Dialects
Dialects
Dialects
To F.B.J. Kuiper
on his 8oth birthday
1. INTRODUCTION
2. Materials for this study
3. Discussion of the materials
4. PARAMETERS
4.1. Geography
4.2. Time
4.2.1 Internal chronology of the texts
2. Development of the ritual
3. Development of thought
4. Absolute dates
5. Relative chronology
5. ESTABLISHING A PATTERN
7.1. khalu
7.2. svid
7.3. Some other particles
7.4. spṛdh : sam.yat
10. CONCLUSIONS
§ 1. INTRODUCTION
It is believed, and quite generally so, that the Vedic language had no
dialects. One usually admits that the archaic poetic language of the Ṛgveda is
a mixture of many dialects which had influenced each other. On the other
hand, the educated speech of post-Ṛgvedic times, found in the prose texts, the
so-called Brāhmaṇas, is regarded as the contemporary, the living language of
the priests and other well-educated men, while the rest of the population
spoke various degrees of early Middle-Indian, i.e., archaic Prākṛts. But this is
as far as one will go.1 My contention will be that even this standard North
Indian Koine, "Vedic," which does not seem to have regional variations
at all, shows traces of the local dialects-- if only one looks carefully enough.
Until now, this has not been done, chiefly because the language is apparently
uniform in all the texts. There are, as has been noted from time to time, a few
words or phrases, like the famous ŚB quotation, he 'lavo he 'lavo, spoken by
the Asuras, which is believed to be from an early Eastern 'Prākṛt' for: he
(a)rayaḥ.2
But otherwise, the sound system and even the phonetical variants of one
particular phoneme are the same, the Sandhis show little variation, the forms
of the noun and verb system seem to be the same throughout the texts, and
the same applies to the syntax.
If there are "Vedic Variants," they are usually attributed to matters of style
or described as late (or post-Vedic) influence of Prākṛt on the Vedic texts.3
Actually, this does not, if one reads the texts carefully, agree with the
testimony of the Vedic texts themselves; there are a number of very clear
statements indicating that the Vedic people noticed and thought about
regional differences in speech:
1
F.Edgerton, Dialectic phonetics in the Veda: Evidence from the Vedic variants, in Studies
in honour of Hermann Collitz, Baltimore (The John Hopkins Press), 1930, p.25-36; -
M.B.Emeneau, The dialects of Old Indo-Aryan, in: Ancient Indo-European dialects, ed.
H.Birnbaum and J. Puhvel, Berkeley and Los Angeles 1966, p. 123-138.
2
Thieme, Pāṇini and the Veda, notes the closeness of Pāṇ.'s bhāṣā and KS, see p. 17,76,80;
cf. below, ann. 65, 239, 280; cf. now Cardona, Pāṇini. A survey of research, Delhi 1980,
p.238 sq.
3
Bloomfield-Edgerton, Vedic Variants, treat only the Mantras belonging to the so-called
Mantra language, see below § 4.2.2; cf. Oertel, Kasusvariationen, SB Aakd. München 1937-
1939. Especially Renou and Caland regarded many of the variations in grammar to be
treated below, as mere variations in the style of Vedic viz. of the various Vedic schools.
5
4
See Weber, Lit.Gesch.p.49, Ind. St. II,309; cf.Thieme, Pāṇ. and the Veda 80; cf. TS 5.3.4.4
with tr. Keith, cf. below ann. 20.
5
About this, author, forthc.
6
Most of these passages have been noticed by W. Rau, Staat und Gesellschaft im alten
Indien, Wiesbaden 1957, p.18. Cf. AV 12.1.45, 'everywhere on Earth there are men of
different speech and customs'; and yatra-āryā vāg vadati KA 8.9 "where the Aryan speech
is spoken" (cf. the similar expression in O.Pers. DB IV 89: āriya- " Aryan language"), cf.
AA 3.2.5; -- the Pañcāla use kuśa instead of darbha, apparently to denote their kings, in
succession to Keśin Dārbhya, who, according to BŚS, was called Śīrṣaṇya Kuśa after
performing the Apaciti sacrifice, see JB 2.100 §133, BŚS 18.38, PB 19.8. Yet, apparently,
this usage is also found in everyday language, see ŚBK 1.2.3.9 kuśā but darbha in the
parallel version ŚBM 2.2.3.11. This passage underlines what will be said below, §4.1., about
the homeland of ŚBK as neigbouring the Pāñcāla area. Passages like TS 7.5.9.2: "all forms
of speech they speak" (at a Sattra), have to be understood, with H.Falk, (Bruderschaft und
Würfelspiel, Freiburg 1986), differently, in terms of the ritual in question. For later texts,
see Patañjali, I p.9, line 25, on dātra, hammati, śavati, and the various words for cow, (cf.
also Pkt. goṇā, gopatalikā).
7
Cf. the names of Keśin Dālbhya; also the various designations of the horse: haya, vājin,
arvan, aśva ŚB 10.6.4.1 etc.; cf. W. Rau., Staat, p. 18.
8
See KS 30.1:181.15, - but only at night!
6
All of these features, however, have either been largely neglected, or have at
least never been investigated in the context of Vedic dialects. Another reason
why one does not have an idea how there could have existed dialects in Vedic
is that one does not really know when the texts were composed or where.
Without an area of composition (or redaction) for a Vedic text, there are, of
course, no dialects.
In fact, the Vedic texts seem to have been composed at an unknown time in
an unknown area (of N. India); in other words, even after some 150 years of
studying the texts, a dark mist still covers the whole Vedic period, which
makes it very difficult to make out who did what, where, and at what time.
The only point usually admitted is the relative chronology of the texts (see
below § 4.2.1, 4.2.5), and even in this area there is no general agreement. I
believe that we can finally move a few steps further. I have tried to localise as
many texts as possible in the Fel. Vol. Eggermont (Louvain 1986/7). The
absolute dates of the texts remain in balance, if we do not take refuge in
external evidence like the Mitanni agreement of ca. 1380 B.C., 7hich mentions
the Vedic gods, or the often discussed date of the Buddha and the age of the
older Upaniṣads.
The results are summarised in the maps and in the tables provided below
where the texts are dated according to the linguistic developments found in
them.10
9
See K. Hoffmann, Aufs. p. 581; the two other cases of RV kur- are 'popular' forms as well.
10
See: Wackernagel, Ai.Gr.I, and the add. of L.Renou; Renou, Histoire de la langue
Sanskrite, Lyon-Paris 1956; K.Hoffmann, Inj., and: Aufs., passim; see especially, J.Narten,
Die Sprache 14 ; cf. also Gonda, Old Indian, Leiden 1971 and OLZ 1977, 205-207; cf.
author, WZKS 24, p.22-24.
7
* 1st: there are regional differences in Vedic (as will be described in detail
in the main part of this paper).11 Unfortunately, this has not been followed up
so far.
* 2nd: these regional differences are not static throughout the Vedic
period, but many of them show developments both in time and space: certain
local peculiarities -- often innovations! -- spread to the next level of texts.
They do not always do so in an Eastern direction, as one might think, with a
view to the history of settlement of N. India, but also in other directions.
When a larger number of such quickly spreading innovations are compared,
a few centres of innovation emerge. It will be interesting to see where they
are situated and what could have been the reasons for the diffusion of
innovations.12
* 3rd: when one studies these variations and their spread in space and time,
the surprising result is a correspondence of the geographical area of some
Vedic schools (śākhās) with that of certain Vedic tribes and with some
archeologically attested cultures. This, ultimately, allows to date the texts for
the first time (see below, § 10.5).
If some of the features mentioned above are local peculiarities, i.e., dialect
characteristics, then the question arises: is there a relationship with the
various early Middle Indian dialects and with the other Prākṛts? The recent
book of O.v. Hinüber on early Middle Indian provides, as far as features
common to Vedic and Middle Indian have been identified until now, an ample
discussion of the relationship between Vedic and the Prākṛts.13 The point has
been discussed earlier by M.B.Emeneau. He concentrates, however, on the
11
P.Thieme is, as far as I can see, the first who has noticed that such regional differences
are clearly mirrored in Pāṇini's knowledge of Vedic texts, and that Pāṇ. is very close to
(N)W texts, KS and PS, see Pāṇ. and the Veda, p.75; cf. now Cardona, Pāṇ., p. 238 sq.; cf.
also K.Hoffmann, Aufs. p.470, about dialect differences in the caste language of the
Brahmins.
12
If true, this alone should be sufficient to disperse the doubts of Caland, Renou regarding
the use of linguistic criteria, found in various Vedic texts, for determining the relative dates
of these texts, as summed up by Minard, Trois Enigmes II, §717-727. For the spread of the
Vedic tribes and their culture to the East, see Rau, Staat, p.12 (where the data are not used
for this purpose, cf. author, Fel. Vol. Eggermont); for the movement towards the South, see
MS 4.7.9:104.14 "people move southwards, conquering," ŚB 2.3.2.2 on Naḍa Naiṣadha who
daily carries Yama (death) southwards, cf. also ŚB 5.3.3.3 : one gets food in the South;
Brāhmaṇic splendour is found in the North, cf. above (ann. 6) on the best speech, KB 7.6.
Cf. finally, JB 2.352 "one brahmin follows the other": tasmād brāhmaṇo
brahmaṇsyānucaro bhavati.
13
O.v. Hinüber, Das altere Mittelindisch im Überblick. SB Akad. Wien 1986, § 7-11.
8
similarities between the Ṛgveda and Pāli. Such discussions must, I think, be
supplemented by the type of evidence to be presented in detail in this paper;
one has to collect items mainly from the texts that precede the Pkt.s, i. e., from
the Middle and Late Vedic texts, and cannot directly compare the Ṛgveda
with the later Prākṛts. It is the Middle Vedic period that saw the diffusion of
Old Indian and of early forms of Middle Indian all over Northern and
Western India.14 The RV territory, however, still was restricted to the Panjab
and its immediate surroundings.
It is not always easy to select materials that are accessible for such a study.
Whatever criteria one wishes to employ for the selection, at present only some
easily accessible materials can be used, like those contained in Vishva
Bandhu's Vedic Word Concordance (VPK), Wackernagel's Altindische
Grammatik, and the grammars of Whitney and Renou, those found in the
introductions to text editions, etc. This means that a thorough investigation
can only be made of the words listed alphabetically in VPK, such as the
spread of a particular word, or of a combination of two words (especially in
the case of particles), which is already much more time-consuming. A
comprehensive study of a particular case ending or of a verb form is not
possible with this tool. Unless one finds the time to read a l l the texts for the
present purpose only, one either has to restrict oneself to an impressionistic
test (as, for example, with the exact number of cases of certain verb or noun
endings), or one has to rely on the grammatical descriptions and the
occasional statistical counts (which, however, do not always include all major
Vedic texts). The lack and the unavailability of complete data will
occasionally be felt in the sequel.
For example, some materials,8like the occurrence in the texts of the opt. in -
īta of thematic verbs,15 are not easily accessible so long as complete lists of
such forms do not exist. These data will, in the future, have to be found in
computer-based data systems which will easily allow one to trace, select as per
14
Panjab to the borders on Bengal, and South to Gujarat and, apparently Vidarbha (Berar,
N.Mahrashtra, acc. to JB ).
15
On -īta, see below ann. 22; see Renou BSL 41 p.11 sqq., and K.Hoffmann, Aufs.371;
Aufrecht, ed. AB p.429, Wackernagel, Ai.Gr., I, German ed. p. XXX = Renou, intro. p. 14,
with ann. 198, further: II.1, p.89; Keith, ed. AA p.172.; Keith, transl. AB, p.46, KB p.75;
also in BŚS, see Caland, Über BŚS, p. 42 "Die meisten also im spateren Teile des Werkes";
and in BhŚS 9.5.3, 10.7.15, 5.16.18, see ed. Kashikar, p. LX; for the DhS, see
S.K.Bharadwaj, Linguistic Study of the Dharamasūtras, Rohtak 1982, p. 119 sqq..
9
chapter, per text or text level, those forms necessary for a certain
investigation.16
As for the criteria to be used in this study, a few remarks have to be added.
16
Earlier statistics help occasionally, notably those of Whitney and his school: see
Whitney's and Avery's statistical accounts in JAOS.
17
See especially the treatment of Mantra variants in J.Narten's and K. Hoffmann's works,
and cf,. already Oldenberg, ZDMG, 42, p. 246 and Keith,TS transl. p. CLIX sqq.
10
RV ḷ- for older -ḍ-; MS,RV -ch-, KS -śch- ), and these, again, have to be
distinguished from later medieval developments which affected the
form of texts, like MS ñch for correct Vedic cch < t+ś or Vāj. viṣṇṇu for
viṣṇu, ppra for pra, etc.
There remain the real Vedic peculiarities, like the changes in the occurrence
and distribution in the texts of r/l or a confusion of s/ś, etc., which are very
difficult to trace without a computer data base.19
Other examples of this type are the loss of the modi of the aorist in middle
Vedic, the loss of the Ṛgvedic case ending in -ebhiḥ, etc. Some other materials
are, again, inaccessible for the time being, like the exact distribution and
18
Cf. author in WZKS 23/24; StII 1; 8/9; VI. Suppl.Bd. ZDMG, 1985.
19
Cf. Ved.Var I, p.134 -154, where only Mantra texts are dealt with.
20
For a further discussion, see author, forthc.; note the higher tones of the Kurus at ŚB vs.
ŚBM 3.2.3.15 tasmād atrottarāhi vāg vadati Kuru-Pañcālatra; ŚBK 4.2.3.15 tasmād
atrauttarāhai vāg vadatīty āhuḥ Kuru-Pañcāleṣu ca Kuru-Mahāvarṣeṣv iti. Note that the
Kāṇvas do not mention the Pañcālas, probably because their territory partly overlaps that
of the Pañcālas (see §4.1). - Note that pathyā svasti in ŚB and KB = North, but = East in
MS; in KS, KpS, TS, AB = all directions? This is important in the context of the eastward
movement of the IA tribes. JB, s.v., has no comparable sentences; cf. ann. 4 and 6.
21
See note Renou, Monographies skts.I: for example, the subjunctive is rarer in KB; or
subj. is found in ŚBK where ŚBM has the future in ŚBM, see Caland ŚBK, p.73. This
indicates that subj. was on the way out in t h i s function, but cf., on the other hand, the
large increase of the number of hypercharacterised subj. in ŚB; see below, §9.6.
11
occurrence of the opt. in -īta (AB,KB, Up.s, BŚS, other Sūtras, later language,
cf.O.v.H, Überblick, § 444 ).22
* Syntax provides some useful materials as well, especially the varying use of
particles, the persistent occurrence of tmesis even in some of the later texts
(Śrautasūtras: BŚS ), etc.
Typical examples in Vedic are the spread of khalu or svid, or of the various
combinations of (u) (ha) (vai), or phrases like eṣā...sthitiḥ; brahmavādino
vadanti, etc.
22
So far attested at: AB 3.19.10, 3.45.7, 4.7.3, 6.21.12 (thus also in the older parts of AB!);
KB 4.4, 19.10, KU 3.8, ChU 6.14.1; PrU 5.1; BŚS very frequent, BhŚS, etc.; note also –iyuḥ
for –īyuḥ in AB,KB.--- Could this be an Eastern development: KB < AB < BŚS, due to the -
e- preterite, see O.v. Hinüber, Überblick § 445 and MSS 36 p. 39 sqq.? In that case, did one
want to make a distinction of forms with e-pret. from the forms in -ī-/-i- of the Opt.? Cf.
analogical forms, O.v. Hinüber, §435: TB sanem instead of saneyam, Pāli labhe (cf. Renou,
Gramm. Skte.§282); cf. also the confusion arising in late Vedic of forms with the augment
a- (meaning "vorzeitig,"/ "pluperfect" value), and the other pret. forms, see below, § 5.2). -
Bronkhorst, in his treatment of Śvetaketu, makes wrong use of some of these data; he
simply attributes the usage of -īta in various texts to the same synchronic level (ĀpDhS =
AB,KB!), without paying any attention to the problems of textual layers, problems of
composition of the texts, redactional activity, dialect spread, and geographical distribution
(AB,KB, Taitt. Sūtras: BŚS, ĀpŚS, BhŚS, etc.). Pure 'diachronic' treatment and
speculation, especially with texts of unknown date and unknown authors, is fruitless.
23
See author in WZKS 23.
24
"this is XY, though...." which carries no meaning whatsoever, except to leave the caller,
psychologically, room to state what he/she wants.
12
* Finally, the occurrence of rare words and their diffusion in the various
areas of Vedic Sanskrit can be studied.
A problem in this context is25 the definition of 'dialect'. When the situation
of Old Indo-Aryan is compared with that of the Old (and Middle) Iranian
languages and dialects, Vedic appears to be a uniform language. It is the
educated speech of the Brahmins, socially separated from popular speech,
only traces of which appear since the RV in such forms as jyotiṣ, words with
'popular' l, etc.26
This applies especially to the post-RV texts, i.e., the Middle Vedic texts. The
earliest occurrence of Middle Indian inscriptions (Aśoka +) and texts (Pāli
canon +) shows that popular speech existed in various dialects, the earlier
forms of which must have concurred with Vedic Sanskrit.27
Their influence can occasionally be traced (see above §1 ), yet even though
some such 'popular' elements do occur, the question is: how many of them
possibly c a n appear in the texts? Popular and local forms generally are
avoided in poetry and in learned 'theological' discussions, like that of the
Brāhmaṇa texts. An exception is provided, as is well known, by the two
Artharvaveda Saṃhitās which contain a great number of rare or otherwise
unknown words, like the names of various sorts of snakes etc., but these texts,
too, have been reformulated by priestly poets.
Yet even given such influence from the more popular forms of Vedic, from
other Old Indian dialects (lost to us), and from the early forms of Middle
Indian occasionally visible in the texts, Vedic seemed to be too uniform, and
the cases of divergence from the norm to be too few, to justify a division of
Vedic Sanskrit into various dialects. Below, I will try, on the basis of
examples from various categories of grammar and style mentioned above, to
show that this can indeed be done.
25
As has also been felt by many participants in this conference.
26
Summed up O.v. Hinüber, Überblick, § 7-12.
27
This has been noticed from time to time, see the summary by O.v. Hinüber, Überblick, §
7 -9.
13
In spite of this, the language of the North had a prestige of its own during
the late Brāhmaṇa period; people went there to study it or liked to listen to
Northern speakers.28
When variations occur in the Vedic texts, they can both reflect the locally
underlying forms as well as represent such more or less widely spread
prestigious forms of, e.g., the Kuru-Pañcāla or the Northern language. These
features have to be distinguished from the special features of a particular
Vedic school which has carried certain peculiarities of phonetic nature
through its whole canon, e.g., Taittirīya súvar for 'normal' Vedic svàr, or
Kapiṣṭhala yunaymi for the usual yunajmi (see below).
In order to distinguish such forms from general and 'real' Vedic ones, one
has to study the tradition of the texts in question as per school, from the late
Vedic period to the Middle Ages, and has, then, slowly to "peel off" the
various layers of textual changes like: medieval writing mistakes (MS ñch <
cch, i.e Maitr. [t ch])29; medieval pronunciation and school habits like viṣṣṇu
< viṣṇu in ŚB; influences of the Prātiśākhyas and of late Vedic orthoepic
diaskeuasis. Finally, one has to establish the authentic form of a text (as
opposed to its original form at the time of composition of the text in question,
i.e., during the Vedic period).30
In the sequel, I will try to show that there was something like a Vedic Koine,
but that this "educated Sanskrit" of the Brahmin community, which they
used, as it is attested for Uddālaka Āruṇi,31 in their disputations, from Madra
(Panjab) to Videha (Bihar), existed in many local varieties based on the
various forms of Old Indo-Aryan and of the underlying Prākṛt dialects
spoken in the particular area. Unfortunately, we have access to only o n e
28
See KB 7.6, and cf. Thieme, Pāṇ. and the Veda.
29
See Lubotsky, IIJ 25
30
For the doubling of consonant in ŚB, VS viṣṣṇu, ppra, etc. see Indian editions and some
of the MSS; this probably goes back to Prātiśākhya influence; apparently, Uvaṭa on this
topic has been misunderstood by medieval scholars.
31
See ŚB 11 and BAU
14
testimony that does not form part of the Vedic canon and yet is closely linked
to it, namely Pāṇini's grammar which records many instances of his local
North-Western educated speech, the "bhāṣā," the probable predecessor of the
Middle Indian Gandhārī.32
§ 4. PARAMETERS
We can now proceed a few steps further, I believe, and consequently, I have
tried to localise as many texts as possible (see Fel. Vol. Eggermont).
32
Note that in one case Pāṇini has a comparatively late form: 6.2.70 maireya < madireya ::
*madirā-, which even in Gandhārī would be possible only much later (see, in this volume,
the article of G.Fussman). Does maireya come from another Prākṛt dialect? cf.O.v.
Hinüber, Überblick, p. 94 §170.
33
See K.Hofmann, Aufs., and P. Tedesco, Language 19, p. 12. Note that bhal- explained by
Tedesco from smar- points to non-Eastern origin of this section: there is no anaptyxis, see
O.v. Hinüber, Überblick §243
15
the geographical data contained in them have only scarcely been studied and
have even less been put to use. K.Mylius, in his studies on the ŚB and the
geographical milieu of Vedic texts, is an exception. A few tentative
localisations had already been made by Weber, Caland, and others.34
SUMMARY OF LOCALISATION:
VSK Kosala (E. Uttar Pradesh), probably excluding the Vatsa country
between Gaṅgā and Sarayū
VSM Videha (N.Bihar), later also S.of the Ganges, on the Andomatis
(Tons? South of Allahabad), see Arrian, Indikë 4.4
Up.s : same area as their schools: AitU, KU, JUB; ChU = more to the
East than PB: "rivers lfow eastw. and westw."; BAUK=ŚBK,
BAUM=ŚBM,
ĀŚS Videha
ŚŚS Pañcāla
LŚS probably in Lāṭī, S.Gujarat
JŚS =JB area
BŚS =BaudhB: in the Vatsa country between Gaṅgā and Sarayū
VādhŚS =VādhB: Pañcāla country on the Gaṅgā
BhārŚS Pañcāla country, on the Yamunā,
ĀpŚS Pañcāla country, opposite of the Matsya
HirŚS Pañcāla country, on the Gaṅgā
VaikhŚS a late text, probably S. Indian
35
See Jaina texts on Saṇḍilla country which J.C. Jain, Life in ancient India as depicted in
the Jain canons, Bombay 1947, locates North of Kāśī.
17
These restrictions applied, one can, in many cases, be fairly certain about the
actual boundaries of the tribes, cf. the statement of ŚB about the Sadānīrā as
the boundary of the Kuru-Pañcāla and Kosala-Videha (ŚBK), viz. that of the
Kosalas and Videhas (ŚBM). The territory of a Vedic school mostly coincides
with that of a particular tribe; this is a point not really noticed so far. (It helps
to explain the confusion in Arrian about the Indian "tribes ," the
Kambistoloi, Madyandinoi of the Indike and the Kathaioi of the Anabasis.)
The boundaries on the maps thus are fairly authentic.
In a few cases, we simply do not know. For example, the exact boundary
between the Kurus and Pañcālas cannot be established on the basis of the
Vedic texts alone. Again, the boundary between the Kaṭha and the Taitt.
territory is not clear at all; it should coincide more or less with the boundary
between the Kuru and the Pañcāla, which in itself is unknown. The same
applies to the Taitt./Vājasaneyi border; most probably it was formed by the
Sadānīrā river which divided the Kuru-Pañcālas from the Kosala-Videhas
(acc. to ŚBK) or the Kosalas from the Videhas (acc. to ŚBM). The river itself
has not been identified with certainty.37 The solution depends on which one of
the ŚB versions is to be taken as authorative. Again, the Eastern border of
the Jaim. territory is unclear; probably the jungle territories of N. Madhya
Pradesh should be excluded, thus a large part of the land to the South of the
Ganges and to the East of the Chambal (=Hvṛṇinī?), i.e., the area inhabited
by Śaphala and Cedi (modern Bundelkhaṇḍ, Baghelkhaṇḍ).
36
See R.L. Singh, India: A Regional Geography, Varanasi 1971, 194-195; N.D. Guhati, in
B.C. Law, Mountains and Rivers, Calcutta 1968, p.348 sqq., cf. O.v. Hinüber, Arrian, p.
1098.
37
See R. Salomon, Adyar Libr. Bull. 42, 1978, p.32 sqq.; he regards the Gaṇḍakībāhu
(Chotī Gaṇḍak) as the original Sadānīrā, (on the basis of the ŚBM passage)
18
This criterion has been used very little so far in Vedic studies, outside of the
ŚB (and AB). However, even a very brief look at the texts, which will be
classified below as the level 3, the Yajurveda Saṃhitās, teaches that a number
of safe conclusions can be made from a brief comparison of the contents of
these texts.
All YV Saṃhitās follow a similar pattern: those (KṛṢṇa YV) texts which
mix Mantras and Brāhmaṇas usually start with the Mantras of the New and
Full Moon and of the Soma rituals. One can, therefore, ask whether this is
not the oldest core of YV ritual. Interestingly, the Brāhmaṇa portions dealing
with these two rituals in the Saṃhitās are found only as appendices to the
treatment of other rituals.38
The mantras of these texts are partly derived from the RV, i.e., before the
redaction of this text; many variants, similar to those found in AV and SV,
are found (see Oldenberg, Prolegomena). Other mantras, especially the short
prose sentences which accompany every action in the ritual, are "new," i.e.,
post-Ṛgvedic, at least in the form they are recorded in the YV. However, they
resemble each other closely enough in all schools allow to suppose a common
38
I will treat this in detail in: The Veda in Kashmir, forthc.; for the time being, see author,
Das Kaṭha Āraṇyaka, diss. Erlangen 1972, introd.; cf. already Oldenberg, Prolegomena on
the smnall Mantra Saṃhitās dealing with the New and Full Moon sacrifice, The Soma
ritual, the Agnicayana, etc. in TS, MS.
19
origin. It can also be noted that as far as the form and the development of
Mantras are concerned, the YV often forms a block opposite the two AV texts
(ŚS,PS) and opposite the SV (JS, Kauth. / Rāṇ. SV); each of these Vedas
forms a block of their own. Therefore, an Ur-YV, as well as Ur-AV and Ur-
SV, could be reconstructed, as far as the form of the mantras is concerned.
The actual contents and the order of the contents of these reconstructed texts
are open to discussion.39
At some period following the RV, a number of Mantras from the RV and
others from an unknown, separate priestly tradition were joined to form the
corpus of the Adhvaryus, the main "acting" priests. Apparently, Ṛgvedic
hymns had such a high prestige already that they were necessarily
incorporated into the YV texts, to enhance the status of the Adhvaryu ritual.
In a way, the Adhvaryus formed their own small Saṃhitās:
Dārśapaurṇamāsa/Soma Saṃhitā and the rest of the rituals in separate small
Saṃhitās constituting the Mantra portion of MS, KS, TS (cf. Oldenberg,
Prolegomena). This goes hand in hand with the development of the Ṛgvedic
hotṛ ("pourer (of ghee)" > "reciter of Ṛgvedic hymns". All of this
restructuring of post-RV ritual necessitated a complex re-arrangement of
texts, rituals, and priestly functions; it took place between the end of the
Ṛgvedic period and the collection of the YV Mantras, as well as the
emergence of early, but lost, Brāhmaṇa-like prose texts, (see K.Hoffmann,
Aufs. p.509 sqq.), and, in my opinion, in Kurukṣetra under the early Kuru
kings (like Parikṣit and Janamejaya Pārikṣita).41
39
Cf. author on AV, in Prolegomena to the AV, forthc., and StII 8/9 on the Caraka texts.
40
The ideology behind this myth will be treated separately; cf., for the time being, author,
FS.W.Rau, esp. ann. 104.
41
See author, The realm of the Kurus, forthc.
20
While the stage was set at that time and the YV Mantras, as well as the lost
Br., were composed, the ritual developed for a long time afterwards, all
through the YV Saṃhitā and the Brāhmaṇa periods. It culminated with the
reformulation of all rituals in Brāhmaṇa form in ŚB and, at about the same
time, in Sūtra form in BŚS.
The stages of this development can be followed; however, we know too little
yet about its starting point, i.e, the Ṛgvedic ritual, and about the relative age
of the various YV texts (e.g., the age of the Vājapeya section in MS, KS, TS,
etc.), to allow this criterion to be used in this investigation. It will be of more
use in the future for counter-checking, when the several sets of dialect traits
will have been worked out.42
42
It has to be noted that a proper procedure for evaluating the growth of the classical
Śrauta ritual has not evolved yet; cf., for the time being, the review of Gonda, The Mantras
of the Sautrāmaṇī..., in Kratylos 26, 1982/3, p. 80 sqq. A better procedure would include:
(1.) the establishment of the nat5re of RV ritual viz. of its traces in the text; (2.) a separate
study of the YV Mantras, the order and contents of which is often more archaic than that of
the Br. portions; (3.) a comparison of the various YV Saṃhitā prose texts with the earlier
material; (4.) a study of further developments in the Br.s and the early Sūtras (VādhŚS,
BŚS).
43
A typical example is the idea of rebirth: is it old, Ṛgvedic, or only Upaniṣadic? Cf., for the
time being, author, 31st CISHAAN, Tokyo 1983.
21
The absolute date of the texts remains in balance,44 if we do not want to use
external evidence like the Mitanni agreement of ca. 1380 B.C., which
mentions the major Vedic gods, or the occurrence of iron45 (first attested in
AV), or the frequently discussed date of the Buddha (who died ca. 480 or 380
BC?)46 and the age of the older Upaniṣads, which is usually linked to this date.
(The exact date is, however, not a serious problem in the present context.)
Patañjali (ca.150 B.C.) presupposes the bulk of Vedic literature, as does his
predecessor, Kātyāyana, to a great extent. What Pāṇini knew of Vedic texts
has already been established by P. Thieme, - without the practical indexes
one can use nowadays (if one is only patient enough to do so). Even Pāṇini
knows of younger Brāhmaṇas and quotes the words upaniṣad and sūtra,
which are attested to in the sense of "literary genre" only in late Brāhmaṇa
and Up. texts (cf. below §10.5) .
Especially illustrative and worthy of mention are the following cases: most
Upaniṣads are technically in fact part of the Āraṇyakas of the schools that
they are attributed to, see author, JNRC I. Or, e.g., TA is a composite text
44
The latest summary in: Mylius, Zur absoluten Datierung der mittelvedischen Literatur,
Festschrift Ruben, Berlin 1970, p.421-31; cf. also W. Rau, Zur ind. Altertumskunde.
45
Note that the first occurrence of Iron in the AV forms a date ad quem (or post quem) for
the Mantras of the AV, at ca. 1150 B.C., see author, Persica 10, p.92, with ann. 122-124; for
a collection of data found in Vedic texts which may be compared with archeological finds,
see various works by W.Rau, all quoted in his last book on the subject, Zur ind.
Altertumskunde, Akad. Mainz, Wiesbaden 1983.
46
See O.v. Hinüber, Überblick, summing up the discussion, § 6; see H.Bechert, The date of
the Buddha reconsidered, Ind. Taur. 10, 29-36.
47
Cf. Gonda, The ritual Sutras, p.471, 496; cf. Minard, Trois énigmes, II §717 sqq.
22
with the very young TA 1 < KaṭhB, another KaṭhB piece (TA 2), some older
Mantra and Brāhmaṇa chapters (TA 3-6, among which 4-5 form the actual
Taitt.Ar. treating the pravargya ritual), and finally the older Upaniṣads (TA
7-9= TU 1-3) and the younger MahUp. (TA 10). Therefore, this text can
neither be classified as Br. nor as Ar. nor as Up. Notably, VādhB
(Anvākhyāna) is still treated as if it were part of the Śrautasūtra of this rare
school (in spite of StII, p. 75 sqq.). In fact, it is a sort of Anubrāhmaṇa of the
Taitt. school. To this category should be added: some parts of KaṭhB (in
fragments), the Br. portions of BŚS (18), ŚŚS (a parallel of AB 7), and GB
(Anubr.of the lost Paipp.Br.). MS 4 9 should be treated together with TA 4-5,
ŚB 14.1-3 and KathĀ as the Āraṇyakas of these schools. Note that MS 4.9
even has some ity eke quotations! As is well known, VS agrees with ŚB only
up to Ch. 25, the rest are various, partly Up.-like, additions, notably the Īśa
Up. in VS 40. Here we find a YV Saṃhitā and its Upaniṣad in one "book."
Note that according to Caland, parts of VS are abstracted from ŚB only.48
When speaking about such categories as "Saṃh.prose," etc., the parts of the
texts mentioned above should be lined up with their proper text level. Some
texts are put in the group representing the genre (BŚS in the Sūtra section
rather than in the late Br. period). I list the Vedic texts in the following, up-to-
date scheme.49
RV
Ṛgveda Saṃhitā (Śākala)
48
See AO 10, p.132, cf. below, ann. 91
49
Only a few later texts, like the many Pariśiṣtas, the later Up.s, and some Sutras are
excluded from the list; the vertical order of the texts is roughly representative of the time of
their composition.
23
perhaps Māṇḍ.)
AB KB PB JB
Aitareya-Br. Kauṣītaki- Pañcaviṃśa-Br. Jaiminīya-Br.
1-5 old Br. (=Tāṇḍya-Br.,
---------------- Mahā-Br.)
6-8 new ṢB
Ṣaḍviṃśa -Br.
(=TāṇḍBr.,26)
AA KA
Aitareya-Ār. Kauṣītaki- ChU JUB
contains: Ār. conts.: Chāndogya-Up. Jaiminīya-
Upaniṣad-
Ait.Up. KU MB Brāhmaṇa,
Aitareya-Up. Kauṣ.Up. Mantra-Brāhmaṇa contains:
Kena-Up.
---------------------------------------------------------------------------------------------------
SŪTRAS: Maśaka-Kalpa Sūtra
Kṣudra Sūtra
VāsDhS GautDhS
Vāsiṣṭha Gautama
Dharmasūtra DhS.
various Pariśiṣṭas
__________________________________________________________________
__________________________________________________________________
---------------------------------------------------------------------------------------------
MŚS VārŚS *KaṭhŚS BŚS Vādh. Bh. Āp. Hir. VkhŚS KŚS VaitS *ĀgŚS-
Mānava (almost Baudhā- Bhārad- Hi- Kātyā- Āgastya
Śrauta compl. yana vāja raṇya- yana ŚrS
Sūtra lost) Śr.S. ŚS keśiŚrS. ŚrS (lost)
Vārāha Vādhūla Āpastamba Vaikhānasa Vaitāna S.
Śr.S. ŚrS ŚrS ŚrS
(uned.) (very late)
25
MGS VārGS KGS/LGS BGS *VādhGS BhGS ĀpGS HGS VkhGS PGS
/ĀgGS
Mān.Vār Kaṭha/ Baudh.Vādh. Bhār. Āp. Hir. Vaikh. Pāraskara
-Gṛhya Laugākṣi- Āgniveśya Sūtra
śūtra GS GS GS GS GS GS GS GS
KauśS *PaiṭhGS
(probably
surviving
in Orissa)
LATE UPANIṢADS:
MU KU MNU IU various
Maitr. - Kaṭha- Mahānārāyaṇa- Īśa- AV-Up.s
Upaniṣad Up. Up. Up. Praśna,
Māṇḍ.-
etc. Up.s
---------------------------------------------------------------------------------------------------
50
See J. Narten, Sprache 14, p.115., ann. 13; cf. K. Hoffmann, Aufs., Index s.v. Vedisch:
Chronologisches (p.702); Minard, Trois enigmes, on textual chron.: I, §231 (gen.-ai, vāva);
26
§ 4.3.1. ṚGVEDA
The Ṛgvedic language stands apart from the following stages in many
respects, and is perhaps better characterised as the last stage of a long period
of Indo-Aryan poetry than as the b e g i n n i n g of Vedic literature. Many
words that occur in RV have cognates or direct correspondences in Avesta,
while these no longer appear in post-Ṛgvedic texts. Another point of interest
is the development of the so-called cerebrals, the retroflex sounds, which
abound in book 8 but are rare in books 3, 4, and 5. A chronology of the
various Ṛgvedic books has been attempted by W. Wüst and others; most
valuable, again, will be an evaluation of the linguistic data, such as the
frequency of the injunctive which disappears quickly in post-Ṛgvedic texts.51
This level includes the Mantras (in verse) and the prose texts of the
Atharvaveda (PS,ŚS), the Ṛgvedakhilāni (RVKh), the SāmavedaSaṃhitā (=
RV, including some 75 new Mantras), and the Mantras of the Yajurveda
(both verse and prose). These texts constitute a separate type of Vedic,
largely unstudied and unrecognised as a distinct entity. These texts have
come down in part from the RV, but have been altered considerably during
§ 597b (subjunctive acc. to Renou); cf. also, author, WZKS 24, p. 22-26 and
Fel.Vol.Eggermont.
51
See in general, Wackernagel, Ai. Gr.I and Renou, introd. gen. with additions; -
K.Hoffmann's statistics, Injunktiv, which closely agree with the results of Wüst (who had
based himself on quite different data). According to both, the books of the Ṛgveda are
arranged in this way:
W.Wüst (see below,§5.4): 9 4| 3,5,7| 2,6| 8 10
K.Hoffmann, Inj., p. 36: 4-6-2-1-7-5-10-3-9-8.
Arnold, Vedic Metre, p.16 sqq. cf. p.48, has a somewhat differing estimate of the age of the
books; this is based, however, on a development (partly, supposed) of the Vedic metre only
(see p. 19 sqq.).
52
For the various forms of the verb kṛ, see K.Hoffmann, Aufs.575.
27
the period of the "free floating" mantra tradition, either by gradual linguistic
change, unnoticed by the transmitters, or by misunderstanding and
subsequent re-interpretation (perseveration). In the AV, the YV, and the SV,
the Mantras have first been collected and gradually codified by a process of
orthoepic diaskeuasis. They have lead a life of their own, conserving a
particular fixed form within the tradition of a particular Vedic school.
Sometimes parts of such texts were taken over by other schools, and then
changed according to their requirements of ritual, etc. In such cases, the texts
were partly changed according to the peculiarities of the new school, and
partly not at all.
The Mantras may differ considerably from the surrounding text53 which, in
the YV Saṃhitās, belongs already to the next level, to that of Saṃhitā prose.
This type of language, which is first found at AV (ŚS 15, PS(Or) 18,27-43) and
in the Nivids and Praiṣas of the RVKh, is the oldest Indian prose; it is
characterised by a number of developments which separate it from the
language of the RV by a considerable margin of style and also of time, the
exact extent of which is still unknown. RV 1o already shows some
developments (kuru for kṛṇu, the many sorcery hymns, etc.) which link it to
the AV; these developments, however, had taken place completely by the time
of the AV and YV Mantras. The loss of a whole catagory of the verb system,
that of the injunctive, is apparent in the AV and YV Mantras. In AV, only
some 50 forms54 still show a contemporary, 'living` use of this category. The
same applies to the allegro forms of kṛ; only the "popular" forms (karomi,
kurmaḥ, etc.) are found in non-RV contexts.55 A shibboleth is the
replacement of viśva-- "all" by "sarva" which now comes to mean both
"whole" and "all."
§ 4.3.3. SAṂHITĀ-PROSE.
This type of Vedic is again separated from Mantra type Vedic by a number
of developments. Again, a certain amount of time was necessary to effect this
change. During this intervening period, the first prose texts explaining the
ritual were composed, but they have not come down to us, except for some
rare fragments.56
53
Cf. Oldenberg, ZDMG 42, p. 246, cf. Keith, TS transl. p. CLIX sqq.
54
See K.Hoffmann, Injunktiv, p. 36, 106
55
See K. Hoffmann, Aufs. p. 575 sqq.
56
See K.Hoffmann, Der Mantra yan navam ait, Aufs. p.509 sqq. and cf. the contents of the
lost Caraka-Saṃhitā, StII 8/9 which predated MS,KS.
28
Thus, while the Mantras had already reached a shaky equilibrium and a
certain sanctity, their explanations were developed gradually, to be collected
and codified at first in the extant YV Saṃhitās. There they take the form of
Brāhmaṇa style explanations and discussions (which are not to be confused
with the Brāhmaṇas, the next level of texts). The developments which
characterise the YV-Saṃh.s are: the complete loss of the injunctive as a living
category; the modi of the aorist (subjunctive, optative, imperative) disappear;
of those, only the inj. with mā and the precative remain.57 Another
innovation is the development of periphrastic aorist forms, both for the aor.
ind. as, very rarely, for the aor. precative. These forms were so curious to
Pāṇini that he made a list of those he knew from the Vedic texts current in his
time and his area of North-Western India.58
The many tales occuring in the YV-Saṃh.s are told in the imperfect only.
The imperfect expresses, as it is well known, the past action longer separated
from the time of the narrator, while the aor. expresses the immediate past
(just as taught by Pāṇini, 3.2110 sqq.).
This level of texts comprises the Brāhmaṇas proper, i.e., those of the RV,
YV, SV and AV. Actually, it should be divided into two sub-levels, that of the
earlier and that of the later Brāhmaṇas. On the other hand, the older
Upaniṣads (like BAU, ChU, JUB) should be included here, as well as the
Vādhūla-Anvākhyānas and some of the oldest Śrauta Sūtras, like
Baudhāyana, Vādhūla, and parts of the ŚŚS and JSS. The exact classification
of all of these texts is still a problem, one which has largely been unnoticed.
One should, perhaps, establish yet another sub-level for the Anu-
Brāhmaṇas. Among these are the Vādh.Br (Anvākhyāna), the Gopatha-Br.
(as Anubr. of the lost Paipp.Br.), and the later parts of the KaṭhB, which has
come down to us only in fragments.
Typical for the Brāhmaṇa level of texts is the complete loss of the
periphrastic aorist, interesting in view of Pāṇini's (relative) date.
Compounds like yat-kāma- "having a particular wish" occur for the first
time, and adverbs like sāyam "in the evening" are used as a base to form new
57
No modi of aor., only precative: K.Hoffmann, Aufs., p.469, 502, 587.
58
On the periphr. aor., see K.Hoffmann, p. 469 ann. 5; Pāṇ.'s collection is important for his
date post quem, for he lived after the period of the YV Saṃh.prose; for his lower limit, see
below, ann. 102, 358.
29
ones in other case forms: sāyāt, sāye.59 In another late Br. text, ŚB, a large
number of the so-called hyper-characterised subjunctives occur.60 Finally,
some of the later portions of the older Upaniṣads (ChU 6) show the
considerable influence of a more popular form of spoken Sanskrit.61
This last level of Vedic comprises the bulk of the Śrauta and Gṛhya Sūtras.
In some of them, the content rather than the language is Vedic. The
Upaniṣads that are later than BAU, ChU, JUB, KauṣU, TU, AitU also belong
here (i.e., texts like the Kaṭh U, Maitr.U, etc.). Late Up.s (of Sectarian origin)
have to be excluded, of course, since they are definitely post-Vedic.62
The change from late Vedic to early classical Sanskrit must be investigated
separately, as well as the similar forms occuring in Epic Sanskrit. What
language did the authors of the Sūtras have in mind? Certainly, the grammar
of these texts has been 'corrected' later on, to some extent, according to
Pāṇini's rules, as these texts were regarded only as the work of human
authors, as Smṛti, in opposition to the revealed texts, Śruti; however, even
some earlier forms of Vedic, e.g., RV Sandhi, were changed by later
redactors. - Some of the Upaniṣads, like the famous Kaṭhopaniṣad, exhibit a
larger number of forms based on Middle Vedic, like a pronunciation [bhoti]
for bhavati, as the metre indicates.
Finally, after the last level of Vedic, there is Epic Sanskrit, with its loss of the
subjunctive, the complete breakdown of the Vedic verb system, etc.63 Its
59
On yatkāma-, see K.Hoffmann, Aufs.133 sq.; on sāyam, p.344, ann. 2.
60
See K. Hoffmann, Aufs. p. 30-31, and cf. the investigation by L. Renou, Monogr. Skt.; for
ŚB, cf. also Minard, Trois enigmes I § 2, who sums up the "stylistic/linguistic differences"
of its 14 books.
61
Cf. P. Tedesco Lg.19, p. 12 sqq.: smāryase > (ni.)bhālayase.
62
For the Muṇḍ.Up., see R. Salomon, WZKS 25,1981, p. 91 sqq.; and the same, in another
article of the present volume. Cf. also Epic forms like vṛṇute < vṛṇoti in later Up.s: KU,
MaṇḍU, ŚvetU, see J.Narten, Sprache 14, p.127. Cf. further the Vedic viz. post-Vedic
features in four newly publ. Up.s: N. Tsuji in Fs. Belvalkar.
63
See Holtzmann's gramm. investigation of the Mahābhārata and K. Meenakshi, Epic
Syntax. New Delhi (Mehar Chand) 1983, pp.XI , 231; cf. Review of H.H. Hock in IIJ
(forthcoming).
30
They differ from Vedic in many ways. First of all, their different character
is known to the Vedic texts themselves. For example, at AB 7.18, the priests
answer to Śruti stanzas used in the Rājāsūya with o, but with tathā after a
non-Śruti stanza. The content of the many Yajñagāthās is a historical one;
they tell about the deeds of kings who had offered the Aśvamedha sacrifice,
etc. In this regard, they look like predecessors of the Epic, especially when
they speak about the Pārikṣitas.
64
See P. Horsch, Die ved. Gāthā- u. Ślokaliteratur, Bern 1966; cf. Renou, Histoire de la
langue Skt., p. 38 and Fs. Weller, p. 528 sqq.(cf. ann. 357).
65
See See P.Thieme, Pāṇ. and the Veda, p.76, cf. p.17, 80.
66
Patañjali calls them gau-: "The Bāhīka is an ox," see A.Wezler, Paribhāṣā, p.248 sq.; cf.
StII 10, p.234; this nickname may have been derived from the designation of one of the
Panjabi neighbours of the Kurus, the Mahāvṛṣa ( note vṛṣa 'bull'; cf. the name of the main
wife of the king mahiṣī).
67
Such names are revealing. Note that the West is bāhīka, the East asurya (ŚB), the SE
with its Muṇḍa tribes is udantya "foreign," the South has the foreign looking tribe Maraṭa
(PS; cf. Kīkaṭa already in RV); the extreme North (Himalayas) is inhabited by the Kirāta.
The Kuru-Pañcālas form the Centre. Cf. also the Majjh.Nikāya on such border peoples like
the Yona, Kamboja, tr. p. 149.
31
immediate surrounding dialects, cf. for example, his remark on the names of
wells (!) North of the Beas!68
The overall linguistic situation, therefore, might have looked like this during
the late Vedic period:
antiquated
high/ (Ṛ g v e d a )
literary
Skt. M a n t r a l a n g u a g e : AV,SV,RVKh,YV mantras)
contem-
porary Middle/Late V e d i c d i a l e c t s
high/lit.
Skt. (YV Saṃhitā prose/ Brāhmaṇas/ Upaniṣads / Early Sūtras)
68
Pāṇ. 4.2.74, see P.Thieme, Pāṇ.and the Veda, p. 77.
69
Cf. on KB 7.6 and the language of the North, see P.Thieme, Pāṇ. and the Veda p.80; cf.
the comments on the Yajñagathās by Renou, Fs. Weller, p. 534: partly non-Vedic, partly
non-Pāṇinean.
32
pop. P r ā k ṛ t d i a l e c t s
speech
W: DRAVIDIAN: S: D r a v i d i a n lg.s
~Brahui (Andhra, etc.)
__________________________________________________________________
Against this background sketched so far, we can now begin to observe and
register some of the divergencies in the Middle Vedic texts. I largely refrain
from dealing with the Ṛgvedic evidence as this text is clearly much older and
also geographically limited to the Panjab and its immediate surroundings. A
Mantra text like the AV knows of the N. Indian plains of the doab/ U.P.;
Mantra language in general still have as their centre the area "where the
rivers flow westwards and eastwards," i.e., the Kuru country on the Beas,
Sarasvatī and the Yamunā (see Fel. Vol. Eggermont).
The following investigation first deals with a few typical divergencies found
in the various texts which tend to establish a dialect pattern (§ 5-6); then more
cases typically distinguishing one text or school from another are adduced (§
6-7), and this is enlarged upon by the study of some individual words, and by
additional materials taken from syntax and style. Finally, the relationship of
Vedic dialects with the early Prākṛts is investigated (§9).
33
*************
§ 5 ESTABLISHING A PATTERN
§5.0. Introduction
First of all, I will try to show that certain peculiarities are to be found only in
some Vedic texts, i.e., in a certain geographical area only (see the maps).
A calculation of the size of the various texts has been made in akṣaras, on the
basis of a few typical pages per text, and taking into account the relation of
prose :: verses;71 in a second stage, the size of the various Vedic texts has been
compared in percentage to that of the RV which is set at 100%:
____________________________________________________________
RV72 397.265 akṣaras RV = 100 %
ŚS 176.389 ŚS = 44.44 %
PS 274.560 PS = 69.11 %
MS 392.619 MS = 98.83 %
KS 337.808 KS = 85.03 %
TS 281.569 TS = 70.87 %
71
Some of the texts that have been printed interspersed with commentary only (like TA,
TB) have been difficult to calculate. Also, the exact relation of Mantras :: prose is difficult
to estimate. The percentages given here should therefore be taken with caution; it will be
better to rely, for the moment, on the absolute figures of occurrences of a particular
phenomenon; the exact percentages wil have to be recalculated once a computer data base
of the various Vedic texts is available, and exact figures can be obtained more easily.
72
Note the various new abbreviations introduced here for the convenience of distinguishing
various levels of texts:
PSk = PS calculated here acc. to ed. Raghu Vira, i.e. without some parts of PS 18 (Yama
hymns).
MSp + MS Saṃhitā prose (different from Mantra portions)
KSa = Aśvamedha portion of KS (book V, perhaps < TS)
ABo = AB older part: 1-5
ABn = AB newer part: 6-8
TBk = TB, Kāṭhaka portion: TB 3.10-12
TAk = TA, Kāṭhaka portion: TA 1-2
JBa = JB, Agnihotra portion: JB 1.1-65, younger than the rest
JBc = JB in ed. Caland (Auswahl)
ŚBMo = older portion: ŚBM 1-5
ŚBKo = older portion: ŚBK 1-7
ŚBMw = Western (Śāṇḍilya) portion: ŚBM 6-10
ŚBMn = newer portion (originally < ŚBK): ŚBM 11-13
ŚBMa = Āraṇyaka portion: ŚBM 14.1-3
ŚBMu = Upaniṣad portion: ŚBM 14.4-9
VādhB = Brāhmaṇa (Anvākhyāna) portions of the so-called Vādhūla Sūtra, which actually
consist of at least two separate texts, the VādhB and the actual VādhŚS, see StII 1, p. 75
sqq.
BŚSb = Brāhmaṇa portion of BŚS= book 18 (and some section in 17); ŚŚSb + Br. portions =
15.17-27, occasionally Br. portions in other ŚS, too (not used here).
35
AB 137.413 AB = 34.59 %
ABo 1-5: 92.002 ABo= 23.15 %
ABn 6-8: 45.411 ABn= 11.43 %
AA 29.896 AA = 7.52 %
KB 112.320 KB = 28.27 %
@KA
TB 285.474 TB = 72.10 %
TBk 26.918 TBk= 6.77 %
TA 207.658? TA = 52.27 %?
TA 1-6: 169.470? TA1-6:42.65 %?
JB 430.920 JB 108.47 %
JBc 87.127 JBc 20.41 %
JBa 1.1-65: 24.360 JBa 6.13 %
PB 130.124 PB 32.75 %
It is well known73 that in Brāhmaṇa texts the gen./abl. sg. of fem. nouns in -ā,
-ī is not formed with -(ā)yāḥ but in -(ā)yai, viz. of -i not in -eḥ but -ai.
The reason for this development does not so much seem to be based on a
shift in syntactical usage, e.g., because the Gen. takes dative function74
(Speyer, Ved.Syntax, § 71),75 but rather, the homonym form will be due to a
collapse of forms into a homonym shape under certain Sandhi conditions.
While this trait is usually thought to be limited to the Brāhmaṇa texts, and
to occur first in TS,78 the first few traces of this development can already be
found in the Mantras of a few Saṃhitā texts.79 In fact, the first two instances
of a gen. asyai occur in the AV (Śaun.); AV 3.25.6 and 4.5.6 have this form in
all available MSS, while the corresponding passages, PS 4.6.6 and RV 7.55.5,
have sarve. Is this a late change, affecting only AVŚ? One can compare also
RVKh 2.6.18, a (later) appendix to the older part of the Śrīsūkta; see
Scheftelowitz, RVKhil., p. 78.
73
See Caland, On a paragraph of Vaidic Syntax, AO 5,1926, p. 49-51; he does not, however,
regard this peculiarity as having chronological value, cf. tr. PB p.XIX; see in detail, Wack.-
Debr.III, §15d,p.39; p.135; §68 a‚; p.150 §75a; Caland, Über BŚS p.45; Oertel KZ 63, p.206
; cf. Whitney, Skt.Gram., § 364d, 307h,336g,363c; McDonnell, Ved. Gr. for Students, p.88,
ann.2, p.89, ann.1; Bloomfield, Ved. Var. III, p. 57 (cf. 61-62); Keith, TS introd. p.145;
Minard, Trois enigmes I, §231a; "no chron. value"; also O.v. Hinüber, Überblick, p.150
§334. The same applies, of course, to the u-stems which are not taken into account here: see
Wack.Debr.,Ai.Gr. p.150 §75; dhenoḥ ŚBK 4.1.2.12, but dhenvai ŚBM 3.1.2.21; also of fem.
stems in -ū: vadhvai in ĀpMP, see Ved. Var.III §144; cf. Wack.-Debr.III 1 §97 sq
74
For abl. fem. -ās in dative function, see Edgerton, Ved.Var.III 39-40, §15d in mantras;
Oertel KZ 63,206: PB 18.5.9 prajāyāḥ; ŚB 14.9.4.18 = BAUK 6.4.19 itarasyaḥ; in KS, MS
rarely -ās as dative.
75
asya is often used "as instr.": Speijer, Ved. Syntax, §69 asya bhavati, JB 1.46.1 asya
spaṣṭam syāt, ŚB 6.2.2.39 tad asya-atrāptam bhavati.
76
Cf. Wack.I, §338, §285 b‚A); for the older Sandhi with -y-, see below §6.7.
77
Note also that, for ex., the i-stems never use -aye as gen./abl.
78
By Brāhmaṇa, most scholars errouneously or, at least, vaguely, intend the Brāhmaṇa
(prose) sections of the YV Saṃhitās (level 3) and the Brāhmaṇas proper (level 4).
79
Materials (complete for the older texts?) in Wack-Debr., Ai Gr., III, 1 p. 40 §15.
37
In addition to these stray findings, the first group of cases occurs in another
Mantra text, namely in the Mantras of TS, the text of a central North Indian
school, of level 2 but not in the parallel versions of MS, KS; TS 1.1.13.3 reads
(note -ai for the abl., pā with abl.): pāhí prásityai / dúriṣṭyai / duradmanyái /
dúścaritād, but KS 1.12 and KpS 1.12 read praśityāḥ, duriṣṭyāḥ,
duradmanyāḥ; MS 1.1.13 lacks the mantra.80
While MS is somewhat older than KS, this Saṃhitā generally does not agree
with TS in the usage of -ai; KS, in fact, participates only in the late book V,
the Aśvamedha chapters, which Bhave regards as having been taken over
from the Taitt. school in a wholesale fashion.81
This distribution of the evidence thus leads to the surmise that these forms
are indeed later intrusions into the Mantra texts, with the exception of the
Taitt. texts, as will be seen presently. In the next two text levels, that of YV
Saṃh. prose (lv.3) and Brāhmaṇa prose (lv.4), the evidence clearly points
toward the Taittirīyas as the originators, or at least as the centre of the
diffusion of this phenomenon.
The case of AB is more complicated. It is well k.own that books 6-8 of this
text are later. However, the distribution of -ai vs. -āḥ does not completely
agree with these divisions. Aufrecht, ed. AB, p.427, mentions: -ai in the
younger books 7.27, 8.2, add: 8.15, but also twice in the older books where one
would not expect the form: at 1.27, 4.27. On the other hand, the Western
form in -āḥ appears, as expected, in the Western books 1.9, 1.23, 3.14, 4.10,
but also in the Eastern books 6.3, 6.32, 7.27. (For an explanation, see below.)
80
The older texts, like KS, etc., have -ās: KS 1.12:7.20, and in the parallel passage, MS. -
Cf. -yai in: TB 3.3.9.9, VS 2.20, ŚBM 1.9.9.20. - The list of cases of (-a)yāi of ā and ī stems,
viz. -yai of i stems, provided for TS by Keith, TS tr. p. CXLV sq. contains, upon checking,
no example from a Mantra portion.
81
See his dissertation, Die Yajus des Asvamedha, Bonn, p. 55 sq.; cf. Edgerton, Ved. Var.
§143: KSa ādityai pājasyam < TS.
82
Oldenberg, Prolegomena, p.380 sq., 491 sq., 510; see below, on this person, ann.97. Note
the same tendency as in RV Pp. to use -ai for the gen. in the Padapāṭha of MS, see Ved. Var.
III §152, which may be indicative of an earlier date than usually assumed for this text; also
in TS-pp., see Keith, TS transl., p. CXLV.
38
83
In ādityai pājasvam, see Ved. Var. III §143; MS has -āḥ in the same Mantra; note that
Bhave, Die Yajus des Aśvamedha, holds that KSa stems from TS.
84
For example, 4 §28b dīkṣāyai, etc.
.85
For example: etasyai 2.11; see Caland, Über BŚS, p. 45: etasyai, chāgāyai etc.; abl.:
uttarāyai śroneḥ, etc.
86
See Kashikar, ed. BhŚS, p. LX.
87
See Garbe, ed. ĀpŚS III, p. VI, cf. Keith, tr. TS p. CXLVI.
88
See Gaastra, ed. JŚS, p. 26.
89
Caland, ed. ŚBK, introd. p.37: dhenvoḥ ŚBK, dhenvai ŚBM; cf. ann. 91
90
For example, 4.17.8 asyāḥ... vidyāyā, 3.1.2 tasyai; for the Up., see Fürst, KZ 47, 14 sqq.
91
Caland, AO 10, 132 sq., see ann. 48.
92
On a few divergent Kāṇva forms in -ai see Edgerton, Ved.Var.III, §137 sqq.
39
Therefore, the occurrence of -ai forms in VSM, which at first glance seems
to be early, may in fact only be due to the redaction of the text. This can show
the way to an understanding of the few occurrences in the Padapāṭha and
AVŚ as well; the Śaunaka version of the Atharvaveda has a more Eastern
homeland than the Paippalāda text.93 ŚS may very well have been influenced
by the Taitt. forms predominant in Central N. India (Pañcāla country).94
It can be seen that the sudden appearance and predominance of the Aitareya
Ṛgvedins in the East replaces an older Eastern RV. That this Eastern text
was a reality is evident from the statement of ŚB 11.5.1.10 about the
Purūravas hymn having only 15 stanzas, as opposed to the extant Śākala
version with 18.
93
See Fel. Vol. Eggermont.
94
Or by the few divergent Kāṇva forms if they indeed are as early as the redaction of ŚS.
95
Cf. the Saggala of the Greeks, see author, Fel. Vol. Eggermont; Śākala ritual at AB 3.43;
for Pāli Assalāyana, cf. Mylius, Fs. Ruben.
96
See ann. 222.
97
In this connection, note that Pāṇini knows of the Eastern grammarians and of Śākalya,
but also about the Vṛji, later on a confederation of Videha tribes otherwise first known in
the Pāli sources (as Vajji/Vṛjji). The Vṛji (sic 4.2.131) of Pāṇ.'s time, however, still seem to
reside in the Panjab, as they are mentioned together with the Madra, see below, ann. 320.
40
1. The origin and first emergence of the gen. fem. in -ai instead of -āyāḥ
occurred with the Taittirīyas. This took place as early as the late Mantra
period, and is clearly evident in the Saṃhitā prose of TS.98
3. The phenomenon spread to whole dialect areas only during the Brāhmaṇa
period:
Secondly, the Baudhāyanas are of interest. Caland has pointed out that
Bodhayana (or Baudhayana) had originally been a Kāṇva, who later became
a Taittirīya, and subsequently the first Sūtrakāra. If this is true, he must also
98
Note that Pāṇini knows of Tittiri as promulgator of mantras (*tena proktam, 4.3.101 sq.),
but does not mention the fem. gen. in -ai. As usual in his grammar, this could either mean
that he did not know of the gen. in -ai, i.e., it had not yet developed at all, and that the
occurrences in the Mantras of TS are due to later changes, or it could mean that Pāṇini did
not accept these forms as good (Vedic) Sanskrit. He certainly should have mentioned the
earlier occurrences in the older part of AB, as he knew the text, and it was composed and
transmitted near the Beas, an area where P. even teaches the names of wells! - The forms in
-ai therefore are most probably post-Paṇinean - at least in AB (TS may be a different case
altogether, see above) -, and were introduced into AB 1-5 only at the time of redaction of
the text in E.India.
41
have changed this dialect.99 We know that even a Kosala prince should not
speak like the Easterners (see above §1); in the case of Bodhāyana, this would
mean that a Brahmin from Kosala should also emulate the speech of the
Kuru-Pañcāla Brahmins, in this case, that of the Pañcāla Taittirīyas.100
While the origin and the spread of the gen.in -ai101 is a good example of the
influence of a centrally located innovative area, the following case, that of the
spread of the narrative perfect, is a late phenomenon that began in the East
and subsequently moved westwards very haltingly, so that it did not reach
Pāṇini at all, but still affected, in late Vedic, the Western Kaṭha and Maitr.
texts.102
Since Whitney's investigation104 into the use of the imperfect and perfect in
the Saṃhitās and Brāhmaṇas, we know that the older texts, i.e., the
Yajurveda Saṃhitās and some of the Brāhmaṇas, use the imperfect to tell
stories, legends, etc., a feature corresponding to Pāṇinis's rules (3.2.110 sqq.).
However, the younger Brāhmaṇas, especially the ŚB, tell such stories in the
perfect tense.105
---------------------------------------------------------------------------------------
99
Note: precisely as he had to codify both the Taitt. and the Vāj. material, or at least had to
compare and reconcile them.
100
Or rather, his disciples and redactors used this dialect. BŚS often quotes Baudhāyana;
this means, of course, that the text was redacted by his pupils viz. his school of ritual.
101
Earlier scholars were not convinced that -ai could be used for purposes of dating the
texts, see, Caland, tr. PB, p. XIX, ann.1 "no criterion of posterity or priority" [of PB vs.
JB]; cf. AO 5, p.51; similarly on the use of the imperfect vs. perfect, PB tr. p. XX; cf. Renou,
above ann. 12.
102
Note: This is important for Pāṇini's date! He cannot be of the late KaṭhB period; cf.
ann. 358.
103
See Delbrück, impf. + vai, perf. + ha, see also Oldenberg, Prosa; Keith, AA transl. p.60,
172.
104
Whitney, PAOS May 1891 (JAOS 15), and: On the narrative use of imperfect and
perfect in the Brāhmaṇas, TAPA 23, pp. 5-34; Keith, introd. to transl. of TS, RV Br., AA.
105
The older texts (MS, KS, KpS, TS, TB, TA, AB 1-5; ŚB 6-10, KaṭhB) have preserved the
use of the imperfect, while the younger texts make use of the perfect (ChU, BAU, ŚB 1-5,
11-14; AB 6-8; Vādh.B., as well as a few very late portions in such texts as the KaṭhB = TB
3.10.11). Cf. Keith, transl. of TS, p. CLIII sq., transl. of Ṛgveda-Br., p. 85 sqq.; Oldenberg,
Zur Geschichte der altindischen Prosa. Mit besonderer Berücksichtigung der prosaisch-
poetischen Erzahlung. Abh. d. Kgl. Gesellsch. d. Wiss. zu Gött., Phil.-Hist.Kl.Bd.XV Nro.6,
Berlin 1917, p.25 ssq.
42
MS 2237
35
KS (unedited TS 1900
at the time, 27
similar to MS)
ŚBMo(1-5) 1107107
941
--------------
(6-9) 1504 }
TBk 1426 148 } 1823
25 (10) 319 } 254
106 }
-------------------------------------------------
TAk & 5 (11) 198 }
136 258 }
4 (12) 123 } 453
ABo(1-5) KB 263 65 } 349
929 149 (13) 132 }
27 26 }
-------------------------------------------------
106
Cf. Keith, tr. TS, p. XCVII, CI, CII, and RV Br., p.86.
107
Caland, ed. ŚBK, p. 70, counts 205 perfects in ŚBK where ŚBM has impf.; and 77 impf.
in ŚBK where ŚBM has perf. Altogether, the Kānva books (1-7) have 1265 perf., as against
993 imp. in the corresponding Mādhy. books (1-5). Cf. also Minard, Enigmes I, § 118,
168b, 441c; II, §140, 163b, 180, 717, 147b, 474a.
108
Note that KaṭhB (in Kāṭhakasaṃkalanam, ed. Sūrya Kanta, Lahore 1943, repr. Delhi)
generally still uses the impf. as narr. tense, as does KaṭhA; it is only these late chapters
(dealing with some special forms of the Agnicayana) which employ the narr. perf.
109
The ratio would be higher if the Up. (BAU) would have been counted separately; note,
however, that even the Ar. (ŚB 14.1-3) already employs the narr. perf.
43
(5 ch. mixed)
BŚS (Br.ch.in stories)
7 ch. impf./
8 ch. perf.
PB 1433 ŚSS (note the perf.
11 in Śunaḥśepa legend)
-----------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------
0.76 % PB
14.31 ŚBMw
? KS
1.42 TS (19.69 ŚBMn 13)
1.56 MS (33.22 ŚBMw 10)
1.75 TB (52.84 ŚBMn 12)
2.9 TA
2.9 ABo 23.41 JB
111
? KaṭhA 28.90 KaṭhB (>TBk,TAk)K
36.76 JUB
56 KB
77 ŚBMn
85 ŚBMo 1-5
134 ŚBa/u 14
153? ŚBKo
9.86 % ŚBMw (6-9) 158 ABn 6-8
443% ChU112
---------------------------------------------------------------------------------------------------
110
Was ChU already composed in the East? Cf. the supposed movement eastwards of part
of the Kauth. (PB), see ann. 223, 319, 334.
111
KaṭhA uses the impf. in narration; the percentage should be close to MS, TS, and ABo
112
The figure is surprisingly high. Was ChU already composed in the East? (note that the
text mentions that the major rivers flow east- ands westwards); cf. below, ann. 319, 334, etc.
for indications that late Kauth. texts (KB) were composed/redacted in the East; cf. also the
SV Kṣudrasūtra with Br. passages.
44
Taken at face value, these statistics seem to indicate that a number of texts,
namely PB, TS, MS,113 the older AB (1-5), and even the Brāhmaṇa texts of the
Taittirīyas (TB, TĀ), have virtually no narrative perfects.114
Oldenberg, (Prosa, p. 25 sqq.) pointed out the use of the perfect in narrative
prose in combination with the particle ha. This agrees with one of the
functions of the perfect, namely to state facts, and with the occurrence of the
perfect in texts otherwise relating in the past tense (MS, KS, TS). He believes
(ann. 2) that the change to a usage relating in the perfect tense had its origin
in a wish to indicate, during a narration, the (present) r e s u l t of one action,
or, in fact, one after another (which required the perfect in the YV
Saṃhitās).116
Notably, all of these texts which use the imperfect in narrative sequences are
of a clearly Western origin, except for TS-TB-TĀ, which stem from the
central area (Pañcāla). Interestingly, ŚB 6-10, i.e., books which have been
known, since Weber's investigation in middle of the last century, to have been
imported from a more Western area to Eastern India, show a very low
percentage of perfects as well, especially when compared to the Eastern books
1-5.
The Eastern parts of ŚB clearly constitute the area of the usage of the
narrative perfect. However, the nearly equal distribution of perfects and
imperfects in the earlier books of ŚBM (1-5) indicates that even these texts
113
KS unfortunately has not been counted, but judging from experience, I think that KS
comes very close to MS/TS cf. the perfects: 31.2 :3.8 tad u ha smāhur Dārteyāḥ <> MS
4.1.3:5.4, KpS 47.2; KS 32.2:20.19 atha ha smāha Kapivano Bhauvāyanaḥ; KS 25.7:112.7
atha ha smāha-Argalaḥ Kāhoḍiḥ (cf.Kahoḍa); KS 26.10:135.3 atha ha smāha Āruṇa
Aupaveśiḥ; KS 34.17:47.2 etad dha vā uvāca Vāsiṣṭhaḥ; KS 26.9, end: uvāca Śyāparṇas
Sāyakāyano 'ṣāḍham Kaiśinam...(The Kuntis conquer the Pañcālas).
114
I quote part of the following section from my article in the Fs. U. Schneider.
115
Unless one wants to be especially archaic, cf.the development of the use of tenses in
recent modern German. While the brothers Grimm, adapting folktales, still use, together
with North (Low) German and Dutch the past tense actively, colloquial modern High
German largely avoids this tense and uses perfect instead.
116
Cf. Oldenberg, Prosa, p. 27 ann. 2, on the preference for sa hovāca, te hocuḥ, etc.; cf.
Delbrück, Syntax: impf. + vai; Whitney, TAPA 23.
45
were composed on the basis of an earlier version using the imperfect. It was
only by the time of the assembly of the materials and the actual composition
of this Brāhmaṇa (during the late Br. period), that the usual narrative tense
became the perfect, at least in the East.
Yet the use of the perfect can be found in the Centre as well. KB, a text of
the Pañcālas, and closely related in ritual to the Baudhāyanas, shows the
perfect prominently, though a little less in degree (56%) when compared to
the typically Eastern books 1-5 of ŚB (85%). It is interesting to note,
however, that although KB contains more or less the same material as AB 1-5,
albeit in reformulated form (a situation similar to the relationship of
MS/KS:TS, PB:JB, etc.), it differs greatly in its use of the perfect from the
older parts of AB.117
If one therefore surmises an origin of the narrative perfect in the East at the
time of the Brāhmaṇa prose (level 4), then it is obvious that this usage had
spread to the Central area by the time of the late Brāhmaṇa (KB), and had
also heavily affected an originally Central, and subsquently Southern, text,
JB, which varies to a great extent in the use of the narrative tenses.
It must be noted here that JB is based on a mostly lost text, Śāṭyāyana Br.,
which was composed in the (Kuru-) Pañcāla area.118 Some indications of an
originally Central, rather than Southern, location of JB are:
-- It has both the traditional formulas about the contest119
of the gods and the Asuras (devāś cāsurāś ca saṃyattā
āsan / āsuḥ :: aspardhanta/ paspṛdhire).
-- It fluctuates in the use of the tenses of narration.120
-- It prefers (Central) ha (vai) to (Eastern) u hai vai; but
there are cases of u ha vai as well.
-- It shows the (early) Central (and late Eastern) genitive
fem. in -ai.
-- It has the (late) North-Western and
Eastern Central ḷ- for -ḍ- (RV of Śākalya's time,
117
Why is there a difference between TB, TĀ, and KB, altough they belong to the same
geographical area? Was the formulation of KB late, like VādhB, or are Taitt. texts
intentionally traditional, cf., the use of suvar instead of svar, etc., see Kuiper IIJ 30.1 and §
6.5; see the summary on this and related topics, below §10.2 .
118
See Festschr. Eggermont; cf. Caland, tr. PB, p. XVIII: "Perhaps the original
Śāṭyāyanaka... was taken over by the Jaiminīyas...."
119
See author in Festschrift U. Schneider, Freiburg 1987 and Fel. Vol.Eggermont.
120
Caland, in: Over en uit het JB, p. 20; similarly, Oldenberg on the Brāhmaṇas as a 7hole,
Prosa p.27: "bald werden bestimmter(e) oder unbestimmter(e) Motivierungen der
Tempuswahl sichtbar, bald verschwimmt alles." See already Whitney, TAPA 23, summary
46
AB, KS, KpS, JS, JB, VSK, ŚBK, see below § 6.3).
All of these features taken together make JB a very interesting text, both
from a literary as well as from a linguistic viewpoint, but this has not really
been noticed as yet. Indeed, there are many cases in JB where even stock
phrases like "the gods and the Asuras were in conflict / contested," are
related in the perfect tense; in other cases, the impf. has been retained. This
later overlay resulted in an almost irregular usage of the perfect/imperfect. 121
The figure for JB is: 23.41 % of perfetcs compared to impf., which places this
text, as expected for an originally Central, now Southern text, between ŚBMw
(14%), and the late KaṭhB (28%); note that a later text of the Jaim. school,
their Ar./Up., JUB, has already 36% of perf., while the Central KB has 56%
and the Ar./Up.of the Kauth., ChU, has an unprecendented 443% (which
makes it either very late or points to a composition in the (Central) Eastern
area; note the supposed movement of parts of the Kauth. towards Videha.)122
This text requires further discussion. It has long been known that the
original AB contained only pañcikās 1-5, and that the rest, AB 6-8, (note the
name 'pañcikā'!) is a later addition. These chapters deal with materials not
contained in the parallel text, KB, i.e., the rituals concerning the king, like the
"coronation" ( the unction ceremonies of the Rājasūya and Abhiṣeka) and the
duties of the royal priest (purohita). Also, the geographical horizon of AB 6-8
is much wider than that of AB 1-5, which clearly was composed in the West,
in the Kurukṣetra area.
121
Interchange impf.<> perf is studied, in some detail by Whitney, TAPA 23; cf. further,
author, Fs.W.Rau; AB 2.1 āyaṃs, abibhyur KB 6.15; JB 1.42 (Varuṇa speaks), see
Bodewitz, tr.JB 1.1-65, p.108, ann. 22 (Oertel).
122
But see ChU bhal- from smar which points to a Western/Central origin of this passage,
see above, ann.33.
123
Note that especially those chapters of KaṭhB that deal with the late Br. concept of a
second death, punarmṛtyu, are affected; see below, § 8.2 !
47
If one compares this with such details as the prominence of the Aśvala Hotṛ
at Janaka's court (in BĀU, ŚB), the emergence of the Āśvalāyana Sūtra in the
East (cf. the Pāli texts on Assalāyana), or the detailed knowledge in the later
AB of the udantya tribes (Śabara and other Muṇḍa peoples),124 it is evident
that this portion of the text was composed in Eastern India.
While the diffusion of the narrative perfect had reached the Western
territories of the Kaṭha school by the time of the late Brāhmaṇas, this
development largely excluded such Western texts as the comparatively late
PB (o.76 %, but KaṭhB > TBk 28.9 %) and did not reach Pāṇini's bhāṣā (in
the extreme North-West).
124
Note that these tribes are, for the first time, made part of an Indo-Aryan realm; in the
Śunaḥśepa story of AB 8.18, this is expressed in the guise of their origin as sons of the Ṛṣi
Viśvāmitra. Though they still are looked down upon, their inclusion into the power base of
the Magadha kingdom would agree with the politics of the future empire.
125
Pers. pronoun yuvām, etc. for Vedic yuvam, see Caland, introd. PB.XX; Wack.III.2 p.
463, Aufrecht, AB, p.428, Caland, Over en uit het JB, p.16 sqq. Further material, below
§6.6
48
style of their Brāhmaṇa, just as the Taittirīyas did in their Āraṇyaka. But
they did not or could not pay attention to such small details as the correct
length of vowels in pronouns (nom. avam / acc. avām, etc.) and let the
younger forms slip in.126
Lastly, it must be noted that although the spread of the narrative perfect has
reached the Kaṭha and Maitr. schools at the time of the composition of their
latest texts (parts of KaṭhB, MU, and quite surprisingly, ChU), this usage has
not made the 'jump' over the Bāhīka territory of the Panjab; Pāṇini still
teaches the older Vedic use of tenses (3.2.11o sqq.).127
early: (KS)129
Sah. MS 1 % TS 1 %
(level 3)
126
Such as dugdhe, tanūm etc., see Caland tr. PB. p. XIX sq. and cf. below § 6.6. Note that
Caland, again, does not regard the use of the impf. vs. perf. as a useful criterion for
establishing the priority of JB vs. PB, see tr. PB, p. XX; he was hindered, however, by the
usual handicap of Vedic scholars, namely their lack of knowledge on the geographical
localisation of the texts and an only limited view of the various levels of Vedic language, see
above ann. 12.
127
See above § 5.2; note that Oldenberg, Prosa, p.25, saw a progressive development in the
use of the perfect in narration from TS - AB 1-5 - (parts of) ŚB .
128
The usage of the narrative perfect goes together with that of the particle ha, see
Oldenberg, Prosa, ann. 105.
129
Unfortunately, countings for the Kaṭhas and Kāṇvas (but cf. Caland, ed. ŚBK, p. 70 sq.;
see below, ann. 107), Vādh., Baudh. etc. are still missing.
49
early ABo 3 % TB 2 %
Brāhm.
(lv.4) (TĀ 3 %)
later JB 23 %
Br., JUB 29 %
KB 56 % ŚBw 14 %
Up.s, ŚBKo 153 %? ŚBo 85 %
ŚBn 77 %
early VādhB mixed
Sūtras BŚS "
(lv.4) ŚŚS ŚBa/u 134%
***
It must also be mentioned that there is, in some later Vedic texts, some
confusion in the use of the perf./impf. This indicates that the
authors/redactors were not entirely sure any more about the use of both
tenses. This is most typical for JB, but it also occurs in texts like ŚBM and
ŚBK. Several theories have been advanced to explain the apparent confusion.
A detailed discussion can be found in the Festschr. U. Schneider.130
Whitney (TAPA 23, p.19 sqq.) stressed the fact that speeches relating
something about the past which are inserted into a story told in the perfect,
usually (though not without exeption) use the impf. (ŚB, JB, etc.). Caland
thought that JB exhibits a 'hierarchical' or 'mythological' imperfect which
was used in order to relate happenings in the mythical past, while those of a
more recent (pseudo-) historical past were be told in the perfect.(Caland,
Over en uit het JB, p.20.)
Oldenberg (Prosa, p.25 sq.) observed that the imperfect is used when the
speaker wants to recall a personal remembrance. In my opinion, most if not
all such cases can be subsumed under a catagory "pluperfect" meaning
(vorzeitig), as has first been observed by Caland: He noticed that ŚB (K, M)
shows signs of confusion in the usage of the aor. and the perfect, see intr. ed.
130
In the rest of this section, I again quote from the notes of my article contributed to the
Fs. U. Schneider, which deal with the frame story of the Cyavana legend in JB and ŚB.
50
ŚBK, p. 71 sqq., p.70 sqq., with this interesting observation on the functions of
tenses in ŚBK: original use in ŚB of the impf., found at the end of a tale told
in the perf., "in pluperfect meaning." However, this "still requires special
investigation."131
In my opinion, this comes close to the use of the impf. in JB; a mythological,
hierarchical (or historical) past is intended. It is summed up in impf.: "this
or that had happened at that time...." However, even this assumption does
not explain a l l the usages of the impf. in JB. We have to reckon with the
retention of an older usage of a narrative imperfect in parts of the text. The
older formulation made at the time of the composition of the Śāṭyāyana Br.,
which preceded JB, has survived frequently, so that in some stories there is a
gradual "slip" towards the perfect, and a sudden reversion to the imperfect in
other parts of the same story; see, for example, the Cyavana legend of JB.132
The later destiny of the various past tenses can be summed up briefly.133
The development is connected with the emergence of the post-Vedic Epic and
"classical" Skt. (other than Pāṇini's North-Eastern bhāṣā). It is known that
Epic / Class. Skt. did not directly develop from (a particular) Vedic dialect,
but that various regional features were intermingled.
In this case, the Eastern feature (typical is ŚBM, AB 6-8), with an opposition
aorist : perfect, is fused with the Central/Western one (TB, AB 1-5) with a
retention of the older opposition aorist : imperfect : perfect, and the Southern
one (JB) with a new opposition aor. : perf.134
Probably this development in later Vedic and in Epic/Class. Skt. is only the
effect of a complete restructuring of the tense system which took place at the
same time in popular speech, i.e., Prākṛt.135 The beginnings of this
restructuring are visible in the YV Saṃhitās (loss of the modi of the aor. and
131
See Whitney, TAPA 23 p.25-26, Minard I §168 b note 471a.
132
Cf. author in: Fs. U. Schneider, cf. also Whitney, TAPA 23, p. 22 with examples of such
"slips" in ŚB 12.9.3.7 sqq.; - cf. also the conditional in Pāli; and some missing augments in
AB, KB, see Keith, RV Br. p. 74; on the other hand, some unwarranted augments are
inserted in other texts, cf. C. Caillat, Fs. Schneider, cf. ann. 290; cf. further, van Daalen,
Valmīki's Sanskrit, p. 83 sq.; S.K. Bharadwaj, Ling. Stud. in the DhS, Rohtak 1982, p. 122;
Satyavrat, The Rāmāyaṇa, a linguistic study, Delhi 1964, p. 223 sq.; E.D. Kulkarni, ABORI
24, p. 83-97.
133
Cf. the notes in my article in the Festschr. U. Schneider.
134
JB, however, has many instances of the impf. for ancient legends or for the summing up
of an ancient happening/distant personal remembrance.
135
For this purpose, the grammar of the so-called Yajñagāthās should be scrutinised. How
far do they differ from the later Epic and from Pāṇini?
51
development of the precative); the aor. as one of the past tenses develops, in
Saṃhitā Prose, a periphrastical aor. (see K.Hoffmann, Aufs. 469, ann. 6),
which then disappears in the Brāhmaṇas.
At this moment, actual restructuring of the use of the past tenses sets in (see
above). A distinction is made between the value of the augmented forms
(impf., aor., conditional) in "pluperfect meaning" (vorzeitig) and the
unaugmented forms (perf., pres., future, subj., opt., imp.).136
The effects of this development are to be seen clearly by the time of early
Middle Indian. Subsequently, the perf. is found only in a few remnant forms
of Pāli; it has almost disappeared in Middle Indian. The impf. is extinct,
except for a few remnants in Pāli which have been classified with the
aorists.137 In a situation where both the perfect and impf. tenses disappear, it
is not surprising that the aor.138 has survived in Middle Indian, i.e., in Pāli,
(more rarely in A.-Mg. and in a few cases in J.-Māh.)139
In late Vedic, the aorist had retained its function, i.e., relating something
that has happened immediately before the present. Apparently it also relates
(the effect of) a recent happening leading u p to the present; see, for example,
ŚB 1.4.1.8-19: aśakata or ŚB 1.4.10-18, ŚBK 2.3.4.8-15: the conversation
between Videgha and Gotama, which recapitulates their journey in the aorist.
Is this an indication that, in the East, personal experiences, whethe2 recent or
not, could be told in the aorist? (Cf. also Weller, Śunaḥśepa, Ber.d. Sachs.
Akad., Bd. 102.3, p. 72). Contrast this with the opposite situation in some
Vedic texts and in Pāli (pers. experience in impf., past events in aor.),
according to Oldenberg, Prosa, p.25. Perhaps this, too, was a regional
feature at the time of the late Brāhmaṇas. This, however, can only be decided
after careful study of similar occurrence.
136
Note that there is no functional distinction between augmented and unaugmented forms
in Pāli, but that this is a remnant of older forms only, regulated acc. to the length of the
form and its origin in one of the aor. types; but cf.now C. Caillat, in Fs. U. Schneider, see
above ann. 132.
137
āsīt > āsī, see O.v. Hinüber, Überblick §479; cf. also Oldenberg, Prosa, p. 25 ann. 2, who
compares a similar distribution for Pāli: avoca (relating an event of the past) / avaca
(personal remembrance).
138
Which even in late Vedic is mainly used in a preterite function (i.e., to tell recent events),
and thus separate from impf/perf.
139
O.v. Hinüber, Überblick, p.192 § 477-488, esp. § 478.
52
The results of the last section can now be counter-checked with the evidence
provided by some of the formulaic sentences so frequently found in the
Saṃhitās and Brāhmaṇas, notably, the standard phrase: "The gods and the
Asuras were in contest...."
This means that materials are discussed now which could be understood as a
mere matter of style. It will be seen, however, that even in such traditional
formulas, the gradual development of the Vedic language asserts itself. These
sentences provide useful indications as to the tenacity of the hieratic language
of the Brahmins on one hand, and the general changes in Old Indo-Aryan and
the underlying Middle Indo-Aryan (Prākṛt) on the other.
The use of the perfect tense in narrative passages during the late Brāhmaṇa
period affected even the traditional sentence, "The gods and the Asuras were
in contest," which begins so many tales. It is found in the following versions:
devāś cāsurāś ca aspṛdhanta/saṃyattā āsan:: ... paspṛdhire, saṃyattā
āsuḥ140
This phrase occurs in the oldest Saṃhitā texts (MS, KS, TS), and there it is
invariably told in the imperfect, which is normal for narrative passages in
these early texts. However, there is a slight difference in style; MS uses the
old Indo-European word spṛdh,141 while KS and TS use sam.yat, lit. "to form
a line (as in battle)."142
140
Note that other school particularities of style are retained as far as word usage is
concerned, see below, §§ 7-8.
141
Engl. sport, German (sich) spurten, cf. in Iranian, GAv. spəәrəәd Y 53.4
142
Note the case in the Saṃh., where the gods search for the hidden Agni. They take each
others' hands, form a 'police line'; see P.Thieme, Mon. Nyberg, 431 sqq.: "to take a firm
position".
53
It is interesting to note that the late Aitareyins (AB 6-8) have retained their
school particularity of style, the idiom sam.yat, and have not followed the
Eastern style of using spṛdh. On the other hand, however, they have
conformed with later linguistic development, namely the characteristic use of
the narrative perfect, and have changed the tense of the sentence accordingly.
In order to underline the identity of their school, retention of such special
features as the use of sam.yat was necessary, but the change to perfect was
probably involuntary and automatic. In the present context, it is important to
note that standard phrases were used time and again to begin telling newly
reformulated myths or newly invented stories about the gods; these stories,
though following the old pattern, were constantly changed to suit the
discussion intended, or to fit the demonstration of a particular point of ritual.
The remainder of the texts agree with this scheme in a way parallel to the
general use of tenses, described in the last section. In the following table, the
typical phrases are given:143
__________________________________________________________________
----------------------------------------------------------------------------------------------
The absolute number of cases of sam.yat vs. spṛdh is given below before the
abbreviation of the text, followed by the number of cases of impf. vs. perf.
__________________________________________________________________
s.y/sp. i./p.
__________________________________________________________________
___
early: 16/4 KS 20/0 15/4 TS 19/0
Saṃh. 5/20 MS 25/0
ChU 3/10
KaṭhB 6/7 2/1 ABn 1/2
(= TBk)
0/7 PB 7/0
1/3 ṢB 4/0
MU 3/ 0146
144
The figures of VādhB do not refer to the absolute number of cases but to the number of
stories/sections; of these 14 stories using the perfect, 3 are dealing with the gods; 51 stories
use the impf., 5 have mixed tenses. The relation of sam.yat/spṛdh is not counted here.
145
In BŚS 18.22-49, where many stories are found (3 means number of stories with perf.).
55
__________________________________________________________________
While the two preceding investigations, that into the gen. fem. in -ai and the
one into the diffusion of the narrative perfect, were confined to the post-
Ṛgvedic texts, the following sections allow one to trace the development of
certain particular features back to the Ṛgveda, as well as down to the latest
levels of Vedic language and texts.
The infinitive in -toḥ is found 11 times in the RV, while in the next level of
texts, the Mantra language of the YV Saṃhitās, it is found 16 times (in both
KS and MS, though both texts differ only marginally in size from the RV.)
With the RV text calculated = 100 % and the MS text at 88% of the RV text,
the 16 cases of MS come to 101 % in comparison to RV's 11 = 100%. The use
of -tos has thus decreased by more than half at the time of ŚS (40%), but this
seems to be conditioned by special factors, perhaps of location, as PS has
108%. This infinitive then gains prominence again in the older YV Saṃhitā of
the Kaṭhas (171%) and increase further with the Taitirīyas.
__________________________________________________________________
_____
RV 11x147 PS 7? ŚS 2?
100 % 108% 40%
KS 16
171%
MS 16 TS 19: 141%
101% TB 11: 138%
TA 2: ŚBKo 7 ŚBMo 13: 180%
142%
AB 20: 289% KB 5: 160%
(ABo 7:274%) KA 1: VādhB 2 ŚBMw 3: 58%
(+īśvara)
ŚBMa/u 0
JUB 2: 3% ŚB total 27:160%148
(ABn 9: 697%!)
(+ īśvara)
__________________________________________________________________
However, according, to Oertel, KZ 65, 1938, p.55 sqq., 66: īśvara- + inf. in -
tos is found as follows:
__________________________________________________________________
KS 16
KpS 5 TS 19 ŚBK 13
TB 11
MS 16 TA 2
AB 16 ŚBMo 18149
AA 4 KA 1
JB 30
148
According to Renou, Monogr. Skts. II, p.37 §37;
In ŚS there are apparently only two cases, janitoḥ 19.56.2, and aitoḥ 12.3.55-60;see Renou,
Mon.Skt. II §28; the number of occurrences in PS will probably increase, as the Sandhi is
not always separated in my provisonal, computer-based dict. ab ultimo. For these
infinitives, see esp.: Minard, Enigmes II, § 727b, acc. to Renou, Monogr. Skts. II (1937),
§39: -tos becomes less frequent in this order: MS--> KS (KpS) TS TB AB JB <!> ŚB PB
KB GB; -tavai: less ŚB --> PB; -tum increases from: MS---> TS TB KS JB AB KB PB
...(big interval)... ŚB; cf. Wack.2.2 § 470 sqq: -toḥ: Br.Ār, but rare in Sūtras, -tave: rare in
Br., tavai < tave vai?, -tum is classical (partly -tavai: MIA); McDonell, Gramm. §582, in
RV -tave: 30x, -tavai 12, -tum 5, -tos 6+3; but acc. to Avery, Verb forms, RV 12x; -- for the
Mantra language, see McDonell, §585.4: -tave RV, AV, TS, VS, -tavai RV, AV; Delbr.,
Syntax, p.427: -tavai MS, TS, AB, ŚB, -tave ŚB 2x; -tum MS, TS, AB, ŚB, PB, -am MS, TS,
AB, ŚB, PB; -tos MS, TS, AB, ŚB, PB; - Aufrecht, AB p.430: -tos 15x, -tavai 2; but acc.to
Avery: AB 0.
149
Brunnhofer, Ueber das gegenseitige Verhältnis der beiden kâṇḍagruppen des
Çatapatha-brâhmaṇa nach massgabe der in ihnen verwendeten infinitivformen,
Beitr.z.Kd.d.idg.Sprachen X, p. 252, counts 10 cases for ŚBM 1-5. His conclusion on the
divison of the text is correct; he wants to join ŚBM 1-5 + 11-13, as opposed to the Śānḍilya
part 6-10. Cf. Weber, Ind. Stud. XIII, 266-268, Vorles. über ind.lit., 2nd ed., p. 146 sq.;
Caland, introd. ŚBK, thinks that ŚBM 11-13 is a separate work, originally stemming from
the Kāṇvas.
57
JUB 2
PB 7 (ABn 9)
__________________________________________________________________
While the early books of the Mādhy.ŚB strongly participate in the usage
(180%),151 the Kāṇvas, as usual, make an exception (142%) and tend to agree
with the Western Śāṇḍilya school (ŚBMw 58%, cf. the enigbouring JB, based
on a Central Śāṭy.Br., with only 92 %, also: MS 101 %, ŚS 40%).
Later lexts like TĀ 2x, KĀ 1x, drastically reduce the usage of the inf. in -toḥ.
But the Ait. school continues to use this inf. Just as the late AB (-toḥ + īsvara)
scores 679%, so does AA (-toḥ) with 483%. This seems to be a phenomenon
150
Perhaps this indicates the late time of composition of this Br., cf. the similar figures for
JUB, the late ŚBMn and especially ABn..
151
A text of later redaction, PB, also belongs here, with 194 %.The non-occurrence of -toḥ
in the Ar. and Up.of ŚBM is unexpected.
58
limited to this school which began in the earlier, Western part of AB with
nearly 300 % and continues in the Eastern, younger texts.
The diffusion of -toḥ is thus limited to the Saṃhita and Brāhmaṇa period,
with a preponderance of cases in the Kuru, Pañcāla and Videha area. The
Southern Kuru (MS), the Eastern Central (Kosala) and the Southern areas do
not participate that strongly.
Excursus:
--------
The distribution of -toḥ in the various parts of the RV is of interest; the
forms are given according to McDonell, Vedic Gr., § 587.152 In order to
compare this with another significant factor for dating the RV books, the
frequency of injunctives is given below, in decreasing order, as the usage
disappears; only some 5o new cases surface by the time of the AV.153
____________________________________________________________
RV book: 1 2 3 4 5 6 7 8 9 10
____________________________________________________________
-toḥ: 3x 3 3 - - 1 1 - - 1
____________________________________________________________
Or, arranged according to the probable age of the various books of RV,
according to the investigation by W. Wüst, and measured by the frequency of
injunctives (decreasing order):
____________________________________________________________
acc.to Wüst154: 9 4 || 3,5,7, || 2, 6 || 8 10
(older >younger)
------------------------------------------------------------------------------------------
inj. in books: 4 6 2 7/1 5 || 10 3 9 8
152
Among these, 2 x īśe + dat. inf.: īśe..dātoḥ, 7.4.6, īśe ...yotoḥ 6.18.11.
153
See K. Hoffmann, Injunktiv, p. 110.
154
Stilgeschichte und Chronologie des Ṛgveda; for the inj., see K.Hoffmann, Inj. p.36,
which is based on the count made by Avery, JAOS XI, p.361; cf. also Arnold, Vedic Metre
and Klein, Towards a Discourse Grammar of the Rigveda.
59
§6.0. The preceding sections have indicated, I hope, that some peculiarities of
dialect development can be traced throughout the Vedic period, right down
from the RV, which itself is, of course, a collection of hymns by various
authors and clans and reflects the traditions of many tribes across several
centuries. The above materials also tend to indicate that there were centres of
innovation, i.e. the areas occupied by the Taitirīyas (TS) and Vājasaneyins
(ŚB).
I will now turn to phonetic peculiarities of some major texts which have so
far largely been disregarded. They have, it is true, been noticed for more
than a hundred years and figure as such in the standard accounts of the
language. However, they have not been placed in their proper context and
perspective, as they were regarded to be phenomena limited to the text
transmisson of a particular Vedic school, or even to a particular text.
the last century, record some of the phenomena, they have tended to establish
a uniform Sanskrit text, recognisably Vedic, but close in its "orthography" to
the MSS of classical Sanskrit of the standard Northern tradition of Benares
and Poona.
This is misleading. Nearly all Vedic texts have, because of the whims of the
editors, been normalised to such an extent that it became unlikely that other
scholars paid attention to or even recognised many of the particular traits of a
particular Vedic school. In the same manner, of course, the traits which
agree in several schools were often not detected. What is almost universally
accepted, however, is the misconception of Vedic as a uniform language. This
is, in itself, not surprising; the texts a p p e a r to have no phonetic
differences (see above). What is usually not taken into account, however, is
that the texts were only m a d e to look uniform by the late Vedic diaskeuasis
and canonisation of the redactors which often comes close to the norms of
classical Skt., as first codified in Pāṇini's grammar. In s0ite of this
normalising tendency, made worse by that of the modern editors, a
considerable number of school divergencies can be recognised, some of which
will be treated in the sequel. It will be seen that many of these śākhā
peculiarities do, in fact, form a pattern, and are part and parcel of the major
Vedic dialects.
One such item is the spelling of the long palatal affricate. In classical
Sanskrit, it is written and printed cch, and this is also the pronunciation (in
Vedic recitation) and, consequently, the spelling found in most schools.
However, as is well known, the Ṛgvedic MSS usually write ch, e.g., gachati
instead of classical gacchati, see Aufrecht, preface to RV, 2nd ed., p.VI.
This is not all. The Maitrāyaṇi school participates in this trait. Schroeder
mentioned the spelling in the preface to his edition, p. XLIII, but normalised
to some extent (e.g., in the Sandhi case -t + ch- > cch). The writing ch is also
found in the Vulgate of the Atharvaveda,156 usually called the Śaunaka-
Saṃhitā, see Lanman in: Whitney, AV transl., p. CXXV.157
156
Also in cases like -t ś- > cch, in ŚS written -ch-; cf. also Allan, Sandhi, p. 92.
157
Finally, Ms. C of JB (see ed. L.Chandra JB II, p. XVI) usually writes gachati, etc. This is
a ca. 300 years old palm leaf Ms. from Kerala (Burnell No. 421 = Keith Cat. no.4353).
61
The origin of the cluster ch is known from such forms as gachati:158 Ṛgvedic
metre indicates that -ch- was measured long, and was therefore pronounced
as a cluster, probably something like [śś / śc]159. Interestingly, this is almost
the same spelling that the close relative of the Maitrāyaṇīya, that is the Kaṭha
school, employs in its MSS: śch. Schroeder (introd. to KS) misunderstood
the evidence; he thought it to be a writing mistake as in the original
Kashmirian MSS of KS, written in Śāradā script; both -śch- and -cch- look
quite similar. However, the writing -śch- is consistent in Kashmirian texts.
It is found also outside KS, in KaṭhB, KaṭhĀ, etc., and notably so in the
Kashmirian MS of the Paippalāda Saṃhitā; it regularly occurs even in MSS
of classical Skt. texts that stem from Kashmir. The written evidence160
reflects a much older stage of pronunciation, that of medieval Veda
recitation.161 The pronunciation of -(c)ch- as [śch] is not the present day
pronunciation of this cluster, nor was it the medieval Kashmirian one; c is
pronounced as [ts] and śch as [tsh].162
The Śākalya Śākhā of the Ṛgveda shows a close connection with the Aitareya
school, the older homeland of which was the Eastern Panjab (see §4.1). The
Kaṭha school occupied roughly the same territory as the Aitareyins; PS is
equally a Western text, when compared to the Śaunaka version of the
Atharvaveda, and the Maitrāyaṇīyas settled in the same area, though
apparently somewhat to the South of Kurukṣetra.
This limits the occurrence of this phenomenon to the West of the area of
Middle Vedic texts and schools. In fact, it is a Kuru peculiarity, since all the
158
*gm-sk’e-ti, probably via > gae-ścæ-ti.
159
This cannot be the same pronunciation as in háriścandra- < hariś candraḥ, RV 9.66.26.
160
Cf. also Wack. I,153 sqq. Nachtr. ad 158,28, ad 154,13.
161
Cf. O.v. Hinüber, Überblick. on c/śc, §192-3; yc, yj in Mg.
162
Note also that the medieval Nepalese MSS hardly exhibit the writing ch, but always write
cch as a ligature, even in cases where a word begins in ch- (and is not preceded by vowel).
62
texts mentioned are located in the area of the Kuru tribe and not in that of the
Pañcālas or that of other, more Eastern tribes, like the Kosalas.
§6.2 kś :: khy
The distribution of the pronunciation [kś] for the etymologically correct and
otherwise universally accepted [khy] is a Kuru peculiarity as well;163 it is also
an innovative one that represents a change from a cluster with palatal
semivowel to one with palatal fricative. The change has affected only one
group of texts: the Yajurvedic texts of the Kurus. Among the printed texts,
only the Kaṭha and Maitrāyaṇīya schools participate. The Carakas, however,
are said to have shared this trait as well. This lost Yajurveda śākhā is very
close to the texts of the Kaṭha school, though it is representrative of a
separate, and apparently older, stage of development of the Yajurveda.
Unfortunately, the texts of this school have been lost, except for some
quotations in various Vedic and post-Vedic texts.164 The innovation has not
spread beyond the area of these Yajurveda texts belonging to the Kuru tribe.
Even the Pañcāla texts (TS, KB, etc.) do not participate.165 It is also
remarkable that the innovation is limited stratigraphically to just one type of
text, in this case, the YV. The local RV, AV, and SV texts (AB, PS, PB) do not
participate. However, a similar phenomenon was observed in the case of the
attestation of the inf. -toḥ in the RV-Br., see above §5.4. There as well, the
RV-Br. do not completely share the local developments of the Yajurvedins.
AB participates to slightly more than half of the percentage of the Kaṭha
school living in the same area, but KB has a minimal amount, ca. 10% of the
cases when compared to the Taitt. school of the same area. This may be
attributed to the later stage of the texts, when compared to the YV Saṃhitās,
but note that a Western SV-Br., the late PB, still has 20%. As these
developments run counter to the usual areal spread of a dialect phenomenon,
it will be instructive to pay attention to this in the following cases.166
163
KS, KpS, MS, Carakas, see StII 8/9 p. 209 and Schroeder, ed. MS I p. XLIII, Wack., Ai.
Gr. I, 209.20 and additions p.116; for KpS, see J.Narten, Sprache 14, p.122: aor. akśat; cf.
also Ved. Var.II § 190.
164
For the geographic position of the Carakas, see IIJ 26. For the school in general, see StII
7 and 8/9.
165
Is it also not known to Pāṇini, cf. 3.2.7, 3.1.52, 8.1.57, 2.4.54
166
Cf. the summary in § 10.2.
63
While this may look like the spread of a particular style of speech, the origin
of these peculiarities is still local, and the spread is centrifugal. The "force" of
the change, and therefore the area of diffusion, differs from case to case, as is
well known in dialect studies. In the present case, the innovation could not
spread beyond the Kuru area, nor did it affect texts of the next, i.e., the
Brāhmaṇa, level. That means that we must recognise this peculiratity as an
isolated feature of the early Saṃhitā prose speech of the Kurus.168
The preceding sections have dealt with old, regional features of the Kuru
area which developed in the time of Saṃhitā prose or even earlier. On the
other hand, the so-called Rgvedic -ḷ- is generally regarded as a late feature;
this can be exemplified briefly, as is usually done, by referring to the two
forms īḷe and īḍya- in Śākalya's Padapāṭha text: -ḷ- occurs only in
intervocalic position, thus not in cases where the Middle/Late Vedic
pronunciation of the group -ḍiy- had already become -ḍy- (cf.below: § 6.5
súvar > svàr, śreṣṭha- < *śraiiṣṭha-, etc.). It is only at this late stage that the
change ḍ > ḷ took place. The distribution of this feature is not limited to the
167
Similar developments can be noticed in the use of RV mantras as well. Again, the
various śākhās of a Veda tend to band together against the texts of other Vedas. See e.g.,
PS 7.4 = MS 2.10.4, KS 18.5, TS 4.6.4, cf. RV 10.103, SV 2.1219, or PS 9.5 ~ ŚS 19.6, RV
10.90, etc. (with parallels). This is important for an understanding of the early activities of
the Kuru Brahmins. They composed and varied hymns in the Kurukṣetra area, and this
was further diversified once the schools of the more distant lands emerged.
168
Note also the words restricted to MS and KS, i.e., cases where even the other old YV
Saṃhitā, TS, does not participate, as it is the product of the Pañcāla area; see the list of
Schroeder, ZDMG 33, p. 189 sqq. This must be reinvestigated, yet it still contains many
such words, like veśatvá and others, which outside of MS/Kps/KS are only recorded by
Pāṇini.
64
Ṛgveda, nor to its Brāhmaṇa texts (where it is found in AB, AĀ, and also in
KB). It also occurs in a number of texts where this usually is not realised at
all.
First of all, in the Kaṭha school, in KS, KaṭhB, KaṭhĀ. The manuscripts of
these texts (especially the Śāradā MSS) apply a special symbol, a small
diacritic triangular mark, which is attached to the akṣara for -ḍ-. The editor,
L.v.Schroeder, has neglected this and has thereby misrepresented the
phonetic status of this school to this very day.169
The Kapiṣṭhala sub-school of the Kaṭhas has three cases of ḷ-/ḷh- in its single
continous MS; see ed. Raghu Vira, introd., repr. p. VII, Oertel, SB
München,1934, p.17: samūḷham 2.4, mṛḷayanta 3.8, dūḷabho 5.2 (in Caland's
MS, Utrecht Univ. Libr.?).
Paippalāda school: both the Kashmiri PS (with a diacritic) and the Oriya PS
with a special letter used for intervocalic ḷ- (now found in Maraṭhī, Oriya etc.)
exhibit the retroflex ḷ- instead of the usually printed -ḍ-.
The Jaiminīya school: JS, JB, JUB, sporadically also JGS, JŚS equally show
the retroflex ḷ- instead of -ḍ-.170
The Kāṇva school: VSK, ŚBK. Here the evidence is not so clear, as the MSS
often writes this sound as normal dental -l-; cf. however, Caland on the
occurrence of ḷ- even in MSS of BĀU, (ed. ŚBK, p.467).171
The development from retroflex ḷ- to dental -l- is not unknown, cf. Lüder's
treatment of this sound change, Phil. Ind., p. 546 sqq., (and cf. Balkan Gipsy
phrāl <*bhrāδā- < *bhrādā < bhrātā)172
A development of the Saṃhitā and early Brāhmaṇa period, the change from
-ḍ- > ḷ- first occurred in the Panjab (KS, AB 1-5) and subsequently spread
eastwards into the Central area (only KB!), perhaps also to the Central
*Śāṭyāyana Br. of SV which developed into the Southern JB; the new sound
also made the usual "Kāṇva jump" to Kosala. Note that the Central Taitt.,
169
Also misrepresented by Scheftelowitz, ed. RVKh, p.47; cf. however, WZKS 23, p. 16;
and author, Das Kaṭha Āraṇyaka, diss. Erlangen 1972, partial print Erlangen/Kathmandu
1974.
170
See Burnell, Jaim. Arṣ. Br. p. IX; Caland, JS 33, cf. W. Rau, MSS 42, p.187 sqq.
171
See also, Renou, JA 1948, p. 34, Lüders, Phil. Ind. p. 550 sq.
172
See further, O.v. Hinüber, Überblick, p. §198 sqq.
65
the SW Maitr., and the Kauth. SV (PB) are not affected, nor are the Eastern
texts like VSM, ŚBM, and the (Central/E.) AV (Śaun.) reached. 173
__________________________________________________________________
Kaṭha Paippalādin
Kap.? Aitareyin Kauṣītaki Kāṇva Eastern Ait.
Śākala RV? Śākala RV
Jaiminīya
__________________________________________________________________
******
173
It is remarkable that the Taittirīyas have not taken over the Kaṭha -ḷ- in their loans
from the Kaṭha school, TB 3.10-12 and TA 1-2. On the other hand, they have not
introduced their typical form suvar in these texts. Is this perhaps evidence enough to
assume that the diffusion of -ḷ- is later than the introduction of the Kāṭhaka-Cayanas into
the Tait. School? Note that even an old text like the BŚS has these special cayanas (BŚS 19,
TB 3.10-12, TA 1). On the other hand, TA is so late that it has a number of traits otherwise
known only from the Purāṇas, cf. the name of the Veda compiler Vaiśampāyana, a
Vātsyāyana, etc.see MSS 30, p. 180 ann.13. Does this mean that these Cayanas are very
late? Or is it simply that their formulation is late? In that case, their introduction into
BŚS, ĀpŚS, etc., must also be very late, an interpolation in fact. This is unlikely. Have they
been introduced into BŚS later than into TB/TA? But what about the style of BŚS in these
cayanas? BŚS copies TA I more or less word for word; the text seems to be as old as BŚS in
its redacted form.
66
The remarkable change of -jm- > -ym-, which occurs in a number of Vedic
texts, is little known, and, if so, thought to be limited to the Kapiṣṭhalas, a
sub-school of the Kaṭhas which has come down to us only in a very
fragmentary state.174
The peculiarity of a change from -jm- to -ym- (ajman > ayman, yunajmi >
yunaymi, etc.) is found only in a sub-school of the Kaṭhas, the Kapiṣṭhala-
Kaṭhas (KpS); the so-called Caraka-Kaṭhas (KS) do not exhibit this trait.175
The homeland of the Kap. school seems to be close to that of the Kaṭhas (in E.
Panjab); in the 3rd cent., BC., the Kapiṣṭhalas (Kambistoloi) were found,
according to Arrian, Indikë, in Panjab, at the confluence of the Panjab rivers
with the Ravi.
The change -jm- >-ym- must be comparatively old. It has already been
taken over into the first book of the PB, which contains a small
Mantrasaṃhitā of this SV school.176 Almost all of these Mantras have been
taken over from KpS, which indicates that both schools were in close
proximity during the period that PB was redacted.177 This relationship was
also known some early grammarians like Candra, who uses the compound
Kaṭha-Kauthumāḥ, 178 just as Pāṇini used Kaṭha and Caraka in one rule. We
therefore have to look for the origin of the trait during a few centuries before
or after C.E. and in an area not too far from W.Panjab/Rajasthan.
However, a few cases of this rather strange development occur in other texts
as well. There are traces in Maitrāyaṇī Saṃhitā (cf. author, StII 8/9, 175) and
MS 1.3.14 and MŚS 2.4.4.9 (corrected by van Gelder to pṛṇacmi, but var.
have: -pm-, -ṣm-).179 Furthermore, even the Vulgate of the AV (usually
known as the "Śaunaka") version, i.e., in the form of the text found all over
174
Only parts of its Saṃhitā, virtually identical with KS, and a very short fragment of some
late parts of its Brāhmaṇa, have been found so far. These fragments (see ed. Raghu Vira,
and cf. Schroeder, ed. MS p. XXXVII sqq.) correspond to the Kāṭhaka portion of TB (3.12)
which has been taken over from KathB.
175
See introd., ed. Raghu Vira, p.V; Oertel, on KpS in ŚB München 1934, p.29, Schroeder,
ed. MS I p. XVIII.
176
See Caland, transl. PB XXIV; Parpola, LŚS transl.p.77 sqq., esp. p.88 (PB 1 < RV,AV,
YV, esp. from Kaṭha school). LŚS 2.12.12 also has -ym- in a hapax mantra. - Parpola, LŚS
tr. p. 88 , points out that the Mantra chapter of PB has been added later on, but before the
composition of LŚS. (In the context of a possible redaction of PB in the East see ann. 97,
233, 334 , this would mean an intermediate stage, in the West, unless the chapter came from
the Prācya-Kaṭhas.)
177
Cf. however, ann. 125 on the late Vedic date of the text; it contains some classical forms
like tanūm, dugdhe, etc.; cf. also ann. 250, 125.
178
See Parpola, transl. LŚS/DŚS p.88.
179
For pṛnaymi MS 1.3.14, see StII 8/9 p. 175; cf. Oertel, SB Akad.München 1934
67
N. India., and also its Sūtra, the Kauśika Sūtra, have a few of these cases.
KauśS 42.17 ajmaḥ > aymaḥ, 64.17 anajmi > anaymi,180. Both texts,
interestingly, can be traced back to a Gujarat tradition of the Middle Ages. -
Occasionally, -ym- , is also found in the Or. version of PS (18.76.4:
maymani).181 Finally, another addition can be made from Nepal, where I
once saw this trait in an Agnihotra Paddhati belonging to the Vājasaneyi
school.182
How does this all add up? The few cases in MS, MŚS, PS, ŚS and KauśS can
all be traced back to medieval Gujarat. PB and the other Kauthuma texts are
also prominently found in Gujarat. It is unlikely, however, that the
substitution of -jm- by -ym- in PB I had much of an an impact on the other
schools (Śaun., Paipp., Maitr.). Even if one evokes the Moḍha Brahmins of
Gujarat who were cāturvedins (and of whom I have indeed seen SV, MS, and
AV MSS), this cannot readily explain the diffusion of the feature to all the
texts mentioned above.
The medieval homeland of the Kapiṣṭhalas is still unknown, but may have
been Gujarat. The school has not thus far been traced anywhere in India; I
suspect that its medieval home lies in Gujarat/Maharashtra (or
Orissa/Andhra), as the only extensive MS of this text shows a peculiarity
common to these areas, i.e., pronouncing and therefore also writing the sound
-ṛ- as [ru]. The MS is written in Devanāgarī; Gujarat or Maharashtra thus
would be the preferred choice.183 Then there is a similarity in Sandhi between
MS and KpS, i.e., -as/-e before accented vowel > - ā in MS, occasionally only
in KpS. This indicates a Maitr. influence on KpS transmission, either in
recitation or, more probably, in writing. This again supports a Gujarat
homeland for the medieval KpS. Modern evidence tends to confirm this.184
If, at a certain time, KpS/KpB was one of the major Yajurveda traditions in
180
Cf. also y/j change in: 120.1 samajyāyan > samayyāyam, 133.6 yajñe > jajñe; see
Bloomfield, p. LXI.
181
The reason is that PS originally stems from Gujarat, at a time of 800/1000 AD, cf.
author, ZDMG, VI. Suppl.band, 1985, p. 265 sqq.
182
See StII 8/9 p.209, but note Brāhmaṇo Gurjaradeśād āgataḥ in a colophon of another
MS from Nepal, 13th cent., see author, in: Formen kulturellen Wandels... = Nepalica 2., ed.
B.Kölver, St. Augustin 1986, p. 1987, ann.2.
183
The only other manuscript of KpS , from the former Ft. William at Calcutta, is also
written in Nāgarī, as is the manuscript, which I unfortunately was not allowed to film at the
Benares Skt.Univ. during three visits in 1972/73, of a so-called Kāpiṣṭhala Gṛhyasūtra.
184
See the letter no. 14 by Kanhaiyālāl Bhāīśaṅkar Dave of Pātāṇ (N. Gujarat), in Dr.
Yaśavant Khuśāl Deśpāṇḍe, Vedaśākhā Vāṅmay āṇi Carak Brāhmaṇyācāṃ Itihās, Nāgpur
1961, app.2, p.26. Someone should investigate the Kap. and other (Yajur-) Vedic traditions
of Gujarat (Dave mentions: Kaṭha, Kapiṣthala, Maitr., Caraka).
68
Gujarat, then the occasional intrusion of a peculiar trait of KpS/PB I into the
other Gujarati traditions (MS, PS, ŚS) would not be too surprising.
Even the singular Nepalese case can perhaps be explained in the same way.
There existed, just as with Kashmir, a connection in trade and religious
affairs between Gujarat and Nepal. A 13th-century MS from Nepal mentions
in its colophona a brāhmaṇo Gurjaradeśād āgataḥ; the Jainas, too, seem to
have had some relation with Nepal.185 It may be that some such Brahmin
brought with him the Paddhati concerned; Vāj. texts have indeed existed in
Gujarat since at least the 6th cent. AD.186
To sum up: the early development of this trait is noticeable with the
Kapiṣṭhalas; the pronunciation spread, well before the end of the Vedic
period, to the neighbouring Kauthuma school and its late Vedic Brāhmaṇa
text, PB.187
It is only in medieval Gujarat (the probable habitat of the Kap. school) that
the substitutional cluster -ym- sporadically influenced other texts as well.
Note that the occurrences of -ym- in these texts (PS, ŚS, MS, MŚS, KauśS) are
very sporadic. That the other texts were indeed influenced is explainable by
the particular state of affairs in Gujarat with regard to the transmission of
the Veda by, among some 100 other Vaidika and non-Vaidik Brahmin
groups, the Moḍhas who are Cāturvedins.188
Excursus:
------------
185
Cf. also E.Bender, "The Nepal Connection," forthc. (Lecture at the Int. Conference-
Seminar of Nepalese Studies, organised by S. Lienhard at Stockholm, June 1987)
186
See the copper plate inscr., cf. author, Beitr.z. Südasienforsch. 104.
187
Cf. also the similar development of intervocalic -j- > -y- in the Caraka school, acc. to VS
Prāt., 4.163 sqq., cf. StII 8/9, p.209.
188
See StII, and Beitr.z. S. As. forsch. 104.
189
For the whole question, see Wack.I §188, Ved. Var. II §192.
69
190
Renou, JA 1948, p.38: tanakmi VSK yunagmi VSK, tanacmi VSM yunajmi VSM; he
regards this as an imitation of RV forms in -km- etc., like vívakmi; cf. also Caland, ed.
ŚBK, p. 37: ŚBK jy: ŚBM gy in avanej/gyam.
191
See O.V. Hinüber, Überblick, §251; in inscriptions, j > y since the 2nd cent. B.C., see
§174.
192
See StII 8/9, p. 209.
193
Die Paiśācī und die Entstehung der sakischen Orthographie, in: Studien zum Jainismus
und Buddhismus, Gedenkschr. L.Alsdorf, Wiesbaden 1981, p. 121-127; cf. Überblick, §51,
98 sqq.
194
See O.v. Hinüber, Überblick, §174.
195
Preserved in the National Archives of Nepal, cf. Albīrūnī's note on the first Veda texts
written down shortly before his time (1030 A.D.) in Kashmir. Only the Upaniṣads
(Śaṅkara, etc.), are an exception, as they had been taken out of the Vedic corpus and
transmitted as texts of the Advaitins.
196
Note: Veyālīya, Vetālīya, Vetāliya, etc., in Jaina texts used as the name of a section in the
2nd Aṅga. Varāhamihira uses Vaitālīya as a synonym of Māgadhī (see Weber, Ind.Stud. 8,
70
§6.5 Ṛgvedic and Taittirīya súvar and later Vedic svàr; anaptyxis.
It is well known that the Ṛgvedic group consonant + uv- (Cuv), which is also
found in some other early texts (ŚS, etc.) developed to Cv- in later texts. The
recent study of F.B.J. Kuiper (IIJ 30, 1-8) underlines the fact that we must
deal here with two sets of Cuv groups, one without laryngeal (*Cuv, as in
*kúua) and one with laryngeal (*CúHa, as in *súHar > svàr, or CuHá as in:
*tanuHám > tanvàm, *tuHám > tvám, *tuaHám > tvm).
As far as *CúHa and *Cuhá are concerned, the change to Cvà seems to be in
progress in the Mantra period, as AV evidence indicates. A comparison with
PS and other Mantra texts (VS, Mantra portions of MS, KS, TS) should be
undertaken, but could not be included here.
As for non-laryngeal cases, Kuiper counts [kúua] 32 times in RV, while [kvà]
occurs only 3 times.199 A similar picture emerges from the Śaunaka AV:
[kúua] appears 12 times, [kvà] twice. The change from Cuv > Cv thus seems
to be post-Mantra in these cases. Note also the evidence from Śākalya's RV
text (late Br. period, probably Eastern: Videha), where the Middle/Late Vedic
pronunciation of the group -ḍiy- had already become -ḍy- ; intervocalic -ḍ- > ḷ
as in īḍe > īḷe, but *īḍiya- was already pronounced īḍya, see above § 6.3.
295, and 16, 261 sq.). Paiśāca = Vetāla may thus be an old nickname for persons of this
area, cf. also ŚB asurya for Eastern aboriginals; but cf. Vesālīa, etc., Weber, loc. cit.
197
Oldenberg, RV Noten, 1909, p. 218.
198
Whitney, AV Index, 1881, p.332, Kuiper IIJ 30, 1; Whitney, in his AV index: always
suarvíd, mostly súarga-; mostly suastí, always kúa, exc. once 15.11.2-3; cf. also: both svastí-
and suásti- (RV su-astí, 3x svastí); svhā (AV once, RV 1x sú-āha 3.32.19), kvà once,
otherwise like RV: kúa;
199
Once in book 1, twice in book 2, Kuiper, IIJ 30, 1.
71
In other cases, the pronunciation [Cvà] was already the contemporary one of
the Saṃhitā prose texts. S.Jamison adduces one telling example. KS, KpS
once substitute carman- for tvac- in a traditional explanation of the various
parts of the body, i.e., in a list where tvác was counted originally as dissyllabic
(as it is indeed found in the parallel passage of MS, see IIJ 29,172). As MS
generally is an older formulation of roughly the same material that is
presented by KS-KpS, one may conclude that either the composers of MS still
pronounced this cluster as [Cuva], or that even they had taken it over from
the lost Mantra time, brāhmaṇa-like explanations of the ritual.200
Usually this has been regarded as the survival of the older pronunciation.
However, the occurrence of -uv- is limited even in Taitt. texts to a few, mostly
semantically loaded, examples: e.g., suvar-, suvarga-, but also tanuvam <
*tanuHám, and even uv eva < u eva. It is not found, however, in other
traditionally protected, in everyday words like tvám < *tuHám, svastí, and
even in the part of a traditional list, tvác-. The retention of súvar-, suvargá-,
tanúvam, etc., is therefore a typical teacher's mannerism, a phenomenon
particular to the Taittirīya school.204 Probably they wanted to stress the
'ancient' character of their school in using this pronunciation in exposed
words like súvar. (Note the formula bhūr bhuvaḥ svar.) Note also that
Pāṇini, 4.3.102, knows only of the Taittirīya mantras (*Tittiriṇā proktam) but
apparently does not yet know or does not want to recognise the prose texts of
this school. From scanning Vedic verses, the Taitt. reciters knew, of course,
that some words like svàr, tanvám were to be spoken as [súvar, tanúvam].
The exact reasons for the introduction of these words and the exclusion of
others, like kvà [kúva], from the canon of the Taitt. is unclear. The history of
the RV text, with its long process of orthoepic diaskeuasis, however, teaches
that the decision of one or more particular teachers, with all of their whims,
203
With the exception of a few times in the JB/JUB: JUB suvar 3.14.3-4 (next to svar!);
suvarga 3.14.4. without v.l.; cf. also tanuve 4.32 in a verse; this belongs to the Gāyatrasya
Up. of Śāṭy., i.e to the sister school of the Jaim. (JUB indeed has two Vaṃśas)
204
Kuiper therefore justly regards it as a "school mannerism.... A historical justification
for this distinction cannot be found," IIJ 30, 2; cf. also Ved. Var. II § 773.
73
'tics', and mannerisms, introduced many unusal forms into the text.205 The
outcome is as unpredictable as the development of the spelling/pronunciation
of an English word.
Note, howewer, that there are a few "innocent" cases, in which (Ṛgvedic)
reminiscences have not played a role: uv eva TS, TB, BŚS, ŚBK or nu vāva
ŚBK (nvāva JB), but: nvai TS (!), BŚS, VādhB, KB, and ŚBK(!).206 It is
interesting to note that the Taittirīyas did not introduce this phenomenon
(Cuv) into a text that they borrowed from the neighbouring Kaṭha school; the
so-called aṣṭau kāṭhakāni (TB 3.10-12, TĀ 1-2) do not have Taitt. súvar but
the Kaṭha form svàr. The occurrence of -u- in súvar, suvargáu, and tanuvám
therefore serves as a shibboleth for typical Taittirīya texts.
******
This is one of the clear cases of opposition between the dialectal development
of Middle Vedic and Prākṛt, cf. below § 9 for otherwise far-reaching
similarities. It must be noted, of course, that the "occupational" caste
language of the Brahmins could not be influenced in all its forms by the local
Pkt.s. Their influence is sporadic and unpredictable; some of the major traits
of the underlying local Pkt.s are taken over, but other areas of the grammar
resist the "popular" forms out of necessity; there is always a need to
205
See Oldenberg, Prolegomena.
206
uv eva TS 2.3.7.4, uv eva TB 1.2.25; uv eva in BŚS, see Caland, Über...BŚS, p.51; u(v) eva
in ŚBK, Cal. p.35 sq.; but JB §186 sa u eva; nu vāva ŚBK : nvāva JB 2.11, see L.Chandra,
ed. JB II, p.23 ann.8; cf. nu vāva ŚBK : nv eva ŚBM, see Caland ŚBK, p. 82, § 39(n);
further: nvāvai ŚBK 1.4.2.5, 1.43.2 : nv eva ŚBM. -- On the other hand: nvai
TS,VādhB,BŚS (Caland, Über BŚS, p.50), ŚBK, KB, KS 23.6; nvai AB 1-5 :: vai 6-8; u nvai
ŚBK 1.4.3.2, 4.9.3.15, 7.5.3.3, explained by Caland ed. p.80 (wrongly printed as anvai!);
further, ha tvai TS 7.2.10.2, tvāvā TS, etc. Again, even in Taitt. texts, the anaptyxis did not
work in all such inconspicious cases; apparently, these forms, are mannerisms,too.
207
See O.v. Hinüber, Überblick, §371, §208.
74
*****
All of these developments must, again, be kept separate from the so-called
Eastern anaptyxis found in the Eastern Aśoka inscriptions (e.g., in Orissa)208
and in Ardhamāgadhī.209 This trait is, of course, attested in written form only
since the third cent. B.C. (Aśoka), and indirectly, in some Eastern forms, in
Pāli. The Western and Southern languages (Śaurasenī, Pāli, Mahārāṣṭrī)
tend to assimilate consonant clusters, e.g., apatya: Eastern Aśoka inscr.
(Khalsi, Dhauli in Orissa) apatiya, Western (Girnar) and Northern
(Shabazgarhi) apaca, Pāli (an)apacca (O.v. Hinüber, Überblick, p. 87).
Anaptyxis, therefore, cannot be claimed to explain the forms tuvaṃ, etc.,
found in Pāli and Śaurasenī, and Bhāsa (mentioned above, see Kuiper, IIJ
30). However, the development seems to be foreshadowed by a few
interesting occurrences of anaptyxis in Middle and Late Vedic.
upavasathyá-
upavasath¡ya-
208
Note that the Orissan pronunciation of Vedic texts continues this until today, see MSS
44, p.283 sqq.
209
See: O.v. Hinüber, Überblick, § 75 p.60, §153 p.87; cf. also Sprache d.Buddh. in
Zentralasien: Gāndharī p.31.
210
Cf. Kuiper, IIJ 30, 2; Wackernagel, Ai Gr. I, 202; Ved. Var. II § 784 sqq..
75
On the other hand, the pronunciation Cya, which is, according to the early
testimony of KS (tvak > carman, see above), to be expected in Western texts,
indeed surfaces in the older part of AB212 and the Western Sāmavedic texts:
ṢB, DŚS, LŚS (perhaps already of Gujarat location); in ŚBK, which is almost
always aligned with the Western schools, and finally in GB 1.4.7-8 which has
been taken over from ŚB 12.1.21 sqq. (according to Caland, originally a
Kāṇva text).
The case of the variants śunāsīryà- / śunāsīr¡ya- also supports this analysis:
śunāsīrīya-
TB 1.4.10.2, 1.4.3.9; BŚS 15.12:1, 17.58:1, 17.60:16, 24.10:3,4,...,°paruḥ
21.6:9,11, 25.1:2,5,14; ĀpŚS 8.20.1,5, 8.21.6, 18.9.5, 20.15.3, 22.9.1,...; HŚS
6.8.8, 14.3.18, 17.3.34, 17.4.1,...; KB 6.15, ŚSS 14.9.1,2,7, 15.12.10;
VSK 26.4.4; ŚBK 1.6.3.2,5,8,9, 1.6.4.9, 7.1.2.4;
VSM 24.19; ŚBM 11.5.2.6,9; KŚS 5.11.1,16, 18, 15.1.16, 17; °sthāna 22.7.10;
KāṭhŚS-Saṃk. °28.23; ĀŚS 2.20.1, 9.2.22, 12.4.9;
Pāṇ. 4.2.32
The distribution in this case is quite similar. The Central and Eastern
Central schools have Cīya, but not, in this case and differently from
upavasathya, the Southern Jaiminīyas: TB, Taitt. Sūtras, KB and its Sūtra
ĀŚS, and, somewhat surprisingly, both Mādhy. and Kāṇva Vājasaneyin, and
their Sūtra KŚS; the Kāṇvas exeptionally do not align themselves here with
the Western schools (MS,KS). Note that the only occurrence of the form in
ŚB 11 (and the corresponding part of VS) are from a book that is suspected
211
Śāṇḍilya's ŚB perhaps is a Central text, if we take into account the Jaina testimony of a
Saṇḍilla country in UP, see §4.1.
212
But cf. the different derivation of aupavasathya- AB 7.32.
76
The reason for this distribution may be that even in the West, the form in -
īya may have been accepted at a late stage. Note that even Pāṇini teaches
both possibilities; this may perhaps be supported by the appearance of the -
īya form in KāṭhŚS and ĀŚS. The former, however, is extracted from a single
MS of the commentary on the KŚS, as no MS of the text has been found so
far; the latter text may alredy have been composed in the East, as (part of) the
Aitareyins who had moved to Videha, etc., during the late Vedic period.
On the other hand, the original form expected for the West is indeed found,
just as in the case of upavasathyà-. The Maitr, with their Sūtras, the Kāṭhaka
Śrauta-Sūtra (which confirms the suspicion about the single Kāṭh. -īya form
mentioned just now), the Western Sāmavedins (PB, LŚS). The form is,
however, found in the Central area as well (KB, ŚŚS, later Taitt. Sūtras: BŚS
in a late Prāyascitta section, BhŚS) and even in the South (JB). Interestingly,
even the Easternmost school, i.e, the Mādhy. sub-school of the Vājasaneyins,
has this Western form, and, even more surprisingly, in their genuine chapters
1-5. If this is compared to the Mādhy. form upavasath¡ya, then the true
Mādhy. books have the Western form in -ya and those Mādhy. books derived
from an original Kāṇva text have, in both cases (śunāsīr¡ya, upavasath¡ya),
the Kāṇva form.213
It is the Kāṇvas and the Jaiminīyas, both living on the fringe of the
innovative Central area, that in one case side with the Central texts, and in
the other one with the Western texts. Obviously, many more such
observations will have to be made to arrive at a definite map of the diffusion
of this phenomenon.214
The Central texts, with their lengthening of -iya > īya, point, in any case, to
an underlying pronunciation [upavasathiya], [śunāsīriya] in the Centra, area
and in parts of the Southern and Eastern areas. The words are, it is true,
ones of traditional śrauta sacrifice, and one may explain -īya as the
intentional stress of an antiquated pronunciation [-iya] which had no
counterpart in contemporary pronunciation of the cluster -thy- viz. -rya-.
However, the anaptyxis also occurs in inconspicious groups like uv eva < u
213
This may be of great importance for an understanding of how the Madhy. text evolved!
Cf. Caland's opinion (introd. ed. ŚBK) that ŚBM 11-13 originally were Kaṇva texts.
214
Cf. further Pāṇ. 4.2.32 dyāvāpṛthivīya, marutvatīya, agniṣomīya, vāstoṣpatīya,
gṛhamedhīya.
77
eva Taitt., nu vāva ŚBK, (but nvāva JB, cf. also nvai BŚS),215 and therefore
seems to be sprachwirklich, at least in the Taitt. and some bordering areas
and texts.
Notably, the same holds true in the cases of the Upaniṣadic (traditional)
pronunciation of satya, see S. Jamison, IIJ 29, 166 f.: satya [satiya <> satya]
agrees with the distribution given above; [satiya] is found in the Eastern texts,
BAU 5, AA, KU, while [satya] appears in BAU2 (a Śāṇḍilya section!), and in a
Western/Central Sāmaveda text, ChU.216 The Jaiminīyas, an originally
Central (Śāṭyāyana), then Southern text, have more forms of this kind.
Caland, JS, p.33, adduces: duṣvapniyam, hastiya, kṛtviyo; śipriyandhasaḥ.217
It seems that the later Eastern anaptyxis began to evolve at this time, and
not so much in the East but in the Central area, with some spread towards the
Eastern Central area (Kosala) and the South (Jaim. territory in N. Madhya
Pradesh, etc.). I suspect that the Central schools, like the Taitt., were under
pressure from two factors. The old Vedic pronunciation of the clusters Ciya,
Cuva was retained in Mantras and some "archaic" words (like súvar);
otherwise they succumbed to the pressure from "below," i.e., from the
spreading pronunciation of the (later on, Eastern) anaptyxis in cases like
upavasathīya, śunāsīrīya, and even uv eva.
Still later is the intrusion of late, post-Vedic forms into some of the texts,
interestingly those also otherwise showing indications of a late redaction: as
is well known, Ved. duhe becomes in late Vedic and class. > dugdhe; Ved.
duhre > duhate, Ved. śaye > śete; however, the classical forms occur in such
texts as VSK and PB.218 Renou has shown that VSK is a text with a
comparatively late redaction; it has otherwise strongly been influenced, as far
as the form of the Mantras is concerned, by the RV.
215
Caland, Über BŚS p. 50, see in detail, above, ann. 206.
216
This could be enlarged further by investigating cases of Cy: Ciy, like vamriyaḥ VSK:
vamryaḥ VSM 37.4, suvite VSK, KS, MS, TS, RV: svite VSM 5.5, aghniye VSK, TS: aghnye
MS, MŚS, PB (cf. Renou, JA 1948, p.39), etc. - See in detail, Ved. Var. II § 774-798.
217
For similar cases, see Wack. I p. 200 sqq., e.g. gāyatriya-; cf. patnayaḥ Taitt., JB:
patnyaḥ PB, etc. (Caland, Over JB p. 15 sqq.).
218
See Caland, Over en uit het JB, p.17 sq.; Renou, JA 1948, p.38.
78
It seems that the Kauthuma school, which has no really old Brāhmaṇa texts,
except for ChU (still late Vedic, though with some colloquial traits),
reworked, at a late stage, and some older and unfortunately lost or
untraceable Sāmaveda Brāhmaṇa,226 in close cooperation with the
Kapiṣṭhala-Kaṭhas, perhaps in Gujarat, as one of the Śrautasūtras of this
school, the Lāṭyāyana ŚS, seems to indicate and where the Kap. seem to have
moved in the late/pots-Vedic period227; the home of the author of LŚS
apparently was in Lāṭī, an area of S. Gujarat.228 Note, in this context, that
LŚS/DŚS must indicate the localities on the Sarasvatī by specifying them in
detail, with the help of an inserted clause.229 The author(s) of these texts
apparently lived far away from Kurukṣetra.230
Again, such forms as mentioned above serve as warning signs. One could,
without suspicion, lump a text like PB together with older Brāhmaṇas like
TB, etc., as it seems to conform to their shibboleth-- the use of the impf. as
narrative tense-- if it were not for the late verbal, pronominal, and nominal
forms quoted above.
"Eastern RV"?
224
Cf. above §6.5 on svar.
225
See for example, gen.fem -ai,§ 5.1, etc.; see also § 10.2
226
Note that even their Kṣudra Sūtra contains many Br.-like passages.
- The final redaction may have taken place in the East, cf. the Videha king Namin Sāpya at
PB 25.10.17, and the tradition about a bhāṣika accented PB text, see below ann.
230,317,334.
227
See § 6.4 on the homeland of KpS
228
See Weber and Parpola, Transl. LŚS/DŚS p. 29. Is the split between Kauthumas (LŚS)
and Rāṇāyaṇiyas (DŚS) due to the movement of one subschool to Gujarat (together with
the Maitr., and perhaps the Kap.), while the other went eastwards to Videha (cf.the Videha
king Namin Sāpya in PB 25.10.17)? The Drāhy. (see Parpola.LŚS tr. p, 40), later on are
found in S. India (together with the Taitt.) cf. ann. 230,334. Both schools differ in the use of
-ym- (Kauth.) viz. -jm- (Drāh.).
229
See K. Hoffmann, Ortsnamenparenthese, Aufs. p.123 sq., cf. Parpola, Transl. p. 30,
ann.1.
230
Cf., however, the supposed movement of the Kauthumas to the East, see ann. 223, note
the alledged bhāṣika accents of PB (Śabara); cf. Parpola, transl. LŚS, p. 30, cf. ann. 228.
80
In the Vedic texts, with their long history of oral transmission, and the
changes made by the collectors, redactors, and at the time of the final
canonisation, it is necessary to be attentive at every step to such warning
signals; they occur, however, time and again, if one is careful enough to notice
them. Even if the Kauthumas had intended to produce a "perfect
Brāhmaṇa" text, they would hardly have succeeded in cases like yuyām,
anuvyam, etc.
It is well known that the Vedas have not reached us in their original form.
Just as many other Indian texts, like the Pāli canon, the Mahābhārata, etc.,
they have gone through a period of oral transmission which was followed by a
redaction. In the case of the various Vedic texts, this has been well studied
only for the Ṛgveda, notably by Oldenberg in his Prolegomena.231 The text of
the RV has been transmitted by only one school, the Śākala śākhā. The other
two prominent schools, that of the Bāṣkalas and Māṇḍukeyas, have, for all
practical purposes, been lost. As far as the other Vedas are concerned, this
process has not been studied and understood very well.232 Just as in the case
of the Ṛgveda, however, there are a number of indications which show that
these texts had a pronunciation which was different from the form that the
texts have now, in their post-redactional shape.
One typical example, just as in the Ṛgveda, is the Abhinihita Sandhi, of final
-as/ -e before a-. In the the Vedic texts, in the language Pāṇini uses in his own
231
After what has been said above on the history of some Eastern texts, a detailed
investigation of PB, ŚBK, ŚBM is in order; for PS, see for the time being, ZDMG,
VI.Suppl.Bd., p.256 sqq., 1985.
232
Especially PS, PB, ŚBK, ŚBM should be studied in detail.
81
grammar, and in classical Sanskrit, it results in -o '- viz. -e '-, This looks
like an "elision" of the initial a-, and is usually described thus in Western
grammars. Most Vedic texts, however, often write -o/-e a- next to -e/ '-. This
has not been a real pronunciation, however, and is nothing but a second-hand
attempt by the readctors to restore the intellegibility of the text viz. the metre.
Something like *-ai a- is to be expected as Ṛgvedic pronunciation.233
To suppose a development -az a- > -ai a- for this early period would lead to
nom. sg. like rathe, deve, putre, name, etc. only in Abhinihita Sandhi. This
would be in competition with the normal forms in -o before voiced cons., like
aśvaz vahati > aśvo vahati. Such forms in -e are unknown from Skt., except
for a single case in RV where an older gen.sg. *sūras duhitā > sūraz duhitā >
sūrai dūhitā > sūre duhitā. The development here is one that otherwise is
found in internal Sandhi, e.g. *sasdai > sazdai > saidai > sede. (A similar
development, also one that took place in the post-IIr. period, namely -az +
voiced retroflex cons. > o is found in internal Sandhi: vajh-tā > vazdhā >
voḍhā).
New Mexico Spanish, where exacto > esauto or esaito and of modern
Provençal, where -s unv.C/V-, but -i vC-.
Here we have to take into account the nature of Vedic a, which was of a closed
variety and this different from open ā. This is clear from Pāṇini's last Sūtra
and has recently been shown by K. Hoffmann, Aufs.p. 552 sqq. Pluṭi cases like
TS 3.2.9.5 śosā moda-iva [śsʌ mda-iuəә] @ < śasa madeva [śəәsəә
məәdaiuəә], ŚB 4.3.2.14 othā modaiva indicate that pluṭi lenghtening of a
resulted in [] which is different from normal ā []. - In analogy to -az vC, -az
a- must have developed, at a time when pre-RV/Eastern -az vC > -e vC (sede,
and az D(h)> voḍha), no longer worked, via -au a- [ əәuəә] to -o , as for
example: *devaz asti > devəәuəәsti > devōsti. @
By the time of Pāṇini this had taken place already, as is evident from the
Sandhi employed in his grammar and his own rules at 6.1.72, 6.109 sqq. (cf.
P.Thieme, Pāṇ. and the Veda, p.46 sqq.). A Padapāṭhakāra at Pāṇ.'s time
then had to decide, for every single case of -e/-o, whether the following word
contained an original a- or not. This is simple in cases like *aśvo vahati
(where aśvo av° is impossible), more difficult in some cases where privative a-
or, worse, the question of augment or no augment is involved (see K.
Hoffmann, Inj. p. 146 sqq.). The only way to indicate what was intended was
to insert secondarily from the Padapāṭha the Sandhi form -e /-o found before
vC and to restore the "lost" a- thus: aśvo vahati, devo asti, namo astu. A
pronunciation devosti [devsti]@ left no other choice, even though the early
grammarians noticed that not elision but substitiution of two sounds by one
had taken place.
17: Pāṇ.(?):
after: bho,bhago,adho: -y- + V / vC [*bho-y-atra ]
[*devā-y-iha]
-a/ā -y- + V / vC [*ka-y-āste]
[**brāhmaṇā-y-dadāti]
The last rule apparently negates the form marked above with a double star
and thus forms: [brāhmaṇā dadāti]. The general rule (of Pāṇini ?), i.e.
insertion of -y-, however, has not been followed even in the text of Pān.'s
grammar. The reason is that Pāṇ. teaches the usual, class. form of Abhinihita
Sandhi (-e/o '- ), at 6.1.109 sqq.; this rule applies to Saṃhitā texts (6.1.72). 237
The rules in 8.3.17 sqq. therefore seem to have been entered only to record
the variant pronunciation of some teachers.
It is interesting to note that all of the authorities quoted here are either
Eastern or Central. Gārgya (cf. Gārgī at BAU) is a name the origin of which
is to be searched with more probability in the Centre than in the East (cf.
author, Fs. W.Rau: The case of the shattered head). Śākalya is, by the time
237
See P. Thieme, Pāṇ. and the Veda p. 47 sqq.
84
The two sets of rules concern (Vedic) texts in their Saṃhitā form (see Pāṇ.
8.2.108 with 8.3.16 sqq., and 6. 1.109 sqq. with 6.1.72). The Vedic texts, even
the Eastern ones, cannot be expected to correspond exacatly to forms of early
Middle Indian or to the later Prākṛts. Otherwise, one would expect, for the
East, a special Sandhi rule, taught be Eastern grammarians, to take care of
the usual nom. pl. of i/u stems in -e (Eastern MIA bhikkave, see O.v.Hinüber,
§ 332) Or must we think of an older stratum, valid only for Brahmanical
speech, where -o was Eastern? On the other hand, even the Asuras (using
non-Brahmanical speech!) say in ŚB: "he 'lavo he 'lavo!" 239 This would
mean that even in "Pkt.," -o was the normal form of (nom.) voc.pl. of i/u
stems in the East at the time of ŚB. On the other hand, nom. pl. in -o is
typically Western MIA. - The majority of the Sandhi cases mentioned in
Pāṇ., however, will have come from nom. sg. in -as > -o in Sandhi like devaḥ >
devo, and also from s-stems: namas- > namo. The pressure of these forms
will have also set the rule for the nom./voc. pl. of i/u stems -as > -o, even in the
Brahmanical/'Asura' speech of the East.240 (As another non-Eastern
phenomenon in the Asura quotation note the Sandhi -e '-). It seems that the
Western Sandhi in -o ', -e '- prevailed everywhere. Pāṇ. is, in this case, more
modern than some of his Eastern colleagues.241 The older (Ṛg)Vedic practice
seems to have been closer to Śakaṭāyaṇa. This is evident if the abhinihita
sandhi of the Ṛgveda is reconstructed, for example:
-as vowel- > -ay vowel-. KpS occasionally still changes -e > -ay , -o > -av, -
ai > -āy before vowel: varuṇadhā-y-iti (Ms. eti) KpS 6.8; ruca-y-eṣā 7.5; ta-
y-enam 7.8; vibhāvasa-v-iti 30.3.242 PS also has a few cases like this, e.g., ta-
y-eka PSOr 6.15.8.
Unless the occurrence of this rule in KpS is due to Maitr. influence on KpS
in medieval Gujarat, it must go back to the Vedic period, when both schools
lived in close proximity, in S. Kurukṣetra viz. S. Panjab. A decision is
difficult to reach at this point, as KpS has been edited on the basis of a single
manuscript of unknown provenience. A thorough search in Gujarat would
perhaps turn up more MSS and help to solve the problem. In either case,this
school peculiarity may provide a hint that Pāṇini's general Abhinihita rule (-e
, -o) originated in the Northern part of Kurukṣetra (KS,AB) and not in the
Southern part (MS, MU). This would be very important for a study of the
history of Vedic texts; clearly, more research is needed.
To sum up: The older Sandhi forms still taught by Pāṇini were (occasionally)
preserved by some of the rarer texts, notably those of which a Prātiśākhya
has not been composed (or has not been transmitted during the post-Vedic
period). The "classical" Abhinihita Sandhi taught by Pāṇini at 6.1.109 sqq.
and actually used in his grammar has been introduced into almost all of the
242
See Raghu Vira, ed. intr., repr. p. VI.
243
Schroeder, ed. MS I, p. xxviii; cf.also Lubotsky, IIJ 25.
244
Schroeder, ed. MS I, p. xxxix; Raghu Vira, ed. repr. p. VI.
86
Vedic texts,245 beginning with the RV; this must have happened already at the
time of Pāṇini, as he subsumes this form of Sandhi under his saṃhitā rules.
************
The following paragraphs, while dealing with topics that perhaps look more
like matters of style than dialect variations, are meant to underline the broad
divisions made above and to refine them to some extent; only a small selection
of the facts can be presented in this article.
It is well known from the study of the style of individual authors246 that
particles are a very useful and effective tool in the the process of determining
whether (part of) a text belongs to a particular author, or whether an
anonymous text is composed by one or more authors. This procedure can be
used, with profit, in the study of Vedic texts and their affiliation with the
various schools (śākhās). It can also be applied in the study of texts said to
have been composed by a particular author, like Yājñavalkya.
The figures found for the use of the particle, khalu, show that after a single,
initial appearance in the late RV, the centre of its diffusion lies in the territory
of the Maitrāyaṇīyas.
__________________________________________________________________
RV 1 (10.34.14): 100 %
245
Cf. the statitsics of Ved. Var., II p. 423
246
For example, cf. B.Kölver on Kalhaṇa's Rājataraṅgiṇī, or Schetelich on the Arthaśāstra,
T.Vetter on Śaṅkara's works, etc.
87
PS 0? ŚS 0!
KS 11 : 1293%
(late TS 134 : 1840 %
parts:
32-35)
MS 69 : 681%
ABo 3 : 1295%
TB 38 : 2739%
TA 18 : 3443%
---------------------------------------------------------------------------------------------------
KB 2 : 1061% ŚBK 7: ŚBMo 24 : 3662%
KA 3 : 997 %
(>svid)
ŚBMw 2 : 426%
JBc 9: 4409%
JUB 2: 1252%
ŚBMn 11 : 4211%
247
KaṭhB 1 : 95 % VādhB BŚS
(=TBk) ŚBMa 0 : 0%
ŚBMu 9 : 2476%
ChU 4 : 663%
PB 18 :1934% ABn 0! : 0%
ṢB 3 (6-8)
AA 2 : 455%
MU 2 : 203 %
__________________________________________________________________
It is notable that the oldest texts, i.e., RV, PS, ŚS, (SV, RVKh) do not know
this particle (except for one case in RV 10.34.14). Khalu suddenly makes its
appearance with 69 cases in MS, the oldest surviving YV Saṃhitā. The
neighbouring, and slightly later, Kaṭha school, however, does not follow suit,
except in its later books 32-35 (which belong to the Orimikā < *avarimikā
section), i.e. the old, final book III (before the Yājyānuvākas and the
247
N.B.: VPK does not mention khalu in the Sūtra Vol.s, except for a few token examples
from ĀpŚS, Yāska, etc. The new ed. of the Brāhm. Vol. mentions only a few cases as well;
the older ed. has many more. The figures given here for JB are from the old ed. of VPK
and refer to the JB ed. of Caland only (JBc).
88
Aśvamedha book were added). These chapters deal largely with the Sattra,
with additions to the New and Full Moon sacrifice, and with the Prāyaścittas.
The occurrence of khalu then diminishes in some late Vedic texts, and not
only in the East (ŚBMu): MU 203; ChU 663%, down from PB 1934%, JUB
1252% (< JB 4409%), ABn 0 (< Abo 1295%; this is an unexplainabe figure),
AA still 455%, and even ŚBMa has no cases.
The use of khalu in the older Saṃh. and Br. period is thus restricted to the
two Yajurveda texts of the Maitr. and Taitt. schools. The later Br. texts
reveal the same picture, with the exception of the ŚB; some differentiation
89
must be made here. The Eastern Central Kāṇva school (books 1-7) has only 7
cases of khalu, as compared to the corresponding Mādhy. books 1-5, with 24
cases. It is interesting to note that the Kāṇvas often substitute svid where the
Mādhy. have khalu.248 This distribution again agrees with the overall picture
which the Kāṇva books exhibit. They form a pièce de résistance in the East,
wedged between the Taitt. and the Mādhy. schools.
The originally Western (Śāṇḍilya) books ŚBM 6-10, as well, contain only 2
cases, and therefore stand very close to the Kāṇvas and Kaṭhas, in opposition
to the Maitr./Taitt. schools.249 The later Mādhy. books 11-13, however, have
11 cases, while BAU (book 14, 3-9) has 9 cases. The later ŚBM, of Eastern
origin, thus prefers to use khalu, while its truly Western components avoid it,
although the latter had probably been transmitted to the East by the time of
ŚBM composition.
Finally, it is interesting to note that an originally Western but late text, PB,
makes frequent use of khalu. The text, as we have it now, may, however, have
been redacted in the East. This holds true for ChU (with diminished
percentage of khalu) as well, which belongs to the same SV school but
contains some geographical materials which place it more in the direction of
the Centre or the East. This low figure makes the text more akin to ŚBK
(997%) and especially the "Western" Śāṇḍilya books of ŚBM (426%).250
Unfortunately, this picture cannot be supplemented by the evidence of such
late Br. and early Śrauta Sūtra texts as VādhB and BŚS, for here again, VPK
does not provide sufficient information.
__________________________________________________________________
248
Cf. also Caland, ed.ŚBK p.78 ("vai or svid"); at ŚBK 5.2.1.5 khalu svid ŚBK, but svid
ŚBM.
249
It is important to note that there are more similarities between the Kaṭha/Caraka and
the Western ŚB books; see below.
250
The redaction is very late; probably the text was summarized out of a larger (Prauḍha),
lost *Kauth. Br. referred to above, cf. ann.125, 223,250,290,334.
90
PS 13 : 40%251 ŚS 1 : 5% VSM 2 : 6%
KS 10 : 9%
(KSa 8) TS 1 : 2%
MS 23 : 49%
(mostly
in prose)
The distribution is interesting. After many instances of svid in RV, only the
oldest YV school, the Maitr. continue to use this particle; it modestly survives
in the early Aitareya (ABo), Kauṣītaki, later Taitt. (TB), and earlier Mādhy.
schools (15-26%), but scores better in the earlier Kāṇva books (36%) and the
late ABN (27%), and then resurfaces strongly in the later Mādhy.252 and in
the Jaim. schools (138-285%). This is a late Eastern / Southern alignment
which appears from time to time and needs further investigation (see §10.2).
251
PS is approximate only; the new VPK Br.Vol. has only 1 case for ABo; the old Br. vol.
has 4 for ŚBK; the Sūtra vol. has only a few token cases as well.
252
Caland, ed. of ŚBK, p.108, believes them to have been Kāṇva works originally. The
Kāṇvas apparently have a different style than the Western Śāṇḍilya books: Note that svid
does not occur at all in Śāṇḍ. ŚB 6-10, and equally not in the Śāṇḍ. parts of BAUM (ŚBM
14.4-5), but rather in the later sections (14.6 sqq.).
91
Even in subsections of a text, like ŚBM 1-5, one can distinguish certain styles
in the use of particles. ŚB 1-2, for example, uses na hi tad, while ŚB 3-4 uses
na vai tad.253. It seems possible that we can recognize here the personal style
of the author of these chapters, or, at any rate, that of the compiler of the
chapters in question (who must then be different from the compiler of the ŚB
as such).254 In books 1-4 of ŚB there is, indeed, a difference in subject matter
which makes two authors likely. ŚB 1 treats the New and Full Moon
sacrifices, 2 deals with the Agnihotra, etc., 3 treats the Adhvara (Soma), and
4, Soma (5: rājasūya, vājapeya). Both rituals (Dārśapaurṇamāsa viz. Soma)
are prototypes of many others and may have been composed, in the late Br.
period, by two different persons who based themselves on their older Black
YV predecessors. - Another example is: athātaḥ ŚB, BŚS, KB, AB.
***************************
A few interesting cases involving single words and their spread throughout
various texts, as well as typical (certainly style-oriented) expressions, may be
added here. It is the (still very incomplete) collection of many such individual
cases which will add perspective to the picture of the major grammatical
forms and sound changes delineated above. Only a few can be given here, for
want of space and opportunity to investigate them in detail.255
253
See Minard, Enigmes I, §800.
254
Note that the compiler of ŚB had a good overview of the text, such a good one, indeed,
that he could compare a section in the Soma book with one in the Pravargya book; see
author, Fs. U. Schneider; cf. also Whitney, TAPA 23, who quotes several cases in ŚB where
a passage reappears with the same wording (and the same mixed use of the tenses).
255
I plan to add to the present collection of materials, from time to time, in Journals like IIJ
and StII.
92
It has been mentioned earlier that the traditional phrases which are found in
the introduction to a myth or aetiological tale beginning with, "The gods and
the Asuras were in contest," usually show up in the texts in two varieties, one
using the word spṛdh and the other the compound verb sam.yat.
In the oldest texts, MS and KS, the use of spṛdh is found almost exclusively;
starting with KS, sam.yat makes its appearance and is especially popular with
the Central groups of the Taitt. and the Jaim., but not in the East. (The
distribution, again, is similar to that of vāva and many other grammatical
features mentioned in the preceding paragraphs). When the rates of
occurrence are compared and calculated against each other (sam.yat : spṛdh),
this picture emerges:
---------------------------------------------------------------------------------------------------
early AB 6/3
Brāhm. ABo 4/2 TB 14/2
PB 0/7
ṢB 1/3
---------------------------------------------------------------------------------------------------
When the relative percentage of sam.yat : spṛdh is calculated for each text
(regardless of relative size), we arrive at the following:
93
__________________________________________________________________
KS 400 % TS 375 %
MS 25 %
TB 700 %
AB 200 % KB 100 % ŚBKo 0.04 % ŚBM 0.03 %
PB -0.! % JB O.22 % [ŚBMo,w,u -0!]
256
ṢB 0.33 % JUB 150% BŚSb 200% [ŚBn 0.16 % ]
ABn 200 %
__________________________________________________________________
The centre of usage of sam.yat lies with the Kaṭha and Taitt. schools. As TS
is a s,ightly later version of a YV Saṃhitā than is KS, the origin of the phrase
must be sought with the Kaṭhas or their close relative, the lost Caraka school.
The use of sam.yat is their innovation. MS is still older than KS, but
predominantly employs spṛdh, an old Indo-Iranian and Indo-European term
of racing and competition (OAv. spəәrəәd, Engl. sport, Germ. (sich) spurten,
etc.).257
Note that the older Aitareyins who live in the same area as the Kaṭhas follow
their trend; they keep this up, even after their emigration to the East, with the
same percentage. This stands out clearly, as all the Eastern texts have but a
sprinkling of cases of sam.yat, even the usually Western-oriented Kāṇvas.
This is one of the many cases where the SW Maitr. and the Eastern Vāj. go
together, which is frequent in later ritual and in the use of particular
mantras.258 The group MS-Vāj. is joined, in this case, by the Western and
Southern SV (PB,ṢB,JB). In other words, the innovation sam.yat is limited to
the Yajurveda and Ṛgveda texts. (Note the enormous increase in TB, in
BŚSb, and in the late JUB, although the actual occurrence of cases is limited
here.)
256
Note that all cases of sam.yat in JUB come from one passage, 2.10.1 (= VPK notation:
2.4.1.1), as opposed to two of spṛdh: 1.15.4.1, 1.8.5.1 (VPK).
257
Cf. Schroeder, ZDMG 33, 177 sqq.
258
How was this possible in geographical terms? In later texts, a closer connection between
the two schools is possible, as MS expanded southwards and ŚBM south-westwards; the
Madyandinoi reside, according to Arrian, Indikē 4.4, on the Southern side of the Ganges,
opposite Allahabad. But what about early texts? - The (non)occurrence of Maitr.-related
material must be checked in Jaim. texts.
94
purovásu- PSK 18.11.7; JS 1.25.3 < RV 8.49.1 (-ū-), 4.14.6, 1.16.2, 1.20.9,
1.26.8, 2.7.1, 4.20.6; MS 1.3.9; MŚS 2.4.1.33; TS 3.2.10.2, 3.2.5.1; TB 4.20.1;
BŚS 7.14.34; HŚS 8.6.26; VkhŚS 15.26.10, 15.32.2; ĀpMP 2.4.1.33.
---------------------------------------------------------------------------------------------------
ū RV ū SVK ū SVJ ū AVŚ ū VSK ū VSM
ū RVKh ū PB ū KB ū VaitS ū KŚS
u PSOr (o PSK) ū ŚŚS
KS 0
ū AB
ū AA
ū ĀŚS
---------------------------------------------------------------------------------------------------
o TS
o TB o BŚS
o/u MS o HŚS
o MŚS o ĀpMP
o VkhŚS
ū SVJ
ū/o SVJ
---------------------------------------------------------------------------------------------------
95
The forms in -o- are restricted to the Maitr., Taitt., and Jaiminīya schools.
Unfortunately, the Kaṭha school does not show any case (as with KaṭhA). The
few divergent forms in SVJ must be investigated. Note that AVŚ has forms
only in book 20, all of which come from RV, except for one form in book 14 =
RV 8.1.12. The only quotation from MS which has -u- is also found in the RV
and must have been influenced by this text (MS 4.9.12=RV 8.1.12).259
On the basis of the broad dialectal features established earlier, it will be very
instructive to look at one or two examples which reflect the religious
development of the late Vedic period. One such case, and a very important
one in the development of the idea of rebirth and karma, is punarmṛtyu, "the
recurrent, the second (and third, etc.) death," which is believed to occur after
one's death here on earth.
Mantra/
early
Saṃh. 0
ABo 0 TB 4 : 4 %
early (TB 3.9.22 and TbK)
Brāhm.260 TBk 3 : 12 %
TAk 2
259
For the sound change o/u, cf. Ved Var. II § 717, 721; the single -o- in the Kashmir Ms of
PS is negligable, as u and ū are constantly interchanged in Kashmiri MSS; the
pronunciation is, in all three cases, [o].
260
The passages are: KB 25.1; KA 13.1; TB 3.10.10.5, 3.11.8.52, 3.11.8.6; TA 2.14.1, 2.19.1;
VādhS 3.90; BŚS 2.11:22, 28.4:32; JB 1.6, 13, 232, 252, 46, 245, 246; JUB 3.6.7.7-82,4.12.2.6
; ŚBK 3.1.9 differs from the parallel = ŚBM 2.3.3.9, 10.1.4.14, 10.2.6.19, 10.5.1.4, 10.6.1.4-
9/11, 10.6.5.8, 11.4.3.20, 11.5.6.9, 12.3.4.11, 12.9.3.11-12, 14.4.3.62, 14.6.2.10, 14.6.3.2; BAU
1.2.7, 1.5.22, 3.2.10, 3.3.2; GB 1.1.5, 1.3.22; . KaṭhB (Svādhy.:81.7)
96
---------------------------------------------------------------------------------------------------
PB, ṢB 0
KaṭhB 5 ABn 0 : 0%
(in pt.s = TBk/TAk:5x) AA 2 : 19%
(Kaṭh.Svādhy.Br.= TA 2)
MU 0
GB 2
__________________________________________________________________
This evidence is interesting in many respects. First of all, the older texts do
not know this word, and, apparently, they are also unfamiliar with the
concept; it is found for the first time in the later Brāhmaṇas.261 Even among
these late Vedic texts, the first books of both ŚBM and ŚBK do not use the
word, except for one case at ŚBM 2.3.3.9, where it looks like a later addition
(it has no parallel in ŚBK!), in a passage dealing with one of the many
speculations of the Brahmins and the Kṣatriyas on the nature and secret
import of the Agnihotra sacrifice, so typical for late Br. texts and the
Upaniṣads.262 The same applies to JB, where the word is almost exclusively
found in the late portion, JBa (JB 1-1.65, which also deals exclusively with the
Agnihotra).
261
For the concept, see H.P. Schmidt, Mélanges Renou.
262
See the examples in Bodewitz, The daily morning and evening sacrifice, Leiden 1976.
263
This has been well known since Weber first established the facts more than 100 years
ago; cf. K. Mylius, Untersuchungen, see above, §4.1. ann. 34. Note that ŚB 10 contains
97
It is important to note that the Central schools are hardly affected. The
Taitt. school employs the concept, as far as true Taitt. texts are concerned,
only once in its late TB book 3.9, and otherwise in the aṣṭau kāṭhakāni, an
addition from the lost KaṭhB. Even a late Anubrāhmṇa, VādhB, has only one
occurrence, and a Sūtra of the same period, BŚS, has only 2 cases. Similarly,
the Central Kauṣītakis have only 2 instances in their Br. and Ar. taken
together. Finally, a Western school, the Kaṭhas, exhibit this word in the
decidedly late part3 of their Brāhmaṇa (i.e the parts which correspond to the
fragments preserved in TB 3.9-12 and TA 1-2, and the Svādhy.Br.= TA 2).
The origin of the word (and of the concept) punarmṛtyu is, therefore, in all
probability, to be found with the late Śāṇḍilya tradition of ŚB, e.g., not in the
extreme East of Northern India, but in a more Western region and, perhaps if
old, with their Southern neighbours, as can seen in JBa.264 It is typical for the
esoteric discussions about the meaning of the Agnicayana (both Śāṇḍilya and
Kaṭha) and the Agnihotra. It is to be noted that both rituals were of
immediate concern for non-Brahmins as well; many Kṣatriyas take part in
the discussions about the Agnihotra, a standard topic of the brahmodyas and
other types of public debates.265 The Agnicayana was, due to its cost and the
elaborate rituals involved, of interest especially to the royal families and the
well-to-do gentry.
references to Śāṭyāyani (10.4.5.2, the presumed author of Śāṭy.Br. > JB) and to Śāṇḍilya
(10.6.3, etc.) and Celaka Śāṇḍilyāyana (10.4.5.3).
264
Unless further research shows that ŚB 10, although a Śāṇḍilya book, was composed in
the East by members of the Śāṇḍilya school.
265
On this topic, cf. Fs. W. Rau, The case of the shattered head.
98
__________________________________________________________________
RV 11 : 100 %
266
VPK is not exhaustive here; I indicate suspected ommissions in enumerating the
occurrences of pāpa by + or ++.
99
BŚS 18: 2?
TA 1 3
TA 10 10
TU 1
JB 45++ : 377%
JBa 8 + : 1186% ABn 11 : 874%
JUB 32 : 1821%
AA 1 : 120%
MU 0
GB 5 (mostly from 1 passage)
__________________________________________________________________
After the RV, there is a strong Mantra and YV Saṃh. time use of the word
(30-600%), except for VS (76%) as most cases there are from RV. Most of
these passages have forms of the comparative or superlative only. The word
is used, in a similar percentage in the early and later Br.s (200-350%), except
for an area located in the Kāṇva and Śāṇḍilya territory (only 77-155%). The
then expands rapidly in the late Eastern (556-974%) and especially the
Southern texts (1186-1821%). Note that the later, Eastern part of AB (6-8)
has a comparatively large number of cases as well (874%).
It would be interesting to check which words are used instead of pāpa in the
other texts. It is readily noticeable, however, that the usage is closely linked
to ideas about guilt which are important in the context of punarmṛtyu. In
both cases,267 it is the late Eastern and Southern texts which initially viz.
heavily employ the word; from the East, the fashion then spread westwards,
to reach the late Taitt. (TB, in TB 3) and the Kaṭhas (Kāṭhaka portions of
TB).268
*****************
area (purovasu-) and the Eastern area (if ŚB is, as it seems, a late text:
punarmṛtyu and pāpa). Just like developments in grammar, such matters of
style spread in various directions from the original centre of innovation.
Given what has been established above in §§ 1-8, it will be clear that a
number of developments can be traced in the history of Vedic which are
based on forms of actually spoken Vedic language, and which are not the
result of simple matters of style, as, for example, the fem. gen. in -ai (as
opposed to cases like the use of spṛdh or sam.yat in traditional tales of the
gods).
If this is so, it should not be surprising if some of these Middle and Late
Vedic developments could be found in the various Prākṛts, especially in the
older forms of these Prākṛts, and preferably in those MIA dialects which
occupy the same area as the Vedic dialects in question. Vedic forms that are
comparable to Pāli and other MIA forms have recently been treated by
C.Caillat and O.v. Hinüber.270
In the sequel, I will rely mainly on forms from the recent, up-to-date
treatment of the early Middle Indian by O.v. Hinüber, Das altere
Mittelindisch im Überblick (SB d.Oester. Akad. d. Wiss., phil.-hist. Kl., Bd.
467, Wien 1986, abbreviated O.v.Hinüber, Überblick, in the sequel). This has
the additional advantage that many of the forms listed by Geiger and Pischel
and used by later scholars for comparisons with Ṛgvedic Skt. have been
reviewed, and the forms, wrongly attributed by earlier grammars to the
various Pkt.s, are eliminated from this paper. As has long been noticed, the
269
For a summary of facts see O.v. Hinüber, Überblick, §7-12
270
See Die Sprache der ältesten buddhistischen Überlieferung, hg.H.Bechert, Göttingen,
1980, p. 50; Fs. I.B.Horner, Dordrecht 1974, p.49, ann.49; O.v. Hinüber, Überblick, passim;
and his rev. of Bechert, Die Sprache...IF 88, 1983, p.307 sqq.
101
§9.1. "Ṛgvedic" ḷ-
Kaṭha Paippalādin
(Kapiṣṭhala) Aitareyin Kauṣītaki Kāṇva
Eastern Ait.
Śākala RV? Śākala RV
Jaiminīya ĀŚS
__________________________________________________________________
Pāli: -ḍ- > ḷ- (MSS interchange with -l-)
Inscr. W: Mathura, Sanci, S:Karle, Nasik E: Jaugada
(SE: Amarāvatī,Bhattiprolu)
Old Śaurasenī ḷ- Old AMg. -ḷ-274
Aśoka:275
------------
W: dbādasa,treḍaśa,
terasa ------- E: duvāḍasa , traidasa
N: badaya codasa
271
See, e.g., Emenau's article, The dialects of Old Indo-Aryan, see above §1.; the question is
now summed up by O.v. Hinüber, Überblick, § 9.
272
Cf. also Lüders, Phil.Ind., 546 sqq., Pischel, Pkt. Gramm. §226, 240; Wack., Ai. Gr.
§222. The Northern MSS usually have -l- , as do the N. Kāṇva MSS; see Caland, ed. ŚBK,
p. 12-23 passim. In the South, it is often inversely -ḷ- instead of -l-, thus the "Bhāsa" MSS;
cf. W. Rau, MSS 42, p.187 sqq. In the Orissa PS, a clear distinction is made between ḷ < ḍ
and old l; Pāli has -ḷ-, later, secondarily, -l- (O.v. Hinüber, Überblick: §198); Gāndh. -'ḍ- <
@ -¬-. In Mārāṭhī, ḷ has been written since the 14th cent. (Master, Old Mar. §55); -l-
modern pronunciation often is -ṛ(h); cf. Turner, Coll. papers, pp. 239-250.
273
See O.v. Hinüber, Überblick, §198 sqq., and on numerals, §401-402.
274
For the fragments of Buddhist dramas, see Lüders, Phil. Ind., p. 547.
275
For the forms, see O.v. Hinüber, Überblick, §400-402.
102
Pāli:
ekādasa, bārasa, -teḷasa, paṇṇārasa/pañcadasa
------
ekārasa, pañcadasa/pannarasa (Gramm.)
In some of the words from 11-19, the realisation as r/ḷ/l may have been
influenced by the immediate surroundings (trayodaśa: r..d ṣoḍaśa : ṣ..ḍ. etc.).
The general distribution of -ḷ-, however, is more important in this context.276
The evidence seems to indicate an early diffusion of the feature in Pāli, the
Aśokan inscriptions (Delhi, Radhia, Mathia), Gāndhārī/Niya Pkt., and a later
spread to all Prākṛts. In the numerals, under the special conditions
mentioned, intervocalic -ḷ- is found in Pāli and, perhaps (with -l-), in
Ardhamāgadhī. Note, however, the almost universal change to -r- (continued
in NIA), which seems to indicate a pronunciation [ṛ/ḷ]. It is surprising that
Māgadhī is missing in the list. On the whole, the originally Western -ḷ- of
Vedic, which by the late Br. period had reached the East and South, had
spread to all MIA dialects by the time they were first recorded.
276
See O.v. Hinüber, Überblick, p.103 § 198; cf. Lüders, Phil. Ind., p.546-651, Wack., Ai.
Gr., Nachtr. I §222.
103
The Ṛgvedic nom. pl. -āsaḥ277 is found in Old Avestan as -ānhö, and in Mede
as -āha (as represented in the O. Pers. inscriptions: aniyāha bagāha). The
extension by -as is, therefore, an Indo-Ir. development (*-āsas) which had
(partially) affected some of the tribes but not all, notably not those of a later
wave of immigrants (e.g., Y. Avestan, O. Pers., Post-Ṛgvedic). While the
innovation -āsaḥ is found in the RV, the older form -āḥ is found exclusively in
post-RV texts (except for archaisms and quotations from the RV/mantra
language).
It is the later, mostly post-Ṛgvedic form (-āḥ), that has gained prominence in
all Prākṛts (-āḥ > ā), except for -āse in Pāli verses (see O.v.Hinüber,
Überblick, p.144 §312). This is a new formation in conjunction with the
Eastern nom. pl. -e. Note that it is not found in Aśoka inscriptions or in Pkt.,
as Lüders has shown (Kl. Schr., p.437), apparently with the exception of two
cases in a Western inscr. of Aśoka, that of Delhi Topra.278 It is also
remarkable that -ase does not yet appear as a popular form of Eastern Indo-
Aryan in the "language of the Asuras," the famous exclamation he'lavo at ŚB
3.2.1.23 (ŚBK 4.2.1.18 hailo), which still retains the old voc. pl. -o (< -aḥ) in
both versions; this is a phrase which otherwise shows the Eastern
particularity of r > l, as well as that of -y- > -v-:279 he 'lavo < he 'rayo, "hey,
you strangers/guys."280 It may well be the case that -āse is a fairly late
development.281 In this connection, it is interesting to note that AMg. -ao is
equally a new formation (-a + -o of the cons. stems; see O.v.Hinüber,
Überblick §312; voc.pl. Apabhraṃśa -aho < -a + bho, Überblick §322).
Thus the Madhyadeśa form is again accepted almost everywhere, except for
some remnants in a few Old Pāli verses. This case indicates that the process
of the spread of Madhyadeśa forms was a slow one. Apparently, it did not
reach the East (Pāli verses) until the last few centuries, B.C.
277
Cf. O.v. Hinüber, Überblick, p.144.
278
See M.A.Mehendale, Aśokan Inscriptions in India, Bombay 1948, p. 28 §53 IXb:
viyāpaṭase.
279
See O.V. Hinüber, Überblick §214: E. Aśoka inscr., in Pāli (partly), in AMg, Paiśācī.
280
Cf. ann. 2, 240; cf. P.Thieme, Der Fremdling im RV; in Vedic, cf. the JUB story 3.7.8, see
Fs. W. Rau, The case of the shattered head.
281
Cf. nom. sg.in Aśoka, A.Mg., and Mg. remnants in Pāli; see O.v. Hinüber, Überblick.,
p.127 §296; but pl. according to the pronoun te?, Überblick, p.161 §378 te. - Otherwise, one
may think that perhaps it was indeed the famous "first wave" of Indo-Aryan immigration
into the East which had perpetuated the spread of the Ṛgvedic usage -āsaḥ to the East,
where it remained in use, while the Kuru-Pañcāla form -āḥ gained prominence in the rest of
the Middle Indian dialects.
104
In the RV, the stems in -a have the pronominal ending -ebhiḥ in the instr. pl.
next to the nominal ending -aiḥ. This innovation (cf. O. Pers. -āibiś but Av. -
āiś) spread to all the Prākṛts, while it disappeared in the post-Ṛgvedic texts.
The local Prākṛts, however, have -ehi (Aśoka, Pāli, Pkt.s; see O.v.Hinüber,
Überblick. §189), with the sole exception of some remnants in Pāli, where -ais
> -e (O.v.Hinüber, Überblick, 145 §316). The crux is to determine whether
this form is a Vedic remnant or a Middle Indic innovation, based on the
analogy of other declensions (O.v.Hinüber, Überblick §316).
Comparing the close connection between the various Vedic dialects and the
Prākṛts (see below), it is surprising that the Madhyadeśa form -aiḥ is not
more prominent than its survival in a few limited cases of Pāli. It might seem,
therefore, that MIA -ehi is a new formation.
On the other hand, the Pañcāla innovation of gen. fem. -ai (see below) was
not accepted into Pāli, the representative of early Western Middle Indo-
Aryan. It may be, therefore, that in the case of some nominal endings, the
older state of affairs was preserved by the Pkt.s, i.e., instr. pl. -ebhiḥ (RV
stage) and gen.fem. in -āyāḥ (RV & Mantra stage). If this is true, then the
"first wave of immigration" had already spread this usage to all the areas of
Northern India settled by IA speakers; this usage, then, was already too
established in popular speech to be replaced by Kuru (MS, KS) or
282
Cf. Delbr., Syntax p.214. Note that stories often begin with devā vai..., but that AB has
te devāḥ... 3.22, 3.26., 3.27. Note as well the change from a RV usage devāḥ > devatāḥ; cf.
K.Hoffmann, Aufs., p.213.
283
Cf. C.Caillat, in: Sprache der alt.buddh.Überl., ed. H.Bechert; O.v. Hinüber, rev. of
SÄBÜ in IF 88, 107 sqq.; - for the Mbhār., cf. Holzmann, Gramm. aus dem Mahābhārata,
and K. Meenakshi, Epic Syntax, Delhi 1983. Note the frequent cases in PS and ŚS of taṃ
tvā "[To] you, as such (and such) a person ...," usually misunderstood in the ed. of PSOr.
284
O.v. Hinüber, Überblick, p. 145 § 316
105
The gen. fem. in -ai, known from a number of Vedic texts, can be found in its
Middle Ind. equivalent286 in the oblique form of the -ā stems: Māharāṣṭrī -āe
(also ai, āa,), Pkt. -āe (also- ai,-e),287 Aśoka -āya, (NW and E -āye)288, Pāli -
āya;
If one compares this with the evidence from the Vedic texts, it is obvious that
the geographical location of the following Vedic schools and the Pkt. dialects
agree. Śaurasenī, with -āe, agrees with the Taitt./Kauṣ. form -ai, and Māh. -
āe (also -āi, -e) agrees with Jaim. -ai. (Note, however, that Māh. also has -ai
and -āa, which are regarded as metrical variants only; see O.v.Hinüber,
Überblick § 80).
It seems that the Madhyadeśa innovation (TS, etc. -ai) subsequently spread
to all nooks and corners of the subcontinent where IA was spoken. Or, in
other words, the Middle Indic innovation is first seen in the Madhyadeśa texts
like TS, etc., and then is accepted into other Vedic dialects due to the prestige
of the (Kuru-) Pañcāla Brāhmaṇical language. Note as well that the form
disappears from Epic (and Class.) Skt. which, in this case, continues the
Western Vedic dialects.
It has been mentioned above (§5.2) that the development of the tenses in late
Vedic and early MIA foreshadows a complete restructuring of the tense
system in MIA.289
285
Note that there still remains a remnant case of -ebhiḥ in KapBr. = TB 3.12.3.3, a text
surviving as a fragment only; see Raghu Vira, ed. KpS, p.XIX (repr.), however this is in a
puronuvākya.
286
See Bloch/Master, Indo Aryan, p.135.
287
See O.v. Hinüber, Überblick, p.150 §334; §80
288
See also O.v. Hinüber, Überblick, § 299
289
In this section, I again quote from my article on the origin of the frame story, in the Fs.
U. Schneider.
106
In this case, the Eastern feature (typical is ŚBM, AB 6-8), with a two-fold
opposition aorist : perfect(/impf.), is fused with the Central/Western one (TB,
AB 1-5) with a retention of the older three-fold opposition aorist : imperfect :
perfect, and the Southern one (JB) with a new, two-fold opposition aor. :
perf.(/impf.).
In late Vedic, a restructuring of the use of the past tenses seems to have
taken place (see above §5.2); a distinction is made between the value of the
augmented forms (impf., aor., conditional) in "pluperfect meaning"
(vorzeitig), and the unaugmented forms (perf., pres., future, subj., opt.,
imp.).290
In early Middle Indian, the perf. is found only in a few remnant forms of
Pāli; it has almost disappeared in the other languages. The impf. is extinct,
except for a few remnants in Pāli (like āsī < āsīt), which have been classified
with the aorists as preterite. The situation in Pāli, which developed from the
Buddhist Middle Indian in (0artly) the same area (Pañcāla/Ujjain, etc.) as the
lost Śāṭy. Br. and JB, is still comparable to that of the Jaim. texts. Instead of
an opposition (impf.)/perf. : aor., Pāli has almost no impf. left at all,291 and
rarely a perf. (and then only in the older text level, in the Gāthās;
O.v.Hinüber, Überblick, § 480). The normal past tense is the aorist. When
compared to even late Vedic, Pāli is one or two steps ahead. The survival of
the aor. in Pāli (more rarely in A.-Mg., and in a few cases in J.-Māh.) fits the
situation encountered in the Jaim. texts (JB and JUB) quite well, where the
aor. is the most prominent past tense (next to the perfect).292
One can imagine the following pattern of innovation for (part of) the area
where Pāli developed from Buddhist Middle Indian. As an example, the
Southern text, JB, which was (re)composed/ redacted on the basis of a lost
(Central, i.e., Pañcāla) Śāṭyāyana Br. in Avanti, Bundelkhand, Malva, lends
itself for comparison, as the other Central texts, TB/TĀ, KB are either more
conservative or do not show the same kind of development. JB uses the
imperfect tense to narrate events of a (long distant) past and also for those
290
Note that there is no functional distinction between augmented and unaugmented forms
in Pāli, but that this is a remnant of older forms regulated acc. to the length of the form and
its origin in one of the aor. types. The same uncertainty occurs in Pāli: On prohibitives /
injunctives with mā which employ augmented "injunctives" in Pāli and Epic, see C. Caillat,
Some idiosyncracies of language and style in Asoka's rock edicts at Girnar, in: Hinduismus
und Buddhismus, Fs. U. Schneider, ed. H. Falk, Freiburg, p.97 sq. with lit; cf. ann 132. Cf.
the restructuring of tenses in Young Avestan, see Kellens, Le Verbe Avestique, Wiesbaden
1984, p.431 sqq.
291
O.v. Hinüber, Überblick § 479.
292
Cf. O.v. Hinüber, Überblick, p.192 § 477-488, esp. § 478.
107
Once the new opposition-- (distant) past : near past-- had been established, a
situation could occur when a narrator goes on to tell in the aor.: "(and then)
this happened, and then that happened just now / a day before > at any time
before now." Constant use of this tense (aor.) must have lead to the use of the
aor. as a general past tense (preterite) and to the disappearance of the perfect
in Pāli and other early Middle Indian dialects.293
The gradual disappearance of the subjunctive in Vedic and its survival in the
1st forms of the imperative paradigm have been studied by L.Renou, in his
Monographies sanskrites, Vol. I, Paris 1937. He concludes (p.43) that in
Vedic prose, the subj. was an archaic remnant, a fossil (une formation figée)
which was necessarily on its way out. He continues that its reappearance,
more apparent than real, in the late Br., has no "chronologic" value,294 in this
case, a correct evaluation. His materials indicate the beginning of the gradual
decline of the subj. in the Mantras of the YV (TS, see p.15 sq.).
Saṃhitā prose contains a fair number of cases. TS has 118 cases (of which
1st pers. = 56x, and thus should be disregarded; see p.36). The decline is
more pronounced in the Brāhmaṇas; from 8-9% forms in the old Saṃh.s, the
usage of the subj. falls to 1.5% in AB (78 cases), viz. 2.5% in KB (67x,
mantras, saṃpraiṣa, gātha always excluded). This agrees with an
accelerating simplification of its syntactical usage (p.16 sq.). Notably, the 2nd
(AB 2x, KB 0) and also the 3rd persons (AB 11, KB 7) become rare (p.20).
The 2nd person was probably already regarded as archaic by the authors of
AB (p.21); in ŚB, it is found only in narrative portions. Another Western Br.,
PB, has 27 cases, of which 16 are in the 3rd pers. and 2 in the 2nd (p.37). The
late AA has 20 cases, including 2 in the Sūtra-like book 5 (p.39).
293
See Geiger, Pāli, § 120, 158- 171; in § 162, he is misinformed about the Vedic impf.; cf. C.
Caillat, Pour une nouvelle grammaire du Pāli, Ist. di Indol. d. Univ. di Torino, Conference
IV, Torino 1970; O.v. Hinüber, MSS 36, pp. 39. (cf. also MSS 32, p.65 sqq., KZ 96, p.30
sqq.).
294
One of his favourite, although generally too global, opinions regarding linguistic
variation in the Saṃh. and Br. texts; cf. above ann. 12, which is shared by Caland and
Minard. They all regard such variations as a simple matter of style, not taking into account
the geographical spread and little of the relative chronology of the texts.
108
BŚS, apparently including the Br. portions, has but a few cases: 4 in the 2nd
pers., 3 in the 3rd. Similarly, JŚS has just one case. The Up.s, too, exhibit but
a few instances: AitU 1, ChU 3, BAU more (Renou, p. 40 §54).295
Notably, the Western books ŚB 6-10 again have fewer instances of the subj.
than the other books (p. 37, §47, cf. Minard Enigmes I, §461b), which agrees
with the decline of the subj. in the West and the Centre (AB, KB, PB); ŚBMo
has 385 cases, ŚBMw 155, ŚBMn,a 114. If the occurrences of the subj.
together with nét are calculated, we get a similar result: ŚBMo 97, ŚBMw 39,
ŚBMn,a,u 29 (Renou, p. 38, 25% of the cases). Note the late hyper-
characterised forms: -ā- as a subjunctive marker, even of cons. stem verbs.296
TS 118 (62)
295
See Liebich, Pāṇini, p. 28, 67 and Fürst, Sprachgebr. der Up. p. 19, 59.
296
See Renou, Mon. Skt. I, p. 5, §7-8. Note that hyperchar. -ā- is found first, with some
cases in RVKh, YV Mantra and notably in ŚS, but is less common in the Br. other than ŚB;
Renou notes these forms, in increasing order of occurrence in: TS, AB, PB, JB, VādhB., and
esp. ŚB (p.39); (-ā- occurs also in VādhB).
109
The strong occurrences in ŚB and JB lead us to expect that the subj. would
survive in the early Pkts. and in Pāli (based as it is on a W./S. language
comparable to JB with some older Eastern words and forms, comparable to
ŚB). However, all MIA languages present a much more advanced state of
affairs in the Aśoka inscriptions; remnants and new formations of this verbal
category are found only in certain inscriptions, and then (as expected) only in
Eastern ones: Sarnath huvāti, nikhipātha, see O.v.Hinüber, Überblick, p.172
§413.
However, the supposedly Western forms of Pāli as the representative of
Buddh. Middle Indian (forms like garahāsi) must be explained differently.
Thus the last flowering of this verbal category in ŚB must have been removed
from the early MIA of Aśoka, Pāli, and the other Pkt.s by a wide margin of
time, another confirmation, incidentally, of what has been noted above about
the respective dates of some late Vedic texts like BAU, and the early MIA
texts, and also of Pāṇini who still teaches the subj., though his language is
particularly conservative (see below, on -eṣma, §9.7).
The Vedic forms in -eṣma have their counterparts in the Prākṛts298 as well: -
eṃha is attested in Mg., Śaur. - but not A.Mg! -. In Pāli, some forms (-
aṃhase, -oṃhase), which are found exclusively in SE Asian MSS, may have
297
see K.Hoffmann, Aufs. 465-74.
298
See O.v. Hinüber, Überblick, p. 177.
110
been derived from the precative (or the injunctive; see O.v.Hinüber,
Überblick, p.178 §433).
It is very illustrative to note here that the extreme North-West, i.,e Pāṇini's
language has n o t accepted the Kuru-Pañcāla innovation.299
Pāṇ. still teaches -yāsma in dāyāsma, etc., and therefore stands on the level of
the RV. This is another indication of the extreme conservatism of the
Northern educated speech, cf. KB on Northern speech, quoted above.
In classical Skt. only one infinitive survived, the one in -tum; this agrees
with the major infinitive form of the Prākṛts: Pāli -tum, Śaur.-dum, J.-Māh. -
um 300; it is, however, rarely found in A.Mg.301
In the Veda (and to some extent in Middle Indian) there still was a great
variety of inf. endings (-e,-dhyai,-as, -am,-toḥ, -tave, tavai, etc.), of which the
one in -tos has been studied above, (cf. § 5.4). The dative variety -tavái,
(which developed from the earlier -tave, note the accent)302 is found in a
number of texts,303 quoted here according to Renou, Mon.Skt. and Delbrück,
Syntax; (the numbers followed by ? indicate my uncertainty as to the
inclusion intended by Renou of all cases found in the texts).
__________________________________________________________________
RV 25 ŚS 3?
MSp 14 TSp 2?
ŚBK more ŚBM 37
KSp 8 frequent
than M304 ŚBMo 26305 (16)
299
See already K.Hoffmann, p.470: "indication for dialect differences in the Brahmanical
caste language.
300
See O.V. Hinüber, Überblick §497, p. 198.
301
Cf. also Renou, Mon.Skt. II §39 sqq., cf. Minard, Trois énigm. I § 119c.
302
From -tave-vái, see Thurneysen, Mél.Saussure, p. 233 sqq.
303
cf. already Brunnhofer, Beitr.z. Kunde d. idg. Sprachen, X, p.234-266; Delbrück, Ai.
Syntax, p. 427 sqq., Oertel, Journal of Ved. Stud. I, p. 141 sqq.; Renou, Mon. Skt.2, esp. p.
26 sqq.= §26 sqq.; Oertel, KZ 65, p.71 (īśvara- + °tavai).
304
See Caland, ed. ŚBK p.47, 74, 85, Oertel ZII 5, p. 111.
305
The numbers in brackets are those of Brunnhofer, whose facts Renou, p. 37, calls "ne
sont rien moins que sûres".
111
AB 4? KN 1 ? ŚBMw 1 ( 2) 306
JBc 4 ? ŚBMn 5 ( 4)
JUB 1 ŚBMa 0 ( 0)
Vādhb 10 ŚBMu 5 ( 2)
PB 1
AA 0
Sūtra forms < Br., Saṃh.
---------------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------------
RV 1 ŚS 1
KS 1
MS 3
AB 0 TB 2
KB 1 VādhB 7 ŚBKo 6 ŚBMo 17308
JB 6 ŚBMw 5
JUB 2 ŚBKn 2 ŚBMn 1
It can be seen that both sets of figures, those of Renou and of Gippert do not
oppose each other but, to some extent, supplement each other. Together, they
provide an indication of the development of this infinitive in Saṃhitā and
Brāhmaṇa prose. The figures indicate that in late Br. prose, the inf. in -tavái
survived (relatively) well in the Eastern Centre and the East (even taking into
account the size of ŚB: contrast the large JB!).
.306
-tavai is missing in books 6-10: Renou, Mon Skt.II §38; Minard I 384d, except for one
case in book 9; K.Hoffmann, -tavai brū, Aufs. p.103 : JB,ŚB; cf. Oertel JVS 1,141.
307
See MSS 44, p.27 sqq.
308
Note that most cases of -tavai in ŚBMo do not have parallels in ŚBKo, see Gippert, MSS
44, p.50, ann.5; the opposite, -tavai in K but other forms in M, is rare.
309
Gippert, p. 47 sqq. , with ann. 74.
112
The phrase can also function as an important indicator for the relative date of
Pāli, JB, and BŚS as its continuant is found in Pāli: seyyathā, cf. taṃyathā
(O.v.Hinüber, Überblick, § 375, Minard, Enigmes I §119a.) -- In addition, it
may be mentioned that the tmesis which is prominent in late Vedic texts like
Baudh.,VādhB, is already rarely found in Pāli, see O.v.Hinüber, Überblick §
310
See C. Caillat, in: Die Sprache der ältesten buddh. Überlieferung (see ann. 270).
311
See Minard, Enigmes I, §119a,cf. Caland, ŚBK p.95.
312
See Bodewitz, transl. JB, p.120.
313
For BŚS, see Caland, Über BŚS, § 54 ; VādhB, Caland AO 1, p.9, etc.; this is close to
ŚBK, see Caland, ed. ŚBK p.98; further: Keith, ad AB 7.5, see tr.p.291, n.2 "a sign of
lateness". The collocation is important for a study of the relationships of late Br. texts: sa
ya, sa yathā/yatra etc. should be taken up in detail; -- cf. also the continuant of sentence
initial pronoun, tad in casu, in the Epic, meaning "then, now", cf. Kāṇva atha for Mādhy.
sa! -- For other similarities between Pāli and late Vedic, see author in Fs. U.Schneider (the
splitting of the head, the Gandharva/Yakṣa with an iron hammer in his hand etc.;
cf.foll.ann. on mahāśāla).
113
********
This section has indicated, I hope to have shown, that some of the
peculiarities noted for the various Vedic dialects reappear in the later
Prākṛts, and, in spite of the gap between Late Vedic and attested MIA, often
in the same areas as those of the preceding Vedic dialects. Now that the area
and the time-frame of the Vedic dialects are better known than before, the
various Middle Indian languages and dialects can be compared with the
earlier Vedic evidence much better than possible so far. It is also interesting
to note that forms which are found in the RV and have been compared
directly to the Pkt.s often have undergone a long development, visible in the
various Middle and Late Vedic dialects, until they reached the MIA stage; the
various levels of such developments and their geographical will be easier to
follow in the future.
§10. CONCLUSIONS:
* 2: These regional differences are not static both in time and space
(geographical distribution) but are dynamic: certain developments
spread from an original (often small) area to the surrounding territories
314
Note, however, Ved. tād, found only a few times in RV and in ŚS,PS; it appears in the
Pkts., see O.v.H, Überblick, p.160 §374 and in Aśoka, Khalsi Inscr.; - some additions of
common vocabulary are: mahāśāla- , of Brahmins with a large house(hold), occurs first at:
ŚB 10.6.1.1; 10.3.3.1; 10.6.1.6; Up.: ChU 5.11.1/3; 6.4.5 ; Chāg.24:6/7;Vādh.B 4.89:31, cf.
Pāli mahāsāla DhN 12.2, etc.; - or: śiraḥ + pra.han, śiraṃ + vi.pat in late Br., Up.s, cf. with
Pāli muddhā + phal, see now author, Fs.W.Rau.
114
This does not always occur only in an Eastern direction, as one might
think, (taking into account the history of settlement of N.India by IA
speakers), but also in other directions, as for example southwards, or
from the extreme East towards the West (§5.2)
Especially clear is the example of the diffusion of the gen. fem. in -ai
which originated in a small area of N. India (the Pañcāla land in Eastern
U.P.) and subsequently spread east- and southwards, - without affecting,
however, the West (the Kuru area) and the "North" (Panjab and the
E.Gandhāra area of Pāṇini's bhāṣā).
Innovations:
* khy > kś (Kuru )
* -ḍ- > ḷ- (Kuru)
* -jm- > ym (part of Kuru area)
* CuV > CV (Kuru)
* purūvasu- > puro° (part of Kuru area)
* parāyate, etc. > palā° (Kuru)
* preṅkha > pleṅkha (part of Kuru area)
Retentions:
* [ ch/śch ] retained (Kuru area: E. Panjab/W. Uttar Pradesh),
later > [ cch ]
Sandhi Innovations::
(2) In declination:
(3) In conjugation
Innovations:
* periphrastic aorist (-ām akar, etc.) (Kuru area, only Saṃh. prose)
* spread of narrative perfect (Prācya)
* renewed use of subjunctive,
hypercharacterised subj. (Prācya)
* renewed use of inf.-tavai (Prācya)
* late forms of duh : dugdhe (part of Kuru?, Prācya)
(4) Particles
(5) Style315
Innovations:
* use of sam.yat (part of Kuru area)
* devāsura- (Kuru and beyond)
* punarmṛtyu- (Prācya)
-
* retention of spṛdh (Kuru area and beyond)
* avoidance of pāpa- (Kuru area and beyond)
* renewed use of pāpa- (Prācya)
*********
When one excludes the Panjab, the main area of the Ṛgvedic tribes,
which itself shows many dialect variations,316 three centres of innovation
and subsequent diffusion of dialect features are observable:
315
Certain lacunae, however, have to be identified here: they occur, as has been mentioned
in the introduction, in those areas of Vedic grammar which are not directly approachable
for a detailed investigation so far, i.e. the occurrence of certain endings of verbs and nouns,
their statistical or geographical relations to each other, or the combinations of two or more
particles with each other. Only a data base system, built up with the help of a complete
electronic storage of the Vedic texts, will bring s i g n i f i c a n t progress.-- Other lacunae
concern certain forms and words - especially particles - which are not completely listed in
Vishva Bandhu's Vedic Word Concordance (VPK), most notably in the Sūtra volumes. It is
regrettable that an old Śrautasūtra like BŚS is recorded only so imperfectly.
316
Which are presently studied by S.Insler.
117
the early Brāḥmaṇa text, AB 1-5) and has a few other, typical
developments:
Among the innovations, the following are notable: khy > kś ; -ḍ- > ḷ-;
CuV > CV; parāyate > palā°; purūvasu- > puro°; preṅkha > pleṅkha;
introduction of: n.pl. -āḥ; nom. dual -au; n.pl.ntr. -āni; istr.pl. -aiḥ; - -
yuvam > -vyam; periphrastic aorist (-ām akar, etc.); vāva; use of
sam.yat; devāsura-.
The following innovations only affected a part of the Kuru area (but
often spread beyond it): -jm- > - ym- ; Sandhi -ān V- > a V-;
Some categories viz. words disappear or are on their way out in the Kuru
area: elimination of the older RV n.pl. -āsaḥ, dual -ā, istr.pl. -ebhiḥ;
decline of subjunctive; the injunctive disappears; decline of inf. -tavai;
decline of the use of the particle u.
(2) -- The Pañcāla land (Madhyadeśa, W.Uttar Pradesh) has its share of
innovations; they are later than those of the Kuru area and are
represented by texts like TS, TB, KB.
Sandhi Innovations:
Sandhi -o/au V- : various innovations in Pañcāla dial.
(3) -- The East, primarily Videha (N.Bihar) but to some extent also
Kosala (E.Uttar Pradesh, W.Bihar), are the late Vedic centre of major
innovations; the area is represented by ŚBM and ŚBK, the later part of
118
AB, i.e. AB 6-8, and BaudhŚS, perhaps also by PB if the text indeed got
its final redaction in the East.
Innovations:
tanacmi > tanakmi, etc. ; Sandhi -e/o a- > -a a (Śākalya); Sandhi -o/au V-
: various innovations in Prācya dial.; late forms of pronouns: nom.
vayām, āvām (part of rācya area); late forms of śīrṣan- made from śiras-
(part of the area); diffusion of the narrative perfect; renewed use of
subjunctive; hypercharacterised subj.; renewed use of inf.-tavai; late
forms of duh : dugdhe (part of the Prācya area); renewed use of u in
collocations; sa in sentence initial position; punarmṛtyu- ; renewed use of
pāpa-.
******
Vedic held sway over the Pañcālas for a long time, until it had to give
way to and subsequently was overshadowed by the one that had
developed among the Pañcālas themselves (most notable is the gen. fem.
in -ai). This dialect gained prominence in the late Saṃhitā and in the
early Brāhmaṇa period and strongly influenced the areas East and South
to it. The "Eastern dialect", perhaps best called Prācya in accordance
with Pāṇini and other early sources, emerged into prominence only
during the late Brāhmaṇa period. Interestingly, even the present version
of JB (i.e. the one superseding the originally Central Śāṭy.B) still
criticises (albeit indirectly) a king's son for speaking "like the
Easterners". Subsequently, some of the characteristics of the Prācya
dialect seem to have been strongly preferred, so strongly in fact that they
penetrated further and further westwards until they reached the Panjab
(but not Pāṇini's homeland) during the late Vedic period. Typical is the
diffusion of the narrative perfect which begon in this area. Other
peculiarities, like the use of certain groups of particles, too, quickly
spread westwards. Finally, the East is very important as the late Vedic
centre of redactional activity (Śākalya for the RV, "Yājñavalkya" for the
White YV, etc.)317
In addition to these major dialect areas, there are others about which
we do not know anything or only have some stray facts. These are those
of the
319
Cf. author, On Late Vedic pitch accent, forthc.
320
It is curious that a Western text, ŚB 9, raises this point. The Panjab was overrun by new
waves of immigration at the time of composition of this passage, cf., the role of the Salvas at
JB 2.208 who conquered Kurukṣetra, of the Malla (JB §198) and of the Vṛji (Pāṇini
4.2.131); see above.
321
See author, Persica IX,1980, p. 92.
121
the bhāṣika accent) is lost. Nor do we have the lost Paipp.Br. for part of
the AV. The list can be prolonged. Such lacunae can, in the future, be
worked away to some extent, step by step, if we succeed in filling in the
map of Vedic India with the help of the "lost schools" (cf. Bh. Ghosh:
Lost Br.).322
There remain, however, still a few more problems in the definition and
proper delineation of the Vedic dialects.
The various dialects mentioned just now will always have to be clearly
distinguished from local style, especially the style of a few Vedic śākhās
versus others. Examples have been mentioned and discussed above: the
use of sam.yat versus spṛdh, the new compound devāsurāḥ (to be treated
seperately), and other school mannerisms and peculiarities like súvar for
svàr, etc.
*****
It has been noted above that several peculiarities are not shared by the
various schools (śākhās) of all four Vedas of a particular dialect area, e.g.
that of the Pañcālas, but only some (or even only one) of them, while the
peculiarity infact transgresses the dialect area. In such cases the
peculiarity is often restricted to the schools of only one Veda (cf.above,
passim) and thus found in various dialect areas. It can be noticed that,
for example, both a Western and a Central Vedic school share the same
traits, while those belonging to another Veda in the same area disagree.
This necessitates, in future studies, a careful deliberation of several
factors. The setting provided by time (relative / absolute chronology) and
location (the homeland of a school viz. [part(s)] of a particular text) will
have to be distinguished from the influence of such factors as the
peculiarities belonging and restricted to one of the four Vedas only, or to
one or more of the neighbouring schools, viz. to those occupying the same
area as the text in question.
belonging to the Ṛgveda, and why only later otherwise, in BŚŚ, etc.? Is
the final redaction even of AB 1-5 so late? Was the text revised in the
East, by Śākalya or his school, during t(e late Br. period?
on an old Western SV Br. which used the impf. but probably received its
redaction in the East (note the extra-ordinary high number of perfects in
ChU, a late Kauth. text).
* The use of the infinitive in -toḥ shows some minor variations between
the various Vedas; it is strong in the YV (with the exeption of MS and the
Śāṇdilya books of ŚB), ranging from 140 to 240 %, and in the RV texts:
AB 289%,323 KB 160%. The SV is a little weaker: PB 194%, JB only
92%, JUB 113%. This involves both diachronic and areal features: If PB
has received its redaction in the East then its percentage agrees with that
of the other Eastern texts (ŚBM).
323
If the collection of the inf.-toṃ + īśvara-, made by Oertel, KZ 65, 1938, p. 55 sqq., is
taken as representative (see above § 5.4), AB 1-5 has 7 cases, AB 6-8 has 9; the length of the
two parts of the text is : 157 : 78 pp., 2/3 to 1/3, and thus the number of cases in ABn is
about 2 times higher than expected.
124
***
In future, when many more dialect features have been recorded and
compared and when a much closer grid of interlocking or overlapping
dialectal patterns has been established, we will probably be able to point
out even the origin of the spread of such developments limited to the
schools of one Veda like the Opt. in -īta of a- stem verbs. At present, we
can only wonder whether the origin of this feature is in the AB area (E.
Panjab) or in the KB area (U.P), and whether in the latte2 case, the
Central (Kauṣ.) peculiarity influenced at first only the later text level of
AB (6-8) and subsequently, at the time of its redaction, even the older
parts (AB 1-5): To solve this and similar problems, we will have to learn
more, first of all, about the final redaction of AB which took place in the
East and was executed either by Śākalya or someone else of his school.325
325
Note the same problem with regard to text variants already in the Mantras. This
indicates that right down from post-RV mantras, there existed a "mīmāṃsā" type activity
inside one Veda and that certain texts were changed right down to the Br.period. Only at
the time of the collection/ redaction of a text did the particular form of a Mantra become
"sacrosanct" and was 'pushed through' everywhere in a particular school, especially to
distinguish the text from that of other schools. Note, for example how ŚBM/ŚBK treat the
Caraka quotations quite differently, each text according to its own phonetic rules (see StII
8/9).
126
Since the beginning of this century, the idea of a first and a second wave
of immigration has been discussed, a series of "invasions" (better: a
gradual trickling in, by the movement of certain clans, and ultimately,
tribes), resulting in the outer band and the inner band of New Indo-
Aryan languages.
This is, of course, easily challengeable. Outer band features can have
their origin, e.g. in the Middle Ages; they represent remnants of an older
situation, but do not necesarily date back to the Vedic period, while the
"centre" (the later Hindustani/Hindi/Urdu) developed innovative,
unifying features which just did not reach the outlying regions.326
This again does not agree with the inner and outer band: The Gāndhāri,
Parśu327 and Āraṭṭa but also the Kāśī-Videha should belong to the outer
band while the Kuru-Pañcāla form the (innovative) Centre. Instead, the
text makes a differentiation between the peoples of the Panjab and the
326
A good example of such developments is Japan where the capital and thus the centre of
administration has shifted several times from the Western Kansai (Yamato, Kyoto area) to
the Eastern Kanto (Kamakura, Edo/Tokyo), and back. Innovations which developed
during one of these periods spread concentrically outwards, towards the Eastern and
Western ends of the archipelago, starting from of the capital of the time: they now form
multiple, overlaying patterns of various dialect features. A particular innovation sometimes
reached the ends of the archipelago, but as often, it did not. In such a situation, it naturally
helps to know where the development in question started, -something we still had to find
out for the Vedic period.
327
Regarded by some as Persians, see Cardona, Pāṇini, p. 276; cf. ann. 327,339.
127
Much has been made, in past decades, of the Vrātyas as the early
Eastern immigrants; now H. Falk's book 'Bruderschaft und Würfelspiel'
provides a better interpretation of their character, strictly derived from
the evidence of the texts themselves: The Vrātyas are poor, mostly
younger Brahmins and Kṣatriyas who in search of a "start capital" form
a dark, ominous sodality which demands ransom from the local well-
settled gṛhasthas and even from the kings.328
A closer reading of the texts yields more results for the still very hazy
picture of Vedic history: for example the fate of the Kurus, who have
been overcome by the Salvas (JB 2.206). ŚB and BAUK mention the
uncertain fate of the Pārikṣitas, the royal family of the Kurus: "where
has their glory gone?" Such sentences might, ultimately, provide the clue
for the prominence, in the later YV-Saṃhitā and the Brāhmaṇa period,
of the Pañcālas with their Taitt., Kauṣ., Śāṭy. schools; notable is the
prominence of Keśin Dārbhya and his successors in these texts.329
328
Note the story in BŚS 18.26, cf. H.Falk's transl. in Bruderschaft, p.55 sqq., about the
Vrātyas of the Kurus at the court of the Pañcāla king Keśin Dālbhya. The Kurus
apparently play the role of vrātyas for the Pañcālas (and vice versa?). Is a constant
dichotomy of society expressed by the loose union of the two tribes? Cf. situation as
reflected in religion: the devas and asuras are in constant conflict; note also JB 2,278-9
Keśin Dārbhya (Pañcāla king): his mother and his (maternal) uncle Ucchaiśravas, son of
Kuvaya, the King of Kurus (kauravya rājā): a clear case of intermarriage of the two royal
houses.
329
Cf. the fight of the Pañcālas with the Kuntis, see ann. 113, KS 26.9, end.
330
Note the intention of the story: Gotama Rahūgaṇa is otherwise known only as the author
of Ṛgvedic hymns. To make him the culture hero of the East is as conspicuous as the
128
The story, as told in ŚB, expressively mentions the role of the ritual fire,
called Agni Vaiśvānara in the legend, in making the marshy country of
the East arable and128acceptable for Brahmins. The Māthavas, about
whom nothing is known outside the ŚB, may be the Máthai of
Megasthenes who places them East of the Pazálai (Pañcāla), at the
confluence of the Erénnesis (Son) with the Ganges.331 While the
movement of some clans and their King Videgha from the River
Sarasvatī in Kurukṣetra to the East may coincide with the 'ritual
settlement' of Kosala(-Videha), this is not to be confused with the
wholesale movement of Vedic Śākhās, like the one of Kāṇva, Śāṇḍilya,
and Aitareyin eastwards, to Kosala and Videha.
A final stage is reached only in the Pāli texts: Suddenly, we do not read
about the Kosala-Videha but about a separate Kosala kingdom and a
large Vajji (= Vṛjji) confederation which includes tribes like the
Licchavi, Naya, and the Videha. West of them live the Śākya, Bulinda,
Malla, Moriya and Kalāma. None of these tribes, with the exception of
the Videha and the Malla, is known from the Vedic texts. Interestingly,
JB still locates the Malla in the desert (Tharr, see JB §198; cf. the Malloi
of the Greeks in S. Panjab). Equally, the Vṛji of Pāṇini 4.2.131 are
mentioned together with the Madra; in all probability, they still were
inhabitants of the Panjab at the time. What we see here is, I believe, the
last wave of immigration which overran Northern India in Vedic time
and which came to an end in its easternmost part (at that time), in N.
Bihar. Note the somewhat unusual origin of the Śākya from a marriage
of the sons of King Okkāka with their sisters and compare that even in
Manu, the Nicchavi (sic) still are regarded as only half-orthoprax. This
wave of immigration from the West is fairly late, as no Vedic text
contains any hint of it and it is only Pāṇini (4.2.131 Vṛji, Vṛjika) and the
Pāli canon which provide a clue to it.
sudden replacement of whole schools, notably the Aitareyins, Śāṇdilya, Śākala, and Kāṇvas
into the East.- Cf. the RV name Namin Sāpya as King of Videha at PB 25.10.17,
interestingly described as making a pilgrimage to Kurukṣetra, the holy land of the Veda
and the home of Gotama Rahūgaṇa and Videgha Māthava who in ŚB are the prototypes of
the eastward movement of Vedic orthopraxy.
331
See Arrian, Indikē 4.5 and cf. the commentary by O.v. Hinüber, in: Arrian, Der
Alexanderzug. Indische Geschichte, hg. und übers. von G. Wirth u. O.v.H., München u.
Zürich (Artemis) 1985, p. 1095; cf. also author, Fs. Eggermont.
129
______________________________________________________________
B.C.
1750 - OCP/late Harrapan/
Gandhāra Grave culture
Earliest immigration?
Kāfirī dac : Ved. daśa, Ir. dasa
muṣ/muš : Ved. muṣ
ašpa@ : Ved. aśva, Ir. aspa/asa
----------------------------------------------------------------------------------------------
FIRST I.-A. IMMIGRATION into the Panjab
----------------------------------------------------------------------------------------------
ca. NEW WAVE of IMMIGRATION: ling.closer to YAv.
1180 B.C.
BRW centre at Kurukṣetra:
(iron!) composition of
YV Mantras, PS, AVŚ, Kuntāpa hymns; SV
332
The dates have been taken from archeological finds, acc. to the standard works of B. and
R. Allchin, The rise of civilisation in India and Pakistan, Cambridge 1982, and W.A.
Fairservis Jr., The roots of ancient India, 2nd ed. Chicago and London 1975; see below §
10.4.
130
r is preferred
viśva- > sarva-
progressive disappearance
of injunctive
______________________________________________________________
ca. 900-
PGW Kurukṣetra, U.P.
MS,KS TS
loss of injunctive,
introd. of periphr. aor.
khyā> kśā
Cuv > Cv; Ciy > CyV
khalu
-yuvam > vyam
-jm- > -ym
diffusion of RV inf. -toḥ
decline of inf.-tavai
sam.yat
----------------------------------------------------------------------------------------------
NEW IMMIGRATION:
into Kurukṣetra:
Salvas
Madhyadeśa (Pañcāla) innov.:
--------------------------
gen. fem. -ai
ch/śch > cch
Sandhi innov. of -o/au + V
súvar, uv (eva) Kāṇva, Śāṇḍilya,
further diffusion of Aitareyin,
Kuru peculiarities move to the East
131
Post-Saṃhitā developments:
Aitareyins (Aśvala=
hotṛ of Janaka; Śākala,
Uddālaka in Videha
Paṇini:
archaic precative
knows of Eastern grammarians
and E. forms but excludes Central
and E. Veda texts
-----------------------------
Yugandhara,
Salvi tribe immigration
in the Kuru/ into the East:
Matsya terr. Śākya, Licchavi,
Malla, etc.
333
Is the name of the tribe, Salva, to be compared with Iran. Sauruua (=Ved. Śarva, a
Western, Bāhīka name of Agni/Śiva)? Just as Bharata is a name of Rudra in RV, Śarva
could be taken from the (Ir.) name of the tribe as well.
334
For the post-RV period cf. the role of the Adhvaryus in the development of the Śrauta
ritual and their apologetic myth (the cutting off, by Indra, of Dadhyañc's head, viz. that of
the head of the sacrifice, see Author, Fs. W. Rau, ann.104); a comparison of RV hymns
taken over into the other Vedas establishes that there must have been a period when there
was an Ur-YV, Ur-SV, Ur-AV (cf. already Oldenberg, Prolegomena, and for non-metrical
texts, author, WZKS 24, p.22 sqq., 76 sqq.); these texts were first assembled and composed
in the Kuru country (cf. the role of Kurukṣetra as devayajana, and cf. BEI 2, Sur la voie du
ciel); from there, they spread eastwards, being both gradually changed or consciously
altered (out of the wish to establish separate identities, because of 'political' motives), until
they reached their Pañcāla (TS, Śāṭy, KB, AVŚ), Kosala (ŚBK) or Videha (ŚBM, Bhall.,
Eastern RV) forms. The centre of ritual activity spread eastwards; note the lack of
complete Br. or Ār. texts of the Kaṭhas and Maitr. schools (though both schools developed
Sūtras, probably in their new territories, in Gujarat viz.the East: Prācya-Kaṭha) ; --- The
spread eastwards and the role of the Kāṇvas in the colonisation of the East has to be
studied, (Videgha Māthava from the Sarasvatī, together with a RV author, Gotama
Rāhūgaṇa!); the East was a melting pot of earlier IA and non-IA settlers and took over the
Vedic śrauta orthopraxy only at a late stage (during the Br. period); the Kāṇvas were in
competition with the Central Taitt. school and others, like the ancient school of the
Carakas; this process is, as it were, personified by the adoption of Taitt. ritual by the first
Sūtrakāra, Kāṇva Bodhāyana, who probably wanted to codify ritual "for once and all".
The strange correspondences of the Mādhy. texts with those of the Maitr. have to be
investigated and have to be compared to the same situation relating the Kāṇvas with the
West (Kaṭhas/Carakas?), or the Taitt. with the Kaṭhas, etc. -- Furthermore, the spread
eastwards of the Śāṇḍilya school and its texts (ŚBM 6-10) has to be investigated. They were
imported from the West (at least, from the Central area, cf. the Saṇḍilla country in the
Jaina texts, N. of Benares). Other Western schools were - in an apparently politically
motivated move by (Mahā-) Janaka - imported into the East: The Aitareyins (AB 6-8), and
most probably also Śākalya who redacted his RV in the East, in competition with an earlier
Eastern RV; composition of ĀŚS in the East; -- the probable move of some Kauthumas (cf.
the RV name Namin Sāpya, already a Videha king at PB 25.10.17, cf. the bhāṣika accent of
PB, - just like ŚB - which is lost now in PB) and some Kaṭhas, the later Prācya Kaṭha of the
133
It has been mentioned above that the three centres of innovation coincide
more or less with the territory of the Kurus, Pañcālas and (Kosala-
)Videhas. This is , from the point of view of political history, not too
surprising. It is well known from dialect studies that political boundaries
often coincide with dialect boundaries, cf. e.g. the curious case of Germ.
dial. schlīn, schlën 'schlagen' which straddles the Middle Rhine valley
and coincides with the borders of the old principality (bisdom) of Trier.
All surrounding dialects have schlān or schlön.335
That the territory of Vedic dialects covers that of the political units, like
that of the Kurus, etc., is thus not surprising. In fact, the coincidene of
tribal and dialect territory and the spread of certain dialect peculiarities
agrees well with the political development as we know it on the basis of
Car.Vy., to the East; -- the problem of the Brahmanisation of the "foreign " territory of
Magadha (and Aṅga), and the Southern spread of the Mādhyandinas; the late/post-Vedic
immigration of new tribes into the East (Malla, Licchavi, Śākya, etc.) and the possibility of
an Iranian element among them (note the river names from E. Iran/Afghanistan like
Gomatī, Sarayū in the East, and cf. the Śākyas and their marriage customs, further Balhika
in ŚB, etc.). --- The gaps in the late Vedic geographical attestation of the śākhās can now be
closed to some extent: The SV of the East is unknown but must have been that of the
Bhāllavins as their Br. was recited with the bhāṣika accent, like ŚB. Perhaps this was a sub-
school of the Kauthumas, cf. the notice in Bhāṣika Sutra that PB was transmitted with
bhāṣika accent as well (see ed.Kielhorn, Ind. Stud.10, p.421). By the time of Śabara (Mīm.
Sūtra), the accented tradition was lost already. - Equally, the SV of the Central area was
that of the pre-Jaim. Śāṭyāyanins; the question of the other Vedas in the Jaim. territory is
open: probably, they were partly Maitr., partly Taitt. (note a Taitt. quotation in the late
Pāli texts, the Jātakas), cf. ann. 345; it can now be attempted to close the gaps between the
late Vedic spread of schools and their earliest attestation on copper plate grants (cf. author,
Beitr. zur Südasienforschung, 104).
335
See W.König, dtv-Atlas zur Deutschen Sprache, München 1978, p.142
134
336
This is, more or less, also what H.Oldenberg in the introduction to his book, Buddha,
sein Leben, seine Lehre, seine Gemeinde, extracted from the Vedic and Pāli sources more
than 80 years ago. (His comments about the negative influence of the climate on the
development of Indian mind and the attitude towards an active life are, however, better
forgotten, although this is a much loved topic in contemporary Indian apologeticism.
337
B.and R. Allchin, and W.A. Fairservis.
135
If we tentatively align these texts with the NBP culture, which in terms
of the more advanced material culture of the late Vedic period should not
present difficulties, then a problem arises concerning the absolute dates
of the later Brāhmaṇa texts (and of the early Up.s). These have generally
been aligned with the age of the Buddha, who is usually believed to have
lived from 563-483 B.C. However, H. Bechert recently has cast some
doubt on this date: the Buddha might have lived ca. 100 years later.338 In
fact, as has been pointed out above, the Pāli texts, which were written
down only in the 1st century B.C., but were composed several centuries
earlier, reflect a much later stage in the cultural and political history
than even the late Vedic texts (like the Upaniṣads): in the Pāli texts (like
Dīgha Nikāya) even Magadha and Aṅga are Brahmanical territory, while
the Veda has only a single case (at KA 7.14) where a Brahmin lives in
Magadha, a generally avoided and despised country. Note that there is
no mention of towns in the Vedic texts, nor of writing. Though this may
due to the cultural tendency of the Brahmins who have no use for
writing, as they learnt all their - mostly secret - Vedic texts by heart and
also could preserve their ritual purity better in a village than in a busy
town, both items cannot simply be dismissed. A date of ca. 500 B.C. for
early Up.s ( like BAU, ChU), BŚS, and some late Br. texts like VādhB,339
and late parts of ŚB, AB 6-8 does not seem, to my mind, impossible, - at
least at the present state of our knowledge.
338
See the summary by O.v. Hinüber, Überblick, § 6.
339
Note that BŚS 18.44 perhaps intends the Persians with its term Parśu (see ann. 334, 323);
however, also other tribes of similar names are attested in the area of E.Afghanistan (Gr.
Paryetai, etc.); for this period, cf. H.Kulke, The historical background of Indian's axial age,
forthc.
136
Northern India differ from those of the Vedic texts:340 Again, the Pāli
texts seem to reflect a later stage in the political development. 341
The Pāli texts,342 indeed, know of the complete Vedic corpus:343 the
three Vedas and their transmitters (tiṇṇaṃ vedānaṃ pāragū, DN 1.88;
tevijja Thag 1248, Thīg 65; mantrapāragū Aṅg.N. I p.163,166: 58,59,
etc.), and even the the various ancillary texts like etymology, grammar,
etc. (DN 2.13, MN 2.91.93, Bv.38). The Vedic texts apparently had alredy
been redacted and collected: "the old text of the mantras.." (porāṇaṃ
mantapadaṃ itīhītiha paraṃparāya piṭakasampadāya MN 2, p.169 : 95);
apparently the collection of mantras is called piṭaka in analogy to the the
Buddhist texts. A Padapāṭha seems to have been in existence, as padaka
belongs to the standard description of Brahmins (DN 1.88, MN 2.133, Ap
502, etc.). The names of the Vedas occur in later texts, at MA 3.362, DA
247, AA 2.61 SnA 447: iru-344, yaju-, sāma-(b)beda; otherwise, some
Veda schools are mentioned at DN 1.237.10-18, Addharīya (~
Ādhvaryava, adhvaryu-, YV), Tittirīya (Taittirīya), Chandoka
(Chāndog(y)ā, SV), and Bavharija (Bāhvṛca-, RV) are known; Titt.
brahmacāriya occurs at Vin. 2.162, and two separate Brahmins called
Assalāyana (Āśvalayana) are found at MN 2.147.9-157.17, Pj 2.372.25,
406.26, Ap 480.17. A late text even quotes, almost verbatim, a passage345
from TS.346136
If the late Br. texts are compared with the earlier Brāhmaṇas which are
limited to the Kuru-Pañcāla area, or better, with the YV Saṃhitā texts
340
The only Vedic text that mentions 16 kingdoms, however, without names, is VādhB., see
StII 1, p. 75 sqq.
341
Note that king Ajātaśatru occurs in ŚB and VādhB but as a king of the Kāśīs viz., the
Kurus. Ajātasattu of Magadha is still unknown; cf. also Brahmadatta Prāsenajita of
Kosala, JB §115, with the Kosala king Pasenadi in Pāli; apparently both names were
common in late Vedic as well as at the time of the Buddha (for more ling. correspondences,
see ann. 314, 359). W. Rau, Altertumskunde, p.21, ann.2 regards the Pāli texts as much
later, because of the development of (material) culture they indicate, and consequently
wants to date all the Vedic texts that precede them later than usually thought.
342
Of course, the problem of the redaction of the Pāli canon, at a later date, remains. Yet
the testimony about Vedic schools , at inconspicuous places in the Pāli canon, is valuable.
343
For many of the following passages, see Hillebrandt, Kl. Schr. p. 309 sqq.; cf. now Pali
Tip. Concordance for more examples.
344
There is some S.Indian influence in these late texts; irubbeda in Mil. under has received
its shape under the influence of the form of the word, see O.v. Hinüber, Überblick, § 126
345
Cf. TS 1.1.9.1, quoted in Jātaka VI,212,11*, see O.v. Hinüber, Überblick, p.131 §275.
346
Note also that both late Vedic and Pāli texts contain similar concepts, like the case of the
shattered head (cf. Fs.W.Rau), and often express them with more or less the same words
(e.g. the Gandharva, Ṛtu, Yakkha kūṭahasta, see ann. 314,359.
137
(MS,KS,TS) it immediately appears that these texts know only the area
from E.Panjab to Allahabad / Benares.347 A look at the material culture
of the texts is facilitated by W. Rau's investigations: They provide
evidence for: ayas ("Nutzmetall", copper (bronze which is unusual in
India; not, as usually translated:"iron"), but also iron (śyāma- ayas-),
wattle (and daub) huts, which were easily removable; cultivation of rice,
barley, etc., cattle (horse, etc.), in short, a generally still very simple
material culture.
However, apart from the general agreement of the testimony of the texts
on material culture and archeological finds, there is one more surprising
correspondence. The area of the YV Saṃhitās and of Paippalāda-AV is:
E. Panjab, Kurukṣetra, Haryana, W. Uttar Pradesh up to
Allahabad/Benares (Kāśī). This is precisely the area that is covered by
PGW culture...349 To my mind, the coincidence is too great to be
accidental. If future research will provide more material to support the
identification of the PGW civilisation with that of the post-Mantra and
pre-Late Br. period of texts and of the Kuru-Pañcāla tribes, this will be
of major importance both for dating the texts (see below) and for a
general correlation of Vedic studies with archeology, as well as for the
further interpretation by archeologists of this culture.
347
See Fel.Vol. Eggermont; the only exception is ŚS, with Aṅga which replaces Kāśī of PS.
348
W.Rau is a little more optimistic: he thinks it is possible to compare sacral pottery with
everyday pottery objects which were already thrown on a wheel but regarded as demonic
(asura-like); see also W.Rau, Altertumskunde, p.41 sq.
349
I should like to underline, that I began this whole investigation with no thoughts spent on
centres of innovation, or on political centres, not to speak of archeological regions. Yet the
outcome indicates several centres of innovation which happen coincide with those of the
political centres/tribal kingdoms, -- and with those of archeologically attested cultures, at
least as far as we know them at this moment.- Cf. also W. Rau who, on other grounds, came
to a similar conclusion: Altertumskunde, p.48: "The so-called.../OCP/...agrees best,
according to the area of spread, technical condition, and approximate age, with the pottery
described in the Vedic texts." (my transl.)
138
The historical facts, as gleaned from the texts, agree: The earliest
centre of political power was in the West, in Kurukṣetra itself. Ever
since the late RV, it has been the "holy land" of the Brahmins. This is the
place where even the gods usually sacrifice (devayajana), and for a good
reason: here is the 'centre of Heaven and the Earth'351 , and the political
centre (at Āsandīvant352) of the Bharata/Kuru tribe353 which dominated
the late Ṛgvedic and post-Ṛgvedic period.354
______________________________________________________________
|
WEST CENTRE | EAST
_________________________________________|_____________________
1750- OCP: Ocre coloured/ |
350
See author, E.Iran and the AV, Persica X. - Therefore, I cannot agree with W. Rau when
he says, Altertumskunde, p. 19, that the beginning of the Vedic period might perhaps be
suppressed below the date of 1000 B.C. The introduction of iron alone (which is not yet
mentioned in RV!) but appears in India already in the 12th cent. B.C. and fittingly, also in
the second oldest Vedic text, the AV, is too early for the date proposed by W.Rau (even if a
late redaction of AV is taken into account).
351
JUB 4.26.12, and the unpubl. VādhPiS, see BEI 2, p.223, with ann. 74
352
Note the meaning of this geographical term 'having a/the throne'
353
For details, cf. my article on the Kuru Realm, forthc.
354
Some further speculation may be added, if the other copper hoard cultures in the South
are taken into account: Do they represent the earliest forerays of IA speakers (or of tribes
closely related to them in culture not in language!), which petered out, without much effect?
Only after the consolidation of IA culture in the Kuru-Pañcāla area, did the spread
southwards start again, as ŚB 2.3.22 asserts: Naḍa Naiṣadha (sic) is said to carry Yama
(death) (further) South, day by day. Cf., however, the Aśoka edicts in Pkt. in S. Karṇātaka:
whom does he want to address there in Pkt.in this area? Note that he uses Greek in a Greek
area (Kandahar).
139
Late Harappan |
and |
RV various smaller | (Neolithic)
cultures (Grey Ware, (1250 Kausambi: |
Gandhāra grave culture, OCP) |
etc.) |
(no iron) |
______________________________________________________________
1180 BRW: Black & Red ware |
KURU ( with iron!) |
1st centre Mantras, AV, etc. | (chalcolithic)
______________________________________________________________
900- PGW: Painted Gray Ware 900- Black & Red
MS,KS Ware
PAÑCĀLA
2nd centre (& Grey Ware)
(iron)
TS
(Kaus. 750 B.C.) No PGW!
(Kosala- ) -VIDEHA
500- B.C. NBP: Northern Black Polished Ware spreads over N.India
______________________________________________________________
It goes without saying that even the following tentative dating, based on
the present knowledge in archeology, has some major consequences for
the interpretation of the Post-Ṛgvedic texts: the various texts receive, for
the first time, a position both in time and space. Future research will
show whether each of the following statements is correct or will have to
be adjusted. However, I have no doubt that the general scheme of text
layers and geographical attributions is correct. The preceding chapters
have shown, time and again, that the spatial and temporal relationships
of the Vedic texts agree with the various sorts of evidence brought into
play. In the following table, this is merely extended to include firm dates
which are based both on the evidence of the material culture mentioned
in the texts and on the geographical distribution of the texts and schools.
In the table, some more details regarding the various Vedic texts and
schools, their formation and relative position, and the relevant layers of
some texts as well as some important historical and cultural data have
been added.
______________________________________________________________
|
Panjab | WEST | CENTRE | EAST
| | & |
| | South |
______________________________________________________________
Late Harappan/ | |
1750- OCP: Ocre coloured pot. | |
| |
and various smaller | | (Neolithic)
cultures (Grey Ware, |1250 Kausambi: |
Gandhāra grave culture, | OCP |
etc.) | |
metals: copper, no iron | |
no rice | |
Ṛgveda | |
composition | |
& first fam. collections | |
| |
immigration to the Panjab, | |
of OIA speakers, | |
in several waves, | |
the latest = Bharata | NB. rice is |
141
| early in the |
RV hymns composed | Malwa culture |
| |
gold, silver, ayas (copper) | Kīkaṭa, S. of
mentioned, no iron; no rice | Kurukṣetra =
| later Niṣāda?
______________________________________________________________
______________________________________________________________
- 500 B.C.
(Kosala-) -VIDEHA
3rd centre
KaṭhB KB
JB BŚS ŚBKo ŚBMo
ŚBMw
- 500 B.C.
ABn
PB KA (PB?) AA, AitU
AA 5
MU ĀśvŚS
16 kingdoms are known (VādhB,Pāli)
prominent are: Kosala, Kāśī, Videha,
Aṅga, Magadha?, and:
the older kingdoms of Kuru-Pañcāla, Matsya;
also: Vaidarbha, Trigarta, Salva, Madra,
Gandhāri, Āraṭṭa, Parśu, Sindhu-Sauvīra;
non-IA S.E. tribes like Puṇḍra, Kaliṅga,
Andhra.
355
For Mahājanaka, see Franke Kl.Schr., 379. -- For accents see O.v.H. Überblick,p.90
§159; cf. Bronkhorst, Two traditions, Stuttgart 1986, p.111. - Note that his examples are
from ŚB/BAU! But there are even later texts with accents: Some late RVKh portions (Śrī
Sūkta, even found in Nepalese Buddhist texts!), the Vaiṣṇava stanzas of Vaikh Mtr.Pr., the
last, very late sentences of ŚB itself about Yājñavalkya; (accent was, on a scholarly basis,
used even much later: some acc. Pāṇinean MSS., a Śikṣā was used to accent an AV MS even
some 300 years ago, see introd.to the AV ed. by Śaṅkar Pāṇḍuraṅ Pt.). Therefore, BAU,
when taken isolatedly and compared to the Buddh. rules about chandas language, in order
to establish a contemporanity of its period with Buddh. texts, has no value: the Buddh. rule
could equally mean late Vedic texts, like Vaikh.Mtr.Pr.-- Note also P.Thieme's observation
that Kātyāyana uses the word ādy-udātta- to express a pitch and does not simply use the
pitch itself, as Pāṇini did: That would mean that in ca. 300 B.C. there was no living pitch
accent in the East, - or at least one different (i.e. bhāṣika) from the Western type with 3
tones.
144
500- B.C. NBP: Northern Black Polished Ware spreads over N.India
357
Cf. for example, Renou, Histoire de la langue Sanskrite, p.38: "spécimens d'une poésie ...
non hiératique,... réellement populaire; (ann.)... sont d'une niveau différent de la prose
[brāhmanique]."; cf. ann. 64.
358
Pāṇini's report of forms with the comparative and superlative suffixes as taught by the
Eastern grammarians, 5.3.94, is interesting. The suffixes -tara/tama in composition with
eka- indeed occur only in Eastern Vedic texts: ŚBK 7x, ŚBM 1-5 only 1x, and 1x in ŚBM 12;
otherwise, -tama is found only in very late Vedic texts: KGS, ĀgGS; -tara- appears in HGS,
AVPar, ṢB, again in late texts. This seems to indicate that Pāṇ. lived at the end of the late
Br. or even during the Sūtra period. This conclusion is also reached by K. Hoffmann, Aufs.
p.541 sqq.: Pān. knew the late Vedic Sūtra texts MŚS, VārŚS (or at least, the lost Br. -if it
existed - of the Maitr. school. The mantra in question is used at the Soma sacrifice; MŚS
has, when compared to MS, taken over many mantras it lacked in its Saṃhitā from other
schools, e.g. the Kaṭhas. Perhaps the mantra belonged to the fragmentary KaṭhB. At this
instance, we can only state that MŚS is quoted by Pānini.) - He knew of Eastern forms (see
above) and of the teaching of the Eastern grammarians (Śākalya!) but he did not
acknowledge the Eastern Veda texts (ŚB, VS, nor even TS-prose, see ann. 98). This seems to
indicate that he lived in a period (cf. ann. 58) when late Br. like ŚB were redacted and some
of the earlier Sūtras (MŚS) were composed, and probably before the Vṛj(j)i had moved
eastwards to Bihar (see ann. 97), i.e. before the time of the Buddha, or at least, that of the
composition of the relevant Pāli Suttas mentioning the Vajji in connection with the
Magadha kings.
359
Similarly, at Pāṇ. 4.1.17, according to the Eastern gramarians, Vṛddhi + accented suffix
-yá, as for example (in Kāśikā) *Gārgyāyaṇī from Gārgī; the Veda has only: Gārgyāyaṇa-
in the Vaṃśa of BAUK 4.6.2, an Eastern text, and Gārgyāyaṇī in Kauṣ Up. 1.1, which is a
Central (Pañcāla) text.
146
At the other end of the spectrum, the dialect features found in the
Ṛgveda still await a detailed study (which is under preparation by
S.Insler). As a next step, these features could be compared to those of the
Middle/Late Vedic period, as described to some extent in this paper.
Also, the three forms of Vedic, namely Ṛgvedic, Middle Vedic, and Late
Vedic in their various dialects, could then again be compared to the
archaic features of Middle Indo-Aryan.
* Part of the materials included in this paper were first presented at the
Sixth World Sanskrit Conference at Philadelphia, Oct. 1984, and in
enlarged form, at this conference. Subsequently, more materials have
been added; others, excluded from this paper, will shortly be published
as a first installment in a planned series of articles, Notes on Vedic
Dialects, (see §10.2). - For abbreviations of the names of Vedic texts see
below, §4.2.5., and for special abbr. see ann. 72.
360
Needless to say, I invite colleagues to join forces in this long term endeavour to find out
more about the exact state of dialect features in Vedic and to exchange materials,
preferably in electronic form.
147