On the Origin of the Pali Language
by Eisel Mazard
|
After more 150 years of controversy among Western scholars,
basic questions as to the origins of Pali (i.e., the language
in which the Theravada Buddhist Canon has been transmitted
for over 2,000 years) remain unsettled.
In this short review of the contending theories, Mazard
indicates their overwhelming reliance on early interpretations
of the Ashokan edicts, and the eschewal of important sources
of evidence from extant Pali and Prakrit grammarians.
|
Preview:
On the Origins of the Pali Language
I had avoided the question of Pali's origins in my own researches
as long as it was possible to do so, but, in the composition of
a new edition of the Kacc?yana-vy?karana (accompanied by a general
grammar) I found myself adding various footnotes that relied uneasily
on authorities from the centuries-old debate.
The discovery and decoding of an ever-larger number of ancient
inscriptions in the late 19th century inspired an exuberant generation
of speculative chronological and geographical theories on Indic
languages; the Ashokan edicts especially seemed to furnish a wealth
material for comparative linguistics. The historical span between
Vedic Sanskrit and modern Indian languages had once seemed closed
to scholarship, but with the invention of the category of "Middle
Indo-Aryan" languages (modelled on the surmised relationship
between the edicts and much later Prakrit literature) there arose
the prospect of an internally-consistent, developmental model
of Northern Indian language and literature, linking antiquity
to the present day. As the 20th century began, the euphoria had
come to an end: the expanding knowledge of Prakrit languages,
with the discovery and study of whole grammars (rather than working
from rules and features inferred a few inscriptions and fragmenta
Of the two points that Pischel urges (in our quotation above)
the first seems to have been taken to heart, and a very great
pessimism now prevails where formerly there had been an unfounded
confidence in the speculative categories (including the definitions
of the middle Indo-Aryan languages themselves) derived from the
comparative reading of the inscriptions. Obversely, Pischel's
second point seems to be less understood today than in the articles
of a century ago: there is nearly universal confusion as to the
mutual relation of the Prakrits to the vernaculars on the one
hand, and of the Prakrits to Sanskrit on the other. The prevalent
assumption of our day (encountered among many monks and laity,
academics and amateurs alike) seems to be as follows:
...[P?li] was originally the language of the rural people. The
fact seems to be that, side by side with Sanskrit, the language
of the elite, there was a spoken language which was corrupt Sanskrit
language and, in some cases, Pali. This theory accords well with
the Buddha's instructions to disseminate his teaching in the own
language (sak?ya niruthtiy?) of the people; by people he obviously
meant the common people. In this connextion, it may be pointed
out that, in Sanskrit dramas, the females and low characters generally
speak Pr?krit. That Pr?krit was widely in vogue as a spoken language
of the masses seems to be proved also by the existence of several
dialects of it, e.g., M?h?r?str?, Shaurasen?, M?gadh?, etc. According
to some, Pr?krta is derived from Prakrti or nature. So, the language
that spontaneously developed among the people came to be called
Pr?krit.
[Banerji, Sures Chadra, 1987 Folklore in Buddhist and Jaina Literatures,
pg. 1-2]
It is not my object to refute the passage above, but we should
keep in mind (as we move through our brief overview of the controversy
as to P?li's origins) that, at present, there is a very wide-spread
notion that the earliest Buddhist literature (i.e., the Pali cannon)
was written in "a spoken language of the masses", and
that this is demonstrated from one side with selective quotation
from the Buddhist Suttapitaka, and from the other side with quotations
from Sanskrit grammarians. The impossibility of relying on ancient
Sanskritists for an account of any language other than Sanskrit
is discussed throughout by Madhav M. Deshpande throughout his
1993 book Sanskrit & Prakrit: Sociolinguistic Issues (e.g.,
pg. 2-4, pg. 92-94, etc.). It is highly significant that the views
of Pali grammarians and Prakrit grammarians themselves do not
figure anywhere in the piously-constructed story, quoted from
Banerji above.
"P?li's origins are controversial," writes Bubenik,
"There are essentially two basic views
depending on
whether a greater weight is given to its eastern ('Magadhisms')
or its western features. According to the former hypothesis, P?li
is based on the eastern dialect (Ardha-M?gadh??) and later on
the early literary works composed in it were 'westernized'. The
antithetical view considers P?li to be the literary language based
on the western dialect." In the latter theory, "The
eastern elements (so called Magadhisms) may be accounted for by
the assumption that the 'Urkanon' was compiled in an eastern dialect
similar to the M?gadh?[.]" [Bubenik, Vit, 1996, The Structure
and Development of Middle Indo-Aryan Dialects, pg. 4 & 5]
Junghare's summary of the same controversy divides the views into
no fewer than five camps (rather than Bubenik's two); we here
quote the first three of her list, to deal with the fourth subsequently:
There are several opinions regarding the origin of P?li, both
geographic and genetic. The most popular opinion is that P?li
is M?gadh?, the language of Bih?r where Buddhism arose. [...]
Burnouf and Lassen (1826) have, however, refuted this opinion
on the grounds that some phonological features of M?gadh? are
not found in P?li. For example, every [Sanskrit] r is changed
to l in M?gadh?, but r is changed to l only sporadically in P?li.
Also, the depalatization of sh to, which is present in P?li, is
absent in M?gadh?. Finally, in M?gadh? the [masculine] and [neuter]
nouns ending in a, as well as consonants, take the [nominative
singular] suffix e. In the same noun stems of P?li, the [nominative
singular] suffixes of [masculine and neuter] nouns are o and ang,
respectively.
Some linguists such as Kuhn (1875) and Franke (1902), have considered
P?li to be the dialect of Ujjayini since it stands closest to
the language of the [Ashokan inscriptions found at] Girnar and
since the dialect of Ujjayini is said to have been the mother-tongue
of Mahinda who preached Buddhism in Ceylon. Hence, the Vindhya
region has been considered to be the home of P?li
[Junghare, Indira Yashwant, 1979, Topics in P?li Historical Phonology,
pg. 2]
My reader may be confused as to how "Ujjayini" has
suddenly broken into the discussion of Pali's origins. I will
here provide a long quotation from Woolner's text on Ashoka to
illustrate the relevance of the Girnar inscriptions to this controversy,
and also to indicate that the terms of the debate over Pali's
origins (reported by Bubenik and Junghare above) have been largely
derived from the even earlier debate as to what language (or:
languages) the edicts of Ashoka were written in. The comparative
studies employed to regionalize the variants found in the Ashokan
edicts were applied by simple extension (in every 19th century
argument reported by Junghare) to identify Pali's geographic origin;
in other words, the following quotation summarizes both the process
whereby Western scholars defined what features were and were not
Pali, then affixed those alien influences to regions, and, subsequently,
reversed the equation to regionalize Pali's origins. As before,
it is
The language of the inscriptions was at first believed to be
P?li. The Girn?r version fitted in best with this supposition,
and this notion led to many misreadings, e.g., in P?li r following
a consonant is always assimilated, so the possibility of a form
like putr? was not suspected. It was soon found, however, that
the language of Dhauli and the Pillars differed from Girnar?especially
in having a [masculine nominative singular form ending in] -e
instead of in -o, and in replacing r with l, e.g., l?j?, 'the
king.' These are peculiarities of M?gadh? Pr?krit, hence this
language has been called "M?gadh?," which was all the
more appropriate because Asoka was king of M?gadha.
The Sh?hb?zgarh?, and later the M?nsehr? versions introduced other
features some of which resembled Sanskrit rather than P?li, e.g.,
priya, putra. At first it was thought that these were Sanskritisms
due to the scribe's knowledge of Sanskrit. A similar explanation
could be applied to such forms in the Girn?r version as putr?,
potr?, prapotr?, when these were recognized.
This hypothesis was gradually given up, and such forms are now
regarded as due to the local language in each case, which was
in certain particulars more archaic and therefore more like Sanskrit
than either P?li or the "M?gadh?" of the Eastern inscriptions.
The detailed, systematic study of all the forms with regard to
grammar and phonetics has led to the adoption of the view that
the original version of every edict was in the Eastern dialect
(The language of Asoka's court) and was translated into, or adapted
to, the current language of outlying Provinces more or less accurately;
or in some cases copied with a few variations due to the local
speech. Thus, so far from reading [the Girn?r edict's use of]
priya as a Sanskritism, the authorities now regard Girn?r piya
as a M?gadhism. Of course it is impossible to be always quite
sure whether a particular "M?gadhism" is due to the
inaccuracy of the translator, who copied an Eastern form; or After
making allowances for such "M?gadhisms" it is found
that the dialectic variations of the inscriptions fall into two
main divisions, Eastern and Western. The Western dialect is represented
by Girn?r. Here we have [nominative singular ending] in o, the
use of r, pr, tr, and [locative singular ending] in amhi, to mention
the most salient features.
[Woolner, Alfred C., 1924 & 1997, Asoka: Text and Glossary,
pg. xx-xxii]
What Woolner reports above, as two theories as to the underlying
language of the Ashokan edicts, is precisely the same division
reported by Bubenik as to the Eastern and Western origin theories
for the Pali language. The various theses as to Ujjayini origins
for (or early influence upon) the Pali language basically stand
or fall with the comparative study of canonical Pali's resemblance
to the Girnar and Magadhi edicts. Writing in 1924, the gleam has
long since dimmed from such theories, in Woolner's estimation
(and his reasons for skepticism resemble the earlier opinions
of Pischel or even the later opinions of Bloch [1965]):
[...] It is clear then that Asoka's inscriptions do not furnish
us with a series of authentic records of the local dialects. We
have rather one series in the Eastern official language showing
here and there slight variations due to local speech, especially
at K?ls? and Mysore. Then there is the Western language, probably
of Ujjain, represented by Girn?r and the Sop?ra fragment. Allied
to this, but strongly influenced by Northern phonetics, is the
dialect of the Sh?hbh?zgarh? and M?nsehr?, of which the latter
is more akin to K?ls?, and may represent the official language
of Taxila.
bq.If we use the name M?gadh? for Asoka's language, it must be
remembered that it does not coincide with the M?gadh? Pr?krit
of the [classical] drama[tists] and grammarians. In particular,
Asoka's "M?gadh?" like Shaurasen? had only the dental
sibilant, e.g., sus?s?, "obedience" ([the K?ls? rock
edict] has susus?, sus?s?, susus?) whereas a marked feature of
M?gadh? Pr?krit is that it has only the palatal sibilant (shushshush?).
There are other features of the M?gadh? according to the Grammarians
which are not found in Asoka's language
bq.[...] Again, a court language is frequently not identical with
the local vernacular, and a lingua franca, or language used for
general purposes over a wide area, rarely if ever retains all
the peculiarities of its original home. We know little of the
linguistic history of the various kingdoms in the Ganges valley
previous to the Mauryan Empire, but we may conjecture that the
official language of the Mauryan court and administration would
not be the "broad" M?gadh? of the day, but a 'refined'
form much influenced by the court languages of neighboring kingdoms,
now absorbed. A modified M?gadh? of this kind could be called
Ardha-M?gadh?, though the language of the extant Ardha-M?gadh?
texts is made up of a somewhat different mixture of ingredients.
bq.An enumeration of all the phonetic and grammatical peculiarities
of each group of inscriptions may be to a certain extent misleading.
We have already seen that the occurrence of an Eastern form in
the West or the North may be a "M?gadhism." The vocabulary
and bulk of the inscriptions is limited, so that a phonetic equation
may be illustrated by a single word, or by words of which the
reading is not quite clear, or the interpretation disputed.
[Ibid.]
I will add one further comment on the thesis as to Mahinda's
influence (part of the Ujjain origin thesis, summarized by Junghare
above). Although I would not dispute the grammatical and syntactical
similarities that have been carefully catalogued in comparing
canonical Pali to the Girnar inscriptions (and, NB, it is a separate
question as to whether the latter edicts can be presumed to be
written in "Ujjain" at all), the proposition that the
language of the entire Pali canon was re-invented or perverted
by Mahinda is susceptible to counter-evidence: Mahinda's journey
to Sri Lanka is not the only source of extant Pali texts that
we have today, and whatever influence he had on the language should
therefore be demonstrable from the contrast between the texts
he transmitted and those from other sources (e.g., sources endemic
to India, or exported by various other routes, such as the direct
transmission East to the Mon kingdom). Junghare does not believe
that any signif
We here resume our quotation from Junghare, listing the final
two of five positions on Pali origins that she provides in summary:
...[B]ut linguists Oldenberg (1879) and M?ller (1884) consider
the Kalinga country to be the home of P?li. Their conclusion is
based on the argument that the oldest settlement in Ceylon [i.e.,
the oldest "Aryan" settlement, ignorning the Adivasi
population!?E.M.] could have been founded by the people of Kalinga,
the area on the mainland opposite Ceylon, rather than by people
from Beng?l and Bih?r.
bq.P?li has been considered as Ardha-M?gadh? because there are
similarities between P?li and ?rsa (?rdha-M?gadh?) phonologies
and morphologies, and because ?rdha-M?gadh? differs from M?ghadh?
exactly on the same points as P?li.
[Junghare, Op. Cit. Supra]
The final view listed above is, again, a direct extention of
the debates on the language of the Ashokan inscriptions, and Bubenik
[Op. cit. supra] suggests that the very language "Old Arha-Magadhi"
(treated as distinct from Ardha-Magadhi) was more or less an invention
of convenience to resolve some of the inconsistencies in that
debate:
[...] Having regard to the retention of sa and other particulars,
Professor L?ders is inclined to regard Eastern Asokan as Old Ardha
M?gadh?. Of course, the language does not coincide with the Ardha-M?gadh?
of the Jain Scriptures, which seems to have come into contact
with M?h?r?shtr?. L?j?, "king" ([or, in the genitive:]
l?jine) [as found in what L?ders dubbed Old Ardha M?gadh?], is
very different from [the Ardha M?gadh? spelling] r?y? ([with its
genitive form being] ranno). Many of the common characteristics
of Ardha M?gadh? are absent or occur very rarely.
[Woolner, Op. cit. supra]
We must emphasise that all the voices represented in this controversy
have been slient for over 100 years; contemporary scholars such
as Bubenik have tended to be more skeptical about the hypothetical
categories established by former generations of scholars and therefore
rely less on speculations based in the comparative study of those
categories. From the perspective of many modern Buddhists, it
is simply convenient to imagine that the Buddha preached in "the
language of the masses" (even if they lack any plausible
anwer to the question "Which masses?"?e.g., could the
Bhaddekaratta have been understood by "the masses" of
both Maghadha and Ujjain?) and for scholars working in literary
Prakrits, there is no actual utility in affiliating Pali with
one inscription or another (and the latter is the limit of our
practical ability to regionalize any middle Indo-Aryan language).
We may note, on the doctrinal side, that after the-Buddha-to-be
gave up his throne, the first place he went was Magadha [stated,
e.g., in KN: Sutta-Nip?ta: 3-1]; in the early stages of his religious
and philosophical education, he learned to memorize and recite
the dhamma of various teachers (e.g., ?l?ra K?l?ma, MN:36, Mah?saccakasutta,
paragraph 11), and this was presumably in a Prakrit endemic to
Magadha or its environs. Then (after studying various doctrines
to the point of becoming the equal of his teachers) he continued
to wander in Magadha looking for a place to practice alone [MN:36,
Mah?saccaksutta, Paragraph 22]. The strong canonical identification
of Magadha as the region in which the Bodhisatta was educated
(after leaving his native kingdom, but before his enlightenment),
i.e., where he first learnt both to recite religious poetry and
also to participate in formal, philosophical debate (e.g., in
The overall unity of the extant Pali texts of diverse routes
of transmission suggests that the interaction of Eastern and Western
Prakrit dialects became ossified at a very early period (i.e.,
before Mahinda's journey); I would here suggest that the authoring
of Pali's first grammar by Kaccayana (who evidently knew both
a Prakrit of Maghadhi and at least one language of Ujjain) may
have been the decisive first step in that synthesis of Eastern
and Western features. Without ascribing some such role to Kaccayana,
it is difficult to account for the rapid adoption of Western lingual
elements into a religious literature that so clearly emerged from
the intellectual environment of Magadha (and Pali is remains,
apart from these numerable features, substantially "a Maghadan
Prakrit", as Deshpande describes it [Op. cit. supra.]). To
prove or refute this thesis would entail detailed reading of the
inclusion and exclusion of Western elements in the Vy?karana,
as compar
It is poor form to begin an article complaining about current
scholarship, and to end with a request for more; however, I have
been blunt in stating that my own engagement with this issue has
been forced by my preparation of an obliquely related, forthcoming
publication, and I have primarily written this article in the
hope that I will be referred to some further studies, as yet unknown
to me. As I have no access to any academic library, there are
a few important texts known to me that I have yet to lay my hands
upon (such as D.R. Bhandarkar's 1883 article for the Bengali Royal
Asiatic Society) but I suspect there is more recent and more important
work to be found in the strange country of unpublished theses,
conference papers, etc.; I will gladly consider any further opinion
on the matter that may be delivered to me in reply to this article.
|