Update: 2016-01-27 02:34 PM -0500


Buddhist Hybrid Sanskrit Grammar and Dictionary


by Franklin Edgerton (1885–1963), Sterling Professor of Sanskrit and Comparative Philology, Yale University, Motilal Banarsidass Publishers Private Ltd., Delhi, 1st ed. New Haven, 1953. ISBN: 81 208-0998-x (Vol. 1), ISBN: 81 208-0997 (Set of 2 books).

Scanned from the original book by Dr. Zin Tun, up to p009, and digitized by Daw Khin Wutyi. From p010 onwards, Daw Khin Wutyi has typed from the original book which I (U Kyaw Tun) have brought with me on my trip from DeepRiver to Yangon where Daw Khin Wutyi is based. This TIL edition is edited, with additions from other sources, by U Kyaw Tun (UKT) (M.S., I.P.S.T., USA) and staff of Tun Institute of Learning (TIL) . Not for sale. No copyright. Free for everyone. Prepared for students and staff of TIL Research Station, Yangon, MYANMAR :  http://www.tuninst.net , www.romabama.blogspot.com

index.htm | Top

Contents of this page 

Changes in the course of tradition


Edgerton notes
Edgerton's footnotes are continuously numbered without any reference to pages. However for this TIL edition, I have given the page numbers as well. Thus fn001-02 means, it is footnote #02 found on p001, and fn003-07 is footnote #07 found on p003.
• There is no footnote on p.007


UKT notes 
Néwari speech - the cousin of Bur-Myan
Tell-Tale-Trio: თ Tan, ი In, and ა An (Georgian alphabet)


Contents of this page

Changes in the course of tradition


1.39. There is evidence to show that, morphologically as well as orthographically speaking, BHS texts have undergone increasing adaptation to Sanskrit norms in the course of handing down by tradition.

1.40. This is particularly clear when, as is true in a number of cases, we have different forms of the same text. (p006c1end-p006c2begin) Most complete texts of extensive works are known to us only from Nepalese and Japanese mss. (UKT ¶)

UKT 160125:  Whenever we meet the word "Nepalese", we have to be careful of the two languages which are almost the opposite of each other. Nepal-Bhasa नेपाल भाषा {né-pa-la. Ba-þa}  aka Néwari (spoken by the blood relatives of Gautama Buddha) is Tib-Bur related to Old Magadhi, whereas Nepalese नेपाली {né-pa-li}, (spoken by the conquerors of the kingdom) is IE akin to Skt-Dev. See my note on Néwari speech - the cousin of Bur-Myan.

But some are preserved in other regions; and particularly in Chinese Turkestan, old fragments have come to light of parts of some of the same works known more completely in Nepalese or other later recensions. (UKT ¶)

UKT 160126: My interest in Chinese Turkestan is to relate the Bur-Myan akshara to the Georgian alphabet - through the Silk-Roads. See in my note The Tell-tale Trio: თ Tan, ი In, and ა An.

It was pointed out notably by Lüders (see fn. 16) that the fragments of the ‘Kashgar’ (Chinese Turkestan) recension of SP show many differences from the Nepalese recension, a fact sufficiently clear even from the critical notes to the very unsatisfactory Kern-Nanjio edition. Lüders emphasizes that this is as true of the prose as of the verses, and that the Kashgar form of the prose often contains Prakritisms, while the Nepalese has Sanskritized the forms (e.g. bhāṣiṃsu: abhāṣanta); tho at times the reverse relation exists, so that we must assume some Sanskritization in both recensions, and an original more Middle Indic than either. Lüders is undoubtedly right in taking it for granted that Middle Indic or non-Sanskrit words and forms, in whatever recension they (p006c2end-p007c1begin ) occur, are in invariably older than corresponding Sanskritizations in other versions.

1.41. Similar results are shown by a comparison of the Central Asiatic fragments of Vaj published by Pargiter (Hoernle, MR 176 ff.), compared with Müller’s ed. Of Vaj (from Japanese sources). Examples, all prose (first Pargiter, then Müller’s ed.):
  parinditāḥ … parindanayā 179.3,
  parīn (even in this form the word is not Sanskrit!) 20.4, 5;
  pratiṣṭhihitvā 180.12,
  pratiṣṭhitena 21.9;
  pratiṣṭhihe (3 sg.opt.) 180.14,
  pratitiṣṭhet 21.11;
  viyūbha 180.16, 188.20,
  viyūha 188.19, for vyūha 27.4, 38.6, 7;
  u(d)graheṣyanti 186.8, 22, and 187.9,
  udgrahīṣyanti 33, 17, 34.14, 35.5;
  babhūva with 3 pl. subject 187.6,
  abhūvan 35.1;
  carimikāyāṃ paścimikāyā(ṃ) 187.8,
  paścimāyāṃ (omitting cari) 35.4.
In all these the Central Asiatic version is more Middle Indic, Müller’s text Sanskritized and so presumably secondary. But the contrary is the case with ārādhitā and virādhitā 187.7, instead of ārādhitā, virāgitā 35.3.

1.42. It has sometimes been suggested that the verses retain in most texts a more Middle Indic appearance than the prose because the meter made it harder to Sanskritize them. (UKT ¶)

UKT 160126: Speaking of the meter (in Vedic and in Sanskrit), we should note that one particular Mantra in Rig Veda, the Gayatri Mantra, has had a meter different from many of the mantras in Rig Veda. The word Om had been added to make it conform to other verses. Please note, I have no knowledge of Meter in Poetry and I am writing this note from what I have read on Gayatri Mantra. While reading through the online info from reliable sources such as  Wikipedia, please note that Skt-Dev speaking Brahmin-Poannars {braah~ma.Na. poaN~Na:}, are sold on the idea of Male-Female duality, and that they like to see every thing in terms of sex. And also that they have to make everything take on human forms of males (gods) and females (goddesses). The word Savitr can be interpreted as Universal Power and Knowledge Source. It doesn't have to have a female form of a goddess. After giving it a female form, they to invent a male god to mate with the goddess. See Wikipedia:
- https://en.wikipedia.org/wiki/Gayatri_Mantra 160126

I should not venture to deny that this consideration may have had some weight. Perhaps the greater aura of distinction, secular or religious, which has usually attached to verses in India, may also have been concerned. (The verses alone of the Pali Jātaka are canonical; the prose is mere ‘commentary’.) I would, however, point out that we have definite proof that the BHS verses, too, were not exempt from changes in tradition. Chakravarti’s edition of Ud cites the text of one ancient and fragmentary ms. (provenience not stated) and of several later ones, mostly, it seems, from Chinese Turkestan. I shall record here a selection of the variants. They are interesting as showing that meter was not really a serious obstacle to change. By patching the meter (often with insertion of a ‘patchword’, Flickwort), by rearranging the order of words, or otherwise, a new form, Sanskrit or nearer to it, may replace a more Middle Indic one. If anything really important had depended on it, later redactors could probably have made the verses of SP, for example, look as Sanskritic as the accompanying prose. Indeed, this is just what happened in the BHS works of my third class (see Bibliography). In the forms in which they are preserved to us, their verses (so far as they have any; Divy, for example, has a good many) seem as Sanskritized as the prose. Furthermore, some Middle Indic forms could have been Sanskritized very easily without affecting the meter. Thus the Skt. 3 sg. Optative ending –et is metrically equivalent of Middle Indic –e. Yet the latter is extensively retained in the verses of most texts of the first and second classes, while the prose almost always has –el, if we may trust the mss. and editions (except in Mv; even in Mv –eis must commoner in verses than in prose). – When the same verse is known to exist in Pali, it is sometimes interesting to compare the Pali equivalents of the words cited; this will be done occasionally, tho by no means invariably, in the following list. I cite first the readings of the oldest ms., then of the later ones, which are invariably secondary.

1.43. Earlier and later forms in ms. Readings of verses of Ud.

  vii.12 bhāṣeya, vihinseya; bhāṣeta, vihiṃseta. Pali bhāseyya, vihiṃseyya.
  ix.3 bhāyasi: bibhesi.

  x.5 śraddhāya (instr.): śraddhayā.
x.6 cchindati: chinatti.

  xi.2 parākkrame: parākramet.
xi.3 saṃkiliṣṭā(read ⁰ṭaṃ?) va (m.c. for )yas (read yat) tapaḥ: saṃkliṣṭaṃ vāpi (note patchword!) yat tapaḥ.
xi.7 careya; carate; saṃkalpānaṃ(gen. pl.): ⁰nāṃ.
xi.10 sālaṃ vā māluvotata (=māluvā-otataṃ, for avata-taṃ): sālaṃ vā māluvā yathā. Pali māluvā sālam ivotataṃ.
xi.11 sthero: sthaviro (which here is unmetrical, even in the later ms. Which writes it; its writer doubtless read it sthero (if not thero), the ‘Skt.’ form being purely orthographic).

  xii.3 yāya (instr.): yayā.
xii.4 mārgānāṣṭāṅgikaḥ(i. e. mārgāṇa, gen. pl., aṣṭ⁰)śreṣṭhaḥ satyānaṃ capture padāḥ; mārgeṣv aṣṭāṅgikaḥśreṣṭhaḥ catvāry āryāṇi satyataḥ. Pali maggān’ aṭṭhaṅgiko seṭṭho saccānaṃ caturo padā, supporting the prior reading, all the forms of which are paralleled in BHS. Chakravarti’s emendations are all false.
  xii. 5 yadā prajñāya (instr.) paśyati: prajñayāpaśyate yadā(order changed to accommodate Sktized form to the meter).

  xv.7 ye suptāḥ pratibuddhatha (impv.):suptāśca pratibudhyataḥ(read ⁰ta; stem and ending both Sktized).

  xvi.4 –darśāvī(n. pl.): -darśino.
xvi.5 sa imāṃ(acc. pl.) bhāsate loke (acc. pl.): sa imaṃ(one fragment imāṃ!) bhāsate lokaṃ. Pali so(i)maṃ lokaṃ pabhāseti (more nearly agreeing with the later version of Ud; the older version of Ud is probably older than the Pali).
  xvi.5, 7, 9 abhramukto va(=iva): abhramuktaiva (v.l. ⁰kta iva, unmetr.). Pali abbhā mutto va.
xvi.14 iccheya: icched.

  xviii.3 chindatha, bhavatha (impvs.): chindata, bhavata.
xviii.5 paduma, vṛṃhaye: padmaṃ, bṛṃhayet.
xviii.11 (vyatiro) cati prajñāya (instr.): prajñayā vyatiro-ca(n)te.
xviii.13 vipramuñcatha (impv.): ⁰ta.
xviii.18 prapuṣpa(kāṇi): tu puṣpakāni (so). Pali papupphakāni.

  xix.5 hirīniṣevī: hirīniṣevī hi (patchword). Pali hirīnisedho.
xix.12 bhadraṃ va: bhadram iva (unmetr.).

  xx. 1 viprajaheya: viprajahec ca (patchword). Pali vippa-jaheyya.
  xx.2 prajahe: ⁰hed.
xx. 5 ahirī: ahrīko.

1.44. At least equally interesting are parallel passages in different works, sometimes paralleled also in the Pali cannon, and in that case presumably going back to extremely early Buddhist tradition. In the statement of the first of the Four Noble Truths, presumed to contain the Buddha’s own words, My iii.332.4 has the adverbial instr. saṃkṣiptena, ‘in brief, in a word’, not recorded in Skt., but = Pali (Vin. i.10.29 = SN v.421.23) saṃkhittena, while the LV version, 417.7, substitutes the regular Skt, saṃkṣepāt. (UKT ¶)

UKT 160126: We can relate «saṃkṣiptena» to a combination of {þän-hkaip} 'summary' (UTM-PDMD366), {té-na.} 'therefore' (UHS PMD0450).

In another prose passage Mv ii.283.14 reads purime yāme‘ in the first night-watch’, but the LV correspondent (344.7) reads prathame yāme. The word purima is not Skt., and is used by LV only in verses, never in prose, according to our mss. Yet in such a passage as this, one cannot help wondering if the obviously secondary prathame introduced by some relatively late redactor or copyist; in short, whether the original LV text did not read purime

1.45. Even in a verse, LV 387.18 substitutes pūrvikāṃ for Mv iii.305.19 purimāṃ, here meaning 'eastern', a rather rare use of purima in BHS, and nowhere found in LV, I believe. In verse, Mv iii. 306.6 purastime ‘eastern’ = LV  388.8 pūrvasmin vai (note the patchword, m.c.); Mv iii.309.8 Pthivī Padumāvatī = LV 391.3 Pthivī Padumāvatī tathā (two epenthetic vowels removed, meter again mended by a patchword).

1.46. In general, LV (at least as preserved to us) is secondary and Sanskritizing compared to Mv, in passages (p007c2end-p008c1begin ) which the two texts have in common. Yet it would be wrong to assume that this is universally true. Each case is a separate problem. Neither text borrowed directly from the other; both incorporated materials that were older than either, and sometimes LV seems to have preserved features which more closely resemble the common original than any parallel version, including Pali versions. In the prose account of the Buddha's first sermon at Benares occurs a passage in which he tells his audience, the five monks who had formerly been his followers, how he first realized that he had reached enlightenment: LV 418.13 ff., Mv iii. 333.11 ff., and Pali Vin. i.11.18 ff. = SN v.422.31 ff. The two Pali versions are identical, verbatim; the two BHS versions, closely resembling the Pali in essence, diverge from it and from each other in many details, sometimes one, sometimes the other being closer to the Pali. In the first sentence, LV eṣu catuṣv (so all mss.; note the M Indic form, for Skt, caturṣu, in LV prose!) āryasatyeṣu = Pali imesu catusu ariyasaccesu, but Mv (changing construction) imāni catvāry āryasatyāni. There are many interesting things which a detailed study of the three versions reveals. I shall mention only the verb-form pratijñāsiṣaṃ, LV 418.15 and 19, an unaugmented aorist. In 15 all mss. lack the augment; in 19 only a few inferior mss. have it. This is a non-Skt. feature, and therefore likely to go back to the oldest times. Yet Pali has an augmented form, paccaññāsiṃ, both times. Mv substitutes a present, pratijāne 'haṃ the first time, and prajānāmi (perhaps read paratijānāmi?) the second time; in lines 12 and 15 it has abhyajñāsiṣaṃ (augmented) in a clause which is differently expressed in LV and the Pali.

1.47. I know of no Pali correspondent to the Avalokita (Mv), or (in Śikṣ avalokana (or  ⁰nā), Sūtra which is reproduced at length in Mv ii.293.16-397.7 (following another similar text with the same title). But many of its stanzas are cited in Śikṣ 297.10-308.12, and a few others 89.15-90.3. The Śikṣ extract is only a small selection, yet includes some verses not found in Mv; for this and other reasons it is certain that Śikṣ was not citing from Mv; rather, both cite from a common ultimate original. Most of the verses common to the two texts are composed in a sama-vṛtta meter known to me otherwise only from LV 229.21-234.10; each of the four pādas has the scheme.

_ _ _ _ _ _ _ _ _ _ _ _ x

Both Śikṣ and Mv especially the latter, are very corrupt in this passage. A combination of the two, with the help of the meter, sometimes makes possible a restoration superior to both. (Omitting two of the four shorts which follow the fifth syllable produces a normal BHS triṣṭubh pāda; and this is often the result of textual corruptions, especially in Mv.)

1.48. In these verses, apart from corruptions, the Mv readings are in general more un-Sanskritic, and hence closer to the original, than those of Śikṣ. Many scores of examples would show this; one is phalikha-bhujo, to be read with mss. (except that they have  ⁰khaṃ-) Mv ii.379.5, 'with arms like bars' (see Dict. s.v. phalikha), for which Śikṣ 303.3 has the normal Skt. parigha-bhujo. Nevertheless, Śikṣ is sometimes non-Skt. and clearly original, or closer to it, and Mv Sanskritized and secondary. Examples are:
Śikṣ 304.10,
 yatha Mv ii.383.5 (in meter, also, secondary); kavacita lak
ṣsṇebhiḥ Śikṣ 304.16, varalak-ṣitāṅgo Mv 386.10 (in meter, also, secondary);
Śikṣ 307.3, (a) pithitā Mv 393.6;
 sthita bhonti
Śikṣ 307.15,
 nivasanti Mv 394.19 (vut here both are metrically second-(p008c1end-p008c2begin)ary lacking two short syllables); saṃpramuñcī  (aor., or possibly opt.)
Śikṣ 308.12, so pramuñced Mv 395.23.

1.49. We might go much farther in citing such examples, but I trust that these are enough. Let me summarize the facts about BHS as stated so far.

UKT 160125: In the following paragraphs, Prof. Edgerton summarizes the seven facts about BHS.

1.50. (1) Buddhist Hybrid Sanskrit tradition, as a whole, starts from, or goes back to, an early Buddhist canon, or quasi-canon, which was composed not in Sanskrit but in a Middle Indic vernacular which very probably already contained dialect mixture.

1.51. (2) Some parts of this old canon, or passages from it, are preserved in BHS, sometimes in more than one form. When this is the case, any non-Sanskrit features of form and vocabulary, wherever recorded, are always closer to the original on which they are based than corresponding standard Sanskrit features, wherever recorded.

1.52. (3) The verses of BHS texts of my classes 1 and 2, as presented in our mss., are on the whole semi-Middle-Indic or hybridized. This means that they represent the BHS tradition in its purest form. In texts of class 2, the accompanying prose parts of these texts are nearly (not quite) Sanskritized in phonology and morphology, according to the mss. In vocabulary, the prose is just as Middle Indic as the verse.

1.53. (4) In all BHS works, as presented in our mss. and editions, there are very many words and forms which are standard Sanskrit. These include many forms which cannot possibly have existed, at any time, in any Middle Indic dialect. They represent alterations in the tradition, later in time than the original Middle Indic canon, at least (not necessarily later than the composition of the specific BHS works in which they are recorded). As time went on, the tendency was in general towards ever increasing Sanskritization. Yet the BHS tradition continued to live, apparently for centuries, as a religious language among the Buddhists, or at least some Buddhists, of North India. The hallmark which distinguishes it is the vocabulary, which contains not only technical religious terms, but quantities of purely secular words, which never occur in standard Sanskrit. Very rarely can any serious doubt arise as to whether a particular work should be classed as BHS. Even if its grammar is virtually Sanskrit, or entirely so (tho such a case hardly occurs), its vocabulary will decide. fn008-17

1.54. (5) There is clear evidence that some of these Sanskrit words and forms were substituted for older, non-Sanskrit ones, by later copyists or redactors of the individual work containing them: in other words, that some such works were originally more Middle Indic than is indicated by some, or even all, of the mss. in which they are preserved to us. In SP for example, one recension Sanskritizes some words, another recension others; the original text of SP must have been less Sanskritc than either.

UKT 160125: The word recension means "a revised edition of a text; an act of making a revised edition of a text."
It is just an opinion of the later author who may not even know the deeper meaning of the word. Remember, the various Sanskrit speaking authors were translating Magadhi (the speech of the Buddha - the Tib-Bur language) into Sanskrit for his audience the IE Sanskrit speakers

1.55. (6) There is, further, evidence that in citing or incorporating older materials, any BHS text (if we accept the evidence of its mss.) may be expected to have introduced some Sanskritizations of originally Middle Indic features.

1.56. (7) It is, however, certain that some Sanskrit appearing features are orthographic only; the words were pronounced as in Middle Indic. This is proved by the metrical structure of the verses of BHS texts of classes 1 and 2. How old this misleading Sanskritic spelling is, we have no way of telling; it appears very commonly, tho (p008c2end-p009c1begin ) not invariably, in the mss. of all the specified texts. That the same was true of the same or similar features in the accompanying prose, at least in earlier times, seems a reasonable guess; naturally, there can be no direct proof that the prose was pronounced otherwise than as written.

(p009c1cont ) starting with
1. INTRODUCTION : Plan and methods of this work, the first para 1.57

Contents of this page

Edgerton Footnote

Edgerton's footnotes are continuously numbered without any reference to pages. However for this TIL edition, I have given the page numbers as well. Thus fn001-02 means, it is footnote #02 found on p001, and fn003-07 is footnote #07 found on p003.

UKT: No footnote on page p007
(p008fnc1begin-  )

fn008-17. Among the works treated by me, at most Jm might possibly be questioned. It may be a borderline case, but seems to me to have enough characteristic BHS vocabulary to justify inclusion. It certainly has few traces of Middle Indic morphology. Cf. however, the gerund (p008fnc1end-p008fnc2begin) adhiśrāya Jm 35.5 (§ 35.20); the pples. (an-) āstarita 220.14 and (Aṅga-) dinna 192.12; sātmy eva bhavanti 95.9 (= sātmībhavanty eva); and the characteristic BHS and Pali locution teṣāṃ ... etad abhūt 41.23.
- fn008-17b

Contents of this page

UKT notes

Néwari speech - the cousin of Bur-Myan

UKT 160125: Whenever we meet the word "Nepalese", we have to be careful of the two languages which are almost the opposite of each other. Nepal-Bhasa नेपाल भाषा {né-pa-la. Ba-þa}  aka Néwari (spoken by the blood relatives of Gautama Buddha) is Tib-Bur related to Old Magadhi, whereas Nepalese नेपाली {né-pa-li}, (spoken by the conquerors of the kingdom) is IE akin to Skt-Dev. Presence of velar /ŋ/ phoneme in words and syllables in both Burmese and Néwari shows that the two languages are closely related.

Néwari speech was written in Asokan Brahmi which I insist has evolved in Myanmar akshara. That Néwari is akin to Bur-Myan can be seen from words and syllables that begin with velar phoneme /ŋ/ which is absent in Eng-Lat and Skt-Dev. Néwari was written in Asokan up to about 13th century when it was first changed into Gupta script, through Bhujinmol script, and finally into Devanagari.
See: https://en.wikipedia.org/wiki/Newar_language 160125

What Prof. Edgerton was referring to "Nepalese" was probably Néwari speech in Bhujinmol script.

Go back Newari-speech-note-b

Contents of this page

The Tell-tale Trio : თ Tan ი In ა An

- UKT 160126:

One of my searches concerns how the Georgian Alphabet-Letter writing system has come to include the two (or three) clearly Bur-Myan Abugida-Akshara glyphs, თ Tan, ი In, and ა An . These three, through the phones /t/, /i/ , /a/ are related to the Bur-Myan {ta.}, {i.} , {a.} . Notice that ა An is the inverse of the Virama-sign , from where you can get the idea that it is the vowel-giver (opposite that of the Virama the vowel-killer) to the Letter (mute) to convert it to an Akshara (pronounceable syllable). Apart from my suggestion (conjecture) that there might have been Buddhist monks from the ancient kingdom of Tagaung in northern Myanmarpré, among the Buddhist missionaries of King Asoka, I venture (conjecture) to suggest that natives of Myanmarpré had gone west to become Buddhist monks . Emperor Asoka the Great, the Buddhist emperor of India was a Magadhi speaker like the Gautama Buddha: the Buddha preceded Asoka by a couple of centuries. When we note that the name of the Buddhist linguist-monk Shin Kicsi {kic~sæÑ:}, who was praised by the Buddha himself, has the killed Nyagyi denoted by {Ñ} as the coda, we might assume (conjecture) that there had been monks from Myanmarpré in the entourage of the Gautama Buddha - the Teacher.

Another suggestion (conjecture), is that because the southern Silk-Road had gone through Myanmarpré, through Pagan kingdom and its predecessor the Tagaung kingdom, the trio {ta.}, {i.} , {a.}, might have been well known to the traders of the northern Silk-Road which had gone through Xinjiang (xīnjiāng) aka Chinese Turkestan and ended up in the country of Georgia.

Please note that Chinese Pinyin with which the Chinese calligraphy can be changed into the Latin alphabet is based on the same idea of Romabama {ro:ma.ba.ma} - my invention. I had no knowledge of Pinyin at the time of inventing Romabama a quarter of a century ago. I am still unfamiliar with Pinyin. See Wikipedia:
- https://en.wikipedia.org/wiki/Pinyin 160126

Go back Tell-Tale-Trio-note-b

Contents of this page

End of TIL file