Stefan Karpik’s Light on Epigraphic Pali: More on the Buddha Teaching in Pali is the latest in a long series of articles in Western scholarship that attempt to clarify the nature of Pali in its relation to other Indic languages of the time. Thanks to Jayarava for bringing this to the forum.
I’ve read a bunch of these, and my usual response is, “okay”. I’m no linguist, so I am not really equipped to evaluate the argumentation in detail. I’m quite happy to accept whatever the experts think. But the arguments always seem unpersuasive to me.
In particular, they are highly inflationary, a bit like multiverses in quantum theory, though not nearly so extreme. They posit major series of events, such as the translation of the entire Buddhist canon, or even the existence of whole languages or types of languages (koine), for which there is no evidence apart from inference from some obscure phrases in Pali. As Karpik points out, this seems to be a case ripe for Occam’s razor.
Another distinction he makes is between the “lumpists” and the “splitters”. The latter tend to proliferate new languages based on a few differing features, resulting in endless distinctions between barely-attested or unattested languages. I’ve long felt that this tendency had gone too far and have been more of a “lumpist”, so I’m happy Karpik has made this label available for me!
The strongest argument Karpik makes is for what he calls “Epigraphic Pali”. From about 200 BCE to 300 CE, there are some hundreds of inscriptions across northern India in a fairly consistent language, which Karpik argues is simply a later form of Pali. He takes several examples and restores them, and it is obvious that the restored text is, in fact, simply Pali.
The process of restoration is largely based on the work of the original scholars of the manuscripts, which Karpik cites in detail. The kinds of alterations are:
- Using modern spelling conventions: for example, old Brahmi often did not spell doubled consonants, so a word like putena may be restored to puttena.
- Fixing spelling mistakes: the inscriptions are often careless and typos can be identified, sometimes by comparing with other occurrences of the same word in the same site.
This restores, for example, something like silākaṁmaṁto cha upaṁno to the familiar Pali silākammanto ca uppanno (“produced by stonemasons”).
As is inevitable in such a long-lived, wide-flung language, there is some variation due to language evolution, local dialectical influence, and the like. Karpik shows that in such cases, the forms in Epigraphic Pali are later than those in Scriptural Pali, if we may thus call the language of the Theravada scriptures. An example of this is in the absolutive ending tvā in Scriptural Pali, which we find as ttā in Epigraphic Pali. As Karpik points out, the simplification of tv to tt is well attested in other cases in Pali (eg. sattva to satta). This kind of change, then, is to be expected as part of the normal linguistic evolution of Pali, and needs no further explanation. Moreover, where such changes happen, they evidence later forms in Epigraphic Pali. Such changes tend to happen in a certain direction, sometimes summarized as Grimm’s Laws.
He further argues that certain changes that have been regarded as dialectical influence are instead natural features of Pali. One classic case is the alteration between r and l. The language of the Ashokan edicts uses l where Pali has r (eg. laghula for rāhula). But this feature can be seen in Pali as well, and is obviously a widespread phonetic phenomenon. So its occurrence does not necessarily have anything to do with the Ashokan language.
Not only does he show that the language is Pali, he gives examples of texts that are not Buddhist. One inscription, from 1st century BCE Maharasthra, is a passage of homage to Vedic gods. It should be noted that throughout this period, Buddhism was the dominant religion of the region, and we do not find large numbers of Brahmanical inscriptions until later, coinciding with the decline of Pali and rise of Sanskrit.
One very old inscription, 3rd century BCE Chhattisgarh, is a secular poem celebrating the poets of the spring festival. Its sensual promotion of worldly delights is very un-Buddhist. These examples put to rest the argument that Pali was solely an ecclesiastical language, invented for the purpose of the Pali canon, and used solely for Buddhist literature.
Other inscriptions are Buddhist, usually dedications for stupas and the like, and occasionally short quotes from the suttas.
Regarding the name of the language, Karpik shows that the term Māgadhī is not found at all in the Pali tradition. The canon and commentaries refer to their language by a variety of names. The Vinaya uses the term Ariyaka, by which is means the language of the canon, although I don’t think it is possible to know how broadly the dialectical net is cast here; does it include closely-related languages like Ashokan Māgadhī? I think it probably does. Later text often refer to Pali as the “language of Magadha”, which I suppose is technically different to “Māgadhī”.
But the main point here is that what the Pali tradition means by the “language of Magadha” is simply Pali, whereas what modern scholars mean by Māgadhī is the language of the Ashokan inscriptions. Karpik argues, I think reasonably, that “Magadha” was an exonym that for the Sinhalese Buddhists essentially meant “northern India”, and has no special relation to the specific region of ancient Magadha with which the Ashokan language is connected.
Speaking of the Ashokan language AKA Māgadhī, Karpik points out that, for all the attention it has received by modern scholars, it is a limited dialect, mainly attested in the Ashokan edicts, and which disappears from the epigraphic record after the Mauryan period. It seems likely, in fact, that it was the spoken language of Ashoka himself, as the personal and unpolished nature of the Ashokan edicts indicate that they are likely to be direct dictations from the king.
Under Karpik’s view, then, Pali was the dominant language of northern India from the time of the Buddha for over 500 years. It would have been the language of trade, of literature, of common discourse (sāmañña) in the region. He does not strongly argue for the origins of this language, but indicates that it may have its roots in Kosalan, as the Buddha was Kosalan and spent the majority of his time there. Kosala was the trans-regional trade powerhouse before the ascendancy of Magadha.
Ashokan Māgadhī is a sidebar, the language of an influential but short-lived empire that nonetheless left only marginal traces in Pali. As before, so after, Pali persisted.
We only have limited sources, and should be wary of over-reading them. Just because the bulk of Pali is Buddhist does not mean it was a Buddhist language; it just means the Buddhists were successful in preserving their scriptures. And while Epigraphic Pali prevails in post-Ashokan inscriptions, it does not mean that the brahmins were not present and working in Sanskrit. And, obviously, there were other languages as well, whether related dialects like Māgadhī, which was the ancestor of the Ardhamāgadhī in which the later Jain texts were cast, or non Indo-European languages such as Dravidian and Munda.
While it is appropriate to maintain a careful degree of skepticism with regards to traditional claims, there is no doubt that the tradition gets many things right, and no particular reason why it should be wrong in this respect. The tradition says its language, the “language of Magadha”, was the language of the Buddha, a common tongue understandable across the region of northern India. It turns out that that language is in fact widely attested across northern India for a long period of time. This is a reasonable and simple hypothesis, which enjoys the support of our oldest sources on the matter both scriptural and epigraphic.
Our forum friend Thanuttamo gives a nice survey of scholarly opinion on the matter, and shows why the traditional account should not be so readily discarded.
While Karpik’s article approaches 20,000 words, however, it’s not long enough. I’d really like to see a systematic analysis of all the inscriptions in Epigraphic Pali, or at least a substantial sample of them. It’d be great if there were a searchable online source for these; in fact they could easily be put in a single HTML file. I’d also like to see a more systematic response to the various arguments for either a Māgadhī source for the original Buddhist canon, or the existence of a koine as argued by Brian Levman. I know Karpik disagrees with Levman’s arguments, but he doesn’t discuss them in this paper.
Perhaps when I read the next paper on this topic, I’ll say, “hmm, seems reasonable”. But I have to say, this is the first time in a long time where I’ve really felt like someone is making a solid case. Previous attempts to rehabilitate Pali as the language of the Buddha seemed slim at best. Others have noticed the connection between the language of the epigraphs and scriptural Pali, but have not established their identity in detail. Still others have made more exotic theories, which often seem persuasive if only through the sheer force of linguistic prowess. But the mark of a good hypothesis is that it makes sense of a great many facts with a simple assumption. And in that regard, Karpik’s work stands out. Another mark of a good hypothesis is that it is fruitful, and I hope that this work inspires further research and clarification.