What Is Karma? (Ki Karmeti) - An Exercise in Philosophical Semantics (Edwin Gerow)

E.
GEROW
WHAT IS KARMA (KIM KARMETI)?

AN EXERCISE IN PHILOSOPHICAL SEMANTICS
At a recent conference on « karman and rebirth »*, I was widely

suspected to have made a joke when I observed that the sense of
« karma » most familiar to me was that of the grammatical « direct
object»: ipsitatamam karma12. The «joke » of course presumes a com
monplace: that the senses of « karma » worth enquiring into must have
to do either with a reality principle (as when the world and the sacrifice
are both said to be a « karma ») or a condition of moral or ethical bon
dage (as when «karma » is said to be the mechanism of samsara).
Grammar deals with mere « words »; it cannot capture the seriousness
of the kosmos.
Yet the conference was animated by one of the laudable trends of
current research on India: that indigenous conceptual systems are best
enquired into through terms that are recognizably indigenous3. One,
therefore, runs a certain risk in laughing at those dimensions of a con
ceptual system that are most strikingly « indigenous », such as the fact
that « karma » means also « direct object » and is functionally equivalent
to our term « passive voice ». Our understanding of the term's cosmic
and ethical significance may in fact depend on our explaining that
« coincidence ».
And this is precisely the issue that I want to raise in this paper,
taking as my text Renou's well known reference to the grammatical
karma: «karman " action" (en tant qu’objet direct, objet du verbe
transitif) appartient aux milieux rituels ou le mot, depuis le RV, desi-
1. Workshop on Karma and Rebirth; Lake Wilderness, Washington, Oct. 22-23,

1976. See « Minutes » mimeographed, subsequently cited.
2. « kartur ipsitatamaip karma» Panini (A?tadhyayi) 1.4.49 « ...la chose que
I’agent souhaite atteindre par dessus toute autre (porte le nom de) karman (’’ objet-
direct")... ».
3. Cf. Marriott and Inden, Toward an Ethnosociology of South Asian Caste
Systems, to appear in David, The New Wind..., The Hague, Chicago.
88 £. Gerow
gnait l’acte par excellence, c'est a dire, le rite »4. But is the grammatical
meaning of « karma » simply a function and a reflex of the term’s other
(or more general) meanings, as this quote suggests, or is it also at the
heart of any proper understanding of those more general meanings?
Let us consider how we might inquire into that latter possibility as we
ponder another of Renou's remarks: adherer a la pensee indienne, c'est
d'abord penser en grammarien56 .
An inquiry into indigenous conceptual systems begins, whether we
like it or not, with words, which often we do not even translate —
«karma» (« action/act ») — and is thus in principle a grammatical
problem. The straightforward or first level question, « what does karma
mean? » if successfully answered will put us in command of an indi
genous dictionary wherein are delimited the term karma and its sym
biotic relatives (conceptual: such as dharma, yaga; or etymological: such
as kriya, kartr)\ and as well these terms' usage or capacity to combine
(their « syntax »).
But when we inquire into « karma » we use concepts indigenous to
our own standards of expression; the term may assume meaning and
functions that derive from the syntax and context of our inquiry. Even
when untranslated, the term absorbs senses from the English sentence
context that it does not or cannot have in Sanskrit and thereby loses
even the clarity of being an « unknown quantity ». For example, « Does
fate equal karma, or is karma used to explain fate?»7. By leaving the
term untranslated the « appearance » of a new technical term is created
in English itself: karma, which « appearance » only conceals the analy
tical fatuousness of the question itself. If « true », then indeed « fate
equals fate » and if not. « fate does not equal fate ».
Many of our questions are also naive in the sense that they must
ignore relations that are implicit in the Sanskrit vocabulary; and they
are equally moot in that they can, therefore, be answered «yes» in terms
of one set of lexical contigencies. and « no » in terms of another. One's
karma is indeed one’s dharma, if by « karma » we mean « kartavya »8
and it is equally not, provided we take karma in its siddha (« accom
plished, reified ») sense of « krta», distinguished thereby from one’s
dharma9.
And so from the Gita we take our third text: kim karma kim
akarmeti kavayo'py atra mohitah10.
4. JA, 1941-42, p. 157; reprinted Staal, A Reader on the Sanskrit Grammarians,

p. 464 (Cambridge, 1972). Also cited by Cardona, Panini, p. 231, who, nevertheless, calls
attention to the difficulties in simply equating the terms from the two milieux.
5. IC2, § 1519.
6. Often « duty/caste duty»; « (vedic) sacrifice »; « action/activity »; « agent».
7. Question 18, p. 3, Minutes of Workshop on Karma and Rebirth...
8. As in the Gita and Vedic texts.
9. As in the advaita texts and the physical treatises.
10. 4.16, cited N. N. Bhide, infra note 47.
What is Karma (Kirn karmeti)? An Exercise in Philosophical Semantics 89
This inquiry is about a grammatical problem and is grammatical by

method. It is, therefore, insufficient simply to propose removing such
terminological confusions as a mode of problem-solving. The inter-lan
guage character of the problem is the problem and cannot be eschewed
unless and until we all speak English. Where terms are not translated,
as in the naive question « what is karma »? one language (here Sanskrit)
is in effect made the object of inquiry.in a context where the other
defines the methodology and solution. In making such a distinction
between our « method » and their « reality » what we miss is that in
the present case there are no objects other than words: the « term »
karma is nothing but its use by Indians variously to organize and illu
minate their experience; the problem is signalled for us not by the fact
that the experience is necessarily different or « indigenous » (for there
would be in that case no basis for comparison; no simile for inquiry),
but by the fact that the term or its « translation » is functionless in our
experience, that is, not used to organize our world. This difference of
form (more than of content) suggests that our problem is more akin to
that of the possibility of different grammars; different organizations of
the « same » world: a world that, therefore, not only seems variable but
achieves distinction and precision in its variety. To a certain extent then
the inquiry into karma becomes thirdly an exercise in rethinking the
world (assuming for the moment karma is a crucial term in the Indian
« world »), and an exercise in giving up our own or « native » grammar.
Thus the « grammatical » problem insensibly becomes, for us, an ethical
and philosophical problem lI.
We derive from this introduction the three topics of 1) Lexicon,
2) Syntax, and 3) Kosmos for the main body of this paper: 1) the term
karma is important in Indian theory, which suggests that we examine
first the Indian lexicon; 2) the syntax of the term karma is an important
issue when the term is used in propositions or to form questions, which
suggests that we inquire into the linkages of karma to other terms;
specifically the grammatical « karma » that is itself used to explicate a
sentence model, or syntax; and 3) the semantics of the term karma
as determined by the grammarians may be crucial to our understanding
of its various contexts, or « worlds », which suggests that we examine
whether the wider or «non-grammatical » karma may not itself be
implied by' a grammar that has already been worked out by the gram
marians I2.
11. Cf. Wittgenstein, Phil. Unter. 1.19, Und eine Sprache vorstellen heisst, sich
eine Lebensform vorstellen.
12. Ashok Aklujkar suggests the following caveat at this point: « The theory /
view of karman that the first employment of the term in grammatical discussions
presupposed is to be distinguished from the theory that developed around karman
as a technical term. But speculation on the first is. equally interesting », Yaska, for
example, uses karman in the sense of « action ». See notes 21 and 42 and, in general,
discussions of the ritual origins of Indian technical terminology: cf. Cardona, p. 231.
90 E. Gerow
Each of these reference points is supported by an observation which

in isolation may appear paradoxical, but which when put in the context
of our problematic, encourages us in our inquiry: viz. (1) In the Indian
lexica, the « grammatical » sense of karma is invariably given first, sug
gesting its primary character13. (2) In the Indian grammar, the sentence
or assertion is both an act (in its own right) and takes an act (kriya/
karma) for its content14, which clarifies both the place of grammar
within ritual and the formal model of the grammar itself, which is not
merely actional, but « intentional » without being « volitional » 15. (3) It
is a grammarian, Bhartrhari, who, speculating on the cosmic-mystic
sense of « karma », establishes a coherence among the Indian views of
karma that derives ultimately from the notion’s grammaticality. It is
perhaps Bhartrhari who provided the rationale for the transformation
of a word for « action » (karma) into the status of a world- or reality
principle in Indian speculation, a status that words for « being » enjoy
in our own. And this surely is the general issue that focusses our interest
in the Indian’s « exaggeration » of karmic matters, and expresses best
the alienation we feel in the presence of the notion « karma »; for (to
capture its philosophical seriousness) we ought to be able to translate
it not as to Kpaypa but as to ov 16.
(A) Karma in the lexicon. A term suffused with as many ambi
guities — cultural and contextual — as karma surely needs definition.
In order to hold the cultural ambiguities to a minimum we examine
the voluminous Sanskritic lexical tradition. The fact that Sanskrit is
defined through Sanskrit establishes as purely as possible the Indian
contexts of relevance, and deletes the most troublesome source of ambi
guity: that we naively assume our own language as method.
Most useful is the compendium of Tarkavacaspati, the Vdcaspatyam,
which conflates in six encyclopaedic volumes of ca. 4000 pages the
Sanskritic lexical traditions. Compiled between 1865 and 1875 (?) by a
team of orthodox pundits, it reflects the intellectual fluctuations of the
medieval period without showing the least trace (save in its motivation)
of Western influence. In its third volume approximately fifty (pp. 1724-71)
13. Nils Simonsson points out an exception in the Nyayakosa.

14. kriyavacano dhatuh [Varttika ad P 1.3.1] «the (verbal) root expresses an
act(ivity) ». The verb [al idea] is of course for the grammarians the head of the
sentence, as against Mimamsa and Nyaya. See notes 32 and 39, infra.
15. The most generally applicable rules in the grammar seem conditioned by
and to explicate the speaker's intention, as for example, P 3.2.123 « varttamane lat».
« Les desinences du " lat ” valent quand il s'agit de l'actuel» (Renou, op. cit., p. 206),
or rather « on condition of (wishing to express) presence, the suffixes of the present
(are enjoined) ». It is the Nyaya that seems to have adopted a « volitional» model.
Our discussion of « bhava » in what follows will make this clear.
16. Cf. K. A. S. Iyer, Bhartrhari, a Study of the Vakyapadiya..., Poona, 1969,
esp. ch. 6 « On action » (kriya). Words for «being », when they do figure in the
discussion (as in the Varttika, «bhavavacano dhatuh», or YKska's, «bhavapra-
dhanam akhyatam », are invariably interpreted in actional terms (as Kaiyata, « bha-
vavacanah kriyamatravaci»; Iyer, p. 329).
What is Karma (Kim karmeti)? An Exercise in Philosophical Semantics 91
pages are devoted to karma and compounds beginning with karman-I7;

of these pages two entries occupy the bulk; karma itself (pp. 1730-37)
and the compound « karma-vipaka » (pp. 1742-70). Although the latter
entry is central to other concerns of the karma theorists, detailing 80
or so consequences in later births of various (mostly immoral) behavior
in this one, we pass it by here, as being itself consequential and not
crucial to an understanding of karma.
We find, sub karma, five contextual meanings distinguished in prin
ciple; that is, all the variety of meaning which the term has undoubtedly
had over ca. 35 centuries, is reported under five general headings, which
are (as we would expect rin an Indian undertaking) not historically, but
sastraically, motivated. As noted earlier, a meaning attributed to Vyaka-
rana, or the Paninian grammar, is given in primo loco; follow senses
said to pertain to the Vaisesika system (realistic pluralism or ato
mism, and one of the « six systems »), (3) Hari (i.e., Bhartrhari, the sixth
century linguistic philosopher who established grammar as a vehicle
suitable for the moksamarga), (4) to Mimamsa, under which are grouped
almost all the senses of the term karma pertaining to the Vedic period
and to the ritual traditions, and (5) agriculture (krsi) (though this latter
meaning is for the most part Vedic).
We will now inventory the various senses of karma, before pro
ceeding to the second part of this paper, which will attempt to assess
the significance of the emphasis on grammar (vyakarana) among these
meanings. In doing so we will somewhat alter the order as given in the
lexicon, proceeding perhaps in a way more suited to a Western sense of
order, from the least to the most grammatical meaning.
(5) Agriculture/krsi. At issue here is a sense akin to that of Hesiod's
« Works and Days»: and one which would appear to be closer to a
literal and valueless sense of « act» than any other. But even here the
term karma is related to the terms which enliven its « abstract » uses,
especially « phala », for the « fruits » of ploughing are indeed its conse
quences. «phalam karmayattam » 18 say the mimamsakas, playing on
both senses. And «phalasya karmanispatter lokavat»19. None of the
phrases adduced in establishing this sense unequivocally refers to
« ploughing » in such a way as to rule out an implication of the more
philosophical meaning. In fact, this « literal» sense of karma seems
17. There is often variation in Western texts between these two forms of the
same word. Briefly, « karma », is the fully inflected nominative singular (neuter),
whereas « karman-» is the so-called stem form, without any inflection. It is normal
to cite Sanskrit nouns in the latter form (Nala, not Nala-s), but evidently because
itis confusing to Western readers to think of a longer form as uninflected,
« karman » (and most other n-stems)constitute exceptions to the rule, We will try
to conform to this practice (cf. brahma and brahman).
18. « The fruit depends on the act» (p. 1737).
19. « Since the fruit is the result of the act, just as in the world » (ibid.), said
to be from Jaimini.
92 E. Gerow
necessary only to justify the explanation of its abstract meanings, « tad

yatheha karmaci tai okah kslyate » 20.
(4) The mimamsakas (or «vaidikas»). Under this heading are
grouped the « types » of karma that express most clearly its ritual and
voluntaristic or moral aspects, but exclude its elemental and cosmic
senses. The yaga, or « sacrifice », the karma par excellence, is not defined
as such, but rather classified according to several bi- or tri-partite distinc
tions that indeed sum up the moral teaching of Hinduism. A karma is
first of all (a) either nitya « obligatory» — without reference to an
occasion, as samdhyavandana, or naimittika, «obligatory given an
occasion », such as the life cycle rituals, or kamya, « voluntary » (where
the self is the occasion) such as the asvamedha; (b) secondly, karma may
be classified with specific reference to its consequence or phala, inas
much as a karma may be done without positing a fruit (as presumes the
first type of the former classification) and is called « sattvika »; or may
be done for a fruit and must involve effort (such as the third type) and
is called « rajasa », or have an evil and perhaps unwanted effect, and
be termed «tamasa ». This classification covers a wider field than the
former, including a greater range of moral options, and demons as
agents and perhaps also error. It is also clearly Samkhyan, speaking
historically. The third classification (c) adds the dimension of motivation
or subjective choice to the second according as an action may tend to
better the condition of the agent (and be sukla) or to worsen it (and
be krsna) or be neutral (and be krsnasukla). Presumed here evidently
is the entire theory of rebirth and samsara, which provides a teleology
for the individual actor. The subjective end, as distinguished from the
objective (or phala), is made a variable of distinction, (d) Lastly, karma
is said to be two-fold insofar as it tends to promote (pravrtta) or discou
rage (nivrtta) further action. The former is « kamapurvakam », presump
tive of a desire, the latter is « jnanapurvakam », presumptive of know
ledge. Here the fourth goal moksa, the non-goal, has been added to the
theory as a variable, in such a way again as to deflate the foregoing
distinctions, for all actions tending to their own continuity (in effect, all
of the above) are now limited by an action that is in some sense not an
action at all. We have now brought the Vedanta under our « vedic » rubric.
(2) Vaisesika: The vaisesikas (realists), according to the lexicogra
pher, consider karma not so much in its ethical or psychic dimension,
but as a physical category: karma here means motion, and is specifically
distinguished from volition and will (yatna). In the category schema with
which the Vaisesika is associated, karma in effect becomes synonymous
with « kriya », one of the four basic existential, categories (with jati,
genus or universal class; guna, quality; and dravya, substance or indivi
20. « So just as here [in this transmigration] the world accumulated by acts
perishes » (ibid.).
dual). Yatna, or psychological « motion » is considered, on the contrary,

to fall under the category « guna », as do all the psychic states (« qua
lities » of substance not involving motion). The vaiêsika view of « mo
tion » resembles Aristotle's, in being subdivided into different local types,
of which locomotion (gamanaj is one. The others are rising (utksepana),
descending (avaksepana), contraction (akuncana) and expansion (prasa-
rana). Interestingly, we find no « decay », as this involves a change of
state, and, therefore, is qualitative (ref. guna)21.
(1) Vyakarana: The bulk of the lexical entry on karma is devoted
to this discussion. It is not only the first item treated but is accorded
the greatest expository development (pp. 1730-35). Vistarabhayat we will
note here only the highlights and distinguishing characteristics of this
long section.
Of course, in one sense grammar is entitled to precede other classi
fications, having inherited from, the older Nirukta its most basic func
tion: nirvacana, or the etymological analysis of words. « karman » itself
is suitably explained by reference to its root element (kr-) and suffix
(man-fin]), with notes as to its ganas (word classes conditioning certain
types of rule eligibility). But this, like the etymologies in Webster's, is
meant as prologue to the main entry, and is not to be confused with it.
The original contribution of grammar to the « meaning » of karma
is of course Panini's definition of what we call the accusative case or
direct object. « The direct object is that which [to reach] is most desi
rable to the agent [of the verbal action]22. Clearly we are in a volitional
mode, but unlike that of the mlmamsaka, the stress is less on the effect
(phala), which is after all supra-mundane, and more on the process itself
(vyapara)23. On the other hand the Vaisesika emphasis on objective
motion is incommensurate with the grammarian’s « karma », which is
in the first place the object of a desire.
This type of karma, patterned in a theory of general verbal syntax,
is subject to several sets of distinctions, which further clarify its essen
tial character. First, grammatical karma is said to be threefold (follow
ing Bhartrhari), as it is (a) nirvartyam, or created (ex nihilo, yad asaj
jayate), as « he makes a pot»; (b) vikaryam, or modified (out of an
already existant) as « he makes ear-rings [of gold] »; or (c) prapyam, or
attained: the « object » of the verb of motion (the completion of whose
21. Included among the bhavavikaras of Yaska, jayate 'sti viparinamate vardhate
'paksiyate vinasyati (cited also MBh ad 1.3.1 [NSP, p. 124]). The matter pertains to
the grammatical « vikaryam », infra.
22. Supra, note 2.
23. ka punah kriya / iha / ka punar iha / cesta / ka punaS cesta / vyaparah
(MBh ad Vt « kriyavacano dhatuh », P 1.3.1 [NSP, p. 114], what is « kriya »? striving).
What is « striving »? movement. What is « movement »? process (!). The pretext here
is perhaps the already cited evidence that many legitimate « direct objects » and
other uses of the accusative (as «poison» in the sentence «The King drinks
poison a) are not « desired » at all by their agents.
94 E. Gerow
activity does not modify the object in any way, as with verbs of pure
motion: the « going » does not change the town reached)24. Already we
can see that these principles of definition function to mediate what
appear to be the opposed « karmas » of the vaisesikas and the mimarn-
sakas, the « objective » and the « ethical» karmas.
Referring specifically to the « most desired » [Ipsitatamam] element
of the sutra, the essence of the relation between the agent, or producer
of the effort, and his «result» is summed up thus: taking the first
triad as expressing the varieties of « karma » that are in fact (con-,
sciously desired, we. may add four more, as « karma » is (d) neither
desired nor not desired, as « he touches grass [while going to the vil
lage] », (e) in fact explicitly not desired, as in the example « he drinks
poison », (f) unmentioned (akathitam) by other technical karaka designa
tions, and (g) required by concord (purely and simply [anyapurva-
kam])25. The latter two types of « karma » share the unintentionality of
the previous two types, but find their motivation in the syntax itself
rather than in the indifference or « ignorance » of the so-called agent.
Under « unmentioned » are grouped the double direct objects of San
skrit, as « gam payo dogdhi » (lit.: « he milks the cow milk »). The last
category comprehends those cases of the accusative (and any other case
in similar circumstances) which are not directly motivated by a relation
to the verb, but rather are what they are by virtue of agreement with
a term already so motivated. In Sanskrit and most Indo-European lan
guages, qualifying adjectives take the case (number, etc.) of their noun.
The Indian grammarians realized that this use of « accusative » served
a quite different function than did the accusative of primary reference.
Put simply, the latter explicated a verb-noun relationship (karaka syn
tax), the former related two nominals, and was not a karaka.
(3) Bhartrhari. The « karman » attributed to the philosophical gram
marian Bhartrhari, and accorded a separate and distinct entry in the
lexicon of Tarkavacaspati, cannot be discussed in fact without entering
upon the second topic that we proposed dealing with in this paper:
(B) The significance of karman in and to syntax. Panini’s gloss of
karma «ipsitatamam » is given flesh in the medieval semantic grammars
in the following way: in every act designated by a verbal root (e.g. « to
go » or « to cook ») are two complementary semantic aspects: a func-
tion/process or vyapara, indicating the change per se, and the fruit, or
phala, indicating the tendency or end of that change26. As the commen
taries illustrate, the « function» in cooking is the series of motions
involved, the placing of the coals, the lighting of the fire, the blowing
24. Vakyapadiya 3 (sadhana), 45 (Iyer ed., p. 266).

25. P 1.4.50-51; also Vakyapadiya 3 (sadhana), 46 (Iyer ed., ibid.).
26. Vaiyakaranabhu$anasara 2 (KSS 188, 1969), p. 11; Paramalaghumanjusa (Ba
roda Research Ser., § 7, 1961), p. 89, etc.
What is Karma (Kirn karmeti)? An Exercise in Philosophical Semantics 95
on the embers, etc., whereas the « result» is the moistening of the grain,
In «going», similarly, the «function» is the leaving of the place
occupied, while the « result» is the attainment of the next place. That
these two aspects of verbality are general, and not as might at first be
supposed limited to transitive verbs, is established also by examples:
« to sit» involves both the « function » of flexing the muscles and limbs
in certain sequences, but also the « result » of a certain posture attained.
Yet the distinction between function and result is also useful for discri
minating transitive from intransitive. The two aspects of the verbal
action look to different external substrates for their practical realization:
whereas the vyapara is based typically in the agent (kartr), or in the
examples, in the cook, or the walker, the phala usually finds its substrate
or basis in the object of the verb, the karma, or in the examples, in
the rice or the place reached. In this way, though the mode of analysis
is actional, the karma is linked closely to the «objective» phala,
whereas the more internal or processual aspect of the verbal idea is
associated with the kartr; karma is ipso facto contrasted with « vya
para », with the more intentional side of activity. Further, intransitive
verbs constitute that interesting class of actions whose vyapara and
whose phala cannot be conveniently so distinguished, just as the posture
assumed is equally supported by the body of the agent along with the
set of actions (flexing, etc.) that directly depend on the agent. These
verbs in effect have no « karma », no « external » [direct] obj ect although
they most certainly do have a phala-, it is precisely this relationship of
karma to phala that defines the class of intransitives27.
« Fruit » or « result » understood in this functional sense avoids the
apparent pitfalls of the one-sidedly volitional « ipsitatamam »; in place
of the direct meaning « wished for » is. substituted the inferred meaning
« resultant of ». The willing is indeed a conscious or semantic dimension
of many verbs; where it is only implicit, the verbal process itself is
indistinguishable from the « effort», or the yatna, and is only conven
tionally and topically related to an « agent » as its substratum, [as in]
«the leaf falls ». This suppression of the «true» or volitional agent
shows inter alia that as far as result is concerned, the substratum of
such acts is indeed the karma. The «leaf » might equally be thought
here to undergo the activity. But the conscious will of an agent may
also be present but as an adventitious « quality » that no longer links
the « functioning » with the « result » of the act. In the sentence « The
king drinks poison », the « vyapara » or function of putting to the lips,
swallowing, etc., is linked volitionally to the agent, or king; but the
« fruit» or the result of that drinking, whose substrate is the « object »
(the poison) is in fact independent of the King (needing only the « poi
son » again as a quasi-« agent» to become explicit). Not only is the
27. VBS § 5, p. 68: svarthaphalavyadhikaranavyaparavacakatvam (sakarma-

katvam).
96 E. Gerow
possibility of error26 built into the semantics, but, insofar as the

« doing » supersedes the « willing » of the action, the determinations that
are proper to the agent, such as consciousness, volition, etc., lose impor
tance in an interpretation of the verbal notion. Those proper to the
« karman » ipso facto gain in importance.
The .peculiarity of Bhartrhari's view of « karman » is that he appears
to adopt this position in its extreme form. According to Tarkavacaspati,
Hari differs from the older grammarians in defining kriya as « vyapara »
(function, or process) only and never as yatna, or motivated effort. And
this to avoid the imputation of intransitivity to all syntactic frames, which
would be implied by a view that « action » were wholly or in essential
part volitional process (and ipso facto incorporate the object of that voli
tion) or, put in another way, that a «yatna » or effort be the link between
the « function » and the « result ». Such a view, attributed to the older
vaiyakaranas, indeed induces the later naiyayikas to assert effort (in
their jargon, krti) as the sense of the verb in all sentences, functionally
the equivalent of the root kr-, made thereby into the root par excellence.
But of course Hari, or the view attributed to Hari, looks (away from
this subjectivism [with its obvious problems]) 'to a notion of vyapara
that emphasizes process and sequence (krama) and is thus exempt from
the problems of volitionality. But vyapSra so separated from any agent,
and becoming the exclusive meaning of the root (dhatu) now coincides
with the « result». Hence « karma » as the substrate of the latter and
kriya (= vyapara) become synonymous. To which synonymy the kosa
bears witness28 29.
To see better the structure of distinctions underlying this revision
of theory and its purpose, let us return to the older discussions of
karma, and rephrase them in terms that will lead more clearly to
Bhartrhari's definition. The issue in situating Bhartrhari (always in the
view of the Kosakara, let it be said once and for all) vis a vis the other
schools is that of specifying the relationship severally asserted between
the terms kriya and karma (grossly, « action » and « act», or perhaps
« activity » and « action »)30.
1) Vyakarana. As we saw, the older grammarians were able to
distinguish the terms (indeed their whole notion of syntax depended
upon that distinction) insofar as the expressor of kriya, namely the
dhatu or (verbal) root [though it is misleading to denote the root as
28. For it is « error » and not whether the King really desires the poison that
is the issue,
29. « vyaparamatrarupakriyayam » (p. 1736).
30. The English «translations », like the originals, are in se ambiguous: the
sense intended has more to do with the distinction itself; and in that distinction
the term kriya would have more verbal or processual overtones, the terms karma
more nominal or resultative [cf. discussion of vyapara and phala, supra]. As the
English term « action » itself indicates (without any reference to the Sanskrit), a
mere term taken in abstracto may properly, given a context, fall on either side of
the distinction.
What is Karma (Kim karmeii)? An Exercise in Philosophical Semantics 97
«verbal» at all in Paninian grammar] involved two semantic dimen

which are differently based (for transitive verbs) in the kartr
sions 3132
(agent of the action) and his desired « object » or karman. The dhatu
provides the nexus or linkage between the « terms » of the utterance,
and most directly between the process and the result, which result
is manifest as the karman of the action (insofar as it is externally deter
mined and not merely a semantic dimension of the verbal idea — the
case for transitive verbs).
In this way, the karman by association with the resultative dimen
sion of the verbal idea is contrasted with the sequential or motivational
dimension of that idea. Not only is the karman differentiated from the
kriya by being associated with only one aspect (the phala) of the latter,
it further attains prominence in the analysis of propositions that are
from the Paninian point of view noting more than delimitations or
realizations of the termini whose nexus is the verbal root idea, the kriya.
For reasons that are never announced, but seem embedded in the syn
tactic possibilities of Sanskrit, not only are verbal ideas invariably
twofold (semantically) but in any given sentence, one or the other must
be given « prominence » (assertive or topical primordinacy). Indeed the
transformation of dhatu into assertion invariably involves the selection
of vyapara or phala as the primary «topic » of semantic reference (the
other becoming subordinate just like any of the other « karakas » [loca
tive, ablative, etc.]) We call these assertional alternatives active and
passive voice. As Cardona has demonstrated33 beyond any reasonable
doubt, neither is inherently primary in Paninian syntax (such that one
could be said to be a « transform » of the other); both are indifferently
realized from the semantics of the verbal root idea by a process that
31. [Later termed] «vyapara» and « phala ».

32. P 3.4.69. The fact that such sentences as « sthali pacati » «the oven cooks »
[where in effect not the agent but the locus (adhikarapa) normally expressed by
the locative, is seemingly expressed as agent] constitute problems proves that such
is the case. It is perhaps with a view to avoiding, the problem that Panini adopts
the neutral term « svatantrah » « independent» in defining kartr (1.4.54). Much of
the grammarians' discussion of syntax concerns such « effortless » subjects, made
problematic by the grammarians' own intentional model, with its Vedic overtones;
as well as by the Nyaya's insistence that an « effort» (krti) was the chief meaning
of the sentence. Cf. Cardona (1976), p. 219. Paramalaghumanjusa, p. 103 ff. That there
is deemed to be but one topic, one pradhana (and never more) in the sentence,
reflects the Mimaipsa doctrine of ekavakyata, according to which there is no possi
bility of construing a sentence by correlating equivalent terms and, therefore, among
them no syntax: cf. the KaumudI to Arthasamgraha (NSP ed., p. 38): na hi guna-
pradhanabhavam antarenanvayah sambhavat, dvayoh pradhanayor gupayor va pa-
rasparakank$arahitatvenanvayayogyatvabhavat.
33. «Some features of Paninian Derivations», from History of Linguistic
Thought and Contemporary Linguistics, ed. Herman Parret, Berlin/New York, 1976,
passim and esp., p. 141; « Panini’s Karakas », JIP 2 (1976), note 36.
98 E. Gerow
resembles arbitrary selection34 (again « motivated » by the . speaker's

intention only). And by that selection, emphasis on vyapara gives topical
prominence to the kartr (active) while emphasis on the phala gives
prominence to the karman (passive). Topical prominence is recognized
at the « surface » also by the nominative case, which denotes nothing
but the kartr or the karman that has already been selected as reference/
support for the vyapara or phala at the level of verbal derivation3S.
The karma then defines one of the two options that at the verbal level
(dhatu) remain implicit and can become explict only in the sentence,
only at the level of assertion. Without the distinction kriya/karman/
kartr, this delicate optionality would be impossible to postulate or
explain. Indeed the assertion as such is nothing more than the specifi
cation of the actional qualities of the verbal nexus, which inevitably
involves the selection of one (kartr or karman) as primary (announced
by the verb directly, through its invariably associated terminations).
The Sanskrit examples have become so ubiquitous that we repeat
them: odanam Devadattah pacati (Devadatta cooks rice: active, or « kar-
tari ») and odano Devadattena pacyate (The rice is cooked by Devadatta:
passive or « karmani »). On one level of analysis (call it surface, or the
phonological) the sentences are evidently different, but on another (the
« karaka » or semantic level in the Paninian grammar), it is either very
difficult to determine the sentences' difference, or they are simply not
different (= mean the same thing). That basic-unity is, stated in the
Paninian grammar in terms of the karaka theory of syntactic semantics.
According to the Paniniyas, the words « rice » and « Devadatta » in both
sentences express the same syntactic function, namely « direct object»
(karman) and « agent » (kartr)36. The reason they are differently realized
or expressed in the two sentences has to do with the basic option
before us in forming any verb, discussed above. Since the kriya expressed
by the root (dhatu) is in any case primary, the option, which is general
in all verbs, is also the most basic option of the syntax. Depending
on the choice made, the rest of the sentence is determined automati
cally, as for example:
34. Though it may have many contextual conditions, and have social and stylistic
connotations. Cf. Gonda, Remarks on the Sanskrit Passive, Leiden, 1951, pp. 3-4. The
« volitional » is again superadded to the « intentional ».
35. P 2.3.46 by the « anabhihite » convention (2.3.1), see below, p. 99. On
«bhave», the third possible sense expressed by the verbal suffix according to
P 3,4.69, see infra p. 100.
36. Kaiyata ad 3.2.124: ihayam lakaro lah karmani ca bhave ca ity anena kartr-
karmanor vidhiyamanah gupabhute kriyam prati kartj-karmani pratipadayati. We
ignore the nicety that the nominative merely agrees with the verbal suffix, the
proper expressor of the karaka. Supra, note 35. Aklujkar adds: «It should be noted
that ” agency " is not confined here to animate conscious entities, that is to living
beings ». Cf. Bhartrhari, 3, 7, p. 103 ff.
pac + kartr odana + karman Devadatta +

ACTIVE [agreement: kartr]
pac-ati odanam (acc.) Devadatta + h
(nom.)
pac + karman odana + [agreement: Devadatta + kartr

PASSIVE karman]
pac-yate odana + h (nom) Devadatt[e]na (inst.)
The single convention needed to form correct strings is that of « ana-

bhihite » whereby a term gets its case only if its inherent karaka has
not already been expressed by the verbal termination (-ati or -yate, for
example). The first case [= nominative] thus expresses no karaka per se.
Karakas that cannot be expressed by the verbal termination, for
example, adhikarana (« location ») are indifferently realized in either
model; if we were to add the notion « sthall» « oven » to the sentence
above, it would be formed in both sentences sthall+adhikarana sthall+
am sthaly+am3738 .
Let us now contrast this syntactic karman with the terms found in
the other sastras. 2) The Vaisesika notion is enchantingly simple. Instead
of distinguishing kriya and karman, the physicists identify the terms,
such as in effect karman becomes a word for « motion » purely and
simply; neither kriya nor karman can legitimately be used to express
the « motions » of the soul, sp. « yatna », which is located in the cate
gory guna. The entire grammatical theory, according to the vaisesikas
is founded on category (padartha) mistakes. It shoud be noted, however,
that both the older grammarians and the physicists agree in distin
guishing karma from the semantic area in which « effort » or (pra)yatna
is located: the former by distinguishing karma and kartr (or phala and
vyapara), and the latter by considering « efforts » an affection of the soul
(atman)3B. Karman is not to be found associated with agents or willing.
3) The commonplace that the root (dhatu) expresses an action or
kriya is in the Mimamsa reinterpreted to signify that the root kr- (from
which is formed the noun kriya, as well as all the others that figure
prominently in our discussion), is the « basic » root idea, and is as such
implicit in all roots. This notion is doubtless traceable to the mimam-
sakas' curious « explicitation » of such verbs as « ... yajeta » as « yagam
kuryat». According to Mimamsa the root (in this case yaj- is itself the
direct object or « karman » (in both grammatical and ritual terms!) of
the notion of universal effectivity (bhavana) expressed by the verbal
37. Shortcuts adopted in the derivation. Cf. supra, note 32.

38. The later Nyaya evidently agrees, considering « effort» (krti) [as the chief
meaning of the sentence] to be a property of the (individual) atman, the subject
of the sentence: ...krtiman Devadattafi. See Cakdona (1974b, pp. 245-54; 1976, p. 230).
100 E. Gerow
ending [here optative or injunctive] which is primary and has the

meaning « kriya », and is « translated » by the « root » kr-. The MTmamsa
notions is thus not far in principle from that of the Vaisesika, for in both
the « dhatu » (per se) is karman.
4) In Nyaya, the «vyapara» or verbal process is related to the
underlying agent (kartr) or efficient cause of the action, through the
« yatna » of the latter. The later naiyayikas as we have seen assert that
the universal root kr- is implicit in the verbal ending but in accordance
with their view of kartr-pradhanya is itself dependent on the external
a agent.», expressing nothing but the « effort» of that agent. The naiya
yikas thus raise « effort » to the status of verbal universal, but remain
true to the older Vaisesika in considering « yatna » merely an adjunct
(of individual substances, such as the ubiquitous « Devadatta »). But if
this interpretation is accepted, it will be equivalent to asserting that
the verb kr- is intransitive, just as « prayatate » its putative semantic
equivalent, is intransitive: « he strives »39.
The issue as to what the dhatu itself means does not at first appear
to affect one way or the other the actual derivations of « active » and
«passive» sentences which proceed through the kriya-karaka model
and utilize centrally the notion « karman ». That issue is raised when
this model is addressed from the perspective of the intransitive verb,
that is, one that requires no (explicit) direct object or karman. A verb
like « sit» is in both English and Sanskrit intransitive: we cannot say
*He sits a seat, or *He sits a posture: « He sits » is a complete utterance:
Devadatta aste. Now the question is posed: «how is the “ delicate
optionality" implicit in the verbal " root ” to be conceived for such
sentences? ». The prima facie answer, that such verbs are inherently
« active » is rejected as inconsistent with the intentionality of our verbal
semantics. Instead the Paninian grammar adopts an epicycle, the notion
« bhava »40.
39. VBS 5cd (Kasi 188, p. 62) krno ’karmakatapatter na hi yatno'rtha isyate. On
these various modes of construing the sentence, see Matilal, Indian Theorists on
the Nature of the Sentence, in « Foundations of Linguistics », 2 (1966), and Cardona,
Paraphrase and Sentence Analysis, JIP 3 (1975), passim, and « Karakas », pp. 251-54.
40. lafi karmapi ca bhave cakarmakebhyah: P 3.4.69. « Bhava » is a common
meaning of abstract or «infinitival» (Whitney, § 1145, 1148 bgf.) nouns, as P 3.3.18
« bhave », enjoining the suffix (gh)a(n): « paka », « cooking ». As our later authors
will show, the meaning is not different from the «vyapara» of the root when-
properly understood. Cf. the historical relationship between the ya- passive and the
ya- present in Sanskrit, the latter used primarily for «intransitive » notions: Gonda,
op. tit,, pp. 73-4. « Bhava » is found in the Nirukta (1.1) in the sense of « verbality »
in general, as opposed to the « sattva » of nouns. R. Rocher recognizes the problem
of the passive impersonal very clearly in her article Bhava «etat» et Kriya
« action » chez Panini' (Recherches Linguistiques en Belgique, 1966, pp. 115 ff.), but
because she focusses on the semantic contrast with kriya rather than on the
functional contrast with karman the simple solution presented here escapes her.
What is Karma (Kim karmeti}? An Exercise, in Philosophical Semantics 101
In the sentence given above, there is no real problem:

Devadatta + [agreement: kartr] as + [kartr]
Devadatta + h (nom.) as 4- te
All accept that Devadatta is the agent of the act « sitting ». But as to
the « passive », a basic condition is missing, namely, an external object
(karman) which can be realized (at the surface level) by the nominative
case:
Devadatta + kartr as + [*karman] 0 + [agreement: karman]
Such sentences are in most languages not formable, for the simple
reason that most languages resist subjectless utterances. But in Sanskrit
it came to be possible to form for every active sentence, transitive or
intransitive, a corresponding « passive » or « passive-like » sentence.
Devadatt[e]na as-yate (0) *« by Devadatta [it] is seated ».
In the Paninian karaka formulation, however, this sentence, though
perfectly well formed could not be explained as « passive », for (we must
realize) the term needed: karman, was ex hypothesi not available to the
analysis. Karman is not a possible « meaning » of the termination of an
intransitive verb. The anabhihite convention would be violated if so
explained: as in the above example: if there were admitted a termination
for as- in the sense « karman » then there ought equally to be a term
agreeing with the karman, and subject to the nominative case ending.
To avoid this difficulty the term « bhava » was appropriated from the
older Nirukta as a «meaning» for the verbal termination of such
subjectless sentences:
Devadatta 4- kartr as + bhava.
The « bhava » is realized exactly as would be a corresponding pas
sive (karman)41:
Devatt[e]na (inst.) as 4- yate
but the problem occasioned by calling the ending « karman » is avoided.
The older grammarians are content by this device to have preserved the
« delicate optionality» of the root (kriya) even where karman is not
«available». The newer school (Bhartrhari, et. al.) raise this notion
of « bhava » from the status of an « epicycle » to that of the center of
the verbal universe!
In the neo-grammarians' scheme, to say that the root expresses
« vyapara » only, excluding the sense « efFort » is to make « bhava/bha-
vana » the only possible « meaning » of the root. The root has become
its own asraya («basis ») in the sense that no external karman (or
41. By 3.1.67, « sarvadhatuke yak », cf. Cakdona, « Karakas », pp. 24143.

102 E. Gerow
kartr) may be postulated to subtend its «two » meanings. It is equally

empty to speculate which of the two meanings — vyapara or phala —
is more or less primary (or which is the basis of the other — a concern
that seems to have animated some the the « older » school): vyapara, so
circumscribed, is phala, nothing but its own tendency to eventuate in
such and such: bhava. By this argument, not only has « vyapara » been
shifted away from the external agent and finds its basis along with the
«phala» as its « karman», that very karman now only incidentally
refers to the external object. It is nothing but the « root » grasped as
primary, as « bhava », and it is to it that the termination refers. In effect
karman becomes a synonym of all these terms: vyaparo bhavana saivot-
padana saiva ca kriya42. In syntactic terms it amounts to saying that
the «passive » impersonal has now become the normative mode of
expression.
42. Tarkavacaspati, p. 1736.

The attribution of this line to Bhartrhari in the lexicon raises a number of
thorny historical issues. The argument of the paper, based on a set of distinctions
offered by tradionalists, depends on the intellectual relations inherent in or attri
buted to those distinctions. That they do not always fit with «history » is itself
interesting, and perhaps more significant than the alternative.
In view of the Kosakara, Bhartrhari or his Vakyapadiya is deemed the source
of the view of kriya/karman to which the bulk of this paper is devoted. This is
a true attribution to the extent that the tradition of grammatical speculation
founded by Hari, but continued ably by much later writers, such as BhattojidIksita,
Kaunndabhatta and NageSsa, is taken as the source of reference. The tradition is
taught today as a distinct subject in the Sanskrit colleges (often termed « siddhan-
tavibhaga » of the grammar curriculum), and roughly approximates to the philoso-
phico-semantic branch of grammar, the aim of which was the estabishment of
grammar as a soteriological vehicle alongside such canonical margas as Vedanta,
Samkhya, Bauddha, etc. (cf. Sarvadarfanasamgraha).
But the writers grouped together in this tradition are not always in agreement
amongst themselves, and though they certainly develop and expand the themes of
Hari, they often do so in quite different perspectives. In fact, the historical problem
is neatly summed up in tracing the filiation of this very source.
Unless I am very much mistaken, the line is not Biiartrhari's at all, but occurs
in the karika portions of the Vaiyakaranabhuâpasara (§ 5, p. 62, KSS 188, appa
rently so Cardona, « Paraphrase...», note 18) and is, therefore, likely the work of
BhattojidIksita [there is some dispute in tradition as to whether the karikas as
well as the commentary were written by Kaunpabhatta, but the opinion of modern
pundits (cf. pp. 9-10, ibid.) generally distinguishes the two. The possibility of iden
tity is suggested only by the pronoun confusion in the final line of the initial
colophon «kurve 'ham...» and the initial line of commentary « atra asmabhir iti
sesah »]. Now that is not to say that there is no ground for attributing this point
of view to Hari. But in point of fact, Hari does not appear to be familiar with the
distinction ph ala/vyapara (in such terms: « phala » is used in the context of « upa-
graha» [i.e. senses of the middle voice], clearly derivative of 1.3.72 « kriyaphale»
[svaritanita...]; his stress in enquiring into the meaning of kriya (not dhatu, N.B.:
Vakyapadiya III: kriyasamuddeâ) is on the distinction first offered in the Nirakta
between the sadhya and the siddha (ibid., § 1) which grosso modo characterizes
the difference between verbs and nouns. Although sadhya bears relationship to the
vyapara and the siddha (evidently) to the phala, Hari's interest is in expressing
the irreducibly « verbal» meaning associated with the sadhya aspect, which he does
What is Karma (Kiifi karmeti)? An Exercise in Philosophical Semantics 103
ultimately through the notion of sequence or krama. An act expressed as sequence

is expressed verbally; an act expressed otherwise, is a noun §§ 3, 4 ff.). Verbality
is the « unity » that nowhere « exists » but is imposed on the sequence, such that
« he cooks » is a proper utterance whether the cook is laying the fire, lighting it,
preparing the food, or just sitting waiting for it to boil.
This certainly resembles the view that vyapara only is the meaning of the root,
but in fact is a translation into objective reference of that statement: the sadhya
is the « meaning » of the (verbal) activity.
Further to complicate matters, Tarkavacaspati, in attributing this line to Bhartr-
hari, not only appears to confuse Hart and Bhattoji, he also misapprehends the
meaning of the line, at least according to Kaundabhatta. That is to say, if Kaunda
can be trusted, Bhattoji does not accept the view that vyapara only is the meaning
of the root (but is rather merely asserting the inadequacy of the naiyayika view
that yatna only is the meaning of the root (kr-). The consequence for our treatment
of the karika is of course that Bhattoji must be considered an « older» gram
marian, accepting vyapara and phala as separate meanings of the root, in contra
distinction both to Bhartrhari (putatively) and Nageâ. But it is equally possible
(at least if one accepts the different authorship of karika and sara) that Kaunda
has confounded Bhattoji’s plain intent, for the first half of the verse does not
appear to say the kriya is also vyapara etc. (which would logically have been ade
quate to deny the Naiyayika's exclusive reliance on «yatna»): na yatna iti na
yatnamatra iti (Kaunda).
Nagesa at any rate associates Bhattoji with the view that phala and vyapara
are separable and separate meanings of the root, which view he takes as purvapaksa
in establishing his siddhanta that the root has one meaning only, which we may
indifferently term phala or vyapara depending on the circumstances (phalavacchinne
vyapare vyaparavacchinne phale). (Paramalaghumanjusa, pp. 89-90 [pare tu is taken
as referring to himself: svamatam aha pare tv ityadi: Comm, of Kalikaprasad Sukla,
ibid.}).
By this he means that « phala » and « vyapara » are related to one another as
for example an individual (« cow») and its species (« bovinity»: go-tvam) are re
lated; that is without any construction of the relationship in sentential or predi
cative terms (which would appear possible and necessary if phala and vyapara
were considered two different «meanings » of the root (phalaviî$tavyapara), And
this view is not simply an evasion. We take it as asserting unequivocally the basic
identity of phala and vyapara (call it by either name, but vyapara in the sense
bhava fits best), which achieve distinction only when construed with the karma or
kartr of the verbal suffix, that is with the karma or kartr given topical prominence
in the sentence/assertion [kartrkarmarthakatattatpratyayasamabhivyahara^ ca tat-
tadbodhe niyamaka iti (p. 90)]. This issue is to be distinguished from the one
Cardona raises (« Karakas », notes 35, 45), which seems to involve an incompatible
conclusion. The «older» school of grammarians (Bhattoh, Kaundabhatta) may
indeed have regarded active and passive sentences as «synonymous» (Cardona) :
but the question they ask (if the posI-Nageâ Commentators can be trusted:
cf. Jyotsna. ad PLM, Baroda, 1961, p. 140) namely which of the two verbal meanings
(phala, vyapara) is primary and which is subordinate, demonstrates for our purposes
(whatever their answer) that the meanings were considered two, and indeed were
to be construed somehow. That the vyapara was understood also as «primary»
in the passive sentence evidently reflects the influence of Yaska's dictum: bhavapra-
dhanam akhyatam, etc., and does not affect the point we are making here, that
the «older» school was concerned with the two kinds of sentences as options,
reflecting a choice among the external relations of the verbal termini.
Nagesa, on the other hand, by refusing to consider the phala and vyapara as
fundamentally different, both. avoids the problem of construing their relationship,
and reduces the difference between active and passive sentences to one that concerns
only externals: is purely formal in other words. Compare his «siddhanta» and
definition of transitivity (PLM, p. 100): vastutas tu sabdaâstriyakarmasamjfiakar-
thanvayyarthakatvam sakarmakatvam.
104 E. Gerow
NageSa in any case is at some pains to establish (vis-a-vis the verb «janati»
that the «meaning» of this root is not exclusively grounded on changes in its
«object» (cf. bhinatti), but is in fact grounded on both its «object» and
its « agent» (to avoid the imputation of « karmavat...» by 3.1.87). In such roots,
the «meaning» is grounded equally in both: yatra kartrkarmasadharanarLipain
phalarp sabdena pratipadyate...» (PLM, p. 102).
K. P. Sukla, in his Introduction to the PLM, emphasizes also the duality of
Kauwa's verbal sakti: atra pracinavaiyakaranah phalavyaparayor dhatuli (VBS 2)
iti dvivacanena phalanirupita vyaparanirupita ca dhatoh prthak saktih (p. 34), in
contrast with NageSa's attempt to overcome the duality by the notion of mutual
qualification (as cited above). Sukla moreover wants to move beyond a relative
identity to a total identity (vastutas tu visiste saktir nasti... i.e. « sakti » is in the
visesapa only in that case) which he also finds implicit in NagfAa himself: ubhayamsc
eka khandasas saktir iti na saktidvayakalpanam na va bodhajanakatvasambandhad-
vayakalpanam, pratyuta navyanam api tat kalpaniyaip bhaved iti tattvavidah (ibid.).
The point of view which we find attributed in the Ko§a to Bhartrhari is thus
in effect that of Nagesa. And this we find suitable, inasumch as Nagesa represents
the final flowering of the Indian grammatical tradition (18th century), and deserves
our most profound regard in other respects for his powerful and original argumen
tation. Moreover, if it is indeed to NageSa that we must look for this key to under
standing the grammatical and the philosophical karman, the chronology of the
problem becomes much easier to follow. Instead of being forced to deal with Hari,
who precedes the formulation of certain issues crucial to our understanding of
both the grammatical and the philosophical karman, being both «pre-ÂMKARA»
and « pre-phala-vyapara », we may much more reasonably look to Nagesa, the culmi
nation of our philosophico-gramlnatical tradition, for the necessary insight.
And so with this in mind, I think we may be almost in a position to substitute
NageSa's name for Hari's (in the Ko?a), despite the fact that the words are probably
Bhattoji's. The necessities of the argument, in any case, suggest a closer relation
between Bhattoji and Nagesa than between either and Hari, with Kauxieabhatta
perhaps playing the role of a conservative.
A copy of S. D. Joshi's unpublished Ph. D. dissertation (Harvard, 1960) was
recently conveyed to me by Prof. Daniel Ingalls, to whom I express my deep
appreciation. In it, Joshi covers some of the ground I do in this footnote and
elsewhere: although I learned of the dissertation's existence after this paper was
essentially complete, it is my good fortune to be able to check some of these points
against Joshi’s authority. Although recognizing (p. XXXIV) the «identity» of kriya
and vyapara, Joshi also calls attention to other contexts in which the two terms’
meanings differ. The chief alternative, that « vyapara » designates the verbal process
distributively, whereas « kriya » designates it collectively, is, I think, only a confir
mation of the argument of this paper, for Hari especially, who would analyze any
« specific» or « component» processual «unit» into an infinitely divisible set of
sub-units (p. XXXVI). On another point (p. LVI), Joshi suggests that the gram
marians owe to Gangesa the notion that the root is a composite of two «meanings»:
phala and vyapara. It is in any case the superaddition of the «phala» to the
older notions of Yaska and Patanjali that occasion our discussion. In the perspective
he adopts, the conclusion of this paper will seem like a restatement of the status
quo ante: vyapara only is the meaning of the root, except for the fact that the
argument here proceeds not by exclusion but by inclusion, by attempting to un
derstand even the «bhava» as a (and in fact the) crux of the verb's sentential
function. Joshi's treatment of «bhava» is brief, and quite matter of fact; it is
not possible for me to decide how he might respond to the argument put forth
here. The thesis is especially useful for the manner in which it clarifies the different
sasIras' different constructions of the root meaning (sabdabodha); it is unfortunate
(for this writer especially) that it has remained so long unpublished.
"What is Karma (Kirfi karmeti)? An Exercise in Philosophical Semantics 105
The ubiquity of « passive » syntax in later Sanskrit has often been

commented on43. From the point of view of the semantic analysis of the
root (dhatu), it is not so much the passive (karmani) as the « bhave »
43. Gonda, op. cit.

Various writers on Indo-European have independently concluded to a concept
of the original IE «passive» that resembles the neo-Paniniyas' in sublating the
distinction between the passive and the «impersonal»: « Passivische Idee ist also:
die Auffassung oder Vorstellung eines Geschenes der Wirklichkeit als Handlung, die
an einem Seienden vollzogen wird und deren Urheber vorstellbar sein muss » [Alice
Berger, cited Gonda, p. 76, ft. 103]; Meillet, followed to some extent by Gonda,
traces the «passive» rather to the true impersonal, or agentless (not objectless)
verb. Gonda prefers the term « eventive »: « arbor cadit» as opposed to « agricola
caedit arborem » or « arbor caeditur ab agricola » (Gonda, pp. 75-6). But of course,
these are precisely the intransitives whose «passive» defines the trenchant or
« true » passive also.
The generalization of the passive construction to instransitive verbs is in part a
feature of all Indo-European, but is nowhere carried out with such rigor and uni
formity as in Sanskrit. Still it is informative to examine other older IE languages,
in the light of the argument that it is precisely the pervasiveness of the «imper-
sonalization » of Sanskrit that marks it also as « philosophically » distinctive. Aside
from the usual « subjectless » impersonals, of which all the languages have a small
and usually overlapping inventory: eusotl (it is possible), licet (it is permitted)
(often understood as having sentences for subjects, Goodwin, § 898), the ubiquitous
«meteorological» verbs: advesperascit, oslsi, etc., etc. All these types are easily
illustrated from the modern languages as well: it rains, il pleut, es regnet, showing
the same « dummy » pronoun (cf. Aristophanes', mot: aXXi tIc &ei; Clouds, 368).
aside from these, the classical languages show great variation in extending the
« normal» passive construction to other areas. Greek seems to be the least inven
tive, indeed this is not surprising, for among all older IE, the distinction of usage
between active, middle, and passive is most clearly preserved here. But in two
areas Greek shows traces of our construction: with verbs implying a cognate accu
sative, a «subjectless » passive is frequently made: s^7t£LO7) aurovt; TrapeotxE&tiaTo
(T. 1.46: G § 1240), which is of course nothing more than the suppression of the
« cognate » (now) nominative: a preparation was prepared. The other frequent con
struction is that involving a participle, often the future passive. As Goodwin remarks
« Even an intransitive verb may thus have a passive voices (ibid.)-. va TjpapTpp^v/:
the errors which have been committed.
In Latin, this «participial» usage appears with apparently wider extent, af
fecting in principle all intransitives: diu pugnatum est (Caesar). But against this,
is the widespread use of «passive » participles from intransitive verbs in active
meaning: pransus, having dined, etc. (HB 290). Provoking this latter usage may well
be the similar case of deponents: secutus sum, not *1 am followed, but « I have
followed». Compare the Sanskrit usage of certain past participles in the active:
aharp gramaip gatah (as well as gramo maya gamyate/'gatati). That these usages
concentrate in the participle (past and future) in fact robs them of the kind of
significance we are seeking, for it is everywhere arguable whether in fact the
« passive » participles are properly so considered. Their « passivity », as implied by
Sanskrit and Latin at least, may be a secondary specialization paralleling the deve
loping morphological passive of both languages (Latin, by capturing the old middle
inflexions; Sanskrit, by specialization of the -ya- suffix in present stems).
It is in Old Irish, often more conservative than its better known classical
relatives, that a case truly parallel with the Sanskrit is observed, though perhaps
not so widely distributed over the lexicon. «Tiagair o Chunn co Conchobar»
(Quin, p. 51) is a fairly typical case, the force of which is somewhat obscured by
the usual contextually justified translation: (messengers) are sent from Conn to
106 E. Gerow
or passive intransitive construction that is typical, wherein is empha

sized the phala or result of the action even in those cases when the
phala has no external support (or, « karman »). The deed we might say,
becomes its own doing, or the deed itself is deemed the pradhana or
head of the sentence, quite apart from any « external» support or basis
(given topical prominence in the usual « active » or « passive » construc
tions). If the notion « vyapara » is understood not to require the chief
connotation « effort » (yatna) and thus be linked inescapably with the
kartr, its basis can be only the karman itself, not the « external » karman
but that « within » the verb, asserting the primary verbality of the root
itself: bhava. The deed is primary, and not what is done*44.
Vis a vis the Mimamsa view, the revision attributed to Hari consists
chiefly in a refusal to treat a meaning of the verbal root, the karman,
as secondary. But in conceiving of karman as primary, it cannot itself
be motivated by another bhavana — an « efficiency » that barely conceals
the circularity of interpreting the universality of the verb « kr- » in the
ending while taking it also as its own « direct object», the karman in
every root. What indeed do we make of the root kr- itself? How do we
interpret the verb form kar-o-ti? On the analogy of yajeta, it would
be « karma kuryat»: and this latter form would itself be analyzable
into « karma kuryat », an infinite regress. The new grammarians postu
late the universality of « kr-» as well, but abandon the Mimamsa's
notion that the « root » is secondary -to the termination (and in that
obvious sense is its « karman »). The equation of vyapara, one of the
Conch obar/Conn sends (messengers) to Conchobar. That messengers is not implied

at all is clear from the number of the finite verb: always singular, as in the
Sanskrit: asyate tair atra «They sit here». We have a genuine «bhava» con
struction, which the Sanskrit student would have no hesitation in translating: it
was gone from Conn to Conchobar (scil. by messengers). We have here, as was said,
a finite verb, not a participle, used in the arbitrary singular of the impersonal, but
passive (clearly passive, for 01 shows passives only in the third persons) in form.
The same construction is often seen with the participle, just as in Sanskrit: Doeth
o Ailill ogus o Meidb do chungid in chon (Scela Mucce..., p. 1). It was come from
Ailill and Meidb for asking about the dog (scil. by messengers). Not only verbs of
motion are subject to this construction: other intransitives, as eigithir «it was
shouted » (a cry goes up) show the same treatment, which is quite general in Old
(Middle) Irish. Did the Irish have a «karma» theory? It wouldn’t be the first
parallel between the two cultures that had been observed. Cf. Dillon and Chadwick,
The Celtic Realms, pp. 26-33.
44. But of course « karma » means deed. Still the argument concludes by taking
it as a nomen actionis, a quasi ghan-anta, in the sense « bhave », supra note 40.
Prof. Nils Simonsson calls miy attention to the parallel conclusion of the medieval
Pali grammar Saddaniti: Yasmim payoge yarn kammuno kiriyapadena samanaga-
tikarn katva vina kammena niddislyati kiriyaya padam... tarn tattha bhavatthadi-
pakam... (.). Evam sante pi bhavo nama kevalo bhavana-lavana-pacanadiko dha-
tuattho yeva. «That construction illustrates the sense "bhava”... wherein is expressed
the word for the action without the object, having made an identity of the action
word (verb) and the object (kamma). Such being the case, "bhava” is the mere
sense of the root itself, "be, wash, cook", and the like» (Saddaniti I, pp. 7-8;
ed. Helmer Smith, Lund, 1928),
What is Karma (Kun karmeti)? An Exercise in Philosophical Semantics 107
two « aspects » of root meaning, with the « bhava(na) », in effect restates

the Vaisesika notion that to be « karman » indeed expresses the primary
character of the root's meaning (kriya) rather than its dependent qua
lity. It is as though the Mimamsa formula were turned inside out, the
root remaining « karma» but that « karma» no longer requiring an
external « bhavana » or effectivity to be object to.
Vis a vis the Vaisesika view it is difficult to see that Hari has radi
cally modified it except of course to use it in a new syntactic analysis.
But on the Vaisesika view, though karman was an (external) motion,
it was in no way any more « primary » than the yatna which was an
effect of the atman: both kriya and guna are thought of as correlative
qualities of substance. Similarly, the Vaisesika made little or no mention
of « result» (phala) and this was to be a keystone of the karma syntax.
Hari’s revision has now been stated in terms of its proximate moti
vations; and in the context of the technical distinctions, the possible
interpretations of the term « karman »; we have now to turn to the
question of the wider motivations of these changes, and their signifi
cance in understanding the cultural and cosmic developments of the
term « karman ». It is precisely our contention that the « cultural » kar
man with its cosmic and soteriological overtones, is very little but the
meaning implicit in every sentence (its universal syntactic import) when
read according to the syntactic theory outlined above.
(C) What conclusions do we draw from this exercise in syntactical
definition? We have used the lexicon and syntax as our guides up to
this point, on the ground that they best discriminate those variations
in meaning that are the cultural content and specificity of the notion
«karman». We have attempted further to rationalize the variety of
meanings, to explore their inner relationships, to see in what sense they
imply a logical progression (not NB a historical progression), a progres
sion which in our view is both illustrative of grammatical distinction
and eventuates in a grammatically adequate view of « karman ». But we
have further supposed that the grammatical « karman» is a key to
understanding the wider and more culturally evident « karmani». In
this third part of the paper we have to pursue that question, however
speculative it may appear.
But the Indian tradition is speculative, and we meet it chiefly in
a speculative form; it is much more difficult to clarify the genuinely
«indigenous » at this level4S. Fortunately, the tradition does provide
us with one modern account of the « theory » of karman, kept as pure
as is reasonable to expect under the circumstances, by its expression
in chaste Sanskrit, and by its anti-Western polemical style. In « The
Karma Theory » of N. N. Bhide46 we have a lively survival of traditional
45. This was indeed the problem with which we started, and the reason that
we began with a study of the vocabulary.
46. University of Mysore, 1950. My thanks to Ashok Aklujkar for bringing this
work to my attention.
108 E. Gerow
lore on the subject of karma, not so much unaffected by Western views

(for the author is not deaf and dumb) as successfully integrating Western
views into a traditionally Indian mode of exposition: the dialectic of
purvapaksa and siddhanta. Needless to say the Indian view (so labelled)
is siddhanta. Though the bulk of the work is eristic in tone and devoted
to an exposition of the errors of Western scholarship, we find the por
tions devoted to the traditional « karma theory» very useful for our
purposes.
But we cannot forbear to quote Bhide as he explains why karma
studies are in such an unrelieved state of confusion47. This condition of
moha has four causes (karanani). The first three are perhaps ineluctable,
the fourth we can only hope by this paper not to have compounded:
1) The tradition, says Bhide, provides us with no single thorough expo
sition of the karma doctrine in one work48; 2) in classical times at least,
everyone, even the Buddhists and Jains, accepted the karma doctrine
« with their eyes shut »49 and with such universal acceptance, 3) we find
today no means of verifying the truth or falsity of the doctrine itself50.
The most important cause of confusion, says Bhide, is the fourth, which
merits some vivid, though still chaste, expression, viz., the wild and
irresponsible manner in which Western scholars have attacked the
question. Standards of Western scholarship are deficient on a number
of points: though scholars have expended much effort, they are unable
to understand the essence of oriental culture51; they concentrate on the
words (viz., texts) only52; they join words (viz., arguments) together in
any way at all5354
; they turn the unmentioned into the chief reason they
construct arguments at will55; and finally they are more interested in
establishing their own favorite novelties than the true siddhanta56.
This last sally comes very close to the quick, in expressing, the
contempt with which a traditionalist views the ever-changing quicksand
of Wetsern «truth », which seems always to owe as much to the agenda
of the inquirer as to the nature of inquiry. Hoping to atone for past sins
accumulated, we proceed to Bhide's exposition of the karma doctrine.
47. Citing Gita., 4.16: kim karma kim akarmeti kavayo'py atra mohitah; supra
note 10. « What is action? what is inaction? even inspired sages are confused about
that! ».
48. viâyasyasya kutrapy ekatra granthe sakalyena vivecanain naiva samlak-
§yate: p. 2.
49. prayena sarvair eva bauddhajainadibhir avedikair api karmatattvam nimili-
talocanair ivarigikrtam apratiravam; ibid.
50. tasya [karmatattvasvarupasya] satyatvaip mrsatvam va niscetum pratyak-
sikam pramapam adya nopalabhyam; ibid.
51. pracyasamskrtes saram avagantum aksamah, op. cit., p. 3.
52. kevalaip sabdamatre ’bhinivis|ah; ibid.
53. yathakathamcana sabdan yojayantah; ibid.
54. anullekham eva pramukham pramanarp prakalpya; ibid.
55. yathakamam tarkarp vidhaya; ibid.
56. svepsitatamain siddhantam sthapayantalj; ibid.
Bhide's analysis is shot through with grammatical distinctions and

terminology. Like the Kosakara, he can begin the discussion of « karma »
only with the root from which it is formed: (du)kr(n) [karane], His
immediately subsequent definition of karma (though we must remember
his purpose is to clarify the widest cultural implications of this unique
concept) is strikingly like that above referred to as Bhartrhari’s:
«... kriyavacitvam. Kriya ca parinamarupo vyaparah » [the word karma
is expressive of an action (kriya). An « action » is a process (vyapara)
which has the form of a transformation (parinama)S758 . Further, action
implies an agent, but does not require one; hence karma is the name
of an action viewed generally: of a process merely (without reference
to its agent) having the form of a transformation5S. Having finessed the
agent, and linked the terms karma, kriya and vyapara, Bhide leaves the
purely grammatical arena by following the implications of the term
«parinama» (transformation), introduced as a synonym of vyapara:
akhilasya jagatah pratiksanam parinamitvat karmamayatvam5960 : «the
entire world (jagat: lit. " the constantly moving ”) being every instant
subject to transformation ( — change) consists of karma ». One effortless
(pun!) leap takes us from the root to the universe (at least the universe
of samsara). From this point the exposition assumes a more traditional
cast, a mix of Advaita and Samkhya. Definitions of « world » (jagat) help
in discriminating the one reality not subject to change: brahman, whose
first « act» (karman), nevertheless, eventuates in and estabishes the
conditions for the world (jagat = samsara) ®.
There follows a derivation of the world via maya6162 and a straight
forward genesis of the five indriyas (sense faculties) and bhutas (ele
ments) in the Samkhya manner — which Bhide says has been accepted
by the Vedantins and others. The account becomes interesting again as
we discover this world (bhur nama:), which in addition to being a result
of viparinama: one of the seven known worlds, is characterized as (a)
perceptible, accessible to senses (indriyagocaram) and (b) as comprehen
sive of entities that move and entities that do not move (caracarasahi-
tam)a. The notion of jagat in general becomes particularized as this
world. But the definition remains the same. Agency is bondage.
The other six worlds (bhuvar, svar, mahas, jana, tapas, satya) are
then distinguished in terms of qualities proper to agents: subtlety and
grossness (of cognition/effort). Bhide emphasizes that the worlds inter
57. Ibid., p. 38.

58. Ibid.: parinamarupasya sakartrkakarqkasya va vyaparamatrasya karma-
samjna.
59. Ibid.
60. svantaraviratagatimayatvaj jagad iti; purve namarupe parityajya namaru-
pantarapraptih; ibid.
61. parimitasvarupatvam: «whose nature is delimitation» (or, «being deli
mited»), ibid., p. 39.
62. Ibid.
110 E. Gerow
penetrate, are formed of the same matter ([karma] and imply the same
agent [s]): they are not one atop another like stories of a palace. Gods
are simply more subtle agents63. The worlds interpenetrate; karma is
pervasive. Other characteristics of the processual notion implied by this
analysis are (1) its uninterruptability, (2) its terminability only by
inherent result (phala) (or from the point of view of the agents, the
[1] latency and [2] the certitude [bondage] of karman). There is no
accident in this comprehensive system, for only in karman does the world
achieve definiteness. Accountability and responsibility merge. The ran
dom is literally the unthinkable.
Even moksa is seen as a phala (the negative phala, terminating
phalas) requiring its own karman, great, in fact superhuman, effort
(prayasa) 64. There is no instant release, just as there is no randomness.
Thus Bhide leads us into the ethical domain of karman: falling in the
chain of existence is easy, rising is hard (falling means « transforma
tion » from the subtle to the gross, rising the opposite: both are « pari-
namas ») and to the Indian, equivalent to the categorical imperative:
for the only way in sum to rise is to help (selflessly) each other (those
below you in the chain of existence) (since all are equally bound). But
we do it not to help them, but to acquire merit for ourselves; the ulti
mate selfishness remains for we cannot escape so easily from ourselves,
that is, from the karmani that constitute our « selves ». And equally,
those below owe debts (of gratitude?) to those who have helped them
rise to their present condition (and the interpenetration of all karma
rules out the possibility that we can have risen through selfishness
alone), specifically the three debts that are the cornerstone of the elite's
daily ethic: to the gods, to the fathers, and to the teachers/seers. The
karma that is required of us in recognition of their help which, in turn,
helps us by accelerating the wave of karma produced, is called, in the
case of the Gods, etc., yajna, or yaga, in the case of the pitrs, sons
(expressed as the necessity of pinda) and in the case of the teachers,
study and teaching (continuing the tradition). In all cases, it is by a
specific karma that the effects of karma are modified to our greater
good: kim api viîstam karma kartavyam6566 .
Finally, seen from the ultimate perspective of its own modification,
or its progress through various agents (« bhave »), karma is said to be
threefold: prarabdha (begun), when having been done (= krta) it is
now working itself out through some agent or other; samcita (accumu
lated), when it is done, but has not yet found its « agent » (is latent);
and kriyamana (or samclyamana) when it is now in the course of being
done (in this body)
63. Ibid., p. 40.

64. Ibid., p. 42.
65. Ibid.'. « Let some special karma be undertaken ».
66. Ibid., p. 43.
What is Karma (Ki>}t karmeti)? An Exercise in Philosophical Semantics Hl
In this synthetic account of karma, many of the leading themes of

the various dictionary definitions are marvellously correlated and the
ethical and cosmic «senses» are derived in admirably grammatical
fashion6768from basic meanings that are sua natura grammatical. Merely
by summing up the principles of Bhide's analysis we may assert in what
sense the cultural karma is nothing but the grammatical karma (as inter
preted by the neo-grammarians),
The first, already noticed, is the equation kriya = vyapara = karma.
The second is the pervasive subordination of the kartr (agent) to the
karma: Not only is the verbal idea indifferent to (more basic than)
agency (indifferent to whether or not there is an agent) but the place
of the agent is strictly dependent on a primordial act (karman-jagad-bhu)
which is the field of the agent's activity and achieves differentiation
through the kartr (almost as karana!6a). Moreover, the ethical model of
karma makes clear the sense in which (to the reversal of our perso
nalized views of ethics) the karma uses the kartr, in a sense, to « purify »
itself. Freedom, as in the Gita, involves suppression of the agent, but
not the act! Thirdly, in dramatic terms, activity (kriya), which is kar
man, and not kartr, is given the status of independent or first principle,
capable of subsisting (satta!) on its own (in fact being nothing but con
tinuity [ = change, parinama], and calling into play agents and other
conditions as its modalities of expression (the « world » is the affirma
tion [= modality] of karman).
To conclude, the world achieves the status of a (maha) vakyam69
only when karma is raised to the dignity of an independent principle.
But why is the term karman needed at all? Why would not kriya suffice?
In fact the older sense of karman, as adjunct to a kriya (the « direct
object ») appears to have been sublated into its very principle, by in
fact canceling the older distinction between karman and bhava. And
indeed if the verbal idea is indifferent to agency (as having « vyapara »
inherent within it) it is equally indifferent to result. It is, in Bhartrhari's
terms a krama, a gradation whose unity is a fiction organized by the
using mind focussing on « result ». The only difference between a transi
tive and an intransitive construction lies in externals: in the former a
substratum outside the verbal act was explicitly called for; in the latter
it was not called for, but was present in the very act itself: « to sit».
The karma model in its cosmic and ethical senses equally asserts the
primacy of activity and result over their external substrates: activity
is terminable only by its inherent result, which inevitably (save the case
moksa) is another activity; activity and result are ultimately one if the
67. Grammar is « derivation » (vyakarana) in the Indian sense.

68. Kartjrkaranayos trtiya (1): P2.3.18, « Les desinences du trosieme cas valent
quand il s'agit de I'agent, de I'instrument» (Renou, vol. 1, p. 115).
69. A « grand assertion » or argument comprehending in coherent fashion all
individual assertions (Mimaipsa term. Not to be confused with Samkara’s « maha-
vakya », i.e. « upanisadic citation »).
112 E. Gerow
notion that agents are primary is given up, for it is only in terms of
the agent that activity and result appear different (goal, effort, etc.).
Everything falls into place once the karman of the older grammar is
abstracted from its external support (the direct object) and shown to be
(from the angle of the verbal root itself) nothing but the bhava, or ten
dency to be definite: the vyapara, the process.
It is perhaps carrying this argument into the realms of poetics to
assert that the notion of karma itself is indeed an inescapable function
(and result) of the passivization or impersonalization of the Sanskrit
sentence. But that is in fact what we have concluded. Both Advaita and
Bauddha philosophies overcome « karman » by a process of (sentence)
analysis akin to the one we have examined.
It is not surprising that of the various philosophical formulations
of « agency » and « action », it is the Advaita that most rewards our
examination. Bhartrhari is himself an early advaitin, at least so con
sidered by the tradition, which labels his version « sabdadvaita » — so
to distinguish it from the more authoritative (and later) Samkaradvaita.
In all the Advaitas the key to unlocking the ontological confusion that
compromises our natural experience is proper understanding of agency:
kartrtva. The natural man confounds (identifies) agency and Ego, says
«I walk », « I make mats », « I am cold, hot... » (or rather, confuses
« agency » and « consciousness » — the result being Ego). Advaita here
uses the older Samkhya analytic dualism to great effect. The point of
philosophical re-evaluation of this experience — which also redefines the
nature of things — is to disassociate consciousness and agency: the
« unconscious » (prakrti )becoming the « agent » of all true propositions,
the conscious « subject » being liberated of all predicates. We now assert
two instead of one: « It is cold, hot... » and « I am (...) ». Grammatically
speaking, simple assertive propositions involving personal agents are
no longer possible, and each one has to be analytically dissovled into a
predicate (or content: hot, etc.) and a « subject ». But terms as such
cannot be propositions; in the place of one, we must apply a « dummy »
subject for the real predicate (the English « it » - prakrti) and an empty
predicate for the real subject (purusa/atman, as unqualified « conscious
ness »). In the Paninian analysis of these new propositions, the former
is alone genuinely interesting, as containing all possible real content of
a worldly nature. And it is precisely the « agency » of such propositions
that has disappeared — agency as something other than and opposed
to act, as predicate. We can say, if agency goes, all that remains on
which to ground the assertion is « karmani » (for those propositions
where the act does pertain to an external « object» («the town is
walked to [by... X] ») or « bhave » (for intransitive verbs and existential
predications, where no ground is available ex hypothesi save the act
asserted itself: It is sat down (by... X). We thus see in the proper (true)
advaitic assertions the correct realization of the grammatical categories
themselves. Which came first, the chicken or the egg?
What is Karma (Kivp karmeti)? An Exercise in Philosophical Semantics 113
The « grammatical» problem for advaita is neatly solved by making

all sentences with real content «passive» (karmani/bhave), in fact
«impersonal»; but the question of the « other » sentence remains: is
there a grammatical model also for a « contentless » active assertion?
Notions of « creative energy » (sakti etc.) in the other systems might
seem to fit this need, but in Samkaradvaita at least (probably not in
Bhartrhari's), creative force itself (as in Samkhya) is located on the
wrong side of the division, and in fact corresponds to the maya, nothing
more than the « X » in our putative « impersonal» sentences supra.
Brahman/atman understood in transcendental terms certainly provides
a notion of absolute universality that is inactive and thereby contentless;
perhaps all we can say is that « agency » itself becomes equally imper
sonal when abstracted from all possible predicates. Grammatically, this
side of the Advaita provides more problems than solutions: the « abso
lute » is anirvacaniya.
It is likely the Bauddha sastra that provides us with the most logi
cally satisfying philosophy — one that is in complete accord with the
new «language ». And this is done by simply facing the difficulty just
mentioned, and denying the need for any « active » sentence at all —
contentless or not: the doctrine of nairatihyam. In effect, for the Baud
dha, there is no agent. Change simply is; it is inherent in being; it is
being, it needs no « explanation ». Our « bhave » sentences now become
the only possible expressions: not « I am white / I sit » but « It is
white... / It is seated... ». As Hiriyanna says « Thus according to Bud
dhism, when we say for instance “ It thinks " or " It is white ", we mean
by the “ it” nothing more than when we say “ It rains ” »70. In other
words, the « It» is truly a « samghata », a dummy term — not even
needed in the less analytical Sanskrit, where the «passive » (kriyate)
had been generalized to all verbs — to verbality itself (asyate). (The
peculiarity of « varsati» is thus not that it has no « subject», but that
the verb retains the external « active » form — an archaic survival no
doubt — impeding the expected novelty *vrsyate!). And the «it» does
not imply, does not correspond to a universal positive « subject » (as
in the advaita atman), of which it signifies the absence in « real» sen
tences. This, it seems to me, is nearly an exact replica (in « philoso
phical » or «metaphysical » terms) of the position attributed to the
grammarians and to Bhartrhari. So that Hari appears once again more
kin to the Bauddha, than in fact to the standard (Hinduized) Advaita.
It is not, let us emphasize, the contextual senses of the passive71
that we are concerned with, but rather, the grammatical form itself, and
the interpretation given there to by the Indian philosophical grammar.
70. Outlines of Indian Philosophy, p. 140.

71. So thoroughly classified and discussed by Gonda, op. cit., and with whose
caveats we entirely agree: as for example that the passive can and must be used
in those contexts where it is inconvenient or impolite or impossible to designate
the agent, and is thus not a «pure equivalent» of the active.
114 E. Gerow
Let us, nevertheless, attempt to draw the strands of our three sections
together by illustrating the ethical implications of « passivization » as
a generalization of syntax meaning; put in another way, we will invert
the usual order of « explanation » (and this is good advaita methodology)
so that instead of employing a theory of syntax to give good interpre
tations to the many individual sentences that illustrate it, we will
attempt to express the general sense in every one of those sentences
imparted to it by the form of its expression.
From the beginning, Indian grammar determined the « subjects » of
every sentence in the verb form itself. Panini located the kartr (agent)
and the karman (object) in the verbal termination that indeed identified
the complete form as verb72. The suffix thus appears to determine both
verbality and subject: pacynte odanah (Devadattena), pacati odanam
Devadattah; and to be ipso facto the ground of the sentence syntax. It is
equally true that the sentence spells out the implications of the formal
act-theory (via the karakas) that becomes impliable only when the root
is determined as verb (I mean, the noun « Devadatta » implies in itself
no karaka, no effectuating of the act implied in every sentence, until
its verb is specified); and also the surface realization of the sentence
(the «cases » automatically follow once the verb is determined in its
subject role [kartari or karmani])73.
From this angle, the formula attributed to Hari appears only to
generalize the sentence form: to find a more universal notion of verba
lity than the one which determined verbality only in a choice (kartr
or karman). It in no way alters the pre-eminence enjoyed by the verb
in the sentence syntax, nor its role in determining the sentence's subject.
But if that subject is not to be determined in a choice, that is, by
reference to the external agent or object (result) of the activity, it can
be found only in the universality of that reference, that is, in the vyapara
or phala which underlies the choice and defines the « sense » of rootness.
But if we were to ask, is it the vyapara or the phala that determined
subject-capacity and pre-eminence of the verb, we would not have
achieved any more general stance than the older grammar, we would
simply have replaced one choice with another (albeit in the root now
and not in its ending). The only solution is that the verb itself (the
« root» itself) in virtue of its very nature must be capable of standing
as its own subject, and in so doing is pre-eminent in the sentence. The
ratio karman : phala :: kartr : vyapara is abrogated: the vyapara, the
« processual meaning » of the verb is boldly identified with the verb in
its resultative guise: karman. After all the only purpose served by
72. The « root» (dhatu) per se as we have repeated ad nauseam underlies both
nouns and verbs, which are realized (and differentiated) only by specific suffixes.
73. Cases like « pacaka », « cook », where the nominal suffix is said to have the
sense « agent of» (cf. English -er) are easily accomodated, for in such cases it is
to the subordinate root «pac» that the nominal « agent» looks, not to the inde
pendent (verbal) root elsewhere in the sentence.
making the distinction vyapara/phala was to justify the syntactic distinc

tion kartr/karman! And if we ask, what can this « general sentence »
mean? we must reply with the caveat that we have not so much « passi
vized » the sentence, as we have made it neutral to the distinction
active/passive. The most convincing exemplification of such sentences
is found in the case of the so-called passive in transitives. Here there
simply is no subject external to the verb: none is possible, for intran
sitives have no objects (by definition) that could be made into subjects
of a passive construction. So if such sentences are possible, and not
only possible, but uniformly realizable for every intransitive, then it
cannot be denied that for one class of sentences at least, the verb (in
its « bhava » mode) is the subject in virtue of its processual function,
and not simply as agreeing with or underlying an externally exemplified
karaka-term.
It is a short step to the general semantics of such sentences: the
verb in its verbality (processual meaning: vyapara) is paramount, and
that « meaning » is ipso facto subject of every utterance (karman in the
sense vyapara, viz. «bhava»). Every sentence on the model of the
passive intransitive asserts « karman » as its subject; in fact « karman »
amounts to determining a verb as subject: determining the process as
capable of being referred to. Karman and vyapara are distinct not so
much (now) as aspects of the meaning of (verb) roots, as they illustrate
two levels of grammatical analysis: vyapara captures the sense of the
root semantically, « karman » syntactically.
Now if every utterance asserts a karman which is none other than
the verb process itself, we have, it seems, correctly derived the ground
of the philosophical karma theory from its own grammar; for as N. N.
Bhide asserts, the karma theory is a theory of actional metaphysics
wherein every act (read: fully realized assertion: kriya) devolves from
a primary or pre-existent result (karman) which is nothing but (the
nature of) activity itself (the « world » itself is nothing but vyapara).
The « karma » theory may then be compared to the equivalent theory
in English, that would assert the primacy and universality of the «it»
that we find implicit as subject in every otherwise subjectless sentence:
it is raining: it was seated by Devadatta (Devadattena asyate); and by
virtue of such implicitness, present also in every realized (passive)
subject: the food (it) was cooked by Devadatta. It-ness in this sense is
the « karma » theory, and as English differs in its capacity to form such
sentences (« it was seated by Devadatta» being impossible), so does
the English speaker resist theories of universal « passivization ». And
equally, as Bhide so cogently remarks, the karma theory is so inherent
in Sanskrit that none have deemed it necessary to propound it. What
to us is exotic is to the Indian a pre-condition of speech. « Karma » is
a turn of phrase... Translation becomes a function of the sophistication
of our grammar.
116 E. Gerow
Now the issues we have raised in this paper have been quite care
fully expressed in terms other than those used to formulate the old saws
about the « passive » character of the Hindu mind, or the influence of
passive syntax on the contemplative bias of Hindu philosophy74. We
are not to be taken as giving voice to a new version of these discredited
socio-cultural notions, which seem to assert some behavioral consequence
of a grammatical cause. Our object here has been simply to explore the
relations between two theories (and, therefore, between two inherently
universal structures which have no immediate behavioral implications)
to see in what way they may illumine each other or shed light on the
problems of their formulation. Both indeed are aspects of a large struc
ture of Indian or Hindu intellectuality, and both doubtless have beha
vioral and practical implications. But I do not wish to be thought
dabbling in this area. It is enough if certain very general theories are
seen possibly to be related, and indeed that a philosophical theory of
action may indeed merely restate in objective terms the same theory
of action carefully worked out and assumed in a tradition of highly
literate Sanskrit speakers. Which is the cause and which effect is in this
sense not interesting: what is interesting is the possibility that we speak
our philosophies along with our grammars.
University of Chicago.
74. Compare the observations of H. Nakamura which are moderate enough.

What Is Karma? (Ki Karmeti) - An Exercise in Philosophical Semantics (Edwin Gerow)

Uploaded by

Copyright:

Available Formats

What Is Karma? (Ki Karmeti) - An Exercise in Philosophical Semantics (Edwin Gerow)

Uploaded by

Document Information

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

What Is Karma? (Ki Karmeti) - An Exercise in Philosophical Semantics (Edwin Gerow)

Uploaded by

Copyright:

Available Formats

E.

WHAT IS KARMA (KIM KARMETI)?

At a recent conference on « karman and rebirth »*, I was widely

1. Workshop on Karma and Rebirth; Lake Wilderness, Washington, Oct. 22-23,

4. JA, 1941-42, p. 157; reprinted Staal, A Reader on the Sanskrit Grammarians,

This inquiry is about a grammatical problem and is grammatical by

Each of these reference points is supported by an observation which

13. Nils Simonsson points out an exception in the Nyayakosa.

pages are devoted to karma and compounds beginning with karman-I7;

necessary only to justify the explanation of its abstract meanings, « tad

dual). Yatna, or psychological « motion » is considered, on the contrary,

24. Vakyapadiya 3 (sadhana), 45 (Iyer ed., p. 266).

27. VBS § 5, p. 68: svarthaphalavyadhikaranavyaparavacakatvam (sakarma-

possibility of error26 built into the semantics, but, insofar as the

«verbal» at all in Paninian grammar] involved two semantic dimen­

31. [Later termed] «vyapara» and « phala ».

resembles arbitrary selection34 (again « motivated » by the . speaker's

pac + kartr odana + karman Devadatta +

pac + karman odana + [agreement: Devadatta + kartr

The single convention needed to form correct strings is that of « ana-

37. Shortcuts adopted in the derivation. Cf. supra, note 32.

ending [here optative or injunctive] which is primary and has the

In the sentence given above, there is no real problem:

41. By 3.1.67, « sarvadhatuke yak », cf. Cakdona, « Karakas », pp. 24143.

kartr) may be postulated to subtend its «two » meanings. It is equally

42. Tarkavacaspati, p. 1736.

ultimately through the notion of sequence or krama. An act expressed as sequence

The ubiquity of « passive » syntax in later Sanskrit has often been

43. Gonda, op. cit.

or passive intransitive construction that is typical, wherein is empha­

Conch obar/Conn sends (messengers) to Conchobar. That messengers is not implied

two « aspects » of root meaning, with the « bhava(na) », in effect restates

lore on the subject of karma, not so much unaffected by Western views

Bhide's analysis is shot through with grammatical distinctions and

57. Ibid., p. 38.

63. Ibid., p. 40.

In this synthetic account of karma, many of the leading themes of

67. Grammar is « derivation » (vyakarana) in the Indian sense.

The « grammatical» problem for advaita is neatly solved by making

70. Outlines of Indian Philosophy, p. 140.

making the distinction vyapara/phala was to justify the syntactic distinc­

74. Compare the observations of H. Nakamura which are moderate enough.

You might also like

«verbal» at all in Paninian grammar] involved two semantic dimen

or passive intransitive construction that is typical, wherein is empha

making the distinction vyapara/phala was to justify the syntactic distinc