The Composition and Transmission of Early Buddhist Texts with Specific Reference to Sutras

Hamburg Buddhist Studies 17, Mark Allon, Hamburg University Press (2021)

Foreword

Mark Allon’s work compares different versions of texts in Pali, Gandhari, Sanskrit, and Chinese to understand how they were composed and transmitted.
He argues that early texts were oral compositions intended for verbatim memorization and transmission.
Changes that occurred during their diffusion were generally intentional, not accidental.
Allon contends that new texts were created to address doctrinal needs and maintain coherence, not just by creatively combining existing formulas.
He suggests that different versions of a text are formal, community-sanctioned “editions” rather than records of individual oral performances.

Preface and Acknowledgements

The book originated from a paper delivered at “The Idea of Text in Buddhism” conference at the Hebrew University of Jerusalem in December 2019.
The author thanks several scholars for their comments and assistance, including Eviatar Shulman, Bhikkhu Anālayo, Jens-Uwe Hartmann, and Richard Salomon.
He also acknowledges colleagues who checked Chinese and Tibetan references and thanks Chiara Neri for her support.

Chapter 1 - Introduction

The author’s previous work argued that early Buddhist texts (Pali suttas) were designed for verbatim memorization and transmission, a necessity for the practice of communal recitation.
Despite the need for fixed wording, variations exist between parallel versions of texts, indicating that Buddhist communities did make changes over time.
This work will examine the stylistic features that support fixed transmission, explore how and why texts were changed, and address recent scholarly arguments that favour improvisation over verbatim memorization in the composition of these texts.
“Early Buddhist texts” are defined as canonical sutras and verse collections, with the oldest manuscripts dating to the 1st century BCE, though the core material is likely pre-Aśokan (c. 380-350 BCE).
The author discusses the scholarly debate on whether similarities between parallel texts point to an ancient common source (the author’s preferred view) or later harmonization and borrowing between schools.
For this study, a “sutra” is defined as a specific textual unit from the main canonical collections (nikāyas/āgamas) that reports the words of the Buddha or his followers.
Early Buddhist sutras had two primary functions: to instruct and guide followers by recording teachings and events, and to act as inspiration and propaganda to attract converts and financial support.

Chapter 2 - The Stylistic Features of Sutra Prose and What They Reveal About the Composition and Transmission of these Texts

Early Buddhist sutras predominantly use prose, likely because it was the preferred medium for oral texts at the time and allowed for stylistic features like repetition that aided in transmission.
These texts are not verbatim records but are highly structured, stylized, formulaic, and repetitive literary constructs.
Many sutras share a typical, predictable structure: 1) an introductory setting, 2) a dialogue interchange, 3) the main teaching, and 4) a concluding formula.
The teaching sections themselves are also highly structured, often using a pattern of statement, rhetorical question, list, repetitive definition of each list item, and a final summary.
Prose is constructed using a “building block” approach, frequently employing strings of grammatically parallel words (verbs, nouns, adjectives) that express similar ideas.
These strings are often arranged according to the “Waxing Syllable Principle,” where each unit has an equal or greater number of syllables than the preceding one, creating a crescendo effect. They also feature sound and metrical similarities.
The composition can be highly artificial, with rare words sometimes chosen or created specifically for a string to fit its sound and syllabic pattern.
Some canonical prose is metrical, following a pattern called vedha. The well-known formula describing the Buddha, Dhamma, and Sangha is a prime example, but looser metrical patterns appear in other common formulas as well.
Repetition is used on a massive scale; one analysis showed that almost 87% of a particular sutta consisted of quantifiable repetition.
The use of standardized formulas to describe common concepts, actions, or events is a dominant characteristic.
These stylistic features were designed to facilitate memorization and faithful transmission of the texts, but they also have aesthetic, poetic, and psychological functions, enhancing the material’s impact on the audience.

Chapter 3 - References within Canonical Texts to Texts being Memorized and Recited Communally

Early Buddhist texts contain numerous references to the recitation and learning of texts, providing evidence that they were memorized and transmitted verbatim.
The monk Sona is described reciting all sixteen suttas of the Aṭṭhakavagga from memory, for which the Buddha praises him, suggesting memorization was part of monastic training.
Laypeople also memorized texts, as shown by the householder Hāliddikāni quoting verses from the Aṭṭhakavagga and the Sakkapañha-sutta, and the laywoman Nandamātā reciting the Pārāyana.
Suttas in the Aṅguttara-nikāya state that for the Dhamma to continue, learned monks must master the tradition and teach the suttantas to others so they are not lost.
The Pātimokkha (monastic rules) was learned and recited regularly from early on, with mastery being a requirement for ordaining others. The Vinaya provides detailed rules for its recitation.

Chapter 4 - The Formation of Sutra and Verse Collections, and the samgītikāras and bhāṇakas

Early Buddhist texts like sutras and verses were systematically organized into larger collections (nikāyas/āgamas, Dhammapada, etc.) to ensure faithful transmission.
These collections were structured using principles such as genre, size, purpose, numerical order, and subject matter.
Larger collections were broken down into smaller, manageable sub-divisions like vaggas (groups of ten), samyuttas, nipātas, bhānavāras (recitation sections), and paññāsakas (groups of fifty).
Several organizational principles were used to arrange texts within these sub-divisions:
- Genre: Grouping texts that are all verse, or mixed prose and verse.
- Numerical Principle: Ordering texts by the number of verses they contain or by internal numerical features, a practice also found in pre-Buddhist and Jain texts.
- Subject Matter or Keyword: Arranging texts that deal with a common topic or share a key word.
- Individuals Involved: Grouping texts based on the speaker, the person being addressed, or a class of individuals (e.g., monks, kings).
- Literary Features: Collecting texts that share a common figure of speech, such as a simile.
- Concatenation: Forming pairs or short sequences of texts linked by a shared keyword, subject, or structure.
Some collections contain “artificial” sutras, which appear to have been generated by splitting a single discourse into multiple texts or by applying a formulaic structure to a list of items. This practice created repetition that aided memorization and served as a meditative exercise.
Canonical collections were not just arrangements of pre-existing texts; new sutras were sometimes composed to fit a specific context, develop a theme, or present teachings in a particular order.
Uddānas (mnemonic verses) were used as a para-textual tool to control the contents and order of a collection. These verses list keywords for each text in a division, ensuring its integrity.
Specialists known as bhāṇakas (reciters) were responsible for memorizing and transmitting specific collections. Inscriptions from the 2nd century BCE confirm the existence of bhāṇakas for collections like the Majjhima-nikāya and Saṃyutta-nikāya.
The complex and systematic organization of these texts indicates they were the product of a significant group effort involving communal memorization, recitation, and validation to ensure the preservation of the teachings.

Chapter 5 - The Main Differences Between Parallel Versions of Early Buddhist Texts and Accounting for these Differences

5.1. Introduction

Parallel versions of early Buddhist texts exhibit numerous differences within an overall similarity.
Major types of differences include: missing or added episodes; different sequences of events or teachings; variations in information, wording, and grammar; different names for people and places; and different arrangements of texts within larger collections.
Factors contributing to these changes include: language shifts, oral transmission traditions (bhānaka), geographical isolation, lack of central authority, the formation of different schools (nikāyas), a dynamic view of the Buddha’s word, an emphasis on meaning over wording, and the impact of writing and manuscript transmission.
Changes can be unintentional (e.g., memory lapses, accidental inclusion of glosses) or intentional (conscious modification). Many differences are too significant to be unintentional and required conscious acceptance by the community to become standard.

5.2. Change of Language

Buddhist communities were often willing to change the language of their texts to suit new audiences, as shown by the preservation of texts in Pali, Gandhari, Sanskrit, Chinese, and Tibetan.
Translating between Indian dialects or into Sanskrit involved systematic changes in phonology, morphology, and syntax.
The process of Sanskritization is visible in late Gandhari manuscripts, which show an inconsistent application of Sanskrit features.
Translating verse into Sanskrit was particularly challenging due to metrical rules, often forcing translators to intentionally alter word order, substitute synonyms, or change the meaning to maintain the meter.

5.3. Modification of the Wording

Parallel texts often show significant, intentional modifications in wording, structure, and plot, going beyond simple translation changes.
The Sanskrit version of the Buddha’s last words is more elaborate than the Pali version, including a scene where the Buddha exposes his body to illustrate impermanence, which may reflect an intentional ideological choice.
Sanskrit and Gandhari versions frequently expand upon or clarify phrases that are simpler or more obscure in Pali, such as specifying that homage was paid by touching the Buddha’s feet with one’s head.
Narrative portions of sutras, like the introduction to the Sāmaññaphala-sutta, show creative and intentional development of the plot, with different versions adding characters, dialogue, and events to make the story more compelling.
While it is often assumed that the Buddha’s direct doctrinal statements are more stable, comparisons show that these too can have significant variations, with different versions using different stock phrases or structuring the teaching differently.
Even the Pātimokkha/Prātimokṣa (monastic rules), a fixed and regularly recited text, shows intentional modifications across different school versions, often to clarify meaning or smooth out phrasing.
In some cases, formulas were simplified over time. The standard opening and closing formulas of many sutras are significantly shorter in Sanskrit versions than in Pali, likely originating from manuscript abbreviations that were later standardized.
Similar variations in wording and structure for the same event can also be found within the different collections of the Pali canon itself, likely due to the different functions of the collections or the traditions of distinct reciter groups (bhānakas).

5.4. Rearranging Sutra and Verse Collections and Creating new Sutras and Verses

One of the most significant forms of intentional change was the rearrangement of sutras within collections. Different schools ordered the sutras in their āgamas/nikāyas differently.
Schools expanded their collections by intentionally transferring sutras from other collections (e.g., from the Madhyamāgama to the Dīrghāgama) and sometimes modified the transferred sutras to fit their new context.
There is substantial evidence that communities also created entirely new sutras and verses. Several collections, such as the Sanskrit and Chinese Dīrghāgamas, contain sutras that have no known parallels, indicating they were later compositions unique to that school’s lineage.
The creation of new sutras often involved compiling and systematizing teachings and cosmological information found in other existing discourses.

Chapter 6 - Recent Scholarship on the Composition of early Buddhist Texts, and Initiating and Adapting to Change

Early Buddhist texts underwent intentional changes despite being designed for memorization and oral transmission. This chapter discusses how communities adapted to these changes and responds to recent scholarship on the topic.
The author refutes McGovern’s theory that early texts were improvised in performance using formulas. Instead, the author argues that communities memorized fixed texts but were willing to create new “editions” where meaning was more important than exact wording.
The high degree of repetition in sutras makes them unsuitable for improvised performance, and the fact that even the memorized monastic rules (Pātimokkha) were intentionally changed supports the idea that sutras were also fixed but editable texts.
Formulas themselves vary between traditions and are not the primary memorized unit; they are the means by which a text’s plot, idea, and structure are realized.
The author critiques Shulman’s view that formulas are the primary textual element, arguing that plot and purpose are more fundamental. While agreeing that formulas have aesthetic and literary functions, their primary role is to aid the composition and transmission of fixed texts.
The process of changing texts was likely a top-down decision by senior monastics, with changes being introduced gradually over several centuries.
The difficulty of adapting to a change would have been limited to the single generation of reciters who had to learn the new version.