Vedic Theory of Language

Vedic Theory of Language

Vedic theory of language states that the relation between word and meaning in Sanskrit is “natural”, not “conventional”. This theory is found throughout the Vedic literature, and is most concisely stated by Jaimini: “autpattikastu shabdasyaarthena sambandhas” (Purva Mimamsa Sutras 1.1.5).

Such a theory of naturalness applies only to a language that precedes the invention of writing and the hierarchicalization of society, which are required for the persistence of artificially created words (i.e. words that are conventional and arbitrary).

But the language of the Rig Veda came into existence before writing, and so it preserves the natural relation between sound (“shabda”) and meaning (“artha”). And “classical” Sanskrit, although a written language, derives its sounds and most of its grammar from the earlier Vedic language. Also, the earliest “classical” works (such as the Upanishads, Brahmanas, Vedangas, Darshans, Itihasa, Sutra literature, etc.) likely existed as memorized, oral works long before they were written down.

Thus, the Susiddha project is safe in making the assumption that the natural relation of sound and meaning holds true in all the ancient works of Sanskrit.

In addition to “shabda” (sound) and “artha” (meaning), there are other concepts associated with the Vedic theory of language, such as “sphota”, “nama-rupa”, “nāda”, and the four levels of speech (parā, pashyantī, madhyamā, vaikharī). Authors associated with the Vedic theory of language include Katyayana, Patanjali, Jaimini, Bhartrihari, and Abhinavagupta.

The Susiddha project will further research into the Vedic theory of language, and especially focus on its application to AI and NLP. At this time, it’s too early to say how important a part this theory of language will play in building an AGI Avatar. We suspect it will play a very important part, which justifies research into: how it works; how it can be implemented in an AGI system; and how it can be used to extract usable knowledge directly from the Vedic literature via computational techniques (such as unsupervised deep learning).

Two fields of immense importance to the development of AGI are natural language processing (NLP) and natural language understanding (NLU). One topic frequently mentioned in NLU is “symbol grounding”. In order for there to be “understanding”, the agent must grasp the meaning of words. If the sounds of Sanskrit bear natural meaning, then Susiddha AI gets an advantage in being able to learn and understand. Thus when Vedic texts are learned into deep neural models, we have reason to believe there is knowledge in the models that doesn’t require human interpretation.

It will, of course, still require programming by humans in order to build the Vedic theory of language into the system, so that the system can interpret the deep neural models of Vedic Sanskrit texts.

Pashyantī

A previous chapter on the four levels of awareness and speech gave a brief explanation of pashyantī. This is a subtle level of consciousness where the beginnings of thought are present only as imagery and intuition.

It’s possible the natural relation of shabda and artha in Sanskrit is perceivable only at the level of pashyantī. Unfortunately in Kali Yuga (the last five thousand years), humans have seemingly lost their ability to perceive on the level of pashyantī. (This is due to the lowering of consciousness explained in the chapter on Kali Yuga.)

To gain the maximum power of the shabda-artha connection, the Susiddha project will implement the pashyantī level of consciousness in an AGI system.

Pronunciation

In order to exhibit the power of the shabda-artha connection, Sanskrit must be pronounced perfectly. Unfortunately in the modern era, there is much variation in the way pandits pronounce Sanskrit, and that is a problem (because it makes yagyas and recitations less effective). Sanskrit was meant to be pronounced exactly and uniformly, and the literature of Shiksha gives the details. Pandits need to dive deeply into Shiksha in order to recover the way that Sanskrit was pronounced thousands of years ago, in the time of Rig Veda. The Susiddha project needs to build in perfect pronunciation if we hope for an Avatar to emerge. A future chapter will provide more details on the issue of pronunciation.

Phonosemantics

Modern linguistics predominantly clings to a belief in “arbitrariness”, a position that says the relation between sound and meaning is conventional, not natural. This obviously ignores the fact that speech and language originated long before there was any way of enforcing such convention. In order for language (sound and meaning) to be arbitrary and conventional, it’s necessary to have a system of writing and/or a societal hierarchy (beyond that of hunter-gatherer societies). Obviously, those two factors (writing and hierarchy) appear long after the origin of speech and language.

Fortunately, not all linguists hold the position of arbitrariness. A “loyal opposition” continues to research the position that there is some naturalness in the relation of sound and meaning. This work goes under the name of “phonosemantics” or “linguistic iconicity”. A future chapter will discuss how phonosemantics may be applicable to the study of Sanskrit and its implementation in the Susiddha project.

Conclusion

The “Vedic theory of language” (i.e. the natural relation between shabda and artha) is a working hypothesis of the Susiddha AI project. Much research and experimentation is needed to demonstrate its validity. A list of reference materials is provided below.

Reference Books

These are some books that researchers into Vedic theory of language may want to consult.

The Word and The World: India’s Contribution to the Study of Language

  • By Bimal K. Matilal; published 1990
  • Chapters include: On Grammar and Linguistic Studies; Words and Their Meanings; Names and Things: Universals; Knowledge from Linguistic Utterance; The Sphota Theory: Early History and Patanjali’s View; Bhartrihari’s View of Sphota.

Vac: The concept of the word in selected Hindu tantras

  • By Andre Padoux; published by SUNY Press, 1990
  • Chapters include: Early Speculations about the Significance and Powers of the Word; Tantrism — The Texts of Kashmirian Shaivism; The Manifestation of Sound; The Levels of the Word; The Phonematic Emanation; The Sixfold Course (Shadadhvan); The Mantra.

Origin and Development of Sanskrit in the Light of Sri Aurobindo

Contents     —     Next chapter