Old Kannada () is the earliest attested stage of the Kannada language, known primarily from the inscriptional record going back to the 4th-5th century CE in Halmidi inscription from the reign of the Kadambas of Banavasi. The earliest known literary work in Kannada is the Kavirājamārga (9th century CE). This work describes Kannada poetics as well as grammar to some extent, citing earlier authorities and forms of the language. The Śabdamaṇidarpaṇa of Kēśirāja, dating to the 12th century CE, is the earliest systematic grammar of Kannada, which remains authoritative today.Sastri (1955), p359Narasimhacharya (1988), p19
Sanskrit grammars of Kannada also exist, drawing on the Sanskrit tradition of Pāṇinī. These include the Karṇāṭakabhāṣābhūṣaṇa of Nagavarma II, composed prior to Kēśirāja (12th century CE), and the Karṇāṭaka Śabdānuśāsana of Bhaṭṭākalaṅka Dēva was composed in the 16th century CE.A Grammar of the Kannada Language F.Kittel (1903), pp. 3 Later commentaries on Śabdamaṇidarpaṇa include the 16th century Sūtrānvayaratnamāle of Liṅgaṇārādhya and the 18th century commentary by Niṣṭhūra (or Niṭṭūru) Nañjayya. Despite being called "Old Kannada", this stage of Kannada was frequently revived in courtly literature, as the Śabdamaṇidarpaṇa was the standard for literary Kannada well into the early modern period. Even while Middle Kannada emerged, court poets would continue to use the morphology and style of Old Kannada to generate a "classical" style.
Origin
A 5th century copper coin was discovered at Banavasi with an inscription in the Kannada script, one of the oldest such coins ever discovered.
In a report published by the Mysore Archaeological Department in 1936, Dr. M. H. Krishna, (the Director of Archaeology of the erstwhile Mysore state) who discovered the inscription in 1936 dated the inscription to 450 CE. This inscription in old-Kannada was found in Halmidi village near Hassan district. Many other inscriptions having Kannada words had been found like the Brahmagiri edict of 230 BCE by Ashoka. But this is the first full scale inscription in Kannada. Kannada was used in the inscriptions from the earliest times and the Halmidi inscription is considered to be the earliest epigraph written in Kannada. This inscription is generally known as the Halmidi inscription and consists of sixteen lines carved on a sandstone pillar. It has been dated to 450 CE and demonstrates that Kannada was used as a language of administration at that time.K.V. Ramesh, Chalukyas of Vatapi, 1984, Agam Kala Prakashan, Delhi p10 Dr K.V.Ramesh has hypothesized that, compared to possibly contemporaneous Sanskrit inscriptions, "Halmidi inscription has letters which are unsettled and uncultivated, no doubt giving an impression, or rather an illusion, even to the trained eye, that it is, in date, later than the period to which it really belongs, namely the fifth century A.D." The original inscription is kept in the Office of the Director of Archaeology and Museums, Govt. of Karnataka, Mysore, and a fibreglass replica has been installed in Halmidi. A maṇṭapa to house a fibreglass replica of the original inscription has been built at Halmidi village. The Government has begun to promote the village as a place of historical interest.
Evidence from edicts during the time of Ashoka the Great suggests that the Kannada script and its literature were influenced by Buddhist literature. The Halmidi inscription, the earliest attested full-length inscription in the Kannada language and script, is dated to 450 CE while the earliest available literary work, the Kavirajamarga, has been dated to 850 CE. References in the Kavirājamārga indicate an earlier tradition of Kannada literature, including genres such the bedaṇḍe, cattāṇa and melvāḍu forms during earlier centuries.Narasimhacharya (1988), pp. 12, 17. The 5th century Tamatekallu inscription of Chitradurga, and the Chikkamagaluru inscription of 500 CE are further examples.Narasimhacharya (1988), p6Rice (1921), p13Govinda Pai in Bhat (1993), p102
Phonology
Old Kannada’s phonology is virtually identical to that of Modern Kannada, having five groups of "classed" (vargīya) sounds, comprising contrastive voice, aspiration, and nasal stops, along with "unclassed" (avargīya) sounds, which includes fricatives, approximants, and laterals. Vowels are distinguished as short (hrasva) and long (dīrgha). Kēśirāja and other grammarians classify certain sounds as native to Kannada and those shared with Sanskrit, though not all agree on whether aspirates or the vocalic liquids of Sanskrit (ṛ and ḷ, as well as their long counterparts) are considered native.Kull (1976), 46-64. The ಳ ⟨ḷ⟩ that occurs in native Kannada words (and which may alternate with ಲ in certain places) is called ಕುಳ (kuḷa), while the ಳ ⟨ḷ⟩ that occurs in Sanskrit words is called ಕ್ಷಳ (kṣaḷa). FrontCentralBackshortlongshortlongshortlongClose (ಇ) (ಈ) (ಉ) (ಊ)Mid (ಎ) (ಏ) (ಒ) (ಓ)Open (ಅ) (ಆ) LabialDentalAlveolarRetroflexPalatalVelarGlottalNasalm (ಮ)n (ನ)ṇ (ಣ)ñ (ಞ)ṅ (ಙ)StopVoicelessp (ಪ)t (ತ)ṭ (ಟ)c (ಚ)k (ಕ)Aspiratedpʰ (ಫ)tʰ (ಥ)ṭʰ (ಠ)cʰ (ಛ)kʰ (ಖ)Voicedb (ಬ)d (ದ)ḍ (ಡ)j (ಜ)g (ಗ)Breathybʰ (ಭ)dʰ (ಧ)ḍʰ (ಢ)jʰ (ಝ)gʰ (ಘ)Fricatives (ಸ)ṣ (ಷ)ś (ಶ)h (ಹ)Approximantv (ವ)y (ಯ)Rhoticr (ರ)~ṟ (ಱ)LateralApproximantl (ಲ)ḷ (ಳ)~ ḻ (ೞ) Kēśirāja also classifies certain consonants as śithila ("slack, loose"), namely r, l, ḷ, and ḻ, listing words where these sounds may be dropped or are mandatory.Kēśirāja (1926), sūtras 36-45, 59-60. Sounds that are considered "slack" are those which may be pronounced such that they lose their syllabic weight. For example, sorku ("madness") is ordinarily scanned as having a heavy and light syllable, but because the r is considered "slack", it may be scanned as two light syllables. This feature of Kannada is frequently used to meet different metrical requirements in verse (see Kannada prosody).
Old Kannada, like certain varieties of Modern Kannada, also exhibits nasality. Kēśirāja indicates that the sounds y, v, and l may be nasal or non-nasal, though he has no clear way of marking this in his commentary.Kēśirāja (1926), sūtra 16. Kulli remarks that these are merely allophonic variations of sounds that come from a neighboring nasal consonant.Kulli (1976), 68-69. However, B. Ramachandra Rao points out that the bindu or anusvāra, which is usually pronounced m at the end of words, varies with n word-finally and, m varies with v intervocalically (ex. māva, "mother's brother" also occurs as māma). This may explain why certain instances of y, v, and l are thought to be nasal, especially because certain cognates in other South Indian languages (such as Tamil) point to specific instances where nasality occurs. For example, nāmbu "we (inclusive)" in Tamil is cognate with the Kannada nāvu, which is sometimes pronounced with a nasal vowel. The word māva can also be similarly nasalized.
ಱ (ṟa) and ೞ (ḻa) There are also two letters used in Old Kannada that do not appear in Modern Kannada: ೞ ⟨ḻ⟩ and ಱ ⟨ṟ⟩, which are usually pronounced the same as ಳ ⟨ḷ⟩ and ರ ⟨r⟩, respectively. The letter ಱ ⟨ṟ⟩ is also known as the ಶಕಟರೇಫೆ (śakaṭa-rēphe, "the cart-shaped r"), for its resemblance to a cart. Similarly, the ೞ ⟨ḻ⟩ is also called ಱೞ (ṟaḻa, "the ṟa-shaped ḻa") because the two letters resemble one another. These names help to distinguish them (especially since their pronunciations are identical) from ಳ ⟨ḷ⟩ and ರ ⟨r⟩. These commonly appear in modern printed editions of classical texts and Old Kannada manuscripts, and it is not clear that these consonants were truly distinguished outside of writing. Kēśirāja describes the pronunciation of these letters as follows:Kēśirāja (1926), sūtra 18.
Kēśirāja gives a number of examples of improper alliterations (prāsa) to demonstrate that these sounds are pronounced differently.Ibid. In practice, many poets used these apparently improper alliterations, and in his Girijākalyāṇa, the poet Harihara (13th century CE) rejected the continued distinction of these letters. This suggests that these letters may not have been pronounced differently, or at least not sufficiently so to merit maintaining Kēśirāja's rule. This is even suggested by the fact that Kēśirāja teaches that r and ḻ may alliterate when form part of a conjunct consonant, and that there are instances where ḷ is preferred to ḻ.Kēśirāja (1926), sūtra 24-29.
Rao writes that it is not known whether ḻ and ḷ were still distinguished in speech by the time of the Pampa Bhārata (942 CE), but does suggest that all of these consonants were distinguished in minimal pairs, if only in writing.Rao (1972), 7-8. He also records a number of alternations in the Pampa Bhārata between r, l, ḷ, ḻ, and ṟ.Rao (1972), 24-25. J.S. Kulli similarly writes these letters seem to be regularly confused or switched, though he writes that r and ṟ were definitely distinct.Kulli (1976), 70-76. R. Narasimhacharya also points out that many native words in Kannada are distinguished only by these letters, evidenced by their parallels with Tamil and historical spellings in Telugu.
Grammar Old Kannada nouns are marked for case and plurality. In some cases, nouns must first be formed as an oblique stem by adding an augment (āgama). Gender is not explicitly marked on nouns unless they mark animate beings, which is to say humans or personified figures (ex. deities). Sanskrit nouns, despite having lexical or grammatical gender, are declined as though they were neuter nouns. Semantically, however, they are treated according to their original gender when it marks a human being, and this is usually only apparent if there is a finite verb in the third person (where gender is marked).
B. Ramachandra Rao describes six noun classes based on syllable structure and other characteristics in his descriptive grammar of the Old Kannada work, the Pampa Bhārata.Rao (1972), 47-70. It is worth noting that Kēśirāja does not systematically describe which nouns receive which augment, instead just listing broad categories of nouns by their augment.Kēśirāja (1920), sūtras 105-112. SubtypeCharacteristicsOblique augmentExampleClass I1aRoots that take gender-number markers, including third person pronoun roots ava- and āta-Øtamma(n) ("younger brother") ava(ḷ) ("she")
arasa(r) ("king(s)")
bandhu(gaḷ) ("relatives")1bRoots with sonorant finals -ṇ, -n, y, l, and ḷ; these double following a short vowelkāl ("leg") kaṇṇ ("eye")1cInanimate nouns with disyllabic roots with final -i and -egiḷi(y) ("parrot") tale(y) ("head")Class IIRoots ending in -u or -ū -inkāḍu ("forest") -> kāḍin hū ("flower") -> hūvinClass IIISanskrit nouns and inanimate, native Kannada nouns with final -a -v/dmēgha(m)/mēghad("cloud") mara(m)/marad ("tree")Class IVNative Kannada nouns indicating directions and location -aṇkeḷage ("down/below") -> kelagaṇ mūḍu ("east") -> mūḍaṇClass VRoots denoting number; the oblique of the interrogative ēnu forms in this class -aṟondu -> ondaṟ ēnu -> ētaṟClass VIPronominal roots in the first and second personshort and long roots in different casesnān ("I") ~ nan(n) nāv ("we") ~ nam(m)
Pronouns
Old Kannada has three classes of pronouns: personal, demonstrative, and interrogative. There are no relative pronouns – relative clauses are formed affixing verbal participles to pronouns. Personal pronouns included the first and second person, as well as reflexives, forming their own class of oblique stem, where some cases use a "strong" stem with a doubled final nasal, and others use a "weak" stem with a single nasal. In Old Kannada, variants of the first person pronouns without the initial n existed before being supplanted by the nasal-initial form (not unlike Old Tamil yān "I" being replaced by nān). However, the oblique stem continues to reflect the existence of the vowel-initial form.
+NominativeOblique I ān/nān en(n)- we ām/nām nam(m)- you (sg.) nīn- nin(n)- you (pl.) nīm- nim(m)- self (sg.) tān tan(n)- self (pl.) tām tam(m)-