The Recitation of the Vedas

Saturday, February 9, 2013

Posted Under: Aesthetics, Philosophy (studies)

Written By: Pierre-Sylvain Filliozat. Translated from the French by T.K. Gopalan.

The earliest Vedic hymns testify already to a high degree of linguistic awareness of its authors. It appears for example in the play on the etymology of words. Thereafter, while the language seems to evolve relatively little, the knowledge of its structures develops considerably. Conservatism and linguistic awareness seem to go hand in hand. That could only have increased the distance between the ‘popular’ language of the non-specialists and the language of the experts in religion, since the former is always capable of a spontaneous evolution. In this situation it is important, for following the evolution of Vedic, to describe the preservationist measures which have influenced its destiny, to determine its environment that consisted of other forms of speech, to uncover the diverse and fluctuating influence of these forms of speech whose own evolution was probably much faster.

The first act of preservation is the organisation of the transmission of the Vedic texts, in the absence of writing. The Vedic texts began to be transmitted orally from master to disciple, by word of mouth, the quality of the transmission depending on human memory. The Veda, literally ‘knowledge’, called also ‘śruti’ literally ‘hearing’ is seen as ‘memory-matter’ or ‘auditory data’, not as a book of Scripture. The recitation is the decisive factor of the preservation. Eleven methods of recitation have been codified during the course of the Ages. At least three are of great antiquity. The starting idea is that the same text can be presented in several forms. Each of these forms is learnt independently of the others. Once the reciter has memorised all the eleven, one after the other, he can then compare them and knowing the encoding rule for each see if there is indeed agreement between all, detect an occasional lapse of memory or an alteration when there appears a deviation in a recitation, and correct this fault. In this way will be preserved the integrity of the text. There are three basic forms and eight derivatives.

The first encoding consists in reciting each stanza continuously (Samhitā) without a pause between words, except for the mid-stanza pause. This is the most natural situation, reproducing that of the ordinary language in which the sentence is the most compact string possible. In Vedic that implies the application of phonetic changes at the point where the words are joined to each other. The final syllable of almost every word undergoes two different phonetic treatments depending on whether it is followed by a pause or by another word. A final vowel contracts when followed by an initial vowel of the same sound, such and such final consonant assimilates such and such initial consonant of the following word, etc.

The second encoding consists in reciting the same text with a pause after each word (pada). That results in modification of the final syllable in a number of cases. That is also a great help for the correct understanding of the text. There are at times ambiguities because of several possibilities of dividing the same string of words. For the preservation of the meaning of the text the analysis into words has to be known. For example there is a string şţutaítu which can be analysed into sú-stutā etu ‘may [the goddess Speech], meetly lauded, go away’ or into su-stutā ā etu ‘may she, meetly lauded, come [to us]’ which is evidently the meaning intended. The meaning of ‘to come’ is obtained thanks to the verbal prefix ā qualifying the imperative of the verb i ‘to go’. The verbal prefix disappears in the contractions determined by the continuous recitation. We also understand from such an example the importance that this analysis into words must assume in the eyes of the officiant who counts on the prayer for the successful accomplishment of his ritual, who believes in the effectiveness of this word, who knows that this effectiveness depends on the accuracy of his pronunciation and his comprehension. The fragility of word frontiers in Vedic, the fact that this language does not accord the same treatment to the frontiers of words and to those of morphemes (i.e. between noun or verb stem and suffix beginning with a vowel), and the fact that the demarcation between these two areas is not simple make the task of division of the continuous string into its constituent words very delicate. It is said to have been accomplished by the grammarian Śākalya whose date is not known to us, but of whom it can be said that he was earlier to Pānini and that, in the history of linguistics, his is the earliest existing work of a linguist. By separating the words of the Rgveda, Śākalya truly accomplished the work of a linguist. His analysis demonstrates an extensive knowledge of phonetics and morphology.

A third encoding consists in linking each word in the text to the following word, and giving a pause after each pair so formed. That results in the application to the final syllable of each word two phonetic treatments successively: before a pause and before the initial syllable of the next word. In the case cited above where a monovocalic word disappears in a contraction of vowels in a string, this recitation isolates it in a group of three words. The fixing of this recitation called ‘by steps’ (krama-pāţha) is due to a grammarian of the name Bābhravya and is of equal antiquity. With the use of the particle iti as an indicator of pause and some other rules governing the case of compounds, etc. these three encodings provide much advanced information on the form of the text and of the elements that are useful for a proper understanding of it. They are thus considered as a foundation (prakrti).

Eight other encodings (aşţa-vikrti) are derived from the material furnished by the first three. They do not bring additional information, but can serve as cross-checks for the detection and correction of errors. Each comprises an arrangement rule for individual words and pairs of words. For example the jaţā-pāţha (‘in plaits’) recitation consists in enunciating the first pair, repeating it in reverse order, reciting it in the normal order and giving a pause: ab ba ab | ; the ghana-pāţha (‘dense’) recitation takes the first pair, reverses it, takes the group of the first three words, reverses it, repeats it in the given order, gives a pause: ab ba abc cba abc | , etc

The text of the Rgveda alone consists of 10462 verses and 153826 words. The learning, by hearing alone, of just the continuous recitation (samhitā) takes two years, that of the eleven some twelve years. The child is initiated into it at the tender age of six to eight years, when his memory has the maximum capability of being fed with material. Thereafter the recitation becomes a profession. The first duty of a reciter is to teach it, since that will ensure the transmission. Moreover, a religious value is attached to it. It is a ritual performed in homes and temples, on certain occasions, for various purposes the nature of which has varied from one region to another during the course of history. This ritual is most often financed by a devotee. It is a ritual that is still alive at the present time and the profession is still practised. Whereas the early antiquity of the first three recitations is certain, it is probable that the eight others are not as old. We do not know when they originated. What is undoubtedly true is the success of the oral transmission of the Vedic texts across more than four millennia, even before the appearance of writing, which is not very old in India. We observe that at the present time when the aid of writing is available there are still reciters who can completely do without it. To our knowledge, only the first two recitations, continuous and word by word, have been written down in the past and printed in the 19th century (editio princeps of the Rgveda by Max Mueller). The other recitations have never been committed to writing, apart from small samples (cf. appendix of the recent edition of Satvalekar). A program to have them written by a computer was made for the first time in 1991. These recitations are practised most commonly for the Rgveda and the Taittirīya-Samhitā of the Yajurveda. There are also reciters of the Śukla-Yajurveda (white Yajurveda) and the Atharva­veda, but far fewer in number. The Sāmaveda is not recited by these methods, but is chanted. Both the text and the melody are memorised by hearing. There are also several methods of singing, and it lakes a long time to learn all of them. This tradition is still alive in several regions of India (Tamil Nadu, Kerala, etc.). It is probably the only case in the world of transmission of music from very early antiquity. Some regional modifications may have taken place during the course of history. But we have sound reasons to believe that the proportion of innovation is less than that of preservation.


Extract of the book The Sanskrit Language: An Overview. History and Structure, Linguistic and Philosophical Representations, Uses and Users by Pierre-Sylvain Filliozat (Indica Books, Varanasi, 2000, 2009).

All diacritical signs have not been given