Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, LINGUISTICS ( (c) Oxford University Press USA, 2016. All Rights Reserved. Personal use only; commercial use is strictly prohibited. Please see applicable Privacy Policy and Legal Notice (for details see Privacy Policy).

date: 24 February 2018

Linguistics in Premodern India

Summary and Keywords

Indian linguistic thought begins around the 8th–6th centuries bce with the composition of Padapāṭhas (word-for-word recitation of Vedic texts where phonological rules generally are not applied). It took various forms over these 26 centuries and involved different languages (Ancient, Middle, and Modern Indo-Aryan as well as Dravidian languages).

The greater part of documented thought is related to Sanskrit (Ancient Indo-Aryan). Very early, the oral transmission of sacred texts—the Vedas, composed in Vedic Sanskrit—made it necessary to develop techniques based on a subtle analysis of language. The Vedas also—but presumably later—gave birth to bodies of knowledge dealing with language, which are traditionally called Vedāṅgas: phonetics (śikṣā), metrics (chandas), grammar (vyākaraṇa), and semantic explanation (nirvacana, nirukta). Later on, Vedic exegesis (mīmāṃsā), new dialectics (navya-nyāya), lexicography, and poetics (alaṃkāra) also contributed to linguistic thought.

Though languages other than Sanskrit were described in premodern India, the grammatical description of Sanskrit—given in Sanskrit—dominated and influenced them more or less strongly. Sanskrit grammar (vyākaraṇa) has a long history marked by several major steps (Padapāṭha versions of Vedic texts, Aṣṭādhyāyī of Pāṇini, Mahābhāṣya of Patañjali, Bhartṛhari’s works, Siddhāntakaumudī of Bhaṭṭoji Dīkṣita, Nāgeśa’s works), and the main topics it addresses (minimal meaning-bearer units, classes of words, relation between word and meaning/referent, the primary meaning/referent of nouns) are still central issues for contemporary linguistics.

Keywords: Ancient Indo-Aryan, Dravidian languages, Middle Indo-Aryan, Modern Indo-Aryan, Padapāṭha, premodern India, Sanskrit, Vedāṅgas, vyākaraṇa

1. Overview of Linguistic Thought in Premodern India

1.1 Vedic Literature and Padapāṭhas

The most ancient Indian texts which have come down to us are the Vedas (“Knowledge”),1 composed in Vedic Sanskrit (Old Indo-Aryan; for a general presentation of the stages of Indo-Aryan, see Cardona & Jain, 2007, pp. 6–18). They constitute the foundational corpus of the Vedic religion, the most ancient form of Brahmanism, which is the starting point for numerous doctrines of premodern India.2 These texts, the most ancient of which go back to the second half of the second millennium bce, are different in form as well as in content and share the feature that they are based on a “revelation” (śruti, literally “hearing”). The Vedas include the saṃhitās, which are versified “collections” gathering together hymns, prayers, and ritual incantations, as well as commentaries on these saṃhitās. There are four kinds of saṃhitās: (1) stanzas (ṛc) which make up the Ṛksaṃhitā or Ṛgveda (ṚV hereafter), (2) ritual incantations (yajus) gathered together (with or without commentary) in the Yajuḥsaṃhitā or Yajurveda, (3) songs (sāman) of the Sāmasaṃhitā or Sāmaveda, and (4) spells (atharvan), which constitute the Atharvasaṃhitā or Atharvaveda. The Vedic—and also Brahmanical—tradition staunchly revolves around these four Vedas (cf. Renou, 1947, p. 12).

This Vedic literature was composed and passed down orally.3 This practice left a deep mark on Indian culture, its relation to texts, and above all its beliefs regarding language. Very early, the oral transmission of holy texts made it necessary to develop techniques based on a subtle analysis of language. One of the most ancient techniques involved the accurate description of phonological rules that apply at word junctures (sandhis) to move from continuous recitation (saṃhitā-pāṭha) of a saṃhitā, where rules are applied, to word-for-word recitation (pada-pāṭha), where rules are not applied.4 This implies, on the one hand, an advanced knowledge of phonetics and phonology as well as, on the other hand, sophisticated thinking about how to parse continuous speech.

Consequently (and because no more ancient documents have come down to us), it is traditionally considered that linguistic thought in premodern India began with the composition of the Padapāṭhas, around the 8th–6th centuries bce. The Padapāṭhas’ ultimate goal was to preserve Vedic texts by drawing attention to those parts of the texts which were those most subjected to modifications, such as the word-final position (see Jha, 1992, p. 14). They consist of versions of Vedic texts where some sandhis are dissolved and replaced by pauses; as a consequence, some linguistic units—the padas—are isolated. The Padapāṭhas represent the earliest available grammatical commentaries of the Vedic saṃhitās (Abhyankar & Devasthali, 1978, p. xviii, among others) and the pada-units they identify are the most ancient linguistic category (cf. §2.1.1 for more details).

1.2 Vedāṅgas

The Vedas also gave birth to bodies of knowledge dealing with language, which are traditionally called Vedāṅgas (“Limbs [for preserving the body of] the Vedas”). They differ from the Vedas in that they are based on the “[tradition relying on] memorization” (smṛti). The four Vedāṅgas related to language are traditionally ordered as follows: phonetics (śikṣā), metrics (chandas), grammar (vyākaraṇa), and semantic explanation (nirvacana, nirukta). According to Scharfe (1977, p. 82), the delimitation of these disciplines occurred quite early: “Towards the end of the Vedic period there were thus three branches of linguistic study: phonetics (śikṣā), etymology (nirukta) and grammar (vyākaraṇa); but their oldest systematical works have not survived the hazards of oral tradition.” And in fact one finds in Vedāṅgas’ literature very old notions such as ‘meter,’ ‘metrical feet,’ ‘word,’ and ‘syllable,’ which demonstrate an exceptionally ancient thought regarding language and linguistic units (cf. Deshpande, 2000, pp. 137–138).

Phonetics (śikṣā) aimed at preserving the correct pronunciation and recitation of Vedic texts, as well as at recomposing the continuous version of saṃhitās by applying phonological rules (this is done in works called Prātiśākhyas—cf. footnote 3—which, though they include materials that fall under śikṣā, are separate from śikṣā texts). Ancient Indian phoneticians accurately described the sounds (varṇa, i.e. vowels, svara, and consonants, vyañjana) according to (1) their accent (svara), (2) their duration (kāla), (3) their point of articulation (sthāna), from the glottis (kaṇṭha) up to the lips (oṣṭha), (4) their manner of articulation (prayatna), including different degrees of contact and different degrees of openness, and (5) their phonation (anupradāna), that is to say the quality of air passing through the glottal aperture. Moreover, ancient Indian phoneticians carefully accounted for features of junction (sandhi), as well as features of syllable-structure (length, quantity, tone). For more details, see Allen (1953), Pinault (1989, pp. 304–313), and Deshpande (2000).

Metrics (chandas) gave instructions on the metrical structures of Vedic texts, that is to say set the rules of prosody and described different types of meters (seven major meters and various complex structures).

Semantic explanation (nirvacana, nirukta) consisted of an elucidation of the meaning of difficult words, the goal being to attempt to find out how a word comes to mean what it does (cf. Kahrs, 1998, p. xiv). Only one work belonging to the Nirvacana tradition (which presents itself as a complement to grammar) has come down to us, namely the Nirukta (N)—and its commentaries—which is attributed to Yāska and which was probably composed between the 5th and the 2nd century bce. Yāska’s work is made up of, first of all, the commentary on vocabulary contained in lists of Vedic words known as Nighaṇṭu, lists which were very probably compiled from the padapāṭha versions of Vedic texts. This vocabulary constitutes the oldest Sanskrit lexicon (cf. §1.3). Yāska introduces his work by expounding some grammatical principles. He enumerates and briefly describes four classes of words (pada): nouns (nāman), verbs (ākhyāta), prepositions (upasarga), and particles (nipāta); he explains that nouns have substance (sattva) as their principal meaning, while verbs have becoming (bhāva) as theirs (several modifications of the “becoming” are listed); and he mentions the debate related to the verbal origin of nouns.

Grammar (vyākaraṇa, action noun coming from the preverbed root vy-ā-kṛ-, traditionally interpreted as denoting a separation or a discrimination process of constituents, but it can also be taken as referring to a creation process, which generates, in a diversified way, the linguistic units; see Thieme, 1982–1983, pp. 11, 23–34, and Cardona, 1997, pp. 565–571) imparted knowledge, by describing their formation, of (correct) speech forms (see Cardona, 1997, pp. 543–544). Several schools of Sanskrit grammar developed in India, some of them known only from quotations (see Scharfe, 1977, pp. 124–126). Their exact number is still unknown (it fluctuates between 3 and 20, see Raghavan, 1974, pp. 272, 276), not only because (Indian as well as Western) scholars did not and do not always share a common conception of what is a school of grammar, but also because, for several reasons, much information has been lost over time (see Bronkhorst, 2014).

Schools of Sanskrit grammar can be divided into two categories, the Pāṇinian school and the non-Pāninian schools. Sanskrit grammatical thought was deeply influenced by the Aṣṭādhyāyī of Pāṇini, the earliest complete surviving Sanskrit grammar, which dates from the 5th century bce. During the following centuries, some grammarians decided to follow on the Pāṇinian work, creating in this way the Pāṇinian school, while others more or less moved away from this work (or were considered by the Pāṇinian grammarians as having moved away from it) and the techniques it implies.

The Pāṇinian school is indisputably the most ancient and the longest school of grammar in India (as well as in the world): it begins with Pāṇini’s Aṣṭādhyāyī, the founding treatise of the school (cf. §2.1.2), and is traditionally considered to end with with Nāgeśa (died in 1755 in Benares), a prolific grammarian whose erudite works marked the last renewal of the Pāṇinian school (cf. §2.1.6). Several great scholars figured among this school (cf. §2.1.3 to 2.1.6), and a substantial amount of grammatical literature is linked to it (different kinds of commentaries and subcommentaries as well as independent treatises). The main reasons which explain why some grammarians moved away from Pāṇini’s work are (1) the need for a more practical and pedagogical grammar and (2) a divergence of opinion regarding theoretical issues (such as the description of the formation of vṛttis; on this point, see Cardona, 2008).

The Aṣṭādhyāyī consists in a derivational model of a highly technical nature, made up of approximately 4,000 rules (sūtra) and including numerous metalinguistic rules—metarules (paribhāṣā), rules related to technical terms (saṃjñā-sūtra), and headings (adhikāra). That is how an arrangement of grammatical rules by topic (cf. §2.1.5) emerged, as well as a reduction (or even a suppression) of the metalinguistic tools and the removal of rules teaching purely Vedic forms. As far as is known, grammars arranged by topic first appeared in the Buddhist and Jaina spheres (that is to say, outside the Pāṇinian school, which is of Brahmanical or Hindu tradition), after Sanskrit versions of their canonical texts were adopted.

The earliest attempt to organize grammatical rules by topic is Śarvavarman’s Kātantra (“Small manual”), which probably dates from the 4th century. This “practical” grammar—which is perhaps a recast of a more ancient one, the Kaumāralāta (see Scharfe, 1977, p. 162)—influenced several later grammars, including Kaccāyana’s Pāli (Middle Indo-Aryan) grammar and the Siddhahaimacandra of the Jaina Hemacandra, and gave rise to a rich secondary literature (commentaries, supplements, etc.). The Kātantra school, which maintained its existence both in and outside India for centuries, retained its popularity in Kashmir, Nepal, and parts of Bengal (Scharfe, 1977, p. 163) from the revival of the Pāṇinian school in the early 17th century onwards (cf. §2.1.5).

Another famous Sanskrit grammar arranged by topic, the Cāndravyākaraṇa (“Candra’s grammar”), composed by the Buddhist scholar Candragomin, who is considered to have lived during the 5th century, is the basis of another grammatical school. The Cāndravyākaraṇa is the first great recasting of Pāṇini’s grammar (it also includes some of the additions suggested by Patañjali and Kātyāyana, the two first known commentators of Pāṇini’s work, cf. §2.1.3) as well as the great grammar of Buddhists. As such, it was widely circulated and was mainly preserved in places where Buddhism spread (Nepal, Tibet, Burma, Sri Lanka). For more details, see Belvalkar (1915, pp. 57–62) and Scharfe (1977, pp. 164–167).

Within the Jaina community, three grammars gave birth to three grammatical schools: the Jainendravyākaraṇa of Devanandin (5th–7th centuries?), which is the most Pāṇinian of the non-Pāṇinian grammars (it preserves, among other, Pāṇini’s generative scheme; see Belvalkar, 1915, pp. 62–68; Scharfe, 1977, pp. 168–169), the Śabdānuśāsana (or Śākaṭāyanavyākaraṇa) of Śākaṭāyana, a grammar of the 9th century arranged by topic (see Belvalkar, 1915, pp. 68–73; Scharfe, 1977, p. 169) and the Siddhahemacandra (or Śabdānuśāsana) of Hemacandra (11th–12th centuries), another grammar arranged by topic, but more practical, which covers Classical Sanskrit and Prakrits (Middle Indo-Aryan; see Balbir, 2000, notice 4336). Non-Pāṇinian grammars were also composed outside the Buddhist and Jaina communities, some of them sponsored by princes, such as the Sarasvatīkaṇṭhābharaṇa of Bhoja, a grammar arranged by topic written in the 11th century (see Scharfe, 1977, pp. 187–190).

1.3 Other Sanskrit “Language Sciences”

Thought about language occupied a central position in premodern India, in particular within the Brahmanical sphere, where traditional scholars were first trained in grammar (Vyākaraṇa, cf. §2), Vedic exegesis (Mīmāṃsā), and dialectics (Nyāya). Up to the end of the first millennium, grammar and Vedic exegesis dominated the Brahmanical thinking about language. Around the beginning of the second millennium, along with the emergence of the Navya-Nyāya (new dialectics), the successor to the Vaiśeṣika (systematics) and Nyāya classical schools,5 the dialecticians developed a theory of verbal cognition (śābdabodha) which competed with grammatical and exegetical theories and influenced them (Gerschheimer, 1996, I, p. 3).

The ultimate aim of Vedic exegesis (Mīmāṃsā) is to guarantee the correctness of ritual practice. In concrete terms, this implies the explanation of the meaning of Vedic utterances (in particular, injunctive sentences) and then the resolution of numerous interpretative problems in Vedic texts. This task led Sanskrit exegetes to deal with, among other issues, the words and sentences expressing injunctions, the nature of the word and the sentence in general, the nature of their respective meanings (see McCrea, 2000 and David, 2017 for details, and for general surveys of the discipline Jha, 1942 and Verpoorten, 1987). McCrea (2000, p. 429) writes: “While the primary focus of attention in Mīmāṃsā is on Vedic texts, the hermeneutical principles developed in the course of analyzing these texts are formulated so as to be applicable to language in general, and are recognised as such by scholars in other fields. For this reason, Mīmāṃsā is frequently designated as vākya-śāstra (‘the science of sentences’).”

The new dialectics (Navya-Nyāya) is firstly a theory of the means of acquiring valid knowledge (pramāṇa), among which speech (śabda) plays a prominent role (cf. Gerschheimer, 1996 I, p. ix). Within this frame, Sanskrit dialecticians addressed such topics as the notions of speech and word, the production of verbal knowledge, the meaning of a word, and the nature of the relation between a word and its meaning, as well as the problems caused by tropes and corrupted forms, the acquisition of the relation between a word and its meaning, the sentence, and the syntactic link. For a general presentation of this discipline (Vaiśeṣika, early Nyāya school, and Navya-Nyāya school), see Matilal (1977); for an overview of the naiyāyika theory of speech, see Gerschheimer (1996, I, pp. 43–99).

Lexicography as well as poetics (Alaṃkāra) may also be considered part of the Sanskrit language sciences.

Classical Sanskrit lexicography (Kośa)6 played an important role in Indian scholarship, especially poetry; the aim of classical lexica, which were learned by heart, was to help poets in composition, where synonyms of varying syllable structure are required to satisfy metrical constraints. Two main kinds of lexicon (kośa) were composed: synonymic (ekārtha, samānārtha), where words are classified according to subject (e.g., words relative to heaven, sky, time, thought, sound, etc.), and homonymic (anekārtha, nānārtha), which list words having more than one meaning (for more details regarding lexicography, see Vogel, 1979 and Patkar, 1981; for more details regarding the opposition between synonyms and homonyms or polysemous words, see Aussant, 2014a). Note that from the 11th century on, bilingual and multilingual dictionaries were composed (Sanskrit-Kannada, Sanskrit-Kawi, and Sanskrit-Marathi-Telugu-Persian, among others). Lexica in Prākrit as well as in Pāli (Middle Indo-Aryan languages) were also composed.

Sanskrit poetics (Alaṃkāra) is an erudite discipline that accompanied Sanskrit literary production (mainly kāvya, refined poetry) for nearly two millennia. It addressed, among other questions, the following issues: analysis of the formal, logical, semantic, and pragmatic aspects of simile and other tropes; word classes; word meanings (denotation, metaphor, suggestion); the meanings of sentences, passages, and whole literary works; and language registers. For a general study of poetics, see De (1960), Gerow (1971), and Bronner (2012).

1.4 “Extended” Sanskrit Grammar

Sanskrit grammar, like the grammars of Greek, Latin, and Arabic, has been transferred or “extended” (see Auroux, 1992, pp. 11–64; Auroux, 1994) to languages other than the one it was originally designed to describe. Indeed, some grammatical descriptions elaborated for Sanskrit were used (to varying degrees and in different ways) for the description of various languages, mainly in India (Middle and Modern Indo-Aryan languages as well as Dravidian languages and Persian) but also in Tibet, Southeast Asia (Burmese, Old Javanese), and Central Asia (Buryat), as well as in the Americas (Algonquian languages). If grammatical descriptions elaborated for Sanskrit played a major role in the history of the grammaticization of different languages, it is undoubtedly because they associated an incredibly powerful and sophisticated grammatical model (or analyses, cf. §2.1.2) with a highly prestigious language—first, the language of holy texts, the Vedas; then the language of a technical literature; and then, in the first millennium, the cosmopolitan (literary as well as political) language of an ever-increasing area (see on that point Pollock, 2007). Indeed, it seems that the transfer of Sanskrit grammatical descriptions to languages other than Sanskrit is always linked, in one way or in another, to the power of Sanskrit grammatical descriptions or to the prestige of the language, or both. That being said, the “extension phenomenon” of Sanskrit grammatical descriptions has been achieved in a wide variety of ways.

One of its manifestations is that Sanskrit grammar serves as a “source grammar” for “indirect grammars” (or “transfer grammars” according to Harris, 1954, p. 260). For instance, the grammars of Prakrits (Middle Indo-Aryan languages, cf. §1.5) have been conceived as appendices to Sanskrit grammar: (1) the general structure of their rules is “instead of x (Sanskrit form), one has y (Prakrit form)”; (2) for any linguistic fact which has not been described in the grammar, the user is taken back to the Sanskrit norm (e.g., the last aphorism of the Prākṛtaprakāśa states śeṣaḥ saṃskṛtāt “the rest [is to be inferred] from Sanskrit”; for more details, see Nitti-Dolci, 1938 and Balbir, 2000, notice 4331, notice 4335, notice 4340). Such a description of Prakrits, which is clearly contrastive, assumes that these languages were considered as vikṛtis, “modifications” (probably, at some time, as distortions), of Sanskrit. In the case of Prakrits, it corresponds to a historical development, but that is not the case for other language descriptions. The Pārasīprakāśa of Kṛṣṇadāsa, for instance, which was written at the request of Emperor Akbar (ruled 1556–1605), describes Persian forms from the Sanskrit norm, though neither language is derived from the other (see Scharfe, 1977, p. 196). Another example is the Līlātilakam, a 14th-century poetical manual which devotes two chapters to the grammatical description of Maṇipravāḷam, the medieval literary language of Kerala, which is defined as a mixture of the Kēraḷabhāṣā (the medieval form of Malayāḷam, a Dravidian language) and Sanskrit (cf. §1.6). Though the general organization of the Līlātilakam is deeply different from Prakrit grammars, one finds occasional mention of counterexamples which would be generated if some Pāṇinian rules were applied (i.e. “if one would apply such [Pāṇinian] rule, one would obtain such form which is not considered as correct in the Kēraḷabhāṣā”; see Aussant, 2012). Sanskrit grammatical rules are thus never far away, even in some descriptions of Dravidian languages.

Another manifestation of the “extension phenomenon” of Sanskrit grammatical descriptions is the use of technical devices, terminology, and concepts initially created for the description of Sanskrit. An example of the extension of a Sanskrit technical device is provided by two basic treatises of the Tibetan grammatical tradition, the SCP (Sum-cu-pa) and the TKJ (Rtags-kyi-’jug-pa), two short versified texts, the dating (7th–9th centuries?) and authorship of which are problematic (Miller, 2000; Verhagen, 2000a). Verhagen (2000b, p. 211, 2001, pp. 229–230, 233–235) notes that these texts resort—far less extensively—to ellipsis, that is to say to the tacit recurrence (Sanskrit anuvṛtti) of one or more elements of one grammatical rule in subsequent rules. Regarding grammatical technical terms, one may give the example of Tamil grammatical terminology, which, according to Chevillard (2000d, notice 4351), was often modeled on or adapted from Sanskrit terminology, though some terms or expressions seem to indicate the existence of a former indigenous metagrammatical terminology (cf. §1.6). Sanskrit influence was more or less significant depending on the period. It is particularly noticeable in the Vīracōḻiyam, a Tamil grammar of the 11th century written by a Buddhist Tamil grammarian. Chevillard (2000c, p. 201, 2000d, notice 4353) notes, among other things, that the grammar borrows massively from the Sanskrit metalinguistic vocabulary: “canti, upakārakam, tattitam, tātu, kiriyā patam and ālaṅkāram are in effect the adaptations to the phonology of Tamil of the terms sandhi [‘junction’], upakāraka [‘auxiliary’], taddhita [‘secondary suffix’], dhātu [‘root’], kriyāpada [‘verb’] and alaṃkāra [‘[rhetoric] ornament’].” If one looks at the use of the term tattitam, for instance, one observes (thanks to Jean-Luc Chevillard, 2009, pp. 211–212) that it denotes a linguistic fact close to the one described by the taddhita of Sanskrit grammarians. An example of conceptual extension is provided by the adoption of a semantic classification of the grammatical object, initially formulated by Bhartṛhari, a Sanskrit grammarian of the Pāṇinian school of the 5th century (cf. §2.1.4), by Cēnāvaraiyar, a 13th–14th-century Tamil commentator of the Collatikāram (“The book of words”) of the Tolkāppiyam (cf. §1.6). The Sanskrit description (such as found in the Aṣṭādhyāyī of Pāṇini) distinguishes semantic-syntactic categories from the linguistic forms which express them. Bhartṛhari expounds a sevenfold classification of the karman (“object”), from which Cēnāvaraiyar keeps only three categories which end up having a purely semantic value (since one type cannot be distinguished from another on the basis of Tamil morphosyntactic features; on this conceptual borrowing, see Vergiani, 2013). For more details on “extended” Sanskrit grammar, see Aussant (forthcoming a).

This transfer of Sanskrit grammatical descriptions occurred outside Asia as well. Mention should be made of Leonard Bloomfield, who was deeply inspired by the Pāṇinian descriptive model (see notably his 1933 book Language) and used Sanskrit grammatical concepts in his description of Algonquian languages (cf. Rogers, 1987; Emeneau, 1988).

1.5 The Grammatical Descriptions of Middle and Modern Indo-Aryan Languages

There are a diverse range of records of Middle Indo-Aryan languages (see Cardona & Jain, 2007, pp. 12–18): early Middle Indo-Aryan is attested by Aśoka’s inscriptions (mid-3rd century bce) and by Theravāda Buddhist texts in Pāli; later stages are attested by various literary or grammatical works composed in (or describing the) different Prakrits. According to Cardona and Jain (2007, p. 8), there is evidence of the modern stage of Indo-Aryan as early as the 12th century.

The most ancient grammar of a Middle Indo-Aryan language known to us is the Prākṛtaprakāśa of Vararuci, which was probably written between the 3rd and 5th centuries. This work deeply influenced later Prakrit grammarians, those of the Eastern school, that is to say Puruṣottama, Rāmaśarman and Mārkaṇḍeya, who are his direct successors, but also those of the Western (or Southern; see Nitti-Dolci, 1938, p. 179) school, the master of which is generally considered to be Hemacandra. Prakrit grammars mainly differentiate themselves (1) by the dialect(s) they describe (Vararuci’s Prākṛtaprakāśa primarily describes the Mahārāṣṭrī, the Prakrit par excellence, and devotes a very few sūtras to Paiśācī, Māgadhī, and Śaurasenī; Hemancandra’s Śabdānuśāsana—11th–12th centuries—describes Sanskrit, Mahārāṣṭrī, Śaurasenī, Māgadhī, Paiśācī, Cūlikāpaiśācī, and Apabhraṃśa) and (2) by the way they classify them. The Eastern school of Prakrit grammarians is characterized by the following features: (1) the study of the same languages, which are classified as bhāṣā (dialects mainly used in stage plays by high-ranking characters), vibhāṣā (dialects used in stage plays by low-ranking characters; see Grierson, 1918, p. 516), apabhraṃśa (dialects spoken by cultured persons and used by poets), and paiśācika (dialects used in tales according to Nitti-Dolci, 1938, p. 126); (2) a large part of these grammars is devoted to Mahārāṣṭrī, the description of which is the basis for the description of the other Prakrits; and (3) Vararuci’s description of Mahārāṣṭrī is strictly followed. The unity of the Western/Southern school is less easy to grasp (see Nitti-Dolci, 1938, pp. 179–194). The vast majority of Prakrit grammars are written in Sanskrit and are conceived as appendices to Sanskrit grammars, allowing for Prakrit units—which are considered to be modified forms (vikṛti) of Sanskrit—to be formed from Sanskrit (cf. §1.4). Pāli grammars, though subject to the influence of Sanskrit grammars—Pāṇini’s Aṣṭādhyāyī, Śarvavarman’s Kātantra, and Candragomin’s Cāndravyākaraṇa—do not teach Pāli units as modifications of Sanskrit forms, probably because Sanskrit is less important than Pāli for the Buddhist communities of the Theravāda tradition (cf. Scharfe, 1977, p. 195).

Modern Indo-Aryan languages were given grammatical descriptions at a much later period and mostly on the initiative of foreigners. However, some works deserve to be mentioned, such as (1) the Uktivyaktiprakaraṇa, a bilingual Sanskrit–Old Kosali manual (which gives, for the first time, an overview of Old Kosali grammar), written in the 12th century by Dāmodara (see Salomon, 1982); (2) the Varṇaratnākara by Jyotirīsvara Kavisekharācārya, the first grammar of Maithilī, composed in 1507 (see Bhatia, 1987, p. 21); (3) the grammar of Braj Bhāṣā written (in Persian) before 1676 by Mirzā Khān-ibn-Fakkru-u-Dīn (see Bhatia, 1987, pp. 17–21). A noteworthy point is that some grammars of Modern Indo-Aryan languages written by Indian scholars—whether or not preceded by grammars composed by foreigners—have followed the model of Sanskrit grammars, for example, the Mahārāṣṭraprayogacandrikā (grammar of Marathi) of Veṅkaṭa Mādhava (1827, published in 1970), the Kaśmīraśabdāmṛta (grammar of Kaśmiri) of Īśvara Kaula (1875, published in 1898), and the Mithilābhāṣāvidyotana (grammar of Maithili) of Dinabandhu Jha (1946). This is another manifestation of the “Extended Sanskrit Grammar” phenomenon (cf. §1.4).

Note that several studies on the history of the grammaticisation of Modern Indo-Aryan languages have been carried out, such as Arjunwadkar (1992); Bandyopadhyay (2011); Bhatia (1987); Mone (1927); and Shapiro (2000).

1.6 The Grammatical Descriptions of Dravidian Languages

Dravidian languages (Tamil, Telugu, Kannaḍa, and Malayāḷam) were equally the subject of native grammatical descriptions,7 more or less ancient and more or less autonomous.

Among these Dravidian grammatical descriptions, the Tamil tradition is the most ancient: it emerged at the beginning of the Common Era in South India. The language of description was classical Tamil, and the object of description was mainly poetry (see Chevillard, 2000a). The most ancient Tamil grammatical text which has come down to us is the Tolkāppiyam (5th century?). It consists of three books: (1) the “Book of Letters” (Eluttatikāram), which is devoted to phonetic, phonological, and morphophonological observations and which contains notes regarding writing; (2) the “Book of Words” (Collatikāram), which provides the description of some morphosyntactic facts, such as cases; and (3) the “Book of [poetic] topics” (Poruḷatikāram), which describes the various items of Tamil poetics (see Chevillard, 2000b). The grammatical texts composed after the Tolkāppiyam are either commentaries on the Tolkāppiyam or independent texts, among which some were more innovative (such as the Vīracōliyam, 11th century) and some had greater influence (such as the Nannūl, 13th century). For more details, see Scharfe (1977, pp. 182–183), Subrahmanya Sastri (1997), and the numerous works of Chevillard (such as Chevillard, 2000c). One should note that Tamil grammatical thought has been influenced by Sanskrit grammar (cf. §1.4), particularly in terms of the classification of items (see Scharfe, 1977, p. 181), terminology (see Chevillard, 2000c, p. 201), and concepts (see Vergiani, 2013). For more details concerning the influence of non-Tamil models on Tamil grammar, see Meenakshisundaram (1974).

It is often claimed that the Telugu grammatical tradition started in the 11th century, with the composition by the poet Nannaya of a Sanskrit grammar of Telugu titled Āndhraśabdacintāmaṇi. However, scholars now generally believe that this text was written in the 16th century. Therefore, the very first Telugu grammar written in Telugu would be the Āndhrabhāṣābhūṣaṇamu, composed by Ketana in the 12th century. In both works, great significance is attached to the different kinds of words: tajjā “derived either from Sanskrit or Prakrit,” samā “similar either to Sanskrit or Prakrit,” deśyā (lit. “regional”) “pure Telugu,” and grāmyā (lit. “related to villages”) “rustic language which is not bound by the grammatical rules,” according to Nannaya (see Sarveswara Sharma, 1973, pp. 385–386). Such a classification of Telugu words implies that the language was conceived as a modified form of Sanskrit, like a Prakrit (see Hock, 2016, p. 717). Mention should also be made of the composition, in the 19th century, of Paravastu Cinnayasūri’s Bālavyākaraṇamu, an influential grammatical treatise in Telugu where sūtras are organized by topic (cf. §1.2). For more details on the Telugu grammatical tradition, see Purushottam (1996).

The Kannaḍa grammatical tradition begins in the 12th century with two treatises written by Nāgavarma (who mentions earlier grammatical works which have not survived): the Śabdasmṛti, which is in Old Kannaḍa and which constitutes a part of the Kāvyāvalokana, a poetical work, and the Karṇāṭakabhāṣābhūṣaṇa, which is an independent work in Sanskrit sūtras (see Scharfe, 1977, p. 186). Other works would follow, some composed in (Old) Kannaḍa, such as Keśirāja’s Śabdamaṇidarpaṇa (13th century) and Kṛṣṇamācārya’s Hosagannaḍa nuḍigannaḍi (19th century), which studies the links between Kannaḍa, Sanskrit, and Tamil; another work, the Karṇāṭakaśabdānuśāsana of Bhaṭṭākaḷanka Deva (17th century), was composed in Sanskrit and influenced by Jainendra’s grammar (see Scharfe, 1977, p. 186). For more details on the Kannaḍa grammatical tradition, see Kulli (1991, 1997).

The oldest known grammatical observations related to Malayāḷam are found in a poetical treatise of the 14th century, the Līlātilakam. This text, composed of Sanskrit sūtras, describes—among other things—morphological and phonological characteristics of Maṇipravāḷam, the medieval literary language of Kerala (a mixture of Keraḷabhāṣā and Sanskrit). The defining characteristic of this treatise is that it mainly establishes a relationship of combination between both languages, not a relationship of contrast or modification (see Scharfe, 1977, p. 184; Aussant, 2012): the sentences combine Sanskrit and Keraḷabhāṣā words, which keep their endings according to their respective syntax. Like the combination of the languages which are described, the Līlātilakam blends descriptive models: one observes features coming from Sanskrit as well as Tamil grammars (see Scharfe, 1977, p. 185; Aussant, 2012, pp. 92–99). Some other grammatical works related to Malayāḷam were written after the Līlātilakam, such as the Keraḷa Kaumudi (1878) by T. M. Kovunni Nedungadi, which was conceived as an aid for writing literary works and which equally associates Sanskrit and Tamil grammars, and the Keraḷa Pāṇinīyam (1896), a successful work by A. R. Rajaraja Varma, which, after having presented the history of Malayāḷam (as a Dravidian language influenced by Sanskrit), describes it—in modern Malayāḷam—drawing his inspiration from the Pāṇinian system. For more details on the Malayāḷam grammatical tradition, see Ezhuthachan (1975).

2. Sanskrit Grammar (vyākaraṇa)

2.1 Major Steps

2.1.1 Padapāṭhas

As mentioned in §1.1, it is traditionally considered that linguistic thought in premodern India began with the composition of Padapāṭhas, around the 8th–6th centuries bce. Padapāṭhas represent the earliest available grammatical commentaries of the Vedic saṃhitās (see, for instance, Abhyankar & Devasthali, 1978, p. xviii), and the pada-units they identify are the most ancient linguistic category. No grammatical text composed by the authors of Padapāṭhas is available to us. Thus, it is from the criteria they adopt for the isolation of pada-units that we can reconstruct the linguistic principles they follow and therefore their conception of pada.

The analysis performed by Padapāṭhas is based on “phonological criteria that involve syntactic units” (Cardona, 2012, p. 53): they identify sandhis occurring in specific places within continuous speech and replace them by pauses of two kinds: 1) one pause graphically marked by “|,” which corresponds to a silence two morae in length (e.g. agním |īe | puráḥ-hitam | from agním īḷe puróhitam [ṚV 1.1.1] “I praise Agni set at the fore”), 2) one pause graphically marked by “-,” which corresponds to a silence one mora in length (e.g. agním | īḷe | puráḥ-hitam |). The two-pause distinction reveals a subdivision of the pada-category: the category of external padas, marked by “|,” and the category of internal padas, marked by “-.” This subdivision of the pada class is explained by the distinction between two kinds of junction: a word junction in the first case and a constituent or morpheme junction in the second case. Both kinds of pause are generally made where the phonetic alterations Western scholars call “external sandhi” apply (cf. Macdonell, 1995, p. 20). The places within continuous speech where these phonetic alterations occur were perceived, and then analyzed, as corresponding to the end of a specific linguistic unit identified by the name pada.

The analysis of some sequences, especially compounds, reveals two more features of pada-units. First, as already noted by Cardona (2014, pp. 91–92), the fact that the break is made only at the last junction of the derivation and even where there is no phonetic alteration, as in prajā́-patiḥ (ṚV 4.53.2 “master [-patiḥ] of creatures [prajā́-],” which is composed of the three units pra-, jā-, and pati-), indicates that authors of Padapāṭhas conceived of padas as syntactic units, that is to say as “segments terminating with nominal or verbal endings or which, in a derivational system, at one point contained such endings” (italics are mine).8 This is confirmed by the fact that, if the first member (or both members) of a compound is not used independently in the ṚV, it is not analyzed in Padapāṭhas (see Jha, 1992, p. 173). Hence, Padapāṭhas authors’ conception of padas is based on phonological as well as derivational considerations.9 Second, as shown by Abhyankar (1974, p. 10), breaks in compounds occur when the constituent parts are considered capable of expressing their separate meanings individually. Hence, there is no break in compounds used as proper names, such as viśvā́mitra (ṚV 3.53.9). Therefore, padas are considered as meaningful units.

Pāṇini, who mentions Śākalya, the author of the Padapāṭha of the Ṛgvedasaṃhitā, inherits the subdivision in external and internal padas (cf. Aussant, forthcoming b).

2.1.2 The Aṣṭādhyāyī of Pāṇini

The founding text of Vyākaraṇa is the Aṣṭādhyāyī (“The eight-chaptered”) of Pāṇini (5th century bce). A Brahmin and subject of a satrapy of the Persian Empire, Pāṇini presumably composed his grammar at a time when some of the foundations of the Brahmanical society were being questioned by emerging Buddhism and Jainism. These philosophico-religious trends challenged, among others things, the supremacy of the Sanskrit language, which, in contrast to the Hindus, was not their holy language. Pāṇini’s treatise was the basis for the establishment of a school of grammar of the same name, creating an institution which—at least as far as we can see from the texts which have reached us—largely dominated up to the 18th century. The success of the Aṣṭādhyāyī eclipsed the works of other schools. Last but not least, Pāṇini’s grammar settled, during the centuries following its composition, the norm of a language later named Sanskrit.

This treatise consists of some 4,000 sūtras (“aphorisms”; the term is generally translated by “rules”) which made up the grammar stricto sensu. On a practical level, Pāṇini’s grammar provides abstract procedures for forming words (pada) with affixes (pratyaya); for more details, see Cardona (1980, pp. 234–236). These affixes are directly introduced (under meaning conditions and co-occurrence conditions) in some of the 4,000 rules, unlike most of the bases with which they combine and which are either verbal roots (dhātu)10 or nominal bases (prātipadika). Verbal roots are listed in the Dhātupāṭha (“recitation of verbal roots”), whereas nominal bases are introduced in diverse ways. Verbal and nominal bases constitute the two main formal starting points11 for the derivational process which is found throughout Pāṇinian rules, making it possible to generate more and more complex units up to correct Sanskrit sentences (vākya). For a concrete illustration of sentence derivation, see Pinault (1989). At first glance, the Aṣṭādhyāyī appears to have not been composed to be read from beginning to end; there is neither progression nor continuity in the linguistic facts it describes. Frequently, rules which are to be applied together are found in sections separated by several rules, or even by several sections. As a consequence, the word class definitions are not systematically followed by the operations (or “accidents”) the classes undergo. Such a framework is explained by the fact that, except for the rules which strictly concern totally different groups of units (such as nouns and verbs) and for which the order of application is unimportant, the order of rules’ application is mostly relevant. It is often governed, more than by the linguistic content, by (1) the functions of the rules (general vs. specific [utsarga/apavāda], necessary vs. non-necessary [nitya/anitya], internal vs. external [antar-aṅga/bahir-aṅga]), (2) the fact that the application of one rule must precede the application of another rule, and (3) the avoidance of repetition in the formulation of the rules. Whatever the explanation of the ordering may be, the core of Pāṇini’s grammar concerns the derivation of words (always within the context of a sentence), a topic treated in the third, fourth, and fifth sections, those that are the most consistent from the point of view of the order of the rules.

The Aṣṭādhyāyī of Pāṇini is the first attempt at a complete description of a language—which encompasses, within a synchronic perspective, the sacred language named chandas (Vedic Sanskrit) and the non-sacred common language named bhāṣā (classical Sankrit)—in the form of a generative grammar (cf. Gillon, 2007) characterized by an extremely condensed formulation, a high level of formalism, and very sophisticated metalinguistic tools (terms and devices). An example of Pāṇinian sūtra is the rule iKo yaṆ aCiy, v, r and l are the substitutes of i, u, and before a vowel,” where: (1) the operation of substitution (which is a pivotal operation in the Aṣṭādhyāyī) is indicated by the substitute being marked by the nominative case and the item substituted by the genitive case (this technique makes it possible to avoid the use of a verb like “to replace”); (2) phonemes are denoted by abbreviations (pratyāhāra): iK stands for i, u, , and , while yaṆ stands for y, v, r, and l; thanks to the rule ādir antyena sahetā (1.1.71),12 a certain number of abbreviations (the Aṣṭādhyāyī uses 41 of them) can be formed, like aC, which denotes the groups of vowels, and haL, which denotes the groups of consonants; (3) the right context of an operation is marked by the locative case: aCi “before a vowel” (the left context is marked by the ablative case). For a detailed description of the Aṣṭādhyāyī, see Cardona (1997).

2.1.3 The Mahābhāṣya of Patañjali

Patañjali (2nd century bce) is the author of the Vyākaraṇamahābhāṣya (“The great commentary on the analysis [of words],” more generally named Mahābhāṣya). This monumental work quotes and discusses 4,300 vārttikas (“Remarks on the procedure” or scholia) composed by Kātyāyana,13 the most ancient commentary on Pāṇini’s grammar which has come down to us and which is known only through Patañjali’s work. The Mahābhāṣya, which takes the form of controversies between a student (śiṣya), a master who knows only a part of the topics addressed (ācāryadeśīya), and a master who establishes the final true view (ācārya, siddhāntin), concerns slightly more than 1,700 Pāṇinian sūtras. Through these scholastic debates, the content as well as the validity of the sūtras and of the vārttikas are carefully studied; examples as well as counterexamples are given to illustrate them. The discussion ends with the acceptance or the rejection of Kātyāyana’s amendments; it is not left to the reader to draw the conclusion.

Patañjali is the last member—hence embodies the highest authority—of what is traditionally called the “triad of wise men” (munitraya), the first two members being Pāṇini and Kātyāyana (about this triad and the notion of authority in the Pāṇinian tradition, see Deshpande, 1998). He is unanimously considered the most prominent commentator of the Pāṇinian work. His Mahābhāṣya, which is written in a simple but vigorous prose, constitutes a valuable dialectic instrument; thanks to it, the reader has access to the heart of Pāṇinian thought. Patañjali’s work has been subjected to numerous commentaries, mostly incomplete, with the exception of Kaiyaṭa’s Pradīpa (“The lamp [of the great commentary],” 11th century). The Mahābhāṣya is a fundamental work because, on the one hand, it provides a “state of the art” of the questions related to language addressed at that time in the Brahmanical “milieu”14 and, on the other hand, it contains the seeds of the issues which will be thought of and discussed during the following centuries. For an overview of these issues, cf. Coward and Kunjunni Raja (1990, pp. 115–119).

2.1.4 The Vākyapadīya of Bhartṛhari

Bhartṛhari (5th century) is the author of the Vākyapadīya (“Work dealing with sentences and words”) as well as of the Mahābhāṣyadīpikā (“Light on the great commentary”), the earliest commentary on the Mahābhāṣya of Patañjali, which has survived in part. Bhartṛhari is traditionally depicted as the philosopher of grammar. Considering himself a grammarian (of the Pāṇinian school), he indeed created an original philosophy which borrows various elements from other disciplines of his time (cf. Bronkhorst, 1998c, p. 764). His philosophy, which implies that the study of grammar provides access to salvation,15 goes well beyond the scope of grammar and deeply influenced later thinkers, Buddhist as well as Hindu.

Bhartṛhari is considered to have composed the Mahābhāṣyadīpikā before the Vākyapadīya. In the latter text, he addresses various topics which concern general linguistics (such as the notions of sentence, word, action, tense, gender, and number; the ways of understanding meaning; the meaning of linguistic units; and the phenomenon of self-reference of terms) as well as pure Sanskrit grammar (such as derivation, composition, etc.) in making reference to different positions defended at his time on the same topics. This perspectivist approach (the views of other schools of thought are not brought in for the sake of refutation but in a spirit of accommodation; cf. Subramania Iyer, 1992, p. 75) is a salient feature of Bhartṛhari’s work, though Patañjali already paid attention to various points of view (cf. Cardona, 2009, p. 121).

One of the key ideas in Bhartṛhari’s philosophy is that any whole is more real than its constituents. On the linguistic level, this means that, among the three classes of units which constitute language (varṇa “phoneme,” pada “word,” vākya “sentence”), only the sentence is the primary linguistic unit; phonemes, stems, suffixes, and words are inventions of grammarians. Another key idea developed by Bhartṛhari—but already formulated by Patañjali, though slightly differently and not on the same scale—is that linguistic units can be conceived of as different from the sounds that reveal them (cf. Bronkhorst, 1998b, p. 382) and, as such, they are called sphoṭa. As noted by Subramania Iyer (1992, p. 160), the notion of sphoṭa is the grammarians’ answer to the problem raised by the understanding of a meaning from sounds which are uttered in a temporal sequence (and, therefore, which cannot cooperate to convey the meaning). Bhartṛhari innovates in making sphoṭa the meaning-bearer (cf. §2.2.1), thus inaugurating a long series of debates, inside as well as outside grammar (cf. Bronkhorst, 1998b, pp. 382–383). For an overview of arguments for and against the sphoṭa resorted to in premodern India, cf. Gaurinath Sastri (1980); for more information on sphoṭa, cf. Coward (1980) and Matilal (1990, pp. 77–105).

2.1.5 The Siddhāntakaumudī of Bhaṭṭoji Dīkṣita

Bhaṭṭoji Dīkṣita (late 16th century, early 17th century) is one of the late major figures of the Pāṇinian school, which he helped to renew. He composed various works on different topics, four of which are related to grammar: the Śabdakaustubha (“The kaustubha jewel of words”), which is a commentary on Pāṇini’s Aṣṭādhyāyī ; the Vaiyākaraṇamatonmajjana (“The advent of grammarians’ views”), which consists of a collection of 76 stanzas dealing with syntax, semantics, and philosophy of language; the Siddhāntakaumudī (“The moonlight of the conclusions”), which is a rearrangement of Pāṇini’s Aṣṭādhyāyī with a commentary; and the Prauḍhamanoramā (“The one which delights the spirit of advanced [students]”), an extensive commentary on the Siddhāntakaumudī.

Bhaṭṭoji Dīkṣita is widely known for his Siddhāntakaumudī. This work does indeed represent a turning point in the Sanskrit grammatical tradition: it is the most accomplished arrangement of Pāṇinian rules by topic (or prakriyā, cf. §1.2 “Grammar”). Grammars arranged by topic are mainly organized according to kinds of pada: the rules which introduce constituent units of one kind of pada as well as the rules which teach operations which apply inside the pada’s boundaries are gathered together (contrary to the Aṣṭādhyāyī, cf. §2.1.2). The Siddhāntakaumudī brought the prakriyā method to such a level of refinement—including all the Pāṇinian rules, but also Kātyāyana’s vārttikas and some of Patañjali’s observations—that it was very successful and eclipsed Pāṇini’s grammar itself (Indian scholars who are traditionally trained today still learn grammar through the Siddhāntakaumudī first).16

2.1.6 Nāgeśa’s Works

Nāgeśa (late 17th century, early 18th century) is traditionally considered the last great representative of the Pāṇinian school. He is the author of several works, some related to grammar and some others related to disciplines such as poetry, dialectics, and yoga. His grammatical works include commentaries, such as the Uddyota (“The light [of the lamp of the great commentary]”), which is a commentary on Kaiyaṭa’s Pradīpa, and the Śabdenduśekhara (“The moon crest of words”), which is a commentary on Bhaṭṭoji Dīkṣita’s Prauḍhamanoramā, but also independent treatises, such as the Paribhāṣenduśekhara (“The moon crest of metarules”) which critically examines 133 metarules, as well as the Vaiyākaraṇasiddhāntamañjūṣā (“The chest of grammarians’ conclusions,” in three recensions of different length), which provides a synthesis of the main ideas related to the philosophy of grammar, ideas initiated by Patañjali and largely developed by Bhartṛhari.

Faithful to the Pāṇinian tradition, Nāgeśa nevertheless does not hesitate to make reference to some ideas from other disciplines. For instance, when he describes the word-meaning relation as being threefold—primary (śakti), secondary (lakṣaṇā), suggestive (vyañjanā)—he makes reference to a classification initially found in poetry, and when he says that śakti is threefold—conventional (rūḍhi), derivative (yoga), conventional-derivative (yogarūḍhi)—he makes reference to a classification initially found in dialectics (note that grammarians disagree with both of these views, cf. §2.2.3). In his Sphoṭavāda, an independent treatise on sphoṭa, Nāgeśa defines sphoṭa as “that from which the meaning bursts forth” and mentions an eightfold classification of it: (1) phoneme, (2) word, (3) sentence, (4) indivisible word, (5) indivisible sentence, (6) phoneme-universal, (7) word-universal, (8) sentence-universal (for more details on this classification, cf. Matilal, 1990, p. 104). Nāgeśa indicates that this classification presents the conceptions of grammarians regarding sphoṭa, but in his Paramalaghumañjūṣā, he says that the sentence-sphoṭa is the most important. According to Bronkhorst (1998b, p. 383), Nāgeśa’s vacillations regarding sphoṭa can be explained by the conflict which lies between two grammatical views: (1) grammatical derivations, for several reasons, cannot presuppose that stems and affixes are imaginary units or (2) only the sentence is real.

2.2 Main Topics Addressed

2.2.1 Minimal Meaning-Bearer Units

The search for minimal meaning-bearer units—brought together with the question of their real existence—has been a topic of great interest, for grammarians as well as for thinkers from other schools (cf. Bronkhorst, 1998b, pp. 380–383).

Some thinkers claimed that phonemes are meaning-bearers (cf. Vākyapadīya stanza 2.62: “Just as a minute perceptible object, when associated with something else, is perceived with it, in the same way, a phoneme becomes expressive [of a meaning] when it is associated with other phonemes,” translation by Subramania Iyer, 1977). For Pāṇini and some other grammarians, the minimal meaning-bearer units are verbal roots, nominal stems, and affixes: arthavad adhātur apratyayaḥ prātipadikam (1.2.45) “[The unit] which is meaningful, which is neither a verbal root nor an affix [is called] prātipadika (‘nominal stem’)” (italics are mine). Some other thinkers consider that finished words (śabda or pada) are meaning-bearer units; this view can be inferred from definitions of the sentence such as “[the sentence is] a collection of words” (śabdasaṃghātaḥ) or “[the sentence is] the first word” (padam ādyam), which are quoted by Bhartṛhari in his Vākyapadīya (stanzas 2.1–2). What is interesting to note regarding such conceptions is the analysis of the way words express their meaning within the sentence of which they are part. According to one analysis (called abhihitānvaya), the words of a sentence first convey their own meaning; these meanings subsequently relate syntactically to each other and produce the sentence meaning. According to another analysis (called anvitābhidhāna), the words of a sentence convey their own meaning as well as their syntactic relation to each other (for more details on these analyses, cf. Kunjunni Raja, 1963, chapter 5). According to Bhartṛhari and some later grammarians (such as Nāgeśa, cf. §2.1.6), the minimal meaning-bearer unit is the vākya-sphoṭa, i.e. the sentence-sphoṭa (cf. §2.1.4).

2.2.2 Classes of Words

The analysis of language into units seems to have been fundamental in all traditions of language study. Indeed, in each Vedāṅga related to language as well as in the other Sanskrit language sciences, one finds at least one classification of words (pada).

Classifying words is an activity which is neither self-explanatory nor consistent; the classifier (whether an individual scholar, a school of thought, or a trend) has an epistemological aim, and we must consider the regularities (in other words, the classes) established on this basis. An accurate study (cf. Aussant, 2016a) shows that words (especially nouns) are more often classified according to semantic criteria in the language sciences of premodern India. This can be explained by the importance given to the relation between the word and its meaning; what understanding a word triggers and how it does so constitute a central topic of thought for ancient Indian theoreticians of language. Moreover, and unsurprisingly, grammar resorts to the widest range of word classifications, according to criteria which are (1) formal (for instance, the Pāṇinian distinction between noun—sUB-anta “[a unit] that terminates in a nominal [ending]”—and verb—tiṄ-anta “[a unit] that terminates in a verbal [ending]”—and their 29 subclasses, cf. Aussant, 2016b), (2) semantic, i.e., ontological (nouns denoting a class, a quality, an action, or a substance) and semiological (words having a generic property, a quality, an action, or the wish of the speaker as connotation), and (3) pragmatic (for instance, the distinction between words of spoken language [bhāṣā] and words of sacred literature [chandas]).

2.2.3 Relation Between Word and Meaning

According to premodern Indian thinkers, the “designating relation” (vṛtti) may have two forms: (1) a primary designating relation, called śakti (by grammarians and dialecticians), abhidhā (by exegetes and poeticians), or mukhya; and (2) a secondary designating relation, called lakṣaṇā or gauṇa. Grammarians and exegetes consider the primary designating relation as being innate (autpattika), natural (svābhāvika), and not relying on a soul (apauruṣeya);17 dialecticians, on the contrary, consider it as being dependent on a convention (saṃketa) and divine (according to ancient dialecticians) or human (according to neo-dialecticians). Unlike other thinkers, grammarians maintain that the word has only one designating relation (the śakti), whatever its uses; secondary or figurative meaning does not result from any particular signification function. Some later Pāṇinīyas like Nāgeśa explain the difference between various meanings of a word—which are all considered as primary—by saying that some are well-known (prasiddha) while others are not or are less well-known (aprasiddha), cf. Aussant (2014a, pp. 29–30).

2.2.4 Primary Referent of Nouns

The ongoing question related to the primary referent of nouns—according to all schools of premodern Indian thinkers, with the exception of Buddhist logicians, cf. Kunjunni Raja (1963, pp. 78–94)—was to determine whether this primary referent was particular (e.g., “cow” denotes a specific cow) or universal (e.g., “cow” denotes cowness). Different key issues are tightly related to this question, such as the relation between language and reality (if words denote particulars, one ends up having to assume an infinity of instructions in order to know truly the referent of words; evolutive referents or referents which do not yet exist —“weave a cloth!”—as well as general rules or Vedic injunctions raise several problems as well) or the nature of the primary designating relation (natural or conventional, cf. §2.2.3).

Different views have been claimed by grammarians: some, like Vyāḍi (cf. Scharfe, 1977, pp. 124–126), considered that the primary referent of a noun is the particular (dravya); some others, like Vājapyāyana, considered that it is the universal or the generic property (jāti), while still others, like Patañjali, considered that it is both the particular and the universal/generic property, one being principal and the other subordinate according to the speaker’s intention (vivakṣā). In addition to these three cardinal theses, some other views were discussed by grammarians: (a) the noun would denote the generic property, the individual, and the gender; (b) it would denote these three items plus the number; or (c) it would denote these four items plus the semantic role the noun takes on within the sentence (cf. Aussant, 2014b, pp. 273–275). According to dialecticians of the old school, a noun primarily denotes the particular (vyakti), its generic configuration (ākṛti), and its generic property (jāti); dialecticians of the new school slightly modify this view (cf. Kunjunni Raja, 1963, pp. 70–71). According to exegetes, the noun primarily denotes the class property (ākṛti), which is common to all the particular instances of one class and only to them. For more details about the denotation of (generic) nouns, see Scharf (1996).

Further Reading

Allen, W. S. (1953). Phonetics in ancient India. London: Oxford University Press.Find this resource:

Aussant, E. (2009). Le nom propre en Inde: Considérations sur le mécanisme référentiel. Lyon: ENS Editions.Find this resource:

Cardona, G. (1980). Pāṇini: A survey of research. Delhi: Motilal Banarsidass.Find this resource:

Cardona, G. (1997). Pāṇini: His work and its traditions. Vol. 1: Background and introduction (2d ed.). Delhi: Motilal Banarsidass.Find this resource:

Cardona, G., & Jain, D. (2007). The Indo-Aryan languages. London: Routledge.Find this resource:

Chevillard, J.-L., & Passerieu, J.-C. (1989). La tradition grammaticale tamoule. In S. Auroux (Ed.), Histoire des idées linguistiques. Vol. 1: La naissance des métalangages en Orient et en Occident (pp. 417–429). Liège: Mardaga.Find this resource:

Coward, H. G., & Kunjunni Raja, K. (Eds.). (1990). Encyclopedia of Indian philosophies. Vol. 5: The philosophy of the grammarians. Delhi: Motilal Banarsidass.Find this resource:

Hock, H. H. (2016). Indo-Aryan grammatical traditions (Sanskrit and Prakrit). In H. H. Hock & E. Bashir (Eds.), The languages and linguistics of South Asia: A comprehensive guide (pp. 707–716). Berlin: De Gruyter.Find this resource:

Matilal, B. K. (1990). The word and the world: India’s contribution to the study of language. Oxford: Oxford University Press.Find this resource:

Pinault, G.-J. (1989a). Parole articulée et vérité. In S. Auroux (Ed.), Histoire des idées linguistiques. Vol. 1: La naissance des métalangages en Orient et en Occident (pp. 293–302). Liège: Mardaga.Find this resource:

Pinault, G.-J. (1989b). Travaux à partir du corpus védique. In S. Auroux (Ed.), Histoire des idées linguistiques. Vol. 1: La naissance des métalangages en Orient et en Occident (pp. 303–330). Liège: Mardaga.Find this resource:

Pinault, G.-J. (1989c). Pāṇini et l’enseignement grammatical. In S. Auroux (Ed.), Histoire des idées linguistiques Vol. 1: La naissance des métalangages en Orient et en Occident (pp. 331–353). Liège: Mardaga.Find this resource:

Pinault, G.-J. (1989d). Procédés pāṇinéens. In S. Auroux (Ed.), Histoire des idées linguistiques. Vol. 1: La naissance des métalangages en Orient et en Occident (pp. 354–370). Liège: Mardaga.Find this resource:

Pinault, G.-J. (1989e). Le système de Pāṇini. In S. Auroux (Ed.), Histoire des idées linguistiques. Vol. 1: La naissance des métalangages en Orient et en Occident (pp. 371–400). Liège: Mardaga.Find this resource:

Scharf, P. M. (1996). The denotation of generic terms in ancient Indian philosophy: Grammar, Nyāya, and Mīmāṃsā. Philadelphia: American Philosophical Society.Find this resource:

Scharfe, H. (1977). Grammatical literature. Wiesbaden: O. Harrassowitz.Find this resource:

Zimmermann, F. (1989). Les théories de la signification. In S. Auroux (Ed.), Histoire des idées linguistiques. Vol. 1: La naissance des métalangages en Orient et en Occident (pp. 401–416). Liège: Mardaga.Find this resource:


Abhyankar, K. V. (1974). Veda-padapāṭha-carcā, together with the text of Upalekhasūtram of Bhāradvāja Bṛhaspati edited with explanation in English. Poona: Bhandarkar Oriental Research Institute.Find this resource:

Abhyankar, K. V., & Devasthali, G. V. (1978). Vedavikṛtilakṣaṇa-saṃgraha: A collection of twelve tracts on Vedavikṛtis and allied topics. Poona: Bhandarkar Oriental Research Institute.Find this resource:

Allen, W. S. (1953). Phonetics in ancient India. London: Oxford University Press.Find this resource:

Auroux, S. (1992). Le processus de grammatisation et ses enjeux. In S. Auroux (Ed.), Histoire des idées linguistiques (Vol. 2, pp. 11–64). Liège: Mardaga.Find this resource:

Auroux, S. (1994). La révolution technologique de la grammatisation. Liège: Mardaga.Find this resource:

Arjunwadkar, K. S. (1992). Maraṭhi vyakaraṇaca itihas [A history of Marathi grammar]. Mumbai: Mumbai Vishva Vidyalay.Find this resource:

Aussant, E. (2009). Le nom propre en Inde: Considérations sur le mécanisme référentiel. Lyon: ENS Editions.Find this resource:

Aussant, E. (2012). Pāṇinian features of the oldest known Malayāḷam description. In G. Cardona, A. Aklujkar, & H. Ogawa (Eds.), Studies in Sanskrit grammars: Proceedings of the Vyākaraṇa section of the 14th World Sanskrit Conference (pp. 87–101). New Delhi: D. K. Printworld.Find this resource:

Aussant, E. (2014a). Sanskrit theories on homonymy and polysemy. Bulletin d’Études Indiennes, 32, 13–36.Find this resource:

Aussant, E. (2014b). Les grammairiens indiens du sanskrit et le sens des mots. In S. Archaimbault, J.-M. Fournier, & V. Raby (Eds.), Penser l’histoire des savoirs linguistiques: Hommage à Sylvain Auroux (pp. 273–283). Lyon, ENS Éditions.Find this resource:

Aussant, E. (2014c). La lecture aryasamajiste du Vyākarana. Dossiers d’HEL 6.Find this resource:

Aussant, E. (2016a). Classifications of words in traditional Sanskrit grammar and related disciplines (Unpublished doctoral dissertation). Université Paris Diderot.Find this resource:

Aussant, E. (2016b). Classifications of words in ancient Sanskrit grammars. In C. Assunção, G. Fernandes, & R. Kemmler (Eds.), History of linguistics 2014: Selected papers from the 13th International Conference on the History of the Language Sciences (ICHoLS XIII), Vila Real, Portugal, 25–29 August 2014 (pp. 97–109). Amsterdam: John Benjamins.Find this resource:

Aussant, E. (forthcoming a). Grammaire sanskrite étendue. Histoire Epistémologie Langage, 39.2.Find this resource:

Aussant, E. (forthcoming b). Les pada et leurs classifications: Un point d’entrée pour étudier l’émergence et le développement de la tradition grammaticale sanskrite. In É. Aussant & G. Colas (Eds.), Pratiques des scolastiques et scolastiques des pratiques. Paris: EFEO/IFP.Find this resource:

Balbir, N. (2000). Notices from the Corpus des Textes Linguistiques Fondamentaux: Vararuci (notice 4331), Puruṣottama (notice 4335), Hemacandra (notice 4336), Mārkaṇḍeya (notice 4340).

Bandyopadhyay, A. (2011). A history of Bengali grammar. Delhi: Sanskrit Pustak Bhandar.Find this resource:

Belvalkar, S. K. (1915). An account of the different existing systems of Sanskrit grammar. Poona: Aryabhushan Press.Find this resource:

Bhatia, T. (1987). A history of the Hindi grammatical tradition: Hindi-Hindustani grammar, grammarians, history and problems. Leiden: E. J. Brill.Find this resource:

Bhide, P. R. (2015). IC analysis in modern Sanskrit lexicography with special reference to Encyclopaedic Dictionary of Sanskrit on Historical Principles (Unpublished doctoral dissertation). Indian Institute of Technology Bombay, Department of Humanities and Social Sciences.Find this resource:

Bloomfield, L. (1933). Language. New York: Henry Holt.Find this resource:

Bronkhorst, J. (1998a). Patañjali. In E. Craig (Ed.), Routledge encyclopedia of philosophy (Vol. 7, pp. 248–250). London: Routledge.Find this resource:

Bronkhorst, J. (1998b). Indian theories of language. In E. Craig (Ed.), Routledge encyclopedia of philosophy (Vol. 7, pp. 379–384). London: Routledge.Find this resource:

Bronkhorst, J. (1998c). Bhartṛhari. In E. Craig (Ed.), Routledge encyclopedia of philosophy (Vol. 7, pp. 764–766). London: Routledge.Find this resource:

Bronkhorst, J. (2014). Deviant voices in the history of Pāṇinian grammar. Bulletin d’Études Indiennes, 32, 47–53.Find this resource:

Bronner, Y. (2012). Sanskrit poetics. In R. Greene & S. Cushman (Eds.), The Princeton encyclopedia of poetry and poetics (4th ed., pp. 1244–1250). Princeton: Princeton University Press.Find this resource:

Cardona, G. (1980). Pāṇini: A survey of research. Delhi: Motilal Banarsidass.Find this resource:

Cardona, G. (1997). Pāṇini: His work and its traditions. Vol. 1: Background and introduction (2d ed.). Delhi: Motilal Banarsidass.Find this resource:

Cardona, G. (2008). Theoretical precedents of the Kātantra. In M. Kaul & A. Aklujkar (Eds.), Linguistic traditions of Kashmir: Essays in memory of Pandit Dinanath Yaksh (pp. 300–367). New Delhi: D. K. Printworld.Find this resource:

Cardona, G. (2009). Bhartṛhari and Patañjali: Traditions preserved. In M. Chaturvedi (Ed.), Bhartṛhari, language, thought and reality (pp. 119–161). Delhi: Motilal Banarsidass.Find this resource:

Cardona, G. (2012). Pāṇini and Padakāras. In F. Voegeli, V. Eltschinger, D. Feller, M. P. Candotti, B. Diaconescu, & M. A. Kulkarni (Eds.), Devadattīyam: Johannes Bronkhorst felicitation volume (pp. 39–61). Bern: Peter Lang.Find this resource:

Cardona, G. (2014). Segmentation of Vedic texts: Padapāṭhas. Bulletin d’Études Indiennes, 32, 87–100.Find this resource:

Cardona, G., & Jain, D. (2007). The Indo-Aryan languages. London: Routledge.Find this resource:

Chevillard, J.-L. (2000a). Les débuts de la tradition linguistique tamoule. In S. Auroux, E. F. K. Koerner, H.-J. Niederehe, & K. Versteegh (Eds.), History of the language sciences: An international handbook on the evolution of the study of language from the beginnings to the present (pp. 191–194). Berlin: Walter de Gruyter.Find this resource:

Chevillard, J.-L. (2000b). Le Tolkāppiyam et le développement de la tradition linguistique tamoule. In S. Auroux, E. F. K. Koerner, H.-J. Niederehe, & K. Versteegh (Eds.), History of the language sciences: An international handbook on the evolution of the study of language from the beginnings to the present (pp. 194–200). Berlin: Walter de Gruyter.Find this resource:

Chevillard, J.-L. (2000c). Les successeurs du Tolkāppiyam: Le Nannūl, le Vīracōliyam et les autres écoles. In S. Auroux, E. F. K. Koerner, H.-J. Niederehe, & K. Versteegh (Eds.), History of the language sciences: An international handbook on the evolution of the study of language from the beginnings to the present (pp. 200–202). Berlin: Walter de Gruyter.Find this resource:

Chevillard, J.-L. (2000d). Notices from the Corpus des Textes Linguistiques Fondamentaux: Tolkāppiyanār (notice 4351), Iḷampūraṇar (notice 4352), Puttamittiranār (notice 4353), Kuṇavīrapaṇṭitar (notice 4354), Pavaṇanti Munivar (notice 4355), Mayilai Nātar (notice 4356), Cēn̲āvaraiyar (notice 4357), Naccinārkkiniyar (notice 4358), Cuppiramaṇiya Tīṭcitar (notice 4359), Civañāna Munivar (notice 4360).

Chevillard, J.-L. (2009). Elementary grammatical devices in classical Tamil grammar. In P. Hummel (Ed.), Epilanguages: Beyond idioms and languages. Paris: Editions Philologicum.Find this resource:

Coward, H. G. (1980). The sphoṭa theory of language. Delhi: Motilal Banarsidass.Find this resource:

Coward, H. G., & Kunjunni Raja, K. (Eds.). (1990). Encyclopedia of Indian philosophies. Vol. 5: The philosophy of the grammarians. Delhi: Motilal Banarsidass.Find this resource:

David, H. (2017). Les définitions de l’énoncé (vākya) dans la tradition sanskrite: Entre grammaire et exégèse. Langages, 205, 27–41.Find this resource:

De, S. K. (1960). History of Sanskrit poetics. Calcutta: K. L. Mukhopadhyay.Find this resource:

Deshpande, M. M. (1998). Evolution of the notion of authority (prāmāṇya) in the Pāṇinian tradition. Histoire Épistémologie Langage, 20(1), 5–28.Find this resource:

Deshpande, M. M. (2000). Indian theories on phonetics. In S. Auroux, E. F. K. Koerner, H.-J. Niederehe, & K. Versteegh (Eds.), History of the language sciences: An international handbook on the evolution of the study of language from the beginnings to the present (pp. 137–146). Berlin: Walter de Gruyter.Find this resource:

Emeneau, M. B. (1988). Bloomfield and Pāṇini. Language, 64(4), 755–760.Find this resource:

Ezhuthachan, K. N. (1975). The history of the grammatical theories in Malayalam (part I). Trivandrum: Dravidian Linguistics Association.Find this resource:

Gaurinath Sastri. (1980). A study in the dialectics of sphoṭa. Delhi: Motilal Banarsidass.Find this resource:

Gerow, E. (1971). Indian poetics. Wiesbaden: O. Harrassowitz.Find this resource:

Gerschheimer, G. (1996). La théorie de la signification chez Gadādhara. Paris: Publications de l’Institut de civilisation indienne, Diffusion De Boccard.Find this resource:

Gillon, B. (2007). Pāṇini’s Aṣṭādhyāyī and linguistic theory. Journal of Indian Philosophy, 35(5), 445–468.Find this resource:

Grierson, A. G. (1918, July). The Prakrit Vibhasas. Journal of the Royal Asiatic Society, 489–517.Find this resource:

Harris, Z. (1954). Transfer grammar. International Journal of American Linguistics, 20(4), 259–270.Find this resource:

Hock, H. H. (2016). Indo-Aryan grammatical traditions (Sanskrit and Prakrit). In H. H. Hock & E. Bashir (Eds.), The languages and linguistics of South Asia: A comprehensive guide (pp. 707–716). Berlin: De Gruyter.Find this resource:

Jha, D. (1946). Maithilī-bhāṣā-vidyotana. Darbhanga: Maithili Sahitya Parishad.Find this resource:

Jha, G. (1942). Pūrvā-Mīmāṃsā in its sources. Benares: Benares Hindu University.Find this resource:

Jha, V. N. (1987). Studies in the Padapāṭhas and Vedic philology. Delhi: Pratibha Prakashan.Find this resource:

Jha, V. N. (1992). A linguistic analysis of the Ṛgveda-Padapāṭha (Pre-Pāṇinian grammatical traditions, Part I). Delhi: Sri Satguru Publications.Find this resource:

Kahrs, E. (1998). Indian semantic analysis: The Nirvacana tradition. Cambridge, U.K.: Cambridge University Press.Find this resource:

Kaula, Īśvara (1898). Kaśmīraśabdāmṛtaṃ, A Kāçmīrī Grammar Written in the Sanskrit Language by Īśvara Kaula [G. A. Grierson, Ed.]. Calcutta: Royal Asiatic Society.Find this resource:

Kulkarni, N. R. (1995). A grammatical analysis of the Taittirīya-Padapāṭha (Pre-Pāṇinian grammatical traditions, Part II). Delhi: Sri Satguru.Find this resource:

Kulli, J. S. (1991). History of grammatical theories in Kannada. Trivandrum: International School of Dravidian Linguistics.Find this resource:

Kulli, J. S. (1997). Grammars-Kannada. In V. I. Subramoniam (Ed.), Dravidian encyclopedia. Vol. 3: Thiruvananthapuram (pp. 244–246). Trivandrum: International School of Dravidian Linguistics.Find this resource:

Kunjunni Raja, K. (1963). Indian theories of meaning. Madras: Adyar Library and Research Center.Find this resource:

McCrea, L. (2000). The hierarchical organization of language in Mīmāṃsā interpretive theory. Journal of Indian Philosophy, 28, 429–459.Find this resource:

Macdonell, A. A. (1995 [1916]). A Vedic grammar for students. Delhi: Motilal Banarsidass.Find this resource:

Mādhava, V. (1970). Mahārāṣṭraprayogacandrikā: Mūḷa Hastakikhitāvarūna, Sampādita Karūna, Prathamaca Chāpalelā, Marāṭhī Vyākaraṇāvaracā, Saṃskṛta-Marāṭhī Junā Laghugrantha. Veṅkatamādhavakṛta. Pune: Deśamukha.Find this resource:

Matilal, B. K. (1977). Nyāya-Vaiśeṣika. Wiesbaden: O. Harrassowitz.Find this resource:

Matilal, B. K. (1990). The word and the world: India’s contribution to the study of language. Oxford: Oxford University Press.Find this resource:

Meenakshisundaram, T. P. (1974). Foreign models in Tamil grammar. Dravidian Linguistics Association Publication 15. Trivandrum: University of Kerala.Find this resource:

Miller, R. A. (2000). The establishment of Tibetan linguistics. In S. Auroux, E. F. K. Koerner, H.-J. Niederehe, & K. Versteegh (Eds.), History of the language sciences: An international handbook on the evolution of the study of language from the beginnings to the present (Vol. 1, pp. 203–206). Berlin: Walter de Gruyter.Find this resource:

Mone, M. S. (1927). Marathiche vyakarankar ani vyakaran-prabandhakar [Marathi’s grammarians]. Pune: Shrimant Holkar Sarkar Granthamala.Find this resource:

Nitti-Dolci, L. (1938). Les grammairiens prakrits. Paris: Adrien-Maisonneuve.Find this resource:

Patkar, M. M. (1981). History of Sanskrit lexicography. Delhi: Munshiram Manoharlal.Find this resource:

Pinault, G.-J. (1989). Le système de Pāṇini. In S. Auroux (Ed.), Histoire des idées linguistiques. Vol. 1: La naissance des métalangages en Orient et en Occident (pp. 371–400). Liège: Mardaga.Find this resource:

Pollock, S. (2007). The language of the gods in the world of men: Sanskrit, culture and power in premodern India. Delhi: Permanent Black.Find this resource:

Purushottam, B. (1996). The theories of Telugu grammar. Trivandrum: International School of Dravidian Linguistics.Find this resource:

Raghavan, V. (1974). How many grammars? In S. K. Chatterji et al. (Eds.), Charudeva Shastri felicitation volume: Presented to Prof. Charudeva Shastri on the occasion of his seventy-fifth anniversary by his friends and admirers (pp. 271–278). Delhi: Charudeva Shastri Felicitation Committee.Find this resource:

Renou, L. (1947). Les écoles védiques et la formation du Veda. Paris: Cahiers de la Société Asiatique.Find this resource:

Rogers, D. E. (1987). The influence of Pāṇini on Leonard Bloomfield. Historiographia Linguistica, 14(1/2), 89–138.Find this resource:

Salomon, R. (1982). The Ukti-Vyakti-Prakaraṇa as a manual of spoken Sanskrit. Indo-Iranian Journal, 24(1), 13–25.Find this resource:

Scharf, P. M. (1996). The denotation of generic terms in ancient Indian philosophy: Grammar, Nyāya, and Mīmāṃsā. Philadelphia: American Philosophical Society.Find this resource:

Scharfe, H. (1977). Grammatical literature. Wiesbaden: O. Harrassowitz.Find this resource:

Shapiro, M. C. (2000). The Hindi grammatical tradition. In History of the language sciences: An international handbook on the evolution of the study of language from the beginnings to the present (Vol. 1, pp. 178–181). Berlin: Walter de Gruyter.Find this resource:

Sharma Peri, S. (1973). What the native grammarians say about Telugu language. Zeitschrift der deutschen morgenländischen Gesellschaft (Suppl. 2)18, 384–389.Find this resource:

Subrahmanya Sastri, P. S. (1997). History of grammatical theories in Tamil. Chennai: Kuppuswami Sastri Research Institute.Find this resource:

Subramania Iyer, K. A. (1977). The Vākyapadīya of Bhartṛhari, Kāṇḍa II. Delhi: Motilal Banarsidass.Find this resource:

Subramania Iyer, K. A. (1992). Bhartṛhari: A study of the Vākyapadīya in the light of ancient commentaries. Poona: Deccan College.Find this resource:

Thieme, P. (1982–1983). Meaning and form of the “grammar” of Pāṇini. Studien zur Indologie und Iranistik, 8/9, 3–34.Find this resource:

Vergiani, V. (2013). The adoption of Bhartṛhari’s classification of the grammatical object in Cēṉāvaraiyar’s commentary on the Tolkāppiyam. In W. Cox & V. Vergiani (Eds.), Bilingual discourse and cross-cultural fertilisation: Sanskrit and Tamil in medieval India (pp. 161–197). Pondicherry: Institut français de Pondichéry & École française d’Extrême-Orient.Find this resource:

Verhagen, P. C. (2000a). The classical Tibetan grammarians. In S. Auroux, E. F. K. Koerner, H.-J. Niederehe, & K. Versteegh (Eds.), History of the language sciences: An international handbook on the evolution of the study of language from the beginnings to the present (Vol. 1, pp. 207–210). Berlin: Walter de Gruyter.Find this resource:

Verhagen, P. C. (2000b). The influence of the Sanskrit tradition on Tibetan indigenous grammar. In S. Auroux, E. F. K. Koerner, H.-J. Niederehe, & K. Versteegh (Eds.), History of the language sciences: An international handbook on the evolution of the study of language from the beginnings to the present (Vol. 1, pp. 210–214). Berlin: Walter de Gruyter.Find this resource:

Verhagen, P. C. (2001). A History of Sanskrit Grammatical Literature in Tibet, Vol. II: Assimilation into Indigenous Scholarship, Handbuch der Orientalistik, Leiden, E. J. Brill Co.Find this resource:

Verpoorten, J.-M. (1987). Mīmāṃsā literature. Wiesbaden: O. Harrassowitz.Find this resource:

Vogel, C. (1979). Indian lexicography. Wiesbaden: O. Harrassowitz.18Find this resource:

Zwartjes, O. (2011). Portuguese missionary grammars in Asia, Africa and Brazil, 1550–1800. Amsterdam: John Benjamins.Find this resource:

Zwartjes, O., & Pytlowany, A. (forthcoming). Pre-modern descriptions in Dutch of languages in India: Ketelaar’s grammar of Hindustani (1698) and Rüell’s grammar of Sinhala (1700).Find this resource:


(1.) In the Indian tradition, the term designates a means whereby one gains a particular kind of knowledge, that is, a text.

(2.) “Premodern India” means India before the arrival of European theories and practices (see Pollock, 2007, pp. 8–10).

(3.) These texts were put in writing at a recent date (cf. Al Birûnî’s record in the 11th century, which mentions a Veda recently transcribed in Kashmir).

(4.) Note that the general procedure adopted in Prātiśākhyas (treatises on phonetics) is to consider the posited padapāṭha to be the source for deriving the saṃhitāpāṭha by phonological rules. Prātiśākhyas indeed give rules on how to construct the saṃhitāpāṭha version of Vedic texts from their padapāṭha version.

(5.) The foundation text of Nyāya was composed at the beginning of the first millenium. Up to the 11th century, it is the “classical period” of the discipline (opponents are Buddhist logicians). The work of Udayana constitutes the culmination point of this classical period; it effects, among others, the quasi-fusion with another philosophical system, the Vaiśeṣika, which aims at classifying real entities (padārtha). Navya-Nyāya emerges with the Tattvacintāmaṇi of Gaṅgeśa, and from then on the opponents are exegetes. See Matilal (1977) for more details.

(6.) The most ancient lexicon which has come down to us, the Nāmaliṅgānuśāsana of Amarasiṃha, would have been composed around the 6th century.

(7.) As they were the subject of grammatical descriptions by foreigners, including missionaries (see Zwartjes, 2011, Zwartjes & Pytlowany, forthcoming).

(8.) Secondary derivatives, which are analyzed as derived from a syntactical pattern involving inflected forms, are submitted to the same analysis, e.g., gó-mān (ṚV 4.2.5), which is conceived as derived from gāvosya santi “he to whom (’sya) cows (gāvo) belong (santi).”

(9.) Note that Jha (1987, pp. 20–23, 1992, pp. 22–25), Kulkarni (1995, pp. 9 et al.), and more recently Bhide (2015, p. 51) have pointed out the striking similarity which can be observed between the analysis adopted by authors of Padapāṭhas and the one achieved by the Western Immediate Constituents Analysis (ICA)—at the level of words, at least (Jha, 1992, p. 67). The procedure is the following: if the case ending is not segmentable, the segmentation occurs before the previous constituent, e.g., prajā́-vatīṣu (ṚV 7.1.11), where -su is not segmented because it follows a long vowel. If this second constituent is not analyzable, the separation is applied to the previous constituent, etc.

(10.) The class name dhātu also applies to derived verbal bases (formed from verb bases and nominal forms), not included among the primitive bases of the Dhātupāṭha.

(11.) Note that the very first step, in the Pāṇinian derivational system, is a weak semantic level, closely related to syntax: nouns and verbs are derived bearing a clear relationship to the utterance of which they are a constituent (cf. Cardona, 1997, pp. 136–185).

(12.) “An initial item joined with a final marker denotes not only itself but also all intervening items.”

(13.) Patañjali nevertheless comments directly on 468 Pāṇinian sūtras.

(14.) Patañjali’s thought was not influenced by the classical systems of Indian philosophy, “with the possible exception of Sarvāstivāda Buddhism” (cf. Bronkhorst, 1998a).

(15.) Cf. Cardona (1980, p. 300): “Grammar, as a means for discriminating correct (sādhu) from incorrect (asādhu, apabhraṃśa) usage, is a means of attaining ultimate release (apavarga), what we call salvation.”

(16.) With a major exception in modern times: the Āryasamāj schools (cf. Aussant, 2014c) stick to the Aṣṭādhyāyī order.

(17.) This did not prevent grammarians from thinking about issues related to conventional relations, such as proper names and metalinguistic terms. See Aussant (2009) for more details on this point.

(18.) There is a revised and enlarged edition of this work: Indian Lexicography Revised and Enlarged Edition, edited by Jürgen Hanneder and Martin Straube (Munich: P. Kirchheim, 2015).