Freelance ExBuddhist wrote:I'm aware of K.R. Norman's argument (can't we just call him "Ken"?) but I find it more provocative than convincing
What exacly is provocative or unconvincing? Norman seems up-to-date with the facts as far as I can say. He is one of the few scholars whom I have found to be relatively rational and philologically competent, as far as the language is concerned. I do not deprecate the work of others, but many others approach it from a 'Theravada' (ideological) point of view so they feel compelled to comply with traditional interpretations and biases; as opposed to rigorous philological research and considering secure evidence from outside the Theravada/Buddhist traditions.
by contrast, see the chapters on Pali in Deshpande's book "Sanskrit and Prakrit, Sociolinguistic Issues" --the latter (really just a few pages of material) sums up what I would call the conservative conclusions drawn from current evidence.
I dont have access to the book, looked it up on google books for references to Pali. Can you give me the page numbers? He appears to conflate canonical Pali with Magadhi Prakrit (and later Pali). Magadhi Prakrit was a 'dramatic' prakrit of the mid 1st millenium CE (mentioned in the Indian dramas), along with the other prakrits i.e. Sauraseni, Maharashtri, and later Pali (Paisachi). Canonical Pali belongs to the 3rd-2nd century BCE. The post-canonical 'Pali' of Buddhaghosa's commentaries belongs to the middle of the 1st millenium CE. Canonical Pali had no Indian literature as far as I know except the Buddhist canon. Fragments of later Pali however survive in 8th century Jain epics like the Kuvalayamala. However these late and 'artificial Pali' works (which are derived from Sanskrit) cannot be used to analyze the canonical language which is much more complex.
So upon a preliminary observation, Deshpande is mixing up different periods of time (i.e. the Buddha's era, Ashoka's era, and post Ashokan commentarial eras) by tarring them all with the same brush i.e. he seems to view the early period with the biases and notions that were prevalent in a later era. This is by and large what modern Pali scholars also do.
You are correct that the word Pali is sometimes used to mean "the original text" (in contrast to a commentary, or in contrast to other things), and not a language unto itself. It would also be correct to say that the assertion of Pali as a language unto itself was something bound to the creation of its own grammatical tradition (something Buddhists did, indeed, do).
With all of these observations have been stated, I do not think there is compelling evidence that Pali formerly was Gandhari (by which "some form of Gandhari Prakrit" is implicitly meant); occam's razor would suggest, instead, that we suppose there was uneven influence from a number of Prakrit traditions, and that these cobbled together and then (imperfectly) "evened out" as Pali developed a set of standard spellings, orthographic conventions, etc.
Pali script was formerly Gandhari i.e. Kharosthi script in the Pre-Ashokan era as Kharosthi/Gandhari is the earliest script of India. Brahmi (i.e. canonical Pali's script) evolved within a century or so from Kharosthi. All Indian scripts are derived from Brahmi (and by extension from Kharosthi). Brahmi eventually evolved sufficiently enough to be able to represent Old-Indic conjunct constants phonetically (and that was the end of Pali, almost all future texts were composed in Old-Indic). If Pali were a distinct language, there would be no reason for it to die out as suddenly as it was born. All this history of Brahmi evolving from Kharosthi is well known.
Evidence also suggests that the sthaviras in some sense were exiled from Magadha (and they carried the canon in pali with them when they left) to Sri Lanka, but I am not sure if this was a voluntary migration or whether it occured due to the Mahasangha sect gaining control of Buddhism in Jambudipa (India). However the differences in BCE between the Sthaviras and Mahasangha was not to do with language, i.e. the Mahasanghikas also followed the same core Pali/late-vedic canon, but had a different (older) vinaya, while the Theras added new rules to the original vinaya which evidently earned the wrath of the mahasanghikas. The mahasanghikas evidently retaliated to the thera innovations in vinaya by composing new suttas of their own (i.e. the mahayana suttas), but these in turn were rejected by the Sthaviras (i.e. theras). However these historical matters of the early post-Buddha sangha cannot be conclusively stated as we need to rely on sectarian polemical works like the Kathavatthu for evidence.
Gandhari was not a prakrit, it is never mentioned in Indian literature as a prakrit (i.e. a common-era vulgar dialect of sanskrit) or even as a language as far as I know. Both the name 'Gandhari' as applied to this putative language and its identity as a prakrit are 20th century 'western' inventions. Just as the 3rd-2nd century BCE language of the non-phonetic texts of the buddhist canon came to be called 'Pali' language in the 19th century (resulting from a misunderstanding of Buddhaghosa's usage of the term, and fully ignoring the lack of any such BC-era names/grammars for that 'language'), the predominantly Buddhist texts in the Kharosthi script which were found mostly in 1994 in and around northern Pakistan and surrounding areas were artificially named the 'Gandhari' texts (claimed to be in a putative Gandhari language), there is no reference to any Gandhari prakrit or Gandhari language in any literature until the 20th century.
The old (i.e. mid-late 1st millenium CE) grammars of prakrits usually follow a set pattern - they rarely describe prakrits in isolation, Prakrits are always explained in contrast to Sanskrit, they usually devote 7-9 out of 10 chapters explaining sanskrit, and the last few chapters explaining the ways each of the 'named' (i.e. dramatic) prakrits diverge from sanskrit in terms of phonology and morphology. Canonical Pali is not usually classified as a prakrit by the ancient grammarians as its 'derivation' from sanskrit (i.e. the simplified grammatical standard of late-Vedic) was tenuous at best, and it (specially in the canon) preserved old-Vedic spoken forms such as the subjunctive mood, and the various Vedic infinitives (which are preserved in Pali, like tum, tuye, taye & tave) - which were almost entirely ignored by Panini in his simplified/standardized 'sanskrit' grammar. A strong understanding of Vedic would therefore be necessary to make full sense of canonical Pali, and there are ways in which Pali can aid Vedic (and sanskrit) study as well. However, there are no real Vedic scholars today (at least I don't know of any) just as there are no real Pali scholars who can do full justice to the canonical language. Pali is the written form of late vedic i.e. several vernacular Old-Indic dialects, and the Buddha lived in the late vedic era before the sanskrit standard was established by Panini. Commentarial Pali (Visuddhimagga, Sumangalavilasini, Manorathapurani etc) however can be safely explained away as a prakrit (derivative of sanskrit) as it no longer preserved the Vedic grammatical forms of the canon.
There were no 'prakrit traditions' in Ashoka's time (3rd century BCE). The name 'prakrit' as applied to a linguistic group (contrasted with sanskrit) is a mid 1st millenium CE construct usually found in Sanskrit dramas of the same time (where the lower 'i.e. unscholarly' classes stereotypically speak/write in 'phonetic prakrit' and the brahmins and kings who are presumed to be educated in 'correct' language stereotypically speak/write in 'phonetic sanskrit'). In Ashoka's time and earlier on, there simply was no way to write perfectly phonetic Old-Indic (as the script was still gradually evolving towards phonetic representation), so all inscriptions are in middle-Indic in written form but old-Indic in speech.
Just that one sentence proposes a long series of assumptions that are very difficult to support (in terms of Occam's razor) instead of the easy-to-imagine scenario of various people converting to Buddhism (in its first few generations) and cobbling together a canon with imperfect linguistic influences from at least a few different provinces of India, with a few different Prakrit traditions, and then (imperfectly) smoothing out the spellings (etc.) to be more consistent with time (but, of course, the canon is still far from being some kind of "perfectly consistent" work of literature).
To be clear, the core-canon did NOT
originate from many parts of current-day India, linguistically speaking the language of most/all of the canon is north-western in origin, and archaeology has yielded all early buddhist manuscripts from the north-western India as well. The language of Ashoka's inscriptions that is closest to canonical Pali are the major rock edicts at Girnār (i.e. the old Girivrāja/Giribbāja, capital of Magadha) in the modern Gujarat state of western India. Even if we assume that certain suttas/vaggas/nipatas did originate from different parts of India, such differences would probably exist between texts, one would not find expect to find such variations within the same text. Also what Norman mentions there are not dialectical variations but plain errors.
I find Norman therefore to be accurate in his finding (supported by numerous examples in the same paper) that "There was another type of error, however, which was due to deficiencies in the early Indian writing system. In the earliest form of the Brāhmī script, double consonants were not written, and the marks for long vowels were frequently omitted." This is evidently a fact, and the earliest Kharosthi manuscripts prove it. The evidence from outside the Buddhist tradition is even more vehemently in favour of Norman's findings about the deficiencies in the earliest written script(s) of India.
Again, I think you'd be willing to support this as a reasonable supposition for the early emergence of Jain prakrit, also --and you would have no reason to believe that its origin was as a transcription from Gandhari (or any other unique source).
The Jains themselves have been more circumspect about such claims. Indian literature (of whichever religion, or no religion at all) do not mention any such thing as Buddhist Prakrit or Jain Prakrit. The name Jain prakrit is a modern western invention. The earliest jain canonical texts that are still in existence are in a language now called 'Arsha' (i.e. of the Rishis), or alternatively ardha-Magadhi (half-Magadhi) which is linguistically very similar to the texts we now call Pali (they can be considered to be written dialects of the same language on account of their similarity). Both Arsha and Pali are western languages from the Gujarat/Rajasthan area of India, and these are the areas where the Jains still have most of their ancient holy sites. Girnar (i.e. the old Giribbaja, the location of Ashoka's 14 major rock edicts) is still one of the most important Jain sites.
Lurking behind Ken's enthusiasm for Gandhari is the old dream of discovering "an ur-canon", or a "proto-canon"... and, as I've mentioned before, rather too much emphasis on minor errors found in passages of poetry (that, really, do not provide any kind of smoking-gun evidence for the earliest composition of the canon as a whole… they just show that Pali has been an imperfect vessel for transmitting poetry, and that poetry was most likely composed by a variety of authors from a variety of linguistic backgrounds in India, etc., some of it going through stages of translation as well as spelling-reform, etc. etc., in an uneven manner).
I think it is better to leave aside insinuations of old-dreams of an ur-canon etc., as they are not relevant. The theravada tradition as a whole insists on being rooted on the origin ( 'sthāvara/sthāvira' = "adamant/unmoving", which is as much a pejorative term as 'hināyana') so there is no use accusing Norman alone of it. The facts speak for themselves. I have personally found such errors and variations in the prose suttas as well (and not really just between texts as also within texts). If I were to spend a few months looking for them, I expect to find many more instances of these (and other) issues all over the canon. To comment on the language of the canon, one necessarily has to look at it from a philological and more broadly a linguistic standpoint. That is Norman's stated principal concern, not old-dreams of ur-canons etc, see http://www.shin-ibs.edu/academics/_forum/v5.php
As I mentioned in the other thread, Dr. Colette Caillat has established in her famous paper 'La langue primitive du bouddhisme' that the Theravada tradition (while it has well preserved the original 3rd century BCE canon) had serious issues in understanding the language of the canon correctly. As an example she has shown how the word isipatana stands for *ṛṣyavṛjana, a synonym of the word migadāya, with which it is almost always found coupled in the canon. This has been accepted and commented on by later Pali scholars like Thomas Oberlies (in his seminal work on the Pali Canon's grammar); Oskar von Hinüber in his paper 'Linguistic considerations on the date of the Buddha'; Heinz Bechert in his book 'When Did the Buddha Live?' etc.
I'm sorry, but it is impossible to make arguments based on "the absence of evidence" with ancient India: we also have zero extant commentaries on the Vedas, but we're aware that they existed (and new ones were quite possibly composed during this period, but lost forever). A large part of India's literary heritage was literally burned to ashes during the muslim conquest --and there has been plenty of cultural dislocation since.
That is correct, a large part of India's literary heritage was burned to ashes during the Muslim conquests, but I cannot assume (from the lack of evidence) that there may have existed other BCE works in canonical Pali that were so destroyed. However I can claim that the lack of evidence for the existence of computers in the Buddha's time is evidence of their absence (because we know through independent evidence that computers were invented much later). Similarly we know that Pali was not recognized as an independent language even by Buddhaghosa, since he refers to the text of the canon (but not his own commentary) as 'Pali'. Moreover there is evidence that he used Sanskrit grammar to make sense of the canonical Pali, see this article -- http://www.ocbs.org/ojs/index.php/jocbs/article/view/5/5
. The article concludes (rightly in my opinion) that the Sthaviras had their own Sanskrit tradition to interpret the Pali canon and did not depend on Brahmin/Mahayana sources for it (which makes perfect sense if Pali was originally the written form of late-Vedic, and not a separate middle-Indic language that was in ideological linguistic opposition to the language of the Brahmins). That is valuable independent evidence that should be taken into account.
The earliest commentaries on the Vedas from the pre-Buddhist era do still exist, they are called the Brahmanas and are composed in the Vedic language (see http://en.wikipedia.org/wiki/Brahmanas
) and passed down orally as part of the vedic canon, much like the Maha Niddesa and Culla Niddesa commentaries of the Sutta Nipata are handed down as part of the Pali canon (but in writing). If the Brahmanas (vedic commentaries) had been written down in the same script as the Pali canon, they would have become Pali Brahmanas. But the Brahmins (probably looking at what writing did to the language of the Tipitaka) at this time laid down a strong proscription against writing down the Vedic canon, with the result that the oral tradition of the brahmins continued until the modern era and writing the vedas down is still frowned upon in Brahmanical circles.
Medieval (Sanskrit) commentaries of the Vedas also exist, they are called the Sayana Bhasya (the commentary of Sayana). There may have been other commentaries as well which are lost, but that is not relevant. As long as they are found mentioned in ancient sources, their existence can be presumed. Vedic works previously considered lost but mentioned in ancient sources/commentaries have been subsequently found, one of these is the Paippalada Samhita, a rescension of the Atharva veda. But there is no mention of Pali/Magadhi/Gandhari as separate languages (even by any other name) in any BCE literature, whether in Theravada or Mahayana or Jain or Hindu. BCE era literature mention only 2 languages (or two major dialects of the same language), early vedic (chandaso) vs late vedic (laukiko). Sanskrit was the grammar of late vedic. The Pali canon was in pre-Sanskritized late-vedic as it still preserves Vedic forms in its grammar (while sanskrit grammar does not). All the linguistic and literary evidence securely indicates that Pali was the 'non-phonetic' written form of that language, not its spoken form.
We should resist the urge to foist our own constructs (of it being a separate language) to Ashoka's era or the Buddha's era. The theravada tradition (including modern western theravada scholarship), not in this instance alone but in other instances as well, regularly foists later constructs and ideas on the era of the Buddha and the era of the canon. The identity of Pali is a prominent victim of this approach.
Yes, the use of terms like "Buddhist Hybrid Sanskrit" is very artificial, and entirely modern; yes, the meaning ascribed to the word "Pali" (and even the word "Sanskrit" itself) has shifted around a bit from one century to the next (as we look at, again, very limited evidence). But, simply, so what?
Saying, "this term did not have the same meaning then as it does now" does not entail that "this language did not exist at that time", etc., and it definitely does not support any of your notions that make orthography more important than other features of the language (at any stage of the history).
I claim that these middle-Indic languages did not exist not because we have new names for them, but because there is a complete absence of evidence for their existential identity from the very era in which they were supposed to have existed. If the Buddhists (and grammarians) from the Buddha's or Ashoka's time did not think that the language of the canon was unique enough as to merit separate mention, the absence of references to canonical Pali is indicative of it not being recognized as a separate language. While using absence of evidence as evidence of absence may be misleading, what is even more irrational however is to use absence of evidence (of references to Pali) as evidence of presence (of Pali as a language) which is what western scholarship is doing.
We today call the language of the Kharosthi texts Gandhari, there is no evidence that the language of these texts were called Gandhari when they were first written, or that it was called a language by some other equivalent name. We may even compose a grammar for these texts (as all, ancient & modern, Pali grammars were composed for the pali canon), but that is not an evidence that the language was spoken phonetically as per the written text. The canon has undergone phonetic 'transliterations' atleast a few times and the standard (and non-standard) word forms that we find now are not an evidence for the way they were originally spoken. We need to distinguish between texts (as they now exist) and the language (as it was spoken just as it was first written down). We can only know the language when we remove the layers after layers of phonetic innovations that the texts have gone through from their earliest written stage. That is the work of the philologers like Norman and Caillat, and of others who are widely read in ancient Indian literature. But for a start, the language of the canon bears strong linguistic resemblance to Vedic, so the modern Theravadan notion of canonical Pali being a middle-Indic dialect (or even a prakrit which makes even lesser sense) rather than an Old-Indic dialect, should be revised.