Who was Kaccáyana, or, who wrote the Vyákarana?

An excert from the introduction to the book.  All subscribers to the list
are welcome to change my mind before it goes to press!

E.M.
  ---------
Who was Kaccáyana, or, who wrote the Vyákarana?

According to Buddhist tradition, the ultimate source of this textbook is a
set of verses on grammar composed by Mahá-Kaccáyana that were preserved by
memorization and recitation from the time of the Buddha for several
centuries before being committed to writing.  For all of the religion's
recorded history (i.e., commencing much later), oral tradition of the
Kaccáyana-Vyákarana has been the primary mode of learning the Pali language;
subsequent grammars have tended to supplement (rather than supplant) the
Vyákarana's prominent place in monastic pedagogy, and many believe that this
text we have today (the earliest extant Pali grammar) is approximately the
same mode of instruction used by Buddhists for some 2,000 years --although
an earlier origin of 2,500 years remains debatable.  I have heard the doubts
of both monks and secular scholars as to whether the author of the grammar
is in fact the same Kaccáyana as the character depicted in the canon, and
who lived in the era of the Buddha; I will return to the question below, but
for the moment will speak of the figure of Mahá-Kaccáyana in the
suttapitaka.

The character of Kaccáyana is very well recorded in the Buddhist canon, and
the anecdotes pertaining to him (both from the suttas and the commentaries)
have been brought together in one article by Bhante Bodhi:

Bhikkhu Bodhi, Maha Kaccana: Master of Doctrinal Exposition, 1995, The Wheel
Publication No. 405/406, Buddhist Publication Society, Kandy, Sri Lanka.
(Also available as a free electronic resource)

In overview, Kaccáyana strikes the reader as an independent spirit, living
most of his life far removed from the Buddha, wandering, teaching, and
finding solitude in the mountainous forests of the kingdom of Avantí, not
far from his birthplace in the capital, Ujjení.  Although Ujjení seems to
have been connected to the "Middle country" (Majjhimadesa) of India by trade
routes, it is perhaps a sign of how very distant it was in practical term
(i.e., how difficult the journey might have been) that the Buddha never
visited it, refusing invitations to do so from both the king of Avanti, and
Kaccáyana's himself.  For a sense of the geographic distance, we may refer
to Bimala Churn Law, Geography of Early Buddhism, 1932 (reprinted 1979), pg.
22, and the map provided at the end of that volume.

Kaccáyana's relative isolation, with only a few students, on a distant
frontier of the religion, seems to be well illustrated by the events
concerning the ordination of Sona recorded in the Vinaya; in brief, 3 years
passed before Kaccáyana could arrange the presence of 9 other monks to
perform the ceremony, and this resulted in the passing of a special rule
that as few as 5 monks could ordain a new monk in very remote areas.  This
is related in more detail by Bodhi in the article cited above (the PTS
citation for the original is Vin.i.195f).  Kaccáyana's independence from the
Buddha in his travel (and in his opinions) is also shown in his association
with the city of Madhurá (where it seems he repeatedly visited, stayed, and
preached to the locals) --this seems to be the one and only place that we
know the Buddha specially avoided, as he expressed his vexation with the
city in a sutta of the same name in the Anguttara Nikáya.

Kaccáyana's situation as an independent voice for the Buddha's teaching in a
land both lingually and culturally separate from Maghada seems to be
consistent with his rôle as an early grammarian and interpreter.  We may
speculate that his pupils in Avantí would have had a greater difficulty in
unravelling the Buddha's dense, poetical statements in Prakrit (e.g., the
Bhaddekaratta sutta) than a native speaker of Maghadan with some prior
exposure to the literary languages used in the area.  Some scholars would
here object that I am anachronistically describing Pali as a single, defined
language at too early a stage, and might further argue that it emerged out
of various Prakitic influences in the centuries following the Buddha's life;
but in this connection, it is immaterial whether one speaks of a single
Magadha Prakrit spoken by the Buddha, or of a category of many Prakritic
dialects used by his followers (subsequently merging), for all such dialects
would have been equally strange to Avantí, and all equally subject to new
influences from other languages (Prakrit or otherwise) in the centuries
before the canonization of Pali in Sri Lanka.  To illustrate my point with a
more historically definite example, given what we know of the close
approximation between the earliest extant Jain Prakrit and language of the
Pali canon as it is now preserved, it should be clear (as a matter of
historical inference, not of fact) that the followers and monks that the
Buddha attracted from the pre-existing Jain tradition in Magadha would have
been familiar with a tradition of reciting religious poetry in a Prakritic
language not very different from the Buddha's own --whereas distant Avantí
(even if it had its own caste of Brahmins reciting the old Sanskrit of the
Vedas) would not have shared in so closely comparable a Prakrit tradition.

The question remains as to what extent the legendary origin of the text
might be true, and, if it is only partly true, to what extent we can
identify the author of the grammar with the Kaccáyana who was a contemporary
of the Buddha, or any other historical figure with a similar name.

Vidyabhusana, in his introduction, dismisses the traditional account on two
inter-dependent assertions: (1) that the text as we have it today was of
single, simultaneous authorship, and (2) as such, there would be
anachronisms were its origins so early as tradition states (Vidyabhusana,
op. cit. supra, pg. xxviii).  The second point collapses with the first, and
one of the very few pieces of historical information we have about the
origin of the text states that it was the work of several different authors,
possibly living in successive (but unspecified) periods of time; and this
sequence successive authors would be perfectly consistent with the
organization of the text as we have it today, being comprised of three
"Layers", namely, sutta, vutti, and payoga.  Vidyabhusana presents and
summarily rejects this (one and only) piece of historical evidence with very
little explanation:

In the Kaccáyana bheda tíká we find: "The Yoga (Sútra) was written by
Kaccáyana, the commentary by Sanghanandi, the examples were added by
Brahmadatta and the gloss by Vimalabuddhi."  From the manner in which the
sutta, vutti, payoga, and nyása are intimately connected with one another, I
am inclined to believe that the entire work was written by Kátyáyana
himself.  At any rate the sutta etc. were written simultaneously.
(Ibid., pg. xxvi)

Thus, on the bare assertion that "the manner" in which the different layers
"are intimately connected" seems to suggest simultaneous authorship,
Vidyabhusana proceeds to treat anachronisms found among the examples
(payoga) as if they were definite indications as to the date of all layers
of the text.  He proceeds to speculate that the author was a Kaccáyana
(i.e., not the same as found in the canon) who lived in Mathurá during or
after the 3rd century B.C. (Ibid., xxviii).  His identification of Mathurá
(more commonly spelled Madhura) as the putative author's home is quizzical,
and actually undermines his thesis that this particular Kaccáyana and the
character known to us from the canon (i.e., contemporaneous with the Buddha)
are two separate people: Vidyabhusana argues that an example mentioning
Madhura, and another naming a king of Madhura, suggest that the author had a
special association with the place; but the Kaccáyana of the Buddha's time
was indeed especially associated with Madhura, and this is plainly shown in
several suttas stating that he lived in a forest there (such as the
eponymous Madhura sutta of the Majjhima Nikaya).  More fundamentally, one
can hardly say that there is any more geographic significance in these two
passing mentions of Madhura than in the two examples mentioning the
Himalayas (following verses 94 and 274 respectively); the simplest and the
truest thing to say is that we would expect to find toponymns such as
Madhura and the Himalayas even in a grammar composed a thousand years later
in distant Burma or Cambodia, as they are canonical toponyms that any
Palicist (of any era or country) might drawn upon to compose examples.  The
basis for Vidyabhusana's assigned date is the same "anachronism" mentioned
in Mason's introduction (i.e., two contemporaries of Ashoka appearing among
the Vyákarana's examples), the significance of which stands or falls with
the assumption that the examples must have been written at the same time as
the sutta (which is especially doubtful, as the commentary tells us that
they weren't written by the same author!).  My main objection to this line
of reasoning is that it replaces evidence with mere speculation, and so does
nothing to improve on what little we know from legend.

I have an open mind on this question, but most of the arguments that I have
seen on the matter are very deeply flawed.  As one example, E.J. Thomas
writes that the Sutta-Niddesa is such an early text that it was written
while Pali "was then a current language", and that "grammatical analysis"
was only just "becoming necessary for the interpretation of the texts" (E.
J. Thomas, "Buddhist Education in Pali and Sanskrit Schools", The Indian
Historical Quarterly, vol. 2:3, 1926, September, pg. 495-508; see pg. 502). 
Thomas thus has the fixed notion that grammars only arise at a "late"
period, i.e., when the knowledge of a language is in decline, and its
writings are becoming incomprehensible.  It is therefore no surprise that he
assigns Kaccáyana to a very late date: "The Pali grammar of Kaccayana is
later than Buddhaghosa, and belongs to the literature of Ceylon" (Ibid.,
504).  He then quotes Geiger to suggest that influence from (or imitation
of) Sanskrit grammatical concepts is another sure sign of a text's late
origin (Ibid.).

In reply to all this, firstly, there seem to be no grounds to assign an
early date to the Niddesa, and it may be as late as the 1st century B.C., or
even the 2nd century A.D. (Hinüber, Op. Cit., §116-118); what is of more
direct significance to our argument is that Thomas believes the Niddesa is
early precisely because it offers only crude lists of synonyms, and not
fully developed "grammatical analysis".  This is a recurrent problem in Pali
studies, whereby scholars first call something "crude", and then, in the
next breath, call it "early", with no attempt to substantiate the connection
between such perceived simplicity and the timeline of the religion.  After
all, the Sutta-Niddesa is a commentary on a poem, i.e., neither a work on
grammar, nor pedagogy --it is not clear as to why it should offer anything
more than "crude" grammatical notes in passing, if it serves the author's
purpose is discussing the poem.  What is more absurd is the notion that
there was ever any period so early that there would be no need for
"grammatical analysis" in teaching or understanding the Pali language; both
the cosmopolitan spread of Buddhism across lingual regions, and its spread
across caste divisions in early Indian society would have assured that from
the first days of the Buddha's preaching his dense, philosophical poetry
would have been very difficult for the majority of his audience to
understand.  The truth of this is everywhere written in the dialogues of the
suttas (we may again refer to the Bhaddekaratta as a ready example of such
an opaque poem, and the suttas offering competing interpretations thereof
follow upon it: Majjhima Nikaya, No. 131-134).  Above all else, we must keep
in mind the simple fact that literary Prakrits (such as Pali was, or out of
several of which Pali was composed) are not and have never been anyone's
"Mother tongue" ("Páli has never been a spoken language neither in Magadha
nor elsewhere", Hinüber, Op. Cit., §7; see also: Deshpande, Madhave M.,
1993, Sanskrit & Prakrit: Sociolinguistic Issues).  Resultantly, the notion
that Pali would have been grammatically self-evident to native speakers of
any one region (e.g., Magadha) is flawed, and the fact that Buddhism rapidly
spread over such a vast area would renders the point trivial even if it
could be maintained; but to show the absurdity of this line of thinking it
is enough if we simply ask, "When has it ever been the case that grammar and
pedagogy were unnecessary to teaching any written language, be it living or
dead?"  The general point (that Thomas quotes from Geiger) that Sanskrit
influence in grammar and poetics has increased over time (in the history of
Sinhalese literature) is quite true; but it would be absurd to infer from
this that Kaccáyana's text must either show a complete ignorance of Sanskrit
terms, or else be considered a fraud of a later period (especially as the
antiquity of the terms themselves is not in doubt, i.e., Sanskrit grammar is
an older science than its Pali equivalent; Vidyabhusana provides a fine
recapitulation of the early timeline of the grammars of both languages in
his introduction, op. cit. supra).  This approach belies a kind of religious
bias that many secular Palicists secretly harbour, namely, the belief that
the "pure", "early" tradition of Buddhism was completely free of Sanskrit
influence; but we have it plainly stated in the Suttas that many of the
Buddha's converts (including Kaccáyana himself) were formerly Brahmins who
chanted the Vedas, etc., so it would be jejune to suppose that monks of that
"sacred generation" in India were entirely ignorant of Sanskrit.  Certainly,
there is no anachronism in their recourse to it for technical terms, except
where those terms can be shown to have arisen from later developments in
drama, poetics, and so on.

The skepticism that various scholars have expressed as to the origin of
Kaccáyana's grammar is understandable, but their grounds for doubt are
deserving of skepticism in turn.  Three assumptions that I have generally
found tacit in the literature are (1) that it would have been highly
desirable for the true authors of the text to fraudulently assert Kaccáyana
as its author, (2) that the historical record (such as we have it from
Buddhist sources) would be incapable of properly distinguishing multiple
personalities with the relatively common name Kaccáyana, and (3) that Pali
itself is assumed to have been undefined for a long period of organic
development before its canonization in Sri Lanka; resultantly, the existence
of an authoritative grammar at such an early stage of the religion's history
would require re-thinking of what little we know about the early history of
the language.  On only one point can I agree: new thinking is needed about
the origins of the Pali language.

Firstly, if Kaccáyana were not the author of a major grammar, it is not at
all clear to me that he would have any degree of celebrity that a fraudulent
author would want to lend to his own work.  Aside from the fame that this
very grammar won for his name, Kaccáyana appears as a fairly minor figure in
the canon, and it is difficult to argue that he would have served as a
better pseudonym than one of the chief disciples (as, e.g., in the
attribution of the Niddesa to Sáriputta) if some false name were to be used
to lend authority to a grammatical treatise.  While there are historical
examples of celebrated names being falsely attributed to other monks' works,
the common scenario tends to be that the work of many hands (such as a
voluminous commentary) is misrepresented as the work of just one, well-known
author (e.g., many of the commentaries attributed to Buddhaghosa by the
Mahávangsa, etc., see: Hinüber, op. cit. supra, §207, & ch. 5 generally);
but, as we have seen, in the case of the Kaccáyana Vyákarana this scenario
is reversed, for what we have is a concise text that has been duly credited
as the composite work of several different authors, none of whom could
really be considered famous except for the very work in question.  I will
not digress to deal with the second point at any length, except to observe
that the Pali literature has preserved the distinctions between a huge
number of persons known by overlapping names (examples of this are very
numerous: Abhaya, Cunda, Gosála, Kálí, etc.), as well as distinctions
between post-canonical authors of the same name as characters in the canon
(such as the various Moggallánas).  The possibility that there could have
been confusion does not indicate that there actually was any; there is no
evidence to suggest the confusion of names, and there is no confusion within
the little evidence that we have.  In response to the third assumption, I
can say little, as I neither believe that the existence of a grammar is an
obstacle to the historical development of a language (has this ever been the
case?), nor do I think that the current theories about Pali's origins (that
tend to rely on inferences from the comparative reading of the Ashokan
edicts) are so precious as to preclude the possible early origin of the
Kaccáyana Vyákarana (there is further discussion of these matters in the 4th
end-note I've added to chapter 1 of the present volume).  Further, the sutta
of the Vyákarana neither provide so complete a description of the language,
nor represent so definite a resemblance to the written language of the
extant canon, as to imply that the rules must have written after Pali was
canonized; on the contrary, there are many interesting differences between
the language Kaccáyana describes, and what we empirically find in the canon
today.  Mason draws attention to the disparity throughout his grammar (e.g.,
§140, "…no such forms are given by Kaccayano, and they have probably been
introduced into the language since his grammar was written.").

It does not seem to me any less plausible that a few hundred grammatical
rules should have been transmitted by oral tradition for several centuries
(i.e., from the time of the Buddha to being committed to writing) than that
any other text should have made the same transition.  We could as easily
propose that the grammar is a genuine historical artifact, but that the
dialogues of Kaccáyana in the canon are false (i.e., were later inventions
to glorify the grammarian), as to propose that the grammar is falsely
attributed to the author known to us through the canon; the evidence to
support either proposition is the same (i.e., nil), and the assumption that
canonical texts are more difficult to tamper with than non-canonical ones
has no basis but religious sentiment.  I do not find it implausible that
this grammar should trace its origins to one and the same Kaccáyana depicted
in the canon as the Buddha's contemporary --at the very least we must
acknowledge that there is no evidence that would make it more plausible to
assume that the basic verses (the sutta) might have been written by any
other monk.


--
A saying of the Buddha from http://metta.lk/
View Streaming Dhamma Video http://dharmavahini.tv/
Not by silence (alone) does he who is dull and ignorant become a sage; but
that wise man who, as if holding a pair of scales, embraces the best and
shuns evil, is indeed a sage.
Random Dhammapada Verse 268