17. Further properties of sarvanāma (pronominals) – in compounds

We saw in sutra 1.1.27 that Panini defines something akin to pronominals by a list of words starting with sarva, ‘all’ in his Gaņapāţha (list of groups). One of the implications of being a sarvanāma is that the case endings are special: nominative plural masculine sarve (rather than sarvāh), certain oblique cases with special endings like –smai in the dative singular (rather than -āya), -smin in locative singular rather than –e (I don’t quite understand why –smat ablative case is not also cited in the special endings). But in compound words, this doesn’t happen; and Panini devotes the next nine sutras to laying out these exceptions.

1.1. 29 na bahuvrīhau

The words are

Na (0, particle),    bahuvrīhau (7/1, locative singular)

And the meaning is

Na (not) in bahuvrīhi compounds. The anuvŗtti (carried forward) would be:

sarvādīni sarvanāmāni (from 23), and the paraphrase is:

vŗtti: bahuvrīhau samāse (in bahuvrīhi compounds)  sarvādīni  (words from sarva) sarvanāma samjñāni (sarvanāma ‘species’) na bhavanti (are not).

A bahuvrīhi compound is one where two words together describe a third entity. Suppose the word were redhead; if it were a bahuvrīhi, it would mean, not a red head, but a redheaded person or bird. In Sanskrit, they describe this by adding the possessive phrase, “who has, that person”, red head yasya sah. Such compounds are not declined (given case endings) like a sarvanāma even if they end in sarva etc.; they decline like a normal noun. Examples provided are:

Priyavisva, priya ‘beloved’, visva ‘(of) all’, which describes a third entity, hence is a bahuvrīhi compound; but it will be declined "normally”, priyavisvāh (not priyavisve), priyavisvāya (not priyavisvasmai), etc.

Priyobhaya, priya ‘beloved’, ubhaya ‘(of) both’, declined as a normal term, priyobhayāya, etc.

There is however a slight modification of this rule, which is actually in the preceding sutra,

1.1.28 vibhāșā (1/1) diksamāse (7/1) bahuvrīhau (7/1)

Anuvrtti: sarvādīni sarvanāmāni (from #27)

Here vibhāșā means ‘option’, and a diksamāsa is a samāsa ‘compound’ with the word of dik (meaning diś ‘direction’), i.e. a compound with a word of direction as the first element. Examples are

Uttara-pūrva ‘north-east’, which can form oblique case forms either way,

Uttara-pūrvasmai or uttara-pūrvāya, etc. (the  hyphen is just for clarity, not used in the original).

There are further exceptions to the sarvanāma definition:

1.1.30 tŗtīyā-samāse (7/1)

to which we add, by anuvŗtti,

na (from 29),  sarvādīni sarvanāmāni (from #27), giving the sense:

‘compound words of tŗtīyā type with sarva etc. (from the list) are not sarvanāma words’. These tŗtīyā compounds are those with the first word in the third case (tŗtīyā), or the instrumental ending (‘by’ or ‘with’). These are a type of tatpurusha compounds, where the first word qualifies the second (redhead to describe a head, not a pretty girl, redcap to describe a cap, not a bird), and the second component is of the particular type referred here (and specified in rule 2.1.31, but that need not throw us here on a first reading. I think it suffices to know that compounds in –sarva etc. which have a third case ending for the first component will not be a sarvanama.

Examples are given with the second component –purva ‘prior’, one of the sarvādini list, thus:

māsapūrva = māsena pūrva, ‘(by) a month earlier’, which is declined normally,

māsapūrvāya ‘for one who was born a month earlier’ etc., not māsapūrvasmai etc.

Another exception:

1.1.31 dvandve (7/1) ca (0)

‘And in dvandva’, with carry-forward by anuvŗtti as before:

 na (from 29),  sarvādīni sarvanāmāni (from #27). This extends the exclusion to compounds that are co-ordinate, dual, dvandva:

pūrvāparāņām and not pūrvāparesām (possessive plural 6/3 of pūrvāh-parāh ‘the priors and the posteriors’). However, under dvandva, there is an exception:

1.1.32 vibhāșā (1/1) jasi (7/1), and by anuvŗtti we add the words

sarvādīni sarvanāmāni (from #27) dvandve (from 31) na (from 29), meaning:

vŗtti : ‘Optionally, in the nominative plural ending (jas), dvandva compounds in sarva etc. may not be called sarvanāma’. This rather clumsy sentence is actually meant to make the exclusion optional: dvandva compounds in nominative plural case may or may not be excluded from the sarvanama category, i.e.  we should be allowed to make the plural of the example given above either way, pūrvāparāh or pūrvāpare. Note also the species jas, which is denoted Jas with an upper case initial J to indicate that it is an iT, a marker. The ending is –as, of the plural (which transforms to -āh̨ according to other rules which we have not encountered).

The vibhāșā  jasi (option in nominative plural 1/3 forms) cases are extended in sutras 1.1.33, 34, 35, and 36 to other members of the sarva- list (not limited to dvandva compounds). Thus many of these plural forms can be used in either form: prathame/prathamāh 'first' (under rule 33), pūrve/pūrvāh 'front, east, prior' (rule 34), sve/svah 'self, own' (rule 35, but why not svāh?), antare/antarāh ‘outside, anterior’ (rule 36). In each of these rules, Panini specifies that the option is available when the words are used in a certain sense, and not in some other sense, so he puts each subset in a separate rule (otherwise I can see no reason why he could not have bunched the lot in a single list, given his frugality of expression).  In 1.1.35, for instance,

1.1.35 svamagñātidhanākhyāyām,

Which parses into svam (1/1) agñātidhanākhyāyām (7/1), to which is added by carry-forward,

Anuvrtti: vibhāșā  jasi (from #32), sarvanāmāni (from #27).

The meaning (vrtti) is as follows: word forms in svam ‘own’ are sarvanāmāni ‘pronominals’ vibhāșā ‘optionally’ jasi ‘in operations relative to Jas, the nominative plural ending’, (provided that they are) ākhyāyām ‘in the sense of’ a-gñāti ‘not a relative’ or (a-)dhana ‘(not) wealth, property’. What this portends is that sva- words are usually a pronominal (hence take all those special endings like –e, -smai, -smin, -eshām), but in forms where the Jas suffix (nominative plural) comes into operation, they can be optionally pronominal or ‘normal’ (endings like –āh, --āya, -e, -ānām). In #31, apparently, the reference to a compound is not carried forward, so the sva- word can be just by itself:

sve OR svah putrāh  ‘one’s own sons’ (but why not svāh?)

sve OR svah gāvah ‘one’s own cows’ (surely there is no implication that the two are equivalent!).

The above option will be available only when sva is used in the sense of ātman ‘self’ or ātmiya ‘one’s own’, and not if it is used in the other two senses of gñāti ‘a relative’ (cognate with agnate?) or dhana ‘wealth, property’; in these latter contexts sva will always be a sarvanāman, and declined accordingly. (Does this explain the phrase used in Rgveda 1.1.8, sve dame ‘in [Agni’s] own home’, where sva- is used in the sense of ‘one’s own’, and so is declined like a normal noun, and not a sarvanāman which would give svasmin? Further, is the option restricted to nom. pl. Jas, or extended to other cases like the locative singular as in sve dame?)

The above span of sutras (from 1.1.26 to 36) dealt with sarvanama, pronominals, and we can make two observations. One is that somewhere in between (from #33), we had to drop the anuvrtti of dvandve na ‘not in dvandva compounds’, and only carry forward sarvanāmāni 'pronominals’. There is no clear indication of where this sort of change occurs, but it may be implied by choice of a particular case ending, or gender, or number.

The other observation is a question of where these rules come from. It would be unreasonable to assume that Panini is making up all these variations (like a Mozart or a Beethoven) just to please himself; much more likely would be that he is reflecting the prevalent usage. This suggests the extraordinary care with which he, and perhaps his associates and predecessors, have listened to the speech of the people, cultivated though they may be. This is apparently not an artificially crafted perfectly consistent rule-bound language, a vase of silk flowers as it were, but a living idiom, a living plant lovingly nurtured in an earthen pot of natural ingredients with all its quirks and inconsistencies. We could, perhaps, devise some involved explanation of why a particular sarva- word ceases to have a pronominal value in a particular context or environment (such as in a compound), but it appears rather that Panini has reflected the natural shape and contours of the native idiom rather than try to craft an artificial set of rules, and to that extent we may have to be content with a description rather than a satisfactory explanation of the underlying rationale for all these variations.