Issue |
Histoire Epistémologie Langage
Volume 39, Number 2, 2017
La grammaire sanskrite étendue
|
|
---|---|---|
Page(s) | 89 - 102 | |
DOI | https://doi.org/10.1051/hel/2017390205 | |
Published online | 18 April 2018 |
On (the) sandhi between the Tamil and Sanskrit grammatical traditions
University of Hamburg,
Hamburg, Germany
This article provides a few observations on some of the theories that the Sanskrit and the Tamil grammatical traditions share about the contexts in which the sounds of words change and which kinds of changes these sounds can undergo. The study shows that even if Tamil grammarians freely rearranged the Sanskrit material and adapted it to their concerns, it is nonetheless plausible to claim that there was a transfer of ideas from the Sanskrit tradition to the Tamil one also in what we could call the field of phonology.
Résumé
Cet article livre quelques observations relatives à certaines théories que les traditions grammaticales sanskrite et tamoule partagent concernant les contextes dans lesquels les sons des mots changent, ainsi que les types de changements que ces sons subissent. L'étude montre que, même si les grammairiens tamouls réarrangent librement le matériel sanskrit et l'adaptent à leurs préoccupations, on peut néanmoins affirmer qu'il y a eu un transfert d'idées de la tradition sanskrite vers la tradition tamoule aussi dans le domaine que nous appelons « phonologie ».
Key words: phonology / sandhi / Sanskrit / Tamil / Tolkāppiyam / Vājasaneyiprātiśākhya
Mots clés : phonologie / sandhi / sanskrit / tamoul / Tolkāppiyam / Vājasaneyiprātiśākhya
© SHESL/EDP Sciences
Introduction
Sanskrit and Tamil grammars often share common linguistic models.1 For the great part, this is the result of the Tamil tradition tapping into the repertoire of ideas developed to describe Sanskrit. However, one should be aware that Tamil grammarians were not at all slavish. It is actually a rather intriguing exercise that of trying to understand not only what was adopted from one tradition to the other and from which source(s), but also to figure out how the adopted ideas were in turn adapted to account for the peculiarities of Tamil and how these ideas blended with the original thinking of the Tamil grammarians.2
In this article, I will offer a few observations on some of the theories that the two traditions share about the contexts in which the sounds of words change and which kinds of changes these sounds can undergo. Borrowing a linguistic parlance that is both modern and Western, I will deal with phonology, and in particular with the conditions of application of phonological rules, i.e. sandhi rules, and the categories such rules may be grouped in.3 Furthermore, I will also try to highlight the fact that although phonological theories are indeed shared by the two traditions under scrutiny, these are in fact embedded in different linguistic frameworks.
The textual sources I will concentrate on in this study are, from the Tamil side, the Eḻuttatikāram (“Book on Sounds/Letters”), i.e. the first part of the Tolkāppiyam (“The Grammar of Tolkāppiyar”), the foundational grammar of Tamil composed most probably some time during the 1st half of the 1st millennium CE. I will also take into account the two extant commentaries of the Eḻuttatikāram composed by Iḷampūraṇar. (12th c.?) and Naccinārkkiṉiyar (14th c.?). From the Sanskrit side I will – for reasons that will be made clear later on – mostly focus on the Vājasaneyiprātiśākhya (attributed to Kātyāyana), i.e. the Prātiśākhya of the Śuklayajurvedasaṃhitā, along with Uvaṭa's commentary (12th c.).4 En passant, I will also mention other Prātiśākhyas and the Aṣṭādhyāyī of Pāṇini (5th c. BCE?), i.e. the foundational work of Sanskrit grammar, and its first complete commentary, namely the Kāśikāvṛttī of Jāyaditya and Vāmana (7th c.).
The Tamil theory of sound combination
Basic information about the context where phonological rules apply and the kinds of rules characterising Tamil is found in a few rules of the Eḻuttatikāram, in particular in its sub-section called Puṇariyal (lit. “The study of Combination”).5
In rule TEn107 we find the expression puṇar-nilai (“state of combination”), which is used to define the linguistic state where two linguistic items occur one after another, thus creating the condition for the last sound of the first item to combine with the first sound of the following one:
uyiriṟu coṉmu ṉuyirvaru vaḻiyu
muyiriṟu coṉmuṉ meyvaru vaḻiyu
meyyiṟu coṉmu ṉuyirvaru vaḻiyu
meyyiṟu coṉmuṉ meyvaru vaḻiyumeṉ
ṟivveṉa vaṟiyak kiḷakkuṅ kālai
niṟutta collē kuṟittuvaru kiḷaviyeṉ
ṟāyī riyala puṇarnilaic cuṭṭē (TEn107)
The case in which a vowel occurs after a linguistic item (col) ending in vowel, the case in which a consonant occurs after a linguistic item ending in vowel, the case in which a vowel occurs after a linguistic item ending in consonant, the case in which a consonant occurs after a linguistic item ending in consonant – when it is clearly explained that [these four combinations] are such, those two kinds [of linguistic items], namely the standing item (niṟutta col) and the item that occurs referring [to the standing one] (kuṟittu varu kiḷavi), are the mark of a state of combination (puṇar-nilai).
In TEn108 we are informed that the items that can be found in a state of combination are nouns and verbs and that their final and initial sounds (may they be vowels or consonants) can undergo four kinds of puṇar-iyalpu (“modes of combination”), among which three entail tiripu (“change”) and one iyalpu (“absence of change”, lit. “natural state”):
avaṟṟuḷ
niṟutta colli ṉīṟā keḻuttoṭu
kuṟittuvaru kiḷavi mutaleḻut tiyaiyap
peyaroṭu peyaraip puṇarkkuṅ kālum
peyaroṭu toḻilaip puṇarkkuṅ kālun
toḻiloṭu peyaraip puṇarkkuṅ kālun
toḻiloṭu toḻilaip puṇarkkuṅ kālu
mūṉṟē tiripiṭa ṉoṉṟē yiyalpeṉa
vāṅkan nāṉkē moḻipuṇa riyalpē (TEn108)
Among these [states of combination], as the first sound of the item that occurs referring to the standing item joins with the final sound [of the latter], when one combines a noun with a noun, when one combines a verb with a noun, when one combines a noun with a verb, and when one combines a verb with a verb, there are four modes of combination of linguistic items (moḻi), namely the occurrence of three [kinds of] change (mūṉṟē tiripu iṭaṉ) and one [case of] absence of change (iyalpu).
A caveat is here in place. Tolkāppiyar employs at least three different terms, which I have for the time being translated as “linguistic item”, namely col, kiḷavi, and moḻi. These are all used – apparently interchangeably, possibly metri causa? – in the rules under consideration: col and kiḷavi in TEn107, and moḻi in TEn108. Throughout the Tolkāppiyam we can observe that these terms can at times correspond to the English words “word” (whatever its actual definition may be) or “morpheme”.6 TEn108 seems however to induce us to interpret all these terms as meaning (more or less) “word”, since it states that it is nouns and verbs that can undergo a state of combination. In section 3 we will see that terminology allows the commentators of the Eḻuttatikāram to tackle with internal sandhi, a major theoretical concern that is described − but not overtly theorised – in the Eḻuttatikāram.7
Reading on in the list of rules, TEn109 specifies that the three changes that sounds can undergo are the modification of a consonant (mey-piṟitu-ātal), expansion (mikutal), and elision (kuṉṟal):
avaitām
meypiṟi tātaṉ mikutal kuṉṟaleṉ
ṟivveṉa moḻipa tiriyu māṟē (TEn109)
Modification of consonants, expansion, and elision. They say that such are the ways of changing.
TEn112 further points out that, as far as expansion is concerned, this can consist of a single sound (eḻuttu) or a group of sounds (cāriyai), despite the two items in combination may or may not share a syntactic relationship:
vēṟṟumai kuṟitta puṇarmoḻi nilaiyu
vēṟṟumai yalvaḻip puṇarmoḻi nilaiyu
meḻuttē cāriyai yāyiru paṇpi
ṉoḻukkal valiya puṇaruṅ kālai (TEn112)
At the time of combination, the state of a linguistic item in combination that is denoted by case relationship and the state of a linguistic item in combination that is without case relationship can manifest itself by means of two characteristics, i.e. a sound or a group of sounds.8
A further terminological caveat is in place here. I have translated cāriyai as “group of sounds”, but one could also translate it as “morpheme”, in particular in the sense of augment or oblique suffix.9
The commentators' interpretation of TEn107 and TEn108
The set of rules that we have just seen tells us that phenomena pertaining to the combination of speech-sounds, i.e. sandhi phenomena, happen between two linguistic items, and that these are actually words of two kinds, namely verbs and nouns. However, these rules seem not to take into account the fact sandhi phenomena occur also inside words, when roots and affixes combine. Nonetheless, since there are rules that account for these events later on in the Eḻuttatikāram,10 the commentators felt the urge to engage with this theoretical gap, so apparent in the set of rules that are supposed to outline the general framework for the combinations of speech-sounds.
In this respect, Naccinārkkiṉiyar explicitly remarks that the words that can actually be found in puṇar-nilai are also cāriyais and urupus (“case-endings”). In particular, he mentions the word viḷaviṉai (“eagle-wood tree”), which can be split as viḷa- (nominal root) + -v- (glide sandhi) + -iṉ- (cāriyai, or oblique marker) + -ai (accusative ending):
viḷaviṉaik kuṟaittāṉ eṉṟavaḻic cāriyaiyum urupum nilaimoḻiyāyē niṟkumeṉpatu nōkki ataṉai niṟutta colleṉṟum muṭikkuñcollaikkuṟittu varukiḷavi yeṉṟum kūṟiṉār | itaṉāṉē nilaimoḻiyum varumoḻiyuṅ kūṟiṉār |
When one says ‘he cut (kuṟaittāṉ) the apple-wood [tree] (viḷaviṉai)’, aware that cāriyai and case suffix occur as the standing linguistic items, [Tolkāppiyar] defined those [i.e. the cāriyai and the case suffix as] the linguistic items that stands in front and the item that occurs referring to the [preceding] item that it completes. Exactly because of this [Tolkāppiyar] talked about the standing item and the occurring one.
This consideration paves the way to rethink the scope of rule TEn108, which seem to prescribe that the words that can be found in puṇar-nilai are verbs and nouns. In this respect, Naccinārkkiṉiyar intends to include also the other two categories of words that are singled out by Tamil grammar, namely, iṭai-c-cols (“particles”, which according to rule 250 of the Collatikāram also include urupus and cāriyais) and uri-c-cols11:
iṭaiyum uriyuntāmāka nillāmaiyiṟ peyarviṉaiyē kūṟiṉār | iṭaiccollum uriccollum puṇarkkuñ ceykaippaṭṭuḻip puṇarppuc ciṟupāṉmai |
[Tolkāppiyar] mentioned nouns and verbs because iṭai[-col] and uri[-col] do not occur by themselves. When a [linguistic] procedure occurs in which an iṭai-c-col or an uri-c-col are in a [state] of combination, it is less frequent to have a combination [of sounds].
On his part, Iḷampūraṇar at first seems to engage with the same issue more indirectly when commenting upon TEn107:
viḷaviṉaik kuṟaittāṉ eṉpatu avvurupu kuṟittuvaru kiḷaviyai nilaimoḻiyuḷ aṭakki irumoḻippuṇarcciyāy niṉṟavāṟu aṟika |
One should understand how the sentence ‘he cut (kuṟaittāṉ) the apple-wood [tree] (viḷaviṉai)’ occurs as a combination of two words, adjoining the item that occurs having targeted that case ending [i.e. the accusative ending -ai] to the standing item [i.e. viḷa].
Here, we have to understand that the occurrence of the cāriyai -iṉ- is a sandhi phenomenon that can occur only if there is a state of combination, which in this case does occur, albeit between morphemes and not between words. Furthermore, Iḷampūraṇar's remarks about TEn108 clearly reveal that he also intended to expand the scope of the rule to include iṭai-c-cols and uri-c-cols:
iṭaiyum uriyum peyar viṉaikaḷai aṭaintallatu tām āka nillāmaiyiṉ, peyar viṉaikaṭkē puṇarcci kūṟappaṭṭatu |
Since iṭai[-cols] and uri[-cols] do not occur by themselves without joining nouns and verbs, combination is described only for nouns and verbs.
To summarise, the commentators are making explicit a feature of the Eḻuttatikāram that seems to be left unspoken by the text itself, i.e. that both rules of internal and external sandhi are accounted for. In this respect, the wording of TEn108 remains puzzling. If we were to read the Eḻuttatikāram as an autonomous text, i.e. without the help of its commentaries, one could also think that TEn108 may reflect − or, in other words, be a relic of – an earlier stage of the Tamil grammatical speculation that limited itself to the description of phenomena affecting words, without investigating their sub-components and the phenomena their sounds can undergo (see section 4 for the Sanskritic counterpart of this approach). Alternatively, this rule may just serve as a general introduction, which is then de facto surpassed by the concern of the rules dealing with internal sandhi. What is certain is that both Iḷampūraṇar and Naccinārkkiṉiyar are rather explicit in moving beyond the apparently restricted scope of TEn108.
Similarities between Tamil and Sanskrit theories
I would like now to point at two striking similarities that characterise the Tamil and Sanskrit conceptualisation of sounds' combination, namely the definition of the context in which sandhi occurs and the classification of the kinds of sandhi rules.
I will try to do so by focusing on the texts that, to the best of my knowledge, are particularly relevant for drawing a comparison with the Eḻuttatikāram and its commentaries (in particular Naccinārkkiṉiyar's one), namely the Vājasaneyiprātiśākhya and its commentary by Uvaṭa, called Mātṛmodaka.
Puṇar-nilai and saṃhitā
The expression puṇar-nilai, which is used to refer to two words coming one after the other to form the condition of combination of sounds, echoes the Sanskrit term saṃhitā. The Vājasaneyiprātiśākhya in fact teaches that a sound is combined (saṃhitaḥ) to the one preceding it in rule 1.155:
pūrveṇottaraḥ saṃhitaḥ (VP1.155)
A following [sound (varṇa m.)] is combined with the preceding [sound].
Uvaṭa further specifies that the combining sounds are the last and the first of the preceding and the following words respectively and that such condition of combination is called saṃhitā:
ita uttaraṃ saṃhitocyate | pūrveṇa padāntena uttaraḥ padādiḥ [|] saṃhitā yadā kriyate svarato varṇataś ca tadā dvipadasaṃhitocyate | yathā – “iṣé ttvā tvorjé” | kramasaṃhiteyam ||
From here onwards the state of combination (saṃhitā) is mentioned. The preceding word-final [sound] is combined with the following word-initial [sound]. When the state of combination is made by means of pitch-modulations and speech-sounds, then it is called state of combination concerning two words. For instance: iṣé ttvā | tvorjé (Śuklayajurvedasaṃhitā 1.1). This is a krama state of combination/recitation.12
The resemblance with the instructions of TEn107 and 108 is striking, but the overall approach to this issue adopted by the Vājasaneyiprātiśākhya is more complex than what it may appear at first. In fact, the Vājasaneyiprātiśākhya contains another rule, namely 1.158, that defines saṃhitā as a contiguity of sounds:
varṇānām ekaprāṇayogaḥ sa̐hitā (VP1.158)
The state of combination is the union [i.e. the articulation] in a single breath of speech-sounds.
Uvaṭa indicates a way out from this seeming contradiction between rules 1.155 and 1.158. In fact, he states that the latter does not concern metrical texts (ṛcs), but only prose text (yajus), i.e. the prose parts of the Śuklayajurvedasaṃhitā:
evaṃ tāvat pādasaṃhitā ṛkṣu karttavyā | yajuṣṣu tv ayaṃ vidhiḥ [|] varṇānām ekocchvāsoccāraṇayogaḥ pade vā vākye viśrāmaḥ [|] sā ca prāṇasaṃhitā |
Thus, in the case of metrical formulas (ṛkṣu) one must create a state of combination13 depending on the verses [i.e. of the words within the verses (as per rule 1.15714)], but in the case of yajuses [one should follow] the [present] rule. The articulatory employment of a single expiration of sounds [corresponds to] the cessation [of the articulation]15 for a word or a sentence. And that is the state of combination depending on the breath.
We are therefore presented with two possible models accounting for the context of application of sandhi rules: one specifying that the combining items are the sounds found at the edges of words, and one for which the combining items are simply sounds.
Other Prātiśākhyas comply to the former model, as for instance Ṛgvedaprātiśākhya 2.1 saṃhitā padaprakṛtiḥ (“the state of combination has [individual] words as its raw matter”) and Taittirīyaprātiśākhya 24.3 nānāpadasandhāna-saṃyogaḥ padasaṃhitety abhidhīyate (“the conjunction of several words is called state of combination”). The definition of saṃhitā that involves only sounds is instead adopted in vyākaraṇa, i.e. in what is mostly the tradition inspired by Pāṇini's Aṣṭādhyāyī. Rule 1.4.109 of the Aṣṭādhyāyī reads paraḥ saṃnikarṣaḥ saṃhitā (“the state of combination is the most immediate contiguity”), concerning which for instance the Kāśikāvṛttī comments saṃnikarṣaḥ as varṇānām ardhamātrākālavyavadhānam (“the intervention of a half moraic lapse among speech-sounds”). It seems therefore possible to argue that the Eḻuttatikāram was closer to the theoretical model of describing the condition of sandhi application that is relevant for the description of the metrical parts of the Vedas and that is found across the Prātiśākhyas and their commentarial tradition.
An idea similar to that of rule 1.158 is later on reiterated in the third section of the Vājasaneyiprātiśākhya dealing in detail with the rules of sandhi.16 When the state of combination (saṃhitā) occurs (rule 3.1), the sounds of the two consecutive words combine (rule 3.3):
saṃhitāyām (VP3.1)
In the [krama] state of combination/recitation.
padāntapadādyoḥ sandhi (VP3.3)
Combination [occurs] between the final [sound] of a word and the initial [sound] of a word.
In his commentary to rule 3.3 Uvaṭa states that when saṃhitā occurs, the sounds of the two consecutive words combine:
yaḥ kaścid vaidikaśāstrasandhi ucyate sa padāntapadādyor veditavya iti | te sandhayaś catvāro bhavanti − svarayoḥ, vyañjanayoḥ, svaravyañjanayoś ca | svaravyañjanayos tu dviprakāraḥ − pūrvaḥ svaro bhavati paścād vyañjanāni, vyañjanāni vā pūrvāṇi bhavanti paścāt svara iti | svarayor bhavati yathā − ā idam → “edam” | varuṇa iha → “varuṇeha” | vyañjanyor bhavati yathā − sam yaumi → “sa̐yyaumi” | sam vapāmi → “sa̐vvapāmi” | svarapūrvo bhavati yathā − iṣe tvā → “iṣe ttvā”, ūrje tvā → “ūrjettvā” | vyañjanapūrvo bhavati yathā − ut enam → “ud enam” | paribhāṣāsūtram etat ||
Whatever combination concerning the Vedic texts (śāstra) is mentioned, this should be understood as [the combination] between the final [sound] of a word and the initial [sound] of a word. Those combinations are four: between two vowels, two consonants, [and] a vowel and consonants. As a matter of fact (tu), [the combination] between a vowel and consonants is twofold. There is a preceding vowel followed by consonants. There are preceding consonants followed by a vowel. [The combination] between vowels is as such: ā idam → edam; varuṇa iha → varuṇeha. [The combination] between consonants is as such: sam yaumi → sa̐yyaumi; sam vapāmi → sa̐vvapāmi. [The combination] with a preceding vowel is as such: iṣe tvā → iṣettvā; ūrje tvā → ūrjettvā. [The combination] with a preceding consonant is as such: ut enam → ud enam. This is a rule setting the condition (paribhāṣā-sūtra).
Because of its reliance on the concept of pada (“word”) to define saṃhitā, the Vājasaneyiprātiśākhya also offers a definition of that term in rule 3.2. This rule is however particularly succinct as it simply tells us that words are defined as those elements having artha (“meaning” or “purpose”):
arthaḥ padam (VP3.2)
A word has meaning/purpose.
It is thus necessary to rely on Uvaṭa's understanding of the rule to try to extrapolate its full meaning:
saṃhitālakṣaṇam uktaṃ − “varṇānām ekaprāṇayogaḥ sa̐hitā” (VP 1.158) iti | adhunā padalakṣaṇam ucyate | arthābhidhāyi padam, padyate gamyate jñāyate 'rtho 'neneti padam | yady evaṃ nipātasyānarthakasya padasaṃjñā na prāpnoti | naiṣa doṣaḥ [|] upariṣṭād arthabhedanibandhanaṃ padacatuṣṭayaṃ vakṣyati − “nāmākhyātopasarganipātāś ca” (VP 8.49) iti | tatrāsya padasaṃjñā bhavi-ṣyati yathā − “kriyāvācakam ākhyātam upasargo viśeṣakṛt | sattvābhidhāyakaṃ nāma nipātaḥ pādapūraṇaḥ ||” (VP 8.40) iti | sūtrakārasya tv ayam abhiprāyaḥ − padapratirūpakasya padāvayavasya padasaṃjñā mā bhūd iti | ato 'rthagrahaṇam | ihaiva padasaṃjñā yathā syāt − “govyaccham antakāya goghātam” | iha mā bhūt − “godhūmāś ca me” ||
The characteristic of the state of combination is told as ‘the state of combination is the union [i.e. the articulation] in a single breath of speech-sounds’ (VP 1.158). Now, the characteristic of the word is told. A word expressing meaning (artha-abhidhāyin) [is] a word by which meaning occurs, is received, is understood. If so, the technical term ‘word’ does not include the meaningless particles (nipāta). This [here] is not a mistake: later on (upariṣṭāt), he will mention a fourfold set of words that is the common basis (nibandhana) for the difference concerning meaning/purpose: ‘nouns, verbs, prepositions, and particles’ (VP 8.49). In this respect, the technical term ‘word’ will apply to that [nipāta], in fact ‘a verb is an expression of an action, a preposition is a qualifier, a noun is the expression of an item (sattva), and a particle is a metrical filler’ (VP 8.40). This is in fact (tu) the intention of the author of the rules [of the VP]: the technical term ‘word’ should not apply to the appendix of a word that resembles a word. Thus, the interpretation of the meaning of the [rule] (alternatively: of the [word] artha). Here indeed the technical term ‘word’ should be such [to include go in] ‘cow-tormentor, for Yama (?) cow-killer’ (govyaccham antakāya goghātam). Here, [the technical term ‘word’] will not be [such to include go in] ‘and wheat for me’ (godhūmāś ca me).
The last remark implies that traditionally the word-by-word analysis of the Śuklayajurvedasaṃhitā, which in turn is the source for the krama recitation of the same text, considers go in go-vyaccham to be a word (pada), but excludes go in godhūmāś. We will come back later in section 5 to the difference between this definition of “word” and the definition of “linguistic item” we encountered in the Eḻuttatikāram.
The kinds of combinations
The categorisation of the kinds of phonological combinations is a further sandhi-related aspect that is treated in a similar way by both the Eḻuttatikāram and the Vājasaneyiprātiśākhya and their commentaries. As mentioned before, TEn108 teaches that there are three kinds of change and one unchanging state ([…] mūṉṟē tiripiṭa ṉoṉṟē yiyalpeṉa […]). This passage is paraphrased by Naccinārkkiṉiyar as:
tiriyum iṭam mūṉṟu iyalpu oṉṟu eṉṟu muntainūliṟ kūṟiya annāṉku ilakkaṇamumē
Those four rules that are mentioned in the old treatise(s), namely the occasions of change are three and the [unchanging-]state is one.
Furthermore, TEn109 specifies that the three kinds of change are the modification of consonants, expansion, and elision ([…] meypiṟi tātaṉ mikutal kuṉṟal […]).
This categorisation directly resonates with Uvaṭa's explanation of the term saṃskāra, which appears in Vājasaneyiprātiśākhya 1.1:
svarasaṃskārayoś chandasi niyamaḥ (VP1.1)
[This Prātiśākhya is] a regulation concerning pitch modulations and [grammatical] formations (saṃskāra) in Chandas [= Saṃhitā].
saṃskāro lopāgamavarṇavikāraprakṛtibhāvalakṣaṇaḥ |
[Grammatical] formation (saṃskāra) has elision, augment, modification of speech-sounds, and their unmodified state as its characteristic.
The similarities are self-evident in terms of the fourfold classification of sandhi phenomena and I would venture to hypothesise that the use of the term ilakkaṇam in Naccinārkkiṉiyar's commentary is not by chance, but the result of his familiarity with Uvaṭa's commentary (note that ilakkaṇam is simply the Tamil rendition of Sanskrit lakṣaṇa). A possible element that could help corroborate this hypothesis is Naccinārkkiṉiyar's use of the expression muntai nūlil (“in the old treatise(s)”), which explicitly indicates that our commentator thought that this classification was borrowed from other sources.
There is however a difference in the way the word ilakkaṇam/lakṣaṇa is used by the two authors: in Uvaṭa's commentary lakṣaṇa is used as the last member of an exocentric compound (bahuvrīhi), thus meaning something like “having as its characteristic”. On the other hand, Naccinārkkiṉiyar uses ilakkaṇam as a noun, thus the translation “rule”.
Shared models, but different concerns
As we have seen, the Eḻuttatikāram and the Prātiśākhyas define puṇar-nilai and saṃhitā in a similar way, i.e. as the combination of complex linguistic items and not simply as the combination of speech-sounds, as it seems to be instead the case of Vedic prose (as per VP1.158) or vyākaraṇa. As a consequence these texts need to define what those complex linguistic items are. When they do so, we have the possibility to observe that the two grammatical traditions answer different linguistic and theoretical agendas. On the one hand, the Eḻuttatikāram covers both external and internal sandhis (as the Aṣṭādhyāyī also does) and, therefore, needs to define the linguistic items in a way that covers the domains of both word and morpheme. In this respect, we meet a fourfold definition of col in the rules 158 and 159 of the Collatikāram as peyar (“noun”), viṉai (“verb”), iṭai (“particle”), and uri, with iṭai including both cāriyais and case-endings. On the other hand, the Prātiśākhyas limit themselves to describe external sandhi (arguably, with a few exceptions that are not dealt with here) given that they aim at reconstructing the saṃhitā- and krama-pāṭhas, i.e. two of the possible recitations of the Vedic texts that are in turn built upon their word-by-word recitation (pada-pāṭhas). Accordingly, the Vājasaneyiprātiśākhya offers a definition of pada (“word”) that includes nouns, verbs, prepositions, and particles, i.e. all the elements that are singled out in the word-by-word recitation of the Śuklayajurvedasaṃhitā.
If the concern for internal sandhi is a point of convergence between the Eḻuttatikāram and the Aṣṭādhyāyī, the two differ from one another not only because of the already mentioned word- vs sound-oriented definition of puṇar-nilai/saṃhitā, but also for the scope they attribute to this concept for their representation of phonological phenomena. In fact, puṇar-nilai is the condition for the application of all phonological rules in the Eḻuttatikāram, but the same does not hold true for saṃhitā in the Aṣṭādhyāyī. In the latter, not all sandhi rules fall under the condition of saṃhitāyām (“in the state of combination”).17 Most probably the principle of economy is at the basis of such a difference, since the Tamil phonological phenomena are less variegated than the Sanskrit ones and may be effectively accounted for without postulating a non-saṃhitā-related condition for the occurrence of certain sandhis.
The difficulty in drawing a clear-cut line between these theoretical approaches to the description of sandhi conditions is evidence of the fact that even if Tamil grammarians were aware of the content of Sanskrit grammars, they freely rearranged that material and adapted it to their concerns. Nonetheless, it seems plausible to claim that there was a transfer of ideas from the Sanskrit tradition to the Tamil one also in what we could call the field of phonology. It is indeed always dangerous to claim that a certain author was aware of someone else's ideas unless this is explicitly stated. In particular, in the case of the grammars that are here under consideration, there are centuries of oral transmission and a potential unknown plethora of lost texts that link our extant sources. As for our specific study, for instance, although it is difficult to prove that Naccinārkkiṉiyar knew specifically Uvaṭa's text, it seems reasonable to think that Naccinārkkiṉiyar had a certain degree of familiarity with some Sanskrit sources dealing with the theory of sandhi.
Primary sources
- Aṣṭādhyāyī of Pāṇini = Katre 1987. [Google Scholar]
- Collatikāram (Second book of the Tolkāppiyam of Tolkāppiyar) = Chevillard 1996. [Google Scholar]
- Eḻuttatikāram (First book of the Tolkāppiyam of Tolkāppiyar) = Anonymous 1964 (with the commentary of Iḷampūraṇar) and Tēvanēya 1966 (with the commentary of Naccinārkkiṉiyar). [Google Scholar]
- Kāśikāvṛttī of Jāyaditya and Vāmana = Misra 1985. [Google Scholar]
- Mātṛmodaka of Uvaṭa (commentary of the Vājasaneyiprātiśākhya) = Sharma 1934. [Google Scholar]
- Vājasaneyiprātiśākhya of Kātyāyana = Sharma 1934. [Google Scholar]
- VP = Vājasaneyiprātiśākhya. [Google Scholar]
- Ṛgvedaprātiśākhya of Śaunaka = Shastri 1931. [Google Scholar]
- Taittirīyaprātiśākhya = Rangācārya and Sastri 1906. [Google Scholar]
- TEn = Eḻuttatikāram (with the commentary of Naccinārkkiṉiyar). [Google Scholar]
- Abhyankar, Kāśīnātha Vāsudevaśāstri and Govind Vinayak Devasthali (ed.), 1978. Vedavikṛtilakṣaṇa-samgraha (A Collection of Twelve Tracts on Vedavikṛtis and Allied Topics), Poona, Bhandarkar Oriental Research Institute. [Google Scholar]
- Andersen, Henning (ed.), 1986. Sandhi Phenomena in the Languages of Europe, Berlin, Mouton De Gruyter. [Google Scholar]
- Anonymous (ed.), 19643. Tolkāppiyam, Eḻuttatikāram, Iḷampūraṇam, Tinnevelli, The South India Saiva Siddhanta Works Publishing Society. [Google Scholar]
- Blevins, Juliette, 2009. “Phonetic Explanations for Recurrent Sound Patterns: Diachronic or Synchronic?”, Raimy, Eric and Charles E. Cairns (eds.), Contemporary Views on Architecture and Representations in Phonology, Cambridge (Mass.), MIT Press, 325-336. [Google Scholar]
- Chevillard, Jean-Luc, 1996. Le commentaire de Cēṉāvaraiyar sur le Collatikāram du Tolkāppiyam. Sur la métalangue grammaticale des maîtres commentateurs tamoul médiévaux, Volume I, IFP-EFEO, Pondichéry. [Google Scholar]
- ― 2008. Companion Volume to the Cēṉāvaraiyam on Tamil Morphology and Syntax / Le commentaire de Cēṉāvaraiyar sur le Collatikāram du Tolkāppiyam vol. 2: English Introduction, glossaire analytique, appendices, IFP-EFEO, Pondicherry. [Google Scholar]
- ― 2013. “On the Two Lists of ‘Four [Types of] Words’ (nāṟ-col) in the Śāstric Description of Tamil”, Lingua Posnaniensis LV (2), 9-23. [Google Scholar]
- Ciotti, Giovanni, forthcoming 2018. “On Texts and Disciplines: How the Prātiśākhyas are Categorised by Their Commentaries”, Philological Encounters 3 (“Commentary Cultures”). [Google Scholar]
- Katre, Sumitra M. (tr.), 1987. Aṣṭādhyāyī of Pāṇini, Austin, University of Texas Press. [Google Scholar]
- Matthews, P.H., 20072. The Concise Oxford Dictionary of Linguistics, Oxford, OUP [Also available online.] [Google Scholar]
- Misra, Srīnārāyaṇa (ed.), 1985. Kāśikāvṛttī of Jāyaditya-Vāmana (Along with Commentaries Vivaraṇapañcikā-Nyāsa of Jinendrabuddhi and Padamañjari of Haradatta Miśra), Part I, Varanasi, Ratna Publications. [Google Scholar]
- Müller, Max, 1866. A Sanskrit Grammar for Beginners, London, Longmans, Green, and Co. [Google Scholar]
- Rangācārya, Kastūri and Rudrapatna Shama Sastri (eds.), 1906. Taittirīyaprātiśākhya, Mysore (Reprinted: 1985. Dillī, Motīlāla Banārasīdāsa). [Google Scholar]
- Sharma, V. Venkatarama (ed.), 1934. Vājasaneyi Prātiśākhya of Kātyāyana with the Commentaries of Uvaṭa and Anantabhaṭṭa, Madras, University of Madras. [Google Scholar]
- Shastri, Mangal Deva (ed.), 1931. The Ṛgveda-prātiśākhya with the commentary of Uvaṭa, Vol. 2 (Text in Sūtra-form and Commentary with Critical Apparatus), Allahabad, The Indian Press. [Google Scholar]
- Tēvanēya, Ñāṉamuttaṉ, 19664. Tolkāppiyam Eḻuttatikāram; Naccinārkkiṉiyar urai, Tinnevelly, The South India Saiva Siddhanta Works Publishing Society. [Google Scholar]
- Vergiani, Vincenzo, 2013. “The Adaptation of Bhartṛhari's classification of the grammatical object in Cēṃāvaraiyar's commentary on the Tolkāppiyam”, Cox, Whitney and Vergiani, Vincenzo (ed.), Bilingual Discourse and Cross-Cultural Fertilization, Pondicherry, Institut français de Pondichéry & École française d'Extrême-Orient, 161-197. [Google Scholar]
Secondary sources
- Aṣṭādhyāyī of Pāṇini = Katre 1987. [Google Scholar]
- Collatikāram (Second book of the Tolkāppiyam of Tolkāppiyar) = Chevillard 1996. [Google Scholar]
- Eḻuttatikāram (First book of the Tolkāppiyam of Tolkāppiyar) = Anonymous 1964 (with the commentary of Iḷampūraṇar) and Tēvanēya 1966 (with the commentary of Naccinārkkiṉiyar). [Google Scholar]
- Kāśikāvṛttī of Jāyaditya and Vāmana = Misra 1985. [Google Scholar]
- Mātṛmodaka of Uvaṭa (commentary of the Vājasaneyiprātiśākhya) = Sharma 1934. [Google Scholar]
- Vājasaneyiprātiśākhya of Kātyāyana = Sharma 1934. [Google Scholar]
- VP = Vājasaneyiprātiśākhya. [Google Scholar]
- Ṛgvedaprātiśākhya of Śaunaka = Shastri 1931. [Google Scholar]
- Taittirīyaprātiśākhya = Rangācārya and Sastri 1906. [Google Scholar]
- TEn = Eḻuttatikāram (with the commentary of Naccinārkkiṉiyar). [Google Scholar]
- Abhyankar, Kāśīnātha Vāsudevaśāstri and Govind Vinayak Devasthali (ed.), 1978. Vedavikṛtilakṣaṇa-samgraha (A Collection of Twelve Tracts on Vedavikṛtis and Allied Topics), Poona, Bhandarkar Oriental Research Institute. [Google Scholar]
- Andersen, Henning (ed.), 1986. Sandhi Phenomena in the Languages of Europe, Berlin, Mouton De Gruyter. [Google Scholar]
- Anonymous (ed.), 19643. Tolkāppiyam, Eḻuttatikāram, Iḷampūraṇam, Tinnevelli, The South India Saiva Siddhanta Works Publishing Society. [Google Scholar]
- Blevins, Juliette, 2009. “Phonetic Explanations for Recurrent Sound Patterns: Diachronic or Synchronic?”, Raimy, Eric and Charles E. Cairns (eds.), Contemporary Views on Architecture and Representations in Phonology, Cambridge (Mass.), MIT Press, 325-336. [Google Scholar]
- Chevillard, Jean-Luc, 1996. Le commentaire de Cēṉāvaraiyar sur le Collatikāram du Tolkāppiyam. Sur la métalangue grammaticale des maîtres commentateurs tamoul médiévaux, Volume I, IFP-EFEO, Pondichéry. [Google Scholar]
- ― 2008. Companion Volume to the Cēṉāvaraiyam on Tamil Morphology and Syntax / Le commentaire de Cēṉāvaraiyar sur le Collatikāram du Tolkāppiyam vol. 2: English Introduction, glossaire analytique, appendices, IFP-EFEO, Pondicherry. [Google Scholar]
- ― 2013. “On the Two Lists of ‘Four [Types of] Words’ (nāṟ-col) in the Śāstric Description of Tamil”, Lingua Posnaniensis LV (2), 9-23. [Google Scholar]
- Ciotti, Giovanni, forthcoming 2018. “On Texts and Disciplines: How the Prātiśākhyas are Categorised by Their Commentaries”, Philological Encounters 3 (“Commentary Cultures”). [Google Scholar]
- Katre, Sumitra M. (tr.), 1987. Aṣṭādhyāyī of Pāṇini, Austin, University of Texas Press. [Google Scholar]
- Matthews, P.H., 20072. The Concise Oxford Dictionary of Linguistics, Oxford, OUP [Also available online.] [Google Scholar]
- Misra, Srīnārāyaṇa (ed.), 1985. Kāśikāvṛttī of Jāyaditya-Vāmana (Along with Commentaries Vivaraṇapañcikā-Nyāsa of Jinendrabuddhi and Padamañjari of Haradatta Miśra), Part I, Varanasi, Ratna Publications. [Google Scholar]
- Müller, Max, 1866. A Sanskrit Grammar for Beginners, London, Longmans, Green, and Co. [Google Scholar]
- Rangācārya, Kastūri and Rudrapatna Shama Sastri (eds.), 1906. Taittirīyaprātiśākhya, Mysore (Reprinted: 1985. Dillī, Motīlāla Banārasīdāsa). [Google Scholar]
- Sharma, V. Venkatarama (ed.), 1934. Vājasaneyi Prātiśākhya of Kātyāyana with the Commentaries of Uvaṭa and Anantabhaṭṭa, Madras, University of Madras. [Google Scholar]
- Shastri, Mangal Deva (ed.), 1931. The Ṛgveda-prātiśākhya with the commentary of Uvaṭa, Vol. 2 (Text in Sūtra-form and Commentary with Critical Apparatus), Allahabad, The Indian Press. [Google Scholar]
- Tēvanēya, Ñāṉamuttaṉ, 19664. Tolkāppiyam Eḻuttatikāram; Naccinārkkiṉiyar urai, Tinnevelly, The South India Saiva Siddhanta Works Publishing Society. [Google Scholar]
- Vergiani, Vincenzo, 2013. “The Adaptation of Bhartṛhari's classification of the grammatical object in Cēṃāvaraiyar's commentary on the Tolkāppiyam”, Cox, Whitney and Vergiani, Vincenzo (ed.), Bilingual Discourse and Cross-Cultural Fertilization, Pondicherry, Institut français de Pondichéry & École française d'Extrême-Orient, 161-197. [Google Scholar]
The term “grammar” deserves clarification. On the one hand, it is usually employed to translate the Sanskrit word vyākaraṇa and, on the other hand, the Tamil word ilakkaṇam. These are however the names of two very different disciplines, which overlap only in part. Vyākaraṇa studies the formation of correct Sanskrit words (once established the semantic and syntactic context of the sentence in which they will appear). When talking about Sanskrit grammar in this article, it will be therefore necessary to extend the scope of the term grammar to include another discipline called śikṣā, which is the study of the articulation of speech-sounds and their modifications in particular contexts (many tenets of śikṣā are in fact given for granted in vyākaraṇa texts). In the Tamil case, ilakkaṇam is a very broad field that not only includes the study of linguistic matters, but also of literary ones, in particular aesthetic conventions (the study of which falls, as far as Sanskrit scholarship is concerned, under a further category called alaṅkāraśāstra, “study of embellishments”). The section of ilakkaṇam that I cover in this article would correspond to the field of eḻuttiyal (“study of sounds/letters”), which is however an anachronistic label if put in relation to the terminology employed in the texts I will take into analysis. At times, I will also include elements from the field of colliyal (“study of words”).
For more case studies on related topics, see D'Avella in this issue, and Vergiani 2013.
I offer here what is a recent and, in my view, effective definition of phonology provided by Blevins (2009, p. 325–bold GC): “Phonology is the study of sound patterns of the world's languages. In all spoken languages, we find sound patterns characterizing the composition of words and phrases. These patterns include overall properties of contrastive sound inventories (e.g. vowel inventories, consonant inventories, tone inventories), as well as patterns determining the distribution of sounds or features of sounds (stress, tone, length, voicing, place of articulation, etc.) and their variable realization in different contexts (alternations).” The ideas from the Tamil and Sanskrit texts that I will analyse in this article fall under the scope of the last part of this definition. Furthermore, in this article I use the term sandhi in the broad and loose sense that is still commonly accepted in modern Western linguistics. See for instance Matthews' entry for sandhi in his 2007 Dictionary of Linguistics: “Ancient Indian term for the modification and fusion of sounds at or across the boundaries of grammatical units. E.g. short -a + i- fused in Sanskrit, both within vowels and across word boundaries, to -e-.” For some of the issues underling the definition of sandhi, see Andersen 1986, p. 1-2. See also fn. 7 for the distinction between internal and external sandhi.
In a nutshell, Prātiśākhyas are treatises describing the linguistic features of Vedic recitation. For more details on how these texts are described within the traditional Sanskritic scholarship, see for instance Ciotti forthcoming 2018.
Hereafter, I refer to the rules of the Eḻuttatikāram according to the numeration found in Naccinārkkiṉiyar's commentary, thus using the siglum TEn (for Tolkāppiyam Eḻuttatikāram Naccinārkkiṉiyam). Despite the fact that Naccinārkkiṉiyar and Iḷampūraṇar read the same version of the Eḻuttatikāram, there is a discrepancy in the way in which they group its rules.
For a useful lexicon that accounts for the different nuances these terms may assume in the Tolkāppiyam (in particular in its second book, the Collatikāram or “Book on Words”, with the commentary of Cēṉāvaraiyar), see Chevillard 2008.
The expressions internal sandhi and external sandhi are established by Müller (1866, p. 9): “For shortness' sake it will be best to apply the name of External Sandhi to the changes which take place at the meeting of final and initial letters of words, and that of Internal Sandhi to the changes produced by the meeting of radical and formative elements.” See also fn. 3.
As Naccinārkkiṉiyar further points out in his commentary not all linguistic items bear a case relationship with other linguistic items. Nonetheless, this does not prevent phonological rules to apply. For instance, Naccinārkkiṉiyar offers two examples concerning cāriyais, i.e. maka-v-iṉ kai (“hand of a child”), where the cāriyai -iṉ occurs between two items that share a case relationship, and paṉai-y-iṉ kuṟai (“reduction by a paṉai [measure] ”), where the same cāriyai occurs where the case relationship is absent (I assume, very much tentatively, that Naccinārkkiṉiyar understands paṉai as the agent of the action of reduction expressed by the noun kuṟai. In this way the two items would not share a case relationship in the kāraka/veṟṟumai system shared by the Sanskrit and Tamil traditions; see D'Avella in this issue).
Naccinārkkiṉiyar for instance defines cāriyai twice. While commenting on TEn112, he says eḻuttiṉāṟ cāriyai yātalāṉum (“a cāriyai is made of sounds”). Later on, while commenting on TEn118, he says cāriyai yeṉṟataṉ poruḷ, vēṟāki niṉṟa irumoḻiyun tammiṟ cārtaṟ poruṭṭu iyaintu niṉṟatu eṉṟavāṟu (“[it] stands agreeably in favour of two separate words that join one another”; note that here I translate moḻi as “word” on the basis of the examples discussed by Naccinārkkiṉiyar before this very passage, for instance āṭūu-v-iṉ kai “limb of a goat”).
I leave the term uri-c-col untranslated following the thoughtful consideration of Chevillard 2013, p. 13, fn. 19. Tentatively, its scope can be approximated to that of the categories of adjective or adverb.
Here the term krama indicates the krama-pāṭha (“step-recitation”), which is a particular form of recitation of the Vedic texts in words' pairs (e.g. agním puróhìtam | puróhìtaṃ yajñásyà | puróhìtam ítì puráḥ hìtam | yajñásyà devám | devám ṛtvíjàm | ṛtvíjam íty ṛtvíjàm). For more details see for instance Abhyankar and Devasthali 1978.
Uvaṭa remarks in the introduction to this section: adhunā kramaprāptaḥ saṃskāro 'bhidhīyate lopāgamavarṇavikāraprakṛtibhāvalakṣaṇaḥ (“now the [grammatical] formation (saṃskāra) that concerns the krama combination/recitation is taught, which has elision, augment, sound change, and natural state as its characteristic”). For more details on the content of this passage, see 4.2.
Differentiating between rules that happen under the condition of saṃhitāyām and rules that do not allows Pāṇini to ingenuously account for the different outcome of sounds' modification in similar phonological contexts. For instance, ī → iy in front of a dissimilar vowel in internal sandhi (see rule 6.4.77 aci śnudhātubhruvāṃ yvor iyaṅuvaṅau of the Aṣṭādhyāyī), but ī → y in external sandhi (see rule 6.1.77 iko yaṇ aci, which falls under the scope of rule 6.1.72 saṃhitāyām).