Introduction

In this chapter I continue examining the topic of the early Buddhist oral tradition, in particular formal and functional aspects of Pāli discourses that reflect their nature as orally transmitted material. I begin by taking up formal elements, such as sound similarities, the principle of waxing syllables, the frequent use of repetition, and pericopes. Then I turn to functional aspects of the Pāli oral tradition, examining its purposes and the reciters responsible for its performance.

---

Formal Elements – Sound Similarities

A phrase discussed already in detail in a previous chapter is the standard opening to a discourse in the form "thus have I heard, at one time".[n.1] Cf. above p. 51ff. Not only does this opening quite explicitly draw attention to the oral nature of what follows, but even the very form of this formulaic beginning testifies to oral transmission. Among these first few words, found at the beginning of each discourse, sound and metrical similarities can be detected that are used throughout the discourses to facilitate memorization and recitation. Such sound similarities can involve alliteration, repetition of an initial sound, assonance, repetition of a sound found in the middle of a word, and homoioteleuton, repetition of the final sound.

As illustrated in table 9, the two parts of the standard opening of a Pāli discourse, evaṃ me sutaṃ and ekaṃ samayaṃ, each consist of five syllables. The first word in each part is closely similar, evaṃ and ekaṃ, differing only in respect to their second consonant. The words evaṃ, sutaṃ, ekaṃ, and samayaṃ share the -aṃ ending,[n.2] [4] The choice of the accusative ekaṃ samayaṃ instead of the locative ekasmiṃ samaye (cf. Wijesekera 1936/1996: 56) might even be related to the sound similarity this creates with the preceding evaṃ me sutaṃ. and the words sutaṃ and samayaṃ also share the same initial consonant s-.[n.3] [5] Allon 1997b: 195 and 242. In this way, even though these few words are merely the standard prose introduction to a discourse, a closer inspection reveals sound similarities that occur with considerable frequency in other prose sections of the early discourses, especially in listings of similar words or in formulaic expressions.

Table 9: Sound Similarities in the Opening of a Discourse

evaṃ me sutaṃ	ekaṃ samayaṃ
e(v)aṃ	e(k)aṃ
(ev)aṃ (sut)aṃ	(ek)aṃ (samay)aṃ
s(utaṃ)	s(amayaṃ)

---

Formal Elements – The Principle of Waxing Syllables

Another oral feature of the early discourses can be found in the frequent use of strings of synonyms. Such a string of synonyms serves to safeguard against textual loss, since a whole set of similar words stands a much greater chance of being remembered than a single word and also better impresses itself on the audience.[n.4] [6] Oldenberg 1917: 42 comments that the use of such strings of synonyms gives the impression of a certain childlike insistence that ensures that all aspects of a particular matter find expression.

A closer look at such strings or clusters of words brings to light that their members tend to occur in a metrical sequence that follows the principle of waxing syllables, already mentioned in the previous chapter. According to this principle, words with fewer syllables in a series of terms are followed by words with an equal or greater number of syllables. This principle can also be applied to listings and enumerations whose members do not share the same meaning. A few selected examples in table 10 below show how a particular theme can be expressed with the help of a string of terms that follow an ascending syllable count. The crescendo effect that results from the application of this principle is a typical stylistic feature of the early discourses, further enhanced when word sequences arranged in this way also share sound similarities.

Table 10: The Principle of Waxing Syllables

Theme:	Pāli terms:	Syllable count:
old	jiṇṇo vuddho mahallako addhagato vayo-anuppatto	2 +2 +4 +4 +6
growth	vuddhiṃ virūḷhiṃ vepullaṃ	2 +3 +3
fear	bhīto saṃviggo lomahaṭṭhajāto	2 +3 +6
to (mis)-meditate	jhāyanti pajjhāyanti nijjhāyanti apajjhāyanti	3 +4 +4 +5
able to attain	nikāmalābhī akicchalābhī akasiralābhī	5 +5 +6
poor	daliddo assako anāḷhiyo	3 +3 +4
wealthy	aḍḍho mahaddhano mahābhogo	2 +4 +4

The examples chosen in the table above cover only cases of up to five words.[n.5] [7] Examples are from MN 12 at MN I 82,26, MN 16 at MN I 101,7, MN 35 at MN I 231,37, MN 50 at MN I 334,23, MN 53 at MN I 354,36, MN 66 at MN I 450,34 and 451,36; taking a lead from von Hinüber 1994b: 16–30 and Smith 1948: 35. Von Hinüber 1994b: 33 draws attention to similar formulas found also in Jain scriptures, such as naṭṭā, gīa, vāiya, corresponding to nacca, gīta, vādita found, e.g., in MN 27 at MN I 180,6; cf. also Allon 1997b: 266. . If a sequence of words becomes relatively long, the principle of waxing syllables is not applied to the sequence as a whole, but to subunits within the sequence. Such subunits can share a similar nuance of meaning or can belong to the same category. The division into subunits would presumably have the function of setting a rhythm that allows the reciter to take a breath before continuing with the recitation of the text.

A case in point for the formation of such prosodic subunits is the description of various types of talks that are unbefitting and better avoided, presented in table 11 below.[n.6] [8] Taken from MN 76 at MN I 513,23; cf. the discussion of this example in Allon 1997a: 48 and Anālayo 2009k: 740f.

Table 11: Subunits in the Description of Irrelevant Talk

1st subunit, syllable count 4+4+6:

rājakathaṃ, corakathaṃ, mahāmattakathaṃ, talk related to government: "kings, robbers, ministers".

2nd subunit, syllable count 4+4+4:

senākathaṃ, bhayakathaṃ, yuddhakathaṃ, talk related to war: "armies, dangers, battles".

3rd subunit, syllable count 4+4+4+5:

annakathaṃ, pānakathaṃ, vatthakathaṃ, sayanakathaṃ, talk related to requisites: "food, drink, clothing, beds".

4th subunit, syllable count: 4+4+4+4:

mālākathaṃ, gandhakathaṃ, ñātikathaṃ, yānakathaṃ, talk on lay life: "garlands, perfumes, relatives, vehicles".

5th subunit, syllable count: 4+5+5+6:

gāmakathaṃ, nigamakathaṃ, nagarakathaṃ, janapadakathaṃ, talk on localities: "villages, towns, cities, counties".

6th subunit, syllable count: 4+4+5+6+6:

itthikathaṃ, sūrakathaṃ, visikhākathaṃ, kumbaṭṭhānakathaṃ, pubbapetakathaṃ, talk on gossip: "women, heroes, streets, wells, the departed".

The same principle of waxing syllables can also be responsible for the order of members in dvanda compounds. An example would be the pācittiya regulation according to which a monk in particular circumstances should not teach more than a certain number of words, specified to be "six or five", cha-pañca. The sequence of the numerals seems to follow the principle of waxing syllables against the natural ascending order of the numbers five and six.[n.7] [9] Vin IV 21,37. Another case is the expression Dhamma-vinaya, where the reason for Vinaya to stand in second position may well be its syllable count of three against the two syllables of Dhamma. The application of the principle of waxing syllables to dvanda compounds is in fact a rule recognized by Pāṇini.[n.8] [10] Caland 1931: 59–68 quotes Pāṇini 2.2.34 and provides a series of examples where, due to following the law of waxing syllables, the compound members in a dvanda come in a sequence that is in opposition to their natural order.

---

Formal Elements – Repetition

The oral nature of the early discourses also easily impresses itself on their audience due to the frequent occurrence of repetition. When treating a particular topic in its positive and negative manifestations, for example, it is a standard procedure in the discourses that the same passage will be repeated with precisely the same words and formulations used for the positive case, making only the most minimal changes required in order to adjust these to the negative case. The same procedure becomes even more prominent when a series of different perspectives on a particular topic are explored. A treatment of, for example, four different types of persons or modes of acting will use four times nearly the same text in order to achieve its aim.

In addition to the frequent occurrence of repetition within a single discourse, the early discourses also make recurrent use of "pericopes", formulaic expressions or phrases that depict a recurrent situation or event and whose purpose is to facilitate memorization. [n.9] [11] Cousins 1983: 1 refers to the "widespread use of mnemonic formulae" as a typical feature of early Buddhist oral literature. Griffiths 1983: 58 explains that the use of pericopes is "a direct result of the methods by which sacred material was preserved and handed down in the early Buddhist communities; the demands of mnemonic convenience … meant that the units of tradition … had to be … reduced to an easily memorized standard form." Von Simson 1965: 47 compares the function of such pericopes in Buddhist prose to the bones and tendons in the human body, in that both provide stability and support for the other parts. Smith 1987: 598, in an examination of modern oral literature in India, reports the finding that a Rajasthani epic that made frequent use of pericopes (which, according to his description, has the result that "every battle … is the same battle, every journey is the same journey, every meeting the same meeting") was transmitted with considerably greater accuracy than other comparable epics. Smith 1977: 151 explains that the reason for the employment of pericopes and the resulting greater accuracy "may lie in the fact that the epic is not merely sung for entertainment, but has a religious function", a reason that would hold true also for the use of pericopes in the oral transmission of the early Buddhist discourses. Whether they are descriptions of how someone approaches the Buddha or of how someone attains liberation, pericopes will be employed with a fixed set of phrases and expressions, with only the most minimal changes introduced to adapt these pericopes to the individual occasion. These two features, the repetition of passages within a discourse and the use of pericopes throughout a discourse collection, are responsible for the highly repetitive nature of the early discourses.[n.10] [12] In his detailed study of these features in a Dīgha-nikāya discourse, Allon 1997b: 359 comes to the conclusion that about 87% of the text of this discourse involves some form or other of repetition. He concludes (p. 360) that "repetition thus thoroughly permeates every dimension of this class of Buddhist literature."

These various oral characteristics of the early discourses testify to the importance given to precise recall in the early Buddhist oral tradition.[n.11] [13] Allon 1997b: 252 explains that "it is surely easier to remember a sequence of words arranged … according to syllable length", just as "it is easier to remember two different words when they share sound similarities and have the same metrical pattern"; cf. also Wynne 2004: 108–112. As already discussed in a previous chapter, in this respect the early Buddhist oral tradition differs from oral traditions where improvisation is a prominent feature.[n.12] Cf. above p. 68ff. The performance of oral literature of an epic or narrative type often demands innovation and improvisation from the performer, whose task is to present the main elements of a tale in such a way as to best entertain the audience. This type of oral literature is thus freely recreated every time it is told.[n.13] [14] Lord 1987: 71 notes that this involves "never merely memorizing a fixed entity, but … ever re-creating a new version of older forms and stories". In contrast, the purpose of the early Buddhist oral tradition was the preservation of sacred material of a type for which free improvisation is inappropriate. Moreover, recitation was often undertaken communally by the reciters, which leaves little scope for free improvisation.[n.14] Cf. above p. 75.

The emphasis on precise transmission in the early Buddhist oral tradition can even be detected in some errors, where at times in otherwise closely similar Pāli and Sanskrit passages the counterpart to a particular term shows close phonetic similarity but a considerably different meaning. In such cases, it seems as if the attempt of the reciters to remember precisely has preserved formal aspects, even though the meaning has been lost.[n.15] [17] Von Simson 1965: 137f gives the following examples: vivattacchaddo - vighuṣṭaśabdo; brahmujjuggatto - bṛhadṛjugātro; muducittaṃ - muditacittaṃ; aññataro - ājñātavān; sammodi sammodanīyaṃ - sammukhaṃ sammodanīṃ.

As is only to be expected of material that has been orally transmitted over longer periods of time, in spite of the various measures undertaken to ensure correct transmission, variations occur. Such differences not only occur between material transmitted by different Buddhist schools, but can also be found in texts of a single school, such as the Pāli texts transmitted by the Theravāda reciters. One type of difference that tends to occur involves variations in the use of pericopes. Such pericope variations usually affect those parts of a discourse that were added by the reciters in order to provide a background narration to the teachings given by the Buddha or his disciples.

---

Formal Elements – Pericope Variations

A difference in the use of pericopes can be seen, for example, between a discourse in the Dīgha-nikāya and its parallel in the Aṅguttara-nikāya, both of which report the same event, namely a visit paid by the minister Vassakāra to the Buddha in order to find out what the Buddha had to say about the plan by King Ajātasattu to attack the Vajjians.[n.16] [19] Allon 1997b: 39. Whereas the Dīgha-nikāya version describes in detail how Vassakāra got his chariot ready, drove with it, and then descended from it to proceed on foot, its Aṅguttara-nikāya counterpart does not mention Vassakāra's mode of arrival at all, but simply notes that he approached the Buddha. Here are translations of the respective part in the Dīgha-nikāya (a) and Aṅguttara-nikāya (b) versions:

a) Vassakāra … assented [to the order given to him] by Ajātasattu Vedehiputta, the king of Magadha, got the state carriages ready, mounted them, left Rājagaha by state carriage, and went towards Mount Vulture Peak. After going as far as the ground was passable for carriages, he descended from the carriage and approached the Blessed One on foot.[n.17] [21] DN 16 at DN II 73,4.

b) Vassakāra … assented [to the order given to him] by Ajātasattu Vedehiputta, the king of Magadha, and approached the Blessed One.[n.18] [23] AN 7.20 at AN IV 18,4. Allon 1997b: 39 notes that a description of how someone approaches by chariot can, however, be found elsewhere in the Aṅguttara-nikāya collection; cf., e.g., AN 5.50 at AN III 59,27 (King Muṇḍa approaches the monk Nārada), AN 8.12 at AN IV 181,23 (General Sīha approaches the Buddha), and AN 10.30 at AN V 65,9 (King Pasenadi approaches the Buddha), although the description given in these discourses is shorter than the "chariot approach" pericope employed in the Dīgha-nikāya.

Another case where the records of the same event differ in the detail in which they depict how someone approaches the Buddha can be observed by comparing the altogether four discourses that describe the famous last meeting between Māra and the Buddha, in which the Evil One asked the Buddha to pass away.[n.19] [24] Allon 1997b: 62. The Dīgha-nikāya and the Udāna versions report that Māra approached the Buddha, stood at one side, and then addressed the Buddha. The Saṃyutta-nikāya version of the same event does not mention that he stood at one side, but only records that he approached the Buddha. The Aṅguttara-nikāya version does not record any approach at all. Below are the Dīgha-nikāya/Udāna (a), Saṃyutta-nikāya (b), and Aṅguttara-nikāya (c) versions:

a) Not long after the venerable Ānanda had left, Māra the Evil One approached the Blessed One. Having approached, he stood on one side. Standing on one side, Māra the Evil One said this to the Blessed One.[n.20] [25] DN 16 at DN II 104,12 and Ud 6.1 at Ud 63,13.

b) Not long after the venerable Ānanda had left,[n.21] [26] Ce does not have acīrapakkante āyasmante ānande. Māra the Evil One approached the Blessed One.[n.22] [27] Se adds (ekamantaṃ aṭṭhāsi. ekamantaṃ ṭhito kho māro pāpimā). Having approached,[n.23] [28] Be and Se add bhagavantaṃ. he said this.[n.24] [29] SN 51.10 at SN V 260,25.

c) Not long after the venerable Ānanda had left, Māra the Evil One said this to the Blessed One.[n.25] [30] AN 8.70 at AN IV 310,11.

Pericopes also differ when it comes to describing the respectful attitude with which someone listens to a sermon given after a meal by the Buddha or by a monk. For such occasions, the Dīgha-nikāya, the Majjhima-nikāya, the Udāna, and the Sutta-nipāta employ a pericope that describes how the listener(s) take(s) a low seat, an obvious expression of respect.[n.26] [31] The pericope of "taking a low seat", aññataraṃ nīcaṃ āsanaṃ gahetvā, leads from the pericope that describes the giving of a meal to a sermon, e.g., in DN 3 at DN I 109,36 (for further references and a discussion of this variation cf. Allon 1997b: 122). The same pericope can also be found regularly in the Madhyama-āgama, e.g., MĀ 132 at T I 625b17: 取一小床; in this case it is also found in the Tibetan counterpart at D 1 kha 105b3 or Q 1030 ge 97b1: stan ches (Q: chem) dma’ ba zhig blangs te, whereas in the parallel MN 82 at MN II 64,23 the whole episode is not found. Some examples for Sanskrit occurrences can be seen in the Mūlasarvāstivada Vinaya, Dutt 1984a: 265,15: nīcataram āsanaṃ gṛhītvā, counterpart to Sn 3.7 at Sn p. 111,9: aññataraṃ nīcaṃ āsanaṃ gahetvā; Mahāparinirvāṇa-sūtra fragment S 360 folio 187 V5, Waldschmidt 1950: 26: (nīcata)[r](a)[k](a)m-āsanaṃ gṛhītvā, parallel to DN 16 at DN II 126,26, where the low seat is not mentioned; or the Saṅghabhedavastu, Gnoli 1977: 145,14: nīcataram āsanaṃ gṛhītvā, parallel to Vin I 18,9, where the low seat is not mentioned. Similar situations in the Vinaya and in the Aṅguttara-nikāya, however, do not mention a low seat at all.[n.27] [32] Instead of the pericope of "taking a low seat" after the pericope that describes the giving of a meal, only the pericope "sat down at one side", ekamantaṃ nisīdi, leads over to a sermon, e.g., in AN 4.57 at AN II 63,4 (for further references cf. Allon 1997b: 123). This difference is particularly notable in the case of a meal given by Prince Bodhi, as the same meal is recorded in the Majjhima-nikāya and the Vinaya, so that in this case the same event is described once with and once without the taking of a low seat. Here are the Majjhima-nikāya (a) and Vinaya (b) accounts:

a) When the Blessed One had eaten and had removed [his] hands from the bowl, Prince Bodhi took a low seat and sat down on one side.[n.28] [33] MN 85 at MN II 93,9.

b) When the Blessed One had eaten and had removed [his] hands from the bowl, Prince Bodhi sat down on one side.[n.29] [34] Vin II 128,36.

The application of a pericope can at times result in inconsistencies within a discourse. An example is the pericope that describes how the Buddha or a monk gets ready to beg for alms. Since food has to be taken before noon, such preparations are usually made in the early morning, so that this pericope describes how "in the morning" the Buddha or a monk dresses and takes his bowl and robe in order to approach the next village or town.[n.30] [35] Cf., e.g., MN 5 at MN I 31,29: pubbanhasamayaṃ nivāsetvā pattacīvaraṃ ādāya. The frequent occurrence of this pericope has caused it to be also applied to a passage in the Vinaya and the Udāna where it does not fit its context. This passage records how the Buddha was travelling and arrived at a particular place where he was invited to come to the local hall. The villagers then approached the same hall and listened to a discourse by the Buddha that went on well into the night. Even though the circumstances make it clear that the invitation to come to the local hall must have taken place in the late afternoon or evening, the Vinaya and the Udāna nevertheless report that it was "in the morning" that the Buddha followed the invitation by dressing and taking his bowl and robe in order to approach the local hall.[n.31] [36] Vin I 227,10 = Ud 8.6 at Ud 86,13: bhagavā pubbaṇhasamayaṃ nivāsetvā pattacīvaraṃ ādāya saddhiṃ bhikkhusaṅghena yena āvasathāgāraṃ ten' upasankami, followed by a description that the laity heard a discourse from the Buddha and were then dismissed, bhagavā … upāsake bahud eva rattiṃ dhammiyā kathāya sandassetvā samādapetvā samuttejetvā sampahaṃsetvā uyyojesi, where the reference to bahud eva rattiṃ makes it clear that the discourse was given at night time, so that the earlier reference to "the morning", pubbanhasamayaṃ, does not fit the context. Allon 1997b: 141 notes this error and also draws attention to instances where the pericope is properly adjusted to its context, such as in MN 53 at MN I 354,12 or in SN 35.202 at SN IV 183,16, which introduce a similar situation only with nivāsetvā pattacīvaraṃ ādāya, without the specification pubbanhasamayaṃ.

The relatively circumstantial differences noted so far may seem negligible, since they do not affect essential matters. Not all such errors, however, are of such circumstantial character. A more significant variation in the use of the pericopes employed at the conclusion of a discourse can be found between the Saṃyutta-nikāya and the Sutta-nipāta versions of the Kasibhāradvāja-sutta. These two discourses record the same event but differ in their conclusion, as according to the Saṃyutta-nikāya account Kasibhāradvāja took refuge and declared himself to be a lay follower, whereas according to the Sutta-nipāta version he took refuge, requested ordination and became an arahant. The two versions, in the Saṃyutta-nikāya (a) and the Sutta-nipāta (b), proceed as follows:

a) I go for refuge to the venerable Gotama, to the Dharma, and to the community of monks, may the venerable Gotama remember me as a lay follower who from today on has gone for refuge for life.[n.32] [38] SN 7.11 at SN I 173,23.

b) I go for refuge to the venerable Gotama, to the Dharma, and to the community of monks,[n.33] [39] Se adds upāsakaṃ maṃ bhavaṃ gotamo dhāretu ajjatagge pāṇupetaṃ saraṇaṃ gataṃ before labheyyāhaṃ etc. may I receive the going forth in the presence of the venerable Gotama and the full admission … and the venerable Bhāradvāja became one of the arahants.[n.34] [41] Sn 1.4 at Sn p. 15,23. The Chinese parallels SĀ 98 at T II 27b26, SĀ2 264 at T II 466c10, and SĀ3 1 at T II 493b8 agree with Sn 1.4 in reporting that he went forth and became an arahant.

An additional perspective on variations in the use of pericopes emerges when comparing their employment in Pāli discourses with the corresponding usage in the Chinese Āgamas. Taking as an example the Madhyama-āgama collection preserved in Chinese, discourses found in this collection regularly describe that a monk fans the Buddha,[n.35] [42] An example is MĀ 33 at T I 474a19: 執拂侍佛. a circumstance noted only rarely in discourses found in its Pāli counterpart, the Majjhima-nikāya.[n.36] [43] MN 12 at MN I 83,20 and MN 74 at MN I 501,1 report that a monk was fanning the Buddha. On frequent occasions Madhyama-āgama discourses also mention the sitting mat,[n.37] [44] An example is MĀ 9 at T I 430b10: 尼師檀 (with the variant 尼師壇); cf. also Minh Chau 1964/1991: 29. one of the standard requisites of a monk, whereas their Pāli counterparts tend to refer to the same accessory only on rare occasions.[n.38] [45] MN 24 at MN I 147,5 and MN 147 at MN III 277,30.

Another standard pericope in Madhyama-āgama discourses describes how a visitor or a monk will depart from the presence of the Buddha by performing three circumambulations, a circumstance not mentioned in Majjhima-nikāya discourses.[n.39] [46] For example MĀ 132 at T I 623b23: 繞三匝 has as its counterpart in the parallel MN 82 at MN II 56,22: padakkhiṇaṃ katvā. A reference to three circumambulations can be found in DN 16 at D II 163,27, where Mahākassapa pays homage to the Buddha's funeral pyre with three circumambulations. Part of the pericope on three circumambulations has been preserved in a Sanskrit fragment parallel to an occurrence in MĀ 161 at T I 686a18; cf. SHT V 1148 R4, Sander and Waldschmidt 1985: 147: triprada(kṣiṇīkṛtvā). The two collections also differ in their descriptions of how listeners will express their appreciation of the teachings. In a Majjhima-nikāya discourse such a person will exclaim "wonderful, wonderful"; in a Madhyama-āgama discourse such a person will rather inform the Buddha: "I understood, I realized."[n.40] [47] An example is MN 7 at MN I 39,27: abhikkantaṃ … abhikkantaṃ, which has a counterpart in MĀ 93 at T I 576a10: 我已知 … 我已解.

In discourses in the Madhyama-āgama collection, when someone asks the Buddha or a monastic a question, the actual enquiry will be preceded by a request to be given permission to put a question,[n.41] [48] A case in point is MĀ 29 at T I 461b27: 我欲有所問, 聽我問耶? a pericope found only rarely in the Majjhima-nikāya.[n.42] [49] MN 35 at MN I 229,35, MN 109 at MN III 15,23, and MN 144 at MN III 264,30. Notably, although none of these three Pāli discourses has a parallel in the Madhyama-āgama, each has a parallel in the Saṃyukta-āgama, where in each case this pericope is not found; cf. SĀ 110 at T II 35c11 (parallel to MN 35), SĀ 58 at T II 14b17 (parallel to MN 109), and SĀ 1266 at T II 347c23 (parallel to MN 144). A Tibetan version of this pericope can be found in the parallel to MN 90, D 1 kha 88a1 or Q 1030 ge 81a6, in which case this particular pericope is also found in the Madhyama-āgama parallel to the same discourse, MĀ 212 at T I 793b15, but not in the Pāli version.

Other pericopes found in the Majjhima-nikāya are absent from the Madhyama-āgama. One example is the pericope employed regularly at the beginning of a Majjhima-nikāya discourse, in which the Buddha addresses his disciples with "monks", and the monks reply "venerable sir", after which the Buddha announces his topic and proceeds to deliver the discourse.[n.43] [50] Cf., e.g., MN 1 at MN I 1,3: bhikkhavo ti. bhadante ti te bhikkhū bhagavato paccassosuṃ; on the use of this pericope in the Dīgha-nikāya and the Majjhima-nikāya cf. also Manné 1990: 33; on its relative lateness Meisig 1987b: 225.

A closer inspection shows that this pericope does not fit too well with the remainder of the Pāli discourses in which it occurs, inasmuch as the vocative "monks", bhikkhavo, used in this passage, differs from the vocative address "monks", bhikkhave, used in all remaining instances in the discourse.[n.44] [51] MN 1 at MN I 1,5 continues with bhikkhave. On the vocative bhikkhave cf. also Lüders 1954: 13, Bechert 1991: 11, and Collett and Anālayo 2014. Similarly, the first vocative "venerable sir", bhadante, used by the monks, is not the same as the vocative "venerable sir", bhante, found in the remainder of the discourse.[n.45] [52] MN 1 at MN I 1,7 continues with bhante. Since there would be no reason for starting with one particular vocative and then switching to another type of vocative for the remainder of the discourse, this difference suggests that this pericope would have been added during oral transmission.

Although this pericope is not found in the Madhyama-āgama, it does occur in an individual translation. Notably, this discourse stems from a Madhyama-āgama collection.[n.46] [53] T 48 at T I 837c25: "the Buddha said: 'monks!' The monks replied: 'yes, indeed!' The monks listened to the Buddha. The Buddha said … "; 佛告諸比丘, 比丘應曰唯然, 比丘從佛聽, 佛說. According to an introductory remark in T 48 at T I 837c21, this discourse stems from a Madhyama-āgama collection, 出 中阿含.

To help make the above collected information on the formal aspects of Pāli orality come more alive, in what follows I try to gather what the Pāli sources indicate to have been the purposes of oral recitation. I also take up the topic of individuals who, according to these same sources, were involved in carrying out the transmission of the Pāli discourses to later generations.

The oral transmission of the early texts is seen as being as old as Buddhism itself. According to the different Vinayas, soon after his awakening the Buddha sent his first arahant monk disciples out to teach others.[n.47] [54] This account can be found in the Dharmaguptaka Vinaya, T 1428 at T XXII 793a7, in the Mahāvastu, Senart 1897: 415,8, in the Mahīśāsaka Vinaya, T 1421 at T XXII 108a7, in the Mūlasarvāstivāda Vinaya, T 1450 at T XXIV 130a20, in the Sarvāstivāda Vinaya, T 1440 at T XXIII 511a12, and in the Theravāda Vinaya, Vin I 21,1; cf. also SN 4.5 at SN I 105,24 and its parallel SĀ 1096 at T II 288b3. For these disciples to engage in teaching activities, they would presumably have taken some teachings along, teachings they would then have passed on to their disciples.[n.48] [55] Gombrich 1990a: 25.

Such teachings would obviously have been in an oral form. Some of the first monk disciples, like Yasa and his friends, were not brahmins trained in the art of oral transmission, so the material to be taken along must have been relatively easy to memorize. Although the discourses at that time would have been few, already at that time some degree of formalization of these discourses, to facilitate their oral transmission, would most likely have been taking place.

In fact some oral features would already have been in use when the discourses were first delivered, as even today repetition is employed as a tool to drive home a point when giving a speech, whereas such repetition is avoided in a written presentation.[n.49] [57] Williams 1970: 166 argues that "it is possible that the Buddha's teaching methods included repetition and stylized formulae to aid memorization." In this way some formalized features could already have been integral to the discourses when they were first delivered, to ensure that the listeners kept the main points well in mind.

A discourse in the Aṅguttara-nikāya highlights that, even though the listeners may be paying attention when they are being taught, once they leave some of them might quickly forget what they have heard.[n.50] [58] AN 3.30 at AN I 130,29, which compares the case of someone who listens but then forgets it all again after leaving to someone who has different types of seeds on his lap and then gets up quickly, as a result of which the seeds will all be scattered around. In view of the need to avoid that this happens on a wider scale, a to some degree formalized body of oral material could well have already come into existence during the Buddha's lifetime.[n.51] [56] Davidson 1990/1992: 293 argues that the "processes of … consolidation must have begun during the life of the Buddha".

Besides facilitating preaching and teaching, such a formalized body of oral material would also have had the function of creating a sense of unity and communal concord through group recitation. In fact the Saṅgīti-sutta and its parallels explicitly take their occasion from the need to ensure harmony in the face of the strife that according to these discourses had occurred among the Jains after the death of their leader.[n.52] [59] DN 33 at DN III 210,18, DĀ 9 at T I 49c6, and Sanskrit fragment K 484 folio 11 Rc, Stache-Rosen 1968: 17. The fortnightly recitation of the code of rules (pātimokkha) would have functioned as another important manifestation of communal harmony, and a discourse reports that even monks who otherwise lived in complete seclusion would come to join the nearest monastic community on such occasions.[n.53] [60] MN 77 at MN II 8,30 reports that even disciples who excelled in living in seclusion would come for the fortnightly recitation of the code of rules, te anvaddhamāsaṃ saṅghamajjhe osaranti pātimokkhuddesāya; in fact according to Vin I 105,26 a monk should come for the fortnightly recitation even if he is living apart. The relevance of group recitation to communal harmony can also be seen in the accounts of the later saṅgītis, where success in establishing communal harmony finds expression in the performance of a communal recital.[n.54] [61] Tilakaratne 2000b: 175f explains that "the fundamental purpose of the act of saṅgāyana and therefore of the events described as saṅgīti is the assurance of the unity of the Buddhist monastic organisation … in the act of saṅgāyana … the key activity was to recite together … memorization or preservation of the Canon … was not its main purpose … the act of saṅgāyana, first and foremost, was meant to be a public expression of one's allegiance to the organisation which was represented by the Dhamma and the Vinaya", adding that "the recital of the Pātimokkha by the members of the Saṅgha every fortnight serves virtually the same purpose." Witanachchi 2006: 721 notes that "what is relevant in a saṅgīti is not so much the reciting of the text together, but the absence of any discordance"; cf. also Bareau 1955: 134 and Anālayo 2014c: 15f.

In addition to functioning as a tool for the preservation of the teachings and for expressing communal harmony, some discourses indicate that oral recitation in early Buddhism had still other functions. A discourse in the Saṃyutta-nikāya and its parallels in two Saṃyukta-āgama translations report that on one night, just before dawn, Anuruddha was reciting texts by himself.[n.55] [62] SN 10.6 at SN I 209,19, SĀ 1321 at T II 362c10, and SĀ2 320 at T II 480c21. A woman overheard him and told her child to be quiet, in order to avoid disturbing Anuruddha's recitation.

According to another discourse in the Saṃyutta-nikāya, on one occasion the Buddha similarly recited a discourse to himself while being alone and in seclusion.[n.56] [63] SN 12.45 at SN II 74,15: bhagavā rahogato paṭisallīno imam dhammapariyāyam abhāsi; no parallel seems to be known. A monk chanced by and overheard the recitation. The circumstances make it clear that the recitation was undertaken merely for its own sake, without any teaching purpose in mind.

Whereas one might suppose that Anuruddha is shown to be privately rehearsing, the Buddha would not be presented as needing to rehearse his own discourses. In fact, what he is on record as having recited on this occasion is a treatment of the six senses from the perspective of dependent arising (paṭicca samuppāda), a treatment that would have been familiar enough to him not to require any private rehearsing. This suggests that his recitation is best interpreted as a recollection of what he had realized, perhaps similar to the inspired utterances that according to the Udāna he made soon after his awakening, a time when he was also alone and in seclusion.[n.57] [64] Ud 1.1–3 at Ud 1,1 or Vin I 1,1; cf. also, e.g., T 212 at T IV 775c19.

The above passage suggests that the early Buddhist oral tradition also served as a way of meditating or reflecting on the Dharma. This impression is confirmed by a list of five possible occasions for reaching liberation, given in several discourses, according to which recitation can even issue in awakening.[n.58] [65] DN 33 at DN III 241,26, DN 34 at DN III 279,12 (abbr.), AN 5.26 at AN III 22,14, DĀ 9 at T I 51c10, DĀ 10 at T I 53c22, and SĀ 565 at T II 149a6; cf. also Collins 1992: 126f. Coward 1986: 300f explains that "the mere memorization of the text is not judged to be the most important aspect of the oral tradition … by chanting or listening to the rhythmic words of a sacred text, the teaching and inspiration in the words becomes renewed and reinforced. In this sense the oral recitation of a text is a sacramental act." These discourses explain that during recitation a deeper understanding can arise that eventually culminates in attaining liberation. Thus, recitation undertaken for its own sake can function as a means of mental cultivation (bhāvanā) in a wider sense, and as such can become a tool for progress on the path to liberation.

Another occasion for reaching liberation, according to the same discourses, occurs when listening to someone else expounding the teachings. Other discourses indicate that the oral delivery of a discourse can also help the listener to overcome a physical disease. One such instance is the Girimānanda-sutta. According to the Pāli and Tibetan versions of this discourse, the Buddha had told Ānanda to recite a teaching on ten types of perception to a sick monk. The monk recovered from his illness while listening to this.[n.59] [66] AN 10.60 at AN V 112,16 and its Tibetan parallel, D 38 ka 279a1 or Q 754 tsi 295b6; cf. above p. 219 note 47.

Such instances support the impression that recitation, whether performed by oneself or by another, served as a tool for meditation in early Buddhism.[n.60] [70] Kwella 1978: 173 reasons that "the texts repeat very often the same words … the citta … comes to the same subtle pictures ever and ever again … a comparatively high concentration of the mind … will be the … result." A discourse in the Aṅguttara-nikāya and its Chinese parallels make this point explicitly, as they advise to practise recitation to overcome sloth-and-torpor.[n.61] [71] AN 7.58 at AN IV 86,9: yathāsutaṃ yathāpariyattaṃ dhammaṃ vitthārena sajjhāyaṃ kareyyāsi, a recommendation which has counterparts in MĀ 83 at T I 559c13: 當隨本所聞法, 隨而受持廣布誦習 and in T 47 at T I 837a21: 聞 法如所誦法, 廣當誦習. According to the Pāli commentaries, recitation can not only act as an antidote to sloth-and-torpor, but can also help to overcome any type of unwholesome thought.[n.62] [72] Ps II 91,4 recommends reciting with a loud voice to overcome unwholesome thoughts, mahāsaddena sajjhāyitabbo.

In this way, the early Buddhist oral tradition was not perceived only as a means to preserve texts, but also as an integral part of the practice of the path to liberation. In fact oral recitation continued for a considerable time even after the writing down of the texts and is to some degree still practised today, which shows that it serves a greater purpose than just ensuring the preservation of the texts.[n.63] [73] Bechert 1992: 53: explains that the oral tradition continued to exist side by side with written scriptures for many centuries. Malalasekera 1928/1994: 46 notes that the "practice of learning up portions of the Scriptures continued … for a very long time" even after they had been written down. This finds confirmation in the travel records by Fǎxiǎn (法顯), who towards the end of the fourth and the beginning of the fifth century searched in vain for Vinaya manuscripts all over India, as the material was still transmitted purely through oral means, T 2085 at T LI 864b18: 皆師師口傳, although, as pointed out by Demiéville 1951: 247 note 1, the lack of Vinaya manuscripts could also have been at least in part due to the intentionally restricted circulation of such material.

---

Functional Aspects – The Reciters

With this range of purposes, it seems probable that oral recitation of at least some key texts would have been part of the general training of monks and nuns.[n.64] [74] Cousins 1983: 5 explains that "every monk would need a stock of small pieces for chanting when visiting the sick or for recitation after receiving food." The commentaries explain that a monk who wishes to live a life of seclusion in the forest should memorize at least the code of rules (pātimokkha) and two or three recitation sections from a discourse collection.[n.65] [75] Pj II 194,32: pātimokkhaṃ dve tīṇi bhāṇavārasuttantāni ca paguṇaṃ katvā, cf. also Collins 1992: 123.

Such ability is depicted in a passage that records the first meeting between the Buddha and the monk Soṇa Koṭikanna. During this meeting, the Buddha asked Soṇa to recite some teachings,[n.66] [76] Ud 5.6 at Ud 59,20: paṭibhātu taṃ, bhikkhu, dhammaṃ bhāsituṃ (following Se; Be and Ce read dhammo and Ee paṭibhātu bhikkhūnaṃ); cf. also Vin I 196,34. and Soṇa complied by reciting the sixteen discourses of the Aṭṭhaka-vagga, a collection now found in the Sutta-nipāta.[n.67] [77] Ud 5.6 at Ud 59,22: soḷasa aṭṭhakavaggikāni sabbān' eva sarena abhaṇi. On the texts that according to different Vinayas Soṇa recited cf. Lévi 1915. The narrative setting conveys the impression that the Buddha, who had already been pleased by Soṇa's meditative conduct, wanted to see if Soṇa was also able to recite some section of the early Buddhist oral tradition, which would have been considered a complementary qualification of a well-trained monk. This particular incident is also noteworthy in so far as it portrays the Buddha as himself checking to see if oral recitation was correctly undertaken.

Although every monastic disciple might have had to memorize some material, as the oral material grew in size, its preservation must have become an increasingly specialized and demanding task. The Pāli Vinaya reports that the reciter monks would sometimes pass the whole night busily reciting the discourses.[n.68] [78] Vin I 169,6: suttantikehi suttantaṃ saṅgāyantehi, mentioned in addition to dhammakathikehi dhammaṃ sākacchantehi, discussion on the Dhamma by those who teach the Dhamma, and vinayadharehi vinayaṃ vinicchinantehi, investigation of Vinaya matters by the Vinaya specialists; this makes it clear that the first of these three refers to mere recitation of the texts. At times, the concern with oral recitation appears to have become excessive and the discourses consequently voice criticism of those who neglect seclusion for the sake of recitation.[n.69] [79] AN 5.73 at AN III 86,25; cf. also AN 6.46 at AN III 355,6 on the conflict between the monks who emphasized theoretical learning and those who emphasized meditation and, for a study of this discourse, Cousins 2009.

The recitation of the early Buddhist texts is attributed not only to monks, but also to nuns. The Theravāda Vinaya records that the nuns Thullanandā and Bhaddā Kāpilānī were well-learned preachers and reciters.[n.70] [80] Vin IV 254,4, Vin IV 255,4, Vin IV 256,23, Vin IV 285,18, Vin IV 290,4, Vin IV 292,14, and Vin IV 302,21 present the nun Thullanandā as bahussutā bhāṇikā; epithets accorded at Vin IV 290,6 and Vin IV 292,14 also to Bhaddā Kāpilānī. Skilling 2000: 61 note 43, however, suggests taking bhāṇikā not in the sense of "reciter", as the term with this meaning is found only in later texts, but in the sense of "eloquent". In a similar vein, the Divyāvadāna refers to nuns who were knowledgeable in the Tripiṭaka,[n.71] [81] Cowell and Neil 1886: 493,8: bhikṣuṇyas tripiṭā dhārmakathikā; cf. also Skilling 2000: 62. and the Dīpavaṃsa records that nuns in Ceylon were capable at reciting the Vinaya, the five Nikāyas, and the seven works of the Abhidharma.[n.72] [82] Dīp verse 18.13, Oldenberg 1879: 97,6; cf. also Skilling 2000: 64.

In general, the oral transmission of the texts was probably the domain of the monastic disciples.[n.73] [83] Gombrich 1990: 25 explains that, although some lay people knew texts by heart, "only monks and nuns … were so organized that they could hand them on to future generations." In fact, a regulation found in the different Vinayas prohibits a monk or a nun from teaching recitation "word by word" to someone who has not received full ordination.[n.74] [84] These are the pācittiya/pātayantika rule no. 6 in the Dharmaguptaka Vinaya, T 1428 at T XXII 639a5, which prohibits "reciting together", 共誦者; rule no. 6 in the Mahāsāṅghika Vinaya, T 1425 at T XXII 336c20, which prohibits "teaching … to speak the Dharma by sentence", 教 … 說句法, which in the Sanskrit version, Tatia 1975: 19,16, reads: padaśo dharmaṃ vāceya; rule no. 6 in the Mahīśāsaka Vinaya, T 1421 at T XXII 39c22, which prohibits "teaching … to recite the discourse(s)", 教 … 誦經; rule no. 6 in the Mūlasarvāstivāda Vinaya, T 1442 at T XXIII 771c22, which prohibits "teaching the Dharma in sentence and phrases by joint recitation", 同句讀誦教授法者, which in the Sanskrit version, Banerjee 1977: 32,11, reads: padaśo dharmaṃ vācayet (the Tibetan version in Vidyabhusana 1915: 77,3 reads tshig gis chos ’don na lhung byed do); rule no. 6 in the Sarvāstivāda Vinaya, T 1435 at T XXIII 71a7, which prohibits "teaching the Dharma by way of sentence", 以句法教, which in the Sanskrit version, von Simson 2000: 205,3, reads: padaśo dharmaṃ vācayet; and rule no. 4 in the Theravāda Vinaya, Vin IV 14,30, which enjoins that one should not "make recite the Dhamma sentence by sentence", padaso dhammaṃ vāceyya; cf. also Norman and Pruitt 2001: 46,12. The background narration in the Dharmaguptaka Vinaya reports that the promulgation of this rule was occasioned by monks teaching recitation in the manner of brahmins, T 1428 at T XXII 638c22: 如婆羅門誦書聲無異, whereas according to the Mahīśāsaka Vinaya the problem was that the monks who taught recitation came from different parts of the country and thus some did not recite the phrases correctly, T 1421 at T XXII 39c15: 誦讀經偈音句不正; cf. also Lévi 1915: 436f. Thus in these two Vinayas the basic issue at stake seems to be the same, even though they do not use the specification "word by word" in the actual rule. For a comparative study of the different Vinaya accounts of this rule cf. Lévi 1915: 422f and 436–441. This would prevent training the laity in recitation to such an extent that they could fully participate in the preservation of the texts.

Another Vinaya ruling suggests that, nevertheless, the laity also memorized discourses. According to the Theravāda version of this rule, monks are allowed to forgo the travelling restrictions during the rains-retreat period for a variety of compelling reasons, one among them being that a lay disciple asks them to come in order to learn a discourse from him, lest it be lost.[n.75] [85] Vin I 140,36: āgacchantu bhaddantā imaṃ suttantaṃ pariyāpuṇissanti pur' āyaṃ suttanto palujjati. The Sarvāstivāda Vinaya has preserved the narrative leading to the promulgation of this rule differently. According to its report, the reason was not that the monks should come to learn the discourse from the lay disciple, but rather that the lay disciple had forgotten a discourse and wanted the monks to come to teach it to him again.[n.76] [86] T 1435 at T XXIII 174b28: 若先學忘欲誦, 大德來教我受學讀誦問義.

Independent of which version of the story may appear more probable, it seems clear that, at least to some degree, householders were also held to have been involved in memorizing the discourses. An example for this would be the householder Citta, who appears to have been well acquainted with the teachings. As part of a discussion with monks, he is on record for referring to the Brahmajāla-sutta,[n.77] [87] SN 41.3 at SN IV 286,12. The parallel SĀ 570 at T II 151a12 differs, as according to its presentation he referred to the exposition of views without using the name of the discourse. suggesting his familiarity with this rather long discourse, and his knowledge of the teachings was reportedly such that at times monks found it difficult to reply to his deep questions. In fact in some discourses times he takes his turn in teaching the Dharma to monks.[n.78] [88] In SN 41.1 at SN IV 282,28 and its parallel SĀ 572 at T II 152a12, Citta gives an exposition on the topic of the fetters to a group of monks who had been unable to resolve a discussion on this issue. SN 41.5 at SN IV 292,1 and its parallel SĀ 566 at T II 149b14 report how Citta explained the meaning of a verse (found in Ud 7.5 at Ud 76,26) to a monk. The list of eminent disciples in the Aṅguttara-nikāya reckons him as chief among lay expounders of the Dharma.[n.79] [89] AN 1.14.6 at AN I 26,5: etad aggaṃ mama sāvakānaṃ upāsakānaṃ dhammakathikānaṃ yadidaṃ citto gahapati; whereas its counterpart EĀ 6.1 at T II 559c10 extols his superior wisdom, 智慧. Pāsādika 1972: 23 notes that Citta can be considered as a precursor to Vimalakīrti.

Another prominent case is the laywoman Khujjuttarā, who according to the traditional account had memorized the whole of the Itivuttaka collection and thereby played a crucial role in preserving this collection for posterity. The Pāli commentary to this work explains that she passed the discourses she had memorized on to the nuns, who in turn passed them on to the monks.[n.80] [90] It-a 32,15. According to AN 1.14.7 at AN I 26,19, Khujjuttarā was outstanding for "having heard much", bahussutā; the listing of eminent disciples in EĀ 7.1 at T II 560b1 instead reckons her outstanding for her wisdom, 智慧; cf. also AN 2.12.4 at AN I 89,2, which presents Khujjuttarā as an exemplary lay disciple, worthy of being emulated by others.

A closer inspection of the Itivuttaka shows that its discourses are set apart from other discourses by their use of peculiar pericopes. Instead of beginning with "thus have I heard", evaṃ me sutaṃ, discourses in the Itivuttaka begin with "this was said by the Blessed One, said by the arahant, so I have heard", vuttaṃ h' etaṃ bhagavatā vuttam arahatā ti me sutaṃ, a peculiarity also preserved in its Chinese counterpart.[n.81] [91] It 1.1 at It 1,4, with its counterpart in T 765 at T XVII 662b15: "from the Blessed One, I heard these words", 吾從世尊, 聞如是語.

Moreover, the conclusions of Itivuttaka discourses are also unique, as are its transitions from prose to verse.[n.82] [92] The conclusion of a discourse, e.g., in It 1.1 at It 1,16, states that: "this meaning was also said by the Blessed One, so I have heard it", ayam pi attho vutto bhagavatā, iti me sutaṃ (T 765 has not preserved a conclusion to its discourses). The pericope employed to lead over from the prose section to verse(s), e.g., in It 1.1 at It 1,8, reads: "this is the meaning of what the Blessed One said. In regard to this, it was said like this", etam atthaṃ bhagavā avoca, tatth' etaṃ iti vuccati. The counterpart to this transition pericope in T 765 at T XVII 662b20 reads: "at that time the Blessed One, taking up this matter again, spoke in verse", 爾時世尊, 重攝此義, 而說頌曰.

This is noteworthy in so far as these peculiarities seem to have resisted the natural tendency of oral transmission to stereotype the introduction and conclusion of a discourse. The present example complements the above listed examples of pericope variations, where changes appear to have occurred at some point during oral transmission. In the case of the Itivuttaka, however, it seems as if its formal aspects are related to the nature of its first recipient(s) and would presumably have come into being not too long after their original delivery. These formal aspects were then apparently passed on unchanged for many generations of reciters to come, without being adapted to fit the form of other discourses. This reinforces the impression that some degree of formalization of the material for oral transmission would have taken place at a comparatively early stage. If the formalization of the discourses had been undertaken at a late point in time, one would expect the procedure used in the case of other discourses to have been applied similarly to the discourses in the Itivuttaka.

---

Conclusion

The formal aspects of the Pāli discourses – the use of pericopes, the occurrence of metrical and sound similarities, the application of the principle of waxing syllables, and the recurrent use of repetition – testify to the oral nature of these discourses as the final product of a prolonged period of oral transmission. The present form of the Pāli discourses is clearly shaped by the exigencies of oral recitation, something to which the formulaic beginning of a discourse with the phrase evaṃ me sutaṃ explicitly draws attention.

Oral recitation in early Buddhism was not only considered a means for preserving texts for later generations. Oral group performance could become an expression of communal harmony, and individual recitation – besides its obvious public function for preaching purposes – appears to have been put to use privately as a tool for contemplation as well. Those involved in the oral transmission of the texts would have been predominantly monastic disciples, although lay followers also are on record for having memorized sections of the teachings and at times even for acting as teachers. In the case of Khujjuttarā, a female lay disciple apparently performed a rather crucial role in the transmission of a whole collection of discourses.