Did the Buddha Speak Pāli?

Richard Gombrich, Buddhism and Pali, Mud Pie Books, 2018

It’s an interesting question. The Pāli canon preserves the one complete surviving collection of the discourses of the Buddha in an ancient Indian language; the Buddha most have spoken some such ancient Indian language; so did the Buddha speak Pāli? The scholarly consensus for the last few decades has been ‘no’: whatever language the Buddha originally spoke, Pāli is a later literary construction. What’s at stake is not just our sense of proximity to the person of the Buddha, though that is something with strong emotional resonance for many of us – it’s also a matter of judging how close the Pāli discourses might be in sound and concept to their supposed origins in the teaching of the Buddha. Richard Gombrich’s new book, Buddhism and Pali, proposes a bold new hypothesis, that the Buddha did speak Pāli – in fact, Pāli is a kind of argot or specialist language devised by him to pass on his distinctive teachings. 

For many years Professor Richard Gombrich has not only been a creative force in Buddhist Studies scholarship, but also a source of support and encouragement to other scholars. One of his achievements has been to devise and teach a Pali course through which probably hundreds of people have by now got started in studying this ancient Indian language, the canonical language of the Theravāda school of Buddhism.[i] Richard was a patient teacher when I started out learning Pali, in 2006, auditing his course at SOAS, and I am grateful for that good start. So what is Pāli? In his new-ish (2018) book, Buddhism and Pali, from the Oxford publisher Mud Pie Books, not only does Richard answer the question of what Pāli is exactly, but he goes on to make the argument that Pāli is a kind of lingua franca or common language that the Buddha himself developed as a means to teach in a consistent way among the different dialects and languages of ancient north India. Gombrich calls Pāli an ‘argot’: a specialist set of idioms and terminology for communicating the Buddha’s teaching. From the point of view of modern historical scholarship, this is a bold hypothesis. It implies the idea that the Buddha spoke Pāli. Is this true? While Gombrich does not himself try to prove his hypothesis, it turns out that there is some new scholarly work that lends support to his bold claim. I’ll turn to that after a short review.

In Chapter 1 of Buddhism and Pali, Gombrich gives the reader some history. The word pāli originally just means ‘a text’, as in a text or passage in the tipiṭika, the collection of the Buddha’s teachings, as distinguished from a text or passage of commentary.[ii] Hence it was used in the phrase pāli-bhāsa, ‘the language (bhāsa) of the texts’, and this was abbreviated to pāli, which hence became the name of the language. Theravādin Buddhists themselves believed that this language was the one spoken by the Buddha, though they called it magadhī, the language spoken in the ancient country of Magadha, in north-east India. In fact Buddhaghosa, the great Theravādin commentator, argued that magadhī was the mūla-bhāsa or ‘root language’: if children were not taught a different language, they would spontaneously speak magadhī. But, as Gombrich observes, this pious view is actually at odds with the Buddha’s own view, that language is conventional. From the point of view of modern linguistics, Pāli is a Middle-Indo-Aryan language (a ‘prakrit’), based on one or more spoken languages or dialects of the time of the Buddha, which appears to have undergone complex processes of formalisation while it was being transmitted orally and then in writing. Indeed, the Pāli of the Pāli canon appears to have been ‘Sanskritized’ – worked over to make it more like classical Sanskrit. Gombrich himself has in the past described Pāli as an artificial language.[iii]

In Chapter 2, Gombrich gives a lovely, readable summary of the inner workings of Pāli language and literature, based on his own familiarity with teaching it to newcomers. In Chapter 3, he emphasises the role of memorisation in how the discourses were composed and passed on, something which has completely shaped the present form of Pāli language and literature. He uses the Pāli words of the dhamma-vandanā to go into some more depth:

svākhāto bhagavatā dhammo

sandiṭṭhiko akāliko ehipassiko opanayiko

paccattaṃ veditabbo viññūhī ti

His analysis of this formulation is rich, and is designed to show the value of being able to read the Buddha’s teaching in its original language. He also comments on the incredibly repetitious nature of Pāli canonical discourses, following the explanation by Sujato and Brahmali that the repetitions probably go back to the Buddha’s own teaching style, and are not just the result of transmission processes.[iv]

It is in Chapter 4 that Gombrich puts forward his bold hypothesis. He begins by endorsing Sheldon Pollock’s view that the Buddha deliberately chose to teach in local, spoken languages that were not Sanskrit – not associated with the values of Brahmanism. He goes on to argue that the Buddha developed a teaching language that was intelligible across northern India, where a range of different dialects of Middle Indo-Aryan languages were spoken. This language was what became known as Pāli, which is why it contains features from different dialects, variations of grammar and vocabulary, built on the basis of the language of Kosala, where the Buddha did most of his teaching. Later Buddhists preserved this argot, this teaching language of the Buddha and his disciples – but not because there is anything special about Pāli. Rather, part of the ideology of the Buddha’s language is that it is merely the conventional vehicle for his teaching, part of which is that there is no unchanging essence of things. Nothing in the Buddha’s teaching, certainly not his choice of language, should be taken too literally. 

Gombrich’s hypothesis is intriguing, but is there any evidence for it? The prevailing scholarly opinion has been that Pāli is one of the several languages by which the Buddha’s teaching was transmitted by oral recitation, and the Buddha’s own language is unknown, or perhaps varied depending on who he was talking to. In an article of 2019, and citing Gombrich, Stephan Karpik has argued for the very opposite, that the oral transmission of the Buddha’s teachings was in one language, which we know as Pāli, which was the language in which the Buddha taught.[v] Karpik is careful not to overstate his case. It is not that there is any direct evidence for his hypothesis, but he makes a series of strong arguments against the prevailing scholarly consensus of multiple dialects for the early oral transmission. These arguments include the implausibility of ‘translating’ the teachings into these different dialects, given that they may have been mutually intelligible to anyone who moved around in ancient north India, like the various dialects of English in the period before mass media.

Both Karpik and Gombrich have to come up with a way of interpreting an important passage in the Cūlavagga of the Pāli vinaya in which the Buddha rebukes two monks who wish to put his teaching chandaso (which may mean, into Vedic metre or formal verse). Instead, the Buddha says the monks should each learn his teaching sakāya niruttiyā – a phrase often previously understood as ‘in one’s own (sakāya) dialect (niruttiyā)’. Karpik disproves this reading and argues that sakāya niruttiyā probably means a ‘way of speaking’, which he takes to mean the opposite of the elevated register of metrical composition. Whatever the phrase means, there is little support for the idea that the Buddha said that his teaching should be put into various local dialects. Karpik instead advances various arguments in support of the admittedly hypothetical idea that the early Buddhist teachings were preserved in the one language developed by the Buddha for the purpose, what we now call Pāli.

The positive case for Gombrich’s hypothesis seems to be shaping up. But an article by Bryan Levman, published in 2019 following that of Karpik, adds a strong note of caution.[vi] Levman is sympathetic to Gombrich and Karpik and their hypothesis, but he presents a wealth of linguistic evidence to show that the Pāli language as we know it, even in its very earliest forms, is underlain by an earlier language, now only recoverable in part by inference, and it is this earlier language, which is not Pāli, which is the hypothetical teaching language of the Buddha. Levman calls this earlier language a koine, which is another word for the kind of inter-dialectical lingua franca that Gombrich calls an ‘argot’. This koine would have been invaluable for teaching in a large area with many different dialects and variations of vocabulary. But it would have been different from the language we know as Pāli. For example, the Pāli word brāhmaṇa would appear to be a Sanskritized version of the word in the earlier language, given that the Buddha is recorded in Dhammapada v.388 as explaining that a (Pāli) brāhmaṇa is one who is bāhitapāpa, ‘having removed his evil’. But this word-play on bāhita only works if the word brāhmaṇa was originally something more like *bāhmana. Karpik (p.57) argues instead that brāhmaṇa was a loan-word from Sanskrit into Pāli, but this makes nonsense of the Buddha’s word-play. Levman (p.71) argues that the original word was *bāhaṇa. The issue illustrates the difficulty in reconstructing the linguistic processes through which Pāli developed, and the kind of changes to sound and meaning that they imply.

So did the Buddha speak Pāli? Probably not exactly. But there is reason to suppose that he did develop an argot, a koine, a form of language that would have been intelligible across the many dialects of ancient north India, in order to develop his teaching in a way that could be remembered and passed on. And one form in which that language has been preserved is what is became known as Pāli. And since the discourses preserved in Pāli give us the one complete surviving account of the early teachings, even if we no longer have access to the language the Buddha spoke, Pāli is our main witness for that language, representing the best effort of the early Buddhists to preserve the actual words of the Buddha as they remembered them.


[i] The course started life as a 10-day intensive, but has now gone online: https://ocbs.org/courses/.

[ii] Margaret Cone, A Dictionary of Pāli: part III p–bh, Pali Text Society, 2020, pp.450–2. 

[iii] Richard Gombrich, ‘Introduction: What is Pāli?’ in Wilhelm Geiger, Pāli Grammar, trans. Batakrishna Ghosh, rev. K.R. Norman, Pali Text Society, 1994.

[iv] The work, which Gombrich strongly endorses, is: Bhikkhu Sujato and Bhikkhu Brahmali, The Authenticity of the Early Buddhist Texts, Supplement to Vol. 5 of The Journal of the Oxford Centre for Buddhist Studies, 2013.

[v] Stephan Karpik, ‘The Buddha Taught in Pāli: A Working Hypothesis’, Journal of the Oxford Centre for Buddhist Studies, 2019, vol. 16, pp.10–86. This article is long but thorough and readable, and well worth studying.

[vi] Bryan Levman, ‘The Language the Buddha Spoke’, Journal of the Oxford Centre for Buddhist Studies, 2019, vol. 17, pp.63–105. This article is more technical but also highly worthwhile.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s