Studies On Panini's Grammar

STUDIES ON
PANINI'S GRAMMAR
BY
BAREND FADDEGON
VERHANDELING DER KONINKLIJKE AKADEMIE

VAN WETENSCHAPPEN TE AMSTERDAM
AFDEELING LETTERKUNDE
NIEUWE REEKS. DEEL XXXVIII. N0. 1
SECOND PRINTING: 1963

UITGAVE VAN DE N.V. NOORD-HOLLANDSCHE
UITGEVERSMAATSCHAPPIJ AMSTERDAM. 1936
Goldstcker has admirably attacked Bhtlingk. but for Bhtlingk
we forget Goldstcker;
and Whitney has admirably attacked Pl).ini. but for Pl).ini we
forget Whitney.
I adore Bhtlingk because he reveals to us the spirit of Pl).ini.
I adore Pl).ini because he reveals to us the spirit of India.
I adore India because it reveals to us the Spirit. the Spirit.
BIBLIOGRAPHICAL TABLE.
Not e. - The subsequent table does not claim to be a full account of the literature
on PI?ini, for, since the article intends in the first place to show the importance of
explaining P~i by the aid of Pl.linl himself, that is Bhtlingk's edition with its numerous
Indexes, several subjects such as the relation between Pl}inl and Yska and the
PrlltiSkhyas, between P~i and the language of the literary doclHOents had to he left
out and even the commentaries on Pl}ini to be sparingly dealt with.
1. Sou r ces a ndd i c t ion a r i e s.

SP. = Sutra~Ptha, Bhtlingk's edition, p. 1---480. Adhy. = adhyya, p. = pda,
su. = sutra, fin. = end of pda. Often the mere mentioning of three ciphers suffices.
GP. = Gal}a~P~ha, Bhtlingk's edition, p. 95*-145*.
DhP. = Dhtu~Ptha, Br. Liebich's edition. Zur Einfhrung in die indische elnheimische
Sprachwissenschaft. Sitzungsberichte der Heidelberger Akademie der Wissenschaften,
Philos.~histor. Klasse, Jahrgang 1920, 10. Abhandlung. Heldelberg 1920.
MBh. = The Vykaral}a~Mahbh~ya of Patafijali, edited by F . Kielhom, I-lIl.
Bombay 1880--1909.
Ksik, Chowkhamb~edition, Benares 1906.
BrhadDevat, attributed to Saunaka, edlted and translated by A. A. Macdonell. Harvard
Oriental Series V and VI. 1904.
Nirukta. Bibliotheca Indica edition. Calcutta 1882-1891. Vol. I = Nighal}~u, vol.
lI-IV = Nighal.ltu~bh~ya. (The bh1/ya is quoted according to its threefold division as
weU as to lts division in adhyyas and khI].gas),
BRW. = Sanskrit Wrterbuch von Bhtlingk und Roth. Petersburg 1855-75.
MWD. = Sir Monler Monier~Willlams. Sanskrlt~English Dictlonary, Oxford 1899.
2. S tud les a n d 9 ram mar s, c h ron 0 log i c a II y a r r a n 9 e d.

W. D. Whitney. Bopp's Comparative Accentuation of the Greek and Sanskrit
languages. Joumal of the American Qriental Society, 5th volume, bibliographical notices
(p. 195). 1856.
A. Weber. Das Mahbh~ya des Patafijali (Benares 1872). Indische Studlen, 13th
volume. 1873.
Monler Williams. Practical grammar of the Sanskrit language. 4th edition. Oxford 1877.
W. D. Whitney. The study of Hindu grammar and the study of Sanskrit. American
Journal of Philology, 5th volume (p. 279). 1884.
J. S. Speyer. Sanskrit Syntax. Leyden 1886.
O . BhtIingk. Sanskrit Syntax by Dr. J. S. Speyer. Zeitschrift der Deutschen Morgen~
lndischen Gesellschaft, 41. Band (p. 179) . 1887.
Bhtlingk's edition = Pl}ini's Grammatik, herausgegeben, bersetzt, erlutert und
mit verschiedenen Indices versehen von Otto Bhtlingk. Leipzig 1887.
B. Delbrck. AItindische Syntax. Syntaktische Forschungen V . Halle a. SaaIe 1888.
B. Liebich. Zwei Kapitel der Ksik. Breslau 1892.
W . D. Whitney. On recent studies in Hindu grammar. American Joumal of Philology,
14th volume (p. 171). 1893.
J. Wackemagel. AItindische Grammatik I-lIl. (tome III in collaboration with A.
Debrunner). Gttingen 1896--1929.
A. A. Macdonell. Vedic Grammar, Grundriss der Indo~Arischen Philologie und AIter~
tumskunde I 4. Strassburg 1910.
H. Skld. The Nlrukta, lts place in old Indian literature, lts etymologles. Lund 1926.
6 Bibliographical table
V. Trapp. Die ersten fnf hnikas des Mahbh~yam ins Deutsche bersetzt und
erklrt. Leipzlg 1933.
H. E. Buiskool. Piirvatrsddham. Analytisch onderzoek aangaande het systeem der
Tripdi van Pl:l.ini"s A.,~dhyyi. Amsterdam 1934.
P. Thieme. PJ.1ini and the Veda. Studies in the early history of linguistic science in
India. AIlahabad 1935.
3. B ook s i n ei den t a 11 y quo t e d.
L. Binswanger. Probleme der aIIgemeinen Psychologie. Berlin 1922.
C. Cappeller. BIamgha. Mgha's Slsupla-vadha im Auszuge. Stuttgart 1915.
J. N. Farquhar. An outllne of the religious literature of India. Odord 1920.
H. Hirt. Indogermanische Grammatik I-VI. Heidelberg 1921-3i,
J. M. van der Hoogt. The Vedic chant studied in its textual and melodic form. Wage-
ningen (Holland) 1929.
A Berriedale Kelth. The Karma-Mmlps. The Heritage of India Series. London 1921.
F. de Saussure. Cours de linguistique gnrale. 3e dition. Paris 1931.
G. Thlbaut. Laug~ Bhskara's Artha-5alp.graha. an elementary treatise on Mml!1s,
edited and translated. Benares Sanskrit Series. Benares 1882.
M. Winternitz. Geschichte der indischen Literatur. Leipzig I-lIl. 1907-1920.
Not e. T h e t e x t u a I tra d i' ti 0 n 0 f th ie Ga ':l a - P ~ h a. See

Bhtlingk's edition p. 95* (on interpolations), Winternitz 111 p. 396 (on the Gar;taratna-
mahodadhi des Vardhamna) ; Liebich, Zwei Kapitel p. XXVI (on the linguistic
Importance of the G~a-Ptha).
I. THE GRAMMAR OF THE INDECLINABLES.
I. THE INDECLINABLES IN GENERAL.
1-14. Divisions of the indeclinables according to etymology and

function.
15-20. The technical terms of the grammar of indeclinables.
1. Synoptical table of the two passages on the

i n dec I i n a bie s.
The indeclinables (avyayas) are enumerated by P~ini in the passages
I, I, 37-41 and I, 4, 56-98 as follows:
I. Svar-di (I, 1,37).
la. Svar-di-gaTJ.a, SP. I, 1. 37; GP. 254.
b. Nipta, see 11.
2. Words derived from pronouns [and hom nouns and numerals,
see 6] missing a complete declension and ending in a suffix which serves
at the same time as taddhita (suffix of a denominative noun) and as
vibhakti (in th is context, a declinational ending) [or as adverbial ending],
SP. 1. 1. 38; 5, 3, 1-26 rand 5, 3, 27 sqq.].
3. Absolutives and infinitives in m, (tum), and diphthong, in tv,
tos, as, SP. I, 1. 39 sq.; 3, 4, 9-66.
4. Avyayibhvas, indeclinable compounds, in many cases formed by
the composition of a declined noun [or nominal stem] with a preceding or
following indeclinable of the prdi-gaTJ.a, SP. 1. 1. 41 and 2, 1. 5-21.
11. Nipta (cf. I 1 b), SP. 1. 1,37 and 1,4,56.
1. Cdi-gaTJ.a, SP. 1. 4, 57; GP. 85.
2. PrdayalJ, indeclinables of frequent occurrence, 22 (20) in number,
summed up GP. 154, which are sometimes used as karma-pravaaniyas
(infra, 4) and sometimes as upasargas i.e. as a class of gatis (infra, 3),
[and which are of ten compounded with a following noun in order to make
a declinable compound, ei th er tatpur~a or bahuvrihi, and with a following
or preceding nominal stem to make an avyayibhva] , SP. I, 4, 58 sq.,
GP. 154.
3. Gatis, 'close adverbs' to finite verbs (tin) and 'p:efixes' to deverba-
tive nouns and indeclinables (krt). As prefixes the gatis are said to be
'compounded with' (saha) the following krt.
a. upasargas used as gatis; compare English compounds such as 'to
beset', 'to understand', constructions such as 'to strike out', 'to blow
up', and compound participles such as 'well-bred', 'ill-bred', SP.
1,4,60.
8 Studies on Pl].ini's Grammar
b. indeclinables and some noun~cases, and derivatives from nominal

stems used as indeclinables, cf. compounds such as 'to back~bite' ,
'to way~lay'! 'to fulfil'. 'to white~wash', 'to cross~question', 'to fore~
teIl', and constructions such as 'to walk along', 'to look around':
urydi~gaTJl , cvi~c:lc etc., SP. 1,4, 61-79, GP. 31 and 243.
Supplement: the position of the gatis in the senten ce, SP. 1. 4,80-82.
4. Karma~pravacaniyas or 'prdaya1;t' syntactically loosened from the
verb, either requiring an accompanying noun~case or used as 'free adverbs'
without an obligatory construction, SP. 1. 4, 83-98.
Note I. PI).ini 's terminology. - (a) As a rule, the first
member of a compound is said to be 'comp~unded with' the second: for
instance in SP. 2, 1. 4 sup (supplemented from su. 2, cf. Ksik) refers to
the first and saha sup to the second member of a compound. (b) The term
samsa is never applied by PI).ini to the construction of a finite verb with
a close adverb such as pra t~thati (cf. SP. 8, 4, 14) .
Not e 11. A n e x p I i c i t e x c e p t ion t o t h e adhikra. -
The siitras 1, 4, 58--60, where the upasarga used as a gati has two tech~
nical names, form an explicit exception to the adhikra 1. 4, 1 as indicated
by ca in su. 60: cf. su. 2, 1, 23.
2. C har act e r 0 f t het w 0 en u mer a t ion sof i n d e~

cl i n a bIe sin SP. 1. 1. 37 s q q. a n d 1. 4, 56 s q q.
The two enumerations given in these passages closely resembIe classi~
fications, the one being based on etymology and the other on function. A
few changes in the formulation illustrate this:
A . classification accordin~ to etymology, 1. indeclinables the etymology
of which cannot be given. e.g. pra, ca, 2. denominative indeclinables
[with the inclusion of isolated noun~cases], 3. deverbative indeclinables,
4. compound indeclinables containing a noun~stem and an upasarga, [and
5. indeclinables derived from a senten ce and containing particles such as
yath and yvat]. and
B. classification according to function, 1. free or descriptive adverbs,
2, close adverbs changing the not ion of the verb, 3. karma~pravacaniyas
or originally close adverbs which, syntactically loosened from the verb, are
used (either as free adverbs or) in close construction with a noun~case,
4 . interjectionaI. emotional and syntactical~relational particles.
3. T h e cIa s s i f i cat ion 0 f wor d s i n t heB r h a d~

De vat (adhy. 1 sloka 39 c . d and 40 a b, sI. 42-45: adhy. 2 sI.
89-125).
In the Brhad~Devat four word~classes are distinguished, nouns
(nman), verbs (khyta), upasargas (without PI).ini' s functional distinc~
tion of some of them as gatis and karma~pravacaniyas) and niptas (in
the limited sen se of relational and emotional particles) . The upasargas and
niptas are not yet subordinated to a higher genus as in PI].ini, the
The gram mar of the indeclinables 9
denominative and deverbative indeclinables being considered as a form of

nman. Cf. Brhad~Devat, adhy. 1
sI. 43. a~tau yatra prayujyante
nnrthe~u v i b hak t a y a IJ
tan n m a kavayaIJ prhur
bhede v a c a n a ~ I i Ti gay 0 IJ.
sI. 45. kr i y bhinirvrtti~vasopajtalJ
k r d ~ anta~sabdbhihito yad syt
s a f!1 k h y ~ v i b hak t y ~ a v y a y a ~ I i ;1 g a ~ yukto
b h v a s. tad dra v yam ivopa/ak~yalJ.
Although in si. 43 only three aspects of the noun are mentioned, case
(vibhakti) , number (vacana) and gender (liTiga) , yet from si. 45, where
the action (kriy or bhva) denoted by a krt is taken as a thing (dravya) ,
i.e. where the verb in case of a krt has the character of a substantive, it
follows that a noun really has four aspects to be arranged thus: A.
declension, 1. case, 2. number, 3 gender, and B. transition into an inde~
clinable noun~form.
Although Occidental grammar (cf. Macdonell's Vedic Grammar) agrees
with the Brhad~Devat in calling infinitives and absolutives nominal verb~
forms, P~ini' s conception is evidently more correct.
4. Th e s t ra ~ li k e pas s a ges int h e svar~di~garya ( 1 I 1).

GP. 254 no. 40: vad~antam avyaya~saf!1jnaf!1 bhavati: brhmaryavat.
k~atriyavat, cf. SP. 5, I, 115-117; no. 89: tasil~dayas 1) taddhit edhc~
paryantlJ 2), sas~tasi S), krtvasuc suc 4), c~th/au 5), cvyarths 6) ca.
The two glosses paraphrase and extend SP. I, I, 38 ( 1 I 2). The
expres sion edhc~paryantIJ adds to the pronominal adverbs of P~ini (cf.
5, 3, 1 and 2) the adverbs derived from nouns and numerals; the separate
mentioning of c refers to 5, 4, 57 sqq., see footnotes I, 2 and 5; and the
cvyarths, ca lied gatis for syntactical reasons in the Stra~Ptha, are called
svar~dy~avyayas for etymological reasons in the Ga~a~Ptha.
GP. 254 nos. 72-73: k(tv)~(tos)un~k(as)unalJ. krn makra~saf!1dhy~
ak~arnto, 'vyayibhvaS ca repeat and paraphrase SP. I, I, 39-41 (cf.
1 I 3 sq.: deverbative and compound indeclinables).
5. Th e ins t anc e s su mme d u pin th e svar~di~gaT.la

(111).
The instances of GP. 254 fonn three main groups: A. 1-39, 41-45,
indeclinables with accent mentioned; B. 46-71, 74-88, 90-95, inde~
1) ~tas, SP. 5, 3, 7 sqq.

2) ~edhii, 5, 3, 46.
3) ~Sa.s, 5, 4, 42 sq.; ~tas, 5, 4, 44--49; tasi and tasil distinguished by accent:
vsudevats. kUtas.
4) ~ftrtvas 5, 4, 17; ~s, 5, 4, 18.
~) ~, 5, 3, 36-38 and 5, 4, 57~7; ~th 5, 3, 111.
6) I, 4, 61; 5, 4, 50; 7, 4, 26 and 5, 4, 52 sqq.
10 Studies on PI).ini's Grammar
cIinables with no accent mentioned; C. 96---150, instances not given by

"K." (the Pandit~edition of the Ksik, see Bhtlingk' s edition, p. 95*).
Within these groups there are sub~groups marked either by semasio~
logical coherence (e.g . 15-20 adverbs of time) or by similarity of ending
(e.g. 1-5) or by etymological coherence (e.g. 131-136, derivatives from
cira) .
Judging from the majority of cases the garya may be defined as an
illustrative and not exhaustive collection (krti~garya) of descriptive
adverbs, i.e. adverbs of place, rank and time, of manner, circumstance and
causality. On the other hand the first sub~group of the niptas, the cdi~
garya contains particles expressing conjunctionai, emotional and-' in~er~
jectional meanings. If we enter into the details we meet with severil diffi~
culties; in the first place why are the title~word svar and bhir and bhuvar
(nrs. 119 sq.) placed in the descriptive garya, and why are other instances
of sacral and interjectional language repeated in both the garyas: he, hai
(GP. 254 nos. 35 sq., GP. 85 nos. 120 sq.), bata (no. 41 and no. 93).
svadha (57; 41) , va~at (58; 39), om (118; 42)1 And why are the adverbs
in vat twice mentioned (G P . 254 no. 40 vad~antam, and G P. 85 no. 90
when compared with no. 89 iva) 1
Not eLSvar. - Svar, originally a noun , cf. Wackernagel 111 160 d,
p. 313, but as a sacral interjection 'may I obtain heaven!' (cf. Van der
Hoogt, Vedic Chant, p. 86 sub III c) no Jonger feIt as such.
Not e 11. E v i den tin ter pol a t ion sin t h e garya. - Saf!1~
vat (106, 'era beginning in 58 B.C:) and san (K. 41 bis, 'era reckoned
from 593 A .O : ) ilIustrate what little respect the Indian schoolmasters had
for their grammatical lore.
6. Denominative indeclinables (SP. I, 1,38, 1 12).
The suffixes of the cIass SP. I, I, 38 are considered by the SP. to be
at the same time vibhaktis (cf. the adhikra 5, 3, 1) and taddhitas (4, I,
76). GP. 254 in its stra~like passages explicitly extends the cIa ss to all
'dcnominative indeclinables', i.e. derivatives from pronouns, nouns and
numerals with all kinds of adverbial meanings.
7. Deverbative indeclinables (SP. I. 1, 39 sq., 1 13) .
Oeverbative indeclinables or infinitives and absolutives are fully dis~
cussed in the passage 3, 4, 9-66; cf. the analysis of the third adhyya
in 35. The krts ending in am have partly kit~form of root (SP. 3, 4, 12,
e.g. apalUpam and partly ryit~form (SP. 7, 2, 115 sq., and 3, 4, 12 and
22 sqq. bhjam, karam; cf. Oelbrck 229 and 225).
8. I n de cl i n a bIe s f 0 r m e d b y co m pos i t ion (1, I, 41;
1 14).
The principal passage dealing with the avyayibhvas (2, I, 5-21)
forms part of a set of two adhyyas containing the grammar of compounds
(adhy. 2 p. 1 and 2). The following cases are enumerated: indeclinables
formed with a following or preceding indeclinable of the .prdi~garya
The grammar of the indeclinables 11
(ak~apari 2, 1. 10; akumram 11-16); other indeclinables or isolated

noun~cases functioning as upasarjana (yatha, yvat, 7 sq. ), bahis and
words in ~ac 12; pre and madhye 18. In GP. 105 (SP. 2, I, 17) many
examples of compounds resembling an absolute case are registered:
ti~thadgu, 'when the cows stand to be milked' = af ter sunset; su~amam,
'when the season is favourable'; lna~yavam , 'af ter the barley has been
cut'. Similar are the indications of place and circumstance in SP. 2, I, 21
kr~1JagaTigam, 'wh ere the water of the Ganges is black', and in 19 sq.
trimuni 'when three munis meet' (in a proverbial expression ) ; sapta~
gangam 'wh ere the branches of the Ganges are seven' .
9. Case~isolation (GP. 254 nos. 17 sq. diva , rtrau etc.,

GP. 85 nos. 62 sq. velaym, mtrayam). 1 I 1 and 11 1.
An important means for the derivation of adverbs is the isolation of a
case from the rest of the declension ; although many examples based on th is
procedure are mentioned, the principle is nowhere explicitly laid down,
(Cf. SP. I , 1. 38 and 1, 4,83 sq. ).
10. In dec I i n a bIe 5 con 5 i der e d tob e nou i! 5 wit h

apo c 0 pat ede ase ~ end i n 9 a n d ne u ter 5 tem 5 , SP. 1. 2,
47; 2, 4, 18; 2,4, 82- 84 .
In su. 2, 4, 82, avyayd ap-supalJ (to be supplemented with luk 2, 4, 58)
we meet with the rule that af ter an indeclinable the feminine ending a and
the case~suffixes are rejected. Bhtlingk adds to this stra the explanation
'since the indeclinables are noun-stems (prtipadika) according to I, 2,
45 sq. one could be inclined to apply grammatical motion and declension
to them'.
This explanation is far from sufficient. Two details may be added: 1.
when a noun ends in the feminine en ding , the grammarians consider this
to be a contract ion of a pratipadika~vowel a and an additional suffix of
mot ion (SP. 4, 1. 4 sqq.); therefore when s. 2, 4, 82 says th at the
ending a is apocopated the reader must understand that is replaced by a ;
2. when a noun ends in i without a separate masculine stem besides it
(cf. on the other hand SP. 4, I, 5; 4, I , 15 etc.), this i is left untouched
as far as this stra is concerned.
These consequences of sutra 2, 4, 82 receive special importance 1. when
read in connexion with 2, 4, 83 (the final a of a pratipadika functioning
as last member in an avyayibhava is changed into am) and 2. wh en com~
bined with 2, 4, 18 (an avyayibhava is of the neuter gender) and I , 2, 47
(the end-vowel of a prtipadika is shortened in the neuter, e.g. n~ta -sri- ,
nom. sg. m. na~ta~srQ and n. na~ta~sri, Monier W illiams, grammar 190).
With the aid of the stras mentioned sub 1 the form madhyegangam
(SP. 2. I , 18) can be explained, whilst the avyayibhva 'adhistri' (Ksik
to SP. 2, I , 6) is derived according to the sutl'as mentioned under 2.
As Bhtlingk does not even add examples to the main case of 2, 4, S2,
abhyagni (2, I , 14) and trimuni (2, I , 19) may be here mentioned as sUl:h.
12 Studies on PJ].ini' s Grammar
I 1. T hes tra ~ 1 i k e pas s a gei n t h e cdi~g1J.a (G P. 85

no. 146; I II I).
Upasarga~vibhakti~svara~pratirpaksca nipt1;L, the niptas are of ten
externally identical with upasargas, declined nouns, conjugated verbs and
vowels (no. 50-58), but differ from them by their interjectional, either
emotional or imperative colouring .
. The passage, thus, is an elucidation of the ga1J.a, which originally ended
at no. 145.
12. The instances summed up (n the cdi~ga1J.a (GP.

85; 1 11 I) 1) .
Just as in the svardi~ga1J.a, sub~groups may here be distinguished; thus
nos. 1-5 and 13-18 are connected by similarity of function, cf. SP. 8, I.
24 and 30; nos. 37-42 form a group of sacral interjections; the group
59-58, (following the method, of the Sfva~sutra.<; with reference to quan~
tity), teach the interjectional use of vowels and diphthongs 2); nos. 75-80
all end in !inal 0 and nos. I 14-I 16 form an etymologically connected group.
Considering the collection in the light of the stra~like passage quoted
supra I 1 and taking account of the fact that in the Vedic chant pronouns
and upasargas are of ten used as sacral interjections (cf. Van der Hoogt,
p . 16), we must acknowledge that the ga1J.a, although overladen with
interpolations, still obeys the same principle, that is the common interest
in the different forms of emotional language and very likely the peculiar
word~order and intonation connected with conjunctional and modal
particles. Thus we can understand the mentioning of the upasargas ava
and anu (117 sq.) used imperatively, finite verbs as brhi (85) and satike
( 100), noun-cases expressing approval as satyam (106) and (tam (107),
pronomina I adverbs as iha (154) and iddh (108) used imperatively, and
descriptive adverbs as sukam (lOl , 'quickly!' cf. su GP. 254 no. 147). As
to peculiarities of word~order and intonatm, notice that satyam heads a
sentence, sanke is an intercalation; ha, aha (3 sq.) khalu, kila (44 sq.),
sma (49), u (60). tu, nu (86 sq.), kam (124). anga (I3 I) and vai (145)
follow the heading-word of the sentence, most of them being enclitic 3),
others again are often added to the verb (nos. 1-5, SP. 8, L 58).
13. Theprdaya1;L(GP.154,SP.I,4,58)in their different

f u n c t ion s.
(I). As member of a compound of which the other member is a noun .
- The instances of the prdi~ga1J.8, wh en compounded with a following
noun to make a tatpuru~a or bahuvrhi, or with a preceding or following
noun to make an avyayibhva, are not mentioned with a special name in
the section on compounds in general (2, 1, 3-2, 2, fin.), although in later
1) For older collections of niptas see Macdonell to Brhad-Devat 2, 93, vol. Il, p. 58.
2) The repetition of the vowel in instances as i lndra is eVidently a vocalic play.
a) Delbrck, p. 471, Hirt V, p. 332 and cf. SP. adhy. 8 p. 1.
sections the prdaya/:l in bahuvrhis are called upasargas (5, 4, 119; 6, 2,

177; and see 19). As to Iinguistic facts all the prdaya/:l are met with in
compositions with an accompanying noun, although some are seldom and
others frequently met with in th is function.
(2) . As 'close adverb' to a tiT\ or prefix to a krt ('gat :', SP. 1. 4, 60) .
- Most of the prdaya/:l are regularly construed as gatis. however dus
and su are not of ten met with in this function before finite verbs. e.g. dus
caranti (BRW. 111 703, Rmyal).a 3. 2, 25); the dictionaries moreover
mention: sunirvavau (MWO. 1220), su baddhv (BRW. VII 1025.
Mrcchakatik 176, 1 with the suffix ktv. yet no imitation of Vedic idiom.
SP. 7, I, 38). It appears from the rare use of these constructions that
these groups of finite verb with su and dus are analogy-formations af ter
compound ta-participles.
(3). As independent word (karma-pravacaniya) , SP. 1,4,83-98. 1
11 2 and 4. - As to the karma-pravacanyas Pl).ini and his commentators
mention:
the construction with a noun-case the use as 'free adverbs' for :

for :
GP. 154 no. SP. 1,4 s. GP. 154 no. SP. 1, 4 s.
5 anu 84-86 14 adhi 21 pari 93
22 upa 87 17 su 94
21 pari 3 apa 88 16 ati 95
12 89 15 api 96
20 prati 5 anu 21 pari 90 construction with noun-case
19 abhi 91 14 adhi 97 sq.
20 prati 92
Peculiar is the retrograde order of discus sion in the SP., when compared
with the GP.
Not e I. W i der u s e 0 f t h e karma-pravacaniyas i n t h e
Ve d i cid i 0 m. - The dictionaries moreover state the independent
use of ni (Atharva-Veda 10, 8, 7), nis (1.1. 6, 18, 3 etc.), pra (Aitareya-
Brhmana 2, 40) , vi (cum accus., RV.), su. 'good, excellent' in the Veda.
Not e 11. C 0 m par iso n 0 f th e u pas a r 9 a -I i s t sin t h e
G P. a n cl t heN i ruk t a (vol. 11 p. 41; 1, 1. 5 = 1. 3) 1). - The Gana-
Ptha enumerates the upasargas in the order 1. pra, 2. par, 3. apa, 4. sam,
5. anu, 6. ava, 7. nis, 8. nir, 9. dus, 10. dur, 11. vi, 12. , 13. ni, 14. adhi,
15. api, 16. ati, 17. su, 18. ud, 19. abhi, 20. prati, 21. pari, 22. upa and the
Nirukta in the order 1. , 2. pra, 3. par, 4. abhi, 5. prati, 6. ati, 7. su, 8. nir,
9. dur, 10. ni, 11. ava, 12. ud, 13. sam, 14. vi, 15. apa, 16. anu, 17. api,
18. upa, 19. pari, 20. adhi. Here it is wor th while noticing that GP. and
Nirukta of ten give the same 'pairs' of upasargas, thus GP. 21. 22 pari,
Studies on PI).ini' s Grammar
upa = N. 19, 18; GP. 19, 20 abhi, prati = N. 4, 5; GP. 16, 17 ati, su =
N. 6, 7; cf. SP. I, 4 su. 87 sq., 90 sq., 94 sq.; and moreover GP. I, 2
pra, par = N. 2, 3; GP. (7,8), (9, 10) nir, dur = N. 8, 9.
Indeed there must have been much activity on the side of teachers and
pupiIs about these upasargas even since Vedic times!
14. Th e gatis en u mer a t e d SP. 1. 4, 61-79 a n d th e
gatigaT)as referred to (GP. 31 and 243; 1113).
The stras 61-72 contain besides a few neuter words (sat. asat, manas,
adas) as a rule indeclinables; many of the in stances mentioned by GP. 243
look like isolated noun-cases, and the same holds good for stra 75 sqq.;
generally speaking there is a tendency to group according to similarity
of ending.
15-20. The technical terms of the grammar of indeclinables as used

in the A~tdhyyi according to Bhtlingk' s Index (vol. 11
p . 193 sqq.).
15. Th e ter m avyaya, 1 land II.
Scarcely any rule for the avyayas as a genera} class is given in PI).ini's
grammar; the term is either used to distinguish an indeclinable from a
homonymous form of a declinable, e.g. 1,4,67 puro 'vyayam, or to denote
by implied restriction or suggestion from context a species, e.g . an upasarga
and moreover the nipta iti 2, I, 6, the absolutive in tv 2, 2, 11, the
absolutive in am 2, 2, 20, an upasarga 2, 2, 25, an absolutive 2, 3, 69, a
pronominal adverb 4, 2, 104, a descriptive adverb of time 4, 3, 23 etc.
16. T het e r m nipta, 1 II.

According to Bhtlingk's index the term nipta only occurs in a few
passages. In SP. I, I, 14 nipta ekc refers to the old particIe u (Uil GP. 85,
no. 60) and formallyto GP. 85, no. 37-42 (supra 12). In SP. 3, 3, 4
the conjunctions yvat and pur 1) (GP. 85, nos. 32 and 68) are called
niptas in order to distinguish them from the pronomina I and nominal case-
forms. In SP. 6, 3, 136 the term seems to have a vaguer meaning. In
SP. 8, I, 30 conjunctional particles are summed up under the name nipta,
and in fact the notion nipta is the subject of several passages in adhy. 8
p. 1 (s.30-50; 56-57; 58-65) . In all these instances, with the excep-
tion perhaps of 6, 3, 136, the term nipta bears on syntactical and modal
particles ( 2, B 4).
17. Th e ter m s upasarga, gati a n d karma-pravacanya.

( 1 ) . For the upasargas as members of a compound of which the other
member is a noun the term upasarga is met with in the fifth adhyya, s.
5, 4, 85 and 119, both belonging to the samasnta-section. In the adhyyas
6 and 7, which describe how word-components (roots, stems, prefixes,
1) Par in Bhtlingk's edition p. 113* is amisprint.

suffixes) coalesce into words, the upasargas in their adnominal function

within a compound are twice mentioned, and then with the term upasarga
6,2, 177; 6, 3,97 (a rather mongrel stra!). In adhy. 2 p. 1 and 2, on
the other hand, they appear under the general name of avyaya in 2, I, 6
(derivation of avyayibhvas) and 2, 2, 25 (numerical bahuvrihis of the
type upadasa) and under the wide term prdayalJ. in 2, 2, 18, (that
prdayah here means adnominal upasarga results from its coordination
with the term gati; cf. the examples in the Ksik).
(2a) . As prefixes to a krt the prdayalJ. are of ten referred to in the long
section 3, 1. 91-3, 3, 130, which, with the exception of a few prsar"lgika-
stras, treats of the derivation of declinable krts. Here we only meet with
the term upasarga (3, 1. 136; 3, 2, 61. 99. 147; 3, 3, 22. 59. 92.106), placed
in the locative in accordance with SP. 3, 1. 92. Likewise in 6, 2, 110 the
prefix to a krt, in this case a n4thnta, is termed an upasarga.
(2b). As 'close adverbs' the upasargas are often mentioned (1. 3, 76;
2,3,59, which stra in its construction much resembles that of the upapada-
section just mentioned; 6, 1,91 sq.; 7, 4, 23; 8, 1. 38 etc.).
(2e) . Only in a small number of passages (see Bhtlingk's index
p. 225*) is the term gati met with and exclusively in rules evidently formu-
lated for all the gatis of I, 4, 60-82.
(3). When used in the function of 2, B 3, the prdayalJ. arE'
exclusively called karma-pravaeaniya.
Not e. Th e 9 ram mar 0 f th e niptas a n d gatis in adhy. 8
p. 1. - Adhy. 8 p. 1 discusses the sentence-accent and in this connexion
gives rules for both the niptas (conjunctional particles) and the gatis
(with the upasargas as a sub-class).
18-20. Appreciation of PIfini's terminology.
18. T het e r m nipta i n i t s w i d e a n dna r r 0 w s i 9 n i-

f i c anc e.
Except in s. 6, 3, 136, which, followed by the wide rule of su. 137,
has no weight in the matter 1), the term nipta is in fact only used for
modal and conjunctional particles (cf. 2, B 4), a narrow significance
which agrees with the B~had-Devat and the Nirukta (vol. 11 p. 44-83;
1 p. 2 and 3 = I, 4-11) and is likewise consistent with SP. I, 4, I,
kacJrd ek saf!1jn, read in connexion with I, 4, 58 sqq. (upasarglJ.,
gatilJ.) and 83 (karma-pravaeaniylJ.). On the other hand this restricted
meaning seems to be contradicted by I, I, 37 together with I, 4, 56, prg
r-isvarn nipt1;r.
We can thus distinguish (1) a wide significance which includes all
the indeclinables with a relational meaning (conjunctions, modal adverbs,
close adverbs to finite verbs, prefixes to deverbative nouns and indeclin-
1) Moreover, S. 6, 3, 131-137 is most likely an interpolation.

16 Studies on P1).ini's Grammar
ables, close adverbs which have developed into free adverbs, prepositions )
and (2) a narrow significanee in which the term only denotes the conjunc~
tions and modal adverbs. The wide significance is only met with in I, I,
37, and in I, 4, 56 in a certain way, that is mentally, before having sub~
str<lcted (cf. s. 1. 4, 1) from th is wide significanee the special notions
mentioned in I, 4,58 ('pradayal)'), 60 (gati) and 83 (karmapravacanya).
This distinction of a wide and narrow significance appears to be a kind of
logical artfulness chosen for the sake of brevity.
Ktyyana (MBh. I, 4, 58-59 vrtt. 1-3) and all the commentators
tried to improve upon the text by dividing the stra which originally ran
'pradaya upasargl) kriy~yoge' into two successive stras, a reading which
penetrated into the tradition of the mla. In this reading the term nipta in
its narrow sense (2) embraces the modal and conjunctional particles of
s. 57 (GP. 85) as weIl as the prdayal) of s. 58.
19. Ambiguity of the definition of upasarga (1,4,

58 sq.).
In agreement with the Brhad~Devat 1) Pl)ini (1. 4, 58 sq.) defines
the upasargas by the aid of the formula kriy~yoge 'in connexion with an
action', that is in the sense of the Brhad~Devat, (1) 'in connexion with
[the notion of] an action [either expressed or imp lied ] '. However kriy~
yoge mayalso be interpreted as (2) 'in connex ion with [a word expres~
sing] an action, [i.e. either a finite verb or . a deverbative],
The first definition is the wider of the two, for it includes the bahuvrihis,
for instance (MBh, I p. 342 1. 12):
vigatl) 2) secak asmd grmd: visecako 3) grmal),
pragata nyak asmd grmt: pranyako grmal),
According to the second definition the upasargas are exclusively the set
of twenty particles enumerated GP. 154 when used as gati!;. that is either
as close adverbs to finite verbs or as prefixes to deverbative nouns or
indeclinables; e.g. pra in pra 1J.ayati and the tatpurWfa derived from it:
pra1J.yaka (with cerebralisation of n according to SP. 8, 4. 14).
From the use of the term upasarga in 5, 4, 119; 6. 2, 177 and 6, 3, 97 it
follows that P1).ini understands the expression kriy~yoge in its wider
sense (1).
1) The B!,had~Devat, which does not distinguish between upasargas and karma~
pravaCan!!8S, defines (2, 94):
Upasargs tu vijiieyl}
kriy~yogena viT[lsatil},
vivecayanti te hy art1uurz
nmkhyatB~vibhBkti~u. .
2) vi = vigafs; compare the predicative and attributive use of some close adverbs in
English: my father is out, he is well~ff, an u~train.
3) Wlth s; SP. 8, 3, 65 bears according to P~ini's intention on finite verbs and
deverbative indeclinables; on the other hand 8, 3, 99 bears on nominal composition as is
interpreted by the Ksik and proved by the ablative nak~tr!d in s. 100.
The grammar of the incleclinables 17
Not e I. Th e ce r e bra 1 i s a ti 0 n 0 f r.a t th e beg i nni n 9 0 f

th e sec 0 n cl mem b e r 0 f a co m p 0 u n cl. - The cerebralisation of
the n in the bahuvrihi praryasa (5,4, 119 Ksik) follows SP. 8, 4,3, where
saT!ljfi specially means 'nickname' , cf. Wackernagel I 170 b, p. 190.
Not e 11. T her e I a t ion bet wee n t hen 0 t ion s upasarga
a n cl gati. - Although su. 1, 4, 60 gatis ca reminds us in its form of
2, 1, 23 dvigus ca, yet these two stras are opposite in construction. The
Ksik interprets 1, 4, 60 by gati-saT!ljfiaks ca prdayo bhavanti kriy-
yoge and 2, 1, 23 by dvigus ca samsas tatpuru~a-saT!ljfio bhavati, i.e. the
gatis are a genus which includes the prdayal) as a species, but the dvigus
are a species subordinate to the tatpuru~as as a genus. However, since
according to the foregoing pages su. 1, 4, 58 and 59 have to be taken as
one stra and kriyyoge to be interpreted in its wicler sense, the inter-
pretation as given by the Ksik likewise must be so changed that
upasargas ancl gatis become crossing notions:
first member of bahuvrihi
gati or nominal tatpuru~a
gatis summed up SP. 1,4,61-98 upasargas any other word
20. T het e r m karma-pravacaniya.

The term karma-pravacaniya (1, 4, 83) is commented upon in the MBh.
as follows: ' ... karma proktavantal) karma-pravacaniy iti 1). Ke punal)
karma proktavantal)? - ye saT!lprati kriyT!l nhul). Ke ca saT!lprati nhul)?
- ye 'prayujyamnasya kriym hus, te karma-pravacaniyl)'. Thus
according to this interpretation karman = kriy and pravacaniya = 'sug-
gesting' 2), 'not directly mentioning', and consequently karma-pravacaniya
= 'denoting an act ion of which [the kartar] is not expressed', which
means that any karma-pravacaniya can be paraphrased or explained with
the help of a verb.
The term calls forth two remarks. In the first place the definition kriy-
yage and the term karma-pravacaniya express almost the same thing!
Secondly the term karman has here a meaning different from the usual
one in PJ?ini, namely 'that which is expressed by the termination of the
(personal) passive verb' or 'the grammatical object of a transitive verb',
whereas the meaning of 'action' is only met with in the other technical
term karma-dhraya = 'owing a karman to acreditor' = not prosessing
an action = a compound of two members not related to each other by an
intermediate act ion = a compound not based on a kraka-relation (SP. 1,
4 , 23) [nor on a ~a~thi-relation, SP. 2, 2, 8]. but on samndhikaranatva
(SP. 1. 2, 42 and 2, 1,49) .
It is likely th at influences of older schools betray themselves in these
inconsistencies, for terms survive the changes in scientific thought both in
modern Europe and in ancient India.
1) On the active meaning of pravacaniya cf. SP. 3, 4, 68.
2) Cf.pra + V vac = to betray, BRW. VI 624 s.v. 1) in fine.
2
18 Studies on Pnini's Grammar
Il. KARMA-PRAVACANiYA AND THE THEORY OF THE NOUN-CASES.
21-28. The grammar of the krakas. noun-cases and karma-prava-

caniyas. The coherence of the adhyyas 1 and 2.
21. Course of the discussion.
The grammar of the karma-pravacaniyas construed with a noun-case
demands an examination of PI).ini' s treatment of the noun-cases in the
two passages I, 4, 23-55 (the not ion kraka) and 2, 3, I-fin. (the use of
sup or the noun-cases), 22-26, whkh finally leads up to a comparison
of the European notion of preposition and the Hindu terms karma-prava-
caniya and dik-chabda, 27.
22. T hen 0 t ion 0 f kraka (f act 0 r 0 fan act ion),
SP. 1. 4, 23-55.
By krakas PI).ini understands the logical or ideational relations
between a noun and a verb. or more precisely between an object or anything
conceived af ter the analogy of an object and an action or anything con-
ceived af ter the analogy of an action. Six or seven such relations are
distinguished; (a) the apdna (1. 4, 24) the object from where the
movement takes place. or whatever may be conceived by linguistic instinct
as a metaphor of such arelation; (b) the sa1!1pradna (32) the object or
person whom one has in view (abhipraiti) during the act; (c) the kara1)a
(42) that which in causal respect is the most necessary means (sdha-
katama) for carrying out the act; (d) the adhikara1)a (45) the reg ion, the place
or time where and when the act takes place; (e) the karman (49) the object
which is aimed at, sought for (ipsita) by the act. or whatever in linguistic
respect is conjoined to a verb in a similar way; (f) the kartar (54) the
agent who is the centre or executor of the act [whereas the instrument is
dependent on the executor]; (g) the hetu (55) the auctor intellectualis of
an act [whilst the executor in this case is only partly responsible].
EVidently PI).ini tries in th is analysis to separate the ideational aspect
from the linguistic expression. an attempt which the Occidental linguists
of the latter half of the nineteenth century have condemned, misled as they
were by the hope of being able to understand language through the
exclusive study of its phonal and morphological aspect, i.e. its articulative
utterance and the association-system underlying declension and conjuga-
tion, as if the application and imitation of physics and a mechanistic
psychology were the last word of moral science. And so besides the
injustice done to a pioneer of grammar who lived about twenty-five cen-
turies ago by -- ~ssociating with him supposed results of modern grammar,
it is even questionable whether PI).ini has not something still to say to
us 1). As to the notion of the grammatical object one may distinguish
between the causal relation in acts of producing. annihilating or changing
1) Whltney' s unfavourab1e judgment (On recent studies in Hindu grammar 1893.
p. 171 sq.) moreover rests on ao insufficient understaodlog of the notions kraka aod
upapada ( 25).
an object, and the 'intentional' 1) relation in the psychical acts of wishing,

seeing, hearing, fearing etc. In PI).ini's definition the 'intentional' relation
is considered to be primary, which at all events is an interesting conception.
23. T h e kl'aka ~ t h e 0 r y a n d t h e u s e 0 f t hen 0 u n~
cases (SP. 1. .4,23-55 and 2, 3, l~fin.).
When making the kraka~theory the basis of the case~syntax PI).ini
attributes the apdna to the ablative or 5th case (2, 3, 28), the saT!lpra~
dna to the 4th case (13), the karaTJa to the 3d (18). the adhikaraTJ.a to the
7th (36). the karman to the 2d case (2) 'uniess . the karman is already
expressed otherwise' (anabhihite. supplemented from su. 1). i.e. unless
expressed by the termination of a passive verb, and the kartar to the 3d
case or instrumental (su. 18) 'unless the kartar is already expressed other~
wise' (s. 1). i.e. unless the active person or thing is expressed by the
termination of a not~passive and not~impersonal verb.
From the analysis given up to th is point. two conclusions can be drawn.
In the Brst place the nominative and genitive are only indirectly connected
with this kraka~theory (see 2.3.46 and 50. and 25 sq.). Secondly the
order in the two passages 1. 4. 23-55 and 2. 3. l~fin . differs much. for
when attributing to the krakas the cipher of the cases which generally
correspond to each of them and to the kartar the letter q as index. the
order of the kraka~passage appears to be 5. 4. 3. 7. 2. q and the order of
the sup~passage 2. 4, 3. 5, 7. 1,6 (see the stras 2. 13. 18 etc. of 2 p. 3).
The seconQ arrangement is in the main based on the morphological order
( 1. 2. 3 etc~) with three permutations. the postposition of 1 and 6 because
of their standing aloof from the krakas and the interchange of 3 and 4
owing to mnemotechnical reasons, s. 12 mentioning verbs which allow a
dative or an accusative and su. 14 explaining how a dative sometimes
expresses the karman or ipsita (1. 4. 49). namely when the expressed
verbal notion mentioning the auxiliary act 'defines' or 'circumlimits' the
implied verbal notion of the act aimed at 2): for instanee in the sentence
edhebhyo vrajati = edhn hartuT!l vrajati. the expressed notion of the
subsidiary act of going is an upapada to the implied notion of the bringing.
which latter act has the fuel as its karman. Thus the 2d and 4th cases are
brought into close conlection.
Not e. Bhva e x pre s s e d b y t h e vibhakti 0 f a ver b. -
Whereas (1) the termination of the active or middle voice of a causative
expresses specially the hetu (1. 4. 55). (2) the active or middle voice of a
non~causative verb the kartar (54) and (3) the passive form of a transitive
verb the karman (49) S). on the other hand the vibhakti of the passive
voice of an intransitive verb expresses the mere occurrenee or bhva (SP.
L 3. 13 and the examples given by the Ksik: g~yate bhavat, supyate
bhavat. syate bhavat).
1) The term Is borrowed from Brentano's psychology. see Binswanger's 'Probleme'.
t) Cf. f 34.
a) BhtUngk's addition: "der Begriff des Accusativs" Is misleading.
20 Studies on Pl).ini' s Grammar
24. S y nop tic a I tab I e 0 f th e di f f ere n t ar r a n 9 e~

men tso f nou n ~ cas esa n d krakas.
A. Mor p hol 0 9 i c a lor der :

1. Sup~sitra, 4, 1. 2 1234567
2. Construction of the karma~pravacaniyas , I , 4, 83-
98; 2, 3, 8-11 according to its original concep~
tion, cf. 27 257
3. Kraka~relations and $a~thi~relation in tatpuru~as,
2, I , 24-48; 2, 2, 8 234576
4. Syntaxis of noun~cases, 2, 3, l~fin . , cf. 23 2435716
5. Secundary use of the genitive 2, 3, 51---68 (3) 2 4 3 7 q (2)
6 . Kraka~relation in a compound with final am~
absolutive 3, 4, 25-46 237q
B. 0 r der 0 f kraka ~ t h e 0 r y :
1. Definition of the krakas. 1. 4, 23-55 54372q
2. Formal noun~case as first member of a compound,
6, 3, 1-24 5 3 (4) 7 6
C. Uncertain principle:
1. A noun ~caseas free adjunction to am~absolutive 3,
4,47-58 3 (7) 5 2
2. Kraka~relations in tatpuru~as (and the accentuation
of these compounds), 6, 2, 2 372
3. Kraka~relations in tatpuru~as with ta~participle as
second member, 6, 2, 45-48 423
Not e I. A nno t a t ion s tot hes pee i a 1 cas e s. - The
place given to the 1st and 6th cases at the end of the series in A 3 and A 4
is due to the particular syntactical function of these cases ( 23). Further
many deviation from the regular order can be explained by mnemotechnical
motives : (A 4) anuvrtti of bahulaf'!'l chandasi from 2. 3. 62 into su. 63; (B 2)
anuvrtti of tmanalJ from 6, 3, 6 into s. 7; (C 1) anuvrtti of trtiyym
from 3, 4, 47 into 49-51, and of paripsym of s. 52 into 53; (C 3)
anuvrtti of caturthi from 6, 2, 43 into 45 sq.
Not e 11. T h e cIos ere I a t ion s hip bet wee n th e 3 d
a n d 7 t hno u n ~ cas e. - As in the passage B 2 the stras 6, 3.
7 sq. form an associational (prsangika) digression to su. 6. the fundamental
scheme of this section really con ta ins the trtiy and saptami in immediate
succession as found in BI ; in C 1 the two cases appear closely related by
the fact that the saptami is here an alternative substitute for the trtiy ; and
above all C 2 (su. 6. 2. 2) is of special interest both by its contents and its
.redaction: tatpuru~e tulyrtha~trtiy~saptamy~upamnvyaya~dvity~krtylJ.
Not e 111. T her e I a t ion bet wee n t h e kraka ~ a r r a n 9 e~
men t a n d t h e sup ~ ser i e s. - The results arrived at in note 11 allow
the following conclusion: the kraka~arrangement is nothing but the
retrograde form of the sup~series with removal of the saptami af ter the
trtiy because of the close relationship feIt by linguistic instinct between
these two noun~cases.
Not e IV. 0 t her f act s s h 0 win 9 t h e c los ere I a t i 0 n~
s hip bet wee n th e karaT)a a n d adhikaraT)a. - In the section 4, I,
92-4, 2, 91 the author gives the first list of meanings of taddhita~suffixes.
Here we read: tasya (apatyam) 4, I, 92-4, 1, fin., tena 4 p. 2, 1-13,
ta tra (referring to place) 14-20, asmin (referring to time) 21-23, asya
24-35, tasya (samiihah), 37-51, (tasya) 52-54, so 'sya SS, tad asym
57, ssym 58, tad (in accusative~relation) 59, tad asminn asti 67, tena
nirvrttam 68, tasya nivsalJ 69, adrabhavas ca 70 (the last four siitras
bearing on the formation of nouns of localities ). The striking tendency of
grammatical theory to expre$ the relation between the original and
derived nouns with the aid of the instrumental and locative cannot be
denied. An analysis of similar passages would give the sameresult. Finally
we may add that in stra 3, 3, 117 the suffix lyut is said to express both
the karaT)a and the adhikaraT)a.
Not e V. At tem p t t 0 ex p I a i n t h e cIos ere I a t i 0 n~
s hip bet wee n karaT)a a n d adhikaraTJa. Int e r pre t a t ion 0 f
s . 6, 2. 2. - The first fact of importance for the explanation of the close
relationship between the karaT)a and adhikaraTJa is the . semasiological
grouping noticeable in the noun~cases : 1. the predominantly adnominal
case: the genitive, 2. the so~called grammatical cases, i.e. the cases which
intrinsically demand from speaker and hearer the 'psychical realisation' of
the act, the nominative, the dative (abhiprya, the psychical or social effect
of the act) and the accusative (the immediate purport of the act, i.e. either
the object of causation or of 'intentionaJ' relation), 3. the cases which
intrinsically express perceptional relations, the ablative ex pressing the point
from where, the locative ex pressing the spot wh ere, and the instrumental
expressing the means by which the act takes place. In old Greek and
German only the first two groups have survived, in Latin moreover the
last group was kept as one case~form.
Evidently the Hindu grammarians have not been able to discover from
their own language this grouping as it now appears to us af ter the deve;.
lopment of European languages, yet the relationship between the karaT)a
and adhikaraTJa they have been aware of. In order to interpret this
interesting fact, s. 6, 2, 2 demands a closer examination.
For interpreting this stra one must read it in connexion with its
adhikra 'prakrty piirvapadam' and the later stras which limit its sphere
of applicability (such as 6, 2, 139 ' ... krt'. cf. moreover su. 144) . From
the siitras mentioned it follows that s. 6, 2, 2 only refers to tatpuru~as of
which the second member is a noun as such, i.e. a noun which is not of
deverbative nature or at least not feit as such. This noun may be either a
substantive or an adjective. In section 2, I, 24-48 this .kind of tatpuru~a
is treated in general and in 6, 3, 2-20 with reference to the expressed
22 Studies on P~ini' s Grammar
case~form of the first member 1). Af ter reading these sections it becomes
clear that substantives and adjectives are of ten compounded with a noun
in the relation of karaTJa or adhikaraTJa (2, 1 s. 30 sq., 34 sq., 40, 42, 44);
but never or scarcely ever in the relation of apdna (2, 1. 37). So no
blame may be attached to the Hindu grammarians for their emphasising
the semasiological coherence between karaTJa and adhikaraTJa.
25. T h ede fin i t ion 0 f t hen 0 min a t i v e a n d t h e

int e r pre t a t ion 0 f SP. 2, 3, 46.
The translation of the siltra runs: the nominative merely expresses (a)
the significance of the nominal stem, (b) the grammatical gender, (c) an
amount, (d) the [grammatical or the explicitly described] number.
The main idea of the stra is as follows: the nominative expresses only
that which is common to all the declinational vibhaktis. but that which
belongs to each of the other noun~cases separately, namely a kraka~
relation, is missing in the function of the nominative; this noun~case is
simply an upapada, a kind of apposition to that which is expressed by the
conjugational vibhakti of the verb; for in stance in a sentence such as kaviIJ
kvyaf!l karoti. kaviIJ is merely an upapadaof the notion of the kartar. the
pratham singular, of the active verb; and in a sentence such as kvyaf!l
kriyate. kvyam is an upapada of the karman expressed by the passive verb
(cf. 23 Note and 34).
However, the wording of s. 2, 3, 46 demands a closer examination.
A nominative is said to express (a) the meaning of the nominal stem, (b)
the gender and (d a) the grammatical number. Thus in the nominatives
rjiii and rjiiyaIJ the meaning of the prtipadika rjan (cf. Bhtlingk's
annotation to s. 4, 1. 1), of the feminine gender and of the singular or
plu ral number is expressed.
Moreover the nominative like any other case can denote an amount or
explicitly described number, e.g. dt'OTJo vrihiIJ 2). vif!lSatir virIJ.
26. T h e cl e fin i t ion 0 f t heg e n i t i v e a n d t hei n t e r~

pre t a t ion 0 f s. 2, 3. 50.
The su. ~a~thi se~e, 'the genitive is usecl in the remaining case( s)' is
explained by Pataiijali 3) as 'karmdibhyo 4) yo 'nyo 'rthaIJ sa se~aIJ' i.e.
another ideational relation than the six krakas. Thus the primary sphere
of the genitive is its aclnominal function such as the genitive of the
possessor or possession etc . and only in its secundary function it expresses
different krakas.
1) For the t8tpuru~ with ~J..relation. see 2, 2, 8 sqq. and 6, 3, 21 sqq.

2) Bhtlingk's recension of Speyer's Syntax p. 179 sq.
I) MBh. I p. 463 1. 9.
) Patanjali begins the 1i!ralca~series with the karmanl cf. 24 A.
27. Comparison of the passage 1. 4. 83-98 (the

me a n in 9 sof th e karma~pravacaniyas) a n d 2. 3. 8-11 (n 0 u n~
cas e s wit h w h i c hso m e 0 f t h e karma~pravacaniyas a i e
construed).
Of these two passages the second in its arrangement is based on the
first. In order to prove this. it is necessary to examine the composition
of the total section adhy. 2 p. 3 (the meanings of the declinational
vibhaktis). Here the different noun~cases are first defined with the aid
of the kraka to which they belong in principle ( 23) and af ter that
(a) the other syntactical relations in which the noun~case is used are
enumerated and subsequently (b) noun~cases which are their substitutes
in a certain meaning are mentioned; for in stance in the paGsage 28-35
explaining .the ablative s. 28 defines the primary kraka of the noun~case.
su. 29 gives another sp here of it. whilst 30-32 and 34 give the competitive
noun~cases . and 33 and 35 are prs8;Tigika rules.
In the passage 2, 3, 2-12 dealing with the accusative s. 2 gives the
primary kraka~meaning, the karmatva; sutras 4, 5, 8 and partly 12 add
further meanings, and stras fr-7, 9-11 and partly 12 give the competi~
tive noun~cases. Of these sutras the 8th has interest for the theory of the
karma~pravacaniyas as laying down the rule that the karma~pravacaniya is
in general construed with the 2d case; su. 9 attributes the 7th case to
karma~pravacaniyas expressing the notions adhika and isvara, su. 10 sq.
attribute the 5th case to apa. and pari and to all karma~pravacaniyas
expressing pratinidhi and pratidna.
Consulting now the karma~pravacaniya~passage (L 4, 83 sqq.) the series
is seen to begin with such instance.s as require the 2d noun~case (84-87),
then follows upa in the meaning adhika with the 7th case (s. 87 partly).
th en apa, pari, with the 5th (88 sq.). and after a digression (90 sq.)
about karma~pravacaniyas construed with the accusative according to the
general ruk s. 92 mentions prati with the 5th case in the meaning of
pratinidhi and pratidna. The one exception in the correspondence of order
is that adhi (isvare) construed with the 7th case comes last in the karma~
pravacaniya~passage (s. 97) whereas in the section on the vibhaktis it is
joined with upa, likewise requiring the 7th case (yasmd adhikaf!l yasya
ceSvara~vacanaf!l tatra saptami, 2, 3, 9).
Whilst the general dependence of the second passage (2, 3, 8-11) on
the first (1, 4, 83-98) is evident, the deviations are likewise easily under~
stood. The first and also, in the plan of the composition, most origfnal
passage complies both with the old prdi~gaTJa ( 13 no. 3) and the
morphological order of tl?e noun~cases (2-5-7) only su. L 4, 8'1 to~ming
an exception for the sake of mnemotechnical economy. And it is this one
exception again which caused the two cases of the saptami to be joined and
together placed between the dvitiy and pancami in s. 2, 3, 8-11.
24 Studies on PJ}.ini' s Grammar
28. T h e E u rop e a nno t ion 0 f pre pos i t ion c 0 m

par e d wit h t h e karma-pravacaniya a n d dikchabda 0 f H i n d u
9 ram mar; t h e (e x ten d e cl) a d n 0 min a I f u n c t ion 0 f t h e
gen i t i v e ( cf. Speyer' s Syntax, p. 113-134 and specially p. 114 note 1 ) .
Speyer, who describes the Sanskrit language as a European gram.
marian, makes the following rem ark on the notion of preposition with
reference to Sanskrit (p. 114 note 1): 'it is wrong to say that the noun-
case attending on the preposition, is governed by it, for it is not the
preposition, that causes the case, but it is the general bearing of the case,
which is qualified and limited by the preposition'. An in stance like antar
vesmani (Speyer 165) illustrates his conception; here antalJ is addecl as
an upapada to a locative for the purport of specifying the meanll'l.g 'in' or
'at' vaguely expressed by the vibhakti of the noun. Of similar nature is the
use of saha where a single instrumental suffices (1.1. 58 sq., e.g.
RmyaJ}.a 2, 68, 2, asau mtula-ku/e... bharato vasati bhrtr satru
ghnena).
Thus the ,construction of a nouncase with an indeclinable is by origin
and in principle not a reasonless mechanical process of association, but the
noun-case is as a rule chosen by linguistic instinct in agreement with the
general case-meaning, the applicability of which is determined by the
meaning of the avyaya added to the notion of the verb 1), and this holds
good although in some cases an irrational association has its influence, e.g.
in vin with instrumental owing to the contrasting saha. Thus the accu-
sative comprises the meaning of the local aim, the object or spot towards
which something is moving (prati with accusative). In the case of anu
(af ter) the aim, the ipsita, although itself moving, still directs the move-
ment of the following person or object. Similarly, as the accusative
expresses 'what is taken hold of; what space is occupied' (1.1. 54) inde-
clinables meaning ,along' (anu), 'through' (tiras) are construed with the
dvitiy (cf. similar constructions in Greek) . In accordance with this prin-
ciple vi (in Vedic = between) and the dik-chabda antar (= between,
SP. 2, 3, 4) attend on the accusative, not the position wh ere, but the extent
over which being the original meaning. Similarly uttarerya (SP. 5, 3, 35
and 2, 3, 31) 'by the northern way which stretch es along' and parallel
derivatives are attended by the accusative.
However in two cases this principle according to which the preposition
is merely an upapada of the noun-case demands correction, in as far as the
construction is of a different origin, namely (a) in those cases wh ere the
noun construed with the preposition 2) expresses a starting point for our
mental imagination, a fixed centre from which the spacial situation of
something is clescribed, and (b) where as in the Latin 'ubi terrarum' the
1) Cf. the quotation from Delbruck at the end of this section.

2) Here the preposition resembles such deverbative indeclinables as mentioned by
Speyer 200 sqq.
genitive construed with an indeclinable shows a metaphorically extended

use of the partitive adnominal 'g enitive. Examples of case a are the con~
struction of the dik~chabdas in ~ac and ~ (SP. 2. 3. 29; 5. 3. 30; 5. 3.
36-38) with the paficami. of case b all those dik~chabdas (S P. 2. 3. 30
and 5. 3. 27!-41) describing a relative position of one object. towards an
other. e.g. upari (above) with the ~a~thi 1) (SP. 5. 3. 31 and 2. 3. 30) .
adhas (under) 5. 3. 39. puras (before). etc. And for temporal undecli~
nables with partitive genitive see Oelbrck 112.
The facts quoted explain why the Indian grammarians never needed the
notion of 'preposition' for the description of their language; above all they
have realised the close relation between gati and karma~pravacaniya. a
relation weil defined by Oelbrck (p. 432): 'Was die Verbindung mit
Casus angeht. so lsst sich aus dem Sanskrit besonders gut nachweisen.
dass der Casus ursprnglich zu der mit der Prposition innerlich ver~
bun denen Verbalform trat. und dass erst allmhlich sich ein engeres Ver~
hltniss zwischen Casus und Prposition entwickelte. Der Ausdruck. dass
die Prposition den Casus regiere.ist ... hchstens bei (bis) und pur
(puras) 'vor' in ihrer Verbindung mit dem Ablativ angebracht.'
Not e I. A d j e c t i v e sof f a r nes s a n d n e a r nes s c 0 n~
strued with ablative or genitive (SP. 2. 3. 34).-
Instructive for the cases a and b (ablative of the mental starting point and
the extended use of the adnominal genitive) are the adjectives of
farness and nearness in their constructions with ablative or genitive. To
the few ex am pies mentioned by Speyer ( 98) the following may be added:
(Cappeller. Blamagha 3. 10. mfirteva sakti/:t kvacid askhalanti. nitya'!l
Hare/:t sa'!lnihit nikma'!l Kaumodaki.. . the disc Kaumodaki. 'Resembling~
the~Moonlustre . never failing as it is a palpitabie representation of 5akti 2).
[Vi1?nu's Power] and [by innate magic] always near to Hari whenever he
wishes ....
Not e 11. 0 el b r c k 's h y pot hes i s i 11 u s tra t e d b y th e
op pos i t e pro ces sin m 0 der n Eng I i s h. - Using the Sanskrit
terms we may say th at in a sentence such as 'he is laughed at' a karma~
pravacaniya or indeclinable attending on a noun has developed into a gati
or close adverb which changes the notion of a verb.
111. GATI. OR INDECLINABLE USED AS A PREFIX OF A

DEVERBATIVE NOUN OR INDECLINABLE .
29-33. The adhikras of the ku~gati~passage (2. 2. 18-22).

34-37. The notion upapada and the 3d adhyya.
38-39. Bhtlingk's translation of the ku~gati~Pi'1ssage.
1) Thus upari grhasya shows affinity to a construction such as grhasya pr~!he.

2) Farquhar. index s.v.
26 Studies on PJ}.ini's Grammar
40--46.'Bridge'~character of the ku~gati~passage and the pada~bha~

passage. Logical elegance of PJ}.ini's composition. Inter~
pretation of s. 2, 2, 18.
47-49. The cross~references between the ku~gati~passage and the 3d
adhyya. S. 2, 2, 19-22 interpreted.
50-56. The Mahbh~ya on the ku~gati~passage.
57-60. Morphological and phonological consequences of the ku~gati~
passage.
29. T h e a d hik ras 0 f the ku~gati ~ pas s a 9 een u m e~

rat e d.
The leading passage on the gatis as prefixes to a krt consists of the th ree
last stras of the difficult passage 2, 2, 18-22. In order to interpret these
rules it is in the first place necessary to examine the adhikras on which
they are dependent, viz. SP. 1,4, 1 sq., 2, 1, 1; 2, 1, 2-4 and ?, 1,22 . .
30. Interpretation of the paribh~ s. 1,4,1 sq.

Buiskool 1), when explaining the most important paribh~s or rules of
interpretation applied by the commentators to the text, introduces this topic
by the prevalence~paribh~ given as no. 38 in Ngojibhatta's Paribh~~
Indusekhara. This paribh~ is said to regulate the mutual prevalence
between four special paribh~s regulating themselves the mutual pre~
valence between two stras which would be equally applicable in a
special case.
As Buiskool remarks, the methad of PJ}.ini's grammar is a methad of
theoretical development and as such foreshadows, in a highly ingenuous
manner, the methad of historical development characteristic for the modern
European science of etymology. In the same way as a European etymo~ .
logist applies in his construction of an etymology the applicable phonetical
laws in their historical order, sa does PJ}.ini successively submit roots,
stems, suffixes and prefixes to the rules of his strls. Sometimes the
succession of application will be indifferent, but of ten a difference in the
order of application will lead to a different result. In order to escape this
difficulty the commentators have laid down the prevalence~paribh~s to
teach with stra has, in case of uncertainty, the prevalence. The commen~
tators even went so far as to construe a prevalence~principle regulating the
mutual prevalences of those prevalence rules. The whole methad is much
alike to the principles of exegesis formulated by the Purva~MmIJlsists
for the difficult cases where ritualist ic texts contradicted each other 2) .
Indeed the whole matter is sa intricate and sophistical or at least subtle
that it seems impossible to attribute, as Goldstcker did, any of these
scholastic inventipns to the time of PJ}.ini hiplself.
For further research it is of importance to notice th at the viprati~edha~
1) p. 52 sq.
lI) See A. Berriedale Keith, chapter V or G. Thibaut's Introductory Remarks.
paribha~ of 1. 4, 2 shows a great resemblance to the asiddhatva~

paribh~as 6, 1. 86; 6, 4, 22 and 8, 2, 1. all these rules referring to the
mutual relation of stras and their applicability, that these stras without
exception have a restricted field of applicability (e.g. su. 6, 4,22 bears on
6, 4, 23-128; s. 8, 2, 1 on 8 p. 2-8, 4 fin ., su. 6, 1. 86 on a limited
subject), and finally that s. I, 4, 2 itself follows immediately an adhikra
I, 4, 1 bearing as such on I, 4, 1-2, 2, fin. So the reader will be al most
or quite convinced that I, 4, 2 is likewise an adhikra with the same extent
of validity.
Partly then su. 2 is a fuller explanation of su. I, for whilst s. 1 lays
down the rule that in the following three pdas technical terms exclude
each other, s. 2 says more clearly that in section I, 4-2, 2, fin. and
spedally in the pada~bha~passage (1, 4, 14"':"-20) it is a feature of composi~
tion and redaction that definitions are often too wide and must be restricted
by following definitions; e.g. s. I, 4, 17 defines padas as word~stems
before case~endings (except the sarvanmasthnas) and before denomina~
tive noun~suffixes 1), but immediately af ter th is s. 18 de fin es the bha as
a word~stem before such endings and suffixes when beg inning with a
vowel; strictly speaking I, 4, 1 would be suffident to solve the diHiculty,
but I, 4, 2 combined with s. 1 seems to make matters much clearer:
padas are word~stems before denominative noun~suHixes and case suffixes
which are no sarvanmasthnas and moreover begin with a consonant.
However, not only perhaps as a sort of comment on the preceding stra,
but certainly by itself su. I, 4, 2 is an adhikra to I, 4-2, 2 fin. and is as
such particularly important for the interpretation of the ku~gati~passage as
will be afterwards shown 2).
A general remark may be inserted here. Whilst the subtle argumentation
with the aid of prevalence~rules seems to hve been greatly developed
under the influence of the Purva~Mmarp.sa, the stras of P~ini may be
considered to be an improvement upon the general stra~style, for whilst
striving for the same pregnancy as found in the Kalpa~Stras the author of
the A~tdhyay at the same time leaves matters less to the risk and caprice
of oral tradition. On the other hand the work was not put together as a
manual for foreigners, but written for scholars and teachers, thoroughly
acquainted with the language. Common sense and general knowledge are
presupposed, and the system of composition is not a matter of mechanica 1
algebraic calculation.
31. Int e r pre t a t ion 0 f th e ad hik ras u. 2, I, 1.

Adhyya 2 beg ins at the very outset with a pregnant and enigmatic
stra 2, I, 1 samarthal} pac/a~vidhil} = a rule about words (here: declined
nouns 3) or indeclinables 4) as last member of a compound, and noun~
1) See 42 note.
2) See 49.
3) SP. 1. 4, H .
4) SP. 2, 4, 82 and I, I, 62 sq. with Bhtlingk's note to 62.
28 Studies on PI].ini's Grammar
stems 1) or indeclinables as preceding member of a compound) must be fit:

As appears from the examples given in the comments of the Ksik the
sutra is meant as an adhikra of 2, 1 and 2 (theory of composition). The
Ksik paraphrases samartha by sakta, and this required fitness consists
in the expressing of all the word~meanings [which become conscious] by
grammatical analysis: vigraha~vkyrthbhidhne saktaf;. But the Ksik
adds to this a further or other ' interpretation: athav samartha~padSraya~
tvt samarthaf;; here whilst using the ambiguity of the word samartha and
giving this word an indirect bearing on the first member of the fo11owing
compound, the commentator considers the rules about padas to be 'fit', [if
they refer to] padas .fitting in together' [i.e. logica11y connected] . And
th is witty, although difficult formula thus informs the reader that 2, 1 and
2 discuss the rules for composition.
In its interpretation of 2, 2, 18 the Ksik refers to 2, I , 1 as adhikra:
... ku~gati~prdayaf; samarthena (!) sabdntare1J.a saha nityaf!1 samasyante.
32. Interpretation of the adhikra 2, 1,2-4.

Su. 2, I, 2 stands rather apart, giving an accent~rule for a dependent
noun~case placed before a vocative. In as far as such a vocative~group is a
phonal unity, it shows likeness to the samsas, subject~matter of 2, 1 and 2
according to 2, I, 1 and 3. Moreover sup of su. 2 is implicitly supplemented
to s. 4, a circumstance which explains the position of this sutra as
prompted by a wish for mnemotechnical economy.
The meaning of 2, I, 4 is thus clear: saha sup = sup saha sup = sub~
antaf; 2) saha sub~antena samsaf;, a compound (su. 3) is a declinable
noun [standing as first member and] accompanied by another declined
noun [as second member] ' . The first noun loses as a rule its declinational
ending, but is still according to grammatical convention called a sup or
sub~anta, and the fin al noun receives the termination required by sentence~
syntax.
A compound, then, in general is = noun + noun, only by way of
exception it is = indeclinable + noun, or noun + indeclinable, or inde~
clinable + indeclinable. May one of the members be a finite verb? The
Ksik on 2, 2, 18 gives as example yd iri~karti; however nowhere in
adhy. 8 p. I, which treats of finite verbs with preceding gatis or fo11owing
enclitics, the term samsa is used for those close verbal~groups; cf. 8, 4, 14.
33. Th e ter m tatpuru~a (SP. 2, I, 22).

In the European grammars which are favourable to PI].ini and his
system the tatpuru~as are classified as the usually so called compound
nouns (e.g. Monier Williams, grammar p. 327); but PI].ini himself (2,2,
18-22) includes also indeclinables + deverbative nouns or indeclinables as
saf!1skra~, saf!1skrtya.
1) SP. 2, 'I, 71 and 1, 1. 62 sq.

2) Sup = sub-anta (1, 1, 72) which evidently includes here the prtipadikas as
member of a compound.
Not e I. T het e r m s tatpuru~a, karma-dhraya a n d dvigu. _.

The passage 2, I, 22 - fin . offers special interest in connexion with the
adhikara J, 4, 1. The tatpuru~as con ta in as sub-class the karma-dhrayas,
in which the relation of the two members is one of concord, e.g. the relation
between adjective and noun (nilotpala); and these karma-dhrayas contain
as a sub-class, the 'theoretica]' dvigus; e.g. paiica gvaQ becomes as a
'theoretica]' compound * paiica-go-, Erom which are derived the neuter
'complexive' dvigu paiica-gav- 1) and the adjectival dvigu paiica-gu- 2),
'bought with five cows', 3). But whereas the reader would expect the term
karma-dhraya to be ~entioned in su. 2, J, 49, its definition is given in
I, 2, 42 in order to escape the difficulties which would arise from the
adhikara I, 4, 1.
Not e 11. Th e ter m upasarjana . - The term upasarjana being
insufficiently defined by PI).ini (in s. J, 2,43 sq.), it has been necessary
to consult the commentators to arrive at a coherent exposition of the theory
of pradhna and upasarjana. According to the results of this research
(Wackernagel 11 58a IJ) the first member of a dvigu is an upasarjana
and the second the pradhna. However the term upasarjana is not only
met with in the theory of composition, but also of denominative noun-
derivation, where the primary noun is termed upasarjana, as is apparent
fr om SP. 6, 2, 36. Thus in the adjectival dvigus which are derived with luk
of the suffix aTJ (4, I, 88 with s. 83 as adhikra) or luk of thak (5, J, 28
with su. 19 as adhikra) the total 'theoretica]' form * paiica-go- could be
termed the upasarjana with respect to the desired adjectival dvigu. IE it be
allowed to attribute this implicit argumentation to PaI).ini, the shortening
of the second member in the case discussed could be explained by SP. I,
2, 48; the Kasik, however, does not mention forms like dvi-gu-, paiica-gu-
as examples.
34-37. The notion upapada and the third adhyaya.
34. Th e me a n in g 0 f th e ter m upapada.

S. 2, 2, 19 of the ku-gati-passage refers with its t~rm upapada to 3, I,
92, one of the adhikaras of the section 3, I, 95-3, 4, fin . Thus the inter-
pretation of the passage 2, 2, 18-22 requires a discussion of the term
upapada and an analysis of the 3d adhyaya.
According to its etymology the term upapada indicates a secundary or
auxiliary word, and occurrences of this wide meaning are frequent. In
SP. 1. 3, 16 and Kasik an adjunct such as itaretarasya with vyati lunanti
.is called an upapada; in I, 3, 71 the indeclinable mithy is an upapada of
1) Wackernagel 11 p. 306, 117 c ~ and SP. 5, 4, 92; 2, I, SI samhre.

2) With luk of the taddhita-suffix (hak, SP. 5, 1. 28 with s. 19 as adhikra; cf. I,
1, 63 which stra annulates the kit-influence (7, 2, 118) of the suffix (hak. On the
'shortening' (I, 1,48) of go into gu see note 11. For the adjectival dvigus in general see
SP. 2, 1. 51 taddhitrtha- and Wackernagel 11 p. 274, 107 b ADm.
3) SP. 5, 1. 37 tena kritam.
krayate: in SP. 1. 3, 77 and Ksik svam in the sentence svaT!' yajiiaT!'

yajati (or yajate) is an upapada of yajiiam and indirectly of the verb, for
it defines the abhiprya of the act (s. 1. 3, 72). According to SP. 1,4,105
tvam in tvaT!' bodhasi is an upapadaT!' samndhikara1)am of the finite verb,
for it de fin es the madhyam~termination of the verb and stands towards it in
the relation of grammatical concord (samndhikara1)atva 1). In the close
construction of a finite verb with an infinitive or participle ending in ~aka
(SP. 3, 3,10) the finite verb is called the upapada or 'auxiliary' to the main
notion expressed by the krt, as is apparent from s.2, 3, 14, where the
Ksik comments: kriy kriyrthopapadaT!' yasya so'yaT!' kriyrthopapadalJ,.
'Tumun~1)vulau kriyyT!' kriyrthaym' (3, 3, 10) ity e~a vi~ayo lak$yate.
Kriyrthopapadasya ca sthnino 'prayujyamnasya dhtolJ, karma1)i krake
caturthi vibhaktir bhavati. Dvitiypavdo 'yaT!' yogalJ, 2).
But although the grammatical term upapada always possesses one aud
the same general meaning, we may distinguish two special applications,
the grammatical and logical. Thus in the constructions 3, 3, 10: vrajati
bhojakalJ" 3, 3, 11: vrajati pkya, 3, 4, 18: alaT!' krtv, 3, 4, 65: sknoti
bhoktum, 3, 4, 66: parypto bhoktum, alaT!' bhoktum the declinable
parypta~, the indeclinable alam, and the fini te verbs vrajati and saknoti
express 'logica]' auxiliaries, whereas in the majority of cases e.g. nicailJ,
krtv and nicai1;r~krtya (3,4, 59), coraT!'~kram (3,4, 25), anu~krtya, and
dyacara~ (3, 2, 17) the indeclinables are grammatica]' auxiliaries, either
used with an accent of their own, or as a part of a compound.
35. A n a I y tic a I tab 1 e 0 f th e th i r d ad h y Y a.
Divis. p. stra Sub j e c tof t h i r d a d h y Y a.
Intro. Introductory rul e s.

1 1-4 Pratyayas and their accentuation.
A R oot ~ suf f i x e s.
5-32 Derivation of secundary and denominative
roots (dhtu).
B Tense~suffixes ( vikara1)a ).
33-90 Formation of lakras.
C I Derivation from r oot s.

adhik. 91-94 Adhikra~ and saT!'jii~stras, see 36.
I
Ca Morphology and semasio~
10 gy of declinable krts.
.. 1 1 95-132 Krtyas (karman~relation etc. cf. 3, 4, 70).
I
1) AdhikarBr;ra = 'relationshIp of words In a sentl<.Oce whIch agree together' (MWO.)
whilst in 2, 3, 36 = 'location, sense of the locative case'.
") Cf. 23 In fine.
The grammar of the incleclinables 31
Divis. pa. stra Sub jee tof t h i r cl a cl h y a y a.
Ca2 Nomina agentis (kartar~relation. cf. 3. 4. 67)

and some other krts.
.. 2a 133-150 derivation in generaI.
2 1-83 .. in case of preceding upapada.
.. 2f3 84-109 subject continued; adhikara: 'bhte .
84-101. with preceding upapada.
102-sqq. no upapada mentioned. derivation
in generaI.
105. digressive stra about finite verb chan~
dasi.
110-122 prsangika~stras (I). finite verb expressing
the past; (cf. su. 105) .
.. 2r 123 prsangika~stras (11). finite verb; adhikara:
lJartamne.
124-180 subject continued: nomina agentis; [lJartamne].
181-186 Kraka~relations of ~tra~ and ~itra~suffix.
187-188 Kta~participles with present meaning ; (for
kraka~relations of participles see 3.4.70 sqq. ).
3 1-2 Supplement: bhta~ and vartamna~relations
in u1)~di~suffixes .
.. 2<5 3 Krts en ding in ~in; adhikra: bhavi~yati.
4-9 prsangika~stras (111). finite verb express~
ing future.
10-15 (partly prsatigika. IV) . declinable and in~
declinable krts and finite verbs expressing the
future .
.. 2, 16-17 Ghan stem~suffix of nomina agentis without
special time~relation (Ksika to s. 16:
.bhav1fyati' nivrttam). 'Concatenative' group.
.. 2C 18-130 Krts expressing bhva and other kraka~rela~
tions than kartrtva . Concrete and idiomatical
meanings.
Cb 3 131- Moods and tenses of the

4 8 fin i t e ver b (lakras 0 f tin) .
Prsatigika~stras: pa. 3 su. 158. 163. 167.
169-172. 174: grammatical mood expressed
by krts. In su. 167 kla etc. is an upapada with
tum~infinitive. [Por a literary example of the
construction mentioned 3. 4. 2 see C. Cap~
peller. Balamgha. 1. 29].
Divis. p. stra Sub j e c tof t h i r d a d h y Y a .
Cc Mor p hol 0 9 Y a n d s erna s i o~

4 log y 0 fin dec I i n a bIe krts.
.. 1 9-17 Vedic infinitives.
. 2 18-21 Exclusive use of tv~ (ya - ) absolutive.
. 3 22-24 Choice between tv~ and am~absolutive. Use
of finite verb.
.. 4a 25-46 Exclusive use of am~absolutive, (s. 25 and
29 sqq. with noun in karman~relation as upa~
pada [and first member of compound]; su. 37,
41 , 43 with noun in other kraka~relation as
such), the absolutive and the main finite verb
derived from the same root, S. 34 sqq:
.. 4fJ 47- 58 Exclusive use of am~absolutive; noun in
different kraka~relations (3rd, 7th, 5th and
2d) as upapada [and as separate word or as
first member of compound]; roots of finite
verb and absolutive are different.
.. 5 59-64 Facultative use of tv~ and am~absolutives;
indeclinable as upapada [and in both cases as
separate word or as first member of com~
pound].
" 6 65-66 Constructions with tum~infinitive; saknoti etc.,
alam etc. as upapada. ,
Suppl . 67-76 Supplement to Ca-Cc. Kraka~relations or

bhva in the verba I word (krt or ti;l) itself.
o Con j u gat ion a I vibhakti.

4 77-{in . Terminations of the finite verb in the different
voices, tenses and moods. (The sutras 113 sq.
are partly prsatigika)
36.Th e ad hik r a~ a n d saf!1jn~ s tra s 3, 1, 91-94.

91 : dhtol;! , . [what follows in the adhyya concerns suffixes] after a
root'. - The interpretation follows in positive respect from SP. 1, 1, 67
and the contents of the rest of the adhyya, and negatively from the
contents of the adhyyas 4 and 5: grammar of the declinational vibhaktis
(4. 1, 2) , grammatieal Dlotion (4, 1, 3-81) and denominative derivation
of nouns (4, 1, 76-5, 4. fin.) 1) .
1) The stras i. 1. 77-81 have a 'concatenative' character,

92: tatropapadaT!l saptamistham, 'in this section of the work what is put
in the locative is upapada'. - Since the ablative is now used for indicating
the root, a new technical application 1) of the locative is introduced by the
author; see for the general technical meaning, 1. 1. 66. In fa ct this saT!ljfi
covers the passage 3, I, 95-3, 4, 66 ( 35 Ca---c).
93: krd atiti. - The stra gives a definition with the help of the contents
of the adhyya.
94: vsarpo striym. - The sutra only refers to the declinable krts
(3, I, 95-3, 3, 130 = 35 Ca), from which stras, however, we must
exclude the passage beg inning with striym (3, 3, 94) 2).
37. Th e kin d sof upapadas int het h i r d ad hy Y a.
In general one may say that in Ca, with exception of the prsaTigika~
sitras I-IV, the nouns and indeclinables used as first member of decli~
nable compounds, and in the prsatigika~stras of Ca and in Cb the temporal
and modal particles construed with a finite verb are 'grammatica!' upapadas.
whereas the upapadas in Cc are partly prefixes or preceding nominal mem~
bers of composition to krts and as such 'grammatica!' upapadas, and partly
fini te verbs or nouns fulfilling more or less the function of finite verbs, con~
strued with indeclinable krts and as such 'logica!' upapadas. cf. 34 in fine.
38-39. Bhtlingk's translation of SP. 2, 2, 18-22 quoted and
criticised.
38. B h t I i n 9 k 's t r 11 sla t ion a n din ter pre t a ti 0 n
of SP. 2, 2, 18-22.
18: ku~gati~prdaya1;t 3); 'Desgleichen werden ku, ein Gati (s. I, 4,
60 fgg.) genanntes Indeclinabile und pra mit seinen Genossen co mp on irt' .
(kupuru.$al). urarkrta1;t. du~puru~al). supuru$a1;t).
19: upapadam atiti, 'Uesgleichen eine Ergnzung, wenn diese kein Ver~
bum finitum ist'.
(kumbhakra1;t u.s.w. Vgl. 3, I, 92 fgg. zu Upapada).
20: amaivvyayena 4), 'Eine Ergnzung auch mit einem 18declinabile,
aber nur dann, wenn dieses ein Absolutiv auf am ist'.
(svduT!lkram, saT!lpannaT!lkram und lava1).aT!lkraT!l bhutikte. V gl. 3,
4, 26).
21: trtiyprabhrtiny anyatarasym, 'Aucl!, ein Instrumental und ein dar~
auf folgender Casus, aber nicht notwendig'.
(milakopadaT!lSaT!l oder mlakenopadaT!lsaT!l bhuTikte, keSagrhaT!l ader
keSe~u grhaT!l yudhyante. V gl. 3, 4, 47 fgg.).
22: ktv ca 5), ' Desgleichen mit einem Absolutiv auf tv (ya)'.
(uccai1;tkrtya oder uccail) krtv. V gl. 7, I, 37).
1) In one case the locative means 'when used in the meaning of', s. 3, 3, 93.
2) See Thieme p. 27.
3) The stra is to be supplemented with samsaQ (2, 1. 3), tatpurusaQ (2, I, 22) and
nityam (2, 2, 17). Nityam is superseded by anyatat'asym in 21 and 22.
4) saha. supplemented from 2. I, 4.
5) ktv instrumental; saha supplemented.
3
39. D i f f i c u I tie s t 0 w h j c h B h t I i n 9 k .s tra n s I a-

t ion a n d c hoi c eo f e x a m p les 9 i ver i s e.
In the following respects the translation and interpretation by Bhtlingk
afford insufficient or incorrect help to the reader:
1. One cannot even understand from these rul es why such common
forms as prakrtya. upetya etc. are obligatory compounds. for in the exam~
pIes of su. 18 only nouns and no absolutives are mentioned. and s. 22 is
only a facultative rule and moreover does not apply to the case.
2. The expression atir'! in s. 19 'wenn diese kein Verbum finitum ist'
remains unclear with no other help but such a vague reference as 'cf. 3.
1. 92 fgg:.
3. The translation trtiy~prabhrtini in s. 21 by 'ein Instrumental und
ein darauf folgendes Casus' 1) is erroneous; it should be '[ the upapadas
mentioned 3. 4. 47-64:] an instrumental etc:.
4. In the note to su. 22 a cross-reference to 3. 4. 59 should have been
given.
On the whole Bhtlingk has failed to see the place of this passage in
PJ}.ini's plan of composition.
40-46. The ku~gati-passage

and the pada~bha~passage as 'bridges.
Logical elegance in PJ}.ini's composition. Interpretation of
su. 2. 2. 18.
40. C 0 u r s e 0 f dis c u s s ion.
The ku~gati~passage forms a bridge between the third adhyya and
three groups of phonological and morphological rules in the 6th and 7th
adhyya as will be explained in 57-60.
In . similar way the pada~bha~section (1. 4. 14-20) forms a bridge
between the adhyyas on denominative noun-derivation (adhy. 4-5) and
extensive parts of the adhyyas 6---8. As these subjects are very important
with reference to PJ}.ini' s methods and scholarly merits. the discussinn
about them will precede the interpretation of su. 2. 2. 18.
41. The relation between the 6th and 7th
a d h y Y a a n d t hel ast a d h y Y a.
The adhyyas 6---8 form together the grammar of coalescence 2). which
partly bears on morphology. processes such as the mutation of the root.
the insertion of i or i between root and suffix etc . partlyon phonology.
processes of word~ and sentence~accent and the coalescence of the speech-
sounds. Generally speaking the closely connected 6th and 7th adhyyas
explain the coalescence of word-components into words. and the 8th the
coalescence of words into sentence. In detail. however. the discus sion is
divided in such a way that adhy. 6 and 7 contain (a a) the morphological
word~coalescence and word~accentuation. and (a fJ) the vowel~saIPdhi
1) This translation would require -prsbhrtsysQ in the text.

2) Buiskool eaUs it in other words the synthetieal part of P~'s grammar.
both within the word and the sentence 1), this combined treatment being
useful for mnemotechnical economy, and that adhy, 8 treats of (b a) the
sentence~accent and (b f3) the consonantal sa'!ldhi both within the sentence
and between the stem~pada ancl the so~called pada~suffixes .
42. T hem e a n i n 9 sof t het e r m s pada a n d bha, SP.

1, 4,14-20.
The term pada denotes (1) a complete word (sil. 14) inclusive the last
member of a compound 2) i.e. a sup (declined noun), tin (finite verb) or
an avyaya 3) (indeclinable), (2) the prtipadika which represents the word
as preceding member in a compound (s. 14) 4) (3) the stem of a noun
before (a) the declinational endings beginning with a consonant except the
s of the nom. sing. 5) (sil. 17), (b ) the denominative noun~suffixes
(taddhitas) beginning with a consonant except y 6) (Sil. 17) and except
the case mentioned in s. 19, (c) the taddhita~suffixes beginning with y or
vowel, but characterised by a grammatical~technical s 7) (sil. 16). and (d)
the stem of a noun ending in n before the denominative root~suffixes
technically called kya (s. 15).
The term bha denotes the stem of a noun before (a) the declinational
suffixes beginning with avowel except in the sarvanma~sthna~cases :
~am, ~au, nom. plur. ~as , neutr. plur. ~i (s. 18 and cf. SP. 1, 1. 42 sq. ),
(b) the suffixes ~, ~i, ~i of grammatical motion (s. 18 and cf. SP. 4, 1,
3--81 ); (c) the taddhita~suffixes beginning with a vowel or y except those
characterised by a technical s (sil. 18) , (d) the stem of a noun ending in
t or s before a matvartha~suffix (sil. 19) and (e) the Vedic cases referred
to in s. 20 (but missing in the GaJ).a~Ptha).
Not e. Th e a nno t a t ion toS P . 1, 4, 17 b y B h t 1 in 9 k
(edition, I p. 477 "Nachtrge und Verbesserungen" ) . - Bhtfingk
annotates to 1, 4, 17, 'unter den svdayaIJ verstehen die Erklrer alle
Suffixe yoni, 1, 2 bis 5, 3, 70' 8) . The Ksik, however, says sv~di~v iti
su~sabdd ekavacand 9) rabhya, kapalJ pratyay grhyante. The kap~
suffix mentioned here, is not found in 5, 3, 70 but it is the samsnta~suffix
of 5, 4, 151 fgg. 10) , which reference in view of the ambiguity of 11)
comes to this: 'the au thor of the stra here refers to all endings mentioned
1) See 92.
2) A compound in its totality Is never called pada, cf. SP. 8, 4 s. 1 and 3.
3) 2, 4, 82 and 1, 1, 62 with BhtIingk's note; 1, 1, 63 (angasya!) does not bear on
th is casenotwithstanding the rejection by luk (adhikra~stra 2, 4, 58) .
4) 1, 2, 46 (where samsa = preceding member of a compound), 2, 4, 71 and 1, 1, 62.
11) This results by the application of the adhikra 1, 4, 1 sq.
6) See 42 Note.
7) These are the suffixes ~iya. ~iga. ~ga and ~gu called respectively ghas. chas. gas.
gus (Bhtlingk 11 p. 185* s.v. sit) .
8) Cf. Bhtlingk 11 p. 145* GP. 257.
8) Cf. SP. 4, 1, 2.
1.0) BhtIingk 11 p. 158* s.v.
11) Cf. SP. 2, 1, 13 and PJ;ni's technical use of prk.
36 Studies on Pnini's Grammar
in the section beginning 4. 1. 2 and ending with the discussion of the

suffix kap 5. 4. 151-154 rand its negative supplement 5. '4. 155-160]'.
43. Th e pada~bha ~ sec t ion a s a b r i d geb e t wee n

the theory of declension and denominative noun~
derivation (adhy. 4-5) and the theory of word~ and
sen ten c e ~ c 0 a les c e n c e.
The close relation of su . 1. 4. 14-20 to the adhyyas 4 and 5 in their
totality has been shown 42 note; the importance of this passage for the
last th ree adhyyas appears from the fact of meeting with such adhikra~
sutras as 1. bhasya (6. 4. 129. adhikra for 6. 4. 129~fin . ) . 2. padasya (8.
1. 16). which. combined with padd (17) and anudttaTJl sarvam (18) is
the adhikra for 8. 1. 19-8. 1 fin .. and by itself for 8. 2. 1-8. 3. 54. and
3. apadntasya (8. 3. 55) and samnapade (8. 4. 1) .
A few examples as to how the pada~bha~section functions as a bridge
are added:
formation of noun~cases; su. 1. 4. 17 ( 42 pada. 3a) connects the
general su tra of declinational endings with 8. 2. 7: rjan~bhis > rja~bhis
(which remains unchanged notwithstanding 7. 1. 9 owing to 8. 2. 2) ;
formation of the feminine; su. 1. 4. 18 ( 42 bha. b) connects 4. 1. 15
with 6. 4. 148 and 8. 2. 5: kuru~car~ (formed with ta 3. 2. 16) + 1i(i)p >
kuru~cart;
denominative noun~derivation ; su. I, 4, 17 ( 42 pada, 3b) connects
5, 2, 124 with 8. 2, 30 and 8, 4. 53: vc~gmin~ > vk~gmin~ > vggmin ~
(cf. 8, 4, 47); su. 1. 4, 18 ( 42 bha, c) connects 4, 1. 84 (with aTJ to be sup~
plemented from su. 83) with 6. 4, 148 and 7,2, 117: asvapati~~ > svapat~.
Note. The function of su. 1. 4,14 (42 pada, 1 and 2) . -
Su. I, 4, 14 with reference to the word as part of the sentence connects the
seven first adhyyas with the eighth, and in its application to the first
member of a compound forms a bridge between the theory of word~compo~
sition (2 p. 1 and 2) and the last th ree pdas (the T ripdi) of the total
work.
44. Log i cal el ega n c e 0 f P 1]. i n i 's co m pos i t ion.

In the same way as the pada~bha~section connects the 4th and 5th
adhyya with the respective parts in the three last adhyyas. so is the
ku~gati~passage a bridge hetween the 3rd adhyya and the adhyyas 6--7.
Indeed, the more one studies the A~tdhyyi the more one realises the
depth and correctness of the characterisation given by Speyer of this work
as 'ein wohl durchdachtes und nicht nur knstliches. sondern auch kunst~
voIles Lehrbuch' (Winternitz 111 p . 386).
45. T hei n ter pre t a t ion 0 f 2. 2. 18-22 i n gen era 1.

i nco n n e x ion wit hit s a d hik r a (sup saha sup samsal;!)
a n d t hei n ter n a 1 c 0 m pos i t ion 0 f t h e a d h y Y a s 6--8.
The passage 2, 2. 18 sqq. mentions exceptions to the adhikra: sup saha
sup samsalJ, (2, I, 2-4). Immediately the first sup is contradicted by

su. 18 ku~gati~prdayalJ" 'the first member of a compound may be an
indeclinable, either (a) ku or (b) one of the gatis (a prefix to a krt or a
close adverb to a tUL 1)) summed up in I, 4, 61-98 or (c) one of the
in stances of the prdi~gaT)a in as far as possessing other functions than
that of a gati (cf. the examples of the Ksik: su~puru~alJ etc.).
But what about the second sup? Does the introduction of these ku~gati~
prdayalJ, supersede the second sup as weU , or is it maintained? The latter
alternative would be logicaI. the former a pardonabIe licence on the si de
of PI).ini. PI).ini has taken this licence and his ultra~algebraic interpreters
and critics may here accu se him of 'bad workmanship', for otherwise such
usual forms as anukrtya and apetya could not be explained by his grammar.
But if both the sups as restrictions are eliminated does this mean that in
general there are nine combinations possible for forming a tatpuru~a:
sup + sup, sup + avyaya , sup + tin, avyaya + sup, avyaya + avyaya,
etc.? For the first member tin is explicitly rejected in s. 19 upapadam
atin, but why does not the au thor say as weU atin?
As PI).ini gives a grammar and no manual of orthography, a discussion
on a doublet such as atra and attra bears on pronunciation, and his dis~
cussion of the compounds is of a similar nature. Therefore a compound is
not 'that which is written and printed without any interspace' , but is a unity
in phonal and ideational respect; ideational and not only notional is this
unity, since in Sanskrit a syntactical coherence may be sufficient for the
formation of a compound; and phonal is th is unity because unifying rules
of accent and morphology enter into play.
Now, the construction of an upapada(a close adverb or a modal adverb
or conjunction ) with a tin is discussed with reference to sentence~accent
in adhy. 8 p. 1 (e.g. 8, I, 71), but nowhere in this pda is the term
samsa applied to th is construction (see Bhtlingk 11 p. 286* s.v.). On the
other hand an upapada (an indeclinable, or a noun in kraka~relation)
with a foUowing krt (deverbative noun or indec1inable) is discussed with
reference to accent in the passage beg inning with the adhikra 'samsasya'
(6, 1. 223) , see e.g. 6, 2, 49 sq., and with reference to morphology in the
7th adhy., see speciaUy 7, I , 37, wh ere the very term samse is found 2).
Thus from comparing 2, 2, 18 sqq. with the internal composition of the
three last adhyyas it foUows that a tin cannot be the last member of a
compound; and the reason for this different treatment is evidently their
accentuation, for in a samsa of an upapada with a krt one syUable is
udtta by fixed rule, but a verba I word~group such as ri karoti can be
totaUy anudtta (8, I, 68 sq.) or accentuated either iiri karti or r
karoti [cana]. (8, I, 71 and 57 sqq.).
1) It will afterwards be proved that the latter case must be rejected.

2) A fuller discussion of these passages is given 57-60.
38 Studies on Pfu].ini' s Grammar
46. Int e r pre t a t ion 0 f su. 2, 2, 18.

According to the.argumentation given in 45 s. 18 means: ku, a gati
(as prefix to a krt) and an indeclinable of the prdi~gaTJa (as prefix to a
noun) are compounded with a following declinable or indeclinable (nouns,
participles, infinitives, absolutives), and th is composition is obligatory
(nityam, to be supplemented from su. 2, 2, 17). Thus a form such as the
absolutive anukttya must be explained with the help of SP. 7, 1, 37; 6, 2,
139 1) and 2, 2, 18. Similarly all other absolutives, infinitives, pa~ticiples
etc. follow in their accentuation the rules of 2, 2, 18 and 6, 2, 1-6, 3, 24,
e.g. nukartum (suffix tumun 3. 3. 10 and 6. 2. 50), anuktlram (suffix
TJamul 3. 4. 22 sqq.; 6, 1. 193 and 6. 2. 139). nukrta~ (suffix kta. 3. 4.
70 sqq. and 6. 2. 49; karmaTJi!). durgat~ (6. 2. 144; kartari!).
47-49. The cross~references between the ku~gati~passage and the 3d

adhyya. S. 2. 2. 19-22 interpreted.
47. Us e f u 1nes sof a tab 1e 0 f cr 0 s s ~ r e f ere n ces.
The stras 2. 2, 19-22 are so interwoven with the third adhyya that
a close comparison is necessary for their interpretation, and it is this
peculiarity of composition which makes it probable that the A~tdhyyi
was discussed by the aid of manuscripts between master and pupils before
the texts were committed to memory. for only such a debate 2) could make
the memorisation attractive and adequate for its purpost.
Not e. A b b r e v i a t ion s in th eta bIe. - Pos. = positively.
neg. = negatively, dir. = directly; the divisions Ca etc. refer to 35.
48. Tab I e 0 f c ros s ~ r e f ere n ces 0 f 2, 2, 19-22 t 0

th e 3 d a d h y Ya .
(1) Su. 19 refers pos. to Ca = SP. 3. 1, 95-3, 3. 130 (exc. pr s. su.)
(2) .. . neg. to pr~
sa;lgika~su. of Ca = .. 3,2, 110-122; 123 etc.
(3) No s. refers dir. to Cb = 3,3.131-3.4.8
nor to Cc 1-3 = .. 3.4,9-24
(4) Su. 20 refers pos. to Cc 4a = .. 3.4.25-46
(5) S.21 .. Cc 4f3
and CeS = .. 3,4,47-64
(6) S. 22 to CeS = .. 3,4,59-64
(7) No su. dir. to Ce6 = 3,4.65-66.
49. Dis c u s s ion 0 f t hes e par a t e s tra s 2. 2. 19-22.

(l) Su. 19, upapadam atift. Positive part: an upapada (that is a noun~
stem in kraka~relation, or an indeclinable either as prefix or in adverbial
relation) is regularly (nityam) compounded with a following declinable in
1) Bhtlingk '.s wrong in leaving out examples of this kind.

2) On the discussions between coileague grammarians see Weber. Ind. Studien. XIII
p. 406. 'Man zankte sieh zu Zeiten unter einander recht krftiglich'.
The grammar of the indedinables 39
all those cases mentioned in the lirst part of the upapada~section which
bears upon the case (3. 1. 95-3. 3. 130 after substraction of the
prsangika~rules). for there can be no question of a taddhita as second
member. this subject having been fully treated in the preceding section
(2 p. 1 su. 22-2 p. 2 su. 18 inclusive) nor of indeclinable krts because
of the viprati~edha in the following sutra 1). which gives rules for the only
possible indeclinables. the ktv~ and am~absolutives . Examples: prastha
3. 1. 136; kumbha~kra 3. 2. 1; agni~toma~yjin 3. 2. 85; sa,!!parkin 3. 2.
142; idhma~pravrascana. 3. 3. 117 (cf. karmaryi in 116); i~at~kara 3. 3. 126.
(2) Su. 19. negative part: a finite verb (as 'logica!' upapada 34 in
fine) is never compounded with its following pradhna: vrjati bhjakalJ
3. 3. 10; vrjati pk'tIya 3. 3. 11.
(3) No su. refers to Cb and similar prsangika~sitras in Ca. as follows
from the argumentation given in 45. Examples: k~iprm gami~yati 3. 3.
133 (Cb) and iti ha cakra 3. 2. 116 (Ca). - Neither does any sutra
positively refer to Cc 1-3. although negatively some of the cases fall
under su. 2. 2. 19 sqq. If in the Vedic idiom saknoti should be followed by
apalupam no compound could arise according to su. 19 (SP. 3. 4. 12).
neither could any Vedic infinitive in ~se, ~ase etc. (SP. 3, 4, 9 sqq.) form
a compound with a preceding (logical ) upapada according to su. 20 and
22. Whereas ala,!! karot i forms its absolutive as ala,!!~krtya according to
2, 2, 18 ( 46). on the other hand la,!! krtva synonymous with khlu
krtva (SP. 3. 4. 18) remains uncompounded and does not follow 2, 2, 22.
in which sutra the trtiy~prabhrtni are mentioned. i.e. a cross~reference
is given to 3, 4, 47 ( 35 and 48 Cc 4{J). Sutras 3, 4. 20-22 do not
mention upapadas; su. 23 bears on the construction of yad with a finite
verb, which, too. does not allow composition ( 45); gre bhja,!! vrajati
aod similar cases mentioned in 3, 4, 24 would according to algebraic
interpretation fall under 2, 2, 20, but since the synonymous expression
gre bhuktv'tI falls negatively under 2. 2, 21, su. 3, 4, 24 being mentioned
before 3. 4. 47, it is evidently the sutrak!,t's intention to make su. 20
exclusively bear on su. 3, 4, 25 sqq. ( 48 Cc 4a).
(4) Su. 20, amaivvyayena: [an upapada, cf. su. 19] is only then
[regularly compounded. cf. su. 17] with an indeclinable if this be an
absolutive ending in ~am. The positive cases are summed up in Cc 4a:
a
cora,!!karam krosati 3, 4, 25; tath~k8.raTJ1 bhok~ye etc. By the following
stra 2) the field of strict applicability is limited to su. 3, 4. 46 as terminus
ad quem.
(5) Sa. 21. trtiy~prabhrtny anyatarasym, if loosened from its
connex ion with adhy. 3, Cc 4{J and Cc 5 = SP. 3, 4, 47-64, would become
quite unintelligible. for trty~prabhrtni cannot mean the third nOU!l~
case. the fourth etc., but refers to the upapadas mentioned in the passage
3. 4. 47 sqq . Examples of facultative composition: mulakenopad,!!sa,!! and
1) The relation between s. 19 and 20 is an instanee of s . 1. 4. 2. Cf. 30.
lI) Cf. the adhikra~stra I, 4, 2.
40 Studies on PJ;!.ini' s Grammar
mlakopadT!lsaT!l bhufzkte (3,4,47), vraj upardhaT!I, and, vrajopardhaT!l

gal]. sthpayati (49); k~i niktl1].aT!l, and, ak~i~niktlT].aT!l jalpati (54), niciIJ.
karam, and, ncaiIJ.~karam a cak~e (59).
(6) S. 22, ktv ca = [trty~prabhrtiny upapadny anyatarasyaT!l
saha] ktva ca. Trty~prabhrtini rders to 3, 4, 47 sqq., but since the
absolutive in ktv construed with an upapada is not mentioned bdore 3, 4,
59, this facultative rule really bears on 3, 4, 59-64 ( 48, Cc 5). Example:
niciIJ. krtvcak~e and nicaiIJ.~krtyacak~e (59).
7) Absence of siltras rderring to 3, 4, 65 sq. Syntactical constructions
with a 'logica!' upapada as saknoti krtum, laT!l krtumare never com~
pounded according to 2, 2, 18-22, mentioning ~am and ~ktva as the only
indeclinable krts of the upapada~section (3, I, 95-3, 4, 102) fit to become
a second member of a compound. - On the other hand gatis as upapadas
are regularly compounded with infinitives and adsolutives of all kinds
according to s. 2, 2, 18, for the rules of su. 20-22 only rder to the
upapadas of the corresponding passage in the third adhyaya.
By observing thus the strict parallelism between SP. 2, 2, 18-22 and
the 3rd adhyya the former passage becomes perfectly clear, and not the
slightest blame can be attached to P~ini's technical expression, the only
lmplicit admission being that s. 18 supersedes both the sups of the general
adhikra: 'sup saha sup sa msa IJ. ,.
50-56. The Mahabh~ya on the ku~gati~passage .
50. H in d u a n d Eu rop e a n men t a I i t y. Th e Mah~

bha~ya a n dit s co mme nt s 0 n SP. 2, 2, 18 sqq.
In the preceding pages the research was prompted by the wish to get
an insight into P~ini' s peculiar genius and his methods and mannerisms
of composition, to understand the differences between Hindu and European
mentality based as the latter is on ancient Hellenism. The aim of the
ancient Hindu commentators, on the other hand, was to understand the
laws of their language, Pa~ini's composition~form being still their own
manner of expression. They correct, give examples and additions, but on
the other hand rarely analyse the text in the way of a European philologist.
In view of a comparison which may be desirabIe between the explanation
given above and Pataiijali' s commentary the main points of his reasoning
are summarised in the following pages, where prominenee is given to those
passages which support or seem to contradict the above argumentation.
51. S tra 2, 2, 18, v rtt i kas I, 3, a n d 4, a n d a par t

of Pat a ii j a I i 's ad dit ion.
Vrtt. I, prdi~prasafzge karmapravacaniya~prat~edhaIJ., 'on occasion of
the pradayaIJ. being discussed it is said th at the karma~pravacaniyas are
excluded [from the discussion)'. Cf. SP. 1. 4, 83---98 and 2, 1. 5-21. One
of the examples of the Bhl?ya: vrk~aT!l prati vidyotate vidyut.
Vrtt. 3, siddhaT!l tu kv~n~sv~ati4ur~gati~vacant, 'but [the rule iS]
The grammar of theindeclinables 41
correct. since [instanees can bel mentioned for ku and the gatis . su. ati.
dus'. [One of the examples]: ku~brhmaT)aly..
Vrtt. 4. prdayaly. ktrthe. 'pra etc .. when implying the meaning of a
participle in kta'. pragata cryaly. = prcryal;!. - Page 416 1. 15 sqq .
quotation from the school of the Saungas on the subject.
Addition by Pataiijali (p. 417 1. 3) . avyaya'!l pravrddhdibhil;! samas~
yate '" punal;!~pravrddham, bahir~bhavati etc.
52. Pro eli tic s a n den eli tic s c a II e d mem b ers 0 f

co m p 0 u n d s b y Pat a ii j a I i [a 9 a ins t P J:t i n i's ter min o~
logy]. p. 4171. 4. sub SP. 2. 2.18.
Ivena vibhakty~alopal;! piirvapada~prakrti~svaratva'!l ca: v~sasl wa,
kany iva. PJ:tini only discusses the enclitic use of iva after a verb (8. 1.
57sqq.).
Udttavat tiri gatimat cvyaya'!l samasyata iti vakt~vyam: anuvy~
calat. anuprjvisat, yat~pariynti 1); the rules bearing on this case are given
SP. 8 p. 1. The terminology of the twogrammarians is different. the
linguistic facts to which they refer are the same.
53. Sutra 2.2.19. vrtt. 1-4.

Vrtt. 1 and 2. 'upapadam atiri'~iti tadartha~prati~edhal;!? kriy~prati~
~edho va? (cf. SP. 3. 3. 10 and 49 no. 2).
(Pataiijali' s comments). MBh. I p. 417 1. 12 tadarthasya .. , - Kasya?
- tiri~arthasya. and 1. 15 athav vyaktam eveda'!l pathitavyam 'upapadam
akriy' iti. - Athkriyeti ki'!l praty udhriyate? - krakogatal;!, hrako
gatal;!. - Naitat kriyavci . .:.- Ki'!ltarhi? - Dravya~vci.
Since not only vrajati krakal;!. but also gatal;! krakal;! cannot be com~
pounded. su. 19. algebraically interpreted. is not quite correct; this is the
starting~point for the commentator's reflections.
Vrtt. 3 and 4. ~a~thi~samsd upapada~samso viprati~edhena; na v.
sa~thi~samsasybhvd upapada~samsal;!.
The vrttikas and Pataiijali's comments discuss the relation between
2. 2. 8-17 and 18-22. especially su. 19. Rja~puru~a is eVidently a
~~thi~samsa and stambe~rama (cf. 3. 2. 13 with supy upapade as supple~
ments from 3. 2. 4 and 3. 1.92; and see 6.3. 14) an upapada~samsa; but
both terms would be applicable in a case as kumbha~kara, (for genitive~
relation see 2. 3. 65 and 2. 2. 16. and for upapada~relation 3. 2. 1 ). Thus the
vrttika~krin considers whether the difficulty could be solved with the
vipratisedha~paribh~ (1. 4. 2) etc.
54. Bot h t hesups 0 f t h e a d hik r a sup saha sup

samsal;! are s u per s ede din t h e pas s a 9 e 2. 2. 18 sqq.
according to Pataiijali (cf. 45).
MBh. p. 417 1. 18 (comm. to SP. 2. 2. 19 vrtt. 2): eva1]1 tarhi siddhe
1) Yat proclitic?1
sati, yad 'atiTi' iti prati~edhaT[l sasti, taj jfipayaty cryo 'nayor yogayor
(18 sq.) nivrttaf!l sup supeti ... Yady etaj jnpyate kenedni1fl samso
bhav4gati? Samarthena (SP. 2. 1. 1 and ~ 31 in fine).
55. Th e int e r pre t a t ion 0 f eva in 2, 2, 20.

MBh. p. 418 I. 19. Ida,!! tarhi prayojanam: amaiva yat tulya~vidhanam
upapadaT[ll) , tatraiva yatha syd; am cnyena ca yat tulya~vidhnam
upapadam 2), tatra m bhid iti: agre bhojam, agre bhuktv.
Not e. Cri tic i s m 0 n Pat a ii j a I i 's int e r pre t a t i 0 n 0 f
2, 2, 20. - In this passage tulya~vidhnam is a bahuvrhi and amaiva is
dependent on its first member tulya, the formula amaiva yat tulya~
vidhnam upapadam refers to SP. 3, 4, 25-58, wh ere syntactical rules
for the exclusive use of the am~absolutive are given; an upapada (he says)
can only becompounded with an indeclinable am ~absolutive that in its
syntactical context does not permit the competing of a ktv~absolutive,
and the composition is obligatory (nityam); or in other words: an upa~
pada can only be compounded with an absolutive which according to
syntactical conditions may only have the am~form , and then etc. Pataiijali' s
sophistry consists in the double function attributed here to the limitative
particle 'only', amaiva yat .. . , tatraiva . In the same way, as it seems, the
commentators explain samartha/J padavidhi/J (SP. 2, 1. 1; cf. 31) by
samarthnT[l padnT[l vidhi/J samartha/J; but really the cases are not the
same; in 2, 1, 1 the commentators attribute a stylistic figure to PJ?ini; in
2, 2, 20 Pataiijali distorts the text.
56. T hef 0 r mof s c i ent i f i c t hou 9 h t o f P J? i n i

a n d h i s c 0 mme n t a tor s.
In Ktyyana and Pataiijali we can already recognise the technical
reasoning, the dialectics of the Pi1rva~MimIp.s and the philosophical
schools, where the doubts and difficulties are formulated as dilemmas and
where the solution of one dilemma starts the formulation of a new dilemma
and so on in a long stream of debate, and where even the accuracy of
expression seems to acquire mathematical exactness. Yet however
interesting this form of scientific expression may be in the history of Hindu
thought, we should go astray in attributing this pseudo~algebra and
dialectics to PJ?ini and his contemporaries. But being on our guard against
such false surmise we can learn much from these commentators, for how
fully they have shown in th is passage their grasp of the narrow relation
between the ku~gati~passage and the third adhyya, a point where even
Bhtlingk has fallen short.
1) SP. 3, 4. 25-58; cf. 35. C c 4.

2) According to examples given the commentator means: SP. 3. 4, 24. Although
Pataiijali is right in substracting this stra from the influence of 2, 2, 20, his method is
far removed from modem philology. Cf. 49 no. 3 in fine.
57-60. Morphological and phonological consequences of the ku~gati~

passage.
57. Mor p hol 0 9 i cal con s e q u e n c e 0 f SP. 2, 2, s . 18

a n d 22. See SP. 7, 1,37.
SP. 7, 1,37, 'samse 'nanpurve ktvo lyap', where the Ksik comments:
samse 'nan~purve ktv ity etasya lyab ity ayam ddo bhavati.
Examples of the Ksik: compounds with a gati according to 2, 2. 18:
pra~krtya, pra~hrtya (SP. 1,4.60); with an upapada according to SP. 2,
2. 22 in connexion with 3, 4. 59-64 ( 35, Cc 5. and 49 no. 6):
prSvata~~kftya. 3, 4. 61; nn~krtya and dvidha~krtya. 3, 4, 62.
Por the indeclinable of negationand disapproval nan (with the forms a~.
na and an. SP. 6, 3, 73-77) see the commentators.
58. P h 0 nol 0 9 i c a I con s e q u e n c e 0 f t h e ku~gati~

pas s a ge: t h e prakrti ~ a c c ent u a t ion 0 f t hef i r s t m e m~
b e r 0 f a c 0 m p 0 u n d, S . 6, 2. 49-53.
Sa. 6. 2. 49. gatir anantara~ 1) [udtta~ 6, 1. 159; tatpur~e 6, 2, 2.
cf. 33. ~kte, 6, 2. 45; karmaTJi 2) 6, 2. 48. - and with regard to a bisyllabic
gati. 6, 2. 1 prakrty purvapadam, the adhikra of the section to which the
stra belongs]. 'A gati. but only the last before the krt, is normally
accentuated in a tatpuru~a when a kta~participle expressing the gram~
matical object (= which has passive meaning) functions as second
member'.
Although the Ksik only mentions examples with the prdaya~: prkrta
prhrta. abhyuddhrta, the rule is generaJ. e.g. lamkrta 3) (1. 4. 64).
The Ksik adds the note ththdi~svarpavdo yoga~. 'the ru Ie is an
exception to s. 6. 2. 144'; i.e. the two rules cross each other. an active
participle in ~ta with prefix is according to 6, 2, 1H oxytonon; example:
durd~gat~.
Por the interpretation of stras 50-53 see Bhtlingk. The examples
borrowed from the Ksik for su. 50 are again too limited in character. cf.
av~~kartum (SP. 1. 4. 61; GP. 31 no. 37). In s. 6. 2, 52 ancatau is the
locative of the fini te verb ancati representing the root anc. which is used
as a krt according to SP. 3. 2, 59; 35. Ca 2a.
59. P h 0 nol 0 9 i c a I con s e q u e n c e 0 f t h e ku~gati~

pas s a ge: th e prakrti ~ ace ent u a t ion 0 f th e sec 0 n d
mem b e r 0 f a co m p 0 u n d. S. 6. 2. 139.
S. 6. 2, 139, gati~karakopapadt krt [prakrty, 6. 2. 137; udtta~,
1) One would expect anantar.

2) Cf. 23.
a) Wackemagel II p. 226, 93b a.
4) Wackemagel does not mention this case, 11 p. 226, 93 a.
Studies on Pnini' s Grammar
6. 1. 159] . The siltra evidently refers in its redaction to the ku-gati-passage

and consequently the term upapada refers likewise to the upapada-section
of the third adhyya. The term ku. found in 2. 2. 18 has been left out in
6. 2. 139. because in the ku-gati-passage ku only refers to compounds such
as ku-puru~a . where the second member is a noun. Had Pt:tini paid
particular attention to a compound such as ku-car-. he would have done
so in sil. 3. 2. 17 (cf. e.g. purah-sar- in 3. 2. 18) . And the term prdayah
of 2. 2. 18 has been left out. because only the prdaya/:l with gati-character
are here considered. Although in the upapada-section the kraka-upapadas
are discussed together with the avyaya-upapadas (cf. 3. 2. 1 karmaTJi,
4 supi. 15 adhikaraTJe etc. and on the other hand: 61 upasarge 'pi. and 3. 3.
126 i~ad-du/:l-su~u). yet the term kraka is added here for clearness. as
upapada by itself would be too wide (cf. 34) and would not refer
distinctly enough to the third adhyy-a . The interpretation thus runs: 'a
compound containing a krt as its second member is accentuated on this
member prakrty. if (1) a gati. (2) an [upapada expressing a] kraka-
relation or (3) [any other] upapada [discussed in the 3d adhyya]
precedes.
Examples: praktuaka (2. 2. 18). idhma-pravrscana (2. 2. 19 and 3. 3.
117) . i~at-kra (2.2.19. and 3.3. 126).
60. P h 0 nol 0 9 i c a I con s e q u e n c e 0 f t h e ku-gati-

pas s a ge: s h i f tin 9 0 f t h e udtta tot hel ast s y II a bIe
of the compound. Sil. 6.2.144-151.
Sil. 6. 2. 144. shifting of the accent to the last syllable takes place when
a krt ending in one of the suffixes (1) -tha. (2) -atha. (3) -gh(a}ii. (4)
-k(ta}, (5) -(a}c, (6) -(a}p, (7) -itra, (8) -k(a) is preceded by a gati,
a karaka-[upapada] or any [other] upapada. cf. 6. 2. 139. 59.
Examples with a gati. La. (4) visu~k (still considered according to
Pt:tini's method of grammatical fiction to be *-su~-t. cf. SP. 8.2.51). the
word is oxytonon. because the kta-participle here expresses the kartar and
not the karman . cf. 6. 2. 49 and 58 ; (8) pra- vr~ 1) . the root vr~ is here
applied to the male animal and the suffix k ( a) expresses the kartar (SP.
3.1 . 135; 3. 4. 67. and 35. Ca2a).
Examples with a [kraka] -upapada. La. (3) k~tha-bhed (name of an
instrument. 3. 3. 121. cf. 3. 3. 117 and Ksik on sil. 121; 35, Ca 2C;
( 8) go-vr~, Icharivr~ 1).
For sil. 145-151 see .Bhtlingk.
1) Liebich. Zwei Kapitel etc . p. XXVIII: Urn so dankbarer werden wir P~ini sein.
dass er es nicht verschrnht hat. auch solche Worte uns aufzubewahren. die er nicht in
Texten . .. fand. sondern bei den Hirten auf dern Felde oder selbst bei den Wrfelspielern
in der Schenke. - The people thernselves. however. used these words in a Prkrit form.
cf. Skld. Nirukta. p. 1331
IV. GAT I AND NIPATA, OR THE CLOSE ADVERBS OF THE FINITE VERB
AND THE MODAL AND CONJUNCTIONAL PARTICLES.
61-68 . Analytical table of adhy. 8 p. 1. Worth of PJ?ini"s work.
61. Th e th e 0 r y 0 f th e cl 0 s e . a d ver b s a n d th e
m 0 d a I a n d con jun c ti 0 n a I par t i cl e s, SP. 8 p. 1.
The theory of the gatis as close adverbs and the niptas is given by
PJ?ini in connexion with the theory of enclisis and proclisis. In the notes
to the following analytical table of adhy. 8 p. 1 references are given to
Whitney's article of 1856 ('Bopp's Comparative Accentuation of the Greek
and Sanskrit Languages') and the chapter 'Betonung der Satztheile'
in Oelbrck' s 'Altindische Syntax'.
62. A n a I y ti c a I tab I e 0 f ad h y. 8 p . 1.
o i v~ sitra I Sub j e c tof a d h. 8 p . 1.
A Amre4ita
1-15 Word~repetitionand its accentuation; bahuvrhis and
karmadhrayas formed by repetition.
B Anudttatva 0 f wor d sin t hes ent e n c e

adhik. 16-18 General adhikra of the pda: every [syllabIe] of a
word anudtta; conditions required: 1. not placed at
the beginning of a verse~line; 2. placed af ter another
word (exceptions su. 69 sqq.).
Not e. Padd =
nvasnd: an enclitic word cannot
come immediately af ter a phonetical stop. From this
follows th at 'the verb is always accented if it stand at
the head of its own particular clause in the sentence
(Whitney p. 216 1. 17).
Ba V 0 cat i v e, (Wh. p. 214).
Bb Per s 0 n alp ron 0 u n s.

20-23 Positive sutras: the enclitic forms.
24-25 Negative stras: the accentuated forms obligatory
before ca, v, ha, aha, eva.
26 Facultative rule.
Bc 27 Par tic 1 e sof bI a m e and [painful] repetition

(GP. 78) af ter finite verb.
46 Studies on Pa~inr s Grammar
Divis. stra Sub j e c tof a d h. 8 p . 1.
Bd Finite verb
.. 1 28 Main rule: [padd = nvasnd] anudtta
.. 2 Exceptions: a c c ent e d are
2a a finite verb af ter a finite verb (Wh. p. 217; D. 23.1).
.. 2(3 29 a periphrastic future (D. p. 35 note 1).
.. 2" 30-50 a finite verb 'combined' (yukta) with certain niptas;
these are partly conjunctions (Wh. p. 215 1. 13. D.
27). partly modal particles of emotion. Of ten the
meaning is added.
30. yad. yadi etc.; 31-35. naha. satyam. ariga. hi.
36---38. yvad. yath; when accentedand when enclitic
(cf. 66).
39-43. tu. pasya. pasyata. aha. aho; pur; na nu.
44-45. kim; for influence of upasarga see 66.
46. ehi manye.
47-50. af ter jtu. indefinite pronoun (kas cid etc.
heading the sentence-; after ho. utho heading the
sentence when verb immediately follows; etc. (For a
case of a similar kind with the nipta atho see D. 23,3).
.. 2d 51-54 a Irt. a lot in a second sentence, certain conditions
being fulfilled; and lot af ter hanta (D. 26); for
influence of preceding upasarga see 66.
Ba! 55 V 0 cat i v e. In combination: m + finite verb +
vocative. in calling from distance vocative accented.
Bd 2E 1
56 - 58 A c c-ent e d mor e 0 ver are:
a fin i t e ver b. f 0 II 0 wed b y yad. hi. tu in
Veda; a finite verb. followed by cana. cito iva (for a case
of a similar kind with the nipta id see D. 24). a
particle of blame or [painful] repetition, the verbal
taddhitas of degree etc . the same finite verb repeated;
and before ca. v. ha. aha. eva (D. 27). For the
influence of a preceding gati see 66;
.. U 59-62 a fin i t e ver b. c 0 m bin e d wit h (yukta) ca.
v. ha ('k~iyym). and aha ('k~iyym' and 'vini~
yoge') - or without ca or ha. but combined with eva
- in the first of two coordinate sentences (D. 25);
63-65 facultative rules bearing on the verb of the first of
two coordinate sentences (D. p. 41);
Bd 21] 66 af ter a form or derivative of the [stem of the] relative
pronoun (cf. s. 30-50).
Divis. stra Subject of ad h . 8 p. 1.
pras. 67 PrsaTigika-rule. Noun as second membe! of com-

pound, anudtta af ter words of praise.
Be 68-69 Finite ver b, and the group cloSE; adverb with

verb, anudtta af ter words of praise and before certain
words of blame.
Bf 70-71 Close adverbs when udtta or anudtta in

dependence on following close adverb or on the accent
of the verb (D. 28) .
-
Bg 72-74 Words following a v 0 cat i v e; vocative as
apposition of vocative. (The rules follow from the
principle padd [= nvasnd]. cf. Wh. p. 217 last
paragraph) .
63. P I). i n i a nd Del b r c k co m par e d .

What is it that strikes us most in Pl).ini, the ancient pioneer of gram-
mar and auditive witness of the linguistic facts, wh en we compare him with
Delbrck, investigator trainoo by modern methods of research and a great
reader of literary documents? What strikes us then, when taking account
of difference and agreement, is the trustworthiness of Pl).ini, his power of
observation, his love of language, his deep intellectual and emotional
understanding .
64. W by is t h e adhikratva 0 f s tra 8, 1, 17 1 i m i t e d

b y t h e c 0 mme n t a tor s t 0 8, 1, 68 i n c I u d e d ?
The Ksik annotates su. 8, 1, 17 by the words padd ity ayam adhi-
kral} prk 'kutsane ca supy agotrdau', which is right since stra 69
contains an exception to the condition required by the adhikra and the
following sutras 70 sqq. are not directly connected with the adhikra itself.
but must be read in connexion with special rules of the section Ba sqq. (see
e.g. the examples given by Bhtlingk and in the Ksik to su. 72).
65. Th e co m pos i t ion 0 f a d h y. 8 p . 1.

The synoptical table of 62 shows us that the construction of the pda
is based in the first place on logical principles, the main division being A.
mrec;lita and its accent, and B. the sentence-accent, which latter subject is
divided into the vocative and cognate facts. the personal pronoun. and the
finite verb in connexion with prefixes and particles.
Peculiar is the distribution of the rules concerning the vocative: Ba (su.

19) . Ba (s. 55) and Bg (su. 72-74) 1). It seems that PJ}.ini has used these
rules and likewise the rule on the particles of blame (Be. s. 27) and the
prsangika~rule on nominal compounds (s. 67) as a means for articulating
the principal discussion of the verb (Bd 1-2b. 2e-2f). Be-n.
66. T her u les 0 f t hes e c t ion s B e and B f s y mbo I i~

c a ll y e x pre s s e d.
In order to symbolise the rules of su. 68-69 and 70---71 the following
symbols are accepted: prij = accentuated expression of piij; kts =
accentuated expression of kutsana; g and ga = accentuated and un~
accentuated gati; tin and tin = the stressed and enclitic form of the finite
verb. The symbolic expressions are then
su. 68-69 pj (ga) tin ( 1)
(ga) tin kts (exc.: su. 27) (2)
sU. 70-71 to single titi corresponds (ga) g tin (3)
tin (ga) ga titi (4)
However. in some cases ru Ie (4) does not hold good and is replaced
by the subsidiary rule:
to single tin corresponds (ga) g tin (5)
For this shifting of accent see su. 38. H . 53. 57 sq.
67. Th e ter m s gati a n d upasarga i n ad h y. 8 p . 1.

The term upasarga is met with in su. 38. Hand 53. the term gati in
57. 68 and 70; the reader gets the impression of there being scarcely any
difference.
68. Wor t hof P J}. i n i .s wor k.

The worth of PJ}.ini' s work for us is not in its linguistic information.
however valuable many a detail may be. but in its fitness as a training
for our understanding the Indian methods of composition; and specially as
a scientific work it offers in this respect great advantages. When thirty
years ago I was writing my first book of research devoted -to Sankara's
Git~bh1?ya the critical and sarcastic remarks made on the Bhagavad~git
by Bhtlingk and Hopkins appealed more to me than now owing to
insufficient insight in the synthetic procedures of the Hindu mind. The
foregoing pages have touched this subject incidentally; the second part
of this article will be entirely given to it.
1) For the vocative cf. SP. 2. 1. 2 and 6. 1. 198.

11. THE METHOOS AND MANNERISMS OF COMPOSITION
IN PANINI'S ASTDHYYI.
69-93. The fundamental notions of Hindu gramm~r. Main and

subsidiary principles of PJ}.ini's form of composition. Ten~
dency towards dichotomy in the main divisions.
69.Th e H in d u a n d Eu rop e a n sc i e n c e 0 f 9 ram mar.

Hindu grammar di Hers from its European sister~science partly by its
fundamental notions, partly by its literary composition and its form of
description, of which the algebraic anubandha~system and further mnemo~
technicaI means such as the use of adhikras and in general of anuvrtti in
a continuous series of sutras form characteristic features.
Whereas European grammar in its traditional form is based on two sets
of notions, the notions bearing on sentence~analysis such as subject,
predicate, object, etc. and the notions of the parts of speech, substantives,
adjectives, etc., Hindu gram mar is in this respect rather undeveloped,
operating on the one hand with notions that lack systematisation such as
upapada, kraka, samndhikara1J.atva, vise~a1J.a and dividing on the other
hand the words into two wide categories: A. atiTi (either used as a complete
word or as first member of a compound) and B. tiTi (the finite verb), and
sltbdividing the former category into Aa sup. declined word, corresponding
to the European notions of substantive, adjective, pronoun and numeral,.
and Ab avyaya, containing not only the particles, conjunctions, adverbs,
prefixes but also the deverbative indeclinables i.e. the absolutives and
infinitives.
In the first part of these studies all these notions have been amply
explained and examined; the following pages deal with the form of literary
composition.
Not e. T h e c a r din a I n u mer a I sas dec I i n a bie s. -
Whereas in the European languages the cardinals with a few exceptions
are indeclinable, they were in Sanskrit all subjected to declension, a
linguistic feature which gives PJ}.ini's simple classification a peculiar fitness
for the language which he describes.
70. M a i n p rin c i p les 0 f P J}. i n i 's f 0 r m sof I i ter a r y

co m pos i t ion: m n e mot e c h n i c a I e con 0 m y, log i c a I
d i v i s ion a n das s 0 c i a t i v e d i 9 r e s s ion.
These principles may be considered to be primary in PJ}.ini' s technique
of composition: (1) mnemotechnical economy, (2) logical division, and (3)
associative digression (prsaTigika).
.
50 Studies on PI).ini's Grammar
What strikes the European student most in reading PI).ini is the

prevalence given to the first principle above the demands of logical division.
A typical example of th is characteristic has been met with in the term
pada, defined in such a way that a combined treatment of sentence-
coalescence and consonantal assimilation within the word becomes
possible 1 )
The associative digression, which often interferes with the log ic of
division, can be either due to what precedes or to what follows , or it arises
from the wish to link together two logically divided subjects. Sometimes,
moreover, a section is not limited to the subject expected by the reader
according to the context, but at the same time deals with subjects which
in the author's mind are coherent with it.
As an example of preceding digression may be mentioned su. 2, I, 2,
for whilst adhy. 2 pa. 1 and 2 treat in general of the compound, the sutra
mentioned gives the rule that a noun-case dependent on a following
vocative is considered with reference to accent to be as it were one word
with th is vocative.
As an example of following digression may be quoted 8, 4, 57. where
unexpectedly a ru Ie is given for final vowel in a context dealing with
final consonants, mainly because of the anuv~tti of vvasne which thus
could be used.
The in stances of concatenative digression are very numerous; see for
instance passage 4, I, 77-81, which forms paJ;'t both of the section on
grammatical motion and of the section on denominative noun-derivation.
As an example of too wide discussion may be quoted I, 3, 2-9, for,
whilst the introductory su. I, 3, 1 dhtaval) and the total discussion of
I, 3, 12-fin. show that the ma in subject of the pda is the verb and its
active and middle voice, yet the sutras 2-9 do not only give rules for the
anubandhas of the Dhtuptha, but treat the anubandhas in general.
71. Sub s i d i a r y f 0 r m s o f co m pos i t ion a n d a r-

rangement: (1) pirvatrsiddhatva, (2) distinctive insertion,
(3) 0 rat 0 r i c a I d i v i s ion a n d (4) ins e r ti 0 n bye me r-
gen c y .
The technical principle of arrangement called pirvatrsiddhatva is
adopted in, adhy. 8 p. 2-4; see Buiskool's publication.
A very striking form of arrangement is the 'distinctive insertion', which,
unexpectedly, introduces and leaves a subject foreign to the context, an
illogical licence here and there made use of in order to mark the articulation
of the composition; see for instance the definitions of sa'!lhit and
avasna at the end of adhy. 1 p. 4, and the peculiar way in which the
rules for the vocative are distributed over adhy. 8 p. 1 (su. 19, 55 and
72-74) 2).
1) See 42 sq.
2) See 65.
The methods and mannerisms of composition in P~ini' s A~tdhyyi 51
Then, we may mention the author's inclination to announce in the last

stra of a pda, more or less directly, the subject of the following pda
(cf. 8, 2, 108, sa1[lhitym , and 6, 1, 223 [anta~] samsasya [udtta~].
the positive rule to which 6 p. 2 gives the exceptions. Likewise we meet
here and there with the opposite form of arrangement, when the au thor
at the beginning of a pda gives a few additions to the foregoing pda
(e.g. 3, 3, 1-2 and 3, 4, 1-8) 1). In all these cases one might think of
oratorical devices for giving emphasis or rousing interest.
As a fourth subsidiary form of composition may be reckoned the'in-
sertion by emergency', in which case the author places, indifferent where,
one or two stras which for some reason or other could not be placed
in the contex! required by logic; e.g. the definition of karma-dhraya is
given in 1, 2, 42 instead of 2, 1, 49 (or af ter 2, I, 23) , to avoid the
difficulties which would arise from su. 1, 4, 1.
72. T h e l 0 9 i c a I d i v i s ion s h 0 win g a t end ene y

t war d sdi c hot 0 m y.
0
A fact of peculiar interest in P~ini' s composition is the tendency towards
dichotomy in .the main divisions of his work. As th is dichotomy repeats
itself three times, the internal arrangement of the A~tdhyyi can be
symbolised by the formula : IAa, lAb, !Ba, !Bb, IIAa, IIAb, IIBa, IIBb.
The analytica I .tabIe of 73 makes this arrangement dear, whilst the
following sections con ta in its full discussion.
Not e. A b b r e v i a t ion sus e din t het a bIe 0 f 73. -
Acc. = accent, an. = analysis, art. = articulation, cj. = con jugation,
cp. = composition of words, de. = dedension, dr. = derivation, i. =
introduction, id. = indedinable, mo. = morphology, ph. = phonology,
tg. = technical-grammatical development.
73. Con cis e a n a I y tic a I tab I e 0 f the A ~ t d h y Y i.
DiViS 1 adhy. .!
I SUb j Summary of the A~tdhyyi.
I 1- 5 Theory of the ultimate components of
language, or the analytical part of grammar.
IA 1-2 The fundamental notions of grammatical

science.
IAa 1 The notions bearing on phonology, word-

dassification, technical-grammatical de-
velopment and the theory of function and
I I form.
1) See 35.
52 Studies on Pl).ini's Grammar
DiViS.1 adhy. subj. S u m mar y 0 f the A ~ t d h y Y i.
p.lllp.2 The pdas land 2 of adhy. 1 are in the

1 p.: beg inning parallel and gradually diverge.
1. 1-19 ph. The mutation of roots and stems; the anu~
2. 1-40 i>andhas bearing on mutation. cf. 77 sq.
1. 20-43 an. General analysis; the parts of speech and
2.42-46 their characteristics (conjugation. declen~
sion. indeclinability; composition).
1. 44- fin. tg. Technical~grammatical development (ga-
ma, dda, lopa).
2.47-73 dc. Motion and declen!Oion. The plu ral forma-
tion as a kind of lopa (su. M .eka-se~alJ. ) .
p.3 cj. The krakas and bhva expressed by the
p.: finite verb; the abhiprya of the voices.
4. 3-20 dc. The fundamental terms of f1exional (esp.
dr. declinational) and derivational morpho~
logy; pada-bha-section;
4.21-55 dc. the numerical and kraka-relations of the
noun.
4.56-98 id. enumeration of the relational indeclinables.
4. 99-108 dc. the term vibhakti (declinational and con-
cj. jugatinal derivation) .
lAb 2p.: cp. The theory of composition; indeclinability

1 and 2 and declinability; relations between the
p.: members.
3 dc. The noun-cases as expressions of kraka-.
p.: upapada-. and vise~ar:ta-relations .
4. 1-31 cp. Number and gender of compounds.
4.35-57 cj. Heteroclitic systems of conjugation (desa
of dhtu).
4.58-85 cp. The theory of luk (slu) with reference to
etc. composition. derivation. tense-formation.
and the derivation of indeclinables.
IB 3-5 The theory of derivation.
IBa 3 dr. Theory of deverbative derivation; derivation

cj. of secundary roots. formation of tense-
stems; derivation from roots: declinable
krts, tiTis and indeclinable krts. conjuga-
tional vibhaktis.
The methods and mannerisms of composition in PJ].ini's A~tdhyyi 53
D i v is. adhy. subj. Summary of the A~tdhyyi.

I
lBb 4 and 5 de. Theory of denominative noun~derivation;
dr. declinational vibhaktis; motion; morphology
and semasiology of taddhitas: declinable
taddhitas, indeclinable taddhitas, sam~
snta~suffixes .
II 6-8 ph. The theory of phonological and morpho~

mo. logical coalescence, or the synthetic part of
I
I grammar.
I I
UA 6-7 The theory of word~coalescence and vocalic
sa'!1dhi in general.
6, 1. 1- i Articulative-phonological and morphologi~

157
.
cal processes defined in character but not
defined with reference to 'place (i.e. in~
dependently of samsa, aflga, bha or pada).
IIAa 16,1.158- acc. Theory of accent with reference to word

1 6,2, fin. and compound, [asamastasya padasya, 6. I ,
158-fin. , and] samasasya. 6 pa. 2. Accent
placed on stem or suffix, on first or second
I I member of compound.
I
BAb 6, 3, 1- art. Articulative~phonology and morphology of
7,4, fin. compounds and derivatives:
0.3, 1-6, 3, fin. Compound; processes in
the preceding member (adhikra: uttarapade,
cf. SP. 1. 1. 66), vowel~lengthening .
6, 4, 1-7, 4, fin . Derivation, the section
6, 4, 129-175 referring to bha~stems only.
I
lIB 8 ph. The theory of sentence~coalescence and of

mo. consonantal sa'!1dhi in general: pada'.
lIB a 8 p. 1 acc. I Theory ofsentence~accent: 8. 1. 1-15.

The accent of mrecjita. 8, 1, 16~fin . Accent
of vocative,' enclitic pronouns, finite verb.
etc.
1
5i Studies on Pt:tini' s Grammar
D i v i s. adhy. sub j. S u m mar y 0 f th e A 1? t d h y Y i.
IIB b p a . 2-4 art. Articulative~phonological sentence~coales~

cence. Changes in the pada and specially
at the end of a pada (words in the sentence,
and pada ~stem before suffixes) . Sa1[lhit~
processes in words and word ~groups . Phone~
tical processes in consonant~groups.
74. The internal composition and exterior

d i v i s ion c 0 m par e din tab e II i c f 0 r m.
Divis. adhy. adhy. adhy. Divis. adhy. adhy. adhy.

I 1-5 II 6-8
IA 1-2 IIA 6-7
IAa 1 IIAa 6 pa. 1-2
lAb 2 BAb 6,3-7.4
IB 3-5 IIB 8
IBa 3 11 Ba 8 pa.
IBb 4-5 BBb 8 pa, 2-4
75 .
Met h 0 dof r e s e arc hwi t h r e f ere n c e t o t h e
ad h y Y a s l a n d 2.
Before determining the place of the adhyyas 1 and 2 in the plan of the
composition of the total work it is desirabIe and even necessary to analyse
these adhyyas internally, since their construction offers great difficulties
and many an opportunity for testing the general principles formulated in
the sections 70 sq. In this analysis ( 76 sqq . ) ma in groups and sub~groups
are distinguished, the contents, if necessary, commented upon, and notes
on the function in the composition added. Preliminarily to a subsequent
summary we may say that this function can be of three kinds: (1) the
passage treats of a general subject, giving for in stance the definition of a
fundamental grammatical notion , (2) it has the character of a bridge
between a stra or series of stras found in the adhyyas 3-5 (indicated
in agreeinent with 73 sq. as 'div. I B' ) and a stra or passage in adhy.
6-8 (= 'div. 11') , or (3) if has a more independent character, in so far as
the discussion forms a unity in itself and may be compared to similar
discussions in div. IB. Numerous and important sections of the last kind
will prove the adhyyas 1-2 to be coordinate with the adhyyas 3-5.
76. Liebich's hypo thesis as to the .original begin~

n i n g 0 f t heS tra ~ P t h a.
The often repeated, but rather superficial comparison of Pnini's method
of theoretical development to a 'classical chemical experiment' has misled
The methods and mannerisms of composition in Pr:tini's A~tdhyayi 55
Liebich to surmise that the stra 'dhtolJ.' (3, 1,91) was the original begin-
ning of the total work , The fact that s, 2, 2, 18-22 is so closely inter~
woven with the 3d adhyaya that the former passage could not be under-
stood without the latter, and the latter, if read without the former, would
miss the bridge leading up to its effects as described in div. II 1), is
already sufficient to prove Liebich's hypothesis to be erroneous.
77. A n a I y sis of adhy. 1 pa. 1.

1. I, 1-19. The mutation of roots and stems, S. 1-6, gU1)a and vrddhi;
the anubandhas (su. 5) k and Ti forbidding any strengthening of roots
(cf. s. 4) in deverbative formations; the passage has of ten a bridge-
character; e.g. kta introduced in div. I B (3 , 4,71 etc.) is by 1. 1,5 exempt
from the rule of div. 11 (7, 3, 84). S. 7-10 de fine phonological terms;
su. 11-19 define the pragrhyatva and enumerate the cases.
1, 1, 20-43. General analysis. Classes of words (20,23 sq., 27 sqq., 37)
and of suffixes (22, 26, 42 sq.); technical names and indications of these
groups and rules bearing on them (e.g. 28-36). Su. 20 deals with conjuga-
tion, su. 22 with derivation , the stras 23-36 with declension, su . 37-41
with indeclinability and 42 sq. again with declension. The irregularities of
arrangement: 10. su. 21 to be interpreted as an associative insertion (cf.
da-d-malJ.) or rather as a distinctive stra ( 71 no. 2) separating su. 20,
deverbative formations from su. 22-36, denominative and declensional
derivation ; 20. s. 42 sq., placed after the treatment of the indeclinables
and so characterised as 'insertion by emergency' (cf. pada-hha- passage
1,4, 14-20, the adhikara-sutra 1, 4, 1 and 71 no. 4).
1, 1, 44-67. Technical-grammatical development: substitution (desa,
45-63), augmentation (gama, 46 sq. ) and rejection (lopa , 60-63)
included; anubandhas, technical terms and forms of expression referring to
these phonological and morphological alterations (64-67). Irregularities
of arrangement: su. 44 is a distinctive insertion; su. 45 is placed before
46 sq. without perspicuous reason.
I , I, 68-75. Further technical expressions. S, 73-75, introducing the
term vrddha , have special interest for the theory of denominative noun-
derivation (adhy. 4, see Bhtlingk, 11 p. 277* s.v.); the entire passage
68-75 is placed as an associative digression af ter 44-67.
1 p. 1 with its many paribha~- and saJpjfi-rules has on the whole a
general character.
78. A n a I y s is 0 f ad h y. 1 P . 2.
1, 2, 1-26. Mutation and accentuation. The anubandhas n, 1), n, k, p
and the rules of mutation: iiit- and nit-suffixes treated as nit-suffixes
( 1-3), apit as nit (4), lie/. apit as kit (5). Zit as kit (6). ktv (with resto red
kit). san, lin with sic, and sic (7,8-10,11-13,14-17; anuv!'tti of
tmanepade~u in (11-17); superseded kit (ktv 18, ni~th 19-21. ktv
1) See 29-60.
56 Studies on Pl).ini' s lirammar
and ni~th 22); facultative kit (ktv 23-25, ktv and san, 26 1 . -
The
arrangement is mainly based on the principle of mnemotechnical economy,
and partlyon parallelism; notice the use of anuvrtti in several stras; the
passage is quite a piece of virtuosity.
I, 2, 27-40. Vocalic quantity and vowel-modulation (accent); the sutras
34-38 about ritualistic recitation are an associative insertion and probably
an interpolation; sa'!lhitym does not refer to a ritual text (cf. Bhtlingk' s
translation) but to coherent speech or recitation (in antithesis to avagraha,
broken up or discontinued reading for exegetic purposes ) .
As the correlation between mutation and accent was known to Pl).ini,
the sections I, 2, 1-26 and 27-40 cohere and take up again the subject
of 1 p. 1 (su . 1-19).
I, 2, 41. Distinctive insertion; apfkta as term for a suffix of one sound,
e.g . the s as a declinational or conjugational termination.
I , 2, 42-46. Grammatical terms referring mainly to composition: karma-
dhraya (42), upasarjana (43 sq.), prtipadika (45 sq .); the definition of
karma-dhraya is an 'insertion by emergency' ( 71 no. 4), the sutras
43 sqq. follow according to associative or even thema tic coherence. More-
over the terms upasarjana and prtipadika are made use of in the following
passage (47 sqq.).
I, 2, 47-73. Morphology and semasiology of mot ion and declension.
Shortening (hrasvatva, 47 sq., 50) of final vowel in noun becoming neuter
or upasarjana (0 > u, etc., see I, I, 48); s. 49 and 51, rejection of suffix;
s . 51 sq ., general rul es about vyakti and [opa; grammatical concord. (The
stras 53-57 form a11 interpolation, see 79). S. 58--63, rules on gram-
matical number; su. 64, definition of riual and plural; s. 65--68, 70 sq.,
72 sq., 'objective' and grammatical concord with reference to family-names
and words ex pressing relationship, to certain pronouns, to herds of cattle.
The succession of the sutras is mainly regulated by logical division:
general rul es about both vyakti and vacana (51 sq.), idiomatical expressions
with reference to vacana (58--63), and social rank (vfddha and yuvan)
and gender. The stra defining dual and plural (64, cf. SP. I, 4, 21 sq.)
is placed before the group 65 sqq., as likewise dealing with se~atva.
Character of pda 2. Pda 2 shows a certain parallelism to p. 1; thus
in pda 2 the discussion of vocalic mutation with the addition of the
definitions of vowel-modulation or accent is taken up first: 1. I, 1-19/1
1. 2, 1-40; then some remarks are made on the compound, whilst the
corresponding passage in p. 1 treats of the word in general: I, 1, 20--43/:
I, 2, 42-46; and thirdly the discussion of nominal gender and number
uses the notions desa and [opa with their indications as explamed in p. 1
(I, J, 44-67 11 1. 2, 47-73). However, p. 2 mits last part (47-71) has
moreover the character of a closed and independent discussion of nomina I
gender and number (cf. 75 no. 3).
1) Bhtlingk translates the technical indication ral by 'ein Consonant mit Ausnahme
von h. y und v', read .... von y und v'.
The methods and mannerisms of composition in Pt:tini's A~tdhyyi 57
Not e I. E x am p 1 est 0 su. 1, 2, 47. - As Bhtlingk adds no

examples to th is rather important stra, some instances of its application
ale added here:
formation of the neuter of adjectives: na~ta-sri1:z (nom. sg. m. and
fem.) > na~ta-sri (idem neuter), Monier Williams, Grammar 190;
formation of an avyayibhva of which the second member is a noun
ending in long vowel or diphthong: * anu-gar'lg- > * anu-ga;lga- (2, 4, 18
and 1,2,47) > anu-gangam (2, 4, 83 in combination with 4, 1, 2; 7, I, 24
and 6, I, 197) .
79. T hei n ter pre t a 't ion 0 f t hei n ter pol a t e d p a s-

s ag e I , 2, 53-57.
The sutras I , 2, 53--'-57 are neither sufficiently clear in Bhtlingk' s
translation nor in the commentary of the Ksik . A few suggestions and
surmises are here added.
The Mahbh!?ya (I p. 229) explains tad in s . 53 as referring to the
different artificial sa~jiis such as ti. ghu. gha. bha (adhy. 1 p. 1 su. 64,
20, 22 and p. 4 s. 18); the opponent says such sa~jns should not be
have been taught because a sa~jii ought to show clearness and bear
weight (tat-prm1Jyd). but the right theory is. 'instruction makes the
term' (sa'!lFina'!l sa'!ljii) .
The Ksik and Bhtlingk connect tad of su. 53 with vyakti-vacane 1 )
in su. 51. For the following of the commentary of the Ksik the reader
will receive some help by the information that the indication lup ('change
of meaning with no adding of suffix and no changing of gender and
number') is met with in the following sfitras:
4, 2, 4 sq. name of time derived from name of asterism ,
4, 2, 81 sqq. .. empire inhabitants,
4, 3, 166 sq. .. fruit .. tree,
5, 2, 105 .. spot sikat and sarkar ,
5, 3, 98, a [nick- ]name of an individual person based on similarity,
5, 3, 99, proper name given to an idol (vsude va etc . ) which was made
by a devalaka as a means for obtaining alms ,
5, 3, 100 a symbolical expression used as a proper name; e.g . devapatha.
The Ksik, thus, differs from the Mahbh!?ya, in taking sarpjfi in the
meaning of 'proper name' and directs the polemics against 1, 2, 51 in
connexion with SP. 4, 2, 81 sqq ., 5, 2, 105 ; 5, 3, 98 sqq., ignoring the two
cases 4, 2,4 sq . and 4, 3, -166 sq. in its argumentation . It further interprets
yoga in su. 54 and 55 (cf. yuktavat in su . 51) in the technical meaning of
the etymologists (MWD. p. 856: 'the connexion of a word with its root,
original or etymological meaning ' ) and as antithesis to rdhi (1.1. , p. 885:
'the more amplified or popular or conventional meaning ' ).
1) See Wackernagel I p. LXII nate 4.

58 Studies on Pl).ini' s Grammar
The interpretation of s. 53-55 by the Ksik can therefore be para-

phrased as follows: the theory of lup as applied to the interpretation of
proper names should have been left out, since the proper names are in
themselves authoritative (53) and there is no need of any etymologicalor
learned interpretation (54), for if we could apply such an interpretation
to some of the proper names, how could others exist which do not allow
such explanations? (55).
Whilst Bhtlingk attributes the whole set of. stras, 1, 2, 53 sqq. (with
the exception of 57) to an ancient antagonist, the Ksik gives in th is
respect no information about 53-55, but takes Pl).ini himself to be the
author of 56 sq.
In s. 56 pradhna is explained by the Ksik as pradhnopasarjane
(the main and subsidiary member of the compound, which together express
the main or intended idea) and pratyaya as prakrtipratyayau (the original
stem or root and the suffix, which together express the speaker' sintention ) ;
the further interpretations of this commentary, however, may be put aside
as not convincing.So then, in my opinion, the whole sutra is to be attributed
to an antagonist, and its translation runs: the exposition of the significations
of pradhna [and upasarjana] and of the pratyayas [i.e. such lists as 4, 1,
92-4,2,91; 4, 3, 25-4, 3, fin; etc.] must not be taught, since something
e1se [than the instruction, namely idiomatical use] is authoritative with
reference to meaning.
To su. 57 Bhtlingk annotates (Nachtrge und Verbesserungen, I p.
477) : 'Da Pl).ini weder kla noch upasarjana erklrt, kann dieses
Stra .. . 1) seine Entgegnung auf die 1, 2, 53 fgg . gegen ihn gerichteter
Angriffe sein'. Pl).ini's argumentation thus could be paraphrased as follows
'I am right in treating the subjects objected to in 53-56 since other gram
marians discuss similar subjects. However, the statement that Pl).ini
nowhere discusses kla and upasarjana is incorrect, and the whole argu-
mentation attributed to Pl).ini, is childish and futile.
In my opinion a text-mistake or incorrect expression is the cause of all
difficuIties, for kla and upasarjana do not form a coherent unity and the
latter term, moreover, is already included in pradhna of s. 56. IE we now
accept kalopapade to have been the odginal reading, this compound would
exactly refer to the 3d adhyya (cf. 3, 2, 84, bhiite; 3, 2, 123, vartamne;
3, 3, 3, bhavi~yati, and the adhikra-sutra, 3, 1. 92, upapadam ... ).
Thus we have arrived at a coherent interpretation, which can be sum-
marised in the two following theses:
1. The interpolation consisted originally only in s. 53, an afttagonistic
opinion current in the old grammarian schools and weIl explained and
refuted by Pataiijali. lts insertion af ter 1, 2, 52 is due to the artificial
(krtrima) term lup used in su. 1. 2. 51.
2. Afterwards the term salpjii was misunderstood. and a long series
1) Blot out als in the edition.

The methods and mannerisms of composition in Pr;tini' s A~tdhyyi 59
of objections were interpolated by school masters, who did not understand

the worth of Pr;tini's researches.
80. Th e not ion pramnatva i n I i n 9 u i s tic sc i en c e.

P r;t i n i a n d h iso P pon ent sin s U. 53~57.
The relation between Pr;tini and his opponents with reference to gram-
matical pramnatva can be iIIustrated by the history of Iinguistic science
in Europe.Before the foundation of comparative Iinguistics the aim of the
grammarians as pupiIs of the old Greek sophists and philosophers was to
teach how to use one's language weil' and effectively, the grammar of the
eighteenth century still being a normative science. But by the discovery of
the coherence of the Indo-"E uropean languages the interests of the
scientific Iinguists became directed towards the historical development of
language and with the help of phonetics as a branch of physics and the
mechanistic psychology of Herbart they tried to find the factors of th is
development. It was, as a comparison by De Saussure precisely expresses
the state of things, as if botanists were studying the morphology of a
plant by taking only lengthwise diagrams of its branches and forgetting to
look at any transverse diagram. It is true, the writers of schoolgrammars
c1ung to the old methods, and some scholar or other Iike Delbrck tried to
unify old and new methods by writing a comparative compendium of
syntax, which, however, neither by its psychological premisses could
explain much, nor satisfy the demands of full description feit by specialists.
So the posthumous work of De Saussure came to light, in which the
au thor tried to unify and correct the old unsatisfactory methods. Next to
the historical treatment followed by the nineteenth century schol ars we are
in need of a descriptive method, which no more strives at laying down
rules born in the theoretical brain of a legislative grammarian, but searches
the rules that are unconsciously and instinctively active in the minds of the
people as a Iinguistic unity.
And it is this theoretical interest which we of ten can detect in Pnini's
researches, but miss in the schoolmaster's spirit and easiness of his
opponents, who interpolated su. 1, 2, 53-57.
81. A n a I y sis 0 f ad h y. 1 p . 3.
The 3d pda of the Ist adhy. has only one subject: the discussion of the
krakas (e.g . kartar, s.. 14; karman, su. 13; abhiprya, cf. the terms
'tmane' - and 'parasmai' -padam ) and bhva (sU. 13). as expressed by the
conjugational terminations . The anubandhas used in this connexion by the
Dhtu-Ptha are an additional anudtta-vowel "after a consonantic root
and of n af ter a vocalic root for the indication of the 'media tantum' (1. 3.
12) and an additional svarita-vowel or Ti for the indication of the 'genere
mobilia' (1, 3, 72) whilst the absence of these anubandhas 1) indicates the
1) This means that roots en ding in a consonant receive as activa tantum the addition
of an udffa-vowel ; See Liebich. DhP. 25.
60 Studies on PJ].ini' s Grammar
'activa tantum' (1, 3, 78). The subject is announced by su. 1 'dhtava!)'.

From this follows that in the passage 2-9 the anubandhas of the Dhtu~
Ptha take the first place and the anubandhas of the Stra~Ptha only
receive attention by associative extension (cf. 70). Bhtlingk, therefore,
misunderstood PJ].ini's intentions, wh en he left out all dhtu~examples in
his comments to su. 3; cf. Vi~k, DhP. 2, 38, Liebich's edition p. 16; Vd~p
2. 50, 1.1.; Vdhe~t I, 951, p. 14 etc., and Liebich 36---43. Further, we
notice that the stras 3, hal antyam, 5, dir iii~tu~4ava!), and 9, tasya lopa!),
form the framework of the passage, whilst su. 4 and 6---8 are insertional.
The stras 10-11 form ei th er an associative or distinctive insertion.
With regard to its character p. 3 may be described as a closed and
independent discussion (cf. 75 no. 3) of one subject.
82 . A n a 1 y sis 0 f a d h y. 1 p . 4.
I, 4, 1-2. Adhikra~sutras regulating the mnemotechnical constructin
of I, 4, 1-2, 2, fin .; cf. the interpretation given of these stras 30.
1. 4, 3-9. Technical terms bearing on the declension of the i~, h u~ and
i~stems; the passage has the character of a bridge between the general
stra of the declinational endings (div. IB, SP. 4, I, 2) and div. 11 , see the
cross~references in Bhtlingk' s translation.
1,4, 10-12. Associative or distinctive insertion. Definition of light and
heavy syllabIes, (cf. SP. I, 2, 27 on vocalic quantity) . IE the insertion is
due to associative influence, th en the distinction of i~ and i~stems etc. has
led up to it.
l, 4, 13-20. Definitions and general rules about afzga, pada and bha. The
term ariga bears on all forms of derivation with the inclusion of conju~
gation and declension. Pada has (1) a wide meaning ('word') and (2) a
limited technical meaning with reference to dedension and denominative
derivation of roots and noims; bha is limited to declension and denomina~
tive derivation of nouns. The passage has the character of a bridge between
div. IB and div. 11 (cf. 40-46).
1,4.21-22. Definition of plural and dual (cf. SP. 1,2,64) as an intro~
ductive associative insertion before the discussion of the krakas.
I, 4, 23-55. Discussion of the noun-cases as adjuncts to averb; this
theory of the krakas is closely related to the syntax of the noun~cases
given in 2 p. 3, forming with this section a discussion of closed and inde~
pendent character.
I, 4, 56---98. The theories of the avyavas, especially the upasargas, gatis
and karma~pravacaniyas. The section has a 'general character', being im~
portant for the theory of the noun-cases (2 p. 3), the formation of krts
with a gati as upapada (adhy. 3) and sentence-accent (8 p. 1).
1,4, 99-108. Morphological definitions of the active and middle verbal
terminations (cf. SP. 3, 4, 78) and participIes, of the grammatical number
of finite verb and noun; the term vibhakti. Concord in 'person' between
The methods and mannerisms of composition in PI).ini's A1?tdhyyi 61
finite verb and subject; the notion upapada . The section, dealing both with
conjugation and declension, forms the thematical ending of p. 4.
1, 4, 109-110. Saf!1hit and avasna defined. 'Oistinctive insertion'.
Thepada is characterised by the great number of subjects touched upon:
morphology of declension (3-9), terms of grammatical analysis (13-20),
theory of the krakas (23-55) and avyayas (58-98) ;vibhaktis of verb
and noun. In this feature the pda is parallel with 2 pa. 4.
83. A n a I y sis 0 f ad h y. 2 pa. 1 a n d 2.

The two pdas, dealing with the theory of the compounds, have the
character of a closed and independent discussion.
Divis. stra Sub j e c tof a d h y. 2 (p . 1 a n d 2)
adhik. 1. 1-4 Adhikra~stras.
A 1. 5-21 The indeclinable compound (avyayi~bhva).
B 1. 22- In general: the declinable compound (with the ex~

2. {in. ception of prsa;lgika~stras).
Ba 1. 22- Tatpuru~a (one member determines the other; as a
2.22 rule upasarjana precedes)
" i 1. 22 sq. Introductory; the dvigu as species (cf. 33 note I) .
.a 24-48 Kraka~relation between the two members.
.. fJ 49-71 Krma~dhraya; (s. 52: dvigus)
.. r 72- Idiomatical and peculiar tatpuru~as; first member
2. 7 expressing the situation of a part (Engl. 'fore~arm')
or the fraction (Engl. 'ha' penny' ). Preceding mem~
ber governs subsequent member: prpto jivikm >
prpta~jivika~; mso jtasya (cf. Ksik and Speyer
128 Rem. 2) > msa~jta~; the negation com-
pounded with noun or krt: abrhmaTJa-, akurvant~);
i~ad with a noun that is no deverbative .
.. cS 2. 8-17 $a~{hi~COmpounds; 8-9 positive rules. 10-16
negative rules. 17 exception to an exception .
.. E 2. 18-22 Tatpuru~as with avyaya or upapada (3rd adhy.) as
preceding member, cf. 29-60.
Bb 2. 23-28 Bahuvrihis.
Be 2. 29 I Dvandvas.
62 Studies on Pl)ini' s Grammar
Divis. stra Subject of adhy. 2 (p. 1 and 2)
C 2. 30-38 Order of the members of the compound. General

rule about the place of the upasarjana, and the
exceptions 30 sq.; order in dvandvas (Be), bahuvr~
his (Bb) and the karma~harayas (subclass of Ba).
Notice: 1. the separation of karma-dhrayas (38)
from tatpuru~as (30); 2. the retrograde order of
dvandva etc. Por 2. 2, 35 cf. SP. 6, 3, 9 sqq. and
Speyer 224 Rem. 2).
84. W a c k' e rna gel' s d i v i s ion 0 f t h e co m p 0 u n d s.

Wackernagel' sdivision of the compounds (Altind. Gramm. 11 1) is
mainly based on Pl)ini's treatment of the subject, with only a few
alterations:
A. Dvandva , p, 149-173.
B. Determinativkomposita (tatpuru~a), p. 174-272
a. mit Nomen verbale als Hinterglied (cf. specially SP. 2, 2. 18 sqq.
and 3rd adhy.), p. 174 sqq.
b. mit Adje1.<tiv als Hinterglied, p. 232
c. mit Substantiv als Hinterglied. p. 241.
C. Bahuvrhi, p. 273-307.
D. Komposita mit regierendem Vorderglied, p. 308-321.
E. Unregelmssige Verbindungen. p. 321-329.
Critical remarks. - As to Wackernagel's division between Bb and Be,
compare such rules as SP. 2, 1. 30 gUTJ.avacanena. - In class D Wacker~
nagel distinguishes two sub~classes a. Prverbien und Adverbien regieren
Hinterglied; b. Verbalform als Vorderglied. With reference to class Da
the author himself recognises that we have to do here with a kind of
bahuvrhis; consequently th ere are two classes of bahuvrihis with preceding
karma-pravacaniya:
. the class of adhijya~ 1) (said of the bow to which the string is
fastened), Ef':rems (of the mother in whom the child lives);
p. the class of anu/oman 2) (going along. anu. the growth of the hair),
EVfjf'O~ (being in the country. living amidst the people).
Both classes have dhtu~lopa in common. but differ in the direction of
the relation expressed by the karma-pravacanya. But besides Wacker-
nagel' s inconsistency in his classification it seems doubtful whether a
karma-pravacaniya may be said to govern a noun-case ( 28).
To his class Db ('Komposita mit regierender Verbalform als Vor-
derglied') Wackernagel observes (11 1. p. 315 120a): 'Von den
1) Wackemagel 11 p. 280 110 a; Hirt. Idg. Gr. V 232 in fine.

2) Wackemagel 11 p. 308 118; Hirt V p. 384 235.
The methods and mannerisms of composition in Pf.lini' s A~tdhyyi 63
Grammatikern nicht als besondere Kategorie anerkannt, aber in allen

Perioden des Altindischen belegt'. We may, however, consider these
compounds to be closely related to Wackernagel' s class Ba (Pf.lini' s krt-
compounds of the 3rd adhyya). although differing from them in the order
of the members (cf. Hirt, Idg. Gramm. V p. 394 238). For instance
compounds such as jahi-stamb (h) a- or QXfXllXO~ could be described as
tatpurw~as with reversed order.
With these notes I hope to defend Pf.lini's conception in its historical
value, but it is far from my intention to depreciate the researches by
Von Schroeder and Jacobi on the historical origin and development of the
compounds.
85. A n a I y sis 0 f ad h y. 2 P . 3.
S . I-fin. Syntax of the noun-cases ( 23 sqq. ) . The section has the
character of a closed and independent discussion .
86. Analysis of adhy . 2 p. 4 .

2, 4, 1-31. Grammatical number and gender of compounds. S . 1-16,
rules on the singular number of [samhara-]dvigus (SP . 2, I, 51, 33
note I) and dvandvas; s. 17-31, rules on the gender of compounds; the
Ksik explains sa in su. 17 as rderring both to the samhara-dvigus and
the singularised dvandvas; su. 18. avyayibhvas taken as neuter for
morphological reasons, see SP. I, 2, 47; su. 19-25, cases enumerated in
which a second, feminine noun becomes neuter, (compounds expressing a
collectivity); su. 26-31, the gender of [dual or plural ] dvandvas and
tatpuru~as . Since Pf.lini gives no rules for the gender of nouns, the section
should be considered to be an appendix to adhy . 2 p. 1 and 2; for the rest
it has the character of a closed and independent discussion.
2, 4, 32-34. Substitution (adda) of stems. Enclitic sterns of the
demonstrative pronouns.
2, 4, 35-57. Root-substitution or heteroclitic conjugation.
2, 4. 58-85 . Various subjects under the common heading luk: su.
58-70, rules on patronymics, dynasty-names etc.; su. 71. loss of case-
en ding a. before denominative root-suffix and {J. in a pratipadika (esp. in
the first member of a compound); s. 72-81, the derivation of the
thema tic present, of different aorist-forms, and the periphrastic perfect;
s. 82-84, the format ion of indeclinables; su. 85, the periphrastic future .
Strictly speaking the sutras 83 sqq . treat of desa, and no longer of luk.
Although in appearance the pada only deals with vyakti and vacana of
samsa, dda of stem and root; and luk, it really contains the discussion of
very many various subjects, being in this respect parallel to adhy. 1 p. 4.
87. Su m mar y; gen era I con c I u s ion s ab 0 ut th e

int e rna I c 0 m pos i t ion 0 f t h e a d h y Y a s l a n d 2.
The detailed analysis given in 77-86 allows the following con-
clusions:
64 Studies on P~in s Grammar
The author has given complete pdas (I p. 3; 2 p. 1-3) to subjects

which demanded an extensive discussion. In the pdas I, 1 and I, 2 the
most fundamental grammatical notions are explained. The shorter dis-
cussions are placed at the end of 1 p. 2 or coUected in the fourth pda of
the first and second adhyya. The last mentioned Eact combined with the
character of l. 4, 109 sq. as a 'distinctive insertion' ( 71 no. 2) proves the
two adhyyas to be unities in respect to composition (cf. 73 sq., IAa_and
lAb) .
Finally, taking into consideration that the ma in subjects of grammar are
its fundamental notions and the theory of declension, conjugation, com-
position. indeclinability and derivation, we can give the following concise
table of the first two adhyyas:
General subjects 1. 1 1.2 1.3 1.4 2,1 2, 2 \2, 3 \2, 4
fundamental notions - -
theory of declension - - - -
" "
conjugation - - -
" "
composition - - -
" "
indeclinability - -
88. T h epI ace 0 E t hef i r s t t w 0 a d h y Y a sin t h e

p I a n 0 f c 0 m pos i t ion 0 f t het 0 t a I wor k.
The discussion of the pada-bha-passage and the ku-gati-passage has
proved that the first two adhyyas must have formed, Erom the very
beg inning a part of the Stra-ptha.
Although several parts of these adhyyas have a general character (e.g.
1 p. 1 and l. 2, 1-46) or form a bridge between div. IB and div. 11, on
the other hand very extensive portions, such as the discussion of the
compounds or the noun-cases, agree in character completely with the
sections which make up adhyayas 3-5. Therefore we may symbolise the
composition of the Aljtdhyyi by the formula: (1, 2; 3, 4-5) . (6-8).
In other words, the first part of the argumentation needed Eor proving the
scheme of division given in 73 sq. has come to a close.
89. Th e di c hot 0 mie a I d i v i s ion 0 f th e a d h y Y a s

6--8. E x pos i t ion 0 E t h e pro bIe m.
Adhyyas 6--8 Eorm a unity of composition when compared with adhy.
1-5, the group 1-5 discussing the components of the words and their
meanings and the group 6-8 explaining the coalescence of word-com-
ponents into words and of words into sentences.
The first division of these three adhyyas into the group 6--7 and the
singJe adhyya 8 is evident, for, although the distribution of the subjects
over these adhyyas cannet be described with accuracy in a short formula,
The methods and mannerisms of composition in P~ini' s Al?tdhyyi 65
yet overlooking the difficulties caused by P~ini's mnemotechnical devices

and his definition of the term pada. we can say in general th at adhyayas
6-7 explain the coalescence of the wood~components into words. and
adhy. 8 the further development of the sentence.
Similarly the dichotomical division of adhy. 8 into 8. 1 and 8. 2-4 is
clear. the first pda treating of the sentence~accent. and the three last
padas (with exception of some prsatigika~stras) of the articulative~
phonological changes of the words as parts of the sentence.
The problem which now remains. is how to prove the dichotomical
division of 6-7. This case is not easily solved. owing to the author's
tendency to attach greater importance to mnemotechnics than to logical
consistence.
90. Th e f i vet hem a tic a I sec t ion sof a d h y. 6 a n d 7.

Putting aside for a while the question wheth~r there is any dichotomy
we can prima facie distinguish five sections:
a. Adhy. 6. 1. 1-157. general phonological rules.
p. Adhy. 6. 1. 158-222. accent of uncompounded words.
y. Adhy. 6. 1. 223-6. 2. fin., accent of compounds 1).
~. Adhy. 6. 3. 1~fin .. morphology of the first member of a compound;
a few prsafl.gika~sitras . (97 sq .. 110) refer to the second member; sU.
111-113 are inserted for the sake of mnemotechnic economy (the anuv~tti
of dirghal;t being made use of); the sutras 131-137 form likewise an
insertion and probably an interpolation.
E. Adhy. 6. 4. 1-7. 4. fin. has for adhikra~sUtra s. 6. 4, 1 afl.gasya,
whereas a subsection (6. 4. 129~fin.) has moreover the adhikra bhasya.

The first thing that strikes us is that section y forms a concatenative link
between p and ~. s,o th at we can combine p + y as theory of accent or
y + ~ as theory of compounds. su. 6. 1. 158 'anudttaf!l padam ekavarjam'
serving as adhikra for the sections p + y. and su. 6. 1. 223 samsasya as
such for y + ~. In the second place it is worth while noticing that the
adhikras afl.gasya and bhasy.a stand iJ! correlation with padasya of 8. 1. 16
provided that pada is taken here in the sense of a noun~stem under certain
conditions. (cf. the pada~bha~section. SP. 1. 4. 14 sqq.). However. since
this correlation is only secondary because of padasya in the first place
denoting the complete word as part of asentence. the question of examina~
tion has only to do with the grouping of the sections a-Eo IE P~ini. when
construing his plan of composition. has been influenced by the supposed
tendency to dichotomy. was then the antithesis between (y + <5) and E.
nominal compositionagaiast nomina I derivation. or that between (P + y)
and (<5 + E). accentuation against articulation. foremost in his mind?
1) Indicating the flrst syllabie of a stem by a. the last by u. the origina1ly udtta
syllabie by i. we can distinguish six types of compound-accentuation: iu-aiu, aiu-aiu etc.
or I a etc. and 11 a etc.
5
66 Studies on PJ).ini' s Grammar
In my OpIDIOn, the decision falls in favourof the latter alternative; in

other words the adhyyas ~7 are divided into:
i. introduction: 6, I, 1-157, general morphological rules,
a. sections (P + r): 6, I, 158-6, 2, fin .. accent of word and compound,
b. sections (b + E): 6, 3, 1-7, 4, fin., morphology of composition and
derivation.
To support th is conception two arguments may be adduced: 1. in this
way the introdutory part, exempt from the dichotomical division, becomes
shorter (a instead of a + P); 2. th ere arises parallelism in the construction
of the adhyyas ~7 and the 8th adhy.: 6, I, 158-6, 2, fin. II 8, 1. theory
of accent, and 6, 3, 1-7, 4, fin. II 8, 2-4, theory of articulative-phono-
logical and morphological coalescence, such stras as 8, 2, 4-6 only being
of a prsangika-nature.
But was the theory of accent in deed thus separated by PJ).ini from the
rest of word-coalescence? And why was it necessary to break the regular
dichotomy of the total composition by placing the stras 6, I, 1-157 as
an introduction?
91.
. A c c ent a s a I i n 9 u i s ti c fa c t.
PJ).ini gives the definition of the pitches of accent (udtta, anudtta,
st'arita and sannatara) in the passage I, 2, 29-40 following the stras
27-28, which define the vocalic quantity. Accent or glottic modulation
and duration are thus likewise qualities of the speechsound. And likewise
accent in the theory of derivation and composition should be taken as one
of the factors in the process of coalescence.
In support of this conception, moreover, the authority of De Saussure
may be quoted (Cours de linguistique gnrale, p. 103) 'Ia syllabe et son
accent ne constituent qu'un acte phonatoire', and 'Ie sighifiant 1) tant
de nature auditive, se dcoule dans Ie temps seul et a les caractres qu'il
emprunte au temps: a) il reprsente une tendue, et b) cette tendue est
mesurable dans une seule dimension, c'est une ligne'.
But are these statements in deed so absolutely true as De Saussure puts
them? The syllable with its accent may be called an articulative, and thus
psychical act, and as such a strict unity. But in biology and psychology
the rule holds good that the unity is strictly a unity, i.e. no mere ;lrith-
metical sum or mechanical conglomeration of parts, and yet the unity goes
together with the existence of the parts. The parts interact mutually and
are likewise determined by the totality.
Thus the syllable is a unity and one psychical art and at the same time
this one act contains a plurality of innervations - innervations of muscles
round oral and nasal cavities, the glottis and the apparatus of expira-
tion; and the accent as produced by such a partial innervation receives also
a certain independence as an auditive unity. The syllables follow one
1) The word is a unity (signe) with an auditive-articulative aspect (signifiant) an~

a notional aspect (slgnifl).
The methods and mannerisms of composition in Pt;tini' 5 A~tdhyyi 67
another and together form the words and the stream of spoken language.
and this one stream is again a strict unity. one psychical act. in each of its
utterances or sentences. And yet in this main current the accents of the
successive syllables form an under~current with a certain independence of
its own. And 50 this accent~current can subject itself to stylistic forms.
and the spoken language becomes a phrase that is sung; and th is song.
although still a unity in the strict sense of the word and therefore one line.
is at the same time the duality of text and melody. And in the same way
as in polyphonic music many melodies simultaneously form many lines.
although the time in which all th is happens is one dimension. 50 does the
current of accent flow onward within the wide current of spoken language.
for. do not let us forget. the science of mechanics. however beautiful and
sublime it may be in itself. does not represent the only shape that human
thought can take. and it is more than time for the students of moral
sciences to desist from a bigoted adoration for physics.
Pt;tini. therefore. was right in defining accent as a quality of the vowel
(or syllable) and at the same time treating it as a linguistic fact with its
own individuality.
92. T hei n t r 0 d u c tor y c har act e r 0 f t h e pass a 9 e

6.1.1-157 and the dichotomical division of adhy. 6---8.
Adhy. 8 p. 2-4 do not only teach the sentence~coalescence but for
mnemotechnical reason~ also i.a. the coalescence of a stem in pada~condi~
tiOll with its suffixes and the substitutions and rejections of consonants
which have to do with this subject. And it is this deviation from the logical
scheme owing to mnemotechnics which again forced Pt;tini to break the
dichotomical division of adhy. 6---7 by the introduction 6. 1. 1-157. in
which he treated of the vocalic sa'!1dhi and other subjects that are indepen~
dent of su eh distinctions as pada. bha. atiga and samsa.
And although a fuller analysis of the last three adhyyas with reference
to the distribution of their rules is desirabIe. for the present the thesis may
be accepted that the composition of the last three adhyyas obeys the
formula: (i=6. 1. 1-157). {(Aa=6. 1. 158-6. 2. fin . ). (Ab=6. 3.
1-7.4. fin)}; {(Ba=8. 1). (Bb=8. 2-4)} : See 73 sq.
93. W a c k e rna gel '5 j u d 9 men ton t h e c 0 m pos i t f 0 n

o f t h e A ~ t d h y Y i.
WackernageI. the grammarian. in showing in 50 many places in his
Altindische Grammatik his great detailed knowledge of the Hindu gram~
matical scienee has done full justice to Pt;tini as a trustworthy and
competent witness oflinguistic facts. But the A~tdhyyi is more than a
depository of facts; as a monument of ancient Hindu science it demands
from us a historiography obeying the rules of historical piety and perspec~
tive. In this respect WackernageI. although in a less offensive manner than
Whitney. faUs short in his judgment on Pt;tini (Altind ~ Gramm. I p.
LXII) , wh en he expresses himself in the following words: 'P~ini selbst

nennt eine Anzahl Vorgnger. Ja sein Werk scheint wesentlich eine Neu~
redaktion eines. vielleicht vor ihm schon mehrfach umredigierten, gram~
matischen Grundwerks zu sein. Ein rationeller Grundplan schimmert
durch ...... Aber dieser Aufbau wird bestndig durch einzelne Regeln oder
ganze Reihen von Regeln durchbrochen, oEfenbar, weil man nach Bequem~
lichkeit an beliebiger Stelle einschob, was bei weiter entwickelter Forschung
neu hinzukam. Auch mochte die Mglichkeit Worte zu ersparen etwa dazu
fhun, Regeln aus ihrem natrlichen Zusammenhang herauszureissen und
anderswohin zu versetzen. '
So then Wackernagel. a great grammarian, could not, as historiographer,
see anything in P~ini save the dimlight of a rational plan and the deep
night of historical chance, miserliness in words, love of ease, a breaking
down of natural connections, the whole a whirlpool of caprice. Indeed,
those ancient Hindu grammarians have committed crime in not being
bom as modern. logical. strenuous and capriceless European scholars,
writers of books as weil arranged as the safes of a bank! However, the
explorer who has in fact done so much to give us a full understanding and
ajuster appreciation of Pnini, really deserves something better than an
ironical tirade. Let me say then that we, Europeans of the twentieth
century, although grateful to the genius of the ancient Greeks, which with
the help of geometrical textbooks and well~balanced political speeches has
taught us the power and beauty of the logical intellect, also realise that
there is something else in the world of thought besides logic. Contemporary
psychology has taught us that creative synthesis does not merely belong to
critica1 intellect and consciousness, but that also our dreams have their laws
of composition, that the mind of the child, though weak in logical dis~
crimination, possesses a gift of construction, which shows itself in the
original touches of its traditional games, and that the artist is artist just
because there is still so much of the child in him.
When, then. in the A~tdhyyi we meet with a wonderful. puerile and
ingenuous anubandha~system, with mnemotechnical interests weighing
heavier than auy log ic in the world, with associative digressions in unlimited
number, then we confess: we adore P~ini because he reveals to us the
spirit of India; we adore India because it reveals to us the Spirit, the Spirit.
APPENDIX.
94-97. Discussion of some technical difficulties. English translation

of French and German quotations.
94. The definition of upasarga ( 19).

According to the Ksik the expression kriy-yoge must be taken as a
supplement to the su tras and specially the adhikra-stra of the passage
I, 4, 60---79. Thus, whilst the passages, quoted in 19, compel us to
interpret kriy-yoge in s. I, 4, 58 sq. by 'in connexion with [the notion of]
an action, [either expressed or implied ] " kriy-yoge. when supplement
to su. 60 has the narrower meaning of 'in connexion with [the notion of]
an action [expressed]' = .in connexion with [a word expressing] an
action, [i.e. either a finite verb or a deverbative]', cf. 17 no. 3 and Note.
Undoubtedly the redaction of the Siitra-Ptha is logically not quite satis-
factory here.
95. Th e m e a n i n 9 0 f t h e pratyhra sup ( 45 a n d

passim).
In the first place sup is the technical designation of (A) the declinational
terminations (SP. 4, 1, 2 and 1, 1,71). Moreover, according to 1, 1,72
sup indicates (Ba) a declined noun, and according . to the rules for
pratyaya-lak~aTJa (esp. 1, I, 62) it likewise includes (Bb) the nominal
stem, e.g. as the preceding member of a compound (1. 2, 46 and 2, 4, 71)
and Be the indeclinables (2, 4, 82).
So, when in 2, 1, 4 the rule is given: [sup] saha sup and we read in
2, 1, 6 avyayam as a substitute for the first sup in this formula, th en th ere
is, strictly speaking, no question of an exception. but only of a limitation.
A similar remark holds good for su. 2, 2, 18, wh ere according to algebraic
reasoning it is not qQjte accurate to say (cf. 45 p. 37 I. 7) that the
second sup of sup saha sup is put aside by Pl)ini, rather should we
understand that th is sup, in anti thesis to what precedes, is extended to both
the meanings Ba and Be.
However, inconsistences will necessarily of ten arise owing to the
artificiality and unfitness of algebraic expression for describing linguistic
facts. And so it is no matter of surprise that the Ksik, too, takes su. 2, 2, 18
as an exception to 2, 1, -4 in its interpretation: ku-gati-prdayalJ samarthena
(cf. 2, 2, 1 and 31) Sabdntarena saha nityam samasyante.
96. Th e well-planned construction of the

A~tdhyyi illustrated by the interpretation of
SP. 1, 1. 3.
Th insufficient number of cross-references and the of ten too limited
choice of examples in BhtJingk's translation do not always allow us to se
the ingenuous construction of P"ini' s work.
In order to interpret su. 1. 1. 3 iko gU1Ja~vrddhi . i.e. the gu~a of i and i

(Siva~Stra read in combination with 1. 1. 69). u and . and r. short or
long. e. o. a[r] and their Vfddhi is ai. au. [r] we ought to reckon with the
following facts:
1. the vocalic ! has to be omitted since the only instance Vk!p is treated
by P~ini' s method of theoretical fiction as if it were krp (DhP. 1. 799
etc. ) ;
2. i. i are replaced by e or by ai. and u. by 0 or by au according to
1. 1. 50. Since P~ini takes it for granted that the gu~a and vfddhi of r
contain an a~sound. we may surmise th at his articulation of the r~sound
had an a ~colouring ('neutra]' colouring) . The addition of r is taught by
1. 1.51. For the substitution of r by ir. ur. ir. r. cf. 6. 1. 111; 7. 1. 100 sqq.;
8. 2.77;
3. although a is the gu~a of a itself and its vfddhi. yet the sutra is
not formulated as * ako gUTJa~vrddhi in view of such stras as 7. 2. 1
(where Bhtlingk supplements: 'fr ein auslautendes i. u. r nebst ihren
Lngen. cf. 1. 1. 52). and 7. 3. 83 sq.
97. Eng 1 i s h tra n sla t ion 0 f Fr e n c h a n d Ge r man

quo t a t ion s 0 n b e hal f 0 f th e H in dur e a der.
Page 33 38. Ku . an indeclinable called gati. and pra with its associates
are likewise compounded. 19. Likewise a supplement if this be no fini te
verb. - 20. Also a supplement with an indeclinable. but only wh en th is is
an absolutive ending in am. - 21. Also an instrumental and one of the
following cases but this is not necessary. - 22. Likewise with an absolutive
in tv .
Page 36 44 i.E. 'a well~planned and not only virtuosic but also ingenuous
text~book .
Page 66 91. 'The syllable and its accent are only one phonetic at.
and the significant being of an auditory nature only flows onward in time
and has the characteristics which it adopts from time. a) it represents an
extension and b) thatextension is measurable in one single dimensiori. it
is a line.
Page 68 93. ' P~ini himself mentions a number of predecessors. His
work. indeed. intrinsically seems to be a new version of a perhaps before
him frequently altered grammatical groundwork . . . A rational basis
glimmers through . But this constructive work is constantly interrupted by
single rules or by whole rows of rules. obviously while what was newly
added by further developed research was at will inserted for convenience'
sake. The possibility to spare word., could also lead towards distorting
rules from their natural connex ion and transposing them somewhere else .
Ex pre s s ion 0 f th ank s. - Dr H. E . Buiskool hasassisted the

author by reading the proofsheets and suggesting here and there an inter-
pretation of his own.
TABLE OF CONTENTS.
Bibliographical table 5
DIVISION I.
The grammar of the indeclinables.
I. The indeclinables in general . . . 7

1-14. Division of the indeclinables according to etymology and
function . . . . . . . . . 7
15-20. The technical terms of the grammar of indeclinables 14
11. Karma~pravacaniya and the theory of noun~cases . , 18

21-28. The grammar of the karakas, noun~cases and karma~
pravacaniyas. The coherence of the adhyayas 1 and 2 . 18
111. Gati, or indeclinable used as a prefix of a deverbative noun

or indeclinable ..... . . .. 25
29-33. The adhikras of the ku~gati~passage (2, 2, 18-22) 25
,34-37. The notion upapada and the 3d adhyya . 29
38-39. Bhtlingk's translation of the ku~gati~passage . . . 33
40-46. 'Bridge' ~character of the ku~gati~passage and the pada~
bha~passage. Logical elegance of PJ).ini's composition.
Interpretation of s. 2, 2, 18 . 34
47-49. The cross~references between the ku~gati~passage and
the 3rd adhyya. S. 2, 2, 19-22 interpreted . . 38
50-56. The Mahbhl?ya on the ku~gati~passage 40
57-60. Morphological and phonological consequences of the ku~
gati~passage . . . . . . . 43
IV. Gati and nipta, or the close adverb of the finite verb and
the modal and conjunctional particles 45
61-68. Analytical table of adhy. 8 p. 1. Worth of PJ).ini's work 45
DIVISION 11.
The methods and mannerisms of composition
in PJ).ini' s Al?tdhyayi.
69-93. The fundamental notions of Hindu grammar. Main and
subsidiary principles of PJ).ini' s form of composition.
Tendency towards dichotomy in the ma in divisions . . 49
72 Studies on PI).ini' s Grammar
APPENDIX.
94-97. Discussion of some technical difficulties. English trans~
lat ion of French and German quotations . . . . . . 69
ANALYTICAL TABLES OF PANINI'S ASTDHYAYI.
Concise table of the entire Aliitdhyyi. 73 . . . . . 51

Full analysis of adhy. 1 and 2. 77-87 . . . . . . 55
Analytical table of adhy. 1 and 2. general subjects. 87 6i
2 p. 1 and 2. 83 61
3. 35 30
8 p. 1. 62 . . 45

Studies On Panini's Grammar

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Studies On Panini's Grammar

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Studies On Panini's Grammar

Uploaded by

Copyright:

Available Formats

STUDIES ON

VERHANDELING DER KONINKLIJKE AKADEMIE

SECOND PRINTING: 1963

1. Sou r ces a ndd i c t ion a r i e s.

2. S tud les a n d 9 ram mar s, c h ron 0 log i c a II y a r r a n 9 e d.

Not e. T h e t e x t u a I tra d i' ti 0 n 0 f th ie Ga ':l a - P ~ h a. See

I. THE INDECLINABLES IN GENERAL.

1-14. Divisions of the indeclinables according to etymology and

1. Synoptical table of the two passages on the

b. indeclinables and some noun~cases, and derivatives from nominal

2. C har act e r 0 f t het w 0 en u mer a t ion sof i n d e~

3. T h e cIa s s i f i cat ion 0 f wor d s i n t heB r h a d~

denominative and deverbative indeclinables being considered as a form of

4. Th e s t ra ~ li k e pas s a ges int h e svar~di~garya ( 1 I 1).

5. Th e ins t anc e s su mme d u pin th e svar~di~gaT.la

1) ~tas, SP. 5, 3, 7 sqq.

cIinables with no accent mentioned; C. 96---150, instances not given by

(ak~apari 2, 1. 10; akumram 11-16); other indeclinables or isolated

9. Case~isolation (GP. 254 nos. 17 sq. diva , rtrau etc.,

10. In dec I i n a bIe 5 con 5 i der e d tob e nou i! 5 wit h

I 1. T hes tra ~ 1 i k e pas s a gei n t h e cdi~g1J.a (G P. 85

12. The instances summed up (n the cdi~ga1J.a (GP.

13. Theprdaya1;L(GP.154,SP.I,4,58)in their different

sections the prdaya/:l in bahuvrhis are called upasargas (5, 4, 119; 6, 2,

the construction with a noun-case the use as 'free adverbs' for :

15-20. The technical terms of the grammar of indeclinables as used

16. T het e r m nipta, 1 II.

17. Th e ter m s upasarga, gati a n d karma-pravacanya.

1) Par in Bhtlingk's edition p. 113* is amisprint.

suffixes) coalesce into words, the upasargas in their adnominal function

18-20. Appreciation of PIfini's terminology.

18. T het e r m nipta i n i t s w i d e a n dna r r 0 w s i 9 n i-

1) Moreover, S. 6, 3, 131-137 is most likely an interpolation.

19. Ambiguity of the definition of upasarga (1,4,

Not e I. Th e ce r e bra 1 i s a ti 0 n 0 f r.a t th e beg i nni n 9 0 f

gatis summed up SP. 1,4,61-98 upasargas any other word

20. T het e r m karma-pravacaniya.

21-28. The grammar of the krakas. noun-cases and karma-prava-

an object, and the 'intentional' 1) relation in the psychical acts of wishing,

24. S y nop tic a I tab I e 0 f th e di f f ere n t ar r a n 9 e~

A. Mor p hol 0 9 i c a lor der :

25. T h ede fin i t ion 0 f t hen 0 min a t i v e a n d t h e

26. T h e cl e fin i t ion 0 f t heg e n i t i v e a n d t hei n t e r~

1) For the t8tpuru~ with ~J..relation. see 2, 2, 8 sqq. and 6, 3, 21 sqq.

27. Comparison of the passage 1. 4. 83-98 (the

28. T h e E u rop e a nno t ion 0 f pre pos i t ion c 0 m

1) Cf. the quotation from Delbruck at the end of this section.

genitive construed with an indeclinable shows a metaphorically extended

111. GATI. OR INDECLINABLE USED AS A PREFIX OF A

29-33. The adhikras of the ku~gati~passage (2. 2. 18-22).

1) Thus upari grhasya shows affinity to a construction such as grhasya pr~!he.

40--46.'Bridge'~character of the ku~gati~passage and the pada~bha~

29. T h e a d hik ras 0 f the ku~gati ~ pas s a 9 een u m e~

30. Interpretation of the paribh~ s. 1,4,1 sq.

paribha~ of 1. 4, 2 shows a great resemblance to the asiddhatva~

31. Int e r pre t a t ion 0 f th e ad hik ras u. 2, I, 1.

stems 1) or indeclinables as preceding member of a compound) must be fit: