Skip to content

The extralinguistic origin of Pāṇini’s syntactic categories and their linguistic accuracy


– Sergiu Al-George (1922-1981) –

The term karaka – which occurs first in Pāṇini’s Aṣṭādhyāyī – is doubtlessly one of the profoundest concepts of the great Acarya. The validity of this concept within the analysis of language facts entails the validity of the whole Pāṇinian system as well. However, the studies that have been devoted to it have certainly provided insufficient commentary, and a reconsideration of this central term might prove particularly illuminating at a time when contemporary linguistics has embarked on the road of syntactic analyses.

Of the many problems that this concept raises, the present article will restrict itself to one aspect only, namely that of its linguistic value as compared to its extralinguistic origin.

In a reference to an assertion of mine – that the karaka system has an extralinguistic starting point[1] –, G. Cardona assimilates it to Rosane Rocher’s assertion on the same question, namely “that these definitions have nothing to do with grammar”, concluding that both would imply the opinion that “Pāṇini was at best a confused linguist, at worst an exemplary of that much misused «mystic» India”[2]. The American scholar overlooks the fact that by “extralinguistic” I meant the Vedic ritual categories, the speculations on the symbolic experience of the archaic Indian world.

The problem raised by the above-mentioned assertion thus pertains to the compatibility between the categories of intelligibility of the archaic world and those of Pāṇini’s linguistic system.

We consider the filiation between grammatical science and ritual science in India to be a well-established fact. This filiation is unquestionable, be it for the single reason that the Indian tradition itself considers grammar as an aṅga, “part” or “auxiliary” of ritual science. Moreover, B. Liebich’s[3] and L. Renou’s[4] works have minutely demonstrated the ritual antecedence of various Indian grammatical terms. Renou’s article “Les connexions entre le rituel et la grammaire en sanskrit” is perhaps one of the most fascinating pieces of writing among the legacy of the late French scholar. Dealing with the designation of cases by numerical indexes as well as with the karakas, the great Renou confidently asserts that both belong to the ritual milieu. We cannot help quoting him on the karakas:

Karman… appartient aux milieux rituels où le mot, depuis le RV, désignait l’acte par excellence, c’est-1-dire le rite. De même kartṛ qui dans les deux domaines est l’agent[5].

Yet, neither Liebich nor Renou discuss or even raise the problem whether their ritual origin is detrimental to the linguistic accuracy of the respective concepts. As for us, we have shown that the karaka system is not arbitrary, that it converges with highly modern linguistic conceptions, thus representing a possibility of sentence analysis which could be confronted with the traditional Western subject/predicate categories[6].

The thesis of the ritual, hence extralinguistic origin of Indian grammatical categories is only apparently incompatible with their linguistic accuracy. If, however, the overall thrust of the structural conceptions were ignored and an incomplete appreciation were made, based on only one of the theses, i. e. that language, being a structure, can be studied in itself, without any reference to extralinguistic factors, then the ritualistic origin of Pāṇini’s syntactic categories would be incompatible with their linguistic accurancy.

However, ritual categories are not alien to language. F. de Saussure, the father of structuralism, showed that rite – being a symbolic act – belongs to a semantic system[7]. After him, L. Hjelmslev pointed out the need for a larger conception of language, which should include as well structures other than those of sound and meaning, and the fact that philology

should form a subdivision of a larger science of sign systems in general, which would be the real theory of language in the structural sense of the word[8].

Lévi-Strauss[9] showed that not only rite, but myth as well is a semantic system, and devised a method of myth analysis starting from the categories of language furnished by structural analysis. By confronting mythical and linguistic structures, the great French ethnologist took the same route as the world’s first linguists, the Indian linguists, although in the opposite direction.

The emergence of Indian linguistics was thus possible because India, as no other culture, had a ritual science which singled out the categories of symbolic thinking. The ritual origin of a linguistic category is not a drawback, but a great advantage: the analysis of the symbolic act accedes to the most general categories of expression, beyond the cultural conditioning of the language, and uncovers the relations and structures of the abyssal zones of the consciousness[10]. In using categories borrowed from the analysis of ritual to analyze verbal expression, ancient Indian linguistics took over from a universal semiotics a system of categories which it applied to a particular, historically and culturally determined semiotic system, that of the Sanskrit language.

As we have shown elsewhere[11], this explains why many modern structuralist concepts and descriptive techniques were anticipated in the texts of Indian grammarians.

It is for the same reason that Indian traditional linguistics is better appreciated and understood nowadays than a century ago, although the discovery of Indian grammatical concepts at the time determined the birth of scientific linguistics in Europe. Indian linguistic analysis has been structural from its early emergence, in the analysis of the categories of symbolic expression.

European grammatical science, since the Greco-Latin antiquity until the appearance of structuralism, had not succeeded in acceding to the universal categories of linguistic expression, because it had constantly concerned itself with its outer form, which is historically and geographically determined. To all this, one should add the superimposition of various logical and ontological categories alien to language. This resulted in a vicious circle which not even the 19th century science of language – in spite of its having non‑Indo‑European grammatical structures at its disposal as well – could break. There was only one way to break this deadlock: to go beyond the phenomenal appearance, in order to capture the inner form of linguistic expression. This was only possible by introducing a functional criterion: “La forme ne peut être reconnue et définie qu’en se plaçant sur le terrain de la fonction”, Hjelmslev[12] had argued. Structuralism came into existence in the West when this functional criterion was introduced and laid at the basis of all of its further concepts.

de Saussure’s assertion that language is form, not substance, meant a break from the Aristotelian substantialist tradition. Indian ritualist thinking, as any archaic thinking, is not dominated by a substantialist view, being, on the contrary, an eminently formal one. According to the archaic mentality, any object can be substituted for another inasmuch as they both fulfil the same function within an analogous structure; things are not considered from the point of view of their substance only, but, particularly, from that of the function they fulfil, of the system of relations and participations they are included in. Phenomenal reality is understood and validated by its analogy with ultraphenomenal reality, which is more real than the real. Structuralists conceive the abstract form of language in the same way in which the background of mythical reality was conceived: in both cases, the abstract form has a totalizing character, covering all the possibilities of the outer form, a system in which the oppositions, variants and distributions of the data sensibly organized are inventoried and, hence, co-present. Compared to the abstract form, the outer one appears impoverished, being merely a manifestation of the former. The relation between the abstract reality manifesting itself and the manifested sensible one is a whole/part relation. The concept of manifestation in structural linguistics can hardly be distinguished from the same concept in the archaic thinking, where it sums up the whole ontology.

In Indian archaic thought, the supreme mythic reality is the sacrifice, the symbolic act thus becoming the paradigmatic reality lying behind every object, act or process, be it the creation of the world or the human psycho-physiological experience.

In India, the semantic nature of the supreme mythical reality is confirmed by the presence of a consistent doctrine of the Logos (aksara, vac, śabdabrahman) originating in the Veda, developed in the speculations of the Upanisads and Darśanas, and further refined in the later speculation of the Tantras. The philosophy of Indian grammar is closely linked to this “semantic emanationism”.

In his syntactic analysis of the sentence (vakya), Pāṇini goes beyond the surface forms of the cases, to an abstract and much more comprehensive form: the karakas. The chain units of the discourse, the verbal and nominal inflexional suffixes, as well as the non-inflexional ones, are referred to their functions, represented by karakas. Consequently, Pāṇini was the first to speak about the idea of function.

The set of the karakas represents an analysis of the act, a set of the categories according to which the representation of an act or a process can be achieved. In Indian traditional linguistics, the agent, the object, the instrument, the location, the bestowal and the ablation are a semantic segmentation of the idea of action (kriya), the sense of any sentence. Running through the set of categories resulting from the analysis of the sacrifice, the most expanded symbolic act, as it was made by H. Hubert and M. Mauss[13], mainly on the basis of the Vedic sacrifice, it is not difficult at all to establish a correspondence with the set of the karakas: the sacrificer is the agent (kartṛ), the victim is the object (karman), the sacrifice area and the propitious moment are the location (adhikarana), while the ablation (apadana) and the bestowal (sampradana) correspond to the magic transfer itself. As for the institutor of the sacrifice (yajamana), he would correspond to hetu.

For the correspondence between the karakas and the sacrifice categories as it was understood by the Indians themselves, we find it very significant that Abhinavagupta’s Tantrasara[14] mentions the identification of the six karakas with the Lord (īśvara) by means of the correspondence between the former and the elements of the sacrifice, which is closely akin to the above-mentioned correspondences. These resemblances are a development of Bhartṛhari’s speculation, who assimilates the karakas and śakti, “the power (of brahman)”[15]. If Pāṇini had emptied his technical terms of their traditional content – as Cardona maintains[16] –, the post-Pāṇinians would not have considered the respective terms in a ritual context.

We can consider that the categories of the Vedic sacrifice represent the most expanded form by which any act or process can be represented. In analysing verbal expression by means of these ultimate categories of representation, old Indian linguistics laid the foundations of a representational syntax in the sense in which it was defined by M. Sandmann. Representational syntax is according to him a picture of reality completely different from the cognitional one, based on the subject/predicate relation[17].

The syntactic controversy over the value of the subject/predicate categories is not yet settled. Through most of its representatives, structuralism has rightly rejected these categories, which linguistics has been questioning since its emergence as a science. N. Chomsky, however, reviving the Aristotelian tradition of Port-Royal, has again made it actual, at the same time denying any grammatical value of the representational categories[18].

We would like to emphasize one aspect which brings out the validity of Pāṇini’s syntax as against that of the Port-Royal school. When Indian culture irradiated in non‑Indo‑European linguistic areas, its missionaries were faced with the task of elaborating grammars, just as the missionaries of the European culture were to be, many centuries later. The Indian missionaries were equipped with Pāṇini’s categories, the Westerners with those of the Port‑Royal grammar. Both grammatical systems pertained to Indo-European linguistic structures. Although Pāṇini’s system was much older (and based on the description of one language only) than the Port-Royalist one (which was based on several ancient and modern languages), it seems that Pāṇini’s system was more efficient than the Port-Royalist one. This at least in the case of the Tibetan language which is representative for a whole group of languages.

The agent of the active verb, the patient of the passive verb, as well as the person or the thing participating in a verbal idea expressed by an intransitive or state verb, which can be rendered in Indo-European languages by a single grammatical form, that of the nominative – thus creating the category of the grammatical subject, – are dissociated in Tibetan. Here, the process is no more dominated by the idea of person, being conceived in itself, impersonally; accordingly, as it has been repeatedly remarked, there is no “subject”, in the sense of Indo‑European languages, to be opposed to a “predicate”[19]. The agent of a transitive verb (or of one conceived transitively) is marked by the suffix kyis / gyis / gis / ‘is, the same as that of the instrument, whereas the name of a person or object associated with an intransitive or state verb is suffixless and represents – according to H. Maspero – an “object”, rather than a “subject”[20]. The national Tibetan grammar, elaborated with the help of Indian pandits, adopted with slight modifications the karaka system. In the grammar ascribed to Thonmi Sambhota, the suffix kyis, etc. has the role of expressing the “agent” function, byed-pa-po (the translation of the sk. kartṛ); as for the suffixless form, it represents “the thing in itself, only”, (ṅo-bo-tsam), being as such deprived of any grammatical function. The qualification “the thing in itself, only” renders the Sanskrit term arthamatra, which is used by the Buddhist grammarian Candragomin (II, 1, 93), in accordance with Pāṇini (II, 4, 36), to indicate that the nominative suffix expresses a functive, and not a function.

The karaka system, without subject/predicate equivalents, thus offers a more adequate framework for the description of syntactical relations in Tibetan or, in J. Bacot’s words,

le cadre de la grammaire sanscrite a moins faussé le tibétain que ne l’a fait le cadre des grammaires européennes[21].

The framework of Sanskrit grammar was, as a matter of fact, the universal frame of symbolic expression, a frame at once immanent to and transcending speech phenomena.

In conclusion, we would like to emphasize that, in order to fully grasp the significance of Pāṇini’s amazing discovery of the universal categories of language in such a remote epoch, and on the basis of one language only, we must see him in the context of the Indian world. The opposition linguistic/extralinguistic is not the same in India as in Europe, because in India linguistic structures are not alien to religious, philosophical, even logical structures. Indian culture has a unitary character given by the organic continuity from the earliest speculation – the mythical-ritualistic one – to the other speculations originating in it. The science of language descends directly from the science of symbolic expression, and the ontology and logic found in the Darśanas are largely structured by it, although later on the influences were to be mutual.

In spite of the surprising and consistent anticipations of our modern structuralist linguistics in Pāṇini’s thought, it would be a big mistake to believe that his conceptions and descriptive methods, although an expression of lay (laukika) thinking, are those of a linguist conceived after our modern image, i.e. not only a structuralist or theorist, but also a positivist, severed from the co-ordinates of traditional Indian thought.

 [1]. See supra, p. 78.

 [2]. G. Cardona, “Pāṇini’s syntactic categories”, JOIB 16 (1967), 3, p. 214.

 [3]. B. Liebich, “Zur Einführung in die indische einheimische Sprachwissenschaft. II. Historische Einführung und Dhatupatha”, SHAW, Phil-hist. Kl., 1919, no 15.

 [4]. L. Renou, “Les connexions entre le rituel et la grammaire en sanskrit”, JA 233 (1941-1942), pp. 105‑165. When I was writing my study “Le sujet grammatical chez Pāṇini”, I had no access to the respective article, as I mentioned p. 78, n. 41, and thus advanced a common thesis without knowing anything about his article, except its suggestive title.

 [5]. L. Renou, op. cit., pp. 156-157.

 [6]. See supra, p. 79.

 [7]. F. de Saussure, Cours de linguistique générale, Paris 51960, p. 33.

 [8]. L. Hjelmslev, “Structural analysis of language”, Studia linguistica 1 (1947), 2, p. 76.

 [9]. C. Lévi-Strauss, Anthropologie structurale, Paris 1958, p. 248.

 [10]. According to C. Lévi-Strauss, “l’ensemble de ces structures formerait ce que nous appelons l’inconscient” (op. cit., p. 224).

 [11]. See supra, pp. 85-88.

 [12]. L. Hjelmslev, Communication du 30 septembre 1937 sur les rapports entre forme et substance linguistiques, BCLC 4 (1937-1938), pp. 3-4.

 [13]. H. Hubert et M. Mauss, “Essai sur la nature et la fonction du sacrifice”, Mélanges d’histoire des religions, Paris 21929, pp. 1-130.

 [14]. Italian transl. by R. Gnoli, Essenza dei Tantra, Torino 1960, pp. 214-215.

 [15]. VP III, 7 (sadhanao), 1-2. See D. S. Ruegg, Contributions 1 l’histoire de la philosophie linguistique indienne, Paris 1959, p. 8, n. 1; P. Ch. Chakravarti, The philosophy of Sanskrit grammar, Calcutta 1930, p. 221.

 [16]. G. Cardona, “Pāṇini’s syntactic categories” p. 212: “What Pāṇini has done, then, is to take terms like sampradana and empty them of their traditional meanings, in order to use them for his grammar”.

 [17]. M. Sandmann, Subject and predicate. A contribution to the theory of syntax, Edinburgh 1954, p. 142:

In a sentence such as The king arrives in London it is of little interest to say the king represents a S[ubject] and the other words together a P[redicate], but it is of paramount importance to recognise an actor (the king), an action (arrives) and the relation between the acting actor and the place (in = relation, London = place).

 [18]. N. Chomsky, Syntactic structures, The Hague 1957, p. 100.

 [19]. H. Maspero, in A. Meillet et M. Cohen (eds) Les langues du monde I, Paris 1952, p. 539; J. Bacot, Grammaire du tibétain littéraire I, Paris 1946, pp. 49-50.

 [20]. H. Maspero, ibid.

 [21]. J. Bacot, Une grammaire tibétaine du tibétain classique. Les ślokas grammaticaux de Thonmi Sambhota avec leurs commentaires, Paris 1928, p. 4.