Follow by Email

Sunday, December 16, 2012

Should we emend Sanskrit metri causa?

Patrick Olivelle, in his 1998 article on the Journal of Indian Philosophy (Unfaithful transmitters), suggests that, in general, one should be more than cautious in emending the received text. This applies, he maintains, also to emendations metri causa (because of faulty metre). Why?

1. As explained by Max Müller (1879: lxxii, quoted in Olivelle, p. 179): "The metrical emendations that suggest themselves are generally so easy and so obvious that, for that very reason, we should hesitate before correcting what native scholars would have corrected long ago, if they had thought there was any real necessity for correction". And, Olivelle could add, if generations of paṇḍits, much more learned than we are, did not emend, they must have had a good reason for not doing it.

2. Several metrical faults can be explained away. Olivelle refers to Alsdorf's (1950, mentioned at pp. 179--180) attempt to explain faults in metre through non application of sandhi or through prakrit pronounciations of bhavati, iti, iva, etc.

As for 1., please read Michael's comment on the distinction between dead and living traditions here (would Olivelle/Max Müller say the same about a codex unicus reporting an odd spell?).

As for 2., as much as it can sound convincing in the case of Epics and of other genres, I cannot really imagine Jayanta or any other philosophical poet pronouncing a word in prakrit amid a verse in his Nyāyamañjarī. And I recollect having imagined emendations in the text of Vāmanadatta's Saṃvitprakāśa, whenever the metre was faulty.

What do you think? What did you do in your actual praxis?


Jayarava said...

Hi Elisa

I think my earlier comments on emendation apply here as well. By far the most interesting thing about a problem with metre is that it indicates imperfect transmission. Though this assumes that the metre was once perfect and now isn't.

A conjecture about the 'original' text might be helpful if the passage is incomprehensible, but otherwise the changes are telling us a story we're often deaf to.

Transmission of texts in ancient India was a complex process. Texts were constantly edited and redacted. At the moment I'm checking with a friend who is a geneticist to see if it's possible to apply the techniques of genetic analysis to the changes in texts. The whole text is like a genome. Passages are like genes. And they are subject to all the changes that genes are: insertion, deletion, duplication, mutation. Changes, whether random or motivated, accumulate in texts and genomes, and may tell us about the relatedness of texts to each other. I'm hoping that geneticists have an app for this that I can adapt.

We have this fairy story about faithful transmission of texts in India. For some reason we treat each unfaithful transmission as a mistake to be corrected - when it may well have been motivated and not random. I.e. a lot of the 'mistakes' were deliberate changes - despite the metre. And that tells us something about the mindset of the transmitters of the text. Rather than glossing over the editorial activities we should be mining them for what they tell us about the process of textual transmission and the priorities of the traditions we study.

Translation produces a whole new family of texts. The Heart Sutra has been translated dozens of times and I hope to analyse the language of these texts as well. I hope to create a taxonomy, and a method that can be applied on a larger and wider scale.

I have a very strong preference for footnotes over endnotes in this case. Endnotes are a waste of everyone's time. They are a left-over from a time when it was easier typographically to place notes together at the end (a time of individual letter cast in lead!). These days there is no technical reason to use endnotes - pagination is automatic.

I'm finding these question very interesting now that I have a chance to work on manuscripts and see how varied they are; and the kinds of variations that occur.

This comment has been removed by the author.

Some thoughts just rushed:

Among the six Vedāṅgas, chanda is one. And a popular verse from the Pāṇinīyaśikṣā says: tasmāt-sāṅgamadhītyeva brahmaloke mahīyate, i.e. one who studies the Vedas with the aid of all the vedāṅgas, revels in the abode of Brahmā. Also, the following verse: aṣṭavarṣe caturvedī dvādaśe sarvaśāstravit/ soḍaśe kṛtavān bhāṣyaṃ dvātriṃśe muni rabhyagāt, and such stories as an extempore composition of the Kanakadhārāstotram found in Śaṅkaradigvijaya attest Śaṅkarācārya’s skill in all the śāstras. Now Śaṅkara’s repute as a follower and upholder of the Vedic philosophy is too well known to require separate explanation. And if some metrical work, whose authorship is ascribed to Ādi Śaṅkarācārya, is found to have a verse written in faulty meter, hardly any traditional Indian paṇḍita would accept it as a genuine work of Ādi Śaṅkara. Instead they would try to explain away the fault in the following ways:

i. It is either the work of a latter author (a latter Śaṅkarācārya?);

(Though not directly concerned with the problem of faulty meter, I think it will not be out of place here to mention that such a modern Sanskrit scholar as Mm. Gopinatha Kaviraja, because of the ‘careless slip-shod style’ of the Jayamaṅgalā gloss on the Sāṅkhyakārikā, was unwilling to accept it as a work of Ādi Śaṅkarācārya, and instead he suggested that it was authored by one Śaṅkarārya. Kaviraja also mentioned that the mention of the name Śaṅkarācārya, instead of Śaṅkarārya was [perhaps] a scribal error.)

ii. The faulty meter in the manuscript might be a scribal error (and in the same vein, according it a benefit of doubt, it might be corrected);

iii. When it is a fact that there were and there can be such scholars who can compose beautiful verses in very long and difficult meters instantaneously (at least I have witnessed this in the case of Mm. Prof. Manudeva Bhattacharya of Varanasi), it is unthinkable that he would “write” a verse in a faulty meter.

Another interesting incident in this connection is worth sharing:

In Swāmī Karapātrī’s magnum opus, Vedārthapārijāta, by sheer mistake were the words, śrī gaṇeśāyaḥ namaḥ and śrī sarasvatyaiḥ namaḥ printed, instead of śrī gaṇeśāya namaḥ and śrī sarasvatyai namaḥ. This was vehemently criticised by Pt. Viśuddhānanda Miśra and his wife in the book, Vedārthakalpadruma. It was the Jagadguru Śaṅkarācārya of Puri, Śrī Nirañjanadeva Tīrtha who first indentified these as typos in his work, Vedārthapārijātavārtika, and wrote the following interesting verse with a view to providing remedy to wrong words (apaśabda): athavā na visargāste kintu śūnyadvayaṃ kṛtam/ jñātvaiva mandabuddhīnāṃ padabhedapradarśakam// (cf. Amiṭa Kālarekhā [in Hindi] by Parameshwar Nath Mishra, pp. 175-76)

Also, consider the following verse quoted in Nyāyamañjarī 5:

ata eva hi vākyārthaṃ bhāvanāṃ pratijānate /
yathocitaphalāḍhyāñca trayasambandhabandhurām// (passage no. 25, beginning with the prose line: phalam api na vākyārthaḥ siddhāsiddhavikalpānupapatteḥ…)

Now the Benaras edition reads this as ata eva …. pratijānīte. But Prof. Prabal Kumar Sen, in his 2008 Bengali edition and translation of Nyāyamañjarī 5 (Calcutta: Sanskrit Book Depot) prefers the reading pratijānate, and says in a footnote that if we read ‘pratijānīte’, it would lead to a metrical fault (chandobhaṅga). Does this mean that Suryanarayana Śāstri, the editor of the Benaras version of Nyāyamañjarī was less learned than Prof. Prabal Kumar Sen?

In the context of Mīmāṃsā philosophy, it is well known that Kumārila’s failure to make sense of a verse, made Prabhākara interpret it as “suṣṭhūktameva tatra tunā uktam atra apinā uktam itthaṃ sthaladvayepyuktam.” Does this mean that Kumārila was less learned that Prabhākara?

Metrical errors often remain in printed Sanskrit texts and this happens because either the errors sometimes genuinely escape the editor’s notice or they are deliberately ignored (the latter stemming from an unquestionable and blind reverence for all that is traditional).

elisa freschi said...


there are already interesting softwares working in a similar direction, i.e., analysing manuscript variants using cladistic analysis (the same kind of analysis applied to fossiles and resting on the idea of casual variations in the genome). Personally, I think that these systems might help in the case of a large number of manuscript, but that one should hold in mind that humans are final beings (i.e., they work having a purpose in mind) and in this sense differ a lot from the casual producers of random variations. Their purposefullness is all the more relevant the less number of variations and of manuscripts one takes into account.
You can read about it in the issue of the WZKS edited by Philipp Maas (articles by Wendy Phillips-Rodriguez and Philipp Maas).


thanks for the many examples. You seem to make a consistent argument in favour of the possibility of mistakes due to scribal errors (and, hence, in need of being emended). However, what about the many copyists who copied again and again the same mistake without emending it? Why did not they? Because they could understand nonetheless the text? Because they thought they had to just copy what they heard/read? Because the metre is not as fix as we believe?…………?


@ Elisa, in answer to the rhetorical questions you have nicely put, I can only say, "yes."


This also provides, I think, an extremely interesting topic for research: The Social and Intellectual Status of Scribes in Indian History.

nOe said...
This comment has been removed by the author.
elisa freschi said...

@ Sudipta: perhaps better "of copyists" (because they were not necessarily professional scribes).


"Copyist" is much better. Many thanks for the suggestion. This would also mean a consideration of the extent of their reliability, apart from the role they played in "transmitting" and making possible the "re-use" of texts.

Jayarava said...

Just noticed your response. Thanks for the link to WZKS. Very interesting!

I agree that texts are changed in ways that are not analogous to genes - i.e. texts are deliberately edited. Still it will be interesting to see which changes are transmitted and which are not. There are at least 40 versions of my text - understanding the relationships between them must employ some kind of visual representation I think.

A visual critical edition?

There are at least 40 English translations of my text, from at least 3 different languages, as well! I think I've got to the point where I could identify the language of the text the English translation was made from by the unique features.

elisa freschi said...

I am not sure I am following. The fact that there must be a visual representation of the genealogy/relations among the versions of a text does not entail its "biological" model. One can draw a stemma codicum or draw connective lines based on one's critical analysis of the text, without any hard-science-software, isn't it?
As for the idea of a visual critical edition, I dream since some time of a digital edition where one could visualise in the main text or in the apparatus one or the other manuscript. In this way, one could read the Kashmiri version of a certain text, than the Bengali one, the one preserved in a single Jaina manuscript in Jaisalmer and so on.

(BTW: happy new year, I wish you a lot of success for your new projects)

Jayarava said...

Sure it doesn't *entail* a biological model. But my first degree and my first intellectual love was science. The biological metaphor occurred to me while reading the mss., before I knew anything about cladastic or phylogenetic studies of texts. I think in these terms because of my education.

I could sit down a draw a diagram of the 40 or so versions of the text by hand. Indeed such a diagram already partially exists in my imagination - I can see it as I write this. And once (if) I see the remaining mss. it will be even clearer. But the relationships are not simple. I doubt a basic branching tree diagram is not going to be capable of representing the relationships. There will be many cross links - this is where software helps.

elisa freschi said...

Using a software to draw a tree is more than fine. My question is about how to "feed" the software. After all, a software treats all variants as variants, whereas a human being knows that, e.g., ś/s/s are easily confused in some areas and, thus, not as significant as other variants. And the same applies to other sandhi variants/the lack of anusvāra/typical scribal errors related to the change of script/typical aural errors… The evaluation of the weigth of each variants needs, I think, a historical appreciation of the copyist's milieu and background. This being said, I am happy that there are science-lovers who develop interesting softwares and I appreciate chances to co-operate with them.

Jayarava said...

Yes indeed. The issue of feeding the demon!

Licenza Creative Commons
Quest' opera è distribuita con licenza Creative Commons Attribuzione - Non commerciale - Non opere derivate 2.5 Italia.