________________
810
JOHANNES BRONKHORST
Yijing's remarks on the Sanskrit grammarians would at first sight seem to be in disagreement with the above. A closer inspection shows that this is not the case. Yijing, as has been shown in detail by John Brough (1973), made a number of serious mistakes in his account of Sanskrit grammatical literature, confusing both authors and texts. It seems nevertheless clear that he knew (or had heard of) the Mahabhāṣya, which he knew by the name Curni, but which he failed to distinguish from the varttikas which it contains. Confusingly, he appears to use the expression Vṛttisūtra for the värttikas, but ascribes far too great a length to this text (18'000 slokas), which he seems to believe to have been studied independently from the Curni. This great length is no doubt to be explained by the fact, pointed out by Brough, that Yijing was unable to discriminate between the värttikas and the Mahābhāṣya. This leaves us with the question what Yijing may have precisely been referring to when mentioning the duration of studying this text. However that may be, he says the following about it (Takakusu, 1896:175): "Boys of fifteen begin to study this commentary, and understand it after five years. [...] All these books (?) should be learnt by heart. But this, as a rule, applies only to men of high talent, while for those of medium or little ability a different measure (method) must be taken according to their wishes. They should study hard day and night, without letting a moment pass for idle repose." About the Cürni he says (p. 178): "Advanced scholars learn this in three years."
It has to be repeated that Yijing's remarks have to be read with much caution. But assuming his testimony about the way of studying grammatical texts to be by and large correct, there is an obvious contrast with the way he described Vedic learning, which we already considered above. About the latter he says (p. 182): "The Vedas have been handed down from mouth to mouth, not transcribed on paper or leaves. In every generation there exist some intelligent Brahmans who can recite the 100,000 verses." He says no such thing about grammatical texts. In other words, these grammatical texts were not exclusively handed down from mouth to mouth. Indeed, Yijing makes a point of stating that less talented students would not learn them by heart at all. The long duration required for studying the Mahabhaṣya (three years? five years? eight years?) can be explained by the great complexity of its contents, not necessarily by the
latory organs and the extra-buccal process of articulation, they were taught the Vedic words. Nowadays, it is not like this. Having learnt the Veda [the students] are quick to say: 'the Vedic words are known [to us] from the Veda, and the ordinary words from common speech. [So] grammar is useless.' To those students entertaining false notions the teacher teaches this science [of grammar] saying: 'these are the uses, [therefore] grammar must be studied." (tr. Joshi and Roodbergen, 1986:68; modified)
LITERACY AND RATIONALITY IN ANCIENT INDIA
811
effort needed to learn it by heart. Yijing's account contains no hint that the Mahābhāṣya was ever studied the Vedic way.
If we now return to the question of scripts, it is hard for me, not being a specialist, to see how Falk can be sure that the Brahmi script at the time of the Sungas could not yet be used for Sanskrit, since practically all our early evi dence is in other languages than Sanskrit. With the growing popularity of Sanskrit or 'Epigraphical Hybrid Sanskrit' as inscriptional languages all the necessary characters (r, au, na, h, and halanta or virama) as well as consonantal clusters appear in the inscriptions (Salomon, 1998:37),26 It is not evident that this must imply that these or similar characters may not have been used in Sanskrit non-inscriptional writing well before these inscriptions. To this may be added the 'überraschende Tatbestand' (Hinüber, 1990:61) that Brahmi has ligatures which serve no purpose in the Middle-Indic languages for which it is used; ligatures are of course essential for Sanskrit.27
It is important to insist once again on the difference between Vedic memorisation and other forms of memorisation. Many of the non-Vedic feats of memorisation enumerated in Falk's chapter "Berichte vom Umfang des Memorierten" (1993:§ 14.1, pp. 321 f.) concern either texts that have been handed down in rather widely differing versions, thus showing the unreliability of this method of memorisation (cf. Falk, 1993:322: "Häufig mussten sich die Chinesen mit lückenhaften [buddhistischen] Texten zufrieden geben, weil ihren Gästen das eine oder andere Kapitel aus dem Gedächtnis geschwunden war"); or texts that could at any time be corrected with the help of their written versions.
26 Hinüber (1990:61) appears to consider it significant that the Lalitavistara, where it enumerates the list of Brahmi signs, skips the letter r. This same list does however contain ai, au, and ña; see Lal(V) p. 89. It is to be kept in mind that the list in the Lalitavistara is used to inculcate some important truths with the help of words or expressions that have the sound concerned in the first or second place (a: anityaḥ sarvasamskära[b]; a: atmaparahita; etc.). In such a list there is no place for h and halanta/virama, and perhaps not for r (the Sanskrit index to the Abhidharmakosabhäṣya-Abhidh-k-bh(Hi)-contains just nine entries beginning with, none of which may have been suitable). It is furthermore not clear that this enumeration of sounds concerns specifically the Brahmi script. In this context it may be of interest to note that the Kharoşthi script of one of the recently discovered early Buddhist manuscripts from Gandhara has a sign for ṛ (i.e. for kr, see Salomon, 1999:123).
27 Cp. Colas, 1997:127: "la finesse de l'analyse phonétique dont témoignent les premières écritures indiennes (attestées au IIIe siècle avant notre ère) trahit l'intervention des érudits mêmes qui déconsidéraient l'écrit. Le bon sens suggère donc que ces clercs employèrent l'écriture plus tôt que le IIIe siècle avant notre ère, peut-être dans des manuscrits utilisés comme aide-mémoire." On p. 129 Colas expresses his view that the first Indian writing systems must have been created in the circle of grammarians or under their influence.
