Professional Documents
Culture Documents
The Oral Transmission of Early Buddhist Literature PDF
The Oral Transmission of Early Buddhist Literature PDF
Book Review
Kurt A. BEHRENDT, The Buddhist Architecture of Gandhara. Handbuch der
Orientalistik, section II, India, volume seventeen, Brill, Leiden-Boston,
2004 by Gérard FUSSMAN............................................................................. 237
ALEXANDER WYNNE
Two theories have been proposed to explain the oral transmission of early
Buddhist literature. Some scholars have argued that the early literature was
not rigidly fixed because it was improvised in recitation, whereas others
have claimed that word for word accuracy was required when it was recited.
This paper examines these different theories and shows that the internal evi-
dence of the Pali canon supports the theory of a relatively fixed oral trans-
mission of the early Buddhist literature.
1. Introduction
Our knowledge of early Buddhism depends entirely upon the canoni-
cal texts which claim to go back to the Buddha’s life and soon afterwards.
But these texts, contained primarily in the Sutra and Vinaya collections
of the various sects, are of questionable historical worth, for their most
basic claim cannot be entirely true — all of these texts, or even most of
them, cannot go back to the Buddha’s life. There are at least two reasons
for believing this. Firstly, although the different Buddhist sects claim that
their canons were compiled at the first council of Rajag®ha (shortly after
the Buddha’s death), there is a general disagreement about the extent
and classification of this canon. Because of this, Lamotte has commented:
‘It would be absurd to claim that all those canons were fixed at the very
beginnings of Buddhism’.2 And secondly, it is hard to believe that all the
doctrinal teachings of the various Sutra-pi†akas could go back to the same
teacher, or even the same period, for they include diverse and sometimes
mutually exclusive ideas.3
1
I would like to thank Professor Richard Gombrich for reading an earlier version of
this paper.
2
Lamotte pp. 129-130.
3
For discussions of some of the different doctrinal strands, see La Vallée Poussin,
Schmithausen 1981, Bronkhorst 1985 and 1993, Gombrich 1996 (in particular, chapter 4:
‘Retracing an Ancient Debate: How Insight Worsted Concentration in the Pali Canon’) and
Wynne 2002.
4
See Albert B. Lord, The Singer of Tales (Cambridge, Mass. 1960).
5
Cousins p. 1.
THE ORAL TRANSMISSION OF EARLY BUDDHIST LITERATURE 99
Against this theory, others have argued that the early Buddhist litera-
ture is different in many ways from the sort of oral material that is formed
in performance. Norman has pointed out the following:
The great majority of Pali canonical texts, however, are in prose, and com-
plete accuracy of reproduction is required at each recitation. In these cir-
cumstances the findings of modern investigators of oral epic literature seem
to have little relevance.6
In addition to this, Gombrich has pointed out that the peculiar nature
of the early Buddhist literature makes it likely that precise wording mat-
tered in its transmission:
The early Buddhists wished to preserve the words of their great teacher,
texts very different in character from the general run of oral literature, for
they presented logical and sometimes complex arguments. The precise word-
ing mattered.7
According to this view, verbatim accuracy would have been the norm
when the early Buddhist literature was composed and transmitted. If so,
it would indeed seem that stratification of the early Buddhist literature is
possible. But what evidence is there in the early texts to support these dif-
ferent views? The views of Gombrich and Norman seem to be based on
intuition rather than the internal evidence of the literature itself. Cousins
too does not adduce much textual evidence to support his claim, but he
does present some arguments. For example, he proposes that the mate-
rial in the Suttapi†aka was formed by singers performing orally on
‘uposatha day or for the occasion of some sangha meeting’, or ‘when
visiting the sick or for recitation after receiving food at the house of a lay-
man’.8 But these suggestions hardly exhaust all the possible ways in which
the early Buddhist literature could have been recited, and in any case, tex-
tual support for them is noticeably lacking. The only textual evidence that
is presented by Cousins comes in the form of his interpretation of the four
6
Norman 1997 p. 49.
7
Gombrich 1990 p. 21.
8
Cousins pp. 4-5
100 ALEXANDER WYNNE
9
D II.123.30ff (= A II.167.31ff).
10
DOP s.v. apadesa.
11
It is possible that the word pamokkha refers to experts in the Vinaya, if it is an abbre-
viation of the expression vinaye pamokkho. On the latter expression, see Gombrich 1992
pp. 247-251.
12
D II.124.21: amukasmiµ nama avase saµgho viharati satthero sapamokkho.
13
D II.125.5: amukasmiµ nama avase sambahula thera bhikkhu viharanti bahussuta
agatagama dhammadhara vinayadhara matikadhara.
14
D II.125.24: amukasmiµ nama avase eko thero bhikkhu viharati bahussuto agatag-
amo dhammadharo vinayadharo matikadharo.
THE ORAL TRANSMISSION OF EARLY BUDDHIST LITERATURE 101
One, the bhikkhu has understood it correctly.’ Bhikkhus, you ought to con-
sider this to be the First Great Authority.15
This passage does not state what was to be done with the dhamma or
vinaya which was accepted as the Buddha’s words (bhagavato vacanaµ).
But because it is stated that the rejected teachings were to be abandoned,
we can suppose that the opposite was to be done with what had been
accepted as the Buddha’s words, i.e. if it was thought to agree with what
had already been collected under the heading of ‘Sutta’ and ‘Vinaya’, it
was to be added to them. According to Cousins, the passage shows that
there were different methods for collecting the Sutra and Vinaya material.
He interprets the passage as follows:
If something does not match with vinaya (vinaye sandissanti), it should be
rejected. This suggests an established and relatively defined set of vinaya
rules such as we know to have existed from the comparative study of sur-
viving vinaya works of various schools. Similarly something should be
rejected if it does not enter into sutta (sutte otaranti). This is an unusual
expression; it is best interpreted in the light of the Pe†akopadesa tradition
where otara∞a is one of the sixteen haras.
It may there be taken as a particular method of exegesis which links a given
discourse into the teaching as a whole by means of one of the general
categories of teaching. The Pe†akopadesa in fact specifies six possibilities:
15
D II.124.3ff: idha bhikkhave bhikkhu evaµ vadeyya: sammukha me taµ avuso bha-
gavato sutaµ sammukha pa†iggahitaµ, ayaµ dhammo ayaµ vinayo, idaµ satthu sasanan
ti. tassa bhikkhave bhikkhuno bhasitaµ n’eva abhinanditabbaµ na pa†ikkositabbaµ. ana-
bhinanditva appa†ikkositva tani padavyañjanani sadhukaµ uggahetva sutte otaretabbani
vinaye sandassetabbani. tani ce sutte otariyamanani vinaye sandassiyamanani na c’eva sutte
otaranti na vinaye sandissanti, ni††haµ ettha gantabbaµ: addha idaµ na c’eva tassa bha-
gavato vacanaµ, imassa ca bhikkhuno duggahitan ti. iti h’etaµ bhikkhave cha∂∂eyyatha.
tani ce sutte otariyamanani vinaye sandassiyamanani sutte c’eva otaranti vinaye ca san-
dissati, ni††haµ ettha gantabbaµ: addha idaµ tassa bhagavato vacanaµ, imassa ca
bhikkhuno suggahitan ti idaµ bhikkhave pa†hamaµ mahapadesaµ dhareyyatha.
The same four sources of canonical authenticity are found in the Sanskrit fragments of
the Mahaparinirva∞a Sutra, in almost exactly the same words as the Pali text — althought
there is no Sanskrit version of the Pali mahapadesa (Waldschmidt p. 238ff). If we accept
Frauwallner’s theory that Buddhism spread to the far north-west of India because of the
Asokan missions (Frauwallner pp. 22-23: ‘The mission of Kassapagotta, Majjhima and Dun-
dubhissara gave origin to the Haimavata and Kasyapiya. The mission of Majjhantika led
to the rise of the Sarvastivadin. The Dharmaguptaka school is perhaps issued from the
mission of Yonaka-Dhammarakkhita… And the community of Ceylon owes its origin to the
mission of Mahinda.’), the coincidence of the Mahaparinibbana Sutta and Mahaparinirva∞a
Sutra implies that this method of establishing canonicity preceded 250 B.C.
102 ALEXANDER WYNNE
For Cousins, then, the difference between the verbs sandissati (in the
phrase vinaye sandissanti) and otarati (in the phrase sutte otaranti) is
that sandissati means ‘match’ and implies that the ‘Vinaya’ with which
some new teaching was to be matched was relatively fixed, whereas
otarati means ‘enter into’ and implies that the ‘Sutta’ with which a new
teaching was to be compared was ‘not a set body of literature, but rather
a traditional pattern of teaching’. Therefore, Cousins implies that doctri-
nal coherence rather than historical truth was the motivating factor of
those who put together the collection of doctrinal discourses called ‘Sutta’.
Is this an accurate estimation of this passage? The difference between
the verbs used to describe the act of comparing teachings with either
‘Sutta’ or ‘Vinaya’ is certainly of some significance. Cousins’ suggestion
that otarati ought to be interpreted in the light of the Pe†akopadesa defi-
nition of otara∞a makes good sense. It probably means, as Cousins indi-
cates, that the doctrinal content of a new teaching under consideration
was to be compared with the doctrinal content of a body of oral litera-
ture called ‘Sutta’, in one of the six categories of otara∞a.17 Of course this
means that the body of literature called ‘Sutta’ is not a ‘set body of lite-
rature’, for the passage is concerned precisely with the supplementation
of the existing body of literature called ‘Sutta’. But the fact that ‘Sutta’
was not fixed during the time when the method of the four mahapadesa-
s was applied says nothing about how the individual works of that body
of literature were composed and transmitted. In fact, if we follow the
wording of the passage, the implication is that the works comprising
‘Sutta’ were transmitted word for word. We can deduce this because we
16
Cousins pp. 2-3.
17
Be and Ne (D II.66.8) both use the verb osarati. This is probably incorrect, for the
Sanskrit version of the text uses the verb ava + √t®ˆ throughout (Waldschmidt p. 238), and
thus corresponds to the PTS editions which are based mainly on Sinhalese manuscripts.
THE ORAL TRANSMISSION OF EARLY BUDDHIST LITERATURE 103
are told that the ‘words and letters’ (padavyañjanani) of the teaching
under consideration were to be ‘learnt correctly’ (sadhukaµ uggahetva)
before judgement was passed. If attention was to be paid to the words and
letters of proposed teachings, it implies that the content of what was
known as ‘Sutta’ was also transmitted by paying a similar attention to its
words and letters, i.e. that it was transmitted word for word. The passage
therefore shows that the accuracy with which a body of literature called
‘Sutta’ was meant to be transmitted was very high, down to the letter.
It was not a fixed body of literature, for it could be supplemented by com-
paring its already established doctrinal content with the doctrinal content
of new teachings, which could then be added to it. But the early Buddhists
at least attempted to transmit it accurately.
Exactly the same observations apply to the way in which the Vinaya
was formed. It cannot have consisted of an ‘established and relatively
well defined set of vinaya rules’, as Cousins supposes, because the issue
in question seems to have been the supplementation of an existing body
of literature called ‘Vinaya’ by comparing new teachings pertaining to dis-
cipline with it. Contrary to what Cousins thinks, the verb sandissati can-
not mean that the set of Vinaya rules was relatively fixed. Instead, it
seems that the verb sandissati was used because it was the standard verb
used to state that a person conforms to certain ethical or religious practices,18
or that certain practices are found ‘in’ a person or persons.19 It is under-
standable, therefore, that in the passage in question, it is asked if the words
and letters of the teaching ‘conform’ (sandissanti) to the ‘Vinaya’, for this
18
D I.102.10: api nu tvaµ imaya anuttaraya vijjacara∞asampadaya sandissasi
sacariyako ti?
D II.75.27ff (=A IV.17.8, A IV.20.20, A IV.22.7, A IV.22.24, A IV.23.9): yavakivañ
ca brahma∞a ime satta aparihaniya dhamma vajjisu †hassanti, imesu ca sattasu apari-
haniyesu dhammesu vajji sandissati.
M III.163.23ff (=S V.177.19, S V.397.7, S V.345.17, S V.345.29, S V.407.28): saµvij-
jante te ca dhamma mayi ahañ ca tesu dhammesu sandissami ti.
A V.340.31: yan’ imani bhante bhagavata saddhassa saddhapadanani bhasitani, saµvij-
janti tani imassa bhikkhuno, ayañ ca bhikkhu etesu sandissati.
19
Sn 50.18: sandissanti nu kho bho Gotama etarahi brahma∞a pora∞anaµ brahma∞a-
naµ brahma∞adhamme ti?
D III.82.11ff: ye ‘me dhamma akusala… khattiye pi te idh’ ekacce sandissanti… [brah-
ma∞e pi… vesse pi…] sudde pi te idh’ ekacce sandissanti.
A III.221.11ff: pañc’ ime bhikkhave pora∞a brahma∞adhamma etarahi sunakhesu san-
dissanti no brahma∞esu.
104 ALEXANDER WYNNE
20
Gethin 1992 p. 156.
21
E.g. the Dasuttara Sutta and the Sangiti Sutta, on which see below.
22
According to Frauwallner, this would have been after the Asokan missions c.250 B.C.
See n.15 above.
106 ALEXANDER WYNNE
23
Gethin 1992 p. 157.
THE ORAL TRANSMISSION OF EARLY BUDDHIST LITERATURE 107
imagine that they were improvised, let alone performed. And I do not
think that the peyyala sections of the Saµyutta and Anguttara Nikayas
‘appear to read more like guidelines for oral recitation and composition
than a fixed literary text.’ In my experience, the content of a peyyala sec-
tion is usually obvious. Even in the Saµyutta and Anguttara Nikayas,
which use the technique most extensively, the numerous peyyala sections
usually come after one preliminary Sutta which spells out word for word
the pattern which is to be understood for the Suttas that follow. This
hardly allows for free improvisation. For example, p. 359ff of the PTS
edition of the Sa¬ayatanavagga (Saµyutta Nikaya IV) marks the begin-
ning of the Asankhata-saµyutta (S Book XLIII = S IV book IX). This
vagga consists of forty-four Suttas24 squeezed into less than fifteen
pages, precisely because of the abbreviations achieved by the peyyala
method. But the pattern for all the Suttas is given in full in the first and
last Suttas of the vagga, and we are never in any doubt about the con-
tent of the peyyala sections.25 This is in fact the general rule of the
peyyala sections of the Saµyutta and Anguttara Nikayas. It can hardly
have been the case that there were Buddhist monks in ancient India who
chanted pe during communal recitation, when they could not be both-
ered to recite the full version of a Sutta. Nor were the peyyala sections
of the extant texts the product of oral composers, or early scribes, who
shortened texts in order that oral reciters might improvise around the
skeleton structure of the text left. They were simply the product of later
scribes who found the job of writing out lengthy sections of repetition
tedious.
The lack of evidence to support the theory that the early Buddhist lit-
erature was composed by free improvisation means that we should instead
examine the textual evidence for the alternative view in more detail.
A preliminary step in this direction has already been taken with the study
of the passage on the four mahapadesa-s; as we have seen, this supports
the claims of Norman and Gombrich, i.e. that verbatim accuracy was
24
The PTS editor, L. Feer, counts only forty-four Suttas in thie section of S IV, but as
he points out in his introduction to the edition, the total amount of Suttas would come to
1463 if each section was printed in full. This is perhaps the most extreme example of
abbreviation attained by the peyyala method.
25
In fact the full pattern, or most of it, is given in Suttas 1, 12(1), 12(45) and 44.
108 ALEXANDER WYNNE
The word dhamma is often contrasted with the word vinaya in the Pali
canon to distinguish the doctrinal discourses from the ecclesiastic law,28
and so it seems that in this pacittiya rule, the word refers to the doctrinal
teachings included in the Suttapi†aka rather than the Vinaya rules. This
much is obvious, for the law forbids the instruction of a non-monastic in
a certain way: if non-monastics were taught in certain ways by members
of the sangha, they cannot have been taught the Vinaya rules, but only
the doctrinal discourses, i.e. dhamma. And if a non-monastic was not
supposed to recite the dhamma word for word, it suggests that this was
exactly how monastics did recite it. This evidence suggests that the Sutta
portions of the early Buddhist literature were learnt verbatim among the
ordained. While the extent of the material covered by the word dhamma
is not made clear, and although the passage does not rule out the use of
improvisational methods, we have important evidence showing that the
basic literary training in early Buddhism consisted of word for word rep-
etition, and that some portion of the Suttapi†aka was transmitted in this
manner.
27
The Patimokkha p. 46.13 (bhikkhupatimokkha, suddhapacittiya 4 = Vin IV.14.20ff):
yo pana bhikkhu anupasampannaµ padaso dhammaµ vaceyya pacittiyaµ. (Norman’s
translation in Pruitt and Norman 2001 p. 47).
28
See PED s.v. dhamma, vinaya; Oldenberg p. x.
110 ALEXANDER WYNNE
In this case, you ought to approach a bhikkhu whom you consider to be par-
ticularly easy to speak to; having approached him, you should address him
thus: ‘There is a difference among you, Venerable Sirs, over both meaning
and letter. The Venerable Sirs ought to know that it is because of this (amina)
that there is a difference over both meaning and letter. Let not the Venerable
Sirs get into a dispute (vivadaµ).’
Then, you ought to approach a bhikkhu, belonging to the other faction of
bhikkhus on the other side (ekato), whom you consider to be particularly easy
to speak to; having approached him, you should address him thus: ‘There
is a difference among you, Venerable Sirs, over both meaning and letter. The
Venerable Sirs ought to know that it is because of this (amina) that there is
a difference over both the meaning and letter. Let not the Venerable Sirs get
into a dispute.’
Thus what has been badly understood (duggahitaµ) ought to be held [by all
concerned] as badly understood (duggahitato), [after which] the dhamma and
the vinaya ought to be spoken.30
Exactly the same process of regulation is described for all four cases,
although it is not clear how there could be a dispute if two parties agree
on the meaning and letter. The process of arbitration seems to involve a
neutral group of bhikkhus, i.e. the sangha at large, mediating between the
two factions, in each case appealing to the most reasonable or moderate
among them. The speaker for the sangha at large outlines the reason for
the difference (‘it is because of this that there is a difference…’), and
then appeals to what would have been one of the most basic sentiments
of the early Buddhist sangha, that is, not to let a difference break out into
a dispute (vivada), which might possibly lead to schism.31 The brief sen-
tence at the end of the passage is not exactly clear (‘Thus what has been
30
M II.239.7ff: tattha yaµ bhikkhuµ suvacataraµ maññeyyatha, so upasaµkamitva
evam assa vacaniyo: ayasmantanaµ kho atthato c’eva nanaµ byañjanato ca nanaµ. tad
amina p’etaµ ayasmanto janatha, yatha atthato c’eva nanaµ byañjanato ca nanaµ. ma
ayasmanto vivadaµ apajjittha ti. athaparesaµ ekato pakkhikanaµ bhikkhunaµ yaµ
bhikkhuµ suvacataraµ maññeyyatha, so upasaµkamitva evam assa vacaniyo: ayasmanta-
naµ kho atthato c’eva nanaµ byañjanato ca nanaµ. tad amina p’etaµ ayasmanto janatha,
yatha atthato c’eva nanaµ byañjanato ca nanaµ. ma ayasmanto vivadaµ apajjittha ti. iti
duggahitaµ duggahitato dharetabbaµ; duggahitaµ duggahitato dharetva yo dhammo yo
vinayo so bhasitabbo.
31
The early Buddhists were well aware of this danger. For example, in the Sangiti
Sutta, at the death of Niga∞†ha Nataputta a vicious dispute is said to have broken out among
the Jains (D III.210.3: vivada), and in response Sariputta is said to have appealed to the
Buddhist sangha that this should not happen to them.
112 ALEXANDER WYNNE
badly understood …’), but it seems to imagine a scenario where the two
factions have been made to acquiesce to the decision of the mediating
body not involved in the dispute. Although it is said that the letter is ‘tri-
fling’ (appamattakaµ) when there is a disagreement about the letter
alone,32 the same process of mediation and resolution is described. The let-
ter mattered.33
We do not know if this evidence records historical actuality, but the
pragmatic approach envisaged in the text suggests that the early sangha
would have resolved literary disputes in such a way. Of course, it is only
to be expected that some differences were not resolved, but that is beside
the point. The point of interest here is that the text shows that disagreements
about the exact wording of the early literature were potentially serious
affairs. Differences certainly arose, and some of them were probably not
resolved, but this passage shows that a common presupposition accepted
by all Buddhists was that teachings should be transmitted to the letter.
In the light of this evidence, it is hard to see how improvisational
methods could have been used in the transmission of the early Buddhist
literature. The learning of the Patimokkhas down to the letter, instruction
involving word for word recitation of the dhamma, and regulatory processes
which rejected wrong wording all preclude improvisation. We have
also seen that the arguments for the possibility of improvisation are not
convincing. The evidence studied thus far is particularly one-sided. But
to come to a more exact understanding of the matter, we must consider
in greater detail the genre of the early Buddhist oral literature.
32
M II.240.10/16: appamattakaµ kho pana’ etaµ yadidaµ byañjanaµ.
33
On this passage, Gethin (2001 p. 236) comments: ‘Disagreement over attha is a
potentially more serious affair. The solution proffered here seems to be that the two sides
in a dispute over attha should accept that some matters may be hard to grasp (duggahita)
others easy (suggahita).’ But the word duggahita in Pali canonical texts always means
‘badly grasped/understood’, and not ‘hard to grasp/understand’; for example, as it is used
in the Alagaddupama Sutta at M I.133.30. Gethin goes on: ‘I take it that this implies that
since difference of opinion over the satipa††hanas and so on ultimately concern quite sub-
tle matters of practical experience, bhikkhus should guard against attachment to particular
interpretations of their theoretical formulation.’ But the differences envisaged in the pas-
sage nowhere refer to different interpretations of personal experience, and its point was not
to warn against being partial to doctrine seen in the light of personal experience. The matter
is simply about the transmission of sacred literature, in its meaning and letter. As we have
seen, there is a concern not only for semantic accuracy, but also for syntactical accuracy,
and this of course has nothing to do with practical experience.
THE ORAL TRANSMISSION OF EARLY BUDDHIST LITERATURE 113
34
Barton p. 32.
35
A III.107.11ff: puna ca paraµ bhikkhave, bhavissanti bhikkhu anagataµ addhanaµ
abhavitakaya abhavitasila abhavitacitta abhavitapañña, te abhavitakaya samana abhavi-
tasila abhavitacitta abhavitapañña, ye te suttanta tathagatabhasita gambhira gambhirattha
lokuttara suññatapa†isaµyutta, tesu bhaññamanesu na sussusissanti, na sotaµ odahissanti,
na aññacittaµ upa††hapessanti, na ca te dhamme uggahetabbaµ pariyapu∞itabbaµ maññis-
santi. ye pana te suttanta kavikata kaveyya cittakkhara cittavyañjana bahiraka savakabha-
sita, tesu bhaññamanesu sussusissanti, sotaµ odahissanti, aññacittaµ upa††hapessanti, te
ca dhamme uggahetabbaµ pariyapu∞itabbaµ maññissanti.
114 ALEXANDER WYNNE
36
Schmithausen 1992 p. 117.
37
A III.251.1ff (=Pañcakanipata CCIX, Kimbilavagga IX): pañc’ ime bhikkhave adi-
nava ayatakena gitassarena dhammaµ bha∞antassa. katame pañca?
38
A III.251.4ff: gahapatika pi ujjhayanti: yath’eva mayaµ gayama, evam ev’ime
sama∞a sakyaputtiya gayanti ti.
39
Vin II.108.5: tena kho pana samayena chabbaggiya bhikkhu ayatakena gitassarena
dhammaµ gayanti. manussa ujjhayanti khiyanti vipacenti: yath’eva mayaµ gayama, evam
ev’ime sama∞a sakyaputtiya ayatakena gitassarena dhammaµ gayanti ti. (On this, see
Collins 1992 p. 25).
40
Vin II.108.21: na bhikkhave ayatakena gitassarena dhammo gayitabbo. yo gayeyya,
apatti dukka†assa ti.
THE ORAL TRANSMISSION OF EARLY BUDDHIST LITERATURE 115
of improvisation, can have been the means for the transmission of the
early Buddhist literature. In addition, there is evidence that the formal
conventions of oral transmission were affected and moulded by the
communal patterns that operated within the early sangha. The evidence
suggests that communal recitation was the norm, and this has important
ramifications for the transmission of the early literature. Such evidence
is seen, for example, in the Pasadika Sutta, where the Buddha is reported
to have advised communal recitation in response to the Jain quarrels after
the death of Niga∞†ha Nataputta:
Therefore, Cunda, as regards the teachings I have taught to you through
understanding, meeting together again and again, [comparing] meaning with
meaning (atthena atthaµ), [comparing] letter with letter (byañjanena byañ-
janaµ), you should recite communally and not argue, so that the holy life will
be long lasting and endure long, which will be for the benefit of the many,
for the happiness of the many; [do it] out of compassion for the world, for
the purpose, welfare [and] happiness of gods and men.41
This evidence suggests that the early Buddhists used the example of the
Jain quarrels about their literature in order to ensure the accurate trans-
mission of their early teachings. There are important implications if com-
munal recitation was the foremost means of oral transmission in the early
Buddhist sangha, as has been noted by Mark Allon:
[C]ommunal or group recitation or performance requires fixed wording.
It is not possible for more than one individual to perform at the same time
in the manner described by Parry-Lord without producing utter chaos, for
in that method, each individual creates his compositions anew each time he
performs.42
41
D III.127.15ff: tasmat iha cunda ye vo maya dhamma abhiñña desita, tattha sabbeh’
eva saµgamma samagamma atthena atthaµ vyañjanena vyañjanaµ saµgayitabbaµ na
vivaditabbaµ, yathayidaµ brahmacariyaµ addhaniyaµ assa cira††hitikaµ tad assa bahu-
janahitaya bahujanasukhaya lokanukampaya atthaya hitaya sukhaya devamanussanaµ.
42
Allon p. 366.
116 ALEXANDER WYNNE
this involved social conditions very different from those in which improv-
isational and performance based methods could have functioned. The evi-
dence for group recitation is not limited to the Pasadika Sutta — similar
evidence is found in the Sangiti Sutta,43 a name which perhaps indicates
that it was composed at a major communal recitation.44 If this was the
case, it is likely that the same compositional conditions applied to the
very similar Dasuttara Sutta, which would rule out Gethin’s suggestion
that differences in its various sectarian versions were produced by the
oral improvisation of individual chanters.
43
D III.211.3: tattha sabbeh’ eva saµgayitabbaµ na vivaditabbaµ. This is exactly the
same pericope as that found in the Pasadika Sutta, but minus the section saµgamma sama-
gamma atthena atthaµ vyañjanena vyañjanaµ.
44
This point has been made by Cousins (1983 p. 4): ‘So far as I know, it has not actu-
ally been suggested that it may well have been recited at one of the Councils. Yet its name
clearly indicates that it is intended for chanting together and this surely means at a Sangiti.’
45
Schopen 1997 p. 3.
THE ORAL TRANSMISSION OF EARLY BUDDHIST LITERATURE 117
But it is surely naïve to treat the entire contents of the canonical liter-
ature in this way. It cannot be doubted that much of the canonical liter-
ature is normative, but this does not warrant the assertion that it contains
no descriptive or historical elements. In fact, some quite simple steps
can be taken to separate normative from descriptive material. We must
differentiate direct evidence from indirect or circumstantial evidence.
In legal parlance, circumstantial evidence is not the direct evidence of a
witness who claims that he saw or heard something, but is a singular fact
that can be used to infer another fact. In other words, circumstantial evi-
dence is the indirect, unintentional evidence which affords a certain pre-
sumption. In the context of the early Buddhist literature, circumstantial
evidence is not the direct evidence contained in the Buddhist texts, e.g.
that the Buddha said such and such a thing on such and such an occasion
(which may be true or false), but consists of the indirect facts from which
other facts can be inferred. So, for example, the Buddha’s advice in the
Kinti Sutta on how to resolve potentially schismatic disputes is direct
evidence that can neither be substantiated or denied;46 we do not know
if the Buddha said such a thing, or if the sangha followed the ideals said
to have been set out by him on that occasion. This is the normative evi-
dence of which Schopen is so suspicious. But the circumstantial evidence
contained in this passage consists of the fact that the early sangha imag-
ined that there could be disputes about the precise wording of the early
oral literature. From this we can infer that a priority of the early sangha
was the precise wording of the literature, and that efforts were made to
transmit the early Buddhist compositions accurately, to the letter. If this
was not the case, the text would never have said that disputes could arise
over the letter.
In a similar vein, all the other evidence that I have presented is circum-
stantial and from it we can infer that the early Buddhists really did attempt
to transmit the early Buddhist literature with verbatim accuracy. So in
the passage on the four mahapadesa-s,47 the way new teachings were said
to be compared with the existing collections of ‘Sutta’ and ‘Vinaya’ means
that these existing collections must have been transmitted verbatim —
46
See section 5.3.
47
See section 3.
118 ALEXANDER WYNNE
otherwise it would never have been said that the words and letters of the
new teachings had to be learnt correctly (sadhukaµ). The description of
the vinayadhara is surely an ideal,48 but the ideal of learning both the
Patimokkha-s to the letter reflects the fact that this was how the early
Buddhists attempted to transmit it in early times. The Vinaya rule for-
bidding non-monastics permission to be taught the dhamma does not
allow us to conclude that such things never happened,49 but the rule would
not have stipulated that the teaching was not to be ‘word by word’
(padaso) unless that was how monastics were in fact taught. And the
Pasadika Sutta would never have said that when bhikkhus should gather
together to recite communally,50 the recitation should be ‘letter by letter’
(byañjanena byañjanaµ) unless that was how the early Buddhists attempted
to transmit the literature.
The worth of this circumstantial evidence should not be underesti-
mated. But it follows from the rule of drawing inferences from circum-
stantial evidence that the prohibitions against bhikkhus singing the dhamma
in a drawn out voice,51 and the evidence warning that in the future there
will be those who compose poetic sorts of Suttas,52 imply that there were
bhikkhus who sang their compositions in drawn out voices and who com-
posed poetic sorts of Suttas. This must indeed be so, but Suttas which may
have been sung in a ‘drawn out voice’ are not evidence that improvisa-
tional methods were used in the transmission of literature; in any case,
this passage shows that even this prohibited recitation was communal
and therefore not improvisational (the offenders were the six bhikkhus).
Moreover, as far as I can tell, there are no Suttas in the early canonical
collections that could be described as poetic with ‘ornamental syllables
and letters’. And the important issue, surely, is the correct determination
of the methods by which the extant texts were transmitted, not texts which
may have existed. Even if others can find some of these ‘heretical’ Suttas
in the early literature, there can be no evidence that they were composed
through improvisational means.
48
See section 5.1.
49
See section 5.2.
50
See section 6.
51
See section 6.
52
See section 6.
THE ORAL TRANSMISSION OF EARLY BUDDHIST LITERATURE 119
53
Hultzsch p. 173: Vinaye-samukase Aliya-vasa∞i Anagatha-bhayani Muni-gatha Moneya-
sute Upatisa-pasine e cha Laghulovade.
54
Norman 2001 p. xxxiii.
55
See Schmithausen 1992 p. 116-117, Jayawickrama pp. 230-32.
56
Oldenberg p. xl.
57
Schmithausen 1992 p. 115: ‘It must refer to a prose text’.
58
See p. 101 n.15 above.
120 ALEXANDER WYNNE
9. Conclusion
Although the early Buddhist texts include information on their own
transmission, there is no canonical evidence to suggest that improvisation
in performance was a factor in their transmission, and it would seem
that the arguments for it have been overstated. Cousins, for example, has
commented: ‘The kind of divergence and variation in the oral tradition
suggested here is not simply an inference from the pattern of most but
not all forms of oral literature so far studied. It has a much firmer basis.
It is precisely this kind of variation which is actually found in the different
versions of the four nikayas preserved by various sects and extant today
in Pali, Sanskrit, Chinese and Tibetan’.59 But no evidence to support this
claim is presented, and we are forced to conclude that this argument is
precisely an inference based upon ‘the pattern of most but not all forms
of oral literature so far studied’. In a similar vein, Gethin has said that ‘The
earliest Buddhist literature was composed orally and built up around
lists’,60 lists which ‘provide or form a matrix within which she [the com-
poser] can improvise.’61 He argues that the differences between the various
versions of the Dasottara Sutra have been caused by this sort of improvi-
sation. But this argument also seems to be an inference based on the study
of other oral literatures, rather than a study of the early Buddhist evidence.
As far as I can tell, there is in no clear evidence in the Pali canon which
supports these claims, and much that goes against them, as I have attempted
to show.
The evidence on the literary history of early Buddhism presented here
concerns only the transmission and not the composition of the extant texts.
Furthermore, we must admit that this evidence reflects a well developed
literary tradition, and not the conditions which existed at the beginning of
Buddhist composition. So how were the texts composed in the first place?
No one can deny that there must have been a period of free literary
transmission at the beginning of Buddhism. After the Buddha’s death, every
bhikkhu or bhikkhu∞i would have remembered a number of stories about
the Buddha’s life and teaching, some of which they witnessed themselves,
59
Cousins 1983 p. 5
60
Gethin 1992 p. 166.
61
Ibid. p. 156.
THE ORAL TRANSMISSION OF EARLY BUDDHIST LITERATURE 121
and others which they heard second-hand. In the beginning it is likely that
such stories and teachings, based on the collective memory of the early
sangha, were not fixed and circulated freely. Does this mean that it was
a period when improvisational methods were used? Possibly, but if so,
the improvisational methods would have been very different from those
studied by Parry and Lord. The techniques studied by Parry and Lord
presuppose a developed literary tradition which utilises fixed building
blocks, i.e. set phrases and pericopes (strings of words conveying vari-
ous ideas), around which a composition could be strung. These building
blocks would not have been known in a period of free transmission, and
so it seems that the methods studied by Parry and Lord simply cannot have
been used at this time. But what about the situation at the beginning of
a developed literary tradition, which would have been initiated by the
appearance of oral building blocks, and which eventually produced the
extant texts?
We can imagine that in an early period when the composition of such
oral building blocks had begun, there may well have been a period when
they were used by individual singers in improvised performances. Indeed,
Mark Allon has commented that stylistic features of some these pericopes
could have functioned as compositional aids ‘within a tradition of com-
posing material during the performance in an improvisational manner and
in a tradition of composing fixed texts which were to be transmitted ver-
batim.’62 At the same time, he points out that the gross repetition found
in the Digha Nikaya Suttas, although based on the use of these pericopes,
cannot reflect a tradition of improvisational performance.63 But there are
many more Suttas which are shorter and which use the same pericopes
without the same level of repetition — could they be the product of
improvisation in performance? This seems to be most unlikely, for a very
simple reason.
It is clear that the building blocks of the early Buddhist texts must
have been composed in a collaborative effort. Mark Allon’s detailed study
of the pericopes used in the Digha Nikaya shows a literary tradition of
great complexity and sophistication. The mnemonic techniques used to
62
Allon p. 365.
63
Ibid. p. 365.
122 ALEXANDER WYNNE
compose the pericopes from which the extant texts were fashioned can
hardly have been the work of individual composers working in isolation
from one another. It must have been a joint effort. And if we are to sup-
pose that a joint effort was required to compose these building blocks, we
must imagine that the communal factor which determined communal
recitation and word for word transmission would have come into existence
at the very beginning of Sutta composition. In other words, I am sugges-
ting that if there was collaboration from the beginning, it can hardly have
been the case that the collaborators composed oral building blocks and
then went off, leaving the pericopes they had fashioned to be used by
individuals as they liked. Surely it was the case that the various peri-
copes, the building blocks of the early Buddhist texts, were fashioned by
committees in order to turn a growing body of loose material into a form
that could be more easily remembered. It must have been a joint project
involving many participants, and this implies that not only the pericopes
but also their distribution within texts would not have been a matter of
improvisation.
Of course, once a sophisticated literary tradition — one that uses oral
building blocks — is up and running, then it is quite likely that new mate-
rial would have been composed using these building blocks. New com-
posers, perhaps who wished to say new things, would have fashioned
new texts out of the building blocks which had already been composed.
But this does not necessarily mean that these new compositions would have
been subject to an improvisational method. As we have seen, the method
of integrating the new into the old suggested by the passage on the four
mahapadesa-s does not allow any room for an improvisational method.
Any new composition would have been presented to the sangha as a whole,
and, if accepted, transmitted verbatim by the collaborative method of the
early sangha.
Group recitation and word for word accuracy does not mean that com-
position and transmission were carried out in a mechanistic and regi-
mented fashion, producing closed and canonised texts. On the contrary,
as long as the canon was not closed, and as long as oral composition and
transmission continued, some degree of variation could not have been
prevented, especially when Buddhism was spreading to the far corners of
the Indian subcontinent and beyond. Every measure was taken to ensure
THE ORAL TRANSMISSION OF EARLY BUDDHIST LITERATURE 123
that the early literature was as fixed and accurate as it could be under the
circumstances, but it was never written in stone. Variation in the parallel
texts of the different sects was only to be expected, for many variables
existed in the post-Asokan age. The parallel texts of the different sects
show us that the arrangement and even the language of compositions could
vary considerably,64 and from the amount of texts unique to certain sects,
it seems that there was an ongoing composition of new texts which could
be incorporated into the Nikayas/Agamas. In such circumstances, we must
imagine that the ancient guardians of the early Buddhist literature in the
post-Asokan period had a significant amount of redactional authority.
This redactional authority allowed some freedom in dealing with the sacred
literature, and it is because of this freedom that the differences in the par-
allel texts of the different sects appeared.
The model of oral composition and transmission which I am suggesting
can be summed up as follows:
Generally speaking we may envisage things as follows. At the beginning
there is a time of free transmission, during which the text is rendered in
free words from memory. Memorial sentences, mostly couched in the form
of verses, probably came early to the help of the memory. This sort of trans-
mission has always been employed with less important texts. I recall, e.g.
how the Jains fixed down by means of memorial sentences the contents of
the legends which they inserted in their sermons, but left the execution in
detail to the reciter.65 The passage to an established tradition is marked by
the appearance of fixed formulae. These are known to everybody from the
Buddhist and Jaina tradition. Wherever a subject of common recurrence is
treated, it is treated in the same words. Also the descriptions regularly
repeated in the Jaina canon belong to this class. This gradually leads to an
established tradition, which fixes the text in a certain version. But even such
an established tradition is never rigid as with the Vedic texts. Chiefly with
the Buddhists we remark even later frequent modifications of redactional
nature. To these belong the above discussed levelling tendencies, which led
to the uniforming of the verbal expressions of similar texts in the various
64
As pointed out by Allon p. 367.
65
This method may explain the composition of the Jatakas, where canonical verses
are separated by long tracts of non-canonical prose commentary. Such a method of trans-
mission is exactly what one would expect under the conditions of an improvisational model
of free transmission. But the prose passages in the Jatakas that may have been left to the
whim of the reciter are not considered to be canonical in the Theravada tradition, and there
is of course no evidence that they were ever transmitted with a degree of improvisation.
124 ALEXANDER WYNNE
Alexander Wynne,
St John’s College, Oxford.
66
Frauwallner pp. 173-74.
THE ORAL TRANSMISSION OF EARLY BUDDHIST LITERATURE 125
Abbreviations
A Anguttara Nikaya
Be Burmese edition
D Digha Nikaya
DOP Dictionary of Pali (= Cone)
M Majjhima Nikaya
Ne Nalanda-Devanagari-Pali-Series. Kashyap, Bhikkhu J.; Bihar: Pali Pub-
lication Board.
PED Pali English Dictionary (= Rhys Davids and Stede)
PTS Pali Text Society
S Saµyutta Nikaya
Vin Vinaya Pi†aka
Bibliography
Allon, Mark (1997). Style and function: a study of the dominant stylistic features
of the prose portions of Pali canonical sutta texts and their mnemonic func-
tion. Tokyo.
Barton, John (1996). Reading the Old Testament, Method in Biblical Study. Second
edition. London: Darton, Longman and Todd. First edition, 1988.
Bronkhorst, Johannes (1985). ‘Dhamma and Abhidhamma’. Bulletin of the School
of Oriental and African Studies 48, pp. 305-320.
—. (1993). The Two Traditions of Meditation in Ancient India. New Delhi: Moti-
lal Banarsidass. First published, Stuttgart: Steiner Verlag.
Collins, S. (1992). ‘Notes on some oral aspects of Pali literature’. Indo-Iranian
Journal 35 no.2/3, pp. 121-135.
Cone, Margaret (2001-). A Dictionary of Pali. Oxford: Pali Text Society.
Cousins, L.S. (1983). ‘Pali Oral Literature’. In Buddhist Studies — Ancient and
Modern, ed. p. Denwood and A. Piatigorsky, pp. 1-11. London.
De Jong, J.W. (1974). ‘A Brief History of Buddhist Studies in Europe and America’.
The Eastern Buddhist, vol. VII no.1, May 1974, 55-106.
Frauwallner, Erich (1956). The Earliest Vinaya and the Beginnings of Buddhist
Literature. Rome.
Gethin, R.M.L. (1992). ‘The Matikas: memorization, mindfulness and the list’.
In the Mirror of Memory, Reflections on Mindfulness and Remembrance in
Indian and Tibetan Buddhism, ed. J. Gyatso, pp. 149-172. Albany: State Uni-
versity of New York Press.
—. (2001). The Buddhist Path to Awakening: A Study of the Bodhi-Pakkhiya
Dhamma. Second edition. Oxford: Oneworld. First edition, Leiden: E.J.
Brill.
126 ALEXANDER WYNNE
Gombrich, R.F. (1990), ‘How the Mahayana began’. The Buddhist Forum: Semi-
nar Papers 1987-1988, ed. T. Skorupski (London: SOAS), pp. 21-30.
—. (1992). ‘Dating the Historical Buddha: A Red Herring Revealed’. The Dating
of the Historical Buddha Part 2, ed. Heinz Bechert, pp. 237-259. Göttingen:
Vandenhoeck & Ruprecht.
—. (1996). How Buddhism Began: The Conditioned Genesis of the Early Teach-
ings. London: Athlone Press.
Hinüber, Oskar von (1996). A Handbook of Pali Literature. Berlin: Walter de
Gruyter & Co.
Hultzsch, E. (1991). Corpus Inscriptionum Indicarum Vol. I: Inscriptions of Asoka.
New Delhi: The Director General Archaeological Survey of India. First
Published: Oxford, Clarendon Press for the Government of India, 1925.
Jayawickrama, N.A. (1948). ‘The Vaggas of the Suttanipata’. The University of
Ceylon Review, VI. No.4, pp. 229-56.
La Vallée Poussin, Louis de (1937). ‘Musila et Narada’. Mélanges chinois et
boudhiques 5, pp. 189-222.
Lamotte, Etienne (1988). History of Indian Buddhism, from the Origins to the Saka
Era, trans. Sara Webb-Boin, (Louvaine: Université Catholique de Louvain).
First published as Histoire du Bouddhisme Indien, des origines a l’ère Saka
(Louvaine, 1958: Bibliothèque du Muséon).
Bhikkhu Ña∞amoli and Bhikkhu Bodhi (1995). The Middle Length Discourses
of the Buddha: A Translation of the Majjhima Nikaya. Boston: Wisdom
Publications.
Norman, K.R. (1997). A Philological Approach to Buddhism: the Bukkyo Dendo
Kyokai Lectures 1994. London: School of Oriental and African Studies.
—. (2001). The Group of Discourses (Sutta-nipata). Second edition. Oxford: Pali
text Society.
Oldenberg, Hermann ed. (1997). The Vinaya Pi†akaµ: One of the Principal Bud-
dhist Holy Scriptures in the Pali Language. Vol. I. The Mahavagga. Oxford:
The Pali Text Society. First published: Williams and Norgate, 1879.
Pruitt, William and Norman, K.R. (2001). The Patimokkha. Oxford: Pali Text
Society.
Rhys Davids, T.W. and Stede, William (1921-1925). Pali-English dictionary. Lon-
don: Pali Text Society.
Schmithausen, Lambert (1981). ‘On Some Aspects of Descriptions or Theories
of ‘Liberating Insight’ and ‘Enlightenment’ in Early Buddhism’. Studien
zum Jainism und Buddhismus, ed. Klaus Bruhn and Albert Wezler, pp. 199-
250. Wiesbaden: Franz Steiner.
— (1992). ‘An Attempt to Estimate the Distance in Time between Asoka and the
Buddha in terms of Doctrinal History’. In The Dating of the Historical Bud-
dha Part 2, ed. Heinz Bechert, pp. 110-147. Göttingen: Vandenhoeck &
Ruprecht.
THE ORAL TRANSMISSION OF EARLY BUDDHIST LITERATURE 127