Distinction between linguistic and external reality

In his contribution to a recent symposium (Does Asia think differently? –Symposium zu Ehre Ernst Steinkellners), Johannes Bronkhorst answered that yes, there is a substantial difference between “our” thought and the Indian one, in so far as the latter does not distinguish between purely linguistic problems and genuine ones. For instance, Indians care for the status of an object which is linguistically present before its actual existence, such as a pot in “the potter makes the pot”. However, many Buddhist schools seem to aptly distinguish between the two and so do, as far as my knowledge reaches, at least also Mīmāṃsakas. For instance, Rāmānujācārya speaks of karman and kriyāphala as two distinct realities (cf. Tantrarahasya, IV §3.13.2: kriyāphalaśali karma).

V. Eltschinger–IV part

A criticism against Dharmakīrti as presented by V.Eltsch.:

The end of the second chapter (pp. 110-114) is dedicated to the establishment of the validity of whatever has been said by the Buddha through the validity (inferentially established) of what he said in regard to the Four Noble Truths. But how does one know that a reliable speaker of X (=the four noble truths) will be reliable also in a fully different sphere? One does not expect an expert of anatomy to be reliable in regard to codicology! Even the fact that the Buddha must be compassionate, since he communicated to us the four noble truths, is not enough, because he could have been compassionate and willing to help, but still not an expert in a certain area.

V. Eltschinger–on yogipratyakṣa

In the second chapter (“La doctrine métareligieuse de Dharmakīrti–unluckily enough the author does not spell out the meaning of “meta” in this case), p. 88, fn. 66, V. Eltsch. proposes the following verse:

PV. III.285-6:tasmād bhūtām abhūtaṃ vā yad yad evābhibhāvyate. bhāvanāpariniṣpattau tat sphuṭākalpadhīphalam.

But I am not absolutely sure of how could something non-existing be seen in a non-conceptual (akalpa) way. Does not the very fact that it does not exist imply its being of conceptual nature? Luckily enough, V. Eltsch.'s translation adds that it just seems to be non-conceptual.

V. Eltschinger-III part

One of the main qualities of this book is that it stresses Dharmakīrti's statement that one needs a Sacred Text (in V.Eltsch.'s words: une Écriture) in order to initiate an activity. That is, although V.Eltsch. does not put it so explicitly, the other instruments of knowledge are reliable as far as the “material” world is concerned (I cannot pause here on the exact connotation that material would have in a Buddhist milieu and hope that the common sense understanding of it will suffice for the moment). They are reliable insofar as the description of the world as it is is concerned. But, they cannot tell one what he OUGHT to do. This thesis would be shared by Mīmāṃsakas, too. Hence, say the latter, the Veda has to be valid, otherwise there would be no key as to how to achieve (enduring) happiness. Buddhists, on the other hand, are content to say that the Buddhist texts, though not by themselves valid, are inferentially proven to be valid. My question is: where does the thesis of the non-derivability of the ought from the is derive from, within Indian Philosophy? In fact, it is not shared by, e.g., Naiyāyikas, who ground the validity of the Veda exactly in its epistemological validity insofar as our common world is concerned (namely, healing of poisons).

Vincent Eltschinger -II part

At the end of his first chapter, V.Eltsch. inscribes Dharmakīrti's contribution within the coeval Buddhist philosophy. One understands that 

Dharmakīrti dote ainsi le bouddhisme d'une philosophie complète, mobile et réputée ne faire appel qu'aux lumières ordinaires de la connaissance humaine (p. 64).

I am willing to agree with this analysis if it refers to the Buddhist milieu alone. Dharmakīrti was surely trying to enhance the blind faith of the other Buddhists whilst strengthening it with rational arguments. But I am not absolutely sure that he was also rationally convinced that those rational arguments would have been powerful enough to make everyone a Buddhist convert. In sum, rational arguments were thought by Dharmakīri to be powerful enough to make one's faith resist and to be eventually victorious in debates against Mīmāṃsakas and other opponents. But this does not mean that the same arguments could have effected a change of faith in those very opponents. Does V. Eltsch. agree?

Book discussion/Buchbesprechung: Vincent Eltschinger's “Penser l'autorité des Écritures”

The starting point of every discussion on this book is that it is an amazing achievement: more than 160 pp. of translation of Dharmakīrti's Pramāṇavārttika and Svavṛtti which will enhance further studies on Buddhist and non-Buddhist epistemology and, even more important if possible, enable non-Sanskritist readers to have a glance of Dharmakīrti's arguments on the epistemological foundation of one's beliefs and actions.

The first lot of my comments is directed on the long introductory study which is, explains the author, only a methodological device to avoid an over-loaded annotation. In fact, the introductory study is sometimes (most of all in the first chapter: Contexte historique et idéologique –V. Eltschinger is a passionate historian), a page-turner. Sometimes, on the other hand, it resembles more a collection of arguments needed to understand the following translation and hence runs the risk to be either too long (to be a handy compendium for the translation) or too short (if one –as in my case– looks for a philosophical insight into the Buddhist-Mīmāṃsaka debate).

More in detail, my only genuine criticism is about §1.3.3, p. 45, where the author describes the Mīmāṃsā's understanding of the Veda as

Les Smṛti et les Purāṇa ne sont source autorisées de connaissance du dharma que parce qu'ils corroborent une révélation incréée, seule inscrite dans l'ordre même des choses (emphasis of the author).

This would hint at the idea that there is a natural order of things and that the Veda is part of it. On the other hand, I have rather the impression that the Veda is felt by Mīmāṃsakas to be of a complete different nature as our common experience and to be the only possible authority in its precinct. That is, as for dharma, there is no order of things which enables the Veda to be the only authority, rather, the realm of dharma is tantamount to Veda. Moreover, the Veda is an instrument of knowledge (pramāṇa) only as far as its prescriptive part is concerned. Hence, no descriptive knowledge of dharma can be driven out of it, whereas V. Eltschinger's passage and even more his hinting at the order of things seems to imply that the Veda has an epistemological validity in its descriptive rather then prescriptive passages.

Longer discussions on arthabhāvanā in Bhāṭṭa texts

The following quote has also be found in Mīmāṃsākoṣa. I will complete its translation in the next days.

ākhyātopasthitā arthabhāvanā kiṃ kena kathaṃ ity aṃśatrayaṃ krameṇāpekṣamāṇā “kim iṣṭaṃ kuryāt” ityākāṅkṣāyāṃ yāgasya kaṣṭatvena karaṇatvena ca asādhyatayā “yāgena anyad bhāvayet” iti (uttaram). tataḥ “kim idaṃ tat anyat?” ityākāṅkṣāyāṃ svargakāmapadasamarpitaṃ asiddhatayā sādhanāpekṣaṃ svargam eva arthabhāvanā bhāvyatvena svīkaroti. śrutivṛttyā hi padārthabhūtasya svargakāmasya bhāvyatvam ucyamānaṃ viśeṣyabodhe viśeṣaṇe svarge paryavasyati. tathā ca yāgena svargaṃ bhāvayet iti phalitam. yāgakaraṇikā svargaphalikā pravṛttiḥ pravartanāviṣayaḥ iti bodhaḥ. yatra ca samabhivyāhṛtānupasthitaṃ phalaṃ tatra prakaraṇādinā teṣām apy abhāve arthavādena vidhisāpekṣeṇa (rātrisatreṣu), tasyāpy abhāve viśvajinnyāyena phalaviśeṣaḥ kalpyate. yady api puruṣapravṛttirūpabhāvanāniṣpādyo yāgaḥ tatra na karaṇaṃ, tathāpi phalāvacchinnabhāvanāyāṃ viśeṣaṇe karaṇatve (karaṇe) karaṇatvopacāraḥ iti phaloddeśyakapravartakakṛtiviṣayatvam eva atra karaṇatvam. daṇḍena ghataṃ karoti ityādau loke 'pi tathaiva karaṇatvam. kena ityāsyāpi kiṃviṣayiṇī kṛtiḥ svargasādhanaṃ ityarthaḥ ity anye. atra yāgakaraṇatvaviśiṣṭe lakṣaṇā iti prāñcaḥ. vastutaḥ karaṇatvaṃ yāgasya saṃbandho vākyārthaḥ. evaṃ jyotiṣṭomādes tatra abhedenānvayo nirābādhaḥ. itthaṃ ca kathaṃ vinaṣṭena yāgena svargo bhāvyaḥ iti pratītyanavasāne antarā apūrvaṃ kṛtvā iti saṃbadhyate. tataḥ “katham apūrvaṃ yāgena kartavyaṃ” iti apūrvabhāvanāyāṃ itthaṃbhāvena saṃyujyamānam itikartavyatājātaṃ apūrvaprayuktam ity ucyate. tataḥ prakārāpekṣāyāṃ “samidho yajati” ityādivākyaiḥ iṣṭaviśeṣāpekṣaiḥ svādhikāravākyaikavākyatayā “darśapūrṇamāsabhāvanā prayājādibhāvanopetā apūrvaṃ sādhayati” ity avagate prayājādīnāṃ tādarthyena samanvaye paścāt upakāratvaṃ (upakārakatvaṃ) kalpyate. tad ayaṃ bhāṭṭamate bodhakramaḥ. “svargakāmaniṣṭhā yāgakaraṇikā svargaphalikā prayājādītikartavyatākā bhāvanā” ity arthabhāvanābodhaḥ. “vidhiniṣṭhā śaktigrahakaraṇikā stutyarthavādopakṛtā pravartanā” iti śabdabhāvanābodhaḥ śaktikalpanena lakṣaṇayā vā upapadyate. tato 'varuddhā arthabhāvanā niruktapravartanāviṣayaḥ ity ekavākyatayā mahāvākyārthabodhaḥ. ata eva vidhyuparaktā bhāvanā liṅarthaḥ iti bhāṭṭasiddhāntaḥ. vidhiḥ pravartanā taduparaktā tadviśiṣṭā arthabhāvanā liṅpratipādyā ity arthaḥ. arthabhāvanāviśeṣaṇatvenaiva śabdabhāvanāyā anvayaḥ rājādikṛtājñādeḥ tathaivānvayāt. tadanvayena yogyatvena pravartanāyām api balavadaniṣṭhānanubandhīṣṭasādhanatvaṃ labhyate. “na surāṃ pibet” ityādau pratiṣedhavākyārtho 'pi ittham evonneyaḥ evam anye 'pi anvayabodhāḥ svayamūhyāḥ iti. (Mīmāṃsācintāmaṇi, pp. 92-93).

Approximate translation (for the time being, I did not polish the English form of this translation, which is only meant as a guide-line through the Sanskrit text):

The arthabhāvanā lead [to the listener] by verbal forms (ākhyāta), expects three components [answering], in due order [the following questions] “What [has to be brought about]?”, “By means of what?” and “In which way?”

As for the expectation “Which desirable [thing] should one do?”, since the sacrifice cannot be the thing to be realised, since it is [by itself not desirable at all, but rather] grievous, and since it is instrumental, the answer is “through the sacrifice, one should cause to be something else”. Hence, [when one asks] “What is this other [thing]?”, since [heaven] has not been accomplished and needs an instrument to be realised, the bhāvanā accepts heaven alone, delivered by the word «the one who is desirous of heaven» [in the prescription «The one who is desirous of heaven should sacrifice with the Full and New Moon Sacrifices»], as the thing it has to cause to be. […] Hence, the [prescription] ends up by meaning “through the sacrifice, one should cause to be heaven”.

I have no keys as for prāñca. Who are those “Eastern” Mīmāṃsakas?

Besides the problem of arthabhāvanā, it is impressive how such analyses of arthabhāvanā and its components are common throughout Bhāṭṭa texts. I read similar depictions in the Bhāṭṭa pūrvapakṣinaḥ of Rāmānujācārya's Tantrarahasya as well as in Āpadeva's Mīmāṃsānyāyaprakāśa and so on.

Various Bhāṭṭas on arthabhāvanā

As reported in the Mīmāṃsākoṣa:

arthabhāvanā eva kārakaviśiṣṭarūpeṇa vākyārtho mīmāṃsakamate (Mīmāṃsābālaprakāśa p. 73).

arthabhāvanā eva śabdabhāvanābhāvyā (Mīmāṃsācintāmaṇi p. 91).

arthabhāvanā ca pacati pākaṃ karoti iti karotisāmānādhikaraṇyāt sarvākhyātapadavācyā (Mīmāṃsābālaprakāśa p. 74).

arthabhāvanāṃ vidhipratiṣedhau sākṣāt viṣayīkurutaḥ. taddvārā tu tadviśeṣaṇāni kārakāṇi. tad uktaṃ tadbhūtādhikaraṇe “viśiṣṭāṃ bhāvanāṃ prāpya vṛttir vidhiniṣedhayoḥ” iti (Mīmāṃsābālaprakāśa p. 74).

arthabhāvanāyāḥ cetanamātrāśrayatve 'pi śabdabhāvanāyāḥ cetanāśrayatvāyogaḥ (Nyāyasudhā, p. 578).

arthabhāvanāyāḥ sarvākhyātavācyatvam (Somanātha's Mayūkhamālikā, ad 1.2.1, p. 4).

The Nyāyasudhā's quote is interesting in view of the Pārthasārathi-Someśvara debate about bhāvanā as prayatna (implying a cetana agent) or pravṛtti (so that it agent could even be a non-sentient one).

Arthabhāvanā in Kumārila

yadā hi sarvākhyātānuvartinī karotidhātuvācyā puruṣavyāpārarūpā bhāvanā avagatā bhavati, tadā tadviśeṣāḥ sāmānyākhyātavyatiriktaśabdaviśeṣavācyāḥ vidhipratiṣedhabhūtabhaviṣyadvartamānādayaḥ pratīyante. pacati, apākṣīt, pakṣyati, pacet, na pacet iti (TV ad 2.1.1.1, p. 378).

That is,

When indeed the bhāvanā –which has the form of a human activity, can be expressed by the verbal root “to do” [and] is present in all finite verbal forms– is understood, then optative (lit. prescriptions and prohibitions), past, future, present etc. –which are expressed by a certain word (i.e., their specific verbal ending) distinct from the general finite verbal ending [and] are its (the bhāvanā's) characteristics– are [also] understood. For instance, “he cooks”, “he cooked”, “he will cook”, “he should cook”, “he should not cook”.

Arthabhāvanā as puruṣabhāvanā in Someśvara Bhaṭṭa

icchārthāt arthayater ṇijantād arthayata iti kartṛvivakṣāyāṃ “erac” (Āṣṭādhyāyī 3.3.53: ikārāntāt dhātoḥ ac pratyayaḥ syāt) iti ac-pratyayotpādanena arthinaḥ puruṣasya arthaśabdenābhidhānāt, bhāvanāyāś ca puruṣadharmatvāt, dharmadharmiṇoś ca atyantabhedābhāvāt tādātmyaṃ vivakṣitvā arthātmā cāsau bhāvanā ca iti vigrahaḥ kāryaḥ (Someśvara's Nyāyasudhā p. 560).

So, according to Someśvara (which is much closer to Kumārila than Bhaṭṭa Śaṅkara and hence possibly more reliable), “artha” means “puruṣa”. Arthabhāvanā is interpreted by him as a karmadhāraya, since a characteristic (such as the person's initiation of the activity) and the characteristic-substratum (the person herself) are not ultimately different and can, hence, be expressed with such a composite.

Arthabhāvanā in Bhāṭṭacintāmaṇi

In Kevalānandasaraswatī's (sic) impressing Mīmāṃsākoṣa the following explanation can be found:

arthabhāvanā … arthātmatvaṃ arthaviṣayatvaṃ arthāśritatvaṃ vā. tathā ca arthaniṣṭhatvād arthabhāvanā phalotpattyanukūlakṛtinodanādivyāpārarūpā ākhyātatvāvacchinnaśakyā cetanācetanakartṛsādharaṇī kartṛśaktiḥ (gāgābhaṭṭa's bhāṭṭacintāmaṇi p. 90-91). 

that is (I am not translating artha, since its meaning in this connection is exactly what I am trying to understand):

arthabhāvanā [is] the fact of having an artha as its essence, as its content or as its support. In this way, since it rests on an artha, the arthabhāvanā is the potentiality of a doer, which is common to conscious and non conscious doers, liable to be determined by verbal endings [but why -tva?], having the form of an activity such as impelling and being and action favorable to the arousal of the result.

Which does not solve the problem of its name. Then, the author of the Koṣa quotes many other authorities (Kumārila, Bhaṭṭa Śaṅkara, Somanātha's commentary “Mayūkhamālikā” ad Pārthasārathi Miśra's Śāstradīpikā). Most of them do not talk at all about the etymological meaning of artha in arthabhāvanā. The longest discussions on arthabhāvanā in general are to be found, according to the Koṣa, in Gāgābhaṭṭa's Mīmāṃsācintāmaṇi and in Someśvara Bhaṭṭa's Nyȳasudhā (commentary ad TV). The latter also contains an explanation supporting Prof. Kataoka's analysis (see next posts).

śabdabhāvanā

I just found an interesting paper of Prof. Kunio Harikai, (Abhidhābhāvanā, in Journal of Indian and Buddhist Studies 36 (18.2), pp.605-606), describing how abhidhābhāvanā has been interpreted as a tatpuruṣa or as a karmadhāraya by the Bhāṭṭa tradition, that is as 

abhidhīyate 'nena iti abhidhā

or as

abhidhīyate yā sā abhidhā

The first interpretation is to be found in Mādhava's Sarvadarśanasaṅgraha:

abhidhābhāvanā ity atra abhidhāśabdena “abhidhīyate anena” iti vyutpattyā śabda ucyate.

Many other Mīmāṃsakas are mentioned (Pārthasārathi Miśra, Someśvara Bhaṭṭa, Umeśa Miśra, Āpadeva). I hope I will find someone who translates for me the exact references, so that I can discuss them.

Kumārila's obscurity and Vācaspati on it

Why did Kumārila choose not to explain what he meant with “artha” in arthabhāvanā or which kind of compound is śabdabhāvanā? And why did he offer two explanations which are even more obscure, such as arthātmikā bhāvanā and śabdātmikā bhāvanā? The couple has been repeated in mantrādhikaraṇa and bhavārthādhikaraṇa (at least), hence it is not a badly deviced hapax legomenon.

In a passage pointed out by Harikai (1990:209), Vācaspatimiśra comments Maṇḍana's proposal of better discriminating (pravivic-) the prescription and rhetorically asks what is left to discriminate after Kumārila's and his followers' analyses. His answer is:

kṛto 'pi tair viveko 'kṛakalpa eva | “abhidhābhāvanām āhur” ityādisaṅkīrṇaśabdaprayogāt

(Nyāyakaṇikā ad Maṇḍana Miśra's Vidhiviveka, 1907 Banares p. 4, l. 17)

“Although they have done the discrimination, it is almost undone, because of the usage of confused expressions such as “[optative and other endings] express the designation-bhāvanā …”

Obviously enough, Vācaspati is not entirely reliable, since he wants to interpret Kumārila as maintaining the same view proposed by Maṇḍana (according to which the prescription conveys the idea that the action it enjoins is the means to accomplish something desired).

Prof. Kunio Harikai on bhāvanā

Hisataka Ishida, a young and learned scholar of Pramāṇavāda and Mīmāṃsā, kindly accepted to translate for me some pages of Harikai's translation of Kumārila's Tantravārttika, mantrādhikaraṇa. This is what I understood:

–Harikai translated bhāvanā with sokushinryoku, something like “motivating energy”. If I am not wrong, this means that Harikai understands (just like Kṛṣṇa Yajvan, see a previous post) prayojakavyāpāra to refer to “bhāvanā” alone (and not to arthabhāvanā or śabdabhāvanā, with artha and śabda featuring as the prayojakas).

–Harikai translates śabdātmika and arthātmika (adj. of bhāvanā) as “having language as its essence (honshō)” and “having a purpose (mokuteki) as its essence (honshō)”, respectively. In a footnote (fn. 99, p.208), he explains that those translations have been inspired by the bhavārthādhikaraṇa.

–Furthermore, Harikai refers to a definition of Kumārila:

tatrārthātmikāyāṃ bhāvanāyāṃ liṅādiśabdānāṃ yaḥ puruṣaṃ prati prayojakavyāpāraḥ sā dvitiyā śabdadharmo 'bhidhātmikā bhāvanā vidhir ity ucyate

Which brings me again to the problem of the exact understanding of the reference of prayojakavyāpāra (śabda-bhāvanā or just bhāvanā?). Moreover, how to understand śabdadharma? If the śabdabhāvanā “is a characteristic of language”, does this modify our understanding of the compound śabdabhāvanā? I guess that śabdadharma suits good a karmadhāraya interpretation, but does not crash against a tritiyātatpuruṣa (śabdena bhāvanā), too. I will come back to this topic.

"purpose creating bhāvanā"

Further, indirect evidences of bhāvanā being preceded in a compound by the object it produces are apūrvabhāvanā and phalabhāvanā, both common in Prābhākara texts, such as Rāmānujācārya's Tantrarahasya or Śalikanātha Miśra's Prakaraṇapañcikā.

Sanskrit usages: vibhu vs. vyāpaka

Alessandro Graheli suggests me the following distinction between vibhu and vyāpaka:

vyāpāka means “all-pervasive”; vibhu is the consequence, as far as size is concerned, of such all-pervasiveness, hence its opposite is “madhyama” (of middle=normal size). I may suggest: “all-pervasive” and “infinitely-extended”. See also Laine, Journal of Indian Philosophy 1998.

Kṛṣṇa Yajvan on arthabhāvanā

Kṛṇa Yajvan, who was also more or less a contemporary of Āpadeva and Laugākṣi Bhāskara, writes in his Mīmāṃsāparibhāṣā:

bhāvanātvaṃ nāma bhavituḥ prayojakavyāpāratvam. tatrārthabhāvanāyāṃ bhavitur jāyamānasya svargādeḥ prayojakavyāpāratvāl lakṣaṇasaṅgatiḥ. śabdabhāvanāyām api puruṣapravṛttirūpasya bhavituḥ prayojakavyāpāratvāl lakṣaṇasaṅgatiḥ. 

Which, I believe, can be translated as follows:

The fact of being a bhāvanā consists in being the activity inducing (prayojaka) something which is [through that] about to be brought into existence. Among those [activities], the definition suites the “arthabhāvanā” insofar as this is an activity inducing heaven and so on, which is brought into existence, that is, born. The definition also suites the “śabdabhāvanā” insofar as this is an activity inducing something which is about to be brought into existence [and] which has the nature of a human activity (vyāpāra). 

So,

1) prayojakavyāpāra is here equated to bhāvanā (and not to X-bhāvanā), hence, one does not need to presuppose something like śabda-bhāvanā=prayojakavyāpāra.

2) the symmetry between the two holds on only insofar as the bhāvanā part is concerned. The first part of the compounds is not compared.

3) artha is possibly thought of as hinting at svarga.

Obviously enough, this regards only Kṛṇa Yajvan and not necessarily Kumārila. I have already collected different definitions by other Bhāṭṭas.

Paper manuscripts in South Asia

Birgit Kellner asks on the Indology list which are the oldest dated paper manuscripts in India and Nepal, mentioning for India a mns dated 1198 and for Nepal a mns dated 1105. Why did paper manuscripts remain comparatively rare until recent times, then? Was paper too expensive? Were trade-routes from China and Tibet not commonly used? After all, paper was well spread in China and Tibet since much older time.

Later Mīmāṃsakas on arthabhāvanā

I have already mentioned Bhaṭṭa Śaṅkara's interpretation of artha in arthabhāvanā as “puruṣa”. It is such a surprising interpretation, that I would like to know how he came to that. I never read it in any other author and Bhaṭṭa Śaṅkara does not explain in detail why it has to be like that. He rather presents it as if it were self-evident, a hint to the fact that it was an interpretation well-spread in his milieu? Or does the cursory reference refer to the minor importance of the etymological interpretation? In fact, other Mīmāṃsakas just do not bother in etymologically explain the meaning of arthabhāvanā (unlike they do with śabdabhāvanā). So Rāmānujācārya and Āpadeva, the author of Mīmāṃsānyāyaprakāśa. In his excellent translation of this work, Franklyn Edgerton uses “end-efficient-force” for arthabhāvanā, but he does not explain how he came to that translation.

Another possible symmetry between śabdabhāvanā and arthabhāvanā

Kei Kataoka maintains that śabdabhāvanā and arthabhāvanā ought to be symmetrical. Hence, śabde bhāvanā=śabdabhāvanā. arthe bhāvanā=arthabhāvanā. Artha ity ukte kim? puruṣaḥ.
But I am not fully satisfied by this understanding of artha. Moreover, arthātmikā bhāvanā is hardly understandable as puruṣātmikā bhāvanā. On the other hand, the proposal that Kumārila just devised two symmetrical names not thinking of any symmetry between the two is also unsatisfying, especially in so far as Kumārila painstakingly tried to develop two parallel theories for the two bhāvanās, although he had obvious problems with the karaṇa and the bhāvya (which is not at all desirable) of the śabdabhāvanā. See, e.g. Rāmānujācārya's refutation in Tantrarahasya, IV, §3.7.2. 

So, what about the next proposal:
śabdabhāvanā= bhāvanā ca śabdaś ca, sā (karmadhāraya)
arthabhāvanā=bhāvanā ca artha ca, sā (karmadhāraya).
Artha ity ukte “prayojanaṃ, śabdārthaṃ ca”. Kasya arthaḥ? śabdabhāvanāyāḥ. śabdabhāvanāyāḥ arthaḥ puruṣavyāpāraḥ iti arthabhāvanā puruṣavyāpārabhāvanā iti yāvat. ata eva kumārilabhaṭṭena “arthātmikā bhāvanā” iti, “śabdātmikā bhāvanā” iti ca uktam. ācāryaḥ “arthātmikā” iti uccaraṇasamaye “puruṣavyāpārātmikā” iti manyate. “puruṣavyāpārātmikā” ca pārthasārathimiśrasya pustake, rāmānujācāryasyāpi punaḥ punar viditam.