Pali Facts, Fictions and Factions (Review Article)

Journal of the Oxford Centre for Buddhist Studies Volume 22 November 2022 ISSN: 2047-1076 Published by the Oxford Centre for Buddhist Studies www.ocbs.org Wolfson College, Linton Road, Oxford, OX2 6UD, United Kingdom Authors retain copyright of their articles Editorial Board Dr Alexander Wynne (Editor): alexwynne@outlook.com Prof. Richard Gombrich (Editorial Consultant): richard.gombrich@balliol.ox.ac.uk Prof. John Holder: john.holder@snc.edu Dr Tse-fu Kuan: jeformosa@yahoo.com All submissions should be sent to: alexwynne@outlook.com Assistant Editors Dr Nicolas Revire Dr Aleix Ruiz-Falqués Mr J. Rotheray Production Team Operations and Development Manager: Mr Steven Egan Journal production & cover illustration: Mr Ivan de Pablo Bosch (www.ivancious.com) Annual Subscription Rates Students: £20 Individuals: £30 Institutions: £45 Universities: £55 Countries from the following list receive 50% discount on all the above prices: All African countries, Bangladesh, Cambodia, India, Indonesia, Laos, Myanmar, Nepal, Pakistan, Sri Lanka, Thailand, Vietnam For more information on subscriptions, please go to www.ocbs.org  Contents List of Contributors vi Guidelines for Contributors viii Editorial. Alexander Wynne xi Articles The Sahassavatthupakaraṇa II Peter Masefield 1 The Syntax of Disagreement Ole Holten Pind 23 Suicide by Fire: How the Indian Ascetic Kalanos Was Mistaken for a Buddhist Bhikkhu Sujato 27 The Rehabilitation of a Japanese Buddhist Heretic Brian Victoria 46 Suicide: An Exploration of Early Buddhist Values Alexander Wynne 83 REVIEW ARTICLE Pali Facts, Fictions and Factions Stefan Karpik (Bryan Levman, Pāli and Buddhism: Language and Lineage, 2021) 121 BOOK REVIEWS Buddhist Ethics: A Philosophical Exploration. (Garfield, Jay L., 2022) Reviewed by John J. Holder 142 Translating Buddhism: Historical and Contextual Perspectives (Alice Collett, ed., 2021) Reviewed by Sarah Shaw 155  Review Article Pali Facts, Fictions and Factions Stefan Karpik Levman, Bryan G., Pāli and Buddhism: Language and Lineage. Newcastle-upon-Tyne: Cambridge Scholars Publishing, 2021, 462 p., hardback, £67.99, ISBN: 9781527575551 In this nicely printed collection of essays by Bryan Levman, there is useful work on the influence of non Indo-Aryan languages on Pali, on inferences of cultural borrowing, on the influence of Dravidian grammar on Pali and on the original meaning of sati. The essay on the correct pronunciation of the anusvāra/niggahīta was less impressive, and I was not at all convinced by a major thread running throughout this book, viz., Levman’s koine theory, which, I regret to say, I still consider to be fantasy sociolinguistics. This review article is intended to examine and discuss the salient, as well as the contentious, points found in Pāli and Buddhism: Language and Lineage. The influence of non Indo-Aryan vocabulary on Pali This is the largest part of the book, spread across Chapters 2 and 3 (pp. 16–182). Levman’s salient point is (p. 40): “in the case of the IA [Indo- Aryan]-indigenous interaction, pervasive linguistic and structural borrowing do indeed mirror a strong cultural influence”. He identifies several hundred 121  Review Article Pali words incorporated from Dravidian or Munda languages and infers cultural borrowing from them. He shows borrowings into Pali from non Indo- Aryan words in passages concerning: robe-making, their dyeing and their repair (pp. 19–31, 59–73, 140–149); the brahmanisation of the jaṭila, Kanḥa, into the purohita, Asita, at Suddhodana’s court (pp. 45–59); and Dhamma words such as sīmā, piṇḍa, phala, sīla, paṭhati, māla, mūla (pp. 73–79). From the Mahāparinibbānasutta, there are yakkha names attached to shrines (pp. 83–88), toponyms (pp. 88–103), the Buddha’s final meal (pp. 103–107), and funeral rites for the Buddha (pp. 113–124). An appendix has selected derivations of 17 words such as āgāra, sāla, kaṭhina (pp. 152–182). There is also a claim of the importation of indigenous culture into Buddhism in the form of snake/tree worship, funeral practices and political organisation (pp. 44, 51): the Buddha himself was called a nāga (p. 100); sāla, the Sal tree, under which the Buddha was born and died, is claimed as the totem of the Sakya tribe instead of the teak (pp. 52, 164); funeral rites for the Buddha are shown to be non-Aryan (p. 113); the Buddhist order was organised like the tribal assemblies (pp. 80–83). Although the topic is generally interesting, many parts seem redundant. Levman acknowledges (pp. 35, 45, 133) that toponyms and the names of local flora and fauna new to Indo-Aryan immigrants are loanwords and do not necessarily indicate cultural borrowing into Buddhism. Nonetheless there are pages of irrelevant detail on exactly that: thirty pages (pp. 83–103) given over to shrines devoted to yakkhas and toponyms with non Indo-Aryan names, plus sections on mayūra, a peacock (pp. 171–173), and tumba, a gourd (pp. 173–174). The author claims (pp. 45, 133) that there is an exemption for loan words and toponyms if they occur in a specific cultural or religious context, citing Franklin Southworth (2005: 122–123), who argues that religious word borrowing indicates a higher degree of linguistic convergence. While I agree with Southworth, this proposition does not offer Levman an exemption as it does not claim that religious word borrowing indicates actual cultural borrowing. Levman concludes for toponyms found in the Mahāparinibbānasutta, all outside the Buddha’s Sakyan tribal land (p. 88): “The place names […] tell us a great deal about the Buddha’s cultural background”. However, he does not explain what they tell and scepticism must remain. For example, the existence of Latin castra in English place names, such as Manchester and Lancaster, and in Welsh place names, such as Caerphilly and Cardiff, does not mean that Latin is the first language of any 122  Review Article 21st-century British people or that they wear togas. Similarly, throughout the huge sprawling Chapter 3 “The Buddha’s Autochthonous Heritage” (pp. 42–182), there are scattered many etymologies of non Indo-Aryan personal names and words such as kuṭa, mukha, kula, with no obvious connection to Buddhism. It is often hard to follow Levman’s argument; for example, when the gods rain down four kinds of flower and three kinds of incense on the Buddha’s palanquin, Levman states (p. 117): “Virtually all of these flowers and incenses are native words, suggesting that they have some ritual significance in the story”. The words are uppala, padma, kumuda, pundarika, agaru, tagara and candana, for which Levman suggests only non Indo-Aryan origins. Is he suggesting that there were Indo-Aryan alternative names for these plants native to India? It appears that Levman has wrongly inferred cultural borrowing from words for which there was no Indo-Aryan alternative such as toponyms, personal names, and names of fauna and flora. Furthermore, he admits (p. 133) that another reviewer has commented that English has many Latin words, but that does not mean that the English have imported Roman customs. Then he continues (p. 133): “therefore the inference that usage [of non Indo-Aryan terms] means that an adoption of customs may be unproven and perhaps unprovable”. However, that admission has not constrained Levman’s enthusiasm for etymology. Despite these reservations, the reviewer was impressed in some places. Levman’s methodology for etymology (pp. 31–36) comes from several authorities, including Burrow (1946: 13–18) and Witzel (1999: 3–5), with supplements from Levman himself, and seems very sound. Others have noted word and cultural borrowing, but Levman’s unique contribution is that he links the two with several examples. In particular, he presents connected passages of Pali on robe practices, in which the surprising scale of the non Indo-Aryan word borrowing in Pali is evident. Such passages to my mind prove cultural borrowing because of the sheer density of word borrowing for which alternative Indo-Aryan vocabulary must have been available. Overall, Levman is convincing regarding indigenous language and cultural borrowing in robe-practices and some Buddhist vocabulary, but he has greater ambitions. He is laying the groundwork for historians to investigate the proposition of: “an autochthonous origin of Buddhism, appropriated by the Indo-Aryan immigrants and translated into MI [Middle Indic]” (p. 132). Frankly, it is doubtful that he will succeed in this goal because Buddhism obviously also has Aryan influence, which this book does not discuss at all.1 However, the search has been productive. 1  For example, the facts that Buddhism has the third precept of brahmacariya, that Brahmā 123  Review Article The influence of Dravidian grammar on Pali Chapter 4 (pp. 183–209) is, perhaps, the most important contribution in the book. It backs up the early claim made above of structural (i.e., syntactical) borrowing (p. 40) by comparing Buddhaghosa’s opening verses to his Dīgha- nikāya commentary with some verses of the Old Tamil Buddhist epic Maṇimēkalai, both written in South India around the 5th to 6th centuries CE. The Tamil is parsed and translated and comparisons are made regarding: (a) strings of absolutives/participles with a single main verb at the end; (b) participial constructions replacing relative-correlative constructions; (c) constructions of the type, paṭhamajjhānam upasampajja viharati (Geiger 1943/1994, §174.5), which apparently is common to all Indic languages (p. 202); (d) a dative-like genitive; and (e) absolutives used as postpositions. The reviewer found the correspondences to be remarkable, and the author commendably shows that these features are also found in the Pali Canon. I observe the increased use of absolutives as a salient difference in style between canonical Pali and the story-telling of the Dhammapada commentary as well as the Jātakas. Even though Levman does not draw any inferences from such tendencies, he does refer to the Tamilisation of Pali (p. 201), and he may have proved his point successfully. I still have a mental caveat, however, that the languages may have been converging, and wonder if it might also be true to speak of a “Palicisation” of Old Tamil, especially Buddhist Old Tamil. I hope Levman will clarify that aspect in the future. The meaning of sati in the Burmese tradition In Chapter 8 (pp. 310–356), Levman believes the original meaning of sati as “memory” is being lost in Western secular mindfulness practice. To correct this, the author provides the entry for sati in the 24 (so far) volume Pāli–Myanmā Abhidan’ dictionary with a translation of the Burmese, an exploration of the references and an analysis of sati into 30 categories, thus providing a helpful resource for research. He does not recapitulate his 2018 debate with Anālayo in the journal Mindfulness, but aims to provide information on how sati was understood in the Burmese tradition Sahampatī asks the Buddha to teach, that the Buddha’s claim that one is a Brahmin by skilful action instead of by birth, that Lord Sakka, a renaming of the Vedic god Indra, is attending the Buddha on many occasions including his funeral, and so on. Levman partly rows back from this radical proposition by saying (p. 378): “Certainly, Brahmanical influences, especially Brahmin converts to the Buddha’s philosophy, played an important role, but it was not the whole story”. 124  Review Article that gave birth to its modern Western counterpart. He concludes (pp. 355–356) that sati, according to this dictionary, includes a degree of memory and is to be cultivated on the foundations of the Buddhist teachings of sila, samādhi and pañña as encapsulated in the Satipaṭṭhāna and other meditation Suttas. In this book, the author does not go into where that leaves Western secular mindfulness practice. Nasalisation in Pali: how to pronounce buddhaṃ saraṇaṃ gacchāmi Chapter 9 (pp. 357–376) includes a tour of many Sanskrit and Pali grammatical sources whereby Levman concludes, using an odd mixture of romanised Pali and International Phonetic Alphabet (IPA), that buddhã saraṇã gacchāmi should be the correct pronunciation of buddhaṃ saraṇaṃ gacchāmi if spoken slowly in separated speech (p. 375). Note that his buddhã and saraṇã have a tilde, not a macron, which indicate a nasalised vowel in IPA: Levman (p. 367) believes it is like a nasal vowel in French. However, he presents only Sanskrit authorities to support this, and I wonder if this pronunciation is a Sanskritism. At any rate, it was certainly rejected by the Vinaya commentary and other Pali sources advocating a closed mouth anusvāra/niggahīta, as we shall see below. The author claims that for separated speech, the Vinaya commentary advocated buddham saraṇam gacchāmi with final m consonants, and that this is a Sanskritisation or archaism (p. 373) because Middle Indic never has words ending in m. His argument is probably misreading the commentary and certainly is hard to follow. Firstly, Levman has a series of awkward mistakes in this section. For example, three times (pp. 372–373), he has Sv (Dīghanikāya commentary) with no reference where he surely means Sp (Vinaya commentary) mentioned some fourteen pages earlier (p. 359, n. 489).2 Secondly, Levman does not give the Vinaya commentary definition of the niggahīta, including its rejection of the pronunciation pattakallā for pattakallaṃ, which appears to regard Levman’s preferred pure nasal pronunciation as unacceptable for separated speech in formal 2   Levman makes the same mistake of confusing the commentaries on p. 362, but he cross- references to footnote 489, so this is less of a problem; the same mistake of Sv for Sp is found at p. 363, n. 494. Similarly, he refers to the Vinaya commentary, the Samantapāsādikā (Sp), on p. 375, although it is almost certain that he means the Sumaṅgalavilāsinī (cf. p. 358, n. 488), which incidentally is misspelt as Sumaṅgalalvilāsinī (p. 362), while Samantapāsādikā is misspelt as Samantalapāsādikā in the list of abbreviations (p. 383). 125  Review Article Sangha proceedings because it does not have an unopened mouth.3 Thirdly, although Levman is perhaps not unreasonably influenced (pp. 359, 366) by that commentary’s apparent approval of an m sound in separated word pronunciation,4 I believe it should be interpreted differently. Since m requires closing and release of the mouth, could it be that instead the niggahīta was pronounced with the mouth initially open for the preceding vowel then closing and remaining closed without release (avissajjetvā) until the airstream was ended? This would produce in the nasal cavity an aftersound, which is what anusvāra means, and it could also be considered a kind of (incomplete) m sound, although it could not be represented by the IPA symbol [m] or any other. It would meet several criteria of the anusvāra: it has an indeterminate status as not a pure vowel or a pure consonant;5 it is long/heavy (garu); it is 3   Vin–a Samantapāsādikā (Sp 7, 1399–1400): niggahitan ti yaṃ karaṇāni niggahetvā avissajjetvā avivaṭena mukhena anunāsikaṃ katvā vattabbaṃ. [...] vimuttan ti yaṃ karaṇāni aniggahetvā vissajjetvā vivaṭena mukhena anunāsikaṃ akatvā vuccati. [...] suṇātu me ti vivaṭena mukhena vattabbe pana suṇantu me ti vā esā ñattī ti vatabbe esaṃ ñattī ti vā avivaṭena mukhena anunāsikaṃ katvā vacanaṃ vimuttassa niggahītavacanaṃ nāma. pattakallan ti avivaṭena mukhena anunāsikaṃ katvā vattabbe pattakallā ti vivaṭena mukhena anunāsikaṃ akatvā vacanaṃ niggahitassa vimuttavacanaṃ nāma. “Niggahīta (restrained/nasal) means restraining the organs of articulation without release where it should be pronounced with a closed mouth nasally. [...] Vimutta (free/non-nasal) means by not holding still the organs of articulation and relaxing them, it is spoken with an open mouth without making a nasal sound [...]. Where suṇātu me should be pronounced with an open mouth, but suṇantu me is said, or where esā ñatti should be pronounced and esaṃ ñatti is said, the nasal pronunciation with an unopened mouth is called niggahīta pronunciation of vimutta. Where pattakallaṃ should be pronounced with a closed mouth and nasally, the pronunciation pattakallā with an open mouth without making a nasal sound is called vimutta pronunciation of niggahīta” (my translation and emphasis in bold). 4   Levman (p. 359, n. 489) offers this translation of Vin–a Samantapāsādikā (Sp 5, 969): imāni ca pana dadamānena, buddhaṃ saraṇaṃ gacchāmī ti evaṃ ekasambaddhāni anunāsikantāni vā katvā dātabbāni, buddhaṃ saraṇaṃ gacchāmī ti evaṃ vicchinditvā vā makāra-antāni katvā dātabbāni. “If one pronounces buddhaṃ saraṇaṃ gacchāmi in one continuous line, it is allowed to make a nasalization at the end (of each word), and if one pronounces buddhaṃ saraṇaṃ gacchāmi after breaking up the words, then it is OK to pronounce the end of each word as the sound -m”. In my interpretation, here anunāsika means the commonly made [ŋ] sound for ṃ, and makāra here means a sub-division of anunāsika, starting with m, but holding it, thus allowing air through the nose. So, in the context of anunāsika, makāra is shorthand for karaṇāni niggahetvā avissajjetvā avivaṭena mukhena anunāsikaṃ katvā (see above Sp 7, 1399–1400) if this commentary is consistent. 5   Allen (1953: 43, n. 4) quotes the Ṛkprātiśākhya I 5: anusvāro vyañjanaṃ vā svāro vā. I take this to mean: “The anusvāra can be either a consonant or a vowel”. Cf. Deokar (2009: 4): “According to Saddanīti 11: assarabyañjanto pubbarasso ca, which assigns the designation ‘garu’ to a short vowel not followed by either a vowel or a consonant as in ‘sukhaṃ’ and ‘isi’, niggahīta is neither a vowel nor a consonant”. 126  Review Article nasalised (anunāsika); it has an aftersound (anusvāra) in the nasal cavity; it restrains (niggahīta) the organs of articulation without release (avissajjetvā) while the airstream continues in the nose during the aftersound; it has an m mouth position (makāra), but without opening the mouth (avivaṭena mukhena); if a vowel follows niggahīta, it is written as m in connected speech, e.g., evameva instead of evaṃ eva, because release of the mouth in order to say the following vowel actually completes [m]. Levman himself quotes the Saddanīti (p. 363), while Deokar (2009: 3) quotes the Kaccāyanavaṇṇanā 1.1.8 and Thitzana (2016: 123, n. 9) comments on Kaccāyana Pāli Grammar, all confirming that the niggahīta is made with an unopened mouth.6 I infer that the sound described above is what Pali writers meant by an unopened mouth niggahīta, or else there would be no difference between ṃ and m.7 Levman does not consider this sound at all, so I am not convinced by his first preference of a pure nasal vowel in separated speech in Pali, contrary to these four Pali sources referring to a closed mouth. As for continuous speech, Levman would certainly also allow the commonly spoken [bʊ.dʰãŋ sarəɳãŋ gə.tʃʰa:mɪ] with final slightly nasalised vowels and velar nasals, as in English sing. However, he writes it (p. 375) as “buddhaŋ śaraṇaŋ gacchāmi (buddhaṅ śaraṇaṅ gacchāmi)”, which is Sanskrit in idiosyncratic notation and must be a mistake. After his investigations Levman finally concedes (p. 373) that the Buddha would be happy with a variety of pronunciations. This treatment of niggahīta contains sections on the diachronic development of nasality and the influence of non-native Indo-Aryan speakers; it is the longest that I know of, but it is still not a complete survey of this intricate subject. I am not convinced nasality is as clear-cut as Levman presents, and I defer to Allen (1953: 46),8 who, after his own analysis of nasality, comes to no final conclusions: “In view of their [ancient phoneticians] generally high standard of competence it seems fair to assume that the phonetic problem in question was a particularly difficult one [...]”. 6   Levman (p. 363, n. 496) translates avivaṭena mukhena as “with a not-open opening” in the belief it refers to the partial closure of the soft palate to make a nasalised vowel. However, Warder (1995: 2, 4) believes it refers to the closure of the mouth. 7   This pronunciation for slow, emphatic chanting has been heard by the reviewer at Wat Asokaram, Samut Prakan, Thailand in the 1970s. 8   Levman (p. 360, n. 492) refers to Allen (1953: 39, n. 5) but omits to list Phonetics in Ancient India in his references. 127  Review Article Levman’s koine theory The present reviewer argued recently that the Buddha taught in Pali (Karpik 2019a), to which the author (Levman 2019) advanced his koine theory, which I critiqued subsequently (Karpik 2019b). Levman (2020: 110, n. 10) said he would answer my criticisms in the present book, then forthcoming. Alas, it turns out that he ignores many of my points, although Prof. Richard Gombrich comes off far worse when the author states (p. 279): “Gombrich’s book does not provide an argument to justify his view [that the Buddha spoke Pali]”, thus completely ignoring the argument that the Buddha developed a composite dialect containing local variants (Gombrich 2018: 74–82). The koine theory (κοινή, koiné; lit. “common”) is principally argued in a reprint (pp. 236–274) of Levman (2016) and a new chapter, “The Evolution of Pali” (pp. 275–307). His thesis is that: in northeast India, the Indo-Aryan speakers were in a minority even during the time of the Buddha; the Buddha spoke Indo-Aryan as a second language; his Indo-Aryan language was pre- Pali; the pre-Pali was a koine existing in India in his time; Pali is a translation from this koine and other languages; Pali was subsequently Sanskritised extensively; and finally, his teachings in his original language are lost. I aim to show here that each of these claims is suspect and, taken as a whole, the theory is incorrect, overcomplicated and unhelpful. 1. “in north-east India, the Indo-Aryan speakers were a minority even during the time of the Buddha” (p. vii) This eye-catching claim is made without any evidence in the text (pp. vii, 16, 40, 169, n. 259,  371). However, his note 23 on p. 40 makes the banal point: “Initially at least the non-Indo-Aryan inhabitants of the sub-continent formed the plurality of the population”. It is hard to evaluate this argument because Levman has switched from northeast India to the subcontinent, which included areas when Aryans had never penetrated at the time of the Buddha; furthermore, “Initially” could predate the Buddha by centuries. This footnote then references Burrow (1955: 386), Emeneau (1980: 198), Sjoberg (1992: 61), Krishnamurti (2003: 15, 36), and Southworth (2005: 118–122), but this offers no clarity. Burrow refers to “a considerable element of Dravidian speakers”, which could be a considerable minority and is referring to the central Gangetic plain and the classical Madhyadeśa, which is not the northeast per se; Emeneau, Sjoberg, Krishnamurti and Southworth have nothing to say on the matter with no reference to the Buddha’s time or locality on the pages cited 128  Review Article nor anywhere else in these works that I can find. In short, Levman provides no relevant evidence to back up his repeated claim. The reason I found this claim eye-catching is that there is evidence against it; the Aśokan inscriptions had translations in the Northwest into Greek and Aramaic. If Levman’s claim were correct, there would surely have been Dravidian or Munda translations in the Northeast a mere 150 years after the Buddha’s demise, for example, on the Pillar Edict at Lumbinī, his birthplace in Sakyan tribal land. 2. The Buddha spoke Indo-Aryan as a second language Levman (p. 3) claims the Buddha’s people, the Sakyans, spoke Dravidian with a Munda substrate and that Middle-Indic was their second language. His evidence for this is (p. 4): “an infusion of autochthonous values into the Buddhist belief system”. Actually, I accept that there was such an infusion, but that does not mean the Buddha’s first language was necessarily Dravidian or Munda as is implied by Levman. By that logic we would infer from the borrowings of Latin or Greek language, mythology, and philosophy in Britain that the British have Latin or Greek as their first language and English as their second. This is patently not the case and cultural borrowing does not entail the wholesale borrowing of another language. The author further claims (p. 31): “He [the Buddha] could have spoken in both languages [i.e., Dravidian/Munda and Indo-Aryan] at different times and probably did”, for which he cites K.R. Norman (1980: 75), who refers only to different dialects of Indo-Aryan and does not support this idea at all. Levman also states (p. 237): “we can be fairly certain that they [the clans] spoke a non-Indo-Aryan language because most of the place names in the […] republics of the clans are non-IA in origin”; but by that same logic again, there would be no native English speakers in Wales or the Highlands and Islands of Scotland and that is clearly not the case. Levman also claims (p. 3): “there is no reason to believe that the Buddha only spoke in Middle Indic”. I can suggest a reason for it. In the Pali Vinaya, the Buddha pronounces on the disrobing procedure as follows: If he declares his resignation in Aryan to a foreigner and the latter does not understand, his resignation is not valid. If he declares his resignation in a foreign language to an Aryan […] and the latter does not understand, his resignation is not valid.9 9   Vin III 27–28: ariyakena milakkhukassa santike sikkhaṃ paccakkhāti so ca na paṭivijānāti: 129  Review Article This implies the Buddha considered Ariyaka, the Aryan language, to be the default language as it is the only one mentioned; he therefore speaks from the perspective of an Indo-Aryan speaker. This might suggest a situation like Britain, where the majority in Wales and Scotland speak only English and are not bilingual in Welsh or Gaelic. I am not claiming that this passage is conclusive proof, but, when combined with the lack of Dravidian or Munda translation in the Aśokan inscriptions, it is suggestive that the Buddha was very likely a native Indo-Aryan speaker. 3. The Buddha’s Indo-Aryan language was pre-Pali Levman states (p. 9): “Ever since Buddhaghosa announced that the Buddha spoke the language of Magadha (Māgadhī), which he considered identical to Pāli, this has been a controversial subject”. For me, it is controversial inasmuch as nowhere in the Pali Canon or the commentaries is “Māgadhī” mentioned; in fact, the commentaries studiously avoid that term, instead using expressions like magadhabhāsa and māgadhiko vohāro, while the Pali Canon has nothing remotely close to that term. In his commentary to the Vinaya passage above (§2), Buddhaghosa actually defines magadhabhāsa as equivalent to Ariyaka, the Aryan language, not a dialect, such as Māgadhī or Kosalī: Here “Aryan” means the Aryan language, the speech of Magadha; “foreign” means any non-Aryan language, Andha (Telugu), Damila (Tamil) and so on.10 “Magadha” with its capital situated at Pāṭaliputra comprised most of the subcontinent in Buddhaghosa’s time, in the form of the Gupta empire, and also in the earlier time of the Mauryan empire, when Mahinda, Aśoka’s son, brought Buddhism and early commentaries to Sri Lanka in the 3rd century BCE. I have argued for this broader sense of magadhabhāsa and Magadha previously (Karpik 2019a: 20–38); the late Ole Pind (2021) has also criticised the notion that the Buddha spoke Māgadhī. However, Levman (pp. 236–237) adopts the misreading, Māgadhī, and assumes Magadha at its smallest extent without responding to my argument. The author uses the apaccakkhātā hoti sikkhā. milakkhukena ariyakassa santike [...] sikkhaṃ paccakkhāti so ca na paṭivijānāti: apaccakkhātā hoti sikkhā (my translation). 10   Vin–a I 255: tattha ariyakaṃ nāma ariyavohāro Magadhabhāsā, milakkhukaṃ nāma yo koci anariyako Andhadamilādi (my translation). 130  Review Article considerable body of speculation fuelled by that misreading, e.g., Lüders’ Urkanon, Hinüber’s Buddhist Middle-Indic or Norman’s Old Māgadhī (pp. 236– 239), as a justification to insert his own version of the Buddha speaking some form of pre-Pali. Enter the koine. 4. Pre-Pali was a koine existing in India in the Buddha’s time Levman argues (p. 238) that the Aśokan dialects found on the Shābāzgaṛhī and Kālsī rock edicts11 were mutually unintelligible or not necessarily mutually intelligible (p. 292) and therefore a koine would have been needed in the Buddha’s time. I regard this argument as fantasy sociolinguistics for the following reasons: (a) Levman does not respond to my claim (Karpik 2019a: 58–64) that the differences in the Aśokan varieties were overwhelmingly one of accent and were therefore mutually comprehensible, in which case a koine would not be needed to promote understanding; (b) elsewhere (pp. 31, 60, 244, n. 375) the author argues for bilingualism and states (p. 244): “The mechanism which creates these shared features [lexical, phonological and grammatical features common to Old Indic, Dravidian and Munda] is extensive bilingualism [...]”, in which case again a koine would not be necessary; we know that in modern Belgium, Finland and Switzerland where there are respectively two, three and four official languages, a koine has not developed; (c) there is no written evidence for this koine, as might be expected in inscriptions, while on the other hand Epigraphic Prakrit is a reflex of Pali (Karpik 2019a: 52–53).12 11   See: https://www2.hf.uio.no/polyglotta/index.php?page=fulltext&view=fulltext&vid=362 &cid=381523&mid=634131 (accessed on November 8, 2022). 12   Dr Yojana Bhagat, Head of the Department of Pali, University of Mumbai, was asked in email correspondence with the reviewer why this standard inscriptional language for centuries is not called “Epigraphic Pali”. Her answer was that Indian scholars are generally ignorant of Pali. 131  Review Article 5. Pali is a translation from this koine and other languages Levman first claims (p. 31): “the conversion [of autochthonous technical terms] is not interdialectic, but a true translation of a local language into MI [...]”. He does make an exception for terms which had no equivalent in the receiving dialect, but does not consider the possibility of simple word borrowing here. Further, he claims (p. 59): “The Buddha certainly spoke other languages as well, including the language of the Sakya tribe, and one must assume he taught in that idiom, the proof being the large number of loan-words imported into MI”. This is quite illogical; it is similar to claiming that modern English speakers must also speak Latin, the proof being the large number of loanwords borrowed from Latin. He does not address my argument (Karpik 2019a: 12-19) that oral translation of the Buddhavacana was simply impractical, discouraged and unnecessary. 6. Pali was subsequently Sanskritised extensively According to Levman (p. 277): “Sanskritization of the Buddha’s teachings probably began right after his parinibbāna (post ~380 BCE)”. The author regards the pr, kr, tr, and ṣṭ clusters found in the Aśokan Girnār inscriptions13 as Sanskritisations and does not consider the possibility of their being retentions from Old Indic in this particular dialect. Oddly, he offers the existence of Prakritisms being Sanskritised in the Vedas as proof of a general proclivity towards Sanskritisation in Indian culture as if it were significant that Sanskrit was Sanskritised! He does not answer my arguments (Karpik 2019a: 53–58) that Pali has Vedic, non-Sanskritic features, which do not fit in with the Sanskritisation narrative. Instead, Levman quotes numerous scholars (pp. 238–239, 278–279, 290–291) who all claim Pali was Sanskritised, but I regard this as academic groupthink. There is an assumption with most advocates of Sanskritisation, with which Levman (p. 296) agrees, that Pali was originally more like the Aśokan Prakrits. However, I assume that, like Sanskrit and Ardhamāgadhī, Pali is not represented in these inscriptions although it existed at that time and, like its reflex, Epigraphic Prakrit/Epigraphic Pali, it was a formal conservative language, unlike the Aśokan Prakrits which represent the accents of local bureaucrats, messengers and stone-masons (Karpik 2019a: 58–64). 13   See: https://www2.hf.uio.no/polyglotta/index.php?page=fulltext&view=fulltext&vid=362 &cid=381524&mid=634132&level=2 (accessed on November 8, 2022). 132  Review Article To be fair, the author does engage in some technical arguments with the present reviewer, but there are always counter-arguments: a. As proof that Pali is an artificial language, Levman (p. 291) quotes Oskar von Hinüber alleging that katvā and disvā are artificial formations, and quotes Norman having the same issue with disvā and atrajā. From my perspective, their difficulty was misconstruing Magadhabhāsa as Māgadhī, and then trying to derive these forms from an eastern Aśokan Prakrit; as that cannot work, they resorted to artificial formations as an explanation, but Wilhelm Geiger was not so blinkered. Geiger calls katvā and disvā historical forms (§209), and atrajā a folk etymology (§53.2). I imagine he thought katvā < Old Indic (OI) kṛtvā (Geiger §12.1, 53.3), and disvā < OI dṛṣṭvā (Geiger §12.2, and perhaps a unique assimilation of s < ṣṭ).14 Levman further states (p. 293): “Norman argues that this view [Sanskritic forms in Pali are retentions] is simply ‘wrong’ (2006: 96)”. However, Norman bases his argument solely on atrajā, which he sees as a quasi-Sanskritic form, and ignores Geiger’s explanation (and the Pali-English Dictionary’s). Furthermore, backformations are a natural language process, as searching online with the terms “backformation” and “English” will confirm, and cannot prove Sanskritisation. b. Levman (2019: 80–81, n. 13) has already criticised my view (Karpik 2019a: 56–57) that the -tvā absolutive is a retention in Pali and not a restoration. He (pp. 293–294) does not openly dispute my argument (Karpik 2019b: 107–108) that over 13,000 14   In the 1943 edition, Geiger §59.4 notes dissā in Ardhamāgadhī (AMg) and refers to Pischel (1957: §334) who states that the regular form in AMg would be *diṭṭhā; Geiger appears to me to be arguing that there is an analogous, but unknown route in both P. and AMg from dṛṣṭvā to their respective reflexes. In the 1994 edition of Geiger, Norman derives disvā from the non-Pāṇinian form dṛśya via *dissa, which was later Sanskritised. My alternative is that, as Pali does not have the śy or sy cluster, dṛśya went straight from *diśya to disvā, without Norman’s intermediate *dissa, on the analogy of other tvā absolutives (Karpik 2019a: 56–57, 2019b: 107–108). Essentially, I argue that Norman’s “Sanskritisation” was really a natural backformation, much as the once incorrect verb “to administrate”, backformed from the noun “administration” (Latin noun administratio), is now used by some instead of the verb “to administer” (Latin verb administrare). 133  Review Article -tvā and 1,900 -tvāna absolutives in the Tipiṭaka overwhelm the handful of alleged -ttā absolutives in Pali (Pind lists 45, 2005: 499–508), which all have alternative explanations. However, my argument that the retained tv conjunct is also found over 2,000 times in tvam and over 400 in the sandhi tve- is dismissed as a “numbers game”. Levman then makes the puzzling statement (p. 293): “for if one looks at all the -tv- > -tt- assimilations in the canon (e.g. catvara > catur; -tvā > -ttā; tvaca > taca; satva > satta; to name a few), these far outnumber those that remain”. He does not present the results of his searches, however, to justify what for me is a plainly incorrect assertion. To me the fact that the -tv > -tt- assimilation is incomplete in the Tipiṭaka means that Pali was a natural language in which sound changes do not occur instantly in every instance and the sheer numbers argue against Sanskritisation. Levman does admit that the assimilation was not quite complete in Aśokan inscriptions and goes on to say (p.  294): “The commonality of tvam perhaps argues for its retention, but why then was it not kept in the other Prakrits?” My answer to that is that Pali was a conservative, formal language variety in which the tv conjunct persisted to a large extent and was preserved in its pre-Aśokan form in conformity to the Buddha’s wishes. c. I have argued elsewhere (Karpik 2019a: 57) that the Sanskrit brāhmaṇa is a loan word in Pali, not a retention. Levman investigates this and concludes (p. 296): “Of course it is always possible that both terms [OI brāhmaṇa and MI *bāhaṇa] were used alongside each other from the earliest time of the Buddha’s teachings, with the MI form being used in the gāthās and the OI form occasionally employed elsewhere for the reason Norman has suggested: to make it clear to both disciples and Brahmins, whom the Buddha was castigating”. If the word “occasionally” were deleted and speaking of were substituted for “castigating”, I would be in complete agreement with Levman’s conclusion. The use of the Sanskrit form could be a matter of politeness. 134  Review Article d. Levman (pp. 296–297) also discusses my claim (Karpik 2019a: 55–56) that the -bb- geminate being unique to Pali proves that it is archaic. He suggests that -b- and -v- were allophonic and it was merely a scribal convention that only -v- for -bb- is used in Aśokan inscriptions. I too have considered this possibility and also wondered if they are different representations of [β], the voiced bilabial fricative, which sounds halfway between b and v and may have been allophonic, with v for non-native Indo-Aryan speakers in instances like vy-. Despite these ruminations, I still think that my argument stands as: (1) -bb- is not found in Epigraphic Prakrit/Epigraphic Pali either; (2) the Sri Lankan manuscript tradition never alternates with -vv- although it interchanges vy- and by- in initial position; (3) I believe no manuscript tradition has, for example, *bā, *baṇṇa, *bibatta or *vandhati, *vāhu, *vīja, and there are many more examples where -b- and -v- are not interchangeable. I therefore think they were not allophonic, but were on occasions interchanged. e. Finally, Levman (pp. 298–300) does answer my point (Karpik 2019b: 109) that geminates do not undergo lenition by pointing out that non-native Indo-Aryan speakers might not be able to distinguish geminate and single consonants and so might introduce errors into the transmission. His point is valid, but not his conclusion that natural language processes and backformations are better explanations than manuscript errors for the variety of readings found, for example, at Dhp 335. This points to a larger problem with his koine theory: the koine reconstructions are extrapolated from variant readings and there is the issue that manuscript errors could be their basis. 7. The Buddha’s teachings in his original language are lost Levman additionally suggests (p. 59): “The Buddha then spoke and taught in several languages; that the only one that survived is Pali, which is apparently derived from a mixed MI interlanguage […], is just an accident of preservation”. 135  Review Article Later (pp. 292–293), he lays out an unconvincing argument that although the Buddha specifically forbade the use of Sanskrit, his disciples failed him by Sanskritising his teachings given in the koine and losing those in his native Sakyan language. This is not provable or disprovable, but seems unlikely. For, to echo Oscar Wilde in The Importance of Being Earnest (1895), to lose teachings in one language may be regarded as a misfortune, to lose teachings in both seems like carelessness. Levman’s implicit assumption is that because some teachings were extensively Sanskritised into Buddhist Hybrid Sanskrit, all early teachings of the Buddha were extensively Sanskritised. However, there is no robust evidence for the second proposition. 8. A misunderstanding In one passage of the book, Levman cites Karpik (2019b: 110) and states (p. 279) that what I took as natural fortitions in his theory are actually non-natural backformations. The author leaves the impression that I am at fault, not he. However, Levman (2019: 76–78) made a case for a degree of natural language change in Pali and he earlier stated: “this word [kañjiya] is straightforwardly derivable from *ga/u(N)hiya, with the fortition [my emphasis] of g- > k-” (2019: 90), thus using the terminology and notation of natural language change. “Fortition” is in bold to demonstrate that Levman (2020: 110) is completely inaccurate when claiming he does not use this word: “He [Karpik] calls the editing/revision/back-formation/Sanskritization process ‘fortitions’ (although I do not use the word)”. Whether Levman has adapted his theory in response to my earlier criticism of excessive fortitions in his theory or not, clarification is welcome. However, he continues to muddy the waters in this book by using the notation of natural language change for backformations, e.g., *veha > (vedha) > dvaidhā (p. 282), and even calling a backformation from roẏa a “fortition” (p. 288): “Pali preserves roga and pa-loka (idem) with a fortition [my emphasis] of -g- > -k-”; I have no idea why Levman writes “preserves” rather than “restores”, but I believe he means the latter if his clarification stands. To avoid confusion, in what follows, I will notate natural language change as > and revisions/non-natural backformations as →. 9. What use is Levman’s koine theory? The kindest thing that can be said of the koine theory is that it is an alternative explanation to transmission errors for variants in texts. However, it comes at the cost of believing that the majority of Pali words are Sanskritisations (Levman 136  Review Article 2020: 144) and such an extreme position is unnecessary to explain why Pali is as it is. Here are my comments on some types of alleged Sanskritisations: Key: bold = alleged koine form /AMg = Ardhamāgadhī / Aś = Aśokan Prakrit / OI = Old Indic / P = Pali Revision (Levman) Retention (Karpik) Comments OI loka >Aś, AMg loga → loka OI loka > P loka One of many Vedic forms retained in Pali (Karpik 2019a: 53). OI śata >AMg saẏa → P sata OI śata > P sata Retention of a simplified (pp. 286–287) (Geiger §3) Vedic form after OI s, ś and ṣ merged into P s. OI laghu > Aś, P lahu OI laghu > P lahu Levman regards the Pali as a (p. 287) (Geiger §37) failure to restore the original form. I take it that Pali, like all Prakrits, was beginning to simplify aspirates, but left most aspirates untouched when the oral teachings were codified. OI prabhā > *paha → P pabhā OI prabhā > P pabhā Retention of simplified (pp. 285–286)15 (Geiger §53.1) Vedic form after most OI conjuncts became single consonants in Pali. Paha is found once in the Tipiṭaka at D I 233 and could be an accidental lenition in dictation (Karpik 2019b: 110) or a confusion of ha and bha in the Sinhalese scribal tradition (Norman 2008: 189); it is a transmission error. 137  Review Article Revision (Levman) Retention (Karpik) Comments OI veṣṭa > P veḍha> *veha OI vyathā > P vedha Levman’s alleged koine Then at D II 100, S V 153 (Geiger §25.1, 38.4) form exists in only one and Th 143 these readings manuscript of which the appear: I defer to Gombrich (1987) editor, Oskar von Hinüber *veha → vekha who deduces from the (1991: 2), writes: “veha *veha → vega context and Sanskrit remains unexplained *veha → vegha sources that vedha and may be a simple *veha → veṭha (trembling) is correct for error”. The manuscript *veha → veḍha D II 100, S V 153 and tradition appears to have *veha→ vedha Th 143; Levman does not confused different roots *veha → veḷa discuss this work. and meanings; the koine *veha → vesa reconstruction is too wide- At Vin II 136 these readings Vin II 136 has vidha ranging to determine the appear: (buckle), which Norman correct readings. *veha → vidha (1994: 97–98) connects to * veha → vīṭha OI veṣṭa(ka) (covering/ * veha → vītha surrounding). * veha → veḍha (von Hinüber 1991) * veha → veha (von Hinüber 1991) (pp. 280–285) Overall, Levman’s revision/Sanskritisation hypothesis risks turning natural sound changes and transmission errors into speculative pre-Pali reconstructions for no advantage in terms of identifying correct readings. On the other hand, in every case, retention has the greater economy of explanation, satisfying the principle of Occam’s Razor. Retention further explains why Vedic, non-Sanskritic, forms are found in Pali and why advanced Pali forms are found in Epigraphic Prakrit/Epigraphic Pali. That Pali was contemporaneous with the Buddha is the better, parsimonious hypothesis. In conclusion Overall, Levman’s scholarship in this book is at times impressive. Possibly no other scholar can demonstrate a working knowledge of Pali, Sanskrit, Tibetan, German, French, Burmese and Old Tamil, as Levman does here. His 27 pages of references are also a useful and up-to-date resource. However, 138  Review Article with the greatest respect, I have found his koine theory lacking in convincing argument and scholarship, not least in his inaccurate or irrelevant citations of Buddhaghosa, Burrow, Emeneau, Gombrich, Krishnamurti, Norman, Sjoberg, and Southworth. Nonetheless, the author’s demonstration of linguistic and cultural borrowing regarding robe-practices from non Indo-Aryan sources into Pali and Buddhism will, I believe, stand the test of time. For applying this analysis to connected passages of Pali is pioneering work and Levman deserves praise for this. Likewise, his comparison of syntax in Pali and Old Tamil poetry is exceptional. His project (p. 378) of a “Prolegomenon for a Pali Etymological Dictionary of non Indo-Aryan Words” is an extension of this good work and to be welcomed. I very much hope he will follow through on his claim (p. 131) that one could do a whole study of the chronological strata of the Suttas based on their engagement with Brahmanism. References Allen, William Sydney (1953). Phonetics in Ancient India. London: Oxford University Press. Burrow, Thomas (1946). Loanwords in Sanskrit. Transactions of the Philological Society 45: 1–30. ⸺ (1955). The Sanskrit Language. Delhi: Motilal Banarsidass. Deokar, Mahesh (2009). Niggahīta—Theory and Practice. In B. Labh, ed., The Ocean of Buddhist Wisdom (Vol IV). Delhi: New Bharatiya Book Corporation, pp. 1–6 [1st Edition]. Emeneau, Murray B. (1980). Language and Linguistic Area. Stanford: Stanford University Press. Geiger, Wilhelm (1943). Pāli Literature and Language: Authorised English Translation by Batakrishna Ghosh. New Delhi: Oriental Books Reprint Corporation. ⸺ (1994). A Pāli Grammar, Translated into English by Batakrishna Ghosh, Revised and Edited by K.R. Norman. Oxford: The Pali Text Society. Gombrich, Richard (1987). Old Bodies Like Carts. Journal of the Pali Text Society 11: 1–4. ⸺ (2018). Buddhism and Pali. Oxford: Mud Pie Books. Hinüber, Oskar von (1991). The Oldest Pali Manuscript: Four Folios of the Vinaya- Piṭaka from the National Archives, Kathmandu. Stuttgart: Franz Steiner Verlag. 139  Review Article Karpik, Stefan (2019a). The Buddha Taught in Pali: A Working Hypothesis. Journal of the Oxford Centre for Buddhist Studies 16: 10–86. ⸺ (2019b). A Reply to Bryan Levman’s “The Language the Buddha Spoke”. Journal of the Oxford Centre for Buddhist Studies 17: 106–116. Krishnamurti, Bhadriraju (2003). The Dravidian Languages. Cambridge: Cambridge University Press. Levman, Bryan G. (2016). The Language of Early Buddhism. Journal of South Asian Languages and Linguistics 3(1): 1–41. ⸺ (2019). The Language the Buddha Spoke. Journal of the Oxford Centre for Buddhist Studies 17: 63–105. ⸺ (2020). Sanskritization and Pāli. Journal of South Asian Languages and Linguistics 7(1): 105–149. Norman, Kenneth R. (1980). The Dialects in Which the Buddha Preached. In H.  Bechert, ed., Die Sprache der ältesten buddhistischen Überlieferung [= The Language of the Earliest Buddhist Tradition]. Göttingen: Vandenhoek & Ruprecht, pp. 61–77. ⸺ (2008). An Epithet of Nibbāna. In Collected Papers Vol III. Oxford: The Pali Text Society, pp. 183–189 [First Published in Śramaṇa Vidyā: Studies in Buddhism: Prof. Jagannath Upadhyaya Commemoration Volume, Sarnath 1987, pp. 23–31]. ⸺ (1994). Pāli Lexicographical Studies XI. In Collected Papers Vol V. Oxford: The Pali Text Society, pp. 84–99 [First Published in Journal of the Pali Text Society 18(1993): 149–164]. ⸺ (2006). A Philological Approach to Buddhism. Lancaster: The Pali Text Society [2nd Edition; 1st Edition 1997]. Pind, Ole Holten (2005). On the Evidence of Unrecognised Absolutives in the Pāli Canon. Zeitschrift der Deutschen Morgenländisch 155(2): 497–518. ⸺ (2021). Did the Buddha Address the Monks in Māgadhī? Journal of the Oxford Centre for Buddhist Studies 20: 83–106. Pischel, Richard (1957). Comparative Grammar of the Prākrit Languages. Delhi: Motilal Banarsidass [First Published in German as Grammatik der Prakrit- Sprache. Strassburg: Verlag von Karl J. Trübner, 1900]. Sjoberg, Andrée F. (1992). The Impact of Dravidian on Indo-Aryan: An Overview. In E.C. Polomé et al., eds., Reconstructing Languages and Cultures. Berlin: Mouton de Gruyter, pp. 507–529. 140  Review Article Southworth, Franklin C. (2005). Linguistic Archaeology of South Asia. Abingdon: Routledge Curzon. Thitzana, Ashin (2016). Kaccāyana Pāli Grammar, Vol. 2. Onalaska: Pariyatti Press. Warder, Anthony K. (1995). Introduction to Pali. Oxford: The Pali Text Society. Witzel, Michael (1999). Substrate Languages in Old Indo-Aryan. Electronic Journal of Vedic Studies 5: 1–67. 141