6. Sounds of a Colour Flock Together in Pāņini

In the previous article (No.5) we saw Pāņini’s arrangement of the sounds of the language in the 14 Śivasūtra, and also saw some of the 41 or more sub-sets by means of abbreviations termed the Pratyāhāra. To repeat, all the vowels are referred to by the code aC, which indicates the span from the first letter a of Śivasūtra 1 to the end-marker C of Śs 4; all the consonants are covered in the span h^a L, from the initial h^a of Śs 5 to end-marker L of Śs 14.  Everything together, from the initial a of Śs 1 to end-marker L of Śs 14 is aL, the set of all the vowels and consonants.

We noticed that the long vowels ā ī ū ŗ̄ ļ̄ do not occur explicitly in this layout. As anticipated, there is a rule which accounts indirectly for this apparent omission. This is contained in a small 2-sutra jewel in the first pāda, and it also illustrates nicely the use of the code terms (the pratyāhāra) aC and h^a L:

1.1.9          tulyāsyaprayatnam savarņam

5. Back to Basics: Shiva-sutras and the Lists of Vowels and Consonants

In an earlier piece, we came across the basic building blocks of Paņini’s technical jargon, which was to arrange the sounds (letters) in a particular sequence, called the Shiva-sutras, denoting the end of each sub-set by a nonsense letter called an it (what else!), which we show in upper-case in our Roman transliteration. We may select and name sub-sets which may start at any letter in a given Shiva-sutra, and may even go over a Shiva-sutra boundary it into a subsequent Shiva-sutra, finally ending in a given boundary-marker (it). The name is given by combining the initial letter and the final boundary it. Intermediate its do not figure, as they do not have any significance in themselves. We saw two such sub-sets named right at the start of the work, 

1.1.1      vŗddhir ādaic

The set of ‘super-augmented’ vowels consisting of the long sound ā, termed āT with a nonsense ending, and the span of Shiva-sutra 4, ai au (ending in C, the boundary-marking it, hence āT plus aiC equals ādaiC).

The second sutra:

1.1.2          aden̊ guņah

defines the set of ‘singly-augmented’ vowels, consisting of short a (termed aT with a nonsense ending), plus the span of Shiva-sutra 3, e o and the marker N̊ (sounded like a velar nasal ng); aT plus eN̊ equals adeN̊ by the rules of sandhi.

Apparently the entire terminology is built up on these Shiva-sutras and the sandhi rules which govern word-joins. Why Paņini arranged the letters in precisely this sequence will be known only as we delve into his treatment of the rules; what is obvious is that the arrangement is different from the alphabet we learn in school. Further, certain sounds are left out, such as the long vowels ā, ī̄, ū, ŗ̄; the nasals N̊ (velar), Ñ (palatal), Ņ (retroflex alveolar), N (dental-alveolar) and M (bilabial), are used as end-markers or its (hence shown in upper-case here), and there is some repetition (Ņ occurs in two Shiva-sutras, 1 and 6, as end-marker, which may give cause for some confusion unless reference is made to the context; k occurs as a sound k^a in Shiva-sutra 12, as well as an end-marker K in Shiva-sutra 2; etc.).

These are then the 14 Shiva-sutras (Śs):

1.       a i u Ņ (retroflex Ņ!)…obviously the short vowel sounds, a and i and u, Ņ an it symbol to show the end of that sequence;

2.       ŗ ļ K … the liquid half-consonants, ŗ and ļ, K ending the sequence;

3.       e o N̊…two medium-long vowels e and o (diphthongs in English?; there are no short e and o in Sanskrit, though there are in Tamil and other Dravidian languages);

4.       ai au C (C palatal, as in chair)… some  real diphthongs, ai and au, symbol C ending the set (an it);

5.       h^a y^a v^a r^a Ţ (strarting the non-vowels, Ţ the end-marker; the superscript ^a to denote the small a-sound with each consonant, whereas the end-markers do not have it);

6.       l^a Ņ;

7.       ñ^a m^a n̊^a ņ^a n^a M (the nasals);

8.       jh^a bh^a Ñ (some voiced consonants, aspirated);

9.       gh^a ḑh^a dh^a Ş (voiced aspirated, end-marker retroflex spirant Ş);

10.   j^a b^a g^a ḑ^a d^a Ś (voiced unaspirated, end-marker palatal Ś);

11.   kh^a ph^a ch^a ţh^a th^a c^a ţ^a t^a V (unvoiced, some aspirated, some not);

12.   k^a p^a Y (unvoiced, unaspirated);

13.   ś^a  ş^a s^a R;

14.   h^a L.

All the common vowels are referred to by the code aC, which indicates the span from the first letter a of Śivasutra 1 to the end-marker C of Śs 4; the long vowels do not occur explicitly (but there is a Sutra which accounts for this apparent omission). All the king’s consonants are covered in the span h^a L, from the initial h^a of Śs 5 to end-marker L of Śs 14 (but introducing some ambiguity, because Śs 14 itself starts with h^a!).  Everything together, from the initial a of Śs 1 to end-marker L of Śs 14 is aL (literally, all!). There are said to be 41 such subset codes (technical term pratyāhāra) in Paņini’s system, plus a few more here and there. Frequent usage over time will no doubt make these terms old friends to the student. They sure do add to the exotic nature of the meta-language used in the work!

Incidentally, the individual sounds are termed varņa; we already saw that the end-markers are termed iT. 

4. Guņa-Vŗddhi Replacements: Rules within Rules in Pāņini

The world of Panini is a maze of rules, metarules, exceptions and extensions all couched in a secret language that itself follows the rules it is expounding, which yields nothing to the unaided aspirant. This leads to fiendishly involved derivations with plenty of traps and pitfalls for the unwary wanderer. Take the seemingly simple substitution,

1.1.3          iko guņavŗddhī

which we assumed means something like

1.1.3 ‘Of short vowels, (happens) guņa and vŗddhī, or, simply put, replace short vowels (we include the semi-consonants ŗ and ļ along with a, i, and u in this category), with the corresponding guņa (augmented) and vŗddhi (super-augmented) forms. Of course, there must be certain circumstances in which this is done, or else the short vowel forms would eliminate themselves by repeated application of this rule.  It appears that this substitution applies to other rules, scattered through the chapters, which call for replacement by guņa and/or vŗddhi, but without specifying what is to be replaced. Obviously whole words will not be replaced; so when there is no specific target stated up front, we are to assume that it is the short vowels (the iK vowels a, i, u, ŗ and ļ in the Shiva-sutras) )which are to be replaced.

We assume that the specific situation calling for such replacement will be elaborated in subsequent rules. So the genitive case here signifies replacement, not possession. The second term sthāneyogā may mean something like ‘in its place suitable’; I confess I have not figured it out, but I understand it signifies that provided no other interpretation of the genitive case is possible in the given context, it signifies replacement by the terms that follow.

Sutra 1.1.3 is followed by a few more sutras that qualify or explain it further; this introduces the device of extending a rule statement over a number of sutras, a ‘read-with’ device to develop a complex situation. Thus,

1.1.4          na dhātulope ārdhadhātuke

which has to be read in continuation to 1.1.3, iko guņavŗddhī. There is some disagreement on whether we should read na dhātulope or na dhātulopa; I will gloss over this issue for the present and go with the former. 

An ārdhadhātuka is something like an affix to a verb root (dhātu) which is not a tense-person form of verbal ending. A  dhātulopa is a condition resulting in truncation of a dhātu, verb root (here a part of a root). The sutra means (very broadly and coarsely!) something like this: replacement of short iK vowels by augmented guņa or vŗddhi, as ordered in sutra 1.1.3, will not (na) be done if there is an affix that causes truncation (deletion) of the root.  

To make life further difficult, there is more:

1.1.5          kṅiti ca

(ṅ is the velar ng-sound in the series k kh g gh ṅ). This is to read with ‘na’ from 1.1.4, so it says something like: (replacement of short vowels by augmented vowels) will NOT take place also if there is a k g ṅ marker in the affix (which apparently is a marking device, not necessarily indicating the actual presence of the sound k or g or ṅ; further, they’re not referred to as kgṅ but as kṅ apparently because the rules require collapse of such adjacent velars – an artifact of the meta-language having to follow the very rules it is expounding!).

And finally, this span of rules is completed by

1.1.6          dīdhīvevīţām , separated out as

dīdhī vevī iţ ām (possessive case ending, sixth vibhakti)

again to be ‘read with’ the negative na of 1.1.4, which thereby further adds to the exceptions to replacement of short (iK) vowels ordered by 1.1.3. This means there will be NO replacement of iK vowels even if prescribed by some rule, in forms of the verb dīdhī, ‘throw light on’, vevī, ‘go forth’, and something called iŢ which is a verb affix (presumably i  followed by a nonsense sound Ţ to denote that it is a grammatical category, not a lexical (meaningful) word; ‘it’ is a general technical term for such nonsense markers, as at the ends of each sequence of sounds in the Shiv-sutras). These negations or exceptions will obviously come into play in the operation of various rules where the target is the short iK vowels of specific word types or categories, which we will hopefully come across and grapple with if we persist on this journey! The main thing here is that we have seen how a rule may span a number of sutras with its explanations and exceptions.

3. Rules about Rules: Metarules in Pāņini

One of the joys of reading Pāņini (with one retroflex ņ and one normal n!) is how we get to see a brilliant mind at work, busy spinning his webs of logical sequences – from all of 2500 years in the past. Apparently, he is trying to develop decision trees that will account for every known form and variation in his language, and a means to generate new forms that fit into a logical framework. But he also wants to do this in as terse and economical a style as possible.  He makes up technical terms as he goes along, and he has rules about his rules – about what these terms signify in a technical sense. Most of them do not even mean anything in the normal language; others are used in a specific sense depending on the context. There are therefore sutras of different types: there are technical terms and their definitions; there are rules; and there are rules about rules – metarules – which allow him to use his technical terms with the least possible extra explanation. In the terminology of the communications expert, he tells you what he’s going to tell you, then he tells you, then he tells you what he told you! If you still don’t get it, it’s not because he didn’t try.

Here are two such metarules which I managed to understand (to a degree!) in the first quarter (pāda) of the first book (adhyāya).

1.1.49      șașțhī sthāneyogā

This is a metarule which helps to interpret a word in the șașțhī or sixth form (vibhakti) of the noun declension, which is the genitive or possessive ‘case’. We saw this in sutra 1.1.3,

1.1.3          iko guņavŗddhī

where the first word is the genitive case (șașțhī) of the term iK, which represents the short vowels I, u, ŗ, ļ, and the rule meant something like

1.1.3 ‘Of short vowels, (happens) guņa and vŗddhī.’

In other words, you replace these short vowels (of which two may be called semi-consonants) with their augmented and super-augmented counterparts (guņa and vŗddhī) in circumstances which are elaborated in subsequent rules. So the genitive case here signifies replacement, not possession. The second term sthāneyogā may mean something like ‘in its place suitable’; I confess I have not figured it out, but I understand it signifies that provided no other interpretation of the genitive case is possible in the given context, it signifies replacement by the terms that follow.

More guidance is given by

1.1.50      sthāne antaratamaḩ

which must mean something like ‘in the place, the most similar’. That is, the replacement term is chosen that is most similar to the replaced entity, in terms of either place of articulation, or in terms of meaning, sound quality, or duration. The word sthāne may refer to place of articulation, which has precedence over other aspects.

Here’s another metarule to interpret words in the locative case:

1.1.66      tasminniti nirdişţe pūrvasya

The word tasmin, in the locative or seventh (saptamī) case, means ‘in it’; tasmin iti, ‘in it, like’, or a word in locative case. Again, nirdişţe is in locative case, meaning ‘in the direction’, or to stretch it a bit, tasminniti nirdişţe , ‘in (the context of) a direction (starting with) a word in the locative case’; pūrvasya, in the genitive case, ‘of what precedes’. That is, a direction starting with a term in locative case, tasmin, would refer to an operation on the word preceding the word in locative. Such a rule refers to the entity on the left, a left-referencing rule. Given, or in the event of, the situation referred to by the locative, is implied the situation in the preceding entity (nirdişţe itself is one such word in the locative!).

A similar metarule with ablative case, tasmād (‘from it’):

1.1.67    tasmād ity uttarasya

Or, breaking the word-join,

                tasmāt iti uttarasya

and reading in continuation of the preceding rule, taking as read the linking word nirdişţe, this means: given a rule (nirdişţe, locative case) starting with an ablative (tasmāt iti) it belongs to (refers to) the following entity (uttarasya). Such a rule refers to the entity on the right, a right-referencing rule. From the situation referred to by the ablative, follows the situation in the succeeding entity.

 Do we glimpse the shadow of self-perpetuating loops here?

2. The first three sutras! Conditioning of short vowel and half-consonant sounds

After beating around the bush, let’s actually get into the work now and see how it goes. The first two sutras are as good a place to start as any. The first aphorism or formula, numbered 1.1.1 to denote the first formula (sūtra), of the first quarter (pāda), of the first Book (Adhyāya), goes like this:

1.1.1          vrddhir adaich or, more phonetically

vŗddhir ādaic

The second goes like this:

1.1.2          adeng gunah, or phonetically,

aden̊ guņaḩ

The first thing, of course, is that it lays down the rules of Sanskrit as she ought to be spoken, with the rules of Sanskrit as she is spoken. It reminds me of the good old German course in college, Deutsche Sprachlehre fuer Auslaender, Grundstufe Eins, which starts off the adventure with ‘Ich bin Anton Brega’, a sentence we all can understand without knowing German, ever since a famous American stood up and intoned ‘Ich bin ein Berliner’.  But one thing here (of many things!): you can’t read Panini (with the retroflex n!) without the rules of joining sounds, sandhi.  So with our 10^th class memory, we realize that vŗddhir stands for vŗddhiḩ, nominative case singular number (paradigm harih), gunah (with the retroflex n!) is likewise a nominative singular name of something (paradigm ra:mah ra:mau ra:ma:h).  

With our 10^th class Sanskrit, we can recognize the first word vŗ̧ddhih, which is ‘growth’; a vŗddhaḩ is an old-timer.  We also know guņaḩ; a guņa is a quality, like a sadguņa is a good quality, durguņa is a negative quality. Maybe some of us also vaguely remember that guņa and vŗddhi are grammataical terms, representing successive ‘augmentations’ of vowel sounds, somewhat akin to the successive vowel-shift in sequences like sing-sang-sung, ring-rang-rung, called ‘Ablaut’ in good old Anton Brega’s language. So guņa is goody-good, augmented, while vriddhi is ultra-good, doubly augmented. Panini is defining the vowel sequences with these two technical terms. But what are the definitions?

We need the sandhi rules again to make sense of the definitions. The second word, ādaic, actually stands for āt aic (voiceless t becomes voiced d, as in sadguņa, sat+guņa above!). The sequence a:t apparently is a symbol for the sound a:, the first such nonsense name Panini has made up. I guess it could have as well been a simple ā or maybe āchhoo or ārrrgh. Panini uses a simple t to symbolize the sound, so we may write it as āT, the upper-case T representing a symbolic symbol as against a sounded sound. So the first part says ‘the sound ā is a doubly augmented sound’.

What to make of the second part, aic? The ai stands for the sound ai; the ending c is a symbolic one, so the sequence ai  to be written as aiC, the upper-case C showing that it is a code, a symbol, rather than a sounded sound. Here is now a key to understanding Panini’s code: he uses these nonsense words to stand for sub-sets of sounds. He actually gives us a list of sounds (letters) right at the beginning, arranged in a certain sequence with these code symbols like C  interspersed, showing the ends of successive sequences or runs; these are called the Shiva-sutras, formulae given by the great god Shiva himself, or if you like formulae which are eternal, pervading.

These are the first few Shiva-sutras (Śs):

1.       a i u Ņ (retroflex N!)…obviously the short vowel sounds, a and i and u, Ņ a symbol to show the end of that run;

2.       ŗ ļ K … the liquid half-consonants, ŗ and ļ, K ending the sequence;

3.       e o N̊…two medium-long vowels e and o (diphthongs in English?; there are no short e and o in Sanskrit, though there are in Tamil and other Dravidian languages), end marker N̊ (ng sound);

4.       ai au C (ouch!)… some  real diphthongs, ai and au, symbol C ending the set (but not part of it)

So we see that aiC is actually a set that consists of two diphthongs, ai and au, rounded off by the marker C for end-of-list; eN̊ stands for the set of sounds e and o, end-of-sequence set off by N̊; now we see why Panini fits into the computer world so well.  So when he says,

1.1.1 vŗddhir ādaic

or

vŗddhiḩ a:T aiC,

he is actually defining the set of doubly-augmented or amplified sounds, vŗddhi, as ‘a: and ai and au’. Wow!

And similarly, with

1.1.2          aden̊ guņaḩ

or dismantling the sandhi word-joins,

aT eN̊ guņaḩ,

he is saying that the set of singly-augmented or goody-good sounds is ‘a and e and o’. Oh!

You want more?

1.1.3          iko guņavŗddhī

The first ‘word’ is the symbolic iK, representing the sub-set of ‘short’ vowels and half-vowels, i u ŗ ļ, and the end-marker K. It’s apparent here that if you want to make a set spanning more than one Shiv-sutra, the first and the second in this case, you leave out the intermediate end-of-sequence markers (called, incidentally, an iT!). Then this symbol, iK, is put into genitive (possessive) case, singular, really ikaḩ, but by the rules of word-joins, sandhi, put into the form iko meaning ‘of the set i u ŗ ļ’.  

The second term, guņavŗddhī, is a compound word, in dual number. We have to supply a verb in English (but not necessarily in Russian!):  ‘Of short vowels, (happens) guņa and vŗddhi’. This is interpreted as meaning that in place of i u ŗ and ļ, when a substitution is to be done, we will replace them with the appropriate guņa or vŗddhi forms depending on the circumstances. What those circumstances are, is not given rightaway; in one of the master strokes of this opus, a given rule may condition various rules and operations in other parts of the text.

I hope I haven’t made any gross blunders in my interpretation. Happy hunting! 

1. Reading Panini: into the world of Sanskrit grammarians

For readers with some knowledge of Sanskrit, and students of Linguistics in general

Most educated Indians would have had some amount of Sanskrit during their school years.  In my case, for instance, we were more or less forced to take up Sanskrit as the Third Language when we shifted to Kerala, as we could not really take Malayalam. I do remember that Sanskrit was a ‘scoring’ subject, with marks above 90% the norm. Later, in the Central School, too, there was enough of Sanskrit to get more than a passing interest in the language.

Later in life, this interest lies dormant, working in the background, but flaring up occasionally when one sees some volume in a bookstore, or nowadays comes across an article in the internet. Over the years, I have collected a tidy library of books on and in Sanskrit, including the hefty volume of the Rgveda published by the Kapila institute in Bangalore. I have tried reading portions of it, also of the Bhagavadgita and the Upanishads. Always, reading Sanskrit is a work-in-progress, since one has never really sat down to mastering it. The more one reads, the more it transpires that this is the case for most readers and students, even scholars. The nature of the language is such that it yields up endless possibilities, alternative interpretations, and obscure allusions, almost as if by design. The profusion of word forms, for instance the various ‘tenses’ and ‘moods’ of verbs, and the complicated rules of word concatenation, ‘sandhi’, ensure that each sentence is a riddle that has to be worked at. The old masters were adepts, and loved to produce finished pieces that would not yield their secrets easily except to those willing to put in the effort.

Now anyone who has been dabbling in this language, or indeed in Linguistics in general,  would not have failed to come across the name of Panini, that famed grammarian of the pre-Christian era. Like all other authors in ancient India, not much is known about him, except that he must have been one in a long chain of highly accomplished and learned scholars. Classical Sanskrit was described by him in a work of 4000 aphorisms, ‘sutra’, arranged in eight chapters or books: hence known as the ‘Ashtadhyayi’, the eight-chaptered. Scholars are stunned at the brevity of his formulations, the imaginativeness of the technical words he coins, at the elegance of the concept itself. Many draw parallels with computer flow-charts, as Panini apparently tried to account for each form and every variation he knew, through a sequential series of operations on root words or lexemes. Students of Linguistics are told that the very discipline of Phonetics owes its conception to Panini, discovered by the astonished Indo-Europeanists in the middle of the 19^th century with such fabled names as Bopp and Grimm and Jones and Monier-Williams and Max Muller and so on.

 So why is it that so few of us have any more detailed acquaintance with this massive work and this fabulous intellectual heritage of our forebears? All we get is a glimpse of the riches that lie hidden, through the glowing tributes and tantalizing suggestions of its modern relevance, in such works as alluded to above. The only brush with the actual work for me, and I am sure for most Indians with a little Sanskrit in their school days, is probably the technical terms used for the tenses and moods of verbs. I don’t remember the whole verse, but it starts like this: “Lat vartamane, Let Vede, bhute Lang Lung Litasthata…”… and then goes on to coin tongue-twisters like Lrng and Lrung (or maybe not!). We understand the first part of this: the ‘present tense’ is termed Lat, the Vedas use a form called Let, the  past tense ( ‘bhuta’, ‘that which has been’, hence ‘bhoot’ in Hindi for ghosts!) uses different forms called Lang, Lung, etc. At school, we would drill in the forms at least for present tense (bhavati, bhavatah, bhavanti…), past (abhavat, abhavatam, abhavan…) and simple future (bhavishyati, bhavishyatah, bhavishyanti…), apart from ‘imperative’ (bhavatu, bhavatam, bhavantu…), and the alternative forms of the ‘Active’ (Parasmaipada) and ‘Reflexive’ (Atmanepada) aspects. One could only wonder at the huge number of such meaningless labels developed by Panini for other categories, which obviously sound even more strange than Greek or Latin… they are no man’s language, a purely symbolic meta-language, probably the reason for computer geeks’ fascination with it.

Why would Panini use such nonsense terms for the ten0ses or other categories and entities? Why not nice simple explanatory names like the English ‘present tense’, ‘past tense’, etc? It’s probably because Panini was optimizing the terminology for oral recording and transmission, writing media obviously being in short supply.  The strength of the oral drilling is evident in the fact that after all these years, even a poor student like me has those basic forms imprinted on the mind and can reproduce them off the cuff, as above (printed page unseen). To compress a complex, inflectional language like Sanskrit into a manageable set of expressions, that could be readily mugged up and readily alluded to in instantly recognizable formulae, probably required the coining of such outlandish ‘words’ and phrases. So much so, that the venerable Shankaracharya even sniffs at the learned and dry pundits who go around muttering these strange formulae like crows going ‘dukram dikram’, whereas the way to God is through love and the heart…

So with such a tantalizing heritage in our own culture, it seems to me we should be making at least an effort to understand what makes Panini so great, and of course this applies not just to Indians with a little Sanskrit from school, but also all Indo-Europeanists and students of Linguistics in general. So that will be my effort in the following days and weeks… starting from a position of almost complete ignorance, to see how much effort is required to make sense of these mutterings and incantations! I hope fellow-Sanskritists will bear with me if I blunder badly, and I do hope fellow-learners will help out along the way…watch this space!

(Just a thought...I've not used any special symbols for the sounds in the Sanskrit words at this stage. Panini is apparently a bread, but here it's pronounced with the tongue curled up against the roof of the mouth, i.e. in a retroflex manner...usually denoted by a dot under the letter!)