Showing posts with label Yāska. Show all posts
Showing posts with label Yāska. Show all posts

05 December 2008

Yāska and his 'Nirukta'

In an earlier post on sound symbolism I made reference to the Indian grammarian Yāska, and I thought it would be a good idea to flesh out the picture of his method and why it should still be of interest to those interested in mantra.

Despite his subsequent influence, we do not know very much about Yāska. His dates are uncertain but most scholars place him between 700 - 300 BCE. His single surviving work is the Nirukta. The early grammarians were responding to a particular problem which was that the spoken language of the day had drifted substantially away from the almost perfectly preserved Vedic language of the sacred Vedas. This meant that passages of the sacred Vedas had become obscure or even unintelligible. Many passages in the Ṛgveda remain obscure. This is a natural consequence of language change and I have previously noted the example of the noun vahatu which occurs in the first verse of the Dhammapada, but whose meaning was apparently obscure to the commentators, and does not appear in traditional dictionaries. The response of the ancient Indians was to study and systematise their language - the contemporary studies of phonetics, grammar, syntax, lexicography and morphology owe much to the Sanskrit grammarians. The result was Classical Sanskrit - saṃskṛta means something like "crafted". Yāska was particularly interested in some of the words that had become obscure and systematised a set of principles for determining what they might mean.

The Nirukta, following an existing tradition, treats all words as deriving from verbal roots - these are the notional abstracts which underlie words. So from the root √budh, we get via a regular process the verb bodhati (to know). Similarly the past-participle buddha (one who knows), and nouns buddhi (intelligence) and bodhi (awakening) are treated analytically as deriving from the verbal root through a series of logical transformations. For instance in first class verbs the vowel in the root undergoes guṇa or "strengthening" with √budh become bodh; active present tense stems are formed by adding the vowel 'a', and then suffixes indicate person and number: 1st person singular bodhāmi, 3rd person plural bodhanti. the verbal noun. Historically the process must have worked the other way - through analysing a group of related word. An entire language was subjected to a detailed analysis without the aid of writing! It is a work of collective genius.

Some words are more difficult to trace. The verb tiṣṭhati (to stand) for instance is thought to come from the root √sthā. Other strange examples are √gam > gacchati (to go) > gata (gone), √dṛś > paśyati (to see) > dṛṣṭi (a view). So it is possible to come across a word and find that identifying the underlying concept is quite difficult.

As described in Eivind Kahrs 1998 book Indian semantic analysis, the Nirukta proposes three levels of analysis. Firstly there are obvious examples like √budh where the root and it's transformations are known. Secondly there are examples where the meaning is not obvious but one can use grammatical paradigms to work out what sense of it is - such as √gam. Thirdly there are very obscure examples which defy logical analysis. It is in these extreme cases that one must apply what has become known as a nirukta or nirvacana analysis. (Sadly I don't have a definite example of one of these).

This kind of analysis has been liken to etymology - the contemporary study of the way a word changes its meaning over time. So the word "know" comes into modern English from Old English cnawan, and is related to Greek gno- (as in the word 'gnosis'); and the Sanskrit. jña- "know" and comes ultimately from a Proto-Indo-European base *gno- "to know". This approach has allowed scholars to speculate on the existence of a language which must underlie all Indo-European languages - which they call "Proto-Indo-European" - and to specify what some of it's features must have been in order to give rise to the variations we see.

Yāska's procedure was somewhat different. Where the root of a word could be determined and was still obscure in meaning, Yāska employed a system of sound symbolism. That is to say that he employed the knowledge that words which share phonemes, especially initial phonemes, have a much higher likelihood of overlapping in meaning, than two words which do not share phonemes. If one approaches this systematically then it is possible to make fairly accurate guesses as to what a word might mean. Having narrowed the field, one can then use context get closer to the meaning.

Contemporary linguists are loath to accept that a phoneme can carry meaning, but there is no a priori reason to think this, and there is evidence to suggest that it is true. Meaning is of course a vague term - what does meaning mean? It seems to me that there is always a level of ambiguity in verbal communication - the higher the level of structure the more clearly defined the meaning being conveyed. An idea might be conveyed with a word, but then words can be ambiguous, and individual words can related in different ways to themselves and to their referents. A sentence relieves some of this ambiguity, but a complex idea may take a paragraph to express, and a book or even a series of books to fully explicate. At the other end of the scale as we break down words into their component parts we lose clarity - prefixes, suffixes and roots for instance are less clear on their own. Individual phonemes then represent a level below this and carry information with considerable ambiguity, but are not absolutely arbitrary.

So there is every reason for Yāska to resort to this feature of language when other more sure methods have failed him. Remember that he was highly motivated to find the meaning of words because they occurred in the Vedas and had the status of revealed and eternal truths. The loss of meaning in this context is disastrous! Just leaving the meaning obscure was not an option.

Despite the fact that his Nirukta is the earliest surviving text of this type Yāska was not the originator of this method, he was a systematiser. Evidence for the method emerges in the Brāhmaṇa literature - beginning perhaps 1200-1000 BCE. Eivind Kahrs notes example from thr Ṛgveda: uṣā ucchati - "the dawn dawns", which indicates a perception of the underlying connection between the two words despite being spelled somewhat differently. This search for connections - bandhu - is characteristic of the Brāhmaṇa literature and of the Vedic religion generally (see my Mantra, Magic, and Interconnectedness). Perhaps given the central important of bandhu in the Vedic religion it is no surprise that it should have been the approach to revealing linguistic mysteries.

Johannes Bronkhorst has drawn attention to parallels between the Nirukta and Plato's Cratylus. The two may well have lived at the same time, although it seems unlikely that they could have known each other. The main parallel of course is that both Yāska and Plato consider that phonemes can and do carry meaning, and can given clues as to what a word means. I covered this in my Yāska, Plato, and Sound Symbolism although there I illustrated Yāska's method with an example I found in Buddhaghosa's Visuddhimagga. I think now that Buddhaghosa is working at some remove to Yāska, although the sound symbolic aspect is still present and prominent. Buddhaghosa is of course applying the method to very familiar words which would have needed no explanation to Buddhaghosa's audience. Similarly T.P. Kasulis has drawn parallels between the Cratylus and Kūkai's Shōji jissō gi (The Meaning Sound, word reality - see Hakeda, Y. Major Works p. 234 f.).

The reason I think that Yāska is worth knowing about is that the ideas that he helped to systematise and popularise seem related to the way in which words have power in India. We say for instance that mantras are 'sound symbols'. This idea is underpinned by Yāska's theory. The use of sounds which have no apparent semantic content - such as oṃ or hūṃ - may make more sense when we recall that the milieu in which they were used was one in which a systematic study had been made of the way that words that sound alike are frequently related in meaning. I firmly believe that Buddhism is best understood against the background of Indian thought generally, and that to study the history of Buddhist ideas in isolation (which is typical) gives a false impression.

Note: 22 Dec
I didn't say this at the time, but in Yāska's day there were no books, no dictionaries or grammars. One met texts orally, and could only study them once they were memorised. Coming upon an unfamiliar form one had very limited resources - probably only one's guru - to consult. It's important to keep this in mind when thinking about this subject.

  • Bronkhorst, Johannes. 2001. "Etymology and Magic: Yāska's Nirukta, Plato's Cratylus, and The Riddle of Semantic Etymologies." Numen, Volume 48, Number 2, 2001 , pp. 47-203(57)
  • Hakeda, Y. Kūkai : major works : translated and with an account of his life and a study of his thought. (New York : Columbia University Press, 1972).
  • Kahrs, Eivind. 1998 Indian semantic analysis : the nirvacana tradition. Cambridge : Cambridge University Press.
  • Kasulis, T.P. Reference and Symbol in Plato's Cratylus and Kukai's Shojijissogi. Philosophy East and West, 32 (4), Oct., 1982, p.393-405. Available online:

Next Week: brahmacarya - the spiritual life.

image: Vedic text from

10 October 2008

Yāska, Plato, and Sound Symbolism

One interesting features of ancient Indian literature is the way they explain the meanings of words. There are two main methods: by giving series of synonyms, and by giving a series of words which appear to only be linked by using many of the same sounds. The latter method has been likened to the contemporary science of etymology, though usually it is seen as a rather debased or inferior version of etymology. In this essay I plan to give an example of an ancient etymology from the writing of Buddhaghosa, show that it has parallels in the Cratylus dialogue of Plato, and make some comments on these two with respect to sound symbolism especially the work of Margaret Magnus.

First a brief word about ancient epistemology. Foucault, in The Order of Things, points out that in Pre-renaissance Europe knowledge was based on resemblance, and that after the Renaissance knowledge came to be based on difference. In other words we now make sense of the world by looking for points of difference between 'things'. Information is meaningful to the extent that it represents something unique. We don't even think about this most of the time. However in the days of the Buddha and Plato people made sense of the world by looking for how 'things' were similar, looking for qualities in common. So for us a red pen, and a red vegetable are not related; but to the ancients the redness of the two meant that they could be seen to be related. This way of thinking is so foreign to us that we would say that the relationship is a false one. However it does mean that they tended to see the relationships between things, and to perceive everything as being interconnected; while we tend to atomise the world, and fail to understand inter-relatedness. The current crisis in the environment is an obvious consequence of this of this failure. The very way we polarise into right and wrong is influenced by this tendency to understand things in isolation.

In the Visuddhimagga Buddhaghosa offers a variety of same sounding words in order to explain various words. This basic procedure is called nirukti (Pāli nirutti) etymology after the text Nirukta by Yāska, a grammarian from about the 6th or 5th century BCE. In chapter VII Buddhaghosa explains the name/title Bhagavā in a whole series of nirukti etymologies, but in particular with this little verse from the Niddesa (a commentary which is included in the canon):
bhagī bhajī bhāgī vibhattavā iti
Akāsi bhaggan ti garu bhāgyavā
Bahūhi ñayehi subhāvitattano
Bhagavantago so bhagavā ti vuccati

The reverend one (garu) has blessings (bhagī), is a frequenter (bhajī), a partaker
(bhāgī) a possessor of what has been analysed (vibhattavā)
He has caused abolishing (bhagga), he is fortunate (bhāgyavā)
He has fully developed himself (subhāvitattano) in many ways
He has gone to the end of becoming (Bhagavantago) thus he is called "Blessed"
He also suggests that bhagavā can be understood as:
bhāgyavā bhaggavā yutto bhagehu can vibhattavā
bhattavā vanta-gamano bhavesu: bhagavā tato

He is fortunate (bhāgyavā), posssessed of abolishment (bhaggavā), associated with
blessings (yutto bhagehu), and a possessor of what has been analysed
He has frequented (bhattavā), and he has rejected going in the kinds of becoming
(VAnata-GAmano BHAvesu), thus he is Blessed (BHAGAVA)

(Visuddhimagga VII, 56, 57, p.225-226)
One of the things which makes the scientific etymologist doubtful about this approach is the obvious fluidity. In the space of two pages Buddhaghosa has offered two quite different versions of what bhagavā means. Our ideal is to have one explanation for each word. To some extent this is a hang over from what Umberto Eco calls "the search for the perfect language". For centuries westerners believed that in the perfect language (initially conceived of as the language which God spoke to Adam) each word would have a single referent, and each thing would have only one name. What we try to do with language is pin down meaning. Many people are disturbed by the fluid multiplicitous nature of the relationship between words and the world, but actually this is what language is like.

If one is familiar with Sanskrit or Pāli one might recognise that some of these words stem from the same notional verbal root, or are different only in their grammatical relationships. bhāgyavā and bhagavā for instance are both concieved of as stemming from a root bhaga (Sanskrit bhaj). However bhagga is from a different root, bhañj. So a theory of verbal roots cannot account for the relationship. In fact there is no obvious relationship between all of these words except for the the initial sound combination: /bha/. (In phonetics sound units or phonemes are placed between forward slashes, which helps in cases such as the letter c which can ambiguously sound like /k/ or /s/.)

Under the current paradigms of linguistics there is no possible relationship between sounds and meaning - these are denied by definition. So for linguists in general the fact that all these words share an initial sound is irrelevant to what the words mean. This has not always been the case in Western thinking. Plato put forward a partial account of the meaning of words based on the sounds of their letters in his Cratylus dialogue. He says for instance:
"Now the letter rho, as I was saying, appeared to the imposer of names an excellent instrument for the expression of motion; and he frequently uses the letter for this purpose: for example, in the actual words rein and roe he represents motion by rho; also in the words tromos (trembling), trachus (rugged); and again, in words such as krouein (strike), thrauein (crush), ereikein (bruise), thruptein (break), kermatixein (crumble), rumbein (whirl): of all these sorts of movements he generally finds an expression in the letter R, because, as I imagine, he had observed that the tongue was most agitated and least at rest in the pronunciation of this letter, which he therefore used in order to express motion".
There are clears parallels with Yāska's Nirukti method. Here Plato is suggesting that the letter rho (i.e. /r/) lends a quality of energy to words. Most contemporary linguists subscribe to a version of the idea of Ferdinand de Saussure (whose own theories on language were in fact influenced by his study of Sanskrit) which says that the relationship between a word and its referent is arbitrary. So any theory which posits a non-arbitrary link is de facto wrong. Behind the scenes however is a growing list of academic papers which demonstrate that the paradigm itself is unable to account for some obvious non-arbitrary links.

Margaret Magnus has definitively shown, in her doctoral thesis, the initial phoneme of a word has a symbolic function in the word. If one examines all words (with no suffixes) that begin with the same phoneme they fall into a smallish number of areas of meaning. Different phonemes create different clusters of meaning, and these do not overlap very much between phonemes. Her more popular account of what she calls "phonosemantics" is quite fun so rather than quote her research I'll give an example from her book Gods of the Word (full details below) which was written for a general audience. In what follows Magnus is describing, poetically and associatively, the impact of having the phoneme /r/ in a word, especially in the initial position:
"/r/ is active directed force. It is red, rowdy, and roguish. Run! Run! Run! But it is also rational. It does not feel. It reasons and acts. And reacts. If it is headed in the same direction as its neighbours, it can be supportive as rock. But if not, it leads to wrack and ruin. And /r/ is linear. It thinks in terms of 'right' and 'wrong'..."
In order to understand what Magnus is getting at one would need to comb through the dictionary and see that many of the /r/ words fit this picture that she is painting. If you have a spare hour this is a fun thing to do. The relationship is not one to one, it is fuzzy, it is symbolic. One cannot predict what sound that any given language will use for a particular referent. The pattern only emerges when comparing large numbers of words - Magnus was involved in creating an electronic dictionary when the patterns began to appear to her.

My feeling is that this contemporary research sheds light on the method of Buddhaghosa in defining words. It makes more sense when you know that initial phonemes do indeed effect what a word means.

I have long wondered whether this knowledge had any impact on the development of Buddhist mantra. While I have amassed a huge amount of information and many thousands of words of notes, I am still not in a position to assess it. I think it is suggestive that the Buddhist exegesis of mantra often focuses on individual syllables.

At the date of publication I am still largely responsible for the text of the Wikipedia article on sound symbolism or phonosemantics. This could change at any time of course. [Definitely changed 22.4.13] Another summary of sound symbolism can be found on

image: lipreading poster from