“What can be said at all can be said clearly, and whereof one cannot speak, thereof one must be silent.”
Ludwig Wittgenstein. Tractatus Logico-Philosophicus.
Stefan Karpik (2023) has proposed that “serious attention be given to the Theravada tradition that the Buddha spoke Pali” (2023: 41). Both this and an earlier paper (Karpik 2019) make linguistic arguments about the Pāli language, arriving at conclusions that question the existing paradigm on the history of Pāli and its relation to other Prakrits. Karpik then argues that these new conclusions tell that the Buddha spoke Pāli. In this essay, I will review these papers and some related material. In this first section I'll outline a broad response to the claim that we can know what language the Buddha spoke, in the context of some responses to Karpik and a resume of the milieu that he has emerged from. I'll identify some unexamined assumptions that Karpik makes (in common with others in his milieu). In the next section, I'll consider the historicity of the Buddha, then the issue of historicity itself. Finally, I will make some remarks about historical facts that can be gleaned from Pāli texts and then conclude with a summary.
My first response to Karpik (2019a, 2023) is that, while the philological methods that Karpik employs allow him to make interesting and even compelling conjectures about the history of Pāli, these methods do not allow him to infer anything at all about what language the Buddha spoke without relying on some major assumptions that I don't find interesting, let alone compelling. Something I will reiterate below is that the issue of what language the Buddha spoke is entirely extrinsic to the issues of the history of Pāli. Karpik's conclusions are compatible with literally any position on the historicity of the Buddha. However, the historicity of the Buddha is the hill that he has chosen to die on.
The reasons for rejecting his conclusions are obvious. Karpik accepts the modern consensus that the Buddha lived in the fifth century BCE. There is simply no evidence related to the Buddha from this period or within about 500 years of this date. All that we think we know comes from Buddhist scripture composed in a later period and how we interpret scripture depends on which assumptions we make and/or do not make. And such assumptions are not explored in Karpik's articles. The date itself is based on a series of assumptions and speculative interpretations of Buddhist scripture. Moveover, there is no evidence of any language other than Sanskrit and a Northwestern Prakrit being spoken at that time (this evidence comes from the Sanskrit Grammarians Yāska and Pāṇini). It's interesting to see Karpik relying on a consensus on dates, when his project is to undermine another consensus amongst virtually the same small group of scholars.
The simple fact is that there is no evidence from that time period on which we can base a history of the Buddha. This is not to say that the myth of the Buddha as found in Buddhist texts is not important. Nor do I argue that the Buddha did not exist. We cannot base an argument for the historicity of the Buddha on the evidence we have, since it all comes from religious texts composed long after the time in question, and then only according to particular, biased, readings of those texts. We simply don't know.
There has been little response to Karpik (2019a) from academics working in the field already. The notable exception is from Bryan Levman. Levman has been actively publishing on the history of Pāli for some years (2008, 2010a, 2010b, 2016, 2019, 2020a, 2020b, 2021, 2022, 2023). Levman's (2019) critique of Karpik (2019a) on philological grounds is the most extensive response and strongly argues against Karpik and for the utility of the current consensus.
Karpik seems to have been given a "right of reply" to Levman since his rebuttal appears in the same issue of JOCBS. Karpik (2019b) repudiates all of Levman's points and criticises him quite severely for ignoring facts, using faulty methods, and even misunderstanding linguistic technical terms. Note that these are serious accusations in an academic context: Karpik implies that Levman is incompetent.
While I don't entirely follow (or care about) the linguistic arguments, the idea that someone as well versed in this topic as Levman got everything wrong and effectively doesn't understand his area of expertise seems far-fetched. On the other hand, my research on the Heart Sutra shows that such situations in which the "experts" in Buddhist Studies are flatly wrong about everything are certainly possible. So I'm not a priori against the idea, but the proposition that Levman is substantially wrong on the facts is prima facie unlikely. Edward Conze was a charlatan of the first order, but Levman seems on the level to me.
Other responses have been cursory. Mark Allon (2021) mentions Karpik (2019a) in passing, grouping him with Richard Gombrich and others who believe, without evidence, that the Buddha spoke Pāli. Allon, a leading expert on Middle Indic, certainly does not seem to take Karpik's argument seriously. Similarly, Roderick Bucknell (2022), another expert on Middle Indic, mentions Karpik (2019) but only in passing. He seems unpersuaded as well.
In the end, I don't know enough about linguistics to adjudicate on the linguistic issues. I think Karpik could be right. I found his articles persuasive. I also found Levman persuasive and he could be right as well. That said, I think I do understand the historical points that Karpik seeks to make and I note that Levman shares many of Karpik's presuppositions on this matter. It is this historical aspect of Karpik's articles that I will be addressing.
Karpik's contributions have emerged from a particular milieu based in Oxford, UK. Richard Gombrich founded the Oxford Centre for Buddhist Studies (OCBS)—until recently associated with Oxford University—in 2004 to promote the study of Buddhism. Gombrich was also instrumental in founding the Numata chair in Buddhist Studies at Oxford, now held by Kate Crosby.
When Gombrich retired as director of the Oxford Centre for Buddhist Studies (OCBS) in 2020, Wynne was anointed his successor (I gather from Gombrich that he was the only candidate). They co-edited the OCBS journal (JOCBS) in 2019 and then Wynne took over in 2020. Wynne (2006) had already contributed to the "debate" on the Buddha's language, concluding:
"I therefore agree with Rhys Davids, and disagree with sceptics such as Sénart, Kern and Schopen, that the internal evidence of early Buddhist literature proves its historical authenticity." (65)
Wynne (2006: 66) ends on a characteristically pugnacious note: "The claim that we cannot know anything about early Indian Buddhism because all the manuscripts are late is vacuous, and made, I assume, by those who have not studied the textual material thoroughly." Like Karpik, then, Wynne sees the people whose interpretation of scripture conflicts with his as not merely wrong, but as incompetent. He apparently believes that no one could read the same scriptures as he has and come to a different conclusion. Which would be a first in the history of interpreting religious scripture if it were true.
In a more recent JOCBS article, Wynne (2019a: abstract) states that "early Buddhist discourses are largely authentic, and can be regarded as a reasonably accurate historical witness." Wynne certainly proves that this is his belief, but his conclusions are based on a reading of Buddhist texts that assumes their authenticity and the historicity of the Buddha. Wynne (2019b) has also weighed in, via a JOCBS editorial, on the specific topic of the language the Buddha spoke. Again the assumption throughout is that the Buddha is historical and that the Pāli suttas are a "reasonably accurate historical witness".
Also emerging from the Oxford milieu are two notable longer works. An extensive apologetic tract by Therāvadin bhikkhus Sujato and Brahmali (2013), published as a supplement to JOCBS 5, which again assumes the historicity of the Buddha and the authenticity of the the Pāli Canon and then presents evidence that "proves" the authenticity of the Pāli texts.*
* Sujato has recently stated that he is "not Theravādin", though he still uses his Theravāda ordination name, still wears Theravāda robes, and still allows people to refer to him using Theravādin honorifics like "Bhante" and "Venerable". Given that he was kicked out of the lineage that ordained him, one wonders why he persists in the fiction that he is a bhikkhu at all.
And Gombrich's (2018) own contribution, which also supports the idea that the Buddha was probably historical and that Pāli was probably the language he spoke. Gombrich, a good Popperian, leaves room for doubt.
To date, all of Karpik's publications have been in JOCBS under Wynne's editorship.
I will happily stipulate that Karpik (2019a) makes an interesting and persuasive argument for Pāli being the ur-language of the Pāli canon. Similarly for his argument that Pāli was a single language with natural variations rather than a koine or argot; that it need not reflect an artificial language or a mashup of dialects, and that at least some suttas were probably composed in Pāli. I am persuaded of the possibility of a community of Buddhists in India using Pāli in daily life and recording their ideas about Buddhism in that language. The idea that texts were composed in some other language and translated into Pāli does look questionable. Karpik (2023) extends this argument to include the Asoka inscriptions under the heading of Pāli.
What puzzles me is why Karpik, Gombrich, Wynne, and even Levman, all think that their conclusions about the history of Pāli, or even conclusions of this general type, have any bearing at all on the problem of what language the Buddha spoke. Knowing what language the Pāli texts were composed in or knowing the relationship between that language and the language of the Asokan edicts tells us nothing at all about the Buddha. I can’t see that one has any bearing on the other, except when we assume a priori that it does. As Karpik explains, in criticising Levman, when the assumption leads the conclusion:
This is a circular argument known as "begging the question" or petitio principii, where one assumes what one wishes to prove in order to prove it.
Karpik accuses Levman of relying on this informal fallacy. It is obvious, however, that this same fallacy is central to Karpik's historical arguments about "the Buddha". The unexamined assumptions that Karpik appears to rely on include:
(1) that the historicity of the Buddha, qua founder of Buddhism, is not in doubt
(2) that the Pāli literature faithfully records the utterances of the “historical Buddha”
(3) that the Pāli literature can be taken literally
(4) that the Asoka inscription have some clear relation to spoken language in different parts of India at the time.
Let us try to see, then, the role these assumptions play in Karpik's articles.
On the Historicity of the Buddha
As already noted, Karpik’s method leans heavily on the assumed historicity of the Buddha. For example, “The Buddha would have known of the precise transmission of the Vedic texts” (2019: 17). I’m not sure how Karpik knows this and he doesn’t say. My impression is that Brahmins learned the Vedas in private and that their mnemonic methods were not used by Buddhists because they did not know about them. There is no mention of such techniques in the Pāli texts to my knowledge (and as it happens I have comprehensively studied references to Brahmanical religious belief and practice in the four Nikāyas for an unpublished article).
A few pages later: “The evidence suggests a single, intentionally fluid, oral transmission from the Buddha.” (19). I agree that he has made a case for oral transmission, but “from the Buddha” is not a conclusion that he draws from the evidence presented. Rather, “from the Buddha” relies on a background assumption about the Buddha and his role in founding Buddhism. The evidence presented does not speak to this issue at all.
Stories about the kings mentioned in the Pāli are discussed as though they, too, are historical. We see statements like, "In the Buddha’s day, king Pasenadi of Kosala and king Ajātasattu of Magadha had each defeated the other in battle (J II.237)" (Karpik 2019a: 21). Just as for the Buddha, there is no evidence that either Pasenadi or Ajātasattu is historical, and no evidence for battles between them other than stories in scripture.
Note that the source Karpik cites here is a Jātaka story. The Jātaka and Avadāna literature is explicitly allegorical and/or mythological in character and predicated on (the supernatural) idea of the Buddha "remembering his past lives". And yet Karpik's interpretation of this literature is presented as an equally reliable and valid source of historical information as, say, the suttas. Karpik seems to accord this special status to every text that he cites in support of his thesis. And at this point his brutal methodological criticism of Levman starts to look disingenuous, since Karpik himself appears to be unclear on what kind of inferences his own methods can validate.
Another example occurs in Karpik's (2019b) rebuttal of Levman (2019):
In common with MOTT (Multiple Oral Transmission Theory) advocates, Levman gives no account of why the underlying layer was discarded and lost, despite repeated injunctions in the suttas to memorise them to the letter (Karpik 2019: 14-15). (Emphasis added)
Here again, Karpik is interpreting scripture rather than putting forward an argument based on evidence. His argument is that certain religious texts say it should not happen, therefore it cannot have happened. But this reasoning is clearly faulty, even at a common sense level.
Gregory Schopen has noted that where we have archaeological evidence for early monasticism, it almost invariably contradicts the texts. Wynne (2006) argued that Schopen's scepticism—he always sides with archaeology over texts—is "extreme" and takes the opposite view, that the texts are usually trustworthy. At best the conflict between text and archaeology leaves us with unresolveable uncertainty. Note that the scholars who seem loathe to acknowledge this uncertainty are all practising Theravādins or Theravāda-adjacent. Note also that the disagreement seems to take the form of denunciation. The suggestion is always that those who argue that we don't know and cannot know are somehow disingenuous, "extremist", and/or incompetent. While it seldom rises to the level of an ad hominem fallacy, the language used is not consistent with academic standards of discourse.
Karpik's (2019a) discussion then turns to the subject of where the Buddha lived and taught, as though the Pāli texts straightforwardly describe his actual life. Karpik provides four pages of charts of locations attributed to suttas, and simply treats these as factual records of where the Buddha visited. He even notes Schopen’s (2004) article outlining his discovery of a Buddhist Vinaya text that shows that locations were allowed to be made up when they were missing. And, of course, they were/are missing in very many cases.
It is, of course, true that the Buddha is popularly believed to be an historical figure. No one denies this. Interestingly, Bryan Levman shares Karpik's belief on this score. However, as David Drewes (2017) has pointed out, academic historiography has a rather higher bar for historicity than religious or popular belief and, all things considered, the Buddha does not meet this bar.
We are thus left with the rather strange proposition that Buddhism was founded by a historical figure who has not been linked to any historical facts, an idea that would seem decidedly unempirical, and only dubiously coherent. Stuck in this awkward situation, scholars have rarely been able to avoid the temptation to offer some suggestion as to what was likely, or ‘must’ have been, true about him. By the time they get done, we end up with a flesh and blood person – widely considered to be one of the greatest human beings ever to have lived – conjured up from little more than fancy (Drewes 2017: 1).
A straw man argument that we commonly see employed against Drewes, is that he argues that "the Buddha didn't exist". This fake argument was raised, for example, by leading Middle-Indic scholar, Oskar von Hinüber (2019: abstract):
David Drewes reviewed the opinions of a number of western Buddhologists on whether or not the Buddha was a historical person and in conclusion claimed that the Buddha never existed." (Emphasis added)
Actually this is not true. Drewes never makes this claim and what he does say is far more nuanced:
Although the idea that the Buddha cannot be considered a historical figure may seem radical, my argument is really a minor one. Though there has long been an industry devoted to the production of sensational claims about the Buddha, nothing about him has ever been established as fact, and the standard position in scholarship has long been that he is a figure about whom we know nothing. My only real suggestion is that we make the small shift from speaking of an unknown, contentless Buddha to accepting that we do not have grounds for speaking of a historical Buddha at all (2017: 19)
Drewes is writing for academic historians not for religious believers. However, this distinction is often blurred in Buddhist Studies because so many Buddhist Studies scholars are heavily invested in normative Buddhist traditions (e.g. Gombrich, Wynne, Sujato, Brahmali, and Karpik). Academic historians not having grounds to use the term "historical Buddha" is not the same thing as saying "the Buddha never existed". What Drewes says boils down to this: academic historians don't know and we should stop saying "we know".
The specific category error of mistaking an epistemic argument ("we don't know") for a metaphysical argument ("he doesn't exist") is so common in Buddhist thought and academic Buddhist Studies that it ought to have a name. This fallacy poisons all of Edward Conze's work, for example. And most of the commentary on the doctrine of anātman. Highlighting this fallacy and correcting it is central to my own revisionist history project on the Heart Sutra. I believe we would get closer to the truth of Buddhism by abandoning all metaphysical claims related to Buddhism and reframing them as epistemic or phenomenological observations. While this is still a minority view, there are some interesting academic contributions such as Hamilton (2000), Shulman (2008), Gombrich (2009), Heim & Ram-Prasad (2018), and Jones (2022).
Not only does von Hinüber (2019) misrepresent Drewes' conclusion, but his method of validating his own claims consists entirely of interpreting scripture. In one sense, then, von Hinüber's article ought to give Karpik heart, since it shows that even the most educated and highly regarded experts are capable of serious missteps. On the other hand, when we pay attention to what Drewes actually says, it clearly vitiates Karpik's claims to know anything at all about the Buddha. The only (potentially) valid inferences that Karpik draws concern the history of Pāli, but even then he makes a number of unexamined assumptions about when Pāli was spoken. We—i.e. people who write about Buddhist history in academic journals—still don't know if the Buddha was a real person or not. His historicity certainly fits certain religious presuppositions, but the arguments in favour of it all involve interpreting scripture.
Drewes is not arguing for one position over another here. He is arguing that we don't have enough information to take any position. As historians, we may choose to indulge in speculation when evidence is lacking, but this has to be clearly marked as such so as not to confuse readers. An inference drawn from interpreting evidence is significantly more meaningful than speculation based on interpreting scripture or speculation designed to mask a lack of evidence.
Drewes points out that that this distinction is seldom if ever drawn in Buddhist Studies. Certainly, Karpik does not make this distinction. At the very least, speculative conclusions must be hedged ("it appears...", "it seems...", "it may be the case..."). Notably, Gombrich (2018) does this. Karpik's choice of language suggests certainty, i.e. that this is a valid conclusion based on clear evidence. How can anyone be certain that their interpretation of scripture amounts to a fact?
Similarly, Jonathan Walters (1999: 248) notes:
I think it is fair to say that among contemporary historians of the Theravāda there has been a marked shift away from attempting to say much of anything at all about “early Buddhism”… more recent scholars have tended to regard early Buddhism as unknowable.
Walters goes on to demonstrate the kinds of historical facts that can be obtained from studying suttas. They are records of something after all. The argument is over what they are records of and when. Long experience of dealing with religious texts tells us that the parsimonious approach is to take the texts as reflecting the beliefs of the community that wrote down the stories. For example, we could say with some confidence that the authors of the Pāli canon believed that the Buddha was an historical character. But then we have to put this in the context of their belief system, their worldview. Karpik appears to share that Iron Age worldview and treat it as self-evident and this blinds him.
On Historicity
There are numerous facts that can be stipulated for the sake of exploring this issue. There certainly was a period of Indian history, beginning in the seventh or sixth centuries BCE and extending over several centuries, known as the Second Urbanisation; the first urbanisation being the Indus Valley civilisation. The cities named in Pāli suttas correspond in many ways to archaeological sites associated with the Second Urbanisation (though most were only found with the help of the Chinese pilgrim Xuanzang's seventh century travelogue).
All this tells us is that the stories in the suttas were composed after the second urbanisation was well underway, when all the named cities were well established and prosperous. That is to say, some time in the latter half of the first millennium BCE. Since they don't mention Asoka, we may infer (though we don't know) that the composition of new suttas in Pāli had ceased by the mid-third century BCE at the latest. Though composition of Buddhist texts per se in India continued apace while there was life in Buddhism.
The archaeology of the Second Urbanisation has a striking feature that Karpik might have cited in support of his thesis but did not. This is the "two cultures" hypothesis. The exposition of this hypothesis in Geoffrey Samuel (2008) is useful and still the best I have read. Based largely on the distribution of ceramic technology, we see two distinct cultures in Northern India at this time: one in the west consistent with the Brahmin's home territory (the Kurukṣetra or Āryavarta), and one centred on the central Ganges Valley consistent with the cities of the second urbanisation. It seems to me that the relative uniformity of the material culture of the region is a sign that we might expect the kind of linguistic uniformity that Karpik proposes. Since this is evidence from the actual time he wishes to discuss, it is surprising that Karpik overlooks it. Still, none of this evidence supports Karpik's assumptions about the Buddha.
Similarly, the geography described in Pāli suttas, the fauna and flora, are all quite accurate where they pertain to the material world. Of course, the Pāli literature is a religious literature and as such it does not limit itself to describing the material world. Alongside descriptions that appear consistent with a modernist worldview, we can read in detail about places such as Brahmaloka, numerous Devalokas, and Niraya, the Buddhist hell. Brahmās, devas, and asuras are every bit as "real" as human beings in Pāli suttas. Our human world, which is incidentally flat, is said to be comprised of four continents arranged symmetrically around Mt Meru. Alongside descriptions of elephants and cattle, we read about nāga, yakkha, gandhabba, kiṃnāra, and many other supernatural species.
While modern scholars, including Karpik, are apt to exploit this natural/supernatural distinction and interpret natural and supernatural descriptions on different criteria, it’s not clear from the texts themselves that the authors of the texts made this distinction. There is no shift in linguistic register, for example, when describing Sāvatthī or Brahmāloka; or between elephant and yakkha. If we look at the Buddhist traditions of Asia and Southeast Asia, living Buddhists tend not to make this distinction, either.
The worldview of the Pāli authors, like other Iron Age societies we know about, was suffused with supernatural entities and magical forces. Part of the appeal of the figure of the Buddha was his "shamanic" ability to master the supernatural, to travel to a devaloka or brahmaloka and converse with the inhabitants. And so on. The Buddha of the Pāli canon regularly performs miracles and magical feats.
If the Pāli descriptions of the material world were truly "authentic", then we would have to accept the proposition that their descriptions of the supernatural world are also authentic, since the texts themselves don't make any distinction between them.
Karpik and the others who argue for the historical authenticity of the Pāli suttas tacitly bracket out the Pāli texts and passages that don't conform to their view of history and pretend that they don't matter. They also pretend that making such distinctions is uncomplicated, mere common sense. They proceed as though the criteria by which they make this distinction need not even be stated, let alone justified.
The idea that the Buddha is "historical" or that the texts are "authentic" requires a biased and motivated reading of the texts which eliminates anything "non-historical" or "non-authentic" (without ever offering, let alone discussing formal definitions); and the corollary is that whatever is left from this motivated winnowing is "reliably historical and authentic". That is to say, it is only by consciously exercising a modern bias that such scholars can make and sustain historical claims through interpreting this ancient literature.
There are numerous problematic absences in the archaeological record. As already noted, no physical evidence from the relevant period has ever been associated with any person named in the Pāli suttas, let alone the Buddha. If there was ever a king of Magadha named Ajātasatthu, for example, he left no evidence behind: no artefacts, no architecture, no coins, no inscriptions, and he is not mentioned outside of Buddhist scripture. There is no external corroboration of his existence from non-Buddhist literature of the period. Nor of any other character mentioned in the suttas.
Arguments from absence are notoriously weak since "the absence of evidence is not evidence of absence". On the other hand, and this is David Drewes' point, the absence of evidence means we cannot draw any definite conclusions. As historians we must respect such epistemic limits. Where the historical record is silent we are left with uncertainty. Speculating to fill this gap is certainly fun, but taking our own speculations as facts is not consistent with good methodology. Karpik does not seem to understand this.
We can contrast this with the situation for Asoka. His dates are frequently cited as absolute and other events are dated relative to his dates. However, these dates are far from certain. The reconstruction of the names of Greco-Bactrian kings in Edict no. 14 is certainly plausible and even persuasive. Moreover, there are numerous inscriptions whose texts are plausibly attributed to Asoka. There are artefacts from the time period that correspond to a wealthy and powerful king. The pillar edicts must have been enormously expensive to make and suggest the kind of wealth that only an emperor could command. The consensus, based on this evidence, is that Asoka was an historical person who lived in the mid third century BCE (with some error bars). No such evidence for the historicity of the Buddha has ever been presented and Karpik certainly does not add to our knowledge in this respect. Rather, he assumes the historicity of the Buddha and proceeds as though his presupposition is a self-evident truth.
When we look at Buddhist historiography, a lot of it is stuck in the Victorian Imperialist conceit known as the "great man of history fallacy". This the idea that history is a description of the lives of a series of so-called "great men" who shaped their times. This is how Victorian gentleman scholars saw themselves. Enriched by the British Empire (a vast and merciless pirate enterprise dedicated to robbing the world), they saw no value in women, people of colour, or the working classes; these classes of people were simply there to be manipulated and exploited by "great men". History is a canvas, our lives are the pigments, and great men the artists.
In this fallacy, great men operate outside the usual constraints of society, rather in the manner of a Nietzschean übermensch (or its modern equivalent, the self-interested "Randian hero"). This fallacy is universally repudiated by modern historians outside of Buddhist Studies. However, in Buddhist Studies, many authors simply cannot imagine the history of Buddhism in any other paradigm except the great man fallacy. And those who are not focussed on the Buddha are almost invariably fixated on Nāgārjuna or some other magical figure who is imagined as having no connection to Indian history, generally. Buddhism is presented as the story of a series of influential men without any attempt to contextualise them (often because they are not really historical, either).
One result of this overall bias in Buddhist Studies is that differences great and small within Pāli texts, and between them and other early Buddhist texts, are routinely glossed over in favour of the idea of "an underlying unity". And, this "underlying unity", is then supposed to be evidence that points to historicity of the Buddha. I have never understood the "underlying unity" argument since, having read the suttas, it is apparent that no such unity exists. There is far too much pluralism and internal contradiction within the Pāli literature for this argument to be coherent. By contrast, the arguments for the earliness of the Suttanipāta seem rest on on the heterogeneity of the Pāli canon; i.e. because the Suttanipāta (or parts of it) is different, it must be early. So much for "underlying unity" if the past was actually more heterogeneous than the present.
While there are minimal attempts to see the great man, known as "the Buddha", in his social, political, and economic context, such attempts are inevitably in the service of asserting the Buddha's historicity. No attempt is made to consider social, political, or economic factors in the birth of Buddhism, and the fact is that very little such information exists. Karpik doesn't bother with archaeology, even when it would support his case. Indeed, Buddhist historians typically shy away from causal explanations entirely, preferring descriptive accounts that have no explanatory value. Very few Buddhist Studies scholars are interested in explaining Buddhism and its developments, or the relations between Buddhists and other sects. Several scholars (notably Gombrich and Bronkhorst) have discussed the relationship between so-called "early Buddhism" and the religion of the Late Vedic period, but even this often takes the form of speculating about the influence of Brahmanism on the Buddha (rather than on Buddhism). A work like Ronald Davidson's (2002) history of Tantra that discusses socio-political contributions to the emergence of Tantra in Indian religions is extremely rare and thus valuable. But then scholars of Pāli are unlikely to ever look at is, since its outside their silo.
Pāli
Another unexamined assumption in Karpik (2019) is that Pāli is old enough to have existed at the putative time of the Buddha. Karpik accepts the consensus that emerged from the Bechert conference on the dates of the Buddha, which concluded that the Buddha died ca. 400 BCE and thus lived in the fifth century BCE. These dates are entirely based on interpreting normative Buddhist texts and there is no evidence whatever of Buddhism from the fifth century BCE. Evidence of Buddhism begins to appear around the time of Asoka, in the 3rd century BCE.
In fact, the oldest extant Pāli text is from the fifth or sixth century CE (Stargardt 1995), some 800-1000 years after the putative death of the Buddha (based on the Bechert consensus). Buddhaghosa composed his commentaries in Pāli, but he was also from the fifth century. The idea that Pāli existed prior to the fifth century CE is conjectural and largely based on normative Theravāda religious tradition. This is not to say that Pāli is not older, but that there is some uncertainty that must be acknowledged by those who chose to write on this topic. Even if we stipulate the historicity of the Buddha (for the sake of argument) the idea that Pāli goes back to the Buddha's time is still a matter of popularly accepted conjecture rather than a matter of established fact.
By comparison, the evidence for texts written in Gāndhārī is very much older, with some manuscripts and inscriptions dated to the second century BCE. The bulk of the Gāndhārī corpus, such as it is, dates from the early centuries of the Common Era (after which the use of Kharoṣṭhī script ceased in India). The Gāndhārī literature, as fragmentary as it is, is obviously much older and at the same time much more diverse, than the Theravāda canon, since it includes Mahāyāna texts.
For example, we have a partial and fragmented birchbark manuscript of the quintessential Mahāyāna text, Aṣṭasāhasrikā Prajñāpāramitā, written in Gāndhārī using Kharoṣṭhī script that is carbon-dated to ca 70 CE ± ~50 years (Falk & Karashima 2012, 2013). Moreover, there is a Chinese translation of this text dated to 179 CE. Again, this is considerably older than the first evidence for the use of Pāli. But this is still not evidence from the fifth century BCE.
The Chinese never received transmission of a coherent body of literature reflecting a Buddhist canon. A physical canon, in the sense of an actual collection of all the texts in the catalogues, didn't exist in China until after the eighth century CE and then it was a local creation based on centuries of bibliographic scholarship. During the first few centuries of the Common Era, texts arrived in China in piecemeal fashion, seemingly at random. As the trickle became a flood, resulting in thousands of translated texts, still no existing canon or sutrapiṭaka arrived whole. While the Chinese did receive the idea of a canon with traditional categories—sutra, vinaya, abhidharma, śāstra, etc—they did not receive an exemplar of such a thing. This is in stark contrast to the countries proselytised by Theravādins. Burma, Sri Lanka, Laos, and Thailand all received and preserved the same canon of texts.
In the end, the Chinese had to create their own canon, and this took several centuries to attain a satisfactory internal coherence. Tibetans also had to invent their own canon from scratch and received perhaps 10% of the extant Pāli suttapiṭaka and then as individual texts rather than as part of a canon. Notably the extant Gāndhārī manuscripts, copied in the centuries spanning the beginning of the Common Era, don't seem to form a canon either. Gāndhārī Āgama texts were not translated into Chinese until the fourth or fifth century and even then the different Āgama collections arrived and were translated separately. If there was a Pāli canon in India, it seems not to have been available to any Chinese pilgrims. These simple facts are inconsistent with the Theravāda version of history.
So why do scholars continue to cite the earlier existence of Pāli and the Pāli canon as an uncontested fact and (in the case of Wynne 2006) refer to dissenting opinions (like mine) as "vacuous"? As far as I can see this claim is based on interpreting the Mahāvaṃsa, a traditional Theravāda (i.e religious) history probably composed in the fifth century CE in Sri Lanka (i.e. hundreds of years and thousands of miles away from the time and place it purports to describe), but purporting to describe a history going back to the Buddha. As with canonical Pāli texts, there is no distinction between natural and supernatural in the Mahāvaṃsa. Modernist scholars tease out the aspects that don't overtly mention the supernatural and treat them as straightforwardly true. This is a methodological bias. It is anachronistic. to say the least, since it assumes that ancient authors made modern distinctions that are certainly not reflected in the Pāli literature.
As far as I can see, the dating of Pāli is not based on evidence; it is based on a biased interpretation of scripture. Again, this is not to say that Pāli was not spoken in the second urbanisation, only that this is not an argument from evidence. It is speculative and should be clearly marked as such. Such speculations seem more plausible to religieux than they do to historians, for obvious reasons.
Buddhists Studies seems to exist in a methodological vacuum (aka the "silo mentality"). Many scholars appear to think, for example, that Buddhist Studies is not part of Religious Studies and shares no methods or theory with the broader field. While it is true that early Buddhism specialists now routinely study the Chinese Āgama translations, this is largely in the service of interpreting Pali and little or no attempt is made, at least by Pāli scholars, to understand Chinese Buddhism or Chinese culture. Having had to make some attempts in this direction, I can only say that after 10 years I have barely scratched the surface.
Meanwhile, Jonathan Silk (2015, 2021, 2022) has raised serious doubts about the idea that philological methods developed to interpret the Bible are straightforwardly applicable to Buddhist texts. But then Silk mainly writes about Mahāyāna texts rather than Pāli, so Pāli scholars simply ignore him. Most scholars of Early Buddhism appear to think that Buddhism is exceptional, even unique, and best studied in isolation from questions of history, anthropology, sociology, and ethnology.
Anyone who has studied Pāli grammar knows that it is a composite language. Grammatical suffixes are (mercifully) simpler than in Classical Sanskrit, but there are numerous alternative forms of declensions, such as ablatives in -ā, -asmā, and -amhā. Pāli shows clear influence from at least two Middle Indic languages. For example, forms like seyyathā (Skt tadyathā) or yebhuyya (Skt yad bhūya) do not conform to the general rules of Pāli phonology. Se and ye derive from Sanskrit pronouns tad and yad; and in Pāli we expect, and generally find, so and yo). Such forms are currently explained as coming from the "Māgadhī Prakrit" since parallels are found in the Asoka Edicts associated with Magadha.
Karpik suggests that a good analogy for the varieties of language spoken in the North India ca 400 BCE would be US versus British English. This clearly does not work for extant Gāndhārī and Pāli texts written down some centuries later (i.e. the actual evidence). The relationship between these two is more like that between the Scandinavian languages. A Swede and a Norwegian can converse without too much difficulty and both can read Danish. However, they struggle to understand spoken Danish. Similarly, a working knowledge of Pāli is not sufficient to read Gāndhārī (I've tried), and as spoken languages the two were probably mutually unintelligible. One has to specifically learn Gāndhārī in order to understand it.
Pāli also shows signs of influence from Sanskrit, both in loan words such as brāhmaṇa and in Sanskritised grammatical inflexions. The Brahmanical influence on Buddhism is obvious, and easily explained by pointing out that many of the legendary followers of the Buddha are said to be Brahmins, not least Sāriputta and Moggallāna. It's also evident that Buddhists felt they had to compete with Brahmanism to some extent, and hence Pāli suttas are constantly pointing out the faults of (non-Buddhist) Brahmins. Such critiques are far more common and more thoroughgoing than, say, critiques of the Nigaṇṭhā sect. Here again, the Nigaṇṭhā sect is identified with Jainism, but never referred to as such in Pāli, another speculation often treated as an established fact.
Some attempts seem to have been made in antiquity to standardise the language of the suttas, but some parts of the Suttapiṭaka seem to have failed to undergo this same process. For example, we find numerous “Māgdhisms” in parts of the Suttanipāta. While the retention of odd inflexions is asserted to be evidence of antiquity, it is equally plausible to me that the text is the same age as all the rest, but simply escaped the rather clumsy standardisation we see elsewhere. While it may have been canonised late, reflected in its status as a miscellaneous text, this does not make the Suttanipāta "early".
Moreover, despite the emic view, the Theravāda sect itself does not really go back to the mythical First Council (weirdly, these councils are routinely treated as historical, even by sceptics). An etic view of the Theravāda tradition tells us that is a late an offshoot of the Vibhajjavāda movement and has undergone repeated reinvention. The ordination lineage of Sri Lanka died out twice and had to be reintroduced from Burma. The Sri Lankan Theravādins embraced both Mahāyāna and Tantra before Medieval purges created the reformed movement that we now think of as Theravāda. This movement is largely focussed on Abhidhamma thought as expressed in Buddhaghosa’s Visuddhimagga and later commentarial works composed in Sri Lanka and Burma. There is nothing very “original” about Theravāda Buddhism. Like other Buddhist sects, Theravādins moved away from reliance on buddhavacana; preferring teachings closer to their own time. We should also note that while the rubric “Theravāda” is often used in an essentialised, monolithic way, there are Theravāda lineages that don’t recognise each other’s ordinations. We have to be wary of Buddhist modernist claims, even when they come from seemingly orthodox quarters.
Conclusions
Stefan Karpik makes some interesting linguistic arguments, some of which may well change how we view the history of Pāli, though experts in Middle Indic languages seem to be unpersuaded to date. This is not my area of expertise, so I can only wait with interest to see how this field develops. I am certainly open to his conclusions and sympathetic with this aspect of his project.
If my experience with the Heart Sutra is any indication, Buddhist Studies experts (including those focussed on philology) can be completely wrong about important things. Literally everyone was wrong about the Heart Sutra , for example. It happens.
That said, when Karpik shifts from drawing linguistic inferences to drawing historical inferences, his methods are fundamentally flawed and his conclusions appear to simply repeat his own pre-existing beliefs and prejudices. When examined, these assumptions and biases vitiate all of his attempts at revising history in the direction of modernist Theravāda orthodoxy. These assumptions include belief in the historicity of the Buddha, belief that the historical Buddha spoke Pali. We also have to include the two contradictory beliefs that we can take the Pāli literature at face value and that we can, at the same time, exclude all the supernatural elements of that literature. There are more unexamined assumptions about how later evidence may be interpreted as evidence of an earlier time.
What's missing from Karpik's articles is any evidence whatever from the relevant time or place as he defines it, i.e. from Northern India in the fifth century BCE.
Those of us who write about the history of Buddhism must pay attention to the methods of modern historiography. We cannot, for example, simply plough on without any attempt to identify and counter our own manifest biases. Part of the problem is the conceit that an education in philology makes one an expert in historiography, anthropology, and archaeology. It does not. To paraphrase Mary Midgley (1979), in the field of Buddhist Studies there is now no safer occupation than talking bad history to philologers, except talking bad philology to historians.
As noted above, the very great irony here is that Karpik's views on Pāli are compatible with virtually any view on the historicity of the Buddha. It wouldn't make any difference at all to the linguistic argument if Karpik simply dropped the issue of "the Buddha's language" entirely. And it would make such arguments infinitely more plausible if he did.
These problems should have been picked up by an academic editor or in peer-review and addressed prior to publication. Unfortunately for Karpik, his editor shares exactly the same biases and prejudices, so he seems not to have been challenged on what seem to me to be egregious methodological errors. The OCBS may wish to consider whether it wishes to publish an academic journal or some other kind of publication. If JOCBS is an academic journal then academic standards apply. The editor should not use the journal as a vehicle to promote one religious sect or any religious views. Articles with obvious, unaddressed bias should be sent back to be revised, especially if they otherwise merit publication.
I opened with a famous quote from (the young) Wittgenstein: "What can be said at all can be said clearly, and whereof one cannot speak, thereof one must be silent.”
I have endeavoured here to say clearly what can be said. Historians cannot speak of "the language of the Buddha", since we do not and, in all probability, cannot know what language he spoke (or if he was even a real person). We can only speak of the language of the texts that have come down to us. And by Wittgenstein's dictum we must not speak of "the language of the Buddha", except to say "we don't know what language the Buddha spoke". If we wish to speculate beyond the evidence, this must be clearly marked and distinguished from facts, and cannot be subsequently relied on as an established fact.
Assumptions, knowledge, belief, and speculation have to be clearly distinguished and identified for the readers of academic articles. No one reads an academic article to find out what the author believes; we read them to find out what the author can prove.
Finally, I want to emphasise that mine is an epistemic claim, not a metaphysical claim. The message is "we don't know" not "he/it didn't exist". With my historian hat on, I have no opinion on the existence of the Buddha. One may speculate on such metaphysical issues, but one should not try to pretend that such speculations amount to history.
Ironically, given the amount of ink spilled and the apparently strong feelings on the matter, in the end, the issue of what language the Buddha spoke has little historical significance. It appears to be raised only in furtherance of an agenda that seeks to legitimise a religious view of the past. While religious Buddhists lap this up, those of us who participate in the academic discussion of the history of Buddhism have an obligation to pay attention to and use established historical methods.
~~oOo~~
Bibliography
JOCBS = Journal of the Oxford Centre for Buddhist Studies
Allon, Mark. (2021). The composition and transmission of early Buddhist texts with specific reference to sutras. (Hamburg Buddhist Studies 17). Numata Centre for Buddhist Studies.
Bucknell, Roderick S. (2022). Reconstructing early Buddhism. New York, NY, USA: Cambridge University Press.
Davidson, Ronald M. (2002). Indian Esoteric Buddhism: A Social History of the Tantric Movement. New York: Columbia University Press.
Drewes, David. (2017). "The Idea of the Historical Buddha". Journal of the International Association of Buddhist Studies 40: 1-25.DOI: 10.2143/JIABS.40.0.3269003
Falk, Harry and Karashima, Seishi. (2012). “A First‐Century Prajñāpāramitā Manuscript from Gandhāra – parivarta 1 (Texts from the Split Collection 1).” Annual Report of The International Research Institute for Advanced Buddhology 15: 19–61.
——. (2013). “A First-Century Prajñāpāramitā Manuscript from Gandhāra – parivarta 5 (Texts from the Split Collection 2).” Annual Report of The International Research Institute for Advanced Buddhology 16: 97–169.
Heim, Maria & Ram-Prasad, Chakravarti (2108). "In A Double Way: Nāmarūpa In Buddhaghosa's Phenomenology." Philosophy East & West 68(4): 1085–1115.
Hinüber, Oskar von (2019). "The Buddha as a Historical Person." Journal of the International Association of Buddhist Studies 42: 231-264.
Jones, Dhivan Thomas. (2022). "This Being, That Becomes: Reconsidering the Role of the imasmiṃ sati Formula in Early Buddhism." Journal of the International Association of Buddhist Studies 45: 119-155.
Karpik, Stefan. (2019a). “The Buddha taught in Pali: A working hypothesis”. JOCBS 16: 10-86.
——. (2019b). "A Reply to Bryan Levman’s The Language the Buddha Spoke." JOCBS 17: 106-116.
——. (2022) "Pali Facts, Fictions and Factions." [A review of Levman, Bryan G. (2021), Pāli and Buddhism: Language and Lineage.] JOCBS 22: 121-141.
——. (2023). “Light on Epigraphic Pali: More on the Buddha Teaching in Pali.” JOCBS 23: 41–89.
Levman, Bryan G. (2008). "Sakāya niruttiyā revisited." Bulletin D'Etudes Indiennes 26-27: 33-51
——. (2010a). "Aśokan Phonology and the Language of the Earliest Buddhist Tradition." Canadian Journal of Buddhist Studies 6. 57–88.
——. (2010b) "Is Pāli closest to the western Aśokan dialect of Girnār-rev". Sri Lankan International Journal of Buddhist Studies 79-108.
——. (2016). "The language of early Buddhism." Journal of South Asian Languages and Linguistics 3(1): 1-41.
——. (2019). "The language the Buddha spoke." JOCBS, 17: 63-105
——. (2020a). "Sanskritization in Pāli." Journal of South Asian Languages and Linguistics 7(1): 105-149.
——. (2020b). "Sanskritization in Pāli". Journal of South Asian Languages and Linguistics 7(1): 105-149. https://doi.org/10.1515/jsall-2021-2030
——. (2021). Pāli and Buddhism: Language and Lineage. Newcastle-upon-Tyne: Cambridge Scholars Publishing.
——. (2022). "The Influence of Proto-Dravidian on Indo-Aryan Phonology, Morphology and Syntax, Part 1" International Journal of Dravidian Linguistics 50(2): 74-113.
——. (2023). "The Influence of Proto-Dravidian on Indo-Aryan Phonology, Morphology and Syntax, Part 2" International Journal of Dravidian Linguistics 52(1): 14-58.
Samuel, Geoffrey. (2008). The Origins of Yoga and Tantra: Indian Religions to the Thirteenth Century. Cambridge University Press.
Schopen, Gregory. (2004). "If you can’t Remember, How to Make it up, Some Monastic Rules for Redacting Canonical Texts." In Buddhist Monks and Business Matters, Still More Papers on Monastic Buddhism in India, 395-407. Honolulu: University of Hawai’i Press.
Shulman, Eviatar. (2008). "Early Meanings of Dependent-Origination." Journal of Indian Philosophy 36(2): 297-317.
Silk, Jonathan (2015). "Establishing / Interpreting / Translating: Is It Just That Easy?" JIABS 36/37: 205-226.
——. (2021). "Editing without an Ur-text: Buddhist Sūtras, Rabbinic Text Criticism, and the Open Philology Digital Humanities Project." Annual Report of the International Research Institute for Advanced Buddhology 24: 147-163
——. (2022) "Thinking About the Study of Buddhist Texts: Ideas from Jerusalem, in More Ways Than One." Journal of Indian Philosophy 50: 753–769
Stargardt, Janice. (1995). “The Oldest Known Pali Texts, 5–6th century: Results of the Cambridge Symposium on the Pyu Golden Pali Text from Śrī Kṣetra, 18–19 April 1995.” The Journal of the Pali Text Society 21: 199-213.
Sujato and Brahmali. (2013). "The Authenticity of the Early Buddhist Texts." JOCBS supplement to Vol. 5.
Wynne, Alexander. (2005). “The Historical Authenticity of Early Buddhist Literature: A Critical Evaluation.” Vienna Journal of South Asian Studies. 49: 35-70
——. (2006). "The Historical Authenticity of Early Buddhist Literature: A Critical Evaluation." Vienna Journal of South Asian Studies 49: 35-70.
——. (2019a). "Did the Buddha Exist?" JOCBS 16: 98-148.
——. (2019b). "Once more on the language of the Buddha." JOCBS 17: 8-10.
