosophy of the Veda, Chap. I.
2 Cf, Wackernagel, Altind. Gramm., i, pp. ix f.; H. Reichelt, Festschrift Streit-
berg (1924), pp. 238ff.; Macdonell, Vedic Grammar (1yto); Meillet, IF. xxxi.
120ff.; JA. 1910, i, 184 ff; Mélanges Lévi, p. 20; Grammont, MSL. xix. 254 ff. 5
Bloch, Formation de la langue marathe (1920); S. K, Chatterji, Bengali (1926).
Ba4 SANSKRIT, PRAKRIT, AND APABHRANCA
nominal declension. Borrowings from other dialects can here
and there be confidently asserted ; in some cases the forms thus
found may be regarded as of equal age with those of the Rgveda,
as in the case of words in / and jajyhati, with 77h in lieu of ks for
Aryan gzh, but in other instances we find forms! which are
phonetically more advanced than those normal in the Rgveda,
and attest loans either from tribes whose speech had undergone
more rapid change, perhaps as the result of greater admixture
with non-Aryan elements, or from lower classes of the population.
Thus we have irregular cerebrals as in kata beside kyta, hata be-
side karta; ch for ps in krchra; jy for dy in jyotis; i for r in
githira;; busa for brga, and many other anomalous forms. To
localize these dialects is in the main impossible ; the rhotacism
of the Rgveda accords with its western origin, for the same
phenomenon is Iranian. The use of / is later a sign of eastern
connexion, and in one stereotyped phrase, sive duhité, we per-
haps find ¢ for az, as in the eastern Prakrit.
From the language of the Rgveda we can trace a steady
development to Classical Sanskrit, through the later Sarhhitas
and the Brahmanas. The development, however, is of a special
kind; it is not the spontaneous growth of a popular speech un-
hampered by tradition and unregulated by grammatical studies.
The language of the tribes whose priests cherished the hymns of
the Rgveda was subject doubtless to all the normal causes of
speech change, accentuated in all likelihood by the gradual
addition to the community of non-Aryan elements as the earlier
inhabitants of the north, Munda or Dravidian tribes, fell under
their control.? But, at least in the upper classes of the population,
alteration was opposed by the constant use of the sacred language
and by the study devoted to it. Parallels to such restricted
evolution are not hard to find; the history of the Greek Koine,
of Latin from its fixation in the first century B.C., and of modern
English, attests the power of literature ¢o stereotype. In India
1 In some cases, no doubt, forms have been altered in transmission.
2 Cf. W. Petersen, JAOS. xxxii. 414-28 ; Michelson, JAOS. xxiii. 145-9 ; Keith,
Camb, Hist, India, i. 109 ff, Common sense renders Dravidian and Munda influences
inevitable, though proof may be difficult ; Przyluski, MSL. xxii, 205 ff.; BSL, xxiv.
120, 255 ff., xxv. 66. ; Bloch, xxv. 1; Lévi, JA.cciii. 1-56. Prayluski endeavours
to prove Austroasiatic influence on culture; JA. ccv. 101 ff.; ceviii. r ff.; BSL. xvi.
9ST. Cf. Poussin, /ndo-errroptens, pp. 198 fi.; Chatterji, i. 170ff., 199.THE ORIGIN OF SANSKRIT 5
the process was accentuated by the remarkable achievements of
her early grammarians whose analytical skill far surpassed any-
thing achieved until much later in the western world. In the
normal life of language a constant round of destruction and
reconstruction takes place; old modes of expression disappear
but new are invented; old distinctions of declension and con-
jugation are wiped out, but new differentiations emerge. In
Sanskrit the grammarians accepted and carried even farther than
did contemporary vernaculars the process of the removal of
irregularities and the disuse of variant forms, but they sanctioned
hardly any new formations, producing a form of expression well
ordered and purified, worthy of the name Sanskrit which the
Ramayana first accords to it. The importance of the part played
by religion in preserving accuracy of speech is shown by the
existence of a special form of sacrifice, the Sarasvati, which was
destined to expiate errors of speech during the sacrifice, and in
the Mahabhasya of Pataiijali (150 B.C.) it is recorded that there
were at one time seers of great knowledge who in their ordinary
speech were guilty of using the inaccurate yar va was tar va nah
for yad va nas tad va nak, but who, while sacrificing, were
scrupulously exact.
The influence of the grammarians, whose results were summed
up in Panini’s As/adhydyi, probably in the fourth century B.C.,
is seen in the rigid scheme of euphonic combination of the words
within the sentence or line of verse. This is clearly artificial,
converting a natural speech tendency into something impossibly
rigid, and, as applied to the text of the Rgveda, often ruining the
metrical effect. Similar rigidity is seen in the process which sub-
stitutes in many cases y and v for the #y and uv of the earlier
speech. Dialectic influence may be traced in the recognition of
Zin many words in lieu of y, and a certain distinction between the
dialect which underlies the Rgveda and that of Panini is revealed
by the absolute ignoring by the latter of the substitution of / and
h for d and dk! Otherwise the chief mark of progress is the
growth of the tendency to cerebralization, possibly under
Dravidian influence.
In morphology there was elimination of double forms; @ as a
variant for eva in the instrumental singular of a stems disappeared,
3 Ch Laders, Festschrift Wackernagel, pp. 294 ff.