1In The Language of the Gods in the World of Men, Sheldon Pollock writes:

“It is an interesting fact that in six centuries of Pallava rule not a single inscription was produced in which Tamil does any work beyond recording the everyday — remitting taxes, specifying the boundaries of a land grant, and the like. While examples exist in earlier Pallava records of Sanskrit being used to document the everyday world — a function that would become increasingly rare wherever it could be relegated to the vernacular — none exists where the everyday language is allowed to do the work of Sanskrit in a praśasti: the literary work of interpreting and supplementing reality and revealing it in its truth” (2006: 122).

2This statement is repeated later, but with a significant nuance: “For the six hundred years of their existence, with [their] very few exceptions, the Pallavas never spoke literarily in Tamil in their public records” (op. cit.: 291).

3On both occasions, Pollock indeed cites some exceptions. “One marginal exception to the documentary restriction of Tamil in Pallava grants is the account of the election of a new king, Nandivarman II, after the death of Parameśvaravarman II around 730” (op. cit.: 122, n. 15). This inscription is found in the Vaikuṇṭhaperumāḷ temple at Kāñcīpuram (infra, p. 382 ff.). “Aestheticized inscriptions are sometimes found within Pallava domains, as in the two singular verses engraved on a ca. eighth-century temple in Tañcāvūr district (EI 13.134 ff., especially pp. 143, 148). Being completely untouched by Sanskrit in idiom and discursive style, these present a picture of literary-political inscriptions very different from what was to come. They may have been unique experiments, with no prehistory and no future” (op. cit.: 291, n. 12). These verses are in fact engraved on pillars in the Mīṉāṭcīcuntaresvarar temple at Centalai; pace Pollock, there actually are more than two (infra, p. 376 ff.).

4It is no denial of the importance of Pollock’s study to state that in general he understates the Tamil side of the picture. Motivated by this observation, the present paper is an attempt to refine our understanding of the use of Tamil during the Pallava period (ca. 4^th to the 9^th centuries CE). I propose to scrutinise the epigraphic corpus of the Pallavas lato sensu (i.e. Pallava dynastic inscriptions, inscriptions of their feudatories, local inscriptions by local individuals) in their Tamil realm (i.e. the northern districts of present-day Tamil Nadu, which is the region later known as Toṇṭaimaṇṭalam, and the northern portion of the Cōḻamaṇṭalam). I will focus on instances of literary, imaginative or expressive Tamil, and specifically on instances of what could be called political Tamil, that is Tamil used in panegyric discourse. To complement the epigraphic data, I will also pay attention to belletristic Tamil under the Pallavas. After reviewing the relevant data, I will discuss their impact on Pollock’s general theory about vernacularisation in the Tamil cultural sphere. In doing so, I will deal also with the political discourse of the Pāṇḍyas and Cōḻas.

5The evidence of literary, expressive or political Tamil in the Pallava realm is substantially more diverse than the two examples mentioned by Pollock. Orr has already warned us that “we must take exception to Pollock’s statement (...). While inscriptions in which Tamil epithets or verses of praise appear are not common in Pallava territory, neither are they completely nonexistent” (2009: 100–1). Orr found Tamil “expressive” language in some Pallava inscriptions and pointed out that, “Given that the total number of ‘Pallava inscriptions’ is not huge” at least at an early period, “these few inscriptions count for something” (op. cit.: 101, n. 9).

6Orr refers to stone inscriptions: on the northern pillar of the façade of one of the Vallam rock-cut caves, a certain Kantacēṉaṉ is presented as the “servant of Catturumallaṉ (i.e. Sanskrit Śatrumalla, “He who is the wrestler to his enemies”) Kuṇaparaṉ (i.e. Sanskrit Guṇabhara, “He who bears qualities”) Mayēntirappōttarecaru (i.e. the bull-king or the sproutking [i.e. Pallava king] Mahendra)” (catturumallaṉ kuṇaparaṉ may[ē*] ntirapp[ō*]ttar [ē*]caru2 aṭiyāṉ; IP 36. B = SII 2.72. B; first quarter of the 7^th century).3 On the southern pillar (IP 36.A = SII 2.72.A), we find a pair of birudas, Pakāppiṭuku (“He who is an invincible [literally: which cannot be split] thunderbolt” or “He who is a forked lightning”,4 see infra, p. 369, for discussion of the meaning of piṭuku) and Laḷitāṅkuraṉ (i.e. Sanskrit Lalitāṅkura, “He who is a gracious bud”), known to belong to Mahendravarman (ca. 600–630).5 At Tiruveḷḷaṟai, a local chief named Cellikkōmāṉ Mallavāṉ is praised in fragmentary Tamil verses preceded by a Sanskrit titulature and date of Dantinantivarmaṉ, i.e. Nandivarman III, the son of Dantivarman (IP 119 = SII 12.48; ca. 850).6

7Besides these brief statements of praise, we have instances of literary Tamil compositions in Pallava inscriptions. At Tiruttaṇi, a certain Perumāṉaṭikaḷ composed a foundation verse, in veṇpā metre as the colophon of the inscription itself indicates (IP 245 = SII 12.94; end of the 9^th century).7 At Tiruveḷḷaṟai again, an inscriptional verse inviting donations is engraved on a svastika-shaped well (IP 98. B = EI 11.15. B = SII 12.40; first half of the 9^th century).8

8These few inscriptions that “count for something” are actually not isolated examples. Other instances of unequivocal political Tamil in Pallava inscriptions are known (see Fig. 1 for the map with all the examples known to me). I will distinguish two stages in the form of this Pallava political Tamil. The first stage starts around 600 with Mahendravarman I (hereafter Mahendra I). I shall be concerned here with inscriptions from four sites (Tiruccirāppaḷḷi [hereafter Trichy], Taḷavāṉūr, Centalai and Kāñcīpuram) and with the use of Tamil in royal titulature. The second stage, partly overlapping with the first stage since Tamil birudas and titles continued to be used, concerns mainly one king, Nandivarman III (hereafter Nandin III) who reigned for approximately 20 years around 850. For this king I will focus on a court poem, the Nantikkalampakam, and on the use of a specific epithet prefixed to his name in local Tamil inscriptions.

1. Pallava political Tamil (ca. 600–900)

9We encounter the first instances of Tamil used in Pallava political discourse under the reign of Mahendra I (ca. 600–630), in two of the caves he had excavated, at Trichy and at Taḷavāṉūr.

1.1. Birudas in the upper cave at Trichy (ca. 600–630)

10In the so-called upper cave in Trichy (Trichy tāluk and district), besides Sanskrit dedicatory or foundation inscriptions (IP 32, 33 and 35), there are at least sixty birudas engraved at different places on the pillars and pilasters (IP 34 = SII 12.8, with facsimiles, plate I). The classification of birudas as pertaining to political discourse may not be immediately evident, since they are names/epithets mostly preserved without discursive context. However they do tell us something about the self-presentation of their bearers and, in this respect, participate in the political imagery of a dynasty.

11The Trichy birudas are in different languages: some are clearly Sanskrit, others Telugu and Tamil,9 while the language of other birudas has not been clearly identified. Lockwood et al. (2001: 245 ff.) have proposed etymologies from Prākrit, Telugu, Kaṇṇaḍa or Tamil. Scholars influenced by a chauvinistic view of the role of India in South-East Asia have even fancied that some of them were in Khmer.10

12But even though these birudas are not all clearly understood, one point is clear: namely that we see a mixing of birudas in Sanskrit and in Dravidian languages, Tamil and Telugu in particular. This mixture involved the use of different scripts: Grantha letters for Sanskrit and Telugu birudas, and sometimes even for Tamil birudas (see vañjavalava below); for Tamil birudas Tamil letters are used, but some are still undifferentiated from Grantha. As an illustration, let us look at some birudas on the exterior (south) row of pilasters and pillars (see Fig. 2). At the top, on the corbels (facing south), we can read from left to right:

vañjavalava (western pilaster), “He who is a conqueror of the universe?” or “He who is a conqueror of the deceit (or: lie)?” or “He who is a conqueror in his lineage?”11

sarvvana[y]{aḥ}12 (first pillar from the west), “He who is fully wise/prudent”.

saṃkīrṇṇajātiḥ // (second pillar from the west), “He who is of mixed birth” or “He who [has invented a system] of mixed jāti (i.e. rāga)?”13

satyasandhaḥ (third pillar from the west), “He who is in agreement with the truth” or “He whose intention is truth”.

[laḷ]itāṃkuraḥ (fourth pillar from the west), “He who is a gracious bud”.

lakṣitaḥ (eastern pilaster), “He who is distinguished”.

13These are all written in Grantha characters. All are in Sanskrit language except the first one, which is in Tamil. At the bottom, on the southern face of the square base of the four pillars, we have from leftto right the following birudas:

[ḍa][...] ku14 (first pillar from west), “?”

cit[t]i{rakā}[ra]ppu{li} (second pillar from the west), “Lion among wonder-doers” or preferably “Lion among painters”.15

piṇapiṇakku (third pillar from the west), “He who is in disagreement with the devils?”16

kucañāṇa17 (fourth pillar from the west), “He whose bowstring is darbha grass?”18

14These are apparently all Tamil birudas written mainly with a Tamil alphabet, which appears here maybe for the first time,19 some characters still being undifferentiated from the Grantha script (see Fig. 3). For instance the retroflex “ṇ” is very similar in Tamil and Sanskrit birudas. Compare also the “k” in a Sanskrit biruda from the top list and in a Tamil biruda from the bottom list: both types have two vertical strokes. As for the fragmentary first biruda, its first letter may be Grantha and the first part of its compound a Sanskrit word. But the ending (-ku) seems to indicate that the second part of the compound is Tamil (or some other Dravidian language). This could be a biruda made from one Sanskrit word and one Tamil (or other Dravidian) word, where script differentiates the etymology, in contradistinction with cittirakāra° and kuca° which are Tamilised forms of Sanskrit words. These last two words attest for their part that the distinct Tamil letter “ca” had already been invented.20

15As I have said above, there are more than sixty birudas engraved in the upper cave at Trichy. The same kind of nomenclature is available for Mahendra I at two other sites: on the beams of the Pallāvaram caves (some 100 birudas; IP 28 = SII 12.13) and on a pillar, originally found in the Ekāmbaranātha temple at Kāñcīpuram (14 birudas; IP 21 = SII 12.14). Mahendra I thus possesses more than 130 different birudas, in at least three different languages.21 We have seen already that four of his birudas are also engraved at Vallam (IP 36 = SII 2.72; supra, p. 361). In that case only one is not of Sanskrit etymology (Pakāppiṭuku), while the three others are Tamilised forms of Sanskrit birudas usually found in Sanskrit and Grantha (Laḷitāṅkuraṉ, Catturumallaṉ, Kuṇaparaṉ).

16This use of different languages is previously attested on coins of Śātavāhana kings. For instance Mahadevan (2003: 201, 2006: 87) reads the bilingual legend of a coin of Sātakaṇi (2^nd century CE) as follows:

Obverse: rāño vāsiṭhiputasa sirisātakaṇisa, “Of the king Vāsiṭhiputa (i.e. Sanskrit Vāsiṣṭhīputra, “son of a Vāsiṣṭhī”) Sirisātakaṇi (i.e. Sanskrit Śrīśātakarṇin)”. [my translation]

Reverse: aracaṉku vāciṭṭimakaṉku tirucātakaṇiku, “Of the king, son of a Vāciṭṭi (i.e. Sanskrit Vāsiṣṭhī), Tirucātakaṇi (i.e. Sanskrit Śrīśātakarṇin)”. [my translation]

17On the obverse, around the face of the king, is a Prākrit legend in genitive case. On the reverse, around some symbols, a legend of same meaning, in dative case, is provided in a language diversely identified by scholars as Tamil, Telugu or proto-Dravidian.22 For our purpose, it is important to note that in any case a Dravidian language contrasts here with the Indo-Aryan of the obverse and that this is an early case of a vernacular used in public discourse.

18To conclude about Mahendra I’s birudas, it is noteworthy that they were also used as legends on his coins. About twenty of these, in Sanskrit and Dravidian languages, that are known from the stone inscriptions of Mahendra I as well as others not attested elsewhere (possibly because the Trichy list of birudas is lacunary), appear also individually on coins.23 A handful of coins of Mahendra I’s direct successors bear Sanskrit birudas such as Śrībhara and Śrīnidhi.24 There is however no instance of bilingual legends as for the coins of Śātavāhana kings. In any case, these Pallava coins point to the importance of birudas for the self-presentation of the ruler and to the use of Dravidian ones for this purpose.

1.2. Tamil birudas and titles of the Pallavas (7^th to 9^th century)

19Tamil birudas or titles were subsequently used for successors of Mahendra I, even though Sanskrit birudas were far more numerous (due to the existence of lengthy praśastis). In the 8^th and 9^th century Tamil birudas or titles such as Viṭēlviṭuku, Kāṭuveṭṭi or Kāṭavar are commonly found. Pōttaraiyar, a dynastic name, was used since the 7^th century.

20The meaning of Viṭēlviṭuku is a matter of debate. It is found in royal and local inscriptions as well as in the Nantikkalampakam, a court poem dated to ca. 850 (infra, p. 387 ff.). This biruda or title is sometimes associated with Pallava kings, but more often with their vassals and officers.25 For some scholars, viṭuku is a Telugu or Kaṇṇaḍa word (piṭuku, “thunderbolt”) and the whole compound would mean “the crashing thunderbolt”.26 Minakshi (1938: 45) argues that viṭēl is the contraction of viṭai (“bull”) and vel(lu)-tal (“to conquer”) and considers viṭuku as an abstract noun derived from viṭu-tal (“to issue, emit”). The term would then mean “issue (i.e. royal order) [marked with] the victorious bull” and, as a biruda, “He whose issues (i.e. orders) [are marked] with the victorious bull”. Some of Minakshi‘s arguments are sound, for instance her interpretation of IP 93, section J, and her reference to Nantikkalampakam 15 concerning the mention of royal orders (supra, p. 368 n. 25). But we can dismiss her reference to Viṭōlaiviṭuku° in IP 155.47, since this is a misreading for Viṭēlviṭuku° (see correction by Hultzsch, EI 18, p. 11). To add credit to Minakshi’s view, I remind readers that the expression tirumukam viṭu-tal (“to issue a royal order”, literally “to issue a gracious face”) is common in the Tamil portion of Pallava copper-plates, that the seals of Pallava grants are usually adorned with a bull, and that the “victorious bull-banner” (viṭai vel koṭi) of the Pallava king is mentioned in Periya Tirumoḻi 186 = 2.9.6c (p. 317). However viṭuku occurs in other birudas where its meaning seems to be “thunderbolt”.27 Neither of these two interpretations can thus account for all attestations. Other hypotheses have been advanced but are less convincing.28

21As for Kāṭavar (“forester” from kāṭu + avar) and Kāṭuveṭṭi (“forest clearer” from kāṭu and veṭṭu-tal, “to cut”),29 they appear in Tamil epigraphy in the 8^th century.30 In Pallava royal inscriptions Kāṭavar rarely designates Pallava kings, whereas Kāṭuveṭṭi is not infrequent in designations of Pallava vassals, petitioners or executors of grants.31 Both terms are used in local Pallava inscriptions and in inscriptions of other dynasties (maybe as early as the 6^th century).32 Kāṭavaṉ apparently designates the younger brother of Nandin III in Nantikkalampakam 33. The Pāli form Kaṇḍuveṭhin occurs in Mahāvaṃsa 47.7 and 77.79 (that is Cūḷavaṃsa) where, according to Hultzsch (1913: 527), it designates Narasiṃhavarman I. Kāṭavaṉ is found twice in the Tēvāram (7.39.7d and 7.39.9a) and five times in the Periya Purāṇam (1411b, 4045d, 4052d, 4053b and 4179a).

22For Krishna Sastri (1926: 9–10), Kāṭavar designates members of a collateral line of the great Pallavas of Kāñcīpuram, descending from Hiraṇyavarman, a branch that rose to power with his son Nandin II.33 Ramamurthy (1983) elaborated on this view and suggested that there were actually two collateral lines of Kāṭavars, ruling the west of Toṇṭaimaṇṭalam, that is, the Palkuṉṟakkōṭṭam (“district of many hills”) of epigraphy, a mountainous and wooded region.34 This feature of their realm would explain their name, which finally became the name of their kula. For Ramamurthy both lines descend from Bhīmavarman, the younger brother of Siṃhaviṣṇu of the main line, who was entrusted with the Maḻava territory conquered by his elder brother.35 When the main line of Siṃhaviṣṇu came to an end at the death of Parameśvaravarman II, his cousin Nandin II of one of these Kāṭavar lines became king in Kāñcīpuram.36 We can doubt the existence of a second line of Kāṭavars, but what seems certain is that before Nandin II (ca. 730), Kāṭavar is not used for Pallavas reigning in Kāñcīpuram.

23Even though their etymology is clear, one can wonder about the exact purport of the birudas Kāṭavar and Kāṭuveṭṭi. Do they refer to the original home of the Pallavas, sometimes supposed to be related to the tribal Kuṟumpars? Or, as argued by Minakshi (1938: 12 ff.), do they tell us more about the civilising work of the Pallavas of claiming forest land for cultivation? It is also possible, as suggested to me by Charlotte Schmid, that we have in the second of these a reference to the tribal custom or topos attested in the Caṅkam literature of kings cutting the tutelary tree of their vanquished enemies.37

24As for Pōttaraiyar (and its variants Pōttarecaru, Pōttaracar, etc.), it is associated with the Pallavas from the 7^th to the 9^th centuries, in the Pallava epigraphic corpus, but also in inscriptions from other dynasties.38 Pōttu means “male of animals”, especially of cattle, tiger and deer, but also “tender branch of shoot of a tree”, according to the Tamil Lexicon. Pōttaraiyar may thus be understood as the Tamil equivalent of Sanskrit pallavarāja-, for Sanskrit pallava means “sprout, shoot, twig, bud”. This meaning has been played on in Pallava dynastic myths (see for instance IP 77.31, verse 17, where the eponym is born on a litter of sprouts: palla[va*]saṃstareṣu). Pōttaraiyar may also mean “Bull-king” in reference to an animal associated with kingship since Aśoka’s time and figuring on the majority of the seals and coins of the Pallavas. The bull may be considered as a symbol of male or reproductive force, but also, subsequently when around 700 the dynasty turned almost exclusively Śaiva, as a mark of their sectarian leaning.

25To be exhaustive, it is to be added that Nandin II bears the title Perumāṉaṭikaḷ (“venerable great lord”), occurring in IP 93, section K, unfortunately fragmentary.39

26A first observation concerning these Tamil birudas or titles is that they rarely occur in royal inscriptions, i.e. inscriptions directly commissioned by Pallava kings. The narrative inscriptions in the Vaikuṇṭhaperumāḷ temple in Kāñcīpuram (IP 93) and the Tamil portions of copper-plates are, from this point of view, exceptions. This rarity is not surprising since most of the royal Pallava inscriptions are in Sanskrit. Noteworthy also is the fact that these Tamil birudas or titles are also borne by Pallava officers and vassals. It is sometimes unclear if a given title in a given inscription refers to a Pallava king or to a vassal or officer, bearing it on his own or entitled to bear it by his overlord. In literature, on the other hand, Viṭēlviṭuku is clearly a title of Nandin III in the Nantikkalampakam (15, 17 and 78), but this text is relatively late (ca. 850). As for Kāṭavar, it may originally be attached to a collateral and regional line of Pallavas. These Tamil birudas or titles seem thus to be associated to a local Tamil context rather than to the dynastic discourse.

1.3. Anuṣṭubh and veṇpā in Taḷavāṉūr cave (ca. 600–630)

27In another cave commissioned by Mahendra I, in the village of Taḷavāṉūr (Gingee tāluk, undivided South Arcot district, today Viḻuppuram district), we find inscriptional verses in Sanskrit and Tamil praising the royal patron (Hultzsch 1914 = EI 12.27). On the exterior, at the top of the left (western) pilaster near the arm of the western dvārapāla, is a Sanskrit anuṣṭubh (IP 19 = EI 12.27. A, with facsimile = SII 12.10; see Fig. 4), facing south, at a visible place, like most dedicatory Sanskrit stanzas engraved on Mahendra I’s caves (IP 24, 27 and 29). It reads:

daṇḍānatanarendreṇa narendreṇaiṣa kāritaḥ
śatrumallena śaile [’*]smin śatrumalleśvarālayaḥ
“By order of Śatrumalla, the king by whose army (or: authority) the kings were bent, this has been made on this mountain: the Śatrumalleśvarālaya”.

28Inside the cave, on the eastern face of the square base of the left (southern) pillar of the porch in front of the cella, a Tamil veṇpā is found engraved. Its colophon is found on the perpendicular face of the same square portion of the pillar, facing south (IP 20 = EI 12.27. B, with facsimile = SII 12.11 = Peruntokai 972, where the text is thoroughly emended).40 They read (Grantha being marked in bold):

śrī
toṇṭaiyantār v[ē*]ntaṉ nare[ē*]ntirapp [ō*]ttaraiyaṉ41 veṇp[ē*]ṭṭiṉ42 ṟeṉpāl43 mika makiḻntu — kaṇṭāṉ cara mikka veñcilaiyāṉ catturumalleśvarālaiyam44 eṉṟ’araṉukk’iṭam ākav āṅku45
ivvūr [b]rammamaṅkalavaṉ cellaṉ civadāsan colliyatu

“Prosperity!
Narentirappōttaraiyaṉ (i.e. “the bull-king/sprout-king among the kings”), the king with the beautiful garland of toṇṭai, he whose cruel bow is overflowing with arrows, has ordered with great joy that, on the southern side of Veṇpēṭu (i.e. “the white hamlet” or “the white hill”46), there be for Araṉ (i.e. Hara = Śiva) a place called the temple of Catturumalleśvara.
Uttered by Brammamaṅkalavaṉ Cellaṉ Civadāsaṉ of this village”.

29Several points are of interest here. The Sanskrit anuṣṭubh is neatly engraved pāda by pāda and compares well with other cave inscriptions of Mahendra I: we have similarly one verse in Sanskrit at Maṇṭakappaṭṭu (IP 27, gīti), at Makēntiravāṭi (IP 24, nardaṭaka), at Cīyamaṅkalam (IP 29, āryā), and at Trichy (IP 35, āryāgīti). In contradistinction, the Tamil veṇpā is not arranged aṭi by aṭi and is badly engraved with the lines bending to the left (see Fig. 5). From this, we may surmise that it is not an inscription commissioned by the king. This seems to be confirmed by the colophon specifying that a local poet uttered (i.e. composed) this verse. It has to be emphasised that poets of Pallava stone inscriptions in Sanskrit (in contradistinction to the poets of Sanskrit portions of copper-plates) are rarely named.47 Noticeable also is the beginning of the veṇpā, which phonetically echoes the beginning of the anuṣṭubh: toṇṭai echoes daṇḍā°; narendra is repeated in narentira. Finally the veṇpā has a specific Tamil flavour with the mention of the toṇṭai garland: in Caṅkam literature, each of the three major dynasties is associated with a specific flower.48 The toṇṭai is a creeper whose flower is distinctively associated with the Pallavas.

30All scholars referring to this veṇpā considered that it was composed to emulate the Sanskrit anuṣṭubh of the first quarter of the 7^th century. There is less agreement concerning its date.49 Venkatasubba Ayyar (SII 12, p. 7) assigns it to the 9^th century on palaeographical grounds. Srinivasan (1964: 77–78) finds it “reasonable to consider the two [inscriptions] as coeval and complementary to each other”. This author noted that palaeography has its limitations and that we have few Tamil records of this period for comparison.50 He argued also that he does not know of any late Tamil translation of a Sanskrit inscription, but that there is an example of coeval Sanskrit and Tamil inscriptions in a Pāṇḍya cave at Tirupparaṅkkuṉṟam (see IEP 10–11).

31This co-presence of Sanskrit and Tamil verses bearing similar content in different languages and cultural contexts is remarkable. The initiative for composing and inscribing the Tamil veṇpā seems wholly local, like in the case of the Tamil birudas or titles. More precisely it seems that the royally commissioned Sanskrit praise urged some local people to praise the royal patron in Tamil language and literary context (alluded to by mention of the toṇṭai).

1.4. Tamil inscriptions at Centalai (8^th century)

32Royal inscriptions with panegyric Tamil can be found on pillars that are now reused in the Mīṉāṭcīcuntaresvarar temple at Centalai (Tiruvaiyāṟu tāluk, Tañcāvūr district; Subrahmanya Aiyer 1916 = EI 13.10; poems are named by Roman numerals for the pillar and Roman letters for the face of the pillar). As mentioned earlier, Pollock mentions these inscriptions, but does not describe them in detail (2006: 291, n. 12). Actually there are many more than the “two singular verses” he refers to. Four pillars, originally set up in the nearby village of Niyamam and brought to Centalai at an unknown date, are engraved on the four faces of their upper and bottom square portions, and, in one case at least (III.c), also on the middle octagonal portions. It seems that originally there was a total of 32 inscriptions, but due to later constructions, only 29 could be read and edited: one prose inscription introducing the lineage of the Muttaraiyar kings (I. a), four lists of birudas (I.b, II.d, III.a, IV.a), and 24 stanzas.51 Although not dated internally, these inscriptions have been assigned approximately to the first half of the 8^th century from a palaeographical point of view (Subrahmanya Aiyer 1916: 135–36). The Centalai inscriptions are not of a Pallava king but of a member of the Muttaraiyar family, which was allied to the Pallavas at least at this period (see Centalai verses I.g and II.h). The Centalai verses, some of them with a colophon — we thus know the names of four poets — extol the greatness of the patron king Kuvāvaṉ Māṟaṉ alias Perumpiṭuku Muttaraiyaṉ, like verse I. d (a veṇpā) which reads (Subrahmanya Aiyer 1916: 140):

[.....................................p]pa [ō*]ṭik
kaḻuku koḻuṅ kuṭar kavva viḻi kaṭ p[ē*]y52
puṇṇ aḷaintu kaiy ūmpap53 p[ō*]r maṇalūr veṉṟat[ē*]
ma[ṇ]ṇ aḷainta cīr māṟaṉ vāḷ54
pāccil v[ē*]ḷ nampaṉ pāṭiṉa

“The sword of the glorious Māraṉ who has embraced the earth, has been victorious in Maṇalūr battle where (...),55 where vultures, rushing, grasped fat bowels, and where pēys (i.e. demons) with wide-open eyes56 sucked their hands after thrusting them into wounds (or: flesh).
[Poems] sung57 by Vēḷ Nampaṉ of Pāccil”.

33This veṇpā is reminiscent, as pointed out to me by Charlotte Schmid, of the battlefield atmosphere in the Puṟanāṉūṟu, as in this extract from poem 62 (lines 2–6), where the battle scene of a fight between a Cēra and a Cōḻa is described in the following words:

porutu āṇṭu oḻinta maintar puṇ toṭṭu
kuruti ce kai kūntal tīṭṭi
niṟam kiḷar uruviṉ pēey peṇṭir
eṭuttu eṟi aṉantal paṟai cīr tūṅka
paruntu arunt’uṟṟa (...)

“In that battlefield while, to the sound of mourning drums beaten high (?), pēy women with bright-coloured bodies (or: with bodies of shining lustre), having smeared their tresses with their hands red of blood after digging out the wounds (or: flesh) of dead warriors, were dancing and kites were eating (...)”

34The common theme of both poems is obvious: pēys (clearly women in Puṟanāṉūṟu 62) are red-handed (kai) with blood from the flesh or wounds (puṇ) of dead warriors. However the metre is different (veṇpā versus āciriyippā). This shared theme may not prove, as observed to me by Herman Tieken, that Vēḷ Nampaṉ and the other poets of the Centalai verses (see also Centalai verses II. c and III. e for birds feeding on battlefields) knew the Puṟanāṉūṟu. And as observed by Eva Wilden, the fact that the pēys are female in Puṟanāṉūṟu is a later stage in the development of the representation of evil in Tamil (beings like aṇaṅku, cūr, pēy originally not having a specific sex) and this feature — the female sex of pēys and even the pēy topos itself — may be in fact imported in Tamil literature from popular folklore or Sanskrit literature. It could then be that Puṟanāṉūṟu 62 and Centalai’s verse I.d draw on common sources without being directly related.

35However there is also a lexical commonality between the Centalai verses and the Eṭṭuttokai.58 I have not been able to check if this lexical commonality is also formulaic. If so, this would be a firmer argument in favour of the Centalai poet’s knowledge of Caṅkam poetry. But there is another fact which certainly does point in that direction: Centalai verse IV.e, even though not clearly understandable, indicates that at least one poet was familiar with Caṅkam poetry or poetics. It first mentions āyam (group of the female attendants of the heroine in Akam according to the Tamil Lexicon), pūvai (a term that may be related to pūvainilai, a tuṟai in the pāṭāṇtiṇai of Puṟam, mentioned in Tolkāppiyam, Poruḷatikāram 60, as Jean-Luc Chevillard informed me) and kaikkiḷai (one-sided love, an originally Akam, then Akappuṟam, tiṇai) in one coordinated group.59 After that, it apparently describes the sorrow of a heroine as in Akam poetry.60

36Also noteworthy is the lexical commonality between Centalai verse I. d and contemporaneous Bhakti poetry when describing the pēys dwelling in the cremation ground, which is the place of Śiva. In Kāraikkāl Ammaiyār’s Tiruvālaṅkāṭṭut Tiruppatikam, the first stanza reads:

eṭṭi ilavam īkai cūrai kārai paṭarnt’ eṅkum
cuṭṭa cuṭalai cūḻnta kaḷḷi cōrnta kuṭar kauvap
paṭṭa piṇaṅkaḷ paranta kāṭṭiṟ paṟai pōl viḻi kaṭ pēy
koṭṭa muḻavaṅ kūḷi pāṭak kuḻakaṉ āṭumē

“Eṭṭi (strychnine-tree), ilavam (clearing-nut tree), īkai (mimosa-tree), cūrai (jujube-tree), and kārai (canthium-tree) spread everywhere in the forest where corpses have been grasped and their bowels fell on the spurges that surround (or: fence?) the hot burning-ground [there] while [he?-] demons (kūḷi = Sanskrit bhūta) sing and [she?-] demons (pēy) with drum-like wide-opened eyes beat drums, Kuḻakaṉ (i.e. Śiva as young and beautiful) dances”.

37Centalai verse I.d and Kāraikkāl Ammaiyār’s stanza share vocabulary (and in one instance this is even a formula): viḻi kaṇ pēy (occurring also and more complete as paṟai pōl viḻi kaṇ pēy in Tēvāram 1.67.3a, that Subrahmanya Ayyar rendered as “the pēy which has open eyes as big as a drum” even though it is probably a plural) and kuṭar kauva/kuṭar kavva. It is then possible that Vēḷ Nampaṉ knew Kāraikkāl Ammaiyār’s poem, if the latter is really earlier than the 8^th century.61

38So whatever the literary background of Centalai verse I.d, it shows a typical Tamil evocation of the battlefield that is met with in Caṅkam poetry and adopted in Tamil Bhakti poetry for the evocation of the cremation ground. Centalai verse I.d (and also II.c and III.e) clearly resonate within a Tamil literary culture. The same may be true for Centalai verse II.f, which mentions warriors who entered stones (kaṟ puka), a reference to hero-stones, set up as memorials to them, which is a pan-Indian practice also known in Caṅkam poetry.62

39As for the birudas at Centalai, we have two different lists of four birudas. Each list is engraved on two pillars. Two birudas appear in both lists. So there are six different birudas in total. On the pillars I.b and II.d, we read (cf. Subrahmanya Aiyer 1916: 139 and 143; Grantha is marked in bold63):

[śrīmāṟaṉ], “He who is warlike”.

śrīśatrukesarī,64 “He who is a lion for his enemies”.

śrīkaḷvar(a)kaḷvan,65 “He who despoils the despoilers?”66

śrī’atisāhasan, “He who is extremely intrepid”.

40On the pillars III.a and IV.a (for III.a see the estampage reproduced in Mahadevan 2003: 208; for IV.a see Fig. 6), we read (cf. Subrahmanya Aiyer 1916: 145 and 148; Grantha is marked in bold):

śrītamarālayaṉ,67 “He who is an asylum for his relations (or: friends)”.

śrī’abhimānadhīran, “He who is firm in pride”.

śrīkaḷvarakaḷvaṉ

śrīśatrukesarī

41What is at first remarkable here is the alternation of Sanskrit and Tamil birudas in both lists. In the second list it is marked by the use of different alphabets: Grantha and Tamil scripts are well differentiated here, unlike in Trichy; the letters “k” and “r” have two vertical strokes in Sanskrit and only one in Tamil. But in the first list, śrīkaḷvarakaḷvan, which is a Tamil word, is written entirely in Grantha, up to the point of having a final “n” (and not “ṉ”). It contrasts with śrīkaḷvarakaḷvaṉ of the second set which is entirely in Tamil letters except for the prefix śrī-.68 Peculiar also for the Sanskrit birudas and for the Tamil śrīkaḷvarakaḷvan written in Grantha is the final “n”. This stands for the Tamil alveolar “ṉ” and means that the Sanskrit birudas bear the Tamil pronominal suffix of the third person singular (“aṉ”), which is not uncommon for Sanskrit words in Tamil inscriptions, but is usually represented with a Tamil letter.

1.5. The narrative inscriptions in the Vaikuṇṭhaperumāḷ at Kāñcīpuram (ca. 800)

42The last instance of political Tamil — or at least a mixture of Sanskrit and Tamil — before the 9^th century is the second exception mentioned by Pollock, that is the inscriptions provided for some of the narrative sculptures of the southern corridor of the Vaikuṇṭhaperumāḷ temple at Kāñcīpuram (Kāñcīpuram tāluk and district; Krishna Sastri 1923 = IP 93 = SII 4.135, with facsimile, plate II; Fig. 7). These prose inscriptions are exceptional in the sense that they narrate the dynastic events represented on the panels they are attached to.69 They explain the circumstances that lead to the coronation of Nandin II (ca. 725), but were probably engraved toward the end of his reign (ca. 800). Let’s examine the first narrative inscription (IP 93, Section A; Grantha is marked in bold):70

[s]vasti śrī ivvakay[ē*] brah[m]anādi71 ’āka vārāniṉṟa72 pallavavaṃśattuḷ para[m]eśvarapp[ō*]ttar[ā*]yar svarggasthar ’āvatu rājyam utsannam [ā*]ka mattrakaḷu ghaṭakayāru

mūlaprakṛt[i]yu73 k[ā*]ṭaveśakulahiraṇyava[r*]mamah[ā*]r[ā*]jar aṭekka74 rājabharam75 [gra]hakka sama[r*]tthar ubhaya[ku] laparaśuddhar [g]ovalapp[ō*]tt’76 [ā*]kav eṉṟu viṭukapaṭ[u?] [...] p[ō*]nta iṭam /

“Hail! Prosperity! Here is the place where, while the kingdom was in chaos because Parameśvarappōttarāyar of the Pallava lineage — which descends continuously from Brahmā — was residing in heaven, the mattrakaḷ, the ghaṭakayār and the mūlaprakṛti went (...) in order to appoint (aṭaikka) the mahārāja Hiraṇyavarman of the Kāṭaveśakula, telling [him]: ‘Let the one who is able to assume the charge of king and is pure (paraśuddhar, from pariśuddha) in both lineages, be the [G?]ovalappōttu (i.e. “the bud/the bull who is a powerful king?” or “the bud/bull-king”,77 that is, in both cases, the Pallava king)!’”

43The mixing of Grantha and Tamil letters of this narrative inscription, that is the mixing of Sanskrit and Tamil words with proper differentiated alphabets, makes its language better characterised as epigraphic Maṇipravāḷam than as Tamil.78 Unlike other inscriptions in Tamil where we find occasionally a few words or syllables in Grantha when Sanskrit words are used, here a major part of the vocabulary is Sanskrit and written almost consistently in Grantha letters.79 In some cases Tamil letters are used for a Sanskrit word (kula in its first occurrence and rājabharam for instance). Note also the content. It is a narrative of the events that followed the death of Parameśvaravarman II and how a delegation went to ask his remote cousin Hiraṇyavarman to take over the kingdom. The next narrative inscriptions recount that Hiraṇyavarman refused the throne, that his youngest son, Parameśvara alias Pallavamalla (IP 93, section C) proposed himself, that Hiraṇyavarman was at first reluctant to let him go because of his young age, that Hiraṇyavarman was convinced by someone who was apparently his purohita to do so, and that Parameśvara successfully reached Kāñcīpuram where he received abhiṣeka, some regalia and a coronation name (Nandivarman). So, these inscriptions do not provide a panegyric but a political narrative aiming to legitimate the accession of a collateral branch of the Pallava dynasty, the Kāṭavars we have already discussed above (p. 370). As such, and since they are engraved in a royal temple, these are royal inscriptions. This probably accounts for the reliance on Sanskrit terminology.

44These inscriptions at Trichy, Taḷavāṉūr, Centalai and Kāñcīpuram represent the first stage of Pallava political Tamil. Let us sum up their characteristics. First, their literary nature is obvious, especially in the case of the veṇpā at Taḷavāṉūr and of the verses at Centalai, and, to a lesser degree, in the case of the lists of birudas. Second, most of these examples show either an influence from or an interaction with Sanskrit. The Dravidian and Tamil birudas are intermixed with Sanskrit birudas at Trichy, Pallāvaram, Kāñcīpuram and Centalai. Tamil birudas of Mahendra I are written in a mix of specifically Tamil letters and letters not yet differentiated from Grantha. Some of these birudas are partly composed of Sanskrit words and there is hesitation as to the alphabet: Cittirakārappuli in Tamil script and Citrakārappuli in Grantha script are both attested. The veṇpā at Taḷavāṉūr seems to be a new rendering of an earlier Sanskrit anuṣṭubh. The narrative inscriptions of the Vaikuṇṭhaperumāḷ are so full of Sanskrit words that they deserve to be characterised as a record in epigraphic Maṇipravāḷam. At the same time, even if a Sanskrit praśasti tradition acted as a trigger, a regional tradition or a local initiative is also operative. The polyglossia of the birudas of Mahendra I recalls the bilingual legends of Śātavāhana coins. The veṇpā of Taḷavāṉūr is the work of a poet of the very same locality leaning on the Tamil literary tradition (see the association of the Pallava king with a specific flower). Some stanzas of Centalai are thematically Tamil and possibly draw on earlier Caṅkam literature. Pure Tamil birudas and titles (Kāṭavar or Viṭēlviṭuku) are few in number but the ones that do exist are often used and occur mostly in Tamil local inscriptions. As for Kāṭavar, it appears in the Pallava royal epigraphic corpus only when a collateral branch of local importance so named came to occupy the throne in Kāñcīpuram.

45These inscriptions show that Sanskrit surely played an important role in the rise of political Tamil, but less as a formal or thematic model than as a paradigm to be adapted or translated into a specifically Tamil garb. To close this first section on Pallava political Tamil, I should also mention that the local Tamil literary culture may sometimes influence its Sanskrit counterpart. In the Kacākūṭi copper-plates (ca. 750) stanza 26 (vasantatilakā) reads (IP 77 = SII 2.73.61–62):

etā dhanāni daitāni80 yaśodhanāni
jetā kaler vvilasitāni babhūva tasmāt /
netā nayaśya81 dhaṣaṇādhikṛtasya82 mārgga83
pātā jaga[n*]ti parameśvarapotavarmmā //

“From him (i.e. Narasiṃhavarman II) was born Parameśvarapotavarman (i.e. Parameśvaravarman II), who obtained the desired riches — those of glory — , who conquered the seductions of the kali [age], who was a leader in the path of politics as shown by Dhiṣaṇa (i.e. Bṛhaspati), who protected the worlds”.

46What is remarkable here is the rhyming of the second syllable of each pāda. Although theoretically this literary embellishment is not impossible in Sanskrit (cf. anuprāsa), it clearly recalls the etukai of the Tamil poetry.84

1.6. Pallava political Tamil under Nandin III (ca. 850)

47In the middle of the 9^th century a major shiftseemingly occurred concerning panegyric Tamil under the Pallavas. While Sanskrit praśastis were still composed for Pallava kings, this was done almost exclusively in copper-plates.85 Literary Tamil was fostered at the Pallava court itself, and not under some of their feudatories like in Centalai or through local impetus like in Taḷavāṉūr. One Pallava king played a crucial role in this development: Nandivarman III (hereafter Nandin III), who reigned for a little more than 20 years around 850.

48Nandin III is usually — and maybe rightly — considered as the patron of the Pārataveṇpā of Peruntēvaṉār, which is the most ancient Tamil version of the Mahābhārata, a large portion of which has survived.86 What is certain however is that Nandin III is the hero of the Nantikkalampakam, the most ancient kalampakam extant today. Kalampakam, from kalampu (= kalappu, “mixing, mixture”) and akam (“place”) is translated by Shulman as “mixed bag” (2004: 157) or “inner aspect of mixing” (2004: 175). The genre may owe its name to the variety of themes (tuṟai) and metres it allows.87 Shulman (2004: 161) speaks of a “musical and rhythmic tour de force”. The genre name may also be due to the fact that a kalampakam draws on Puṟam and Akam conventions, actually mixing these two worlds clearly demarcated in Caṅkam poetry.88 The mixing goes to the point of deliberately confusing the pāṭṭuṭai talaivaṉ and the kiḷavi talaivaṉ, in contradistinction to the kōvai genre, another pirapantam.89

49Nanti, the pāṭṭuṭai talaivaṉ of Nantikkalampakam, is described in the poem as a Pallava and is associated to the toṇṭai (the flower of the Pallavas) and to the toṇṭai country, later known as Toṇṭaimaṇṭalam, i.e. the Tamil realm of the Pallavas.90 His three major cities are in the Pallava kingdom: Mallai (Mahābalipuram), Mayilai (Mylapore, today in Ceṉṉai) and Kacci (Kāñcī).91 Since Nanti is described as a Pallava victorious in Teḷḷāṟu, it is clear that he is Nandin III, who bore a specific epithet invoking this deed in local Tamil inscriptions (infra, p. 390)92 and occurring once, verbatim, in the Nantikkalampakam.93

50As I have dealt extensively with this text elsewhere (see Francis 2009: 508 ff., and forthcoming a), I will just sum up the important points. We do not know the name of the poet and there are later traditions presenting him as the brother of Nanti. A major problem with the Nantikkalampakam is that some verses were possibly interpolated. According to descriptions of the genre in the Pāṭṭiyals, a kalampakam should contain 90 verses if the pāṭṭuṭai talaivaṉ is a king.94 The Nantikkalampakam consists of 116 stanzas in the Kaḻakam edition. Some scholars concluded on this ground that stanzas were interpolated.95 And indeed the antāti scheme is not consistently observable. Nevertheless one may question the trust in later norms (Pāṭṭiyals), especially in the case of the first specimen available of a genre presumably predating the treatises setting its norms. Without going into further detail, suffice it to say that some of the verses are only of Puṟam flavour and summarise the royal and martial qualities of Nanti. Particularly interesting is verse 65, presumably genuine and not interpolated because it fits in the antāti scheme:

tiruviṉ cemmaiyum nilamakaḷ urimaiyum
potuv iṉṟi āṇṭa polam pūṇ pallava
tōḷ tuṇai āka māveḷḷāṟṟu
mēvalark kaṭanta aṇṇāl nanti niṉ
tiru varu neṭuṅ kaṇ civakkum ākiṉ
cerunar cērum pati civakkummē
niṟaṅ kiḷar puruvam tuṭikkiṉ niṉ kaḻal
iṟaiñcā maṉṉarkk’ iṭam tuṭikkummē
maiyil vāḷ uṟai kaḻikkum ākiṉ
aṭaṅkār peṇṭir
pūṇ mulai muttap pūṇ kaḻikkummē kaṭu vāy pōl vaḷai atira niṉṉoṭu
maruvā maṉṉar maṉam tuṭikkummē
mā mata yāṉai paṇṇiṉ
utiram maṉṉu(m) niṉ etir malaintōrkkē

“O Pallava with golden ornaments, who ruled (or: cherished), without sharing the excellence (or: beauty) of Tiru (i.e. Śrī) and the possession of Nilamakaḷ (i.e. Bhū),
O Nanti, great man who overcame the enemies in the big Veḷḷāṟu with the help of his arms,
If your large eyes where beauty dwells become red, the place where the [enemy] soldiers gather becomes also red;
If your bright-coloured eyebrows (or: eyebrows with shining lustre) quiver, the leftside of the kings who do not bow to your feet (literally: anklets) also quivers;
If you remove your bright (or: unblemished) sword from its sheath, the wives of reluctant [kings] also remove the pearl-ornaments of their ornamented breasts.
When conchs similar to your kaṭuvāy (“harsh voice”, i.e. Nanti’s drum) resound, the determination of the kings who are not on your side falters (quivers);
If you prepare your big intoxicated elephant(s) [for battle], the blood of the kings opposing you congeals”.

51In this verse of Nantikkalampakam there are undoubtedly themes borrowed from the Sanskrit literary tradition, such as the association of goddesses Śrī and Bhū with the king or the description of elephants as intoxicated (mata yāṉai).96 But the vocabulary is mostly Tamil and the metre is the profoundly Tamil āciriyappā, which is also the principal metre of Puṟanāṉūṟu. These features as well as the focus on war victories, the mention of an actual battle (at Veḷḷāṟu),97 and the introduction with the goddesses, are similarly found in the Cōḻa meykkīrttis (i.e. Tamil epigraphic panegyrics) which consist almost exclusively — after the mention of the goddesses — in the enumeration of victories, if not historical at least set in an historical context.98

52As for the practices in inscriptions, the reign of Nandin III is marked by an innovation. As far as I know, he is the first king to bear in local Tamil inscriptions a recurrent and specific Tamil epithet (properly a relative participle clause) prefixed to his name which is itself followed by his regnal year as internal date. This epithet — teḷḷāṟṟ’ eṟinta (see Fig. 8) which means “who vanquished (or: smashed; or: blew, as the wind) [the enemies] at Teḷḷāṟu” — occurs with slight variants (erinta; eṟiñca; teḷḷāṟṟ’ eṟintu rājya[mu]n koṇṭa in IP 141), in approximately 25% of the Tamil inscriptions dated to the reign of Nandin III.99 In most of them the mention of Nandin III has no other purpose than providing a date: the king is not involved in the transaction recorded. These are mostly local inscriptions, so the exact role of the king in imposing this epithet is debatable. But we can guess that he himself provided the impetus in this matter (on meykkīrttis and their seeming imposition by Cōḻa kings at least in the early period, see Francis and Schmid 2010: xlvi).

53If we thus consider verse 65 of the Nantikkalampakam side by side with this epigraphic innovation under Nandin III, we have here the forerunner of the Cōḻa meykkīrttis. In this context I must recall that Cōḻa meykkīrttis, whose first example is that of Rājarāja I a little before 1000, are mainly encountered in local Tamil stone inscriptions, in which the Cōḻa king plays sometimes no other role than providing a regnal date, and that the use of brief epithets, based on the model of teḷḷāṟṟ’ eṟinta, is known for Cōḻa kings since the 9^th century.100 Cōḻa meykkīrttis can thus be considered as a development of the Tamil panegyric of Nandin III: they are a versified development of epithets like teḷḷāṟṟ’ eṟinta and a thematic avatar of a poem like Nantikkalampakam 65.

54In this second stage of the political use of Tamil under the reign of the Pallava Nandin III, we can thus identify a development which is more autonomous from Sanskrit tradition and which assumes the heritage of the Caṅkam literature, a literature that has not been referred to in detail by Pollock. Under Nandin III we have a mature literary form of vernacular Tamil political discourse with the Nantikkalampakam and we also have the prefixing of an epithet to the king’s name in local Tamil inscriptions, both phenomena that, taken together, give us the forerunner of Cōḻa meykkīrttis.

2. Vernacularisation: a literarisation of the vernacular?

55The evidence presented above helps to draw the picture of the rise of “an aestheticized public political discourse in Tamil” sketched by Pollock (2006: 322). It entails that the theory of vernacularisation he advances must at least be qualified concerning the Tamil area. To proceed in this way I will first recapitulate my findings and summarise Pollock’s vernacularisation theory.

56In a first stage Pallava epigraphic political Tamil was indebted to, or at least associated with or triggered by political Sanskrit. Remarkably all the evidence reviewed consists of stone inscriptions as opposed to copper-plates, which are always royal inscriptions. Most of these stone inscriptions are local (Taḷavāṉūr veṇpā, inscriptions with Tamil birudas and titles, Tiruttaṇi) or records of vassals of the Pallavas (Vallam, Centalai, Tiruveḷḷaṟai). Royal inscriptions (lists of birudas of Mahendra I, Vaikuṇṭhaperumāḷ) show less influence from the Tamil literary tradition.

57In the second stage, under Nandin III (ca. 850), a tendency already at work in the first stage — that is, the reliance on specific Tamil literary conventions and themes — is intensified and reaches the Pallava court. Tamil political discourse then becomes rather autonomous vis-à-vis the Sanskrit model or antecedent. This trend was perpetuated with the Cōḻa meykkīrttis, which followed the path that Nandin III paved in a court poem (Nantikkalampakam) and in local Tamil inscriptions. The Cōḻa meykkīrttis are, in a sense, heirs of the Puṟam literature of the Caṅkam corpus with regard to metre, the heroic and martial stature of the king — which cannot be considered as imported from Sanskrit as implied by Pollock (2006: 322) — , the mention of his name and actual victories. Nandin III thus appears as the missing link between Puṟam and the Cōḻa meykkīrttis. Tamil as a cosmopolitan vernacular actually drew minimally on the superposed Sanskrit tradition but maximally, and more and more as time passed, on Tamil antecedents or specific traits. This conclusion now has to be confronted with Pollock’s concept of vernacularisation.

58By vernacularisation, Pollock means “the historical process of choosing to create a written literature, along with its complement, a political discourse, in local languages according to models supplied by a superordinate, usually cosmopolitan, literary culture” (op. cit.: 23). This process has three connected components: literisation, literarisation and superposition (i.e. “presence of a dominant language and literary formation”). The superordinate literary cultures in premodern India are “Sanskrit, but also to some degree Prakrit and Apabhramsha (...), Tamil in some areas of South India, and, much later, Persian, in areas of the north” (op. cit.: 26). Vernacularisation began around 1000, a little earlier for Tamil. In this process, to quote Ali (2011: 280) commenting upon Pollock, “Sanskrit, in other words, taught the vernaculars how to be literary”.

59Although Tamil is considered by Pollock as one exception regarding bhāṣākāvya in the first millennium (since kāvya-like literature existed early on in Tamil) and is included among the superordinate literary cultures, actually Pollock does not draw consequences from the early existence of a Tamil literary tradition. The reason is that Tamil literary prehistory is “unique, and uniquely obscure”, “opaque”, “uncommonly obscure” (op. cit.: 383, 292). Pollock argues that “disentangling fact from fiction in Tamil literary history is complicated, and the more reliable the date (as in inscriptions), the more the Tamil case conforms with the general picture of literary South Asia” (op. cit.: 100). I will not deny that the history of the Caṅkam corpus is a complex and debated matter, but when Pollock presents (op. cit.: 383–385) a description of Tamil in the first millennium, he only mentions facts that go well with his general theory: the late appearance of the legend of the Caṅkam, the written character of Bhakti poetry, the commentatorial writing modelled on Sanskrit, the adaptation in Tamil of Sanskrit epics, the philologisation of language, the conceptual borrowing from the Sanskrit grammatical tradition. Not a word about the Puṟam literature centred on kings, about the tiṇai concept so specific to Caṅkam poetry, about the pirapantams (a specifically Tamil type of work, even though its name is Tamilised Sanskrit) which are the heirs of Tamil Caṅkam conventions, however subverted they may be sometimes. As for the inscriptions, reliable as they are, only two of the Pallava inscriptions discussed above are mentioned by Pollock, and then only in footnotes.

60So for Pollock, as far as the expression of political power in the Tamil country is concerned, it was in Sanskrit, for much of the first millennium (op. cit.: 384). This would have begun to change in the 8^th or 9^th century with Pāṇḍya records “and even more dramatically in the Cōḻa realm around 1000 with the invention of the Tamil meykkīrtti and the ornate historiographical records of Rājarāja I and his successors” (op. cit.: 384). These Pāṇḍya and Cōḻa records are discussed as examples of “cosmopolitan vernacular” (op. cit.: 322–323). Wishing to illustrate that “any new vernacular literature (...) had to acknowledge and come to terms with the superposed model established in the cosmopolitan tradition [i.e. Sanskrit] for everything from lexicon and versification to figures, genres, and themes” (op. cit.: 322), Pollock uses a verse from the Pāṇḍya Cēntaṉ that he dates tentatively to the 8^th century (Vaikai riverbed inscription; EI 38.4 ff. = IEP 4).101 He draws attention to the mixing of Tamil and Sanskrit words it shows — differentiated by the use of two scripts, Vaṭṭeḻuttu and Grantha, which makes this inscription an example of epigraphic Maṇipravāḷam, like the Vaikuṇṭhaperumāḷ narrative inscriptions (supra, p. 384) — and argues rightly that we have here “the language and style of Sanskrit political poetry gradually being domesticated to the ways of the Tamil world”. Pollock goes on to say that the Cōḻa meykkīrttis were developments of “a piece with this sort of Pāṇṭiya innovation” (op. cit.: 323). Meykkīrtti he defines as “an account of ‘true fame’ or ‘personal fame’, a new public literary-political genre that details the genealogy and achievements of a king along with mentions of his biruḍas, or titles [...] marked throughout by a cosmopolitan-vernacular idiom embodied in the term itself, which offers a blending of Tamil (mey) and Sanskrit (kīrti) that epitomises the larger cultural (and political) synthesis under way”. Then Pollock introduces Maṇipravāḷam which “embodied the very process of localization of the Sanskrit universal, in both political discourse and literature, that was occurring across southern Asia from this moment on, with the vernacular at first supplementing the Sanskrit and later taking on an ever-increasing proportion as vernacularization gained power and confidence”.

61Several reasons urge me to not fully endorse Pollock’s views on the vernacularisation of political Tamil discourse.

62First, there is an ancient Tamil literature going back to the first centuries of the first millennium, the so-called Caṅkam corpus known through later anthologies. Pollock quite rapidly dismisses this Caṅkam corpus as being “obscure” or “opaque” as far as its history and development are concerned. Even though the date of any particular poem in these anthologies and the status of this literature — the usual opinion is that a bardic oral tradition was taken over by a literary written tradition responsible for its anthologisation — is debatable, it is nevertheless generally accepted that already several centuries before Pollock’s vernacularisation, that is relatively early in the first millennium, there existed an expressive or imaginative Tamil for the description of kings (that is the Puṟam poetry; see also the theme named pāṭāṇ in poetics).102 However thorny the dating, the consensus is that at least before the middle of the first millennium, Caṅkam poetry, which is a worldly literature comparable to kāvya, was established.103 However late a written aspect may have come to attach itself to this Caṅkam poetry, the anthologisation as well as the later creation of a Caṅkam legend proves that there was an interest for this early literature in the Tamil country, that it was an early revered and known beginning.104

63Second, Pollock’s ensuing view that “there is no doubt that for many centuries Tamil was, if not mute, then certainly not loquacious in the domain of political-expressive, until in the last several centuries of the millennium it began to speak with an altogether new and confident voice” (2006: 292) is an overstatement. The Pallava evidence provided above attests to the relatively early appearance of political, literary, expressive or imaginative epigraphy in Tamil, that is at the beginning of the 7^th century, as opposed to the 8^th century, at the earliest, for Pollock. We may find that the lists of birudas are not very imaginative. But the veṇpā at Taḷavāṉūr and the Centalai verses are good pieces of political poetry.

64Third, if a Tamil cosmopolitan vernacular existed that was like Cēntaṉ’s verse or early Pāṇḍya copper-plates (IEP 7, 11, 16, 61, 79, 90, and Āvaṇam 18.1) modelled on Sanskrit, it is in no way a general rule that Tamil cosmopolitan vernacular relied on Sanskrit. It may be triggered by the antecedent of Sanskrit, but relies heavily on Tamil literary culture. Such was the case at Taḷavāṉūr, at Centalai, and in the Nantikkalampakam. The early Pāṇḍya records could well represent a vernacularisation as represented by Pollock, but they are relatively isolated.105 They can compare with some of the instances of Pallava political Tamil from around 600 CE, triggered, as we saw, by Sanskrit, but this is only one part of the picture. We know also of Pāṇḍya inscriptions entirely in Tamil, literary or political, that do not follow strictly this Sanskrit-modelled thread and are, like the Taḷavāṉūr veṇpā and Centalai verses, true examples of Tamil poetry.106 The innovations under Nandin III show that at least from 850 onwards, the literary reference was increasingly Tamil, not Sanskrit. This reference is inherited by the Cōḻa meykkīrttis. It may be true that without the Sanskrit cosmopolis we would not have cosmopolitan Tamil, but the latter is less and less modelled on the Sanskrit paradigm. It is the cosmopolitan vernacular of early Pāṇḍya copper-plates that could be best described as “unique experiments, with no prehistory and no future” (Pollock, 2006: 291, n. 12), not the Centalai inscriptions.

65Fourth, the definition of Cōḻa meykkīrttis by Pollock (2006: 323) is based on an approximate paraphrase of the Paṉṉiruppāṭṭiyal 198 given in the Tamil Lexicon: in Cōḻa meykkīrttis there is no place for genealogy, which is instead a common feature of praśasti. Cōḻa meykkīrttis cannot be considered as a development of a Pāṇḍya cosmopolitan vernacular. If early Pāṇḍya copper-plates, portions in Tamil are indeed genealogy, these are, I believe, not meykkīrtti (contra Pollock 2006: 323, n. 70; see Francis and Schmid, 2010: xiv). In Cōḻa meykkīrttis Sanskrit words in Grantha are extremely rare and Sanskrit words are usually Tamilised, in contradistinction to what is observable in Cēntaṉ’s inscription and in the early Pāṇḍya copper-plates. In that sense Cōḻa meykkīrttis are by no means instances of epigraphic Maṇipravāḷam. The model for Cōḻa meykkīrttis is Caṅkam literature and the epigraphic example of Nandin III (see Nantikkalampakam 65 and epigraphic titulature).107 The Cōḻa meykkīrttis are less an adaptation in Tamil of Sanskrit models than the continuation of a long thread of royal description in Tamil going back as far as the Puṟam poetry of the Caṅkam corpus through a host of Tamil antecedents, epigraphic and literary.108

66If we try to characterise the political discourse of the early Pāṇḍyas (8^th–10^th century), on the one hand we find Tamil eulogies, modelled on Sanskrit praśastis, in copper-plates (following the praśastis) and in a stone inscription like Cēntaṉ’s. On the other hand, in the same Tamil eulogies of copper-plates, the Pāṇḍyas claim association with the Caṅkam academy and with a Tamil rendering of the Mahābhārata. These copperplates are instances of how the Sanskrit tradition shaped the discourse of power in Tamil, but this was done more profoundly than in the first stage of Pallava political Tamil. The early Pāṇḍyas are in a sense paradoxical: they present themselves as linked to the ancient Tamil literary tradition while at the same time adopting a heavily sanskritised Tamil political discourse. The later Pāṇḍyas (11^th–15^th centuries) followed the meykkīrtti model of the Cōḻas, which was a continuation of Puṟam, and were even more faithful heirs of Puṟam: their āciriyappās are more classical, conforming more closely to the Caṅkam type, than the Cōḻa ones.109

67As for the Cōḻas, we have to distinguish two threads. On the one hand we have praśastis, in fact relatively few, mostly on copper-plates. The Cōḻas ruled the territory of the Pallavas (and later the Pāṇḍya territory through a vice-royalty) and their copper-plates follow the Pallava model (that is without an expressive versified Tamil portion), not the Pāṇḍya one. These praśastis are remnants of Pollock’s Sanskrit cosmopolis of the first millennium. On the other hand, we have no less than 40 different meykkīrttis for the 15 Cōḻa kings. These do not, in my opinion, represent a vernacularisation as conceived by Pollock, that is, a literarisation of the vernacular with Sanskrit as model. These Cōḻa meykkīrttis are heirs, through Pallava antecedents, of an ancient Tamil literary tradition. The definition of meykkīrtti given by Pollock as embodying in the term itself two streams is exaggerated since the Sanskrit influence seems finally to be minor.

68So, if vernacularisation as conceived by Pollock concerns only the early Pāṇḍyas, what is the impact of this fact on his general theory? Criticism of Pollock’s concept of vernacularisation has already been expressed from a variety of perspectives,110 but I will concentrate here on Tamil. Since vernacularisation as conceived by Pollock played only a minor role for Tamil, as I have tried to demonstrate, and since there was an ancient Tamil literary tradition, we arrive at the question why, before political discourse in Tamil, Sanskrit was adopted, for centuries, as the language of power in certain sectors of the Tamil area.

69One has to keep in mind that Sanskrit appeared relatively late in the epigraphy of the Tamil country. Sanskrit is used intensively in royal Pallava epigraphy only from the middle of the 6^th century (IP 17) onwards. Earlier Pallava Sanskrit praśastis exist, but mostly in the Andhra region. Only one has been found near Kāñcīpuram (IP 10), but it is in prose, not in verse as Pallava praśastis usually are from ca. 550. It thus seems that before the Pallavas, usually considered as culturally oriented towards the north, there was no need in the Tamil country for a discourse of public power, be it in Tamil or in the Sanskrit, supposing that Puṟam poetry had no public dimensions. In any case, the Pallavas played a decisive role in making the Tamil country part of the Sanskrit cosmopolis.

70Pollock argues that Sanskrit was chosen for its aesthetic dimension (power and culture being inseparable, see Pollock 2006: 2 ff.), not because it was a language associated with Hinduism. But since there was an ancient Tamil literary tradition — as Caṅkam poetry certainly presents a form of the language with significant aesthetic dimensions — why then did Tamilian dynasties not rely on it straight away, instead of waiting until after the Sanskrit discourse of power had been established to choose Tamil? In other words, why is it that Sanskrit served as a model in the case of the early Pāṇḍyas, and as a trigger in the case of the Pallavas? The choice of Sanskrit by the Pallavas may be explained not only by the fact that it enabled the aestheticisation of power through its specific features (on these, see Pollock 2006: 254 ff.), but also because Sanskrit was associated with Hinduism, a rationale Pollock dismisses (op. cit.: 74).111

71Pallava Sanskrit, with the exception of IP 17, is linked to the establishment of Brahmanic institutions (brahmadeya) and Hindu temples (devadāna). When we know the authors of these praśastis, their names reveal they are brahmins.112 It is then conceivable that while dynasties turned away from Buddhism — too moralist, not sufficiently sanctifying the violence of royal duty — towards Hinduism, they adopted Sanskrit because it was the language of that religious culture (even though it could be shared by others) and of the Brahman elite. Pollock shows that Sanskrit was adopted first by foreign dynasties which could emancipate Sanskrit from its religious background, that is, secularise it.113 That may be true, but it doesn’t follow that the same logic operated everywhere, every time. Sanskrit had to be secularised to be able to serve as the language of politics, but once this had been accomplished, it could be adopted also precisely because it was associated with the specific religious specialists that the brahmins are. What I want to say is that in a Brahmanic world, dynasties could choose Sanskrit because it allowed them to pretend to be dharmic kings in the eyes or ears of Brahmans, or even to challenge Brahmanic discourse on kingship in its own language.114

72This suggestion goes against Pollock’s view about religion in the choice of Sanskrit, and against his criticism of the concepts of brahmanisation, sanskritisation, legitimation, ideology and functionalism (op. cit.: 511 ff.). I believe these theories indeed have their flaws, several of which are pointed out by Pollock, but that the analysis of political discourse in Tamil invites us to refine them, not to dismiss them and replace them entirely with Pollock’s aesthetic theory. I contend that kings were engaged in an ideological debate for legitimation not with the people dominated, but with the dominant Brahmanic Sanskrit ideology and with rival dynasties. They chose Sanskrit because, after it had been secularised, it was available for political discourse and because it was the language of those at whom their discourse was directed: the brahmins as source of authority — or of contested authority — and the other competing dynasties. They chose Sanskrit as a political lingua franca permitting a discursive relation with Brahmin ideology and rival dynasties.115

73As Sanskrit developed as a language of political discourse, it triggered expression in the vernaculars. Such expressions surfaced first in the Tamil country, because it already knew a literary tradition to latch onto. The early Pāṇḍyas adopted a political Tamil fitting the model of vernacularisation described by Pollock. In the Pallava realm, a political Tamil relying more on the Tamil literary culture was adopted, at first mainly on local initiative (Taḷavāṉūr; Centalai, from a minor dynasty that never used Sanskrit). This trend finally reached the Pallava court under the reign of Nandin III.116

74In this rise of political Tamil we can discern an early form of regionalism that can be related to the process of “territorialization of Tamil cultural space” and to the creation of a “political regionality in the premodern Tamil world” mentioned by Pollock (2006: 385–86, 413 ff.). In a sense the intrusion of Sanskrit caused this rise: Tamilians took conscience of their difference in an India globalised through the Sanskrit cosmopolis, and wanted to show their Tamilness in the language of public political discourse itself. From the local level this process arrived at the court. It is maybe symptomatic that this happened in the middle of the 9^th century, at a time when the Pallava dynasty was near its end, no longer playing a major role in South India.