Footnotes from the book are added immediately after the sentence in parenthesis and
they start with a footnote number and a colon. My additional explanations are also 
added in parenthesis and start with 'N:'. 

Key for pronunciation of Sanskrit terms (English sound/Slovene sound):

- A, I, U denote long vowels, 
- R, L denote vowels 'r' and 'l' (syllabic  r and l) as in Slovene words 'čvrst' (firm) or 
pre?-proto-Slavic(?) 'vLk' (wolf)
- y - y/j,
- j - j/dž,
- c - ch/č, 
- S - sh/š, 
- z - soft /palatalized sh (here written wiith sh),
- "G, J, N, n" different kind of nasals ("GG, J, n" are in Slovene written with the same letter 'n' 
(ban(G)ka, panj(J), normal n; whereas the retroflex 'N' doesn't exist), 
- "T, D" - retroflex version of the similarr dental sounds 't' and 'd'
- consonant followed by h denotes an aspiraated consonant (ch - ch/č followed by an aspiration, 
ph - p followed by an aspiration, etc...).



George Cardona: pANini - A Survey of Research (1976)
George Cardona: Recent Research in pANinian Studies (1999)


I. INTRODUCTION

I treat in this volume research done on major school of Indian grammarians, pANini's, the 
fundamental work of which is pANini's grammatical treatise called the /aSTAdhyAyI/. Second 
in antiquity among the works fully preserved in this school are the scolia, called /vArttikas/ 
(N: explanatory or supplemental rules), on pANinian rules by kAtyAyana. pataJjali's /mahAbhASya/, 
the work of the third of the three sages /muni-traya/ of the most ancient period, deals both with 
kAtyAyana's /vArttikas/ and also independently with pANinian statements. Subsequently, pANini's work 
was the object of two types of commentaries. Some commentators followed the order of rules as they 
appear in /aSTAdhyAyI/, others reordered these and dealt with them in major sections concerning 
definitions, metarules, rules of /sandhi/ (N: changes of sounds, in speech only, or in writing, 
because of the preceding and/or following sounds: a table -> an apple), etc.

The /mahAbhASya/ itself contains discussions of topics more general linguistic and philosophical 
interest than the interpretation and application of rules. These discussions foreshadow later, 
more extensive treatments of such subjects in semantic treaties, of which three major ones are 
bhartRhari's /vAkya-padIya/, kauNDabhaTTa's /vaiyAkaraNa-bhUSaNa/ and nAgezabhaTTa's 
/vaiyAkaraNa-siddhAnta-(laghu)-maJjUSA/.







III.1 pANini - THE aSTAdhyAyI
III.1.1 INTRODUCTION

The /aSTAdhyAyI/, also called /aSTaka/, is a grammar of Sanskrit. It consists of eight chapters, 
further divided into quarter-chapters, and contains about 4000 rules, called /sUtra/, preceded 
by a catalog of sounds itself subdivided into 14 groups and variously called /pratyAhAra-sUtras/, 
/shiva-sUtras/, and /maheshvara-sUtras/. (11: A /sUtra/ is normally a verbless statement in which 
one is to understand the verb "be") (12: These groups of sounds are arranged in subsets to allow 
the use of abbreviational terms such as 'ac' "vowels" to refer to sets of sounds. According to 
tradition, this catalog of sounds was handed down to pANini by the lord Shiva (maheshvara).)

(N: /pratyAhAra sUtras/ are: 1. /aiuN/, 2. /RLk/, 3. /eoG/, 4. /aiauc/, 5, /hayavaraT/, 6. /laN/, 
7. /JamaGaNanam/, 8. /jhabhaJ/, 9. /ghaDhadhaS/, 10. /javagaDadash/, 11. /khaphachaThacaTatav/, 
12. /kapay/, 13. /shaSasar/, 14. /hal/. The sets are formed by taking a sound in some /sUtra/, 
which is not last (like 'a') and combining it with a marker - last letter of some /sUtra/. Such 
/pratyAhAra/ (abbreviation) stands for all non-marker sounds from the first sound to the last 
sound before the marker (all markers are ommited). So /ac/ denotes a set of vowels, namely a, i, u 
(marker N is ommited), R, L, (k ommited), e, o, (G ommited), ai, au (c ommited). Simmilarly 
/hal/ = {ha, ya, ra, va, ... sa, ha} i.e. consonants, /aic/ = {ai, au} (diphthongs), /eG/  = {e, o}, 
/ik/  = {i, u, R, lR},  yaN = {y, r, l, v}. Marker is techincally called /it/ in Sanskrit.)

In these rules (N: sUtras of aSTAdhyAyI) pANini refers to groups of verb and nominal bases 
in various ways. These bases appear in two ancilliaries to the corpus of rules, a catalog of 
(N: verbal) roots, called the /dhAtu-pATha/, and a catalog of nominal bases called the 
/gaNa-pATha/.

III.1.2 EDITIONS AND TRANSLATIONS, INDICES AND LEXICA OF THE aSTAdhyAyI

There is available a good number of editions of the /aSTadhyAyI sUtra-pATha/ of which some merit 
special mention. ... Pathak-Chitrao 1935..., Sankara Rama Sastri's edition (1937)..., ... Boethlingk
published first (1939-1940) a text of /sUtra-pATha/ from an earlier Calcutta edition. Later 
(Boethlingk 1887a) published what in Europe is considered the standard edition of the text, with 
a German translation. R. Rocher (1965b) has noted some corrections to be made in this work. Renou 
(1969) remarked that Boethlingk's translation contains "certain deficiencies". In fact, Boethlingk's 
version of pANini's grammar, though it merits praise as a pioneer work, hardly serves to give 
a student insight into the way the rules apply and are related to each other. It fails to give 
anywhere near the cross-referencing necessary to show this. For example Boethlingk translates rule 
7.2.115 /aco J-N-iti [vRddhiH 114]/: "/vRddhi/ wird auch fuer den Endvokal eines Stammes substituirt 
vor einem Suffix mit stummen J oder N". No reference is given to any other rule. Yet a good number 
of rules has to be considered in order to interpret this one rule correctly. 

(15: One must consult 1.1.17 /Adir antyena saheta/, to know that /ac/ denotes a set of vowels and 
1.3.3 /hal antyam/ to know that the /c/ of /ac/ is a marker. To know that the vowels 
/a, i, u, R/ etc., denoted directly by /ac/, themselves denote a set of vowels (N: a set of short 
and long voweels, regardless of the accent. If I disregard the accents then /a/ = {a, A}, /i/ = {i, I},
etc...), rule 1.1.69 /aN-ud-it savarNasya cApratyayaH/ has to be consulted. The term /vRddhi/ 
denotes /A, ai, au/ by 1.1.1. /vRddhir Ad aic/. The genitive (N: /aco/ written without sandhi as 
/acaH/ is a genitive of /ac/; genitive denotes a substituend in the rule) and locative (/J-N-iti/ 
is locative of a compounded word /J-N-it/: an element(affix, for instance) denoted by a marker /it/ 
of /J/ or /N/; locative marks a right context of the operation, i.e., the element needed on the 
right side of the substituend) are interpreted by 1.1.49 and 1.1.66. In addition, the term /ac/ in 
7.2.115 denotes not merely vowels but a presuffixal base ending in a vowel, which cannot be known 
without consulting 1.1.72 /yena vidhis tad antasya/. This provides that a term X used to state 
an operation for Y, which X qualifies, denotes the unit Y ending in X. The heading 6.4.1 
/aGgasya/ is valid in 7.2.115, so that this states an operation on a presuffixal base /aGga/ 
ending in a vowel. Further, 1.1.52 /alo 'ntyasya/ must be consulted to see that the substituend 
in 7.2.115 is the final vowel of such base. And 1.1.50 must be consulted to know which of the 
vowels denoted by /vRddhi/ replaces a particular substituend. A translation like Boehtlingk's, 
though it looks faithful to the original pANinian statement, fails to show that metarules and 
definitions indeed apply together with rules which they serve to interpret.

(N: Trying to make it more simple: The sUtra 7.2.115 /aco J-N-iti [vRddhiH 114]/ states literally 
that "in place of a vowel, which precedes the element marked by J or N, there comes a vRddhi letter,
or mathematically:  

{a, i, u, R, ...} -> {A, ai, au}" |with right context of an element marked by J or N 

I will not explain which vowel is substituted with which /vRddhi/ letter, too complicated; 
it would involve definition of some phonetical terms, interpretation of at least 5 new sUtras 
and additional number (unknown to me, as of yet) sUtras which define replacement techniques 
- the general rule, for instance, is that ssets of equal power are treated as ordered sets, 
so the substitution is made first element of the first set with the first element of the second set, 
etc... Only that the above sets don't have equal power!). 

Because of the heading rule 6.4.1 /aGgasya/ "of the base" and interpretation according to 
1.1.52 /alo 'ntyasya/ (all genitives) - literally "of the last /antyasya/ sound /alah/", /al/ 
being a pratyAhara denoting all sounds (/a/ is first letter of the first pratyAhAra sUtra and 
/l/ is the marker of the last pratyAhAra sUtra) - together "of the last sound of the base", 
we know that this vowel /acah/ must be the last letter of a base word. For an example see below.
end-of-comment)

Boethligk's translation gives the impression that the rules such a 7.2.115 can stand for 
themselves as independently interpretable statements, thus obscuring the extraordinary to 
which metarules, definitions and headings must be considered constantly in order to interpret 
any given rule. A translation, "The final (1.1.52) vowel (ac: 1.1.71, 1.1.69) of a presuffixal 
base (1.4.13, 6.4.1) which ends in (1.1.72) a vowel is replaced by (1.1.49) vRddhi (1.1.1, 1.1.50) 
when it is followed by (1.1.66) an item marked with J or N markers" is admittedly cumbersome. 
However it may be necessary to resort to precisely such renditions in order to bring out clearly 
how pANini's rules operate.

Especially since translations of the aSTAdhyAyI are apt to be used not by those familiar with 
pANini's system but by linguists and others who wish merely to become acquainted with what pANini 
did. (N: of course, someone who studied formal language definition methods wouldn't have such 
difficulties; the above conclusion would to his be an obvious fact - so this comment is probably 
more for linguists.) 




III.1.3 CONCERNING THE ORIGINAL TEXT OF THE aSTAdhyAyI
III.1.3.1 EVIDENCE OF PRE-pANinian GRAMMARIANS

In the aSTAdhyAyI pANini mentions by name ten other persons, all presumably grammarians: Apishali, 
kAshyapa, gArgya, gAlava, cakravarman, bhAradvAja, shAkaTAyana, shAkalya, senaka, sphoTAyana. ... 
According to the commentators, pANini's mentioning such other grammarians indicates that 
the operations in question are optional; and the later grammarians indeed substituted 'vA (or)' 
for names. When, however, a rule contains both the name of a grammarian and an item such as 'vA' 
indicating that the operation in question is optional, commentators then say that this name is stated 
honoris causa. However, most modern scholars have recognised that pANini's mention of other 
grammarians serves to convey difference in usage; and when the name of another grammarian is coupled 
with a term such as 'vA', the rule in question states an optional operation recognised by that 
grammarian. In effect, these are mentions of dialectal usages. ...

From the above information it is proper to conclude that other grammarians, predecessors or 
contemporaries of pANini, had, like him, noted particular usages, of which pANini took notice 
in his own work. ... (N: one material I read compared the aSTAdyAyI - the oldest grammar of 
Sanskrit known to us (pANini, so is said by modern indologists, has lived around 500 BCE, but 
there are variations by different scholars, up to 800 BCE, or even more by Indian tradition) - 
to the Boeing 747 which some archeologist dug out on some archeological site. Maybe a dent 
here or a bump there, but just pour in some gas (interpret the rules) and you can go flying 
(define correct usage of Sanskrit, and consequently incorrect usage too). He says that there 
must have been years (few hundred at least IMO) of preceding tradition of linguistic exploration, 
which enabled to build such complicated apparatus as Boeing 747 (or this grammar). Of course, it 
takes someone who is on the similar level (i.e. is aware of developments of modern lingustics 
and computer science - like The Chomsky–Schützenberger hierarchy of classes of formal grammars 
http://en.wikipedia.org/wiki/Chomsky%E2%80%93Sch%C3%BCtzenberger_hierarchy, or language definition 
techniques from computer science), to recognise what the original such a grammar stands for and 
what is its intent.) ...



III.1.5 THE SYSTEM OF aSTAdhyAyI
III.1.5.1 GENERAL INTRODUCTION
III.1.5.1a pANini's DERIVATIONAL SYSTEM - GENERALITIES

pANini's grammar is descriptive, not prescriptive. The rules of the grammar serve to derive forms 
which accord with correct usage. These rules /sUtra/ are also called /lakSana/ 'characteristic, that 
by which ... is characterized' in that they serve as means to characterize, that is, to explain 
by derivation, the forms of correct usage. ... Basic to this derivational system is the distinction 
between bases /prakRti/ and affixes /pratyaya/. The grammar provides for introducting affixes after 
bases (N: 'after' only in the process of derivation; affixes can be suffixes, infixes or postfixes) 
under given conditions to derive items terminating in verbal or nominal endings. These are called 
/pada/. The bases themselves are of two types: verb roots /dhAtu/ and nominal bases /prAtipadika/. 
In addition, bases are either primitive or derived. Primitive verb roots appear in a /dhAtu-pATha/ 
and primitive nominal bases would appear in a lexicon; the /gaNa-pATha/ represents a partial lexicon 
of items which undergo particular operations. Derived bases are gotten by applying rules of grammar. 
Derived verb roots are formed from both primitive roots and nominal forms. Examples of the first are 
causatives such as /kAr-i/ 'make to do, have to do' (/kR/ 'do, make), desideratives such as /cikIr-Sa/ 
'wish to do'; examples of the latter are denoinatives such as /putrIya-/ 'desire a son /putra/ for 
onself', /putrakAmya-/ 'idem'. Derived nominal bases too are gotten from both verb roots and nominal 
forms. Bases such as /pAk-a/ 'cooking', /pAc-aka/ 'cook' are derived from the root /pac/ by introducing 
the primary affixes (called /kRt/) /ghaJ/ (=a) and /Nvul/ (= aka) (N: after introducing affixes, which 
have names /ghaJ/ or /Nvul/, and are marked with markers /J/ and /N/, last vowel of the base changes to 
the /vRddhi/, in this case /a/ of /pac/ changes to /A/ of /pAka, pAcaka/). From nominal forms are derived
compounds such as /rAja-puruSa/ 'king's man' and items sucj as /aupagava/ 'descendant of Upagu', /tatra/
'there', which contain affixes called /taddhita/. ...


III.1.5.1b pANini's SYSTEM: STUDIES OF GENERAL ASPECTS

There has not been published to date (N: cca 1976) in a European language a single work in which 
pANini's total system is set forth clearly and with insight.. ...

pANini's substitution procedure has received renewed attention in recent years. Particularly 
interesting are attempts to formalize pANini's morphophonemic substitution rules. To my knowledge, 
this was initiated by Staal (1965), whose notation has been adopted and refined by Cardona (1969). 
Staal's initial effort is praiseworthy for its attempt to clarify. However, his paper also contains
elementary errors and misrepresentations of what pANini said. Van Nooten has also recently dealt with 
aspects of pANinian substitutions. In the latter article (1970), he suggests a formulaic statement of
replacements such as "au -> o". He proposes to write "x + y -> z /C", which is to be read, "In the 
context C, a successtion of elements x and y is replaced by one element z." He does not, however, 
note the important fact that pANini accounts for such replacements by the interesting stratagem of 
making the substituends their own left and right contexts.

One particular replacement, zero, has attracted special attention. This has been discussed by 
W. S. Allen (1955), who proposed that the linguistic zero was known before the mathematical zero 
(Toporov 1961, Staal 1974). (N: I have read another material, which also states that the Indian 
invention of mathematical zero is a direct consequence the linguistic studies and linguistic zero 
used in Indian grammars). Karunasindhu Das (1984) has briefly discussed, with examples, the various 
types of zero with which pANini operates, designated /lopa, luk, shlu/ and /lup/. ...


III.1.5.2 THE COMPOSITION OF THE GRAMMAR: GENERAL ORGANIZATION, RULE ORDER, TYPES OF RULES
III.1.5.2b RULE ORDER AND THE APPLICATION OF RULES

Some rules of the aSTAdhyAyI are extrinsically ordered (rules of the tripAdI - last three 
quarter-chapters of the last chapter). ...

In other cases the external order of rules is not pertinent to their application: the order of 
application is not the same as the external order of the rules. Here pANini makes use of other 
decision procedures to allow one to select proper rules and apply them correctly at any given 
stage of derivation. For example, in deriving the utterance /ayaja indram/ 'I sacrificed to Indra' 
there is a stage /ayaja-i indra-am/. Here two operations can apply: a+i->e or i+i->I. Only if 
the former applies can the correct utterance be derived: /ayaja-i indra-m -> ayaje indram -> ayaja indram/.
Bracketing is recognized: an operation the cause of which is internal relative to a condition 
which causes another operation applies prior to the latter. In the present instance one starts 
with /((ayaja-)i) indra-m/ and works outwardly from internal brackets (Cardona 1970a). 

Another principle has to do with relation of rules with each other. Given to root /dah/ (to burn) 
followed by the participial affix /ta/, one should derive /dag-dha/ (from /dah+ta/). Two rules 
can apply. By one (8.2.31 /ho Dhah/), /-h/ is replaced by /Dh/ when there follows a consonant 
such as /t/. The other rule (8.2.32 /dAder dhAtor ghaH/) provides that /-h/ which is part of a 
root /dhAtu/ beginning with /d/ (/d-Adi/) is replacey by /gh/. Only the latter should apply here: 
/dah+ta -> dagh-ta -> dagh-dha -> dagdha/. A general principle allows one to choose 8.2.32. 
8.2.31 is a general rule (/utsarga/) applicable to any /-h/. But 8.2.32 is a particular rule 
(/apavAda/) related to 8.2.31: its domain of applications in included in that of the latter. 
If 8.2.32 is not to lack a proper domain of application, it must block 8.2.31. In all such cases
a principle of vacuity is used: a particular rule applies instead of a related general rule, 
from the domain of which it extracts its own domain of application. ...




III.1.5.2c TYPES OF RULES

The major types of rules are:

- definitons /saMjJa-sUtra/,
- metarules /paribhASA/,
- headings /adhikAra-sUtra/,
- operational rules /vidhi-sUtra/,
- restrictions /niyama-sUtra/,
- extension rules /atidesha-sUtra/,
- negation rules /niSedha-sUtra/.



III.1.5.3c MARKERS (213: I have used this term to translate pANini's term /it/.)

pANini uses markers appended to items. He calls them /it/ and /pANinIyas/ refer to them also using 
the term /anubandha/. Originally the grammar was not transmitted in writing but recited. Therefore, 
the markers used to appear as parts of the items which they characterize. In order, then, to show 
which sounds are markers, pANini states a series of rules (1.3.2-8). For example, 1.3.2, 
(/upadeshe 'j anunAsika it/ N: without sandhi - /upadeshe ac anunAsikaH it/) states that nasalized
/anunAsika/ vowel /ac/ in an item when it has been introduced but has not yet entered into 
grammatical operation /upadeshe/ N: original enunciation (i.e. the original form [often having an 
anubandha] in which a root, base, affix, augment, or any word or part of a word is enunciated 
in grammatical treatises) is an /it/. sUtra 1.3.3 (/hal antyam/) provides that the final consonant 
(/hal/) of an item at the same stage (/upadesha/) is an /it/.

Markers are obviously not part of linguistic items as they are actually used. They appear with items 
only in grammar. Hence, they are unconditionally deleted, before items enter into operations. Rule 
1.3.9 (/tasya lopaH/) provides that a marker (tasya 'in place of that - N: which was discussed before, 
namely the markers - /it/') is replaced by zero (lopa). ... (N: with a formula /it->0/)

Markers serve various purposes: to indicate that an item conditions or undergoes certain operations 
or belongs to a certain class, to distinguish homophonous units, to allow reference to a group 
of items by means of a common referend. Moreover, markers are used in forming abbreviations. 
A unit X followed by a marker M denotes not only itself but also all items which are listed between 
X and M (I.1.71: /Adir antyena sahetA/). For example, /ac/ is an abbreviation denoting /a/ and the 
following sounds of the Shiva sUtras, through /au/, which is itself followed by the marker /c/; 
/tiG/ denotes all verb endings beginning with /tip/ (= ti, marker p denotes an accent of -ti) and 
ending with /mahiG/ (= mahi); /sup/ denotes all the nominal endings beginning with /su/ (= s) and 
ending with /sup/ (= su).

(N: the /-ti/ ending marks the 3rd person singular active of the verb: /vadati/ 'he speaks', 
same ending as in Slovene infinitive, the slovene 3rd person singular has lost the ending; 
but dual endings are almost identical /-tas, -thas, -vas/, if these are combined with a same-meaning 
root, we get: /plu/ (to swim) - /plavatas, plavathas, plavAvas/ - plavata, plavata, plavava)

pANiniyan use of sounds as metalinguistic markers is comparable to the use of super- and subscripts 
in written grammars.


III.1.5.3d THE GRAMMAR OF pANini's METALANGUAGE

pAnini uses three case forms for particular purposes. A locative form is used to denote a right 
context of an operation, an ablative to denote a left context, and a genitive to denote a substituend.

III.1.5.3f QUOTATION

In Sanskrit an item is of course used to denote its meaning. To refer to a linguistic item itself, 
one uses the Sanskrit equivalent of quotation marks, the particle /iti/: /agniH/ - "fire" but 
/agnir iti/ - /"agni"/. However, pANini reverses this convention. He provides (I.1.68 /svaM 
rUpaM shabdasyAshabda-saMjJA/) that an item (/shabda/) serves to denote itself (/svaM rUpam/) 
unless it is a technical term (/ashabda-saMjJA/). This new convention is understandable. The 
grammar states rules which generally concern the items themselves. For example, rule VI.3.23 
(/agner Dhak/) introduces the affix /Dhak/ after a nominative form of the item /agni/, not after 
'fire'. ...



III.1.5.7b COMPARISONS WITH MODERN SYSTEMS AND TECHNIQUES

pANinijas such as pataJjali refer to pANini's grammar as a /shabdAnushAsana/ - '(treatise on) 
instruction (concerning the formation of correct) words. Earlier scholars such as Wackernagel (1896), 
and more recently, Thieme (1956) considered, that the grammar dealt exclusively with word formation, 
to the exclusion of syntax. In fact, however, pANini operates with a syntactic system, as 
has been recognised more recently, Cardona (1967), Joshi (1969). Al-George agrees with Wackernagel 
that pANini's principal aim was to provide for a word analysis, but he also notes that pANini's 
kAraka system (N: /kArakas/: /apAdAna/ (removal-ablative), /saMpradAna/ (recipient-dative), /karaNa/ 
(instrument-al), /adhikaraNa/ (locus-locative), /karman/ (object-accusative), /kartR/ (agent-nominative), 
and a special type of agents, a causal agent (/hetu/-agent of the causal verb)) implies a sentence 
analysis. I think some subtlety is requires here. It cannot be denied that pANini does deal with 
syntactic relations and relations among certain kinds of sentences. On the other hand, it is also patent 
that his basic derivational procedure consists in introducing affixes to bases, so that he also deals
essentially with the relations among /padas/ (bases with affixes) which constitute sentences. This
is important for appreciating the limits of his syntactic and semantic interests.

Ananthanarayana (1970) compared pANini's grammar with case grammar. Though case grammar is indeed 
possibly the closest modern analog to pANini's system, there are nevertheless important differences 
between the two. Cardona (1974) has briefly sketched one difference: the notion of subject is absent 
from pANini's system. In my opinion, pANini's approach differs in an essential way from case grammar. 
In the latter, one is forced to posit surface categories as subject or others, depending on the 
"surface phenomena" of individual languages. This is motivated, I think, in the first instance because 
the deep syntactico-semantic categories such as Agent, Object, have in this system, to be defined in 
purely semantic terms. This requirement is, in turn motivated by another attitude, namely that the 
grammarian is seeking to posit universal categories of language. It is, I think, clear that pANini 
did not define his kAraka categories in purely semantic terms. Nor is there any evidence to suggest 
that they were seeking to posit universal categories of language. There aim was straitforward: to give 
a set of rules, which account for correct Sanskrit usage. ...



III.1.5.8e CONCLUSION

pANini's grammar has been evaluated from various points of view. After all these different evaluations,
I think the grammar merits asserting, with Bloomfield (1933), that it is "one of the greatest monuments 
of human intelligence".







III.2 THE mahAbhASya

The earliest extensive discussion of pANini's rules preserved to us now are contained in the /vArttika/
of kAtyAyana, which themselves are known as cited and discussed in pataJjali's /mahAbhASya/. This text, 
is not, however, a true commentary on the grammar. kAtyAyana and pataJjali do not explain all rules, 
with examples to show how they operate. They discuss the validity of rules, how they are stated, their
relations to other rules, and whether some rules or parts of them can be eliminated without harm and
additional rules need to be stated.

III.2.6 METHODS
III.2.6.1 GENERALITIES

The mahA-bhASya is composed in the form of dialogues in which take part a student /shiSya/ who 
questions the purpose /prayojana/ of rules and their formulations, and unaccomplished teacher 
/AcAryadeshIya/ who suggests solutions which are not fully acceptable, and a teacher /AcArya/ 
who states what is the finally acceptable view /siddhAnta/. Commentators also refer to an 
/ekadeshin/ 'one who knows only part /ekadesha/ of the final answer' and a /siddhAntin/ 'one 
who establishes the final view'.

The argumentation involved in these discussions includes the citation of examples /udAharaNa/ 
and counterexamples /pratyudhAraNa/ for rules and also illustrations /dRSTAnta/ showing how 
things procese in grammar in some ways parallel to real life.


III.2.6.3 DISCUSSIONS OF PHILOSOPICAL IMPORT

The mahAbhASya contains discussions of subjects on the threshold of grammar and philosophy, many 
of them concentrated in the introductory section.

A basic premise of grammar is, according to /pANinIyas/, that the relation /sambhanda/ between 
linguistic items /shabda/ and their meanings /artha/ is fixed and permanent /nitya/, not the 
invertion of someone. A question discussed at the beginning of the /mahAbhASya/ is this. What 
is it precisely that one calls /shabda/? Two answers are given. A /shabda/ is that which, when 
articulated, serves to convey an understanding of the meanings. Alternatively, one can understand 
a /shabda/ to be merely sound. That is, any item can be viewed either qua signifier of qua sound complex.

A related question taken up in the /mahAbhASya/ is whether these linguistic items /shabda/ are 
perennial, eternal /nitya/ or susceptible of production /kArya/. Moreover, a distinction is 
made between absolute eternality /kUTastha-nityatA/ such that an item is susceptible to no 
modification whatever, and what is called /pravAha-nityatA/, the perennity of linguistic items 
as used through generations of speakes. ...

Concerning what items signify /artha/, one important question discussed is: does a noun such as /go/ 
(cow) designate an individual thing /dravya/ or a type /akRti/? Commentators consider that this term 
was used in two values by pataJjali: as an equivalent of /jAti/ 'generic property' and to denote a 
form, a particular arrangement of parts /avayava-saMnivesha-visheSa/ (Cardona 1967-68). Sreekrishna 
Sharma concluded that pataJjali used /AkRti/ to mean 'structural form' viewed in two ways: common 
form (=universal) and particular form. He distinguished between /AkRti/ 'structural pattern' and 
/jAti/ 'kind of class', which, he noted, is an abstract concept. ...

The mahAbhASya on rule 4.1.3 takes up the question of gender. The term /liGga/, used in the meaning 
'gender', denotes, in normal Sanskrit, a mark of characteristic. If the term is understood in this 
sense in grammar, then a /liGga/ is a characteristic of males, females and things which are neither. 
A female /strI/ would then be characterized by breasts and hair, a male /puruSa/ by his body hair, 
other by neither. This conception of /liGga/ does not work in grammar, it is noted, so that another 
concept is introduced. Any thing is characterised by different states of constitueent elements of 
properties /guNa/ and these states constitue the gender of things (382: as commentators have noted, 
this view has affinities with the theories of sAMkhya-yoga. Jacobi considered the views put forth 
in the mahAbhASya crude with respect to the sAMkhya-yoga theories. Later, aformal definition of 
grammatical gender was proposed: this is masculine which is referred to by /ayam/ 'this - 
nom.sing.masc.', that is feminine which is refered to by /iyam/ 'this-nom.sing.fem.' and that is 
neuter which is referred to by /idam/ 'this-nom.sing.neut.' This question is related to another topic
discussed, namely whether one is justified in recognising a substance /dravya/ as something distinct 
from its constituent parts or properties. /pANinIyas/ also had a formal definition of a thing: 
anything in the range of a pronoun (, which is used as a variable) is a thing.

The concept of time /kAla/ and its divisions are also discussed in the mahAbhASya. Twho of the 
questions treated are the following. One speaks of actions as taking place currently /vartamAna/, 
in the past /bhUta/, and in the future /bhabiSyat/. Now, one can properly say /ihAdhImahe/ 'we 
are studying /adhimahe/ here /iha/, using the present form of the verb (N: middle form, not active, 
is used here), and the grammar accounts for this by introducing the L-suffix /laT/ on condition 
that an action is characterised as current (3.2.123 /vartamAne laT/). But the sentence /ihAdhImahe/ 
can be proper even if, at the moment, one is not actually studying. Therefore, current time 
/vartamAna-kAla/ is characterised more specifically: once an action has begun and has not ended, 
the entire stretch of time included therein is refered to as current. The other question is this. One 
can say of mountains "The mountains are standing - Gore stojijo" /tiSThanti parvatAH/, as one 
also says of rivers "The rivers are flowing - Reke tečejo" /sravanti nadyaH/. Such sentences are 
derived on condition that an action is referred to current time /vartamAna/. Yet, the mountains 
are always there and the rivers always are flowing, so that it is difficult to see how one can 
speak of current actions here, as opposed to part and future ones. kAtyAyana answers that there 
are indeed time divisions /kAla-vibhAga/ here too. And pataJjali explains as follows. There are 
individuals such as kings, who existed in the past, exist now, and will exist. These different 
individuals perform activities. And it is with respect to these activities that one can speak 
of mountains as having stood at the time of past kings, standing at the time of present kings, 
etc. (384: On the basis of this passage it was also claimed that what one refers to as time 
is merely actions which characterise other actions.)

One recent work (Scharfe 1961a) has been devoted to an investigation of logical concepts, terms 
and techniques found in mahAbhASya. This is a useful collection of materials. Nevertheless, 
Scharfe's work suffers from two faults. First, the author procedes rather mechanically, 
equating terms used in other Indian systems in particular values with homophonous terms which 
occur in mahAbhASya, and he doesn't always note important differences. For example, he says that 
the terms /anvaya/ and /vyatireka/ are used in mahAbhASya with the meanings they have in the 
context of contraposition, that is: a->b /anvaya/ and not_b->not_a /vyatireka/. However, these 
terms are often used in other values in the context of showing that a is the cause of b: 
a->b /anvaya/ and not_a->not_b /vyatireka/. And it is this values terms have in the mahAbhASya 
discussions considered by Scharfe. 

Less ambitious than Scharfe's work, yet by the same token more trustworthy, is the recent 
work by Rama Prasada Tripathi 1963, who discusses the views found in the mahAbhASya and 
other pANinIya texts concerning the means of obtaining and coveying correct knowledge, 
what are called /pramANa/. 

Frauwallner 1959 and Scharfe 1961a have agreed in chiding pataJjali for what they considered 
his lack of philosophical acumen. Frauwallner declared that pataJjali had neither an interest 
in nor a head for philosophical questions and claimed that pataJjali took a topic of discussion 
from an earlier source and mangled it. Scharfe asserted that, in the passage considered, pataJjali 
did not recognise the essential feature of citation, that he argued instead from primitive analogy.
Frauwallner has been answered by Staal 1967 and Scharfe was rebutted by Stall 1963, both I think 
quite justly. Schafe simply misunderstood the discussion in question. Frauwallner either misunderstood
or refused to understand the context of pataJjali's discussion. pataJjali's aim, after all, was not 
to set forth in the mahAbhASya a fully elaborated philosophical system. We do not know whether he 
had in fact arrived at such consistent system. But even if he head, his aim in the mahAbhASya was 
to discuss pANini's grammatical rules, not to set down his own philosophical views in full 
array. (N: Of couse other Indian philosophers did that).

III.3 CHRONOLOGY AND REALIA
III.3.1 THE DATES OF pANini, kAtyAyana and pataJjali
III.3.1.1 INTRODUCTION

pANini, kAtyAyana and pataJjali have been assigned widely divergent dates. Renou gave the 
fourth century BCE as pANini's date on two occasions, but later said that pANini's date was 
either the fourth century of possibly the fifth. In his last treatment of the subject (1969) 
Renou remarked "Generally, I have decided, without absolute certainty, on the 4th or 5th 
centuries BCE." ... Certainly, most competent scholars have favored placing pANini's at a time 
earlier that the 4th century BCE. ...

I have no definite opinion on this problem. However, whatever one's view, this (in "..." 
discussed internal and external evidences) cannot serve as solid evidence for the date of pANini. 
One must, then, depend on relative chronology: if the dates of pataJjali and kAtyAyana can be 
determined, pANini's date can be approximated.

III.3.1.4 THE DATE OF pataJjali

... The evidence is thus not absolutely probative but sufficient to warrant considering seriously
that pataJjali lived in the second century BCE. (407: Another issue which has engaged scholars 
should be mentioned here although its resolution does not contribute directly to the precise 
dating of pataJjali. According to pANinIyas, such as nAgesha, pataJjali was the author not only 
of the mahAbhASya, but also of the yoga-sUtras and the medical work caraka-saMhitA 
(N: all these three works is said by the tradition that they were written by pataJjali). Some 
modern scholars have argues that the author of the mahAbhASya was not the same author as the 
author of the yoga-sUtras. Fairly recently, Janaček 1958, concluded that Jacobi's (1931) 
appeal to a statistical method with respect to the use of vocabulary in the texts in question 
proves nothing. I have no definite opinion on this question. K. Madhava Krishna Sharma (1944-45) 
argued against another traditional identification, namely that pataJjali - who is considered 
to have been /sheSa/, /viSNu's/ serpent - was identical with /sheSanAga/, the author of the 
philosophical text /paramArtha-sAra/.


III.3.1.7 CONCLUSION

The evidence for dating pANini, kAtyAyana and pataJjali is not absolutely probative and depends
on interpretation. However, I think there is one certainty, namely that the evidence available 
hardly allows one to date pANini later than the early to mid fourth century BCE.

VI. TREATISES ON SEMANTICS AND PHILOSOPHY OF GRAMMAR
VI.1 INTRODUCTION

Questions of general linguistic and philosophical import were already treated fairly extensively 
in the mahAbhASya, and these discussions were the bases for later elaborations. There are later 
pANinIya treatises in which such issues are dealt with in full. Not only the issues considered, 
but the view of other schools of thought are also given detailed attention. ...