Early Tamil Epigraphy: From the Earliest Times to the Sixth Century A.D. c22: 3. Language

Early Tamil Epigraphy: From the Earliest Times to the Sixth Century A.D. c22

Language: English

c22.s1 3.1 Language of cave inscriptions — 5 parts
c22.s2 3.2 Dravidian element — 4 parts
c22.s3 3.3 Indo-Aryan element — 7 parts

3. Language

3.1 Language of cave inscriptions

3.1.1 The unsolved problem

c22.s1.t1.1

The pioneering studies, especially by Subrahmanya Aiyer on the language of the cave inscriptions, have established that the language is basically Tamil although with an admixture of Prakrit. However, Subrahmanya Aiyer's own readings led to another controversy: How much of Tamil is present in the inscriptions and what kind of Tamil? In the words of Kamil Zvelebil:

c22.s1.t1.2

How is it possible that there is such a vast difference between this kind of Tamil employed in these records, and the more or less contemporary or/and only slightly later literary Tamil language (*centamiḻ*) of *Tolkāppiyam* and the Sangam classics? This was a problem left unsolved by K.V.S. Ayyar.¹

3.1.2 How much Tamil?

c22.s1.t2.1

Before we look at the problem of the quality of the language of the cave inscriptions, it will be instructive to examine quantitatively the presence of Tamil and Prakrit in it.

c22.s1.t2.2

To the meagre number of four or five Tamil words already recognised by Krishna Sastri, Subrahmanya Aiyer could add only a dozen more—surprisingly few, considering the major paradigm shift effected by his own study (Table 3.¹).

c22.s1.t2.3

Tamil words in the cave inscriptions as read by -

c22.s1.t2.4

Krishna Sastri (1919):²

c22.s1.t2.5

² H. Krishna Sastri 1919:p. 346.

c22.s1.t2.6

| -*uṭai* #, *ūr*, *ēri*, *kuṭu-* / *koṭu-*, *nāṭu* | 5 |

c22.s1.t2.7

\# Incorrectly read.

c22.s1.t2.8

Subrahmanya Aiyer (1924):³

c22.s1.t2.9

³ K. V. Subrahmanya Aiyer 1924:pp. 287-299.

c22.s1.t2.10

c22.s1.t2.11

List of additional words only.

c22.s1.t2.12

| Total | 17 |

c22.s1.t2.13

Table 3.¹. Earliest readings of Tamil words in cave inscriptions (up to 1924).

c22.s1.t2.14

¹ Kamil Zvelebil 1966:pp. 18-19.

c22.s1.t2.15

Kamil Zvelebil compiled a glossary of simple stems occurring in the inscriptions as read by him and computed the relative frequency of Dravidian and Indo-Aryan as follows:¹

c22.s1.t2.16

¹ Kamil Zvelebil 1966: p. 41; see also Index Verborum at pp. 36-41.

c22.s1.t2.17

c22.s1.t2.18

(† Including 6 Dr. and 6 IA additional stems from the later Sittannavasal II inscriptions, which, however, do not materially alter the picture.)

c22.s1.t2.19

Zvelebil reached the following conclusion on the basis of the statistical data compiled by him:

c22.s1.t2.20

The ratio of 47: 43 for IA: Dr. entitles us fully to characterize the language of the records under scrutiny as a hybridized form of language. The grammar of the inscriptions is, however, fully Tamil . . . This, as well as the distribution of lexical items, shows that the language of the records has rightly been characterized as hybrid Tamil. ²

c22.s1.t2.21

² Ibid. p. 42.

3.1.3 What kind of Tamil?

c22.s1.t3.1

It is, therefore, hardly surprising that even while scholars hailed Subrahmanya Aiyer's effort, his readings were treated with reserve.³ The 'defective' orthography, as it was perceived at the time, drew responses in opposite directions as illustrated by the following quotations:

c22.s1.t3.2

³ "They (cave inscriptions) are yet to be satisfactorily read and interpreted." T.N. Subramanian 1957: pp. 1506-1507. The scepticism, even if it is somewhat less at present, still persists; e.g., "They (cave inscriptions) are too short, fragmentary, and not yet quite completely intelligible. . . " Kamil Zvelebil 1990: p. 51.

c22.s1.t3.3

Accepting, in the absence of a more satisfactory explanation, that these Brāhmī documents represent oldest specimens of the Tamil language that we have, we find that the orthography was halting and defective, suggesting that the Tamil language during the 3rd century B.C. was being reduced to writing for the first time.⁴

c22.s1.t3.4

⁴ Sunitikumar Chatterji 1956: pp. 149-150.

c22.s1.t3.5

The legitimate inference seems to be that these votive inscriptions are in a hybrid language containing Tamil as well as Prākṛit words . . . the strange jumble of words belonging to two different languages . . . It is of supreme importance, therefore, to remember that these epigraphs are not of great value to the study of linguistic development.⁵

c22.s1.t3.6

⁵ K.K. Pillay 1956: pp. 178-179.

c22.s1.t3.7

In a short chapter on the language of the cave inscriptions in his book *A History of Tamil Language*, Meenakshisundaran carried out a brief linguistic analysis of the published readings. According to him:

c22.s1.t3.8

The language is something like Tamil or the Proto-South Dravidian. But it must be noted that these inscriptions are in a hybrid language containing Tamil as well as Prakrit words because the Buddhists, who were great scholars in Prakrit and not themselves native speakers of Tamil, had these inscriptions cut in the rocks so as to be understood by the people of the region.¹

c22.s1.t3.9

¹ T.P. Meenakshisundaran 1965: pp. 41-42.

c22.s1.t3.10

Even though Mahalingam made an independent study of the cave inscriptions,² his conclusions were not materially different from those of other scholars of the period:

c22.s1.t3.11

The basically Tamil language used in these inscriptions was different from the Tamil language used in the Sangam literary works and contains good proportion of Prākṛit words. Probably the view that the language employed in many of these lithic records "is Tamil in its formative stages" may not be quite easy of acceptance though it betrays much unsteadiness in its grammatical and other features showing that "it was less homogeneous and regular in its characteristics". Its difference from the literary Tamil from the point of style was obviously due to the fact that "those who have been responsible for the records in question were obviously Buddhist and/or Jaina (and Ājīvika? - T.V.M) monks using a hybrid jargon with great admiration for Pāli and/or Prakrit" ... The language of these inscriptions is that of these religious teachers and their followers in the Tamil country.³

c22.s1.t3.12

³ Ibid. p. 157. The first citation "Tamil in its formative stages" is from K.A. Nilakanta Sastri, *History of South India*, 3rd edition 1966: p. 87. The other citations in this quotation are from Kamil Zvelebil 1966: pp. 48-50.

3.1.4 The language of the inscription is Old Tamil

c22.s1.t4.1

Recent advances in our knowledge of the language of the cave inscriptions summarised in Chapter 2 have rendered the earlier views based on unsatisfactory readings obsolete. The argument of the present study is that starting from accurately copied texts,⁴ and applying the orthographic rules which can be empirically formulated for reading the texts,⁵ it can be demonstrated that the language of the cave inscriptions, despite the presence of Prakrit loanwords, is Old Tamil, not materially different from the language of later Tamil inscriptions or even literary texts, in its basic phonological, morphological and syntactical features.⁶

c22.s1.t4.2

⁴ See section 2.15 on the procedures followed to procure accurate copies of the cave inscriptions.

c22.s1.t4.3

⁵ See Chapter 6 (Orthography).

c22.s1.t4.4

⁶ See Chapter 7 (Grammar).

3.1.5 Dravidian and Indo-Aryan in the inscriptions: the true proportion

c22.s1.t5.1

The extent of change in our understanding of the language of the cave inscriptions may be quantified (Table 3.²) on the basis of the data compiled from the revised and enlarged Corpus included in this volume.

c22.s1.t5.2

² T.V. Mahalingam 1967: pp. 201-298.

c22.s1.t5.3

The Corpus comprises 110 inscriptions, 89 in Tamil-Brāhmī and 21 in Early Vaṭṭeḻuttu. There are 371 simple stems in the Corpus; 256 stems are Dravidian, 101 stems are Indo-Aryan, and the remaining 14 are of doubtful etymology.⁷ Taking the Tamil-Brāhmī inscriptions (Nos. *1-89*) alone into account, there are 307 stems out of which 213 stems are Dravidian, 81 stems are Indo-Aryan, and the remaining 13 are of doubtful etymology. In brief, the proportion of Dravidian: Indo-Aryan in the Tamil-Brāhmī inscriptions has increased dramatically from less than 1:1 as computed by Zvelebil in 1964 to more than 2.5:1 at present; only about one-fourth (in numbers as well as frequency) of the simple stems in the Tamil-Brāhmī inscriptions in the Corpus are of Indo-Aryan origin. It is significant that the proportion remains about the same even when the Early Vaṭṭeḻuttu inscriptions in the Corpus are included. These findings disprove the earlier view that the language of the cave inscriptions is a 'hybrid jargon' dominated by Indo-Aryan.

c22.s1.t5.4

⁷ See Etymological Indexes (Appendices IV to VI) for complete word lists.

c22.s1.t5.5

| Stems in the Corpus | Dravidian (Appendix IV) No. | Dravidian (Appendix IV) % | Indo-Aryan (Appendix V) No. | Indo-Aryan (Appendix V) % | Doubtful Etymology (Appendix VI) No. | Doubtful Etymology (Appendix VI) % | Total No. | Total % |
| In Tamil-Brāhmī Inscriptions (*1-89*): | | | | | | | | |
| Stems | 213 | 69.38 | 81 | 26.38 | 13 | 4.24 | 307 | 100.00 |
| Frequency | 339 | 71.67 | 118 | 24.95 | 16 | 3.38 | 473 | 100.00 |
| In Early Vaṭṭeḻuttu Inscriptions (*101-121*): | | | | | | | | |
| Stems | 43 | 67.19 | 20 | 31.25 | 1 | 1.56 | 64 | 100.00 |
| Frequency | 61 | 71.16 | 22 | 25.88 | 2 | 2.36 | 85 | 100.00 |
| Total for the Corpus: | | | | | | | | |
| Stems | 256 | 69.00 | 101 | 27.23 | 14 | 3.77 | 371 | 100.00 |
| Frequency | 400 | 71.68 | 140 | 25.09 | 18 | 3.23 | 558 | 100.00 |

c22.s1.t5.6

Table 3.2. Relative frequency of Dravidian and Indo-Aryan stems in the Corpus.

c22.s1.t5.7

Dry statistics do not, however, tell the whole story. To convey the 'feel' of the language of the Tamil-Brāhmī inscriptions, a rapid overview of their main lexical and grammatical features, dealing with the Dravidian and Indo-Aryan elements separately, is given below. Detailed documentation has been avoided in this summary as the topics are dealt with more fully in Chapter 7 (Grammar) and in the Commentary on the inscriptions. See also Inscriptional Glossary (Appendix I) and word-lists (Appendices II-VI).

3.2 Dravidian element

3.2.1 Phonemic inventory

c22.s2.t1.1

There is near one-to-one correspondence between the phonemic inventory of Old Tamil and Tamil-Brāhmī; the only exceptions are the absence of the vowel *au* and the *āytam* (*ḵ*) from the inscriptions.

3.2.2 Lexical items

c22.s2.t2.1

(i) *Verbs

c22.s2.t2.2

The Tamil-Brāhmī inscriptions are relatively richer in verbs when compared to similar brief votive inscriptions in Prakrit. 15 verb stems are attested in the inscriptions, all of which are Dravidian in origin.

c22.s2.t2.3

(e.g.) *aṟu* 'to cut', *āku* 'to become', *ī* 'to give', *ēvu* 'to command', *koṭu* 'to give', *cey* 'to make', *toṭu* 'to excavate', *piṟa* 'to be born', *puṇaru* 'to compose' and *vai* 'to endow'.

c22.s2.t2.4

Verb forms occurring in the Tamil-Brāhmī inscriptions belong mainly to the following categories:

c22.s2.t2.5

intransitive, transitive and causative forms; past and non-past tenses; participial and verbal nouns; adjectival participles and the infinitive.

c22.s2.t2.6

(ii) *Adjectives

c22.s2.t2.7

Reflecting the Old Tamil tradition of prefixing attributes to personal and place names, the Tamil-Brāhmī inscriptions possess a rich inventory of adjectives.

c22.s2.t2.8

(e.g.) *irum* 'great', *iḷa(m)* 'young', *kaṭu(m)* 'ferocious', *kuṟu(m)* 'short', *ciṟu* 'small', *cem* 'good', *nal* 'good', *neṭu(m)* 'great', *peru(m)* 'large, great' and *veḷ* 'white, pure'.

c22.s2.t2.9

(iii) *Nouns

c22.s2.t2.10

(e.g.) *ayam* 'tank', *aṟam* 'charity', *aṟuvai* 'cloth', *u(p)pu* 'salt', *uṟai* 'abode', *ū(ṟ)ṟu* 'spring', *eṇṇai* (LT *eṇṇey*) 'oil', *eḻuttu* 'letter', *ēri* 'lake', *kaya(m)* 'pond', *karaṇṭai* 'cave', *kal* 'stone', *kala(m)* 'a measure', *koḻu* 'ploughshare', *cuṉai* 'spring', *nel* 'paddy', *paḷḷi* 'hermitage', *paṉai* 'palm (tree)', *poṉ* 'gold', *yāṟ(u)* 'river' and *vēmp(u)* 'neem (tree)'.

c22.s2.t2.11

(iv) *Toponyms

c22.s2.t2.12

Toponyms commonly occurring with place names in Old Tamil are well represented in the inscriptions.

c22.s2.t2.13

(e.g.) *aṟai* 'rock', *il* 'place', *ūr* 'village' (frequent), *kāṭu* 'jungle', *kuṉṟu* 'hill', *talai* 'place', *tuṟai* 'waterfront', *teṉku* 'southern', *nāṭu* 'country', *pa(ḷ)ḷi* 'hamlet', *malai* 'mountain', and *veḷi* 'field'.

c22.s2.t2.14

(v) *Rare lexical items

c22.s2.t2.15

The Tamil-Brāhmī cave inscriptions are the earliest lithic records in Dravidian. It is, therefore, not surprising that there are words and usages in the inscriptions which are not found even in the earliest layer of Old Tamil. It is, in my view, a mistake to dismiss these features as 'scribal errors'. Some of the rare usages are attested in later Tamil inscriptions indicating their presence all along in the spoken language though not represented in the earlier literary records. Some of the more interesting items not attested in Old Tamil are mentioned below.¹

c22.s2.t2.16

¹ Listed in Etymological Index: Dravidian (Appendix IV). See Commentary on the relevant entries for discussion.

c22.s2.t2.17

*-aṇṇi*: honorific suffix (fem.). The dictionaries describe the word as 'local'. However, the word is undoubtedly ancient as it has been borrowed (along with the corresponding masculine honorific suffix *-aṇṇa*) in the early Prakrit inscriptions of South India.

c22.s2.t2.18

*antai*: a frequent honorific affix (masc.). It occurs also as an independent word in the Tamil-Brāhmī inscriptions and once in a later Tamil inscription, but only as a bound suffix in Literary Tamil. *antai* is arguably the most significant of the 'lost' words in Old Tamil recovered through the decipherment of the Tamil-Brāhmī inscriptions. This has led to two important results in linguistic and historical studies:

c22.s2.t2.19

(a) *antai* as an independent word appears to be the primary kinship term from which cognates like *entai*, *tantai*, etc., are derived systematically by the addition of pronominal prefixes.

c22.s2.t2.20

(b) *-antai* as a bound suffix is seen to be merely honorific (like *-ayyaṉ* or *-appaṉ*) and does not literally mean 'father of' as interpreted in Tamil grammatical tradition.¹

c22.s2.t2.21

¹ See in particular the parallel passages in inscriptions Nos. 65 & 66 and the discussion on 66.2 in the Commentary for clinching evidence in support of the proposed reinterpretation.

c22.s2.t2.22

*-a(p)pa-*: honorific suffix (masc.). The word is attested in later works from the time of *Tēvāram*.

c22.s2.t2.23

*karaṇtai*: 'cave'. The word is listed in later *nikaṇṭus* as *karaṇtai* or *kāṇtai*. The occurrence of *karaṇtai* in Tamil-Brāhmī indicates that it is the correct form and that *kāṇtai* is incorrect reading caused by the near-identical shapes of *ra* and *-ā* in the medieval Tamil script.

c22.s2.t2.24

*karu-iya*: 'which was carved'; adjectival past participle from **karu(v)u* 'to carve'. The verb is clearly related to the nouns *karu* 'mould' (Ta.), 'embossed work, bas-relief' (Ka.) and *karukku* (inscr.) 'bas-relief'.

c22.s2.t2.25

*kolvaṉ* occurs in *poṉ-kolvaṉ* 'goldsmith'. The word appears to be a variant of the literary form *kollaṉ* as indicated by analogous pairs like *kaḷḷaṉ* / *kaḷvaṉ*, *cellam* / *celvam*, etc.

c22.s2.t2.26

*cē-*: 'to make' (LT *cey*). The verb occurs in the forms *cē-iya*, *cēy-a*, *cētavar* and *cēvit-ōṉ*. The *cē-* forms occur fairly commonly in Tamil inscriptions.² In my view, it is not necessary to 'correct' the inscriptional forms like *cēta* as *ce(y)ta* and *cēvittāṉ* as *ce(y)vittāṉ*.

c22.s2.t2.27

² A. Velu Pillai 1976: pp. 5, 143 (examples from inscriptions of 6-9 centuries A.D.)

c22.s2.t2.28

*muḷa-ukai* and *muḷākai*: 'cave'. The context and the obvious resemblance to *muḻai* 'cave' makes the meaning of the two words clear. It is significant that when the nominal suffix *-kai* is removed, we get the forms *muḷa(v)u* / *muḷā*,³ following the known pattern of pairs like *nilavu* / *nilā*, *palavu* / *palā*, etc.

c22.s2.t2.29

³ These words meaning 'cave' are cognate with *muḻai* (D. 4994), but are to be distinguished from *muḻā* / *muḻavu* 'drum' (D. 4989).

c22.s2.t2.30

*muṉṟu*: 'front yard'. The word is clearly related to the literary form *muṉṟil* with the same meaning. The recovery of the basic form *muṉṟu* makes it possible to analyse *muṉṟil* as *muṉṟ(u)* + *il*, rather than as *il-muṉ* in the traditional, but rather *ad hoc*, derivation.

3.2.3 Grammatical elements

c22.s2.t3.1

(i) *Pronouns* (e.g.) Personal pronouns: *em* 'our' (exclusive). Demonstrative pronominal bases: *a-* 'that'; *i-* 'this'. Numeral pronouns: *iruvar* 'two persons'.

c22.s2.t3.2

(ii) *Numerals* (e.g.) *iru-* 'two'; *patiṉ-* 'ten' (oblique); *nūṟu* 'hundred'.

c22.s2.t3.3

(iii) *Interjections* (e.g.) *ītā* 'lo, behold!'

c22.s2.t3.4

(iv) *Particles* (e.g.) *-um* 'also, too'; *-ē* 'emphatic particle'

c22.s2.t3.5

(v) *Grammatical morphemes*

c22.s2.t3.6

There are about 38 types of grammatical morphemes in the Tamil-Brāhmī inscriptions including pronominal (PNG) suffixes, suffix of the oblique case, case-endings, causative suffixes, tense markers, suffixes of the participial and verbal nouns, suffixes of the adjectival participles and suffix of the infinitive.¹

c22.s2.t3.7

¹ See section 7.38 and Appendix VII (Index to Grammatical Morphemes) for details.

c22.s2.t3.8

(vi) *Rare grammatical usages*

c22.s2.t3.9

(a) *Absence of finite verbs:*

c22.s2.t3.10

A remarkable feature of the Tamil-Brāhmī inscriptions is the absence of finite verbs, their place being taken by participial nouns or adjectival participles. (e.g.) *antuvaṉ koṭupitavaṉ* 'Antuvaṉ caused to be given'. *(i)ḷaṅkō ceypita paḷi-i* 'hermitage caused to be made by (I)ḷaṅkō'.

c22.s2.t3.11

(b) *Absence of present tense:*

c22.s2.t3.12

The Tamil-Brāhmī inscriptions have only two tenses, the past and the non-past, and no present tense, agreeing in this respect with Old Tamil. The present tense in Tamil is a later development.

c22.s2.t3.13

(c) *Rare grammatical morphemes:* *-a* occurs as the accusative case suffix in *itta* 'this' (LT *ittai*). *-a* is the most common genitive suffix in the Tamil-Brāhmī inscriptions. (e.g., *ciḻivaṉ-a*, *tiṭi-il-a*, etc.). This suffix occurs only rarely in Old Tamil. *-ā* occurs as a genitive case suffix. (e.g.) *nalliy-ūr-ā* 'of Nalliyūr'.

c22.s2.t3.14

This morpheme is not attested elsewhere in Tamil. However, it occurs as a genitive suffix in Old Kannada inscriptions.

c22.s2.t3.15

*-ā* occurs as an adjectival suffix in *mutā*/*mūtā* (*mutu*/*mūtu*+*ā*) 'senior, elder'. This morpheme is also not attested in Tamil.

3.2.4 Influence of Old Kannada

c22.s2.t4.1

The present study has brought to light the hitherto unsuspected influence of Old Kannada on the Tamil-Brāhmī inscriptions from a period (ca. 2nd century B.C.-4th century A.D.) anterior to the earliest known Kannada inscriptions and literature. See especially No. **49** from Sittannavasal-A (ca. 1st century B.C.), No. **55** from Tirupparankunram (ca. 1st century A.D.), No. **79** from Edakal-A (ca. 3rd century A.D.) and No. **83** from Nekanurpatti (ca. 4th century A.D.). Some of the more interesting lexical items and grammatical usages showing the influence of Old Kannada are listed below.¹

c22.s2.t4.2

¹ See also Commentary on the inscriptions cited in this paragraph.

c22.s2.t4.3

(i) *Lexical items

c22.s2.t4.4

*erumināṭu* is almost certainly the same as LT *erumaināṭu*, the Mysore region (*mahiṣha-maṇḍala*) of Karnataka. The word *erumi* (Ta. *erumai*, Ka., Tu. *erme*, Go. *ermi*) 'buffalo' appears to preserve an ancient dialectal form.

c22.s2.t4.5

*kavuṭi* is the personal name of the nun who is described in the inscription as born in a village in *erumināṭu*. cf. Ka. *gavuḍi*, *gauḍi* 'feminine of *gauḍa*, wife of a *gāvuṁḍa*, wife of a village officer'.

c22.s2.t4.6

*pocil* 'entrance'. The expression occurs as part of the place name *teṉku-ciṟu-pocil* which appears to be the same as *teṉ-ciṟu-vāyil*, mentioned in the later Tamil inscriptions of the region as the name of the *nāṭu* (territorial division) immediately to the east of the hill at Sittannavasal. *pocil* is not attested in Tamil and appears to be related to Ka. *hosilu* (< *\*posil*) 'entrance'. cf. also To. *pōş* 'entrance'.

c22.s2.t4.7

*tāyiyaru* 'mother' (honorific). This is clearly a loanword from Kannada. The word occurs in New Kannada but is not attested in Old Kannada. However, as this inscription from Nekanurpatti is assigned to ca. 4th century A.D. on palaeographic evidence, we have to regard *tāyiyaru* as an Old Kannada word which existed in the spoken language but was not attested in contemporary records.

c22.s2.t4.8

(ii) *Personal names and honorifics

c22.s2.t4.9

Personal names like *āy(c)ca* and *polāl(a)*, and the suffixed honorifics *aṇṇi*, *a(p)pa*- and *a(y)yaṉ* / *aiyaṉ* appear to be more at home in Old Kannada onomastics.

c22.s2.t4.10

(iii) *Grammatical usages

c22.s2.t4.11

*-ā*, occurring as the genitive suffix in some inscriptions is not attested in Tamil, but occurs in Old Kannada inscriptions where it is regarded as more ancient than *-a*.²

c22.s2.t4.12

² See section 7.27.4 for details.

c22.s2.t4.13

*-u*, the euphonic suffix to stems ending in liquid consonants (e.g., *ūru* 'village') occurring mostly in the later inscriptions, is not attested in Old Tamil and appears to be due to the influence of Old Kannada, even though the suffix is attested in Kannada inscriptions only from the middle period (ca. 8th century A.D.).

c22.s2.t4.14

*(iv) Spread of Jainism from Karnataka*

c22.s2.t4.15

The presence of Old Kannada elements in the Tamil-Brāhmī inscriptions corroborates the traditional account of the spread of Jainism from Karnataka into the Tamil country.¹

c22.s2.t4.16

¹ See sections 4.8 to 4.13.

3.3 Indo-Aryan element

3.3.1 Phonology: presence of non-Tamil sounds

c22.s3.t1.1

Only two non-Tamil sounds (*dh* and *s*) occur marginally in loanwords in the Tamil-Brāhmī inscriptions.² It is significant that these are confined only to the Early Tamil-Brāhmī Period (ca. 2nd century B.C. to 1st century A.D.). Actually, the number and relative frequency of non-Tamil characters in the Tamil-Brāhmī inscriptions are far less than those of Grantha letters in the later Tamil inscriptions.

c22.s3.t1.2

² See section 5.5 for details. The rarity of non-Tamil sounds in the Tamil-Brāhmī inscriptions is by itself a sufficient ground to call into question the earlier theory of a predominant Indo-Aryan element in them.

3.3.2 Orthography

c22.s3.t2.1

(i) *Adaptation to Tamil phonetic pattern*

c22.s3.t2.2

The proportion of non-Tamil sounds is relatively much less than what one would expect from the Indo-Aryan element present in the Tamil-Brāhmī cave inscriptions as most of the loanwords are adapted to the Tamil phonetic pattern.³

c22.s3.t2.3

³ See section 6.21 for details.

c22.s3.t2.4

(ii) *Use of single consonants for geminates*

c22.s3.t2.5

The employment of single consonants in the Tamil-Brāhmī inscriptions to represent the doubled consonants in the language is due to the influence of Prakrit inscriptional orthography. This orthographic feature is seen not only in loanwords but also in native Tamil words. However, unlike in Prakrit inscriptions, the use of single for doubled consonants was optional in the Tamil-Brāhmī inscriptions even from the earliest times both in loanwords and native words.⁴

c22.s3.t2.6

⁴ See section 6.17 for details.

3.3.3 Vocabulary

c22.s3.t3.1

The IA loanwords in the Tamil-Brāhmī inscriptions are all nouns.⁵ Apart from personal names, the vocabulary comprises mostly religious or cultural terms. Some of the more interesting items are briefly noticed below.

c22.s3.t3.2

⁵ See the Indexes in Appendices V and VI for word lists and the Commentary on the entries for discussion.

c22.s3.t3.3

(i) *Religious terms*

c22.s3.t3.4

Religious terms especially associated with the Jaina religion are discussed in the sections on Jainism.⁶ The following are some of the other religious terms common to Indian religious tradition.

c22.s3.t3.5

⁶ See sections 4.8 to 4.13.

c22.s3.t3.6

*atiṭṭāṉam* (Pkt. *adhiṭṭhāna*): lit., 'permanent fixed abode' of the itinerant monk who stayed in the cave during the rainy season. The term occurs in the Tamil-Brāhmī inscriptions in the special sense of 'stone beds' carved on the floor of the cave for the use of the monks.

c22.s3.t3.7

*tāṇa* (Pkt. *dāna(ṁ)* 'religious gift').

c22.s3.t3.8

*dhammam* / *dhamam* (Pkt. *dhamma*) 'charity, religious endowment'.

c22.s3.t3.9

(ii) *Cultural terms*

c22.s3.t3.10

These comprise mostly titles and names of some professions:

c22.s3.t3.11

*kaṇatikaṉ* 'chief of scribes'; cf. Sinh. Pkt. *kaṇa* 'scribe', Pkt. *adhika* 'superintendent'.

c22.s3.t3.12

*kaṇaka* (Pkt. *gaṇaka*) 'accountant'.

c22.s3.t3.13

*kāviti* 'an ancient title bestowed on Vaiśyas and court officials'. cf. Sinh. Pkt. *gapati* / *gapiti* < Pkt. *gahapati* < Skt. *gṛhapati*, lit., 'householder', a title of merchant classes and landed gentry.

c22.s3.t3.14

*kuṭumpikaṉ* (Pkt. *kuṭumbika*) 'householder, husbandman'.

c22.s3.t3.15

*ta(c)caṉ* (Pkt. *taccha*) 'carpenter'. The term occurs in the Tamil-Brāhmī inscriptions with the meanings 'carpenter, stonemason'.¹

c22.s3.t3.16

¹ However, see Commentary (73.8) for possible alternative Dravidian etymology.

c22.s3.t3.17

*māraya-* (read *mārāya*) 'an ancient title', probably from Pkt. *mahārāya*.

c22.s3.t3.18

*vaṇikaṉ* / *vāṇikaṉ* (Pkt. *vaṇija* / *vāṇija* < Skt. *vaṇik*) 'trader'.

c22.s3.t3.19

*siri* (Pkt. *siri* < Skt *śrī* ) 'an auspicious affix to personal names'.

c22.s3.t3.20

(iii) *Miscellaneous loanwords from Prakrit*

c22.s3.t3.21

Selected examples are listed below.

c22.s3.t3.22

*āci* (Pkt. *āsi*) 'support, prop'. cf. Ta. *ācu* 'support'. The inscriptional context indicates the meaning of a wooden 'support' for a canopy in front of the cave.

c22.s3.t3.23

*kaṭikai* (cf. Skt. *ghaṭikā*) 'an assembly of learned persons or institution of higher learning or the place of such assembly or institution'.

c22.s3.t3.24

*nikama-* (Pkt. *nigama*) 'merchant guild'.

c22.s3.t3.25

*pāṇita-* (Pkt. *phāṇita*) 'juice of sugar cane, candy, molasses'.

c22.s3.t3.26

*sālakaṉ* (Pkt. *sālaka*) 'brother-in-law', but occurs in the Tamil- Brāhmī inscription with the meaning 'sister-in-law's husband' as indicated by the context (cf. Ta. *cālakaṉ*).

3.3.4 Prakrit grammatical usages

c22.s3.t4.1

While the grammar of the Tamil-Brāhmī inscriptions is almost wholly Tamil, there are a couple of Prakrit grammatical forms occurring in them.

c22.s3.t4.2

*-ō*: Nominative case-ending of masculine singular noun in *-putō* 'son' in *satiyaputō*.

c22.s3.t4.3

*-sa*: Genitive case-ending of masculine singular noun in *utayaṇasa* 'of Utayaṇaṉ'.²

c22.s3.t4.4

² See No. 1, Table 1.6 for the occurrence of the Pkt. genitive ending *-sa* in a coin-legend.

c22.s3.t4.5

The relatively higher frequency of causative verb forms in these inscriptions when compared with Literary Tamil appears to be due to the influence of inscriptional Prakrit.¹

c22.s3.t4.6

(e.g.) *koṭupitōṉ* 'he who caused to be given'.

3.3.5 Prakrit of the cave inscriptions

c22.s3.t5.1

Prakrit, in the widest sense of the term, comprises all Middle Indo-Aryan (MIA) languages including those recorded in the inscriptions of Asoka and the post-Mauryan period, the canonical languages of the Buddhist and Jaina religions, and the Literary Prakrits found in the Sanskrit plays. An attempt has been made in the present study to determine the type of Prakrit from which the loanwords in the cave inscriptions are derived.²

c22.s3.t5.2

(i) *Loanwords not derived from Asokan dialects*

c22.s3.t5.3

Most of the inscriptions of Asoka are in the Eastern Dialect which was the Mauryan court language. However, such distinctive features of the Eastern Dialect as the substitution of *r* with *l* and the use of the nominative singular ending *-ē* are not found in the loanwords. It is now generally accepted that the Tamil caves were occupied by the Jaina monks and have no Buddhist vestiges.³ Thus, the available evidence indicates that the loanwords in the cave inscriptions are not derived from the language of the Asokan inscriptions, though they share some features common to most dialects.

c22.s3.t5.4

(ii) *Loanwords mostly from standard epigraphic Prakrit*

c22.s3.t5.5

The data compiled in the Etymological Indexes (Appendices V-VI) indicate clearly that most of the loanwords in the Corpus are derived from standard epigraphic Prakrit based on the dialect of the central and western regions. The circumstances in which this dialect rose to become the common epigraphical language for almost the whole of India in the post-Mauryan period are well brought out by Richard Salomon in his recent book *Indian Epigraphy*:

c22.s3.t5.6

The causes of the abrupt dialectal shift from east to west undoubtedly lie in political and historical developments, that is, the decline of Magadha as the center of power in northern India after the collapse of the Mauryan empire and the movement of the center of political power in the following centuries toward the west and northwest. Like the eastern dialect under Aśoka, the central-western dialect of the post-Mauryan era was used far beyond what must have been its original homeland. Thus we find inscriptions in this standard epigraphic Prakrit as far afield as Orissa in the east, for instance, in the Hāthīgumphā inscription, while

c22.s3.t5.7

The distinguishing features of standard epigraphic Prakrit as reflected in the loanwords in the Corpus are as follows.

c22.s3.t5.8

i) Nominative masculine singular *-ō* (as in *satiyaputō*).

c22.s3.t5.9

ii) Retention of the distinction between *r* and *l* (as in *āritaṉ*, *sālakaṉ*, etc.).

c22.s3.t5.10

iii) The exclusive use of the sibilant *s*.

c22.s3.t5.11

However, the loanwords are radically transformed by the local language, namely Tamil, and adapted to its phonetic pattern as discussed earlier. Further, there are also minimal influences on the loanwords from two other dialects of Prakrit as noticed below.

c22.s3.t5.12

(iii) *Loanwords from Jaina Prakrit (Ardhamāgadhī)*

c22.s3.t5.13

Considering the exclusive association of Jaina monks with the Tamil caves, it is somewhat surprising that the influence of Ardhamāgadhī (AMg.), the language of the earliest Jaina canonical literature, is minimal on the loanwords occurring in the inscriptions. From among more than a hundred IA stems, only five may be regarded as derived from AMg. based on the special characteristics of that dialect (Table 3.³). The first four loanwords occur in the Tamil-Brāhmī inscriptions, and the last in an Early Vaṭṭeluttu inscription.

c22.s3.t5.14

(iv) *Loanwords from Sinhala-Prakrit* While most of the loanwords in the Tamil-Brāhmī cave inscriptions are naturally from Indian Prakrits, there are a couple of items for which Sinhala inscriptional Prakrit seems to be the proximate source.

c22.s3.t5.15

*kaṇa-* 'scribe'.¹ In the Early Brāhmī cave inscriptions of Sri Lanka, *kaṇa* occurs with the meaning 'scribe'. It is derived from Skt. *karaṇa*.² The form *kaṇa* does not occur in Indian Prakrits in this sense. In later Tamil inscriptions, the corresponding expressions *karaṇa* (> *karṇam*) are directly borrowed from Sanskrit.

c22.s3.t5.16

¹ See Commentary (40.1.a) for discussion.

c22.s3.t5.17

² S. Paranavitana 1970: p. xcv.

c22.s3.t5.18

*kāviti* 'title of a senior merchant and court official'. This expression is ultimately derived from Skt. *gṛhapati* most probably through Sinh. Pkt.* *kāpiti* (> *kapiti*).³

c22.s3.t5.19

³ See Commentary (3.5) for discussion.

3.3.6 Lexical items of doubtful etymology

c22.s3.t6.1

There are a few lexical items in the Corpus, whose derivation from Dravidian or Indo-Aryan is in doubt.⁴ Some representative types are noticed below.

c22.s3.t6.2

⁴ These are listed separately in the Etymological Index: Doubtful Items (Appendix VI) indicating alternative derivations from both sources. Each lexical item is also discussed in detail in the relevant entry in the Commentary.

c22.s3.t6.3

(i) There are a few lexical items where either derivation is equally plausible:

c22.s3.t6.4

(e.g.) *aiyaṉ* 'lord, master'; cf. Dr. *ai* (D. 196a); IA *ayya* < *ārya*. *cantaṉ* 'a personal name'; cf. Dr. *cantam* (D. 2328); IA *caṁda* < *candra*.

c22.s3.t6.5

c22.s3.t6.6

Table 3.3. Ardhamāgadhī loanwords in the Corpus.

c22.s3.t6.7

c22.s3.t6.8

Table 3.4. Sanskrit loanwords in Early Vaṭṭeḻuttu inscriptions in the Corpus.

c22.s3.t6.9

(ii) There are some items which are generally regarded as derived from IA, but for which plausible Dr. etymologies can be suggested:

c22.s3.t6.10

(e.g.) *akaram* 'Brāhman village'; cf. Dr. *nakar* (D. 3568); IA *aggahāra* < *agrahāra*.
*ta(c)caṉ* 'carpenter, mason'; cf. Dr. *tai* / *tacc-* (D. 3473); IA *taccha* < *takshan*.

c22.s3.t6.11

(iii) Borrowing and re-borrowing:

c22.s3.t6.12

(e.g.) Dr. **catiya* > IA *satiya* > Ta. *atiyaṉ* 'a clan name'.

c22.s3.t6.13

(iv) Borrowed into both:

c22.s3.t6.14

(e.g.) Dr. *nāka* (IA *nāga*) appears to be the name of an ancient autochthonous race of South Asia borrowed into both Dr. and IA.

3.3.7 Changeover from Prakrit to Sanskrit

c22.s3.t7.1

Sanskrit supplanted Prakrit in South Indian inscriptions from about the beginning of the 5th century A.D.¹ This changeover is also reflected in the language of the cave inscriptions in the Tamil country. Thus the loanwords in the Tamil-Brāhmī inscriptions (ca. 2nd century B.C.-4th century A.D.) are invariably from Prakrit, while those in the Early Vaṭṭeḻuttu inscriptions (ca. 5-6 century A.D.) are from Sanskrit or survival of borrowings from Prakrit in the earlier age. The transitional character of the Early Vaṭṭeḻuttu inscriptions is best illustrated by the personal names of two monks, *vaccaṇanti* (< Pkt.) and *cantiraṉanti* (< Skt.) appearing in the Jaina inscriptions of about the same period (ca. 6th century A.D.).²

c22.s3.t7.2

¹ The earliest stone inscription in Sanskrit from ancient Tamil country(Tamil Nadu and Kerala) is that of Vishṇuvarman, probably a Kadamba prince, at Edakal in Kerala, assigned on palaeographic evidence to ca. 5th century A.D. E. Hultzsch in F. Fawcett 1901 (Estampage No. 1). See I. Mahadevan 1999 for revised interpretation.

c22.s3.t7.3

² Nos. *115* and *116* in the Corpus.

c22.s3.t7.4

To illustrate, loanwords occurring in the Early Vaṭṭeḻuttu inscriptions in the Corpus (Table 3.4) appear to be borrowed directly from Sanskrit. The same situation obtains in later Tamil inscriptions and in Literary Tamil, with the earlier layer of borrowings from Prakrit and the later layers increasingly from Sanskrit.

Linked texts (2)

தேவாரம் — Tiruñāṉacampantar & Tirunāvukkaracar, Tamil
Tolkāppiyam — Tolkāppiyar, Tamil