Corpus Effects in Music


Date: 16 Nov 1995 06:07:46 GMT
Message-ID: <48ekfi$ov1@spool.cs.wisc.edu>
Subject: Corpus effects in music

The following essay is rather long. It also contains many opinions that will anger and/or irritate many people (so read at your own risk). Unlike some people, I have provided full references to my sources and refrained from using vague cliches. Those with access to a good library can check all my quotations.

The essay summarizes my position on four topics:

The sections are meant to be read in that order. I welcome all sorts of objective and debatable comments (polite or otherwise) by e-mail (I will certainly respond to e-mail, a guarantee that I cannot make for Usenet postings). On the other hand, I do not have the time or the inclination to respond to subjective hand-waving.


The curse of Corpus Entities in Music

Definitions

A corpus is an extant body of work that does not change---it is defined by a set of examples which cannot be added to, deleted from, or modified.

Scriptures are a form of corpus. I cannot write a scriptural verse today; not because I am incapable, but because the term scripture refers to a static body of work (some people may relent a little and say that my verse is in the scriptural style but is not scripture proper).

Corpus entities should be anathema in a living and improvisational system like karNAtaka music.

Corpus rAgas and compositions

There are some rAgas in karNAtaka music that are slowly starting to take on the status of corpus rAgas. These are rAgas which are represented by single compositions (what I will call corpus compositions). Given that composers do not compose new kRthes in this rAga, and given that performers do not sing elaborate AlApanAs in the rAga (indeed, musical practice has shown that the ability to perform AlApanAs in a rAga increases as one learns more compositions in that rAga), indulge in naeravals or svara kalpanAs, the audience starts to identify such rAgas based on the kRthe, not, as they properly should, based on the melodic characteristics of the rAga.

Incidentally, we can already see a limited form of the corpus effect when someone posts a question to RMIC asking what rAga is nagumOmu?. Most people would, if asked to answer the question within a couple of seconds, say AbAErE. Many others might say maDhyamAvathE. A moment of reflection should show us why the question and the answers are so totally ludicrous. We can be absolutely sure that the readers of the question cannot hear the notes of the kRthe; they certainly do not see a notated version either. How can they then know what rAga this kRthe is in? Because they know that there are only a couple of popular kRthes that start with the phrase nagumOmu? But this is precisely the corpus effect in action.

If I wrote a new kRthe that started with the words nagumOmu ganalAEnE, but in the rAga rAEvathE, I would immediately thwart all those corpus identifiers. In the past, I have mentioned why posting just one word from the start of a composition is close to meaningless because composers have, in the past, done precisely what I proposed in the previous sentence i.e. composed new kRthes whose starting phrases were deliberately the same as those of well-known kRthes. patNam subrahmaNya Iyar's aenthanAErcenA in sAvAErE (based on thyAgarAja's SudhDha-dhanyAsE kRthe) is an example.

Of course, we do not yet have widely available means of transporting sound across networks and having it heard by everyone reading the rAga queries. But that only means that the people asking the question should post more than one or two words from the kRthe. If they do not give detailed information, they will get responses that only strengthen the corpus effect.

This (i.e. the identification of a rAga with a single kRthe) is the worst state of affairs that can happen in music. A rAga should be defined by something other than a very specific sequence of notes and lyrics. [It is bad enough that many people identify even popular rAgas by kRthe association, as I noted with the nagumOmu example].

The ill-effects of the corpus phenomenon

One example of a kRthe-rAga identity is that between the kRthe dhAEvE brOva samayamedhAE and the rAga centhAmaNE. Veterans of RMIC may recall the discussions about why different performers sing this kRthe in a rAga that they call centhAmaNE but which sounds very different (ranging from ShaNmuKapreyA-like to rAmapreyA-like). Part of the reason why this is so is that there is nothing else to compare this kRthe and rAga with. Singleton sets are notoriously susceptible to mutation and drifting.

Corpus rAgas and corpus compositions are also very bad for the learning process. Learning and understanding karNAtaka music should be a process of acquiring those skills that allow us to identify melodic aspects in a composition. Identifying rAgas by words in a composition is no different from looking into a large dictionary (I know from past experience that the previous statement infuriates some people. I stand by my position. In my opinion, those who identify rAgas by kRthe are really not identifying the rAgas, and that they are quite ignorant of the significant aspects of karNAtaka music).

I will return to the problem of corpus entities after a brief digression into a question that I am sure that baffled many of the readers of RMIC.


The problem of close rAgas

Question

(Q) What is the difference between dharbAr and nAyakE?
(A) No one knows.

Lack of clear distinctions

Why do I say this? Because no one has yet given me a definitive, specific, objective answer to the question.

Here is a section from Prof.sAmbamUrthe's book on South Indian music (South Indian Music, Book IV, 3rd edition, by P.Sambamoorthy, pp. 383, The Indian Music Publishing House, Madras, 1963---for those interested in full citations):

dharbAr is different from the rAga karNAtaka-kApE though both possess some common prayOgas. In the phrase mrGGrs the frequency of the ga is more than 6/5 i.e. 6/5+. [...]

A rAga closely allied to this is nAyakE. The difference between the two is revealed in the uththarAngga.

Presumably sAmbamUrthe was a knowledgeable scholar of the grammar of classical music. Why then is he so meticulous and detailed in differentiating between dharbAr and karNAtaka-kApE (two very different sounding rAgas, in my opinion), going to the extent of talking about micro-tonal frequency differences in specific prayOgas (something I am sure most people are completely oblivious to), but is at the same time so breezy and vague about the difference between dharbAr and nAyakE (two rAgas whose similarity is legendary to the point of inspiring a large body of associated and, very likely, apocryphal stories).

It is not that sAmbamUrthe does not want to load the reader with technical details. The entire book (and the entire series) is highly technical and is meant as a textbook for students of musicology.

But wait, it gets even better. Here is what the same sAmbamUrthe says in South Indian Music, Book V, 2nd edition, pp. 276, 1963.

[...] dharbAr is a closely allied rAga. Compared to dharbAr, ga and ne in nAyakE are slightly less in pitch.

Of course, there is no indication of how much less in pitch these two notes are (although sAmbamUrthe found it pertinent to quote pitch variations for dharbAr and karNAtaka-kApE).

Continuing on, to Ragas in Carnatic Music, Dr.S.Bhagyalekshmy, CBH publications, Trivandrum, 1990, we find this on the section on dharbAr (pp. 126):

[...] A closely allied rAga to dharbAr is nAyakE.

and in the section on nAyakE (pp. 268):

[...] Closely allied rAga is dharbAr. ga and ne in nAyakE are not in its exact place but sounded slightly higher than its own pitch. [...]

Well, well, well. You really don't say. sAmbamUrthe wants ga and ne in nAyakE to be flattened w.r.t dharbAr, but BAgyalakShmE wants the exact opposite i.e. the pitches to be sharpened.

Undaunted, I went to another source: Raganidhi, B.Subba Rao, Vol I-IV. Here is the section on dharbAr (Vol. II, pp. 46-47):

[...] svaras ga and ne are elongated in avarOha with a special gamaka which is beautiful and characteristic of this rAga. [...]

In nAyakE, the svaras ga and ne are long and have a special charm. In ArOha, ne is taken in vakra sanjcAra (see nAyakE).

Although ga and ne are supposed to be kept elongated only in nAyakE, the temptation to do in dharbAr also cannot be overcome easily. Hence we find avarOha of dharbAr also being rendered like nAyakE with ga and ne elongated.

Here, subbA rAu assists our queries by contradicting himself in the space of a few paragraphs. He first says that ga and ne are elongated in the avarOhaNa of dharbAr and then says that these notes should not be elongated in dharbAr.

All the florid talk about special gamakas and beautiful gamakas are, once again, indicative of the simple fact that the writer has no clue as to what is going on and is groping around for nice-sounding descriptions that fill up some space, seeming to say something significant without actually doing so. subbA rAu is also in disagreement with both sAmbamUrthe and BAgyalakShmE as to the distinction between dharbAr and nAyakE.

In short, if someone contradicts themselves, and if two or more people contradict each other, then it is almost certain that none of them is correct. The following table summarizes the comedy from these well-respected musicologists.

+-----------------------------------------------------------------+
| musicologist   distinction between dharbAr and nAyakE           |
|-----------------------------------------------------------------|
|                uththarAngga (i.e. vague and unusable),          |
| sAmbamUrthe    ga/ne are lower in pitch in nAyakE               |
|-----------------------------------------------------------------|
| BAgyalakShmE   ga/ne are higher in pitch in nAyakE              |
|-----------------------------------------------------------------|
|                ga/ne are elongated and not elongated in dharbAr |
| subbA rAu      ga/ne are elongated in nAyakE                    |
+-----------------------------------------------------------------+

Here is another, albeit less formal and scholarly, incident.

In the movie todi rAgam (starring madhurI sAEShagOpAlan) there is a point where the heroine asks the hero for the exact difference between dharbAr and nAyakE. Instead of answering the question, the hero launches into a highly charged romantic monologue about how astute the heroine is to ask such a detailed and technical question, why she is asking such a question, and blah blah.

The simple answer to why the musical question was not answered is that it would go over the heads of the average movie-going audience. That is just so much nonsense.

Quite likely, sAEShagOpAlan and kunnakkude vIdhyanAthan, the film's producer, do not know the difference, although sAEShagOpAlan has sung dharbAr and nAyakE kRthes on many occasions. Before droves of fans come after me for saying such a thing, I would suggest that they also find out what the exact difference between dharbAr and nAyakE is and let me know of that, along with their flames (which would be well-deserved if and when the flames are accompanied by the difference between the two rAgas in a form that is well-defined, usable, and does not contradict other opinions on the subject). While the fans are at it, I would also like an explanation from them for the fiasco among these musicologists on the difference between dharbAr and nAyakE---I want people to make a categorical statement to the effect that one or more of the above authors was talking crap.

Thirdly, I have on innumerable occasions asked people for the difference between dharbAr and nAyakE. What I have gotten back is one of:

  1. honest admission of ignorance (in my opinion, the best and most commendable response).
  2. something about how I cannot understand the difference even if told (in my opinion, clearly indicative of both the speakers' ignorance and of their arrogance).
  3. some vague, incorrect, or unusable babble.

I am still looking for the distinction between dharbAr and nAyakE and I am fairly confident that I will continue to for a long long time.

Non-existent "differences"

In my opinion, this alleged difference between dharbAr and nAyakE (so subtle that even the best are apt to slip up [yeah right !!]) is simply a device that has been cooked up to distinguish the cognoscenti from the uninitiated. Either there never was any difference, or the difference was not a subtle one and the two rAgas, for some reason, converged to the same current point. People started to cook up post-hoc differences to explain a non-existent phenomenon. And since no one wants to admit that they cannot see any difference, the emperor continues to have new clothes.

It is not enough for people to say that when they hear yOcanA kamala-lOcanA they know that it is dharbAr, and that when they hear kanugOnu sAUKyamu they know (in their bones, as it were) that it is nAyakE. As I have said before, this is nothing more than the corpus effect (i.e. identifying rAgas by existing kRthes rather than by shared and distinguishing melodic characteristics).

Countering corpus effects

Corpus effects are one of the reasons why I am very much in favor of computer-synthesized music as a learning tool in kArNAtaka music. There are two advantages to computer-generated sequences:

  1. It is not simple to instruct the underlying computer program on how to generate a rAga. The difficulty is in part due to the vague and fluffy notions people have about what exactly constitutes a rAga. Being forced to make these notions concrete will make sure that non-existent but alleged distinctions between rAgas will disappear. After all, if it looks like a duck, walks like a duck, and quacks like a duck, why not call it a duck.
  2. Once programmed, students can turn to the computer for training on how to distinguish the subtle nuances between rAgas. A typical examination session would consist of a random sequence of random phrases from two or more close rAgas. Both the student and the teacher will have to identify the rAgas (including the teacher is important).
    Not playing existing kRthes, but choosing random phrases that conform to the syntax of the rAga makes sure that there is no danger of the corpus effect (i.e. the teachers, simply by virtue of the fact that they have heard all of the 5 compositions in a rAga, cannot do better than the students who may not even be aware of these compositions).
    If the teacher does no better than the student, then it is time to look for a new teacher. This will also go a long long way is stemming the blind, hero-worshiping, faith that some people have in their teachers (who can apparently do no wrong). It will also do a lot to stem the arrogance and I know all there is to know attitude among some musicians, because now we have an objective way to determine at least one facet of musical knowledge.
    If both the teacher and the student do no better than a random selector, then it is time to revisit step (1) and start thinking about whether there really is any distinction between the two rAgas.

Now, I know that there are many people who will be scandalized at the very thought of letting a device with the potential for objectivity and invariance (like the computer) into music (regardless of how it might help improve the state of affairs). I am also not saying that the human teacher should be eliminated in favor of a computer. All I am saying is that a properly programmed computer (and we have all the necessary hardware and software as of this very moment) can prevent us from deluding ourselves and falling into traps.

I would certainly love to try the computer experiment on the so- called musicologists, musicians, and critics of karNAtaka music (assuming of course that I can find a consensus on what the difference between dharbAr and nAyakE really is, in order to be able to program the computer). That should help shatter some egos and myths.

It is high time we all accepted that not only is the emperor stark naked, but that he has been this way for more than a couple of generations.


BIravE vs. mAnjjE

What, if anything, is the difference between BIravE and mAnjjE?

The problem with BIravE and mAnjjE is this. BIravE is a popular rAga, known from antiquity, the basis of a large number of compositions by different composers from different time periods, and the subject of development by different performers. mAnjjE is a rare rAga, known only from recent times, close to being represented by corpus compositions, and rarely performed (SyAmA SAsthre's brOvavammA thAmasamAElAE seems to be the only kRthe for which recordings are widely available).

Given the improvisational nature of karNAtaka music, the scope of BIravE keeps expanding with time, and the scope of mAnjjE either remains constant or starts to decrease (fewer and fewer people learn how to sing it; even these few become reluctant to experiment with it, since experimenting implies making mistakes, and thus invites the criticism from the seemingly all-knowing critics). Is it then any surprise that BIravE is so improvised that it encroaches into the territory previously occupied by mAnjjE.

Lack of usable information or consensus

Here is what sAmbamUrthe has to say about mAnjjE in his History of Indian Music (1st edition, pp. 173, The Indian Music Publishing House, 1960).

This is a CAyAlaga type of BAShAngga rAga and traces of BIravE are seen in many places. The individuality of mAnjjE is brought about by the phrase pGRS. The wide shake of the sADhAraNa gAnDhAra in the phrase dpG~ deserves to be noted.

Wow!! How enlightening. Statements like wide shake of are indicative of the fact that there is really no objective difference between the two rAgas---since there is no proper definition of what a wide shake is and what it isn't. The statement about the phrase pGRS is a little more useful since one can learn to listen for it. But I cannot see anything in the rAga lakShaNa for BIravE which prohibits the use of such a phrase. In that case, does the distinction between BIravE and mAnjjE become a matter of counting the number of times pGRS was used in a composition, defining a cutoff point, and then deciding the rAga based on which side of this dividing line the count fell?! Do we need such hair-splitting in music?

What is worse, this emphasis on single phrases (which like the single kRthes of corpus rAgas) will lead to imitative compositions all stressing this one phrase to the exclusion of everything else. The result is a total loss of freedom in improvising the rAga---something that we do not want in karNAtaka music. brOvavammA thAmasamAElAE is an excellent example of this kind of repetition.

The very word CAyAlaga should ring warning bells in our heads---it is very likely indicative of the fact that there is some difference that is postulated, but is really absent.

The analysis of sAmbamUrthe's and BAgyalakShmE's descriptions of mAnjjE is quite illuminating.

Here is sAmbamUrthe (South Indian Music, Book VI, pp. 207):

[...] This is also a BAShAngga rAga with the foreign note incorporated in the scale itself. In these two respects, it resembles the BIravE rAga, but still differs from it on account of the wide shake or dhErGa kampetha of the sADhAraNa gAnDhAra.

[...] jaNta svara phrases are not permitted in this rAga, and this is one of the points of difference between mAnjjE and BIravE. [...]

Note: A rAga called mannjjsa BIravE figures in the compositions of purandharadhAsa. It is possible that this rAga contained traces of mAnjjE. The rAga manjjesa (sic) BIravE is now obsolete.

Here is BAgyalakShmE (Ragas in Carnatic Music, pp. 238-239).

[...] An aekasvara BAShAngga rAga wherein the foreign note is incorporated into the scale itself. Even though in structure it resembles BIravE, it differs due to the shake of dhErGa kampetha gamaka given to sADhAraNa ga. [...] Unlike BIravE jaNta svara, dhAttu svara, prathyAhatha gamaka prayOgas are not permitted in this rAga, which is the main difference between the two. [...] purandharadhAsa has composed in a rAga named manjjuSha BIravE which closely resembles mAnjjE.

In this case (as opposed to the dharbAr-nAyakE fiasco), sAmbamUrthe and BAgyalakShmE agree. Both attribute an extended shake to the gAnDhAra of mAnjjE, the absence of jaNta svaras, and an origin in an unknown and obsolete rAga via a composition of purandharadhAsa. But that is only the surface.

There are lots of things to note here. Although sAmbamUrthe and BAgyalakShmE refer to a dhErGa kampetha gamaka, they do not say what exactly the difference is (there is no rule in the lakShaNa of BIravE prohibiting a dhErGa kampetha gamaka on the gAnDhAra---the fact that mAnjjE uses these is a manifestation of the corpus effect, very much like saying that the rAga malaharE does not have certain prayOgas based on a single gEtham SrE gaNanATha).

Secondly, the statements about the absence of certain svaras and prayOgas in mAnjjE is equivalent to saying that mAnjjE is BIravE sung by a beginner.

The third is the gratuitous reference made to manjjuSha (or manjjsa or manjjesa) BIravE, without any indication of what this rAga is (BAgyalakShmE herself does not describe this rAga, sAmbamUrthe at least admits that it is obsolete). Apparently, we are supposed to know about an already vague and ill-defined rAga based on a completely unknown and un-defined rAga. It is this gratuitous reference (in addition to the similarity of phrasings) which makes me suspect that either BAgyalakShmE was simply parroting what sAmbamUrthe had written, or, more likely, both of them were parroting what someone else had written (without any basis in fact).

Here is what subbA rAu says about mAnjjE in his Raganidhi (Vol. III, pp. 131).

[...] This type resembles the karNAtak rAga BIravE very much. As in BIravE, the cathuSruthe dhIvatha is used in ArOha. Hence this is also a BAShAngga rAga like BIravE. This is a difficult rAga. Hence AlAp and svara singing in this rAga are attempted by only expert musicians. Compositions in this rAga are very few. [...]

And that is it. Does subbA rAu think that saying this is a difficult rAga is sufficient and usable material to distinguish between BIravE and mAngjE.

The following table summarizes the comedy among the musicologists regarding the BIravE-mAnjjE distinction.

+---------------------------------------------------------------------+
| musicologist   distinction                                          |
|---------------------------------------------------------------------|
|                dhErGa kampetha of ga (unusable)                     |
| sAmbamUrthe    absence of jaNta svaras in mAnjjE (corpus effect)    |
|---------------------------------------------------------------------|
|                dhErGa kampetha of ga (unusable)                     |
| BAgyalakShmE   absence of certain prayOgas in mAnjjE (corpus effect)|
|---------------------------------------------------------------------|
| subbA rAu      mAnjjE is difficult                                  |
+---------------------------------------------------------------------+

Does mAnjjE use the anthara-gAnDhAra

Srini Pichumani (in his RMIC posting) says that the use of the anthara-gAnDhAra in mAnjjE serves to distinguish it from BIravE.

This is really a non-sequitur since BIravE and mAnjjE then do not share the same scale and are not CAyAlaga rAgas. If the use of the anthara-gAnDhAra is the crux, then we are not talking about close rAgas at all. sAmbamUrthe does not refer to this gAnDhAra BAShAngga at all, nor do BAgyalakShmE or subbA rAu. If that is so, then dEkShethar's mAnjjE is not the same as the mAnjjE that others seem to refer to when they speak of it (shades of the AbAErE vs. karNAtaka-dhAEvagAnDhArE controversy, and the related debate about whether AbAErE should use d1 or d2).

If BIravE and mAnjjE have identical svaras, and identical Sruthes, and are supposed to be distinguished only by some subtle nuance, then it is my claim that there is no difference at all between these two rAgas.

But it gets even better.

In any case, I went to the primary source of information on dhEkShethar's compositions, the sangEtha sampradhAya pradharSenE. The section on mAnjjE has the full text and notation for the kRthe rAmacandhrAENa sa~rakShethOham, and in not a single place is the anthara-gAnDhAra specified (although all kinds of gamakas and the specific use of the sudhDha and cathuSruthe dhIvatha are clearly specified). If this is so, I question the authenticity of what kalpagam svAmenAThan plays as representing mAnjjE, or perhaps kalpagam svAmenAThan was simply playing an apasvaram (Mark Twain once wrote Beware of following health books, you could die of a misprint. I would like to say Beware of drawing inferences from corpus compositions, you could be misled by an apasvaram). The rAga lakShaNa for mAnjjE in the sangEtha sampradhAya pradharSenE does not mention anything about a gAnDhAra BAShAngga. At best, this is a post-hoc neologism by one or a few people who are desperately trying to find some difference between BIravE and mAnjjE.

mAnjjE has now successfully reached the status of a corpus rAga represented by a corpus composition (brOvavammA thAmasamAElAE is the only safe example). BIravE has taken over any territory previous occupied only by mAnjjE. Result: death of an independent rAga called mAnjjE.


The future status of corpus compositions

I have the following (radical) suggestion to make.

All corpus rAgas should be retired after a certain period of time. They have no place in an improvisational system of music.

New rAgas must, of necessity, start off being represented by corpus compositions. However, unless kRthes and pallavEs, which explore the scope of the new rAga, are composed and performed, the following will happen:

I rest my case.


[ Indian Classical Music | Krishna Kunchithapadam ]


Last updated: Sun Jun 27 17:00:19 PDT 2004
URL: http://geocities.datacellar.net/krishna_kunchith/rmic/rmic.1995.11.16.html

1