On the Authorship of Kaccāyana, the Oldest Pāli Grammar

A. RUIZ-FALQUÉS¹

1. Single or multiple authorship of Kaccāyana

The Kaccāyana (Kacc) Pāli grammar was composed around the sixth century CE, most probably in India, for it was unknown in Sri Lanka until a later date. This work, partly original and partly based on Pāṇini and Kātantra (Kā) is the oldest extant Pāli grammar. It is also the most influential vyākaraṇa text in the history of Theravāda Buddhism. The authorship of this text, however, remains disputed. An old Sinhalese tradition identifies Kaccāyana with Mahā Kaccāyana, the disciple of the Buddha and alleged author of the Nettippakaraṇa and the Mahānirutti (Ap-a 491, 17-21 (ad Ap 531); A I 23, 16-28). This view is today dismissed by most scholars. Even orthodox Theravādin-s find it difficult to accept that an imperfect treatise such as Kacc could have been written by an impeccable arahant such as Mahā Kaccāyana. The name Kaccāyana, therefore, must represent a different author. PIND suggests that Kacc could be the work of one or more compilers rather than authors: “Kacc is a compilation of various hands” (PIND 2012: 75). This is the reason that scholars have described it as chaotic and devoid of any systematic arrangement. Now, when we say “various hands” it is not clear whether we mean a definite or indefinite number of authors. A more or less established consensus among scholars, after D’ALWIS (1863: 104–105), divides the work in three layers of text. In the Pāli grammatical tradition, each layer is ascribed to a particular author. Thus, the sutta text, that is to say, the grammatical aphorisms, are ascribed to a certain Kaccāyana, the vutti (“gloss”, Kacc-v) is ascribed to a certain Saṅghanandin, and the payoga (“examples”) section is ascribed to a certain Brahmadatta. These three texts make the Kacc proper. The Nyāsa of Vimalabuddhi, otherwise known as Mukhamattadīpanī (Mmd) is considered the fourth layer of the corpus, but I will not concern myself with this work, as it has a textual tradition of its own.

In order to illustrate the threefold nature of Kacc, let us examine Kacc 336 sare an “[when deriving a compound], before a vowel an [replaces na]”. The following is the text that we would find in a manuscript:

sare an. nassa padassa tappurise uttarapade sabbasseva anādeso hoti sare pare. anasso. anissaro. anariyo. aniṭṭho.²

Here we can distinguish three sections. The main rule, sare an, is the first. What follows is a gloss giving the context that applies to the rule, that is to say, conditions from previous sutta-s that implicitly “recur” (anuvattanti) or “are still valid” (vattanti), that is why this second section is called vutti. Finally, we have a series of words that exemplify the “application” (payoga) of the rule.

Some sutta-s in Kacc include even another type of section at the end. This section consists of questions and answers about the purpose and scope of the words used in the previous sections. For instance, Kacc 9 parasamaññā payoge:

parasamaññā payoge. yā ca pana sakkataganthesu samaññā ghosā ti vā aghosā ti vā tā payoge sati ettha pi payuñjante. ga gha ṅa ja jha ña ḍa ḍha ṇa da dha na ba bha ma ya ra la va ha ḷa iti ghosā. ka kha ca cha ṭa ṭha ta tha pa pha sa iti aghosā. ghosāghosa icc anena kv attho. vagge ghosāghosānaṃ tatiyapaṭhamā (Kacc 3,1–8).

The initial statement parasamaññā payoge is the sutta, which means “Concepts of others [should be used] when they apply”. The vutti explains that concepts, such as ghosa “aspirate” and aghosa “unaspirate”, defined in Sanskrit treatises, are to be used when there is occasion to apply them (payoge sati). We subsequently find a list of ghosa and aghosa consonants. This part already belongs to the payoga. The last section consists of a question about the purpose of saying “ghosāghosā” and the answer giving the place in Kacc where these two concepts are used [i.e. Kacc 29]. That type of section could also be considered payoga, for it specifies the scope for the application of the sutta. In this case and in other cases, it is very important to observe how the payoga does not comment upon the sutta (allegedly composed by Kaccāyana) but upon the vutti (allegedly composed by Saṅghanandin). In his critical edition of the text, PIND establishes a simpler division that overlooks this fact. He separates the sutta text, that is Kacc, from the rest, which he calls Kaccāyanavutti (Kacc-v). According to PIND, it is practically impossible to determine the authorship, single or multiple, of Kacc. That is why, I think, he has adopted a straightforward and practical policy when editing the text.

In this paper I do not intend to solve the question of the Kacc authorship. I will simply problematise it taking into consideration some facts that have so-far been overlooked.

2. Origin of the fourfold division of the Kaccāyana system

James D’ALWIS was the first modern scholar to raise the question of the triple authorship of Kacc. In his essay on this grammar, he quotes a passage from the “Kaccāyanabhedaṭīkā”, actually meaning, as PIND already pointed out, the Kaccāyanabheda-navaṭīkā (Kacc-bh-nṭ, not to be confused with the old Kaccāyanabheda-porāṇaṭīkā). D’Alwis quotes the entire passage, which deals with the number of sutta-s in every chapter of Kacc and it basically contains a quotation of an older work, now lost, called Kaccāyana-dīpanī. Ariyālaṅkāra, the author of Kacc-bh-nṭ points out that the number of suttas according to Kaccāyana-dīpanī does not match the number of Kacc sutta-s embedded in the Nyāsa. And he adds the famous passage, including another quotation that may come from Kaccāyana-dīpanī as well:

imāni suttāni mahā kaccāyanena katāni vutti ca sanghanandi sankhātena mahā kaccāyaneneva katā payogo brahmadattena kato ti. vuttañ cetaṃ:

kaccāyanakato yogo vutti ca sanghanandināpayogo brahmadattena nyāso vimalabuddhinā ti (Kacc-bh-nṭ 129, 15-30).

D’ALWIS translates:

These aphorisms were composed by Mahā Kaccāyana. The Vutti were [sic] composed by Mahā Kaccāyana himself, (who was also) called Saṅghanandi; – and the illustrations by Brahmadatta.

So it is expressly stated – that ‘The aphorisms were made by Kaccāyana, the vutti by Sanghanandi, the illustrations by Brahmadatta and the nyāsa by Vimalabuddhi’ (D’ALWIS, 1863: 105).

D’ALWIS adds, in a footnote, that the interpretation of the author of Kacc-bh-nṭ, identifying Mahā Kaccāyana and Sanghanandi, is probably a misunderstanding of the source, where clearly four layers of text are implied. D’ALWIS therefore believes that Kacc-bh-nṭ is wrong. Other scholars, such as MALALASEKERA, VIDYABHUSANA, DEOKAR and PIND,³ have also dismissed the interpretation of Kacc-bh-nṭ and prefer what they consider to be the original meaning of the quotation given in Kacc-bh-nṭ, that is, they do not identify Kaccāyana with Saṅghanandin, and they do not believe that the sutta and the vutti are the work of the same author.

A tradition similar to the one quoted by Ariyālaṅkāra is found in the Saddhammasaṅgaha (Saddhamma-s IX, 18-20, 35). This work is a “bibliography” written in the fourteenth century by the Thai monk Dhammakitti (or maybe by Ñāṇakitti, fifteenth century⁴). The passage reads as follows:

kaccāyano kato yogo saṃghanandi pavuttikā, ṭīkā vimalabodhī ca brahmaputto ca kārakā. (SADDHĀNANDA, 1890: 63)

Kaccāyana composed the yoga [“rule”], Saṅghanandin the pavuttikā [“notes”], and Vimalabodhī the ṭīkā [“commentary”], and Brahmaputto the kāraka [“section on factor of action”].

The author of Saddhamma-s had allegedly studied at the Mahāvihāra of Laṅkā (laṅkāramamahāvase⁵) and it is likely that the stanza ultimately comes from Laṅkā, not Burma, but this is difficult to ascertain.

Nandapañña’s Gandhavaṃsa (Gv), a Burmese catalogue written probably around the seventeenth century⁶ but source of many other bibliographies and one of the most consulted in modern Pāli, mentions Kaccāyana as the author of the “Kaccāyanagandho”. Gv does not mention Saṅghanandin, nor Brahmadatta (or Brahmaputta). It however mentions Vimalabuddhi as the author of Mmd.⁷

Ariyālaṅkāra probably had some reasons to believe that Kacc and Kacc-v are the work of the same author, and to consider the payoga as a different work, a latter addition. It seems as if he already expected three authors, and that is why he forces the meaning of the stanza in such a counterintuitive way. Indeed he is quoting the stanza in order to back up his judgement. He might be wrong in his interpretation, but he might still be right in seeing three, and not four, authors in the core text of the Kacc System.

In Western scholarship FRYER argued, as early as in 1882, in favour of Ariyālaṅkāra’s position (although he ascribes the stanza to Ariyālaṅkāra himself). VIDYABHUSANA also maintained that the text known as Kaccāyana was the work of a single person (see previous note). In the following section I will examine some features of Kacc and Kacc-v that can explain why some scholars tend to believe in a single author or editor.

3. Points of structure

It is necessary in the first place to examine the structure of the text as a whole. The structure of Kacc sutta is the same as Kātantra and many other grammars of the so-called Aindra class. Originally the work consists of four books: Sandhi, Nāma, Ākhyāta and Kibbidhāna⁸. Introductory verses are found in the beginning of the 1st, the 3rd and the 4th books. Those in the 3rd book, according to PIND (2013: 146), are later additions. Kacc structure is based on the complete version of Kātantra, including the last book on kṛt suffixes, allegedly authored by a certain Kātyāyana, not by Śarvavarman (SAINI 1999: 18). The title Sandhikappa, which should be the title of the first book only, is found in Kacc manuscript colophons at the end of every chapter (… sandhikappo niṭṭhito), the fact is that tradition consistently refers to this grammar as Kaccāyana (DEOKAR 2008: 10). Taking the title from the introductory stanzas, with slight variations, is a normal practice.

As early as in Vimalabuddhi’s Mukhamattadīpanī (probably around the 10th century CE) we find that Kacc is treated a compilation of four treatises. The original fourfold sectioning developed into an artificial eightfold division already recognised by Vimalabuddhi. Furthermore, Vimalabuddhi’s recension of Kacc (and Kacc-v) is not exactly the same as the recension of the text called simply Kaccāyana in Sri Lanka and Burma. But the textual transmission of Kacc and Kacc-v is the same. They are consistently called Kaccāyana in mss.

The commentarial pattern or method of Kacc-v is the same throughout the four books. We find the same set of (optional) fields:

A paraphrase (vutti)
Examples introduced by tamyathā or equivalent
Counter examples introduced by kasmā or equivalent
Purpose or scope of the sutta introduced by kv attho or equivalent
Comments on anuvṛtti, meaning of ca or vā, etc. for instance: Kacc-v ad Kacc 35: casaddaggahaṇena iheva makārassa pakāro hoti.

Regarding the text of PIND’s Kacc-v specifically, it is considered that:

It is made of different layers in Kacc/-v, each layer is the work of one single author.
It is made of different sections in Kacc/-v, each section is the work of one author.

According to PIND, it is questionable that Kacc-v is the work of a single author and he gives the following explanation:

[T]he sandhi chapter differs from the other chapters in the way it paraphrases each sutta. In contrast to the other chapters which use the verb āpajjate to indicate that a grammatical operation obtains, the sandhi chapter invariably uses pappoti, e.g. lopaṃ pappoti or papponti at Kacc-v 12 through 17, contrasting with lopaṃ āpajjate or āpajjante, e.g. at Kacc 220. They also differ with respect to the formulation of the contrastive sections of the exegesis of Kacc: in the sandhi chapter this section is invariably introduced by kasmā, elsewhere by kimatthaṃ. This difference in style seems to indicate that the commentary on the sandhi chapter has a different author from the rest of the work (PIND 2012: 90).

The difference in terminology seems obvious. But the difference of style in Kacc-v I and II-IV is merely lexical. The meaning of the technical terms is the same. The method of grammatical analysis is also the same. In fact, the only differences PIND accounts for are:

Kacc-v, book I	Kacc-v books II-IV
pappoti	āpajjate
kasmā	kimatthaṃ

In this context, pappoti and āpajjate are synonyms. Furthermore, there are other formulae to express the same idea in Kacc-v. For instance, even within the Sandhikappa, the obtaining of a substitute can be expressed in different ways:

Kacc-v ad. Kacc 32: ekāre hakāre ca pare niggahītaṃ kho ñakāraṃ pappoti vā

Kacc-v ad. Kacc 34: niggahītassa kho sare pare makāradakārādesā honti.

Sandhikappa probably translates Kā’s āpadyate either with acc. + pappoti or with (x)-ādesa hoti or simply (x) hoti. Compare:

Kā, I.4.60 mo ’nusvāraṃ vyañjane

Kā-v makāro punar anto vyañjane pare anusvāram āpadyate.

Kacc, 30 aṃ byañjane niggahītaṃ

Kacc-v niggahītaṃ kho byañjane pare aṃ iti hoti.

This does not necessarily imply that the commentary is written by two different authors. Variation might be a feature of the author’s style, a mixed style that reminds us of the Sanskrit commentator Kātyāyana or the Kātantra commentators. The same applies to kasmā and kimattham which seem to be synonyms.

Finally, Kacc-v uses the expression kv attho “where is it used?” in order to exemplify the scope of a rule. This expression (kv attho) is not different in books I, II and III, but interestingly, it changes to kimpayojanaṃ in book IV.

Compare the examples:

	I	II	III	IV
	Sandhi	Nāma	Ākhyāta	Kibbidhāna
A) Introduces examples of scope of a sutta	kv attho	kv attho	kv attho	kimpayojanaṃ
B) Introduces examples from the canon	kasmā	kimatthaṃ	kimatthaṃ	kimatthaṃ

From this we could conclude that Kacc-v is an aggregate of layers, every layer being the work of different authors according to the section. But despite some variations in terminology, I think Kacc-v proves fairly consistent as a commentary. In fact, VIDYABHUSANA, who was well versed in logic and grammar, had the impression that “the entire work was written by Kaccāyana himself”. We should also take into account that variations and variant readings occur also in Kā and Kā-v, and in other commentaries of long tradition. That may be the result of a textual tradition that transmitted the text in separate books, not as a whole, as it is the case with Kacc/-v. Orality, in the handing down of these treatises, can also be an important factor of lexical variation.

4. Author of what?

The question, however, still remains: author of what? PIND assumes a clear cut difference between sutta and the rest – which is conventionally called Kacc-v. But, as I have shown, this is not what an internal analysis of Kacc-v reveals and this is not the opinion of other grammarians. Ariyālaṅkāra, for instance, sees a unity in the sutta and the vutti “gloss”, whereas he considers the rest as a payoga – and obviously he knows that Mmd is a different work. I think that some features point to a singularity of author of the sutta and the vutti, if not the sutta together with vutti and payoga. I will subsequently offer some examples of the way sutta and vutti are intimately connected and could be considered the work of the same author, an author that was creating a Pāli version of Kātantra and Kātantravṛtti (or perhaps a version of some Buddhist Kātantra such as the Kaumāravyākaraṇa) (see LÜDERs, 1930). I will also show some problems with this view.

Alleged misunderstandings of Kacc-v:

[A] Misunderstanding of vā

Example 1 – vā in Kacc 273

The beginning of the kāraka section in Kacc starts with the definition of apādāna:

Kacc 273 yasmād apeti bhayaṃ ādatte vā tad apādānaṃ

Kacc-v comments upon the sutta:

yasmā vā apeti yasmā vā bhayaṃ jāyate yasmā vā ādatte taṃ kārakaṃ apādānasaññaṃ hoti.

Etc. SENART, ignoring the vutti, translates:

On apelle apādāna (ablatio) [la relation syntactique où se trouve] l’objet dont on s’éloigne ou dont on s’effraye (SENART 1871: 125).

SENART skips the translation of ādatte, and remarks:

Malgré le scholiaste, suivi par M. Kuhn, je ne crois pas possible de dédoubler l’expression bhayaṃ ādatte; il faudrait dans ce cas un premier vā après bhayaṃ (SENART 1871: 125).

SENART also points out that this extra vā in Kacc-v is not to be found in Pāṇini’s or Kātantra’s sūtrapāṭha. He is right: in Kā II.214 there is only one vā. But this missing vā is implicit in Durgasiṃha’s Kā-v. It is evident that Kacc and Kacc-v follow the same text:

Kā II.214 yato ’paiti bhayam ādatte vā tad apādānaṃ

Kā-v ad Kā II.214 yasmād apaiti, yasmād bhayaṃ bhavati yasmād ādatte vā tat kārakam apādānasaṃjñaṃ bhavati.

Kacc-v follows the interpretation of Kā-v. This example shows the three different contexts where apādāna is applied, and they are the ones we find in Kacc-v. This instance seems to indicate that Kacc sutta and vutti are composed together following the model Kā and Kā-v.

Example 2 – vā in Kacc 281

A similar example is found again in the kāraka section (although there are instances of this alleged “over interpretation” all over the work).

Kacc 281 yena vā kayirate taṃ karaṇaṃ

That with which something is, for instance (vā), done, is the instrument (karaṇaṃ).

My translation follows the interpretation of Kacc-v, where this vā has usually the meaning of “for instance”:

Kacc-v ad Kacc 281 yena vā kayirate yena vā passati yena vā suṇāti taṃ kārakaṃ karaṇasaññaṃ hoti.

SENART (1871: 142) says: “Il est difficile de croire que le vā du sūtra ait réellement le sens que semble lui attribuer le scholiaste” (“It is difficult to believe that the vā of the sūtra actually has the meaning that the commentator seems to ascribe to it”) and later on, in the same comment: “Le grammarien pāli aurait cherché à rendre par la particule vā l’intention contenue dans le superlatif du grammarien sanskrit” (“The Pāli grammarian would have tried to express, by means of the particle vā, the intention that is implied in the superlative of the Sanskrit grammarian”). This vā is not found in the Kā or Kā-v. But we know that Kacc and Kacc-v, in the kāraka section, are a faithful copy of Kā and Kā-v. Why should Kacc add a particle devoid of information? A closer look reveals that the particle vā in Kacc 281 actually reflects Kā-v ad Kā II.218: dātreṇa dhānyaṃ lunāti. manasā meruṃ gacchati. tatha paśunā rudraṃ yajate, etc. The author of Kacc knows that the verb kayirate (S. krīyate) means actually any action, as the examples in Kā-v indicate. That is why he adds vā. Kacc sutta seems to be based not on Kā but on Kā through its vṛtti.

[B] Misunderstanding of ca

Example 3 – ca in Kacc 20

Kacc 20 do dhassa ca

Kacc-v dha iccetassa sare pare kvaci dakārādeso hoti. ekaṃ idāhaṃ bhikkhave samayaṃ.

“And dh becomes d”

“dh, when followed by vowel, becomes d sometimes [ca triggers vā= kvaci from Kacc 13]”

The commentary goes on with a list of other instances of consonantic change (vikāra). This is, according to Senart, an “overuse” of the particle ca: “cet abus, que nous rencontrerons fréquemment par la suite, d’un mot ou d’une particule qu’il étend et dénature au point de faire dire à l’auteur une foule de choses, souvent fautives, qui n’étaient nullement dans sa pensée” (SENART 1871: 18). But if we think of the vutti as the gloss only, we realize there is no “abus”. The gloss sticks to the meaning of the sutta. The flaw is in the payoga layer, where the list of extensions (“foule de choses … fautives” i.e. “many wrong things”) is found. If this is a mistake, then, it should not be ascribed to the vuttikāra.

Example 4 – ca in Kacc 366

Kacc 366 tad assatthī ti vī ca

Kacc-v tad ass atthī ti icc etasmiṃ atthe vīpaccayo hoti.

medhā assa atthi tasmiṃ vā vijjatī ti medhavī. evaṃ māyāvī.

casaddaggahaṇena sopaccayo hoti. sumedhā assa atthi tasmiṃ vā vijjatī ti sumedhaso.

Again, we can observe that the “abus” of the particle ca is not found in the vutti “gloss” proper, but in a sort of vārttika comment at the end, which should be considered part of the payoga. This example shows clearly two different layers in what is conventionally called Kacc-v. In one of them we find a rigorous gloss of the sutta, with proper examples. In the second layer we find a correction or extension of the sutta. This correction (called “abus” by SENART) is based on the extension of the rule triggered by the particle ca. Vimalabuddhi (Mmd 317, 8) explains the examples derived from ca separately: caggahaṇaphalaṃ sayam eva vakkhati “the results of the use of ca will be explained independently”. I think it is clear that the first layer of the vutti does not trespass the rule.

[C] Problematic passages

Example 5 – tudampatī in Kacc 341

Kacc 341 jāyāya tu daṃ jāni patimhi

Kacc-v jāyāya icc etāyaṃ tu daṃ jāni ete ādesā honti patimhi pare. tudampati. jānipati.

This passage has been presented as the proof that Kacc-v misunderstands the text of Kacc and therefore they cannot be the work of the same author or editor. Indeed, the vutti reads tu with daṃ, and the result is the artificial word tudampatī. This word is accepted in CHILDERS Dictionary and PED sv. tudampatī as a dual, tu > du (!), and also in other grammars such as DUROISELLE and MULLER. Moggallāna (Mogg-p 187: 26-28) says: yam pana kaccāyanavuttiādisu tudampatī ti udāharan ti nāyam payogo āgame niruttiyañ ca tādisassa payogassāsambhavato, “the example tudampatī that we find in the vutti and other commentaries is not found in the scriptures, because there is no such a usage in the language (nirutti)”. We do find, however, variant readings without tu:

Cᵉ(1) jāyāya pati dampati

Cᵉ(2) jāyā ca pati ca dampati⁹

These Sinhalese variants may be corrections based on Mogg criticisms.

Alternatively, the word tu could be an interpolation fruit of the contamination of the lexicon Abhidhānappadīpikā 242 padas ab:

jāyāpatī janipatī jayampatī tu dampatī (or tudampatī ?)

The Abhidhānappadīpikā-ṭīkā refers to Kacc 341 as an authority:

catukkaṃ patipatinīnaṃ yuge ’‘dārā pume bahutte ca, daṃ kalatre napuṃsake’’ti [Cintāmaṇiṭīkā 16.38] amaramālā, jaṃsaddo tvabyayo dāravacano. tasmā ’‘jampati, dampatī’’ti pi bhavitabbaṃ, idha pana kaccāyanamatenodāhaṭā. jāyā ca pati ca jāyāpati. itarītarayogadvando. jāyā ca pati ca jānipati, tathā jāyampatiādayo, jāyāsaddassa patimhi pare jāni, tudañ ca, jāyañ ca yadādinā.

The particle tu could be an old interpolation into both sutta and vutti. If we remove it, the text makes perfect sense:

*Kacc 341 jāyāya daṃ jāni patimhi

*Kacc-v jāyāya icc etāyaṃ daṃ jāni ete ādesā honti patimhi pare. dampati. jānipati.

It is nevertheless striking that the tradition has accepted the “wrong” reading without doubts. Even Vimalabuddhi reads the example tudampatī. This may point to some meaningful solution that I cannot see. In any case, this faulty example could have survived until today precisely because Vimalabuddhi did not purge it in his edition. Rūpasiddhi and its ṭīkā also accept it. The case is, no doubt, problematic, but in my opinion this example is not enough to prove that Kacc and Kacc-v are the work of a different authors.

Example 6 – mantā in Kacc 7

The following is an instance of Kacc using material not from Kā but from Kā-v.

Kacc 7 vaggā pañcapañcaso mantā,

The word mantā is not found in the Kā equivalent Kā I.1.10 te vargāḥ pañca pañca pañca, but it is actually found in Kā-v ad loc. te kādayo māvasānā varṇāḥ pañca pañca pañca bhūtvā pañcaiva te vargasaṃjñā bhavanti, followed by Kacc-v ad Kacc 7: tesaṃ kho byañjanānaṃ kakārādayo makārantā pañcapañcaso akkharavanto vaggā nāma honti. Notice how Kacc-v translates Kā-v using synonyms: māvasanā = makārantā; vargasaṃjñā = vaggā nāma; varṇāḥ = akkhara-. The word mantā in the sutta and vutti of Kacc seems to derive from Durgasiṃha’s vṛtti. Again, it seems that the author of Kacc and Kacc-v was working directly not only from Kā but Kā-v.

5. Conclusions

Even though it is at present impossible to determine whether the author of Kacc sutta and the vutti is the same, I have shown in this article that there are some reasons to follow the Burmese savant Ariyālaṅkāra and postulate a single authorship for these two works. For two things are clear: first, the text called Kacc-v in PIND’s edition is made, at least, of two layers (identified by the tradition), and second, not only the vutti, but also the sutta text follows the text of Kātantra-vṛtti. It makes sense, therefore, to think that if the author of Kacc sutta had the Kātantra-vṛtti in front, he could have adapted this text as well. This would explain the great consistence of sutta and vutti, a consistence that is lost when we move to the payoga, probably written, as the tradition maintains, by some other grammarian, conventionally known as Brahmadatta.

REFERENCES

Primary Sources


Kacc	Kaccāyana = PIND, 2013.
Kacc-v	Kaccāyanavutti = PIND, 2013.
Kā	Kātantra = DWIVEDI, 1997–2005.
Kā-v	Kātantra-vṛtti = DWIVEDI, 1997–2005.
Gv	Gandhavaṃsa = KUMAR, 1992.
Mmd	Mukhamattadīpanī= Nyāsapāṭh, Yangon, Sudhammavatī cā pum nhip tuik Press, 1933.
Saddhamma-s	Saddhammasaṅgaha = SADDHĀNANDA, 1890.

Secondary Sources

D’ALWIS, J.

(1863). An introduction to Kachchāyana’s grammar of the Pāli language, with an introduction, appendix, notes, etc. London

BURNELL, A. C.

(1875). On the Aindra School of Sanskrit grammarians. Mangalore.

DEOKAR, M. A.

(2008). Technical terms and technique of the Pali and the Sanskrit grammars. Varanasi: Central Institute of Higher Tibetan Studies.

DWIVEDI, J.

(1997—2005). Kātantravyākaraṇa of Ācārya Śarvavarmā, with four commentaries: Vṛtti and Ṭīkā by Śri Durgha Singh; Kātantravṛttipañjikā by Śri Trilocanadāsa; Kalāpacandra by Kavirāja Suṣeṇaśarmā; Samīkṣā by the editor. Sarnath, Varanasi.

KUMAR, B.

(1992). The Gandhavaṃsa: A History of Pali Literature. Delhi: Eastern Book Linkers.

MALALASEKERA, G. P.

(1960). Dictionary of Pāli Proper Names. London: Luzac and Co.

NYANATUSITA.

(2008). Reference Table of Pali Literature, electronic edition by the author: http://gretil.sub.unigoettingen.de/gretil_elib/

PIND, O. H.

(2013). Kaccāyana and Kaccāyanavutti. Bristol: PTS.
(2012). Pāli Grammar and Grammarians from Buddhaghosa to Vajirabuddhi: A Survey. XXXI, 57-124. Journal of Pali Text Society.

SADDHĀNANDA, N.

(1890). Saddhamma Samgaho. 21–90. Journal of Pali Text Society, IV.

SAINI, R. S.

(1999). Post-Pāṇinian systems of Sanskrit grammar. Delhi: Parimal Publications.

SENART, E.

(1871). Kaccàyana et la littérature grammaticale du Pâli. Paris: Journal Asiatique.

VIDYABHUSANA, S. C.

(1901). Kaccayana’s Pali Grammar: Edited in Devanagari Character and Translated into English. Calcutta: Mahabodhi Society.

Ph.D. University of Cambridge. E-mail: ar607@cam.ac.uk ↩
I follow the PTS critical edition (= PIND 2013). ↩
MALALASEKERA (1958: 180) says: “In the Kaccāyanabheda, written by Mahāyasa Thera of Thatōn about the thirteenth century, there occurs a memorial verse: ‘The aphorisms were made by Kaccāyana, the Vutti by Saṅghānandi. The illustrations by Brahmadatta. And the Nyāsa by Vimalabuddhi.’” To the best of my knowledge, that verse is not found in Kacc-bh. VIDYABHUSANA (1901: xxvi) says: “‘The Yoga (Sūtra) was written by Kaccāyana, the commentary by Saṅghanandī, the examples were added by Brahmadatta and the gloss by Vimalabuddhi.’ From the manner in which the sutta, vutti, payoga and nyāsa are intimately connected with one another, I am inclined to believe that the entire work was written by Kātyāyana himself. At any rate the sutta, etc. were written simultaneously.” Even though VIDYABHUSANA is clearly not aware that the Nyāsa is a different text, it is important to keep in mind that this great scholar had the intuition of a single authorship. DEOKAR (2008: 7) and PIND (2012: 71) also distinguish four authors in that quotation. ↩
NYANATUSSITA, 4.3.1. ↩
SADDHĀNANDA, 1890: 90. ↩
NYANATUSSITA, 4.3.3. ↩
Gv, 60: vimalabuddhi nāmācariyo mukhamattadīpanī nāma nyāsapakaraṇaṃ akāsi. ↩
For the so-called Aindra System see BURNELL, 1885 passim. ↩
PIND, 2013: 119 fn 19. ↩