Philosophically astute and meticulously researched, Roy Tzohar’s book A Yogācāra Buddhist Theory of Metaphor describes the evolution of a sophisticated Classical South Asian theory of intersubjective world creation in the absence of ultimately real universals. The idea that language is purely conventional because the enduring objects and subjects to which we take words to refer simply don’t exist in reality is one of the hallmarks of South Asian Buddhist thought, and a good deal of the brilliance of the 6th century Yogācāra Buddhist philosopher Sthiramati’s pan-figurative view hinges on his understanding of the ability of metaphors to convey meaning in the absence of an object that could serve as the primary referent of a word. As Roy presents the upshot of Sthiramati’s theory of language: “all words are metaphors insofar as their nonexistent primary objects are absent from their loci of reference. Hence, a word’s actual referent—that which it designates figuratively—is not an object, but rather the entire causal process that makes the object appear in our consciousness” (Tzohar 2018, 169). This picture of language connects words to the causal processes that constitute reality not through the presence of an essential connection between word and meaning, but precisely through the absence of primary objects of reference.

Roy’s book is a significant accomplishment on many fronts, not the least of which is its deft presentation of various non-Buddhist traditions in dialogue with which the Buddhist Yogācārin picture emerges. In the spirit of this inter-traditional engagement, I’d like to draw on a slightly later South Asian tradition, Pratyabhijñā Śaivism (10th-11th century), to push against one of the core arguments of the Buddhist philosophers in Roy’s study: that there is simply no way to make sense of language use if one holds on to an essential grounding for the relationship between a word, its referent, and its meaning. At one point in the dialectics of his book, Roy paraphrases the basic framework of an essentialist theory of reference as being one where “a relation of correspondence between the designation and the particular object is not the outcome of convention—i.e., it is determined in a manner that logically precedes and is not influenced by any actual language usage, and it is grounded (somehow) in the relations between the essential nature and the object” (Tzohar 102).  This would mean that “the comprehension of meaning would occur without requiring any actual language use, either in thought or speech,” which in turn “leads to many absurdities, most notably the rendering of all language use as an outer manifestation of an incommunicable private ‘language’ whose semantics is determined prior to and independently of any actual language use” (Tzohar 102). These absurdities do sound bizarre, and not like claims most language users would rush to defend. Yet, with the major clarification that the inner language in question would be nonconceptual rather than incommunicable, they’re not far off from the Pratyabhijñā understanding of meaning. 

From Utpaladeva (10th century) onward, Pratyabhijñā authors employ aspects from a number of distinct (and generally incompatible) traditions, both Vedic and Buddhist, to fill out their account of how words come to life. Very roughly, Pratyabhijñā appropriates the Vedic Grammarian Bhartṛhari’s (5th century) levels of vāc (very roughly, language) to provide an ontology of language that has both nonconceptual and conceptual forms. But this tradition does not use Bhartṛhari’s own theory meaning; rather, they rely on the Vedic Mīmāṃsāka understanding of the eternality of the Sanskrit phonemes to ground the possibility of meaningful language use, but then also employ the Buddhist apoha (exclusion) theory of concept formation to explain how the eternal relationship between signifier and signified at the level of the phonemes comes to be expressed in everyday language, which is entirely arbitrary and conventional. 

As one would expect, the details of how such a broad-ranging amalgamation works are very complex, and are beyond the scope of this preface. For our current purposes, a stray moment in the Pratyabhijñā philosopher Abhinavagupta’s (10th-11th century) commentary on the Tenth Discourse of the Bhagavad Gītā offers a helpful window into this tradition’s understanding of the world-constituting role of metaphor.  The well-known narrative arc of the Bhagavad Gītā as a whole moves from Arjuna breaking down at the thought of killing his friends, relatives, and teachers to Kṛṣṇa convincing Arjuna that he should fight. The climax of the narrative comes when Kṛṣṇa grants Arjuna a direct perception of Kṛṣṇa’s all-encompassing, all-pervasive and terrifying cosmic form. Before Arjuna can ask for and receive this direct insight into reality, however, Kṛṣṇa must prepare Arjuna to see his world from a radically new perspective. The last step in Arjuna’s transformation occurs through Kṛṣṇa’s description of his own divine forms, in which he offers a series of apparent metaphors that present himself existing as, within, and beyond the universe. Notably, these metaphors do not simply express that Kṛṣṇa is the best of all things. Kṛṣṇa names himself as everything from risk among those who cheat, to the self that dwells in all beings, to the Himālayas among the mountains, to Arjuna himself among the Pāṇḍavas (van Buitenen 1981, 109–11). The rapidly shifting frames brought about by counterposing so many seemingly incompatible designations are themselves the point of the performance.

[1] ahamātmā ityanena vyavacchedaṃ nivārayaty anyathā sthāvarāṇāṃ himālayaḥ ityādivākyeṣu himālaya eva bhagavānnānyaḥ, GAS ad 10.42 (Abhinavagupta 2002, 236).

As Kṛṣṇa’s final statement—“I support this entire universe with but a single portion of myself!” (van Buitenen 1981, 111)—indicates, although the enumerated namings sound like a series of metaphors, they are simultaneously quite literal truths: Kṛṣṇa is, in fact, everything. As Abhinavagupta comments, however, this statement cannot be reduced to a simple assertion of identity because any strict delimiting of identity necessarily excludes what it is not. At the end of the list of forms, Abhinava remarks: “This statement ‘I am the self’ precludes division. Otherwise, in statements such as ‘Among mountains, I am the Himālayas,’ the Lord would only be the Himālayas, and not the other mountains.” [1]. The ‘division’ precluded here is the idea that Kṛṣṇa and the world are two distinct entities. When such a truth is posed within a restrictive frame, as is the case when Kṛṣṇa contrasts himself as the Himalayas against other lesser mountains, the implied limitations on him cannot be understood literally. However, rather than negating that reality can be named, this technique affirms that every name in fact expresses reality. At least as Abhinavagupta reads it, by performing the contradictions inherent in any attempt to name the ineffable, Kṛṣṇa creates a fusion of human and divine, self and other, immanence and transcendence that leads to a radical transformation of Arjuna’s understanding of himself and his world. The fact that reality, including Arjuna’s current situation, is nothing but Kṛṣṇa both makes possible conventional limitations, but also undoes them at the limit.

Abhinavagupta understands the upshot of this rapid succession of metaphors that collectively indicating the fundamentally figurative nature of reference in a very different way than do Yogācāra Buddhists. Recall that, following the dominant analysis of figurative language in Classical South Asia, Yogācārins understand metaphor as occurring when the primary referent is absent from the locus of reference. Focusing on Asaṅga’s (4th century) analysis, Roy describes:

“Figures, in this picture, are no longer viewed as mere content carriers… Instead, by virtue of their referential interchangeability and repetition, they undermine a monosemic world-word correspondence, and in this sense are performative: reflexively invoking and calling to mind a certain understanding of the nature of language, not through any particular content, but by the mere fact of their repeated presence… For Asaṅga, then, the uniqueness and usefulness of figures lie in their ability to demonstrate through their very existence and employment by language-users that which is otherwise hidden behind habit and convention—namely, the lack of any essentialist grounding for language” (Tzohar 108).

For Asaṅga, as Roy indicates, the fact of figurative interchangeability indicates that there’s no essential connection between word and meaning. And yet, at least as read by Abhinavagupta, Kṛṣṇa’s apparently metaphorical enumeration of his forms in the Tenth Discourse serves precisely to indicate that any meaningful distinction is essentially grounded in Kṛṣṇa himself. Neither the restrictions nor the totality can be understood literally, but that’s not because of an absence of the presence of the primary referent. Rather, what’s absent is an understanding of the totality of meaning out of which distinctions are carved. The key point is that the limitations experienced within the conventional world are both part of the ultimate because there is nothing (including conventional dualities) that is not part of the ultimate, and unable to truly limit the ultimate because the ultimate exceeds any duality. 

Even in passages with significant apophatic elements, Abhinavagupta often returns to some kind of affirmation of the real existence of differentiation. The following passage from another of Abhinavagupta’s texts gives an indication of his complex interweaving of affirmation and negation when trying to speak of the ultimate:

[2] kiṃca śāstram idaṃ samyag bhagavadyogadeśakam // bhagavadyogam advaitaṃ nirdvandvaṃ ca pracakṣate/ tasyopadeśa itthaṃ syād yadi yāvadvibhedavat // saṃbhāvyate tan nirbhajya nirbhajyaiva nirūpyate/ advaite bhairavavibhau yat praveśopaveśayoḥ // ābhyāsikī sthitir nāsti tau hi bhedaikajīvitau/ ataḥ saṃbhāvyanikhiladvaitaśaṅkāvyapohane // gurūṇāṃ ca śiśūnāṃ ca yatnaḥ sarvo vijṛmbhate / ato dvaitam ihāśaṅkyāśaṅkya sarvaṃ pratanyate // tad yāvadgati saṃbhāvya na tu kutrāpy udāsyate / tathā hi yadi nāmṛṣṭaṃ dvaitaṃ tarhy ekam eva sat // cidbrahma tad alaṃ tattvasaṃkhyākalpananirṇayaiḥ / pañcatriṃśatitā kasmāt tattvānāṃ tan nirūpyate // tasmād dvaitasya bhedātmashiter yāvadgati graham / kṛtvā yas tatpratikṣepas tena niḥśaṅkatā bhavet //, MŚV 110cd-117 (Hanneder 1998, 76).

The practice of the Lord is declared to be nondual and beyond divisions. Instruction in it should be as follows: to the extent that one imagines something to possess division, it is explained by analyzing it again and again. A practice for entering and dwelling in the nondual omnipresent Bhairava does not exist, for both [entering and dwelling] are enlivened only by difference. Therefore, all effort on the part of teachers and students reaches its fullest extent by removing the inhibitions about any duality that they have imagined; therefore, having doubted duality again and again, everything is revealed here. That duality should be imagined for as long as one lives, and never set aside. Indeed, if duality is not considered, then ultimate consciousness exists just as one; in that case, what is the point of the determination, conceptualization, and enumeration of the principles of reality? Why describe the thirty-five principles of reality? Therefore, having accepted duality, which is the essence of difference, for all one’s life, one should become free from inhibitions by rejecting it [2].

[3] advaye tattve bhedo ʼpi na na yujyate/ idaṃ hi tat parādvaitaṃ bhedatyāgagrahau na yat//, MŚV 1.123 (Hanneder 1998, 78).

As Abhinava concludes, one should never entirely forsake duality since “within nondual reality, difference is also not not possible. Indeed, this ultimate nonduality is not the grasping or rejection of difference.” [3]. For Abhinavagupta language does not invalidate itself upon reaching its limit. Rather, in affirming its own limitations, language also affirms its own truths. Abhinavagupta’s ineffable ultimate is one about which everything can be said, not just nothing.

[4] bhede tu viśvabhāvānāṃ svasvabhāvavyavasthiteḥ / abheda iti śabdo ʼyaṃ manye bhedayate rasāt //, MŚV 1.124 (Hanneder 1998, 78).

In a rather contemplative moment immediately following this comment that duality is “not not possible” within nondual reality, Abhinavagupta remarks: “However, within difference, because all things are established through their own individual natures, I think that this word ‘nondifferent’ will become differentiated naturally.” [4]. This personal reflection seems to indicate Abhinavagupta’s keen awareness of the problem with holding any one idea—including even the idea of nonduality itself—as the sole, highest formulation of truth. However, just as no truth is the whole truth, no truth is ever entirely false, either. For Abhinavagupta, reality is a question of degree. Erroneous awareness-events aren’t wrong. They’re just incomplete.

Roy’s exploration of Yogācāra accounts of intersubjectivity results in a similarly complex model of what it means for something to be real, and both the Yogācārins and these Śaivas develop the capacity to see reality as a gradient of truths that allows interaction between entities with vastly different epistemic circumstances. However, there is a highly intriguing difference between how these two traditions view the relationship between language and ultimate reality. For the Yogācārins, the ultimate is ineffable in every sense; it is utterly beyond language. But, as we’ve seen, for the Pratyabhijñā Śaivas, the ultimate is most certainly nonconceptual, but it is not beyond language. Moreover, they claim, this must be the case if we are to account for meaning. From Roy’s analyses, we can clearly see that karma plays the role that nonconceptual forms of language play for the Śaivas; different life forms may have both shared and idiosyncratic experiences because they operate in relation to the causal potentials of the dependent nature. These Śaivas reject this move to make karma a defining characteristic of the ultimate. Rather, these Śaivas see causality itself as an emergent structure that comes along with the creation of subject/object duality, which is also the root form of conceptualization, which is also the emergence of time. For both traditions, language and causality are intimately bound up in creating conventional worlds, but they seem to be diametrically opposed on which of these forces is more fundamental: language for Pratyabhijñā Śaivas, causality for Yogācāra Buddhists.

To return, then, to the absurdities with which we started. For Pratyabhijñā Śaiva philosophers, conventional meaning is indeed the externalization (and internalization!) of inherently meaningful nonconceptual language. Reality speaks itself as the divisions, both of objects and of subjects, that constitute worlds. The shift from eternal, nonconceptual language to conventional concept use happens via delimitation, where subject/object structured worlds are carved out of the totality of language, which is identified with the totality of reality itself. Language is not exhausted by conceptuality; the ultimate is beyond concepts but not beyond language. All of these ideas are argued for in a context that happily takes on board a great deal of Buddhist philosophy of language. As Roy himself mused when I first told him about all of this, perhaps the Pratyabhijñā analyses make it seem arbitrary to view ultimate reality as ineffable either because nothing can be said of it, or because everything can be said of it. The pan-figurative view of language seems to be able to support both. For various other reasons, I don’t actually think that this choice is an arbitrary one ⎯ but that is a topic for a different time. 

A final note on how to understand absurdity itself. If nothing is not not possible within nondual reality, then absurdities should be something other than simple impossibilities. Perhaps intuitions about absurdities are faultlines in intersubjective worlds. They’re guides to the limits of the reality within which they appear that are potentially meaningful because reality overflows any limits. Like any delimitation, they can invite us to think through what we’ve excluded in our quest to make sense.

References

Abhinavagupta. 2002. Abhinavagupta’s Commentary on the Bhagavad Gita: Gītārtha-Saṁgraha. Translated by Boris Marjanovic. Varanasi: Indica Books.

Buitenen, J. A. B. van van. 1981. The Bhagavadgītā in the Mahābhārata: Text and Translation. Chicago: University of Chicago Press.

Hanneder, Jürgen. 1998. Abhinavagupta’s Philosophy of Revelation: An Edition and Annotated Translation of Mālinīśloklavārttika I, 1-399. Gröningen: Forsten.

Tzohar, Roy. 2018. A Yogācāra Buddhist Theory of Metaphor. Oxford, New York: Oxford University Press.

Roy Tzohar’s A Yogācāra Buddhist Theory of Metaphor examines upacāra, here understood as ‘metaphor’, with particular reference to 6th century CE Buddhist Yogācāra thinker Sthiramati’s ‘pan-metaphorical’ claim that all language is figurative (and the influences of his claim). Tzohar hopes to provide a ‘conceptual history’ of upacāra (Tzohar 11). He relates this to broader Yogācāra interests such as the three natures (trisvabhāva), storehouse consciousness (ālayavijñāna) and mind-only (citta-mātra) teachings. This study considers the paradox at the heart of Buddhist theories of language: that language is both an obstacle and a vehicle to liberation, playing an active role in both regards. Tzohar mimics the argumentative strategy of the texts in question and leads not with an ontological discussion of whether language corresponds to external objects, but rather with the epistemic quagmire of whether our language can ‘access’ knowledge and concepts. In doing so, he operates within a Buddhist framework that arises from early Indian linguistic discussions of referential meaning. Although there is plenty of philosophical content in this book, theories of language remain Tzohar’s point of reference.

[1] This is Tzohar’s explanation of some contenders’ interpretation. He introduces the term ‘mentalist view’ in reference to Puṇyarāja’s reading of Bhartṛhari (see Tzohar, 50). Here, a mentalist view is that which takes ‘meaning as being determined by whatever cognitive content arises through the word’s denotive power, regardless of the ontological status of the referent’ (ibid).

[2] The problem of incommensurability asks how we can have intersubjective agreement if language is figurative all the way down (i.e. how are we seemingly able to agree with each other about certain experiences/perceptions). If all language is secondary signification, not grounded in the presence of a corresponding external object, then Sthiramati must account for intersubjective agreement about the world and the possibility of incommensurability between speakers. Sthiramati tackles this problem by pointing to the causal nexus.

[3] It is worth noting that the Yogācāra developed a doctrine of eight consciousnesses (aṣṭa-vijñānakāyāḥ), though there is disagreement amongst modern scholars over whether to take the final three as further categories of consciousnesses (introduced in addition to the pre-Yogācāra schema) versus transformations of consciousness. The eighth consciousness, the storehouse consciousness (ālaya-vijñāna) is the repository for karmic seeds (i.e. karma legacy). For an introduction to Yogācāra philosophy and Mahāyāna Buddhist thought more broadly, readers could begin with Mahāyāna Buddhism: The Doctrinal Foundations by Paul Williams (London: Routledge, 2009).

The book is structured in three parts, each containing two chapters. In Part One Chapter One, Tzohar presents two non-Buddhist Mīmāṃsā and Nyāya school understandings of figurative language to contextualise his later discussion of Yogācāra works. This chapter explores two elements of early Indian scholastic discourse: linguistic Realism, whereby language was considered meaningful if it had word-world correspondence, and also the notion of language as eternal – rooted in  Vedic interpretations. In Chapter Two, Tzohar provides another  comparator by presenting Bhartṛhari’s  (450–510 CE), Sanskrit Grammarian and author of the Vākyapadīya (VP), theory of meaning. Bhartṛhari’s ‘mentalist view’ [1] presented a contextualist referential theory of meaning in which ordinary language use does not reach beyond itself to touch reality. The curious reader can find two appendices in Tzohar’s book, both translated excerpts from Bhartṛari’s VP. Parts Two and Three of the book move on to Yogācāra figures. Part Two, Chapter Three looks at Asaṅga (c. 360 CE) and Chapter Four examines Vasubandhu (c. 360 CE), Dignāga (c.480–540)., and the Laṅkāvatāra Sūtra (LAS). The focus on Sthiramati in Part Three (Chapters Five and Six) is the most innovative and ambitious component of the monograph. In exposition of Sthiramati’s Triṃśikābhāṣya (TriṃśBh) and Asaṅga’s Mahāyānasaṃgraha (MS), Tzohar illustrates how Sthiramati built on the work of previous Indian philosophical thinkers and linguists. Sthiramati’s theory of metaphor took a bare bones view of the term ‘metaphor’ as denoting when a word’s primary referent is absent from its locus of reference, which he decides is true for all language about phenomena. Sthiramati tackles the question of intersubjective agreement and incommensurability [2] by arguing that language refers to the causal nexus that produces our illusory experiences and perceptions. All language is therefore figurative – since what we understand as the primary referent is always absent from the locus of reference – yet is still meaningful in effective intersubjective communication for its salvific value through our shared referent of causal sequences. Experiences and objects we deem ‘real’ are functions of the transformation of consciousness (vijñāna-pariṇāma) [3]. 

Having addressed Tzohar’s widely recognised masterwork on Sthiramati I turn our attention to his study of Asaṅga concentrating on Part Two Chapter Three: ‘It’s a Bear…No, It’s a Man…No It’s a Metaphor! Asaṅga on the Proliferation of Figures’ (Tzohar 77-124). This includes Tzohar’s underdiscussed contribution to elucidating Asaṅga’s take on the ineffability paradox.

This chapter highlights two works ascribed to Asaṅga: the Tattvārthapaṭalaṃ chapter of the Bodhisattvabhūmi (TApaṭ, BBh) and its accompanying commentary in the Viniścayasaṃgrahaṇī (VS). Asaṅga is principally concerned with the performative role of figurative language. Tzohar begins with noting the lack of scholarly consensus on the relative dating and authorship of these texts (Tzohar 81-84), providing an example of conflicting approaches in the works of Janice Willis (1979) and Hartmut Buescher (2008). Willis follows the traditional understanding in Buddhist Studies scholarship of the TApaṭ and VS as contemporaneous, complementary projects by Asaṅga. She grounds her analysis in one understanding of the three natures doctrine. By contrast, Buescher sees the VS as a later work that showcases significant doctrinal development since the TApaṭ. He thus analyses TApaṭ without reference to the three natures doctrine outlined in the later VS.  Tzohar hopes to pave a middle way between these positions in his own analysis of TApat and VS. Tzohar reviews without fixating upon the literature pertaining to the relative dating and authorship of these works (Tzohar 79-84) - a general feature of his writing I appreciated throughout the book.

4. A tetralemma is a fourfold figure in Indian Buddhist logic: it enquires into whether something is A, B, both A and B, or neither A nor B. Here, ‘A’ and ‘B’ stand for different possibilities. In this case, ‘A’ and ‘B’ are apprehension through the designation and apprehension through the object designated.

Asaṅga’s primary argument concerns the ineffable nature of reality and the incoherence of essentialist theories of meaning – Asaṅga’s concern is epistemological not ontological. Rather than dwelling upon ascertaining the ultimate essential nature of things, Asaṅga draws attention to the accessibility of this ‘essential nature’ to knowledge via language. His critique wields a dual strategy: pointing to the ‘polysemy of figures – a circumstance in which various metaphors (upacāras) denote the same object’, and ‘systematically scrutinizing all the possible relations between these three elements (the thing, the designation, and an essential nature)’ (Tzohar 85). The latter fuels a threefold argument in the TApaṭ,  then revised as a tetralemma [4] in the VS where Asaṅga spells out the options for ‘the locus of essential nature’: that the essential nature is (1)  ‘apprehended equally’ through the object and its designation (the mentioned ‘argument from polysemy’), (2) apprehended solely through the designation, (3) apprehended solely through the object, and (4) apprehended through neither the object nor its designation (Tzohar 85). Asaṅga uses the common scriptural analogy of a magic show (for example also found in the first chapter of the Saṃdhinirmocana Sūtra (SNS)) to explore the inexpressible nature of reality and the capacity of enlightened beings to see the causal continuum behind illusions (Tzohar 92). These visual elements recur in Tzohar’s monograph, and later in the chapter he recalls how Asaṅga uses a polysemy of figures to argue against essentialist theories of meaning (Tzohar 106). In the VS, Asaṅga describes a magician who creates an illusion that takes on the forms of different creatures in quick succession. These varying forms represent polysemy: figurative language can create multiple designations for the same designatum that is therefore not fixed to only one designation. Thus ‘the semantic-ground will never be entirely dependent on only one designation/figure, as it can always derive its identity from the essential nature related to another figure’ (Tzohar 107). The polysemy of figures shows ‘their referential interchangeability - is viewed as diagnostic of the “free-floating” nature of language, whose words are not tied to essentially existent things’ (Tzohar 108, italics original) though this language epistemological claim makes no ontological claims about such existent things. Thus figurative language, as illustrated through the analogy of magical creation employed in the VS, challenges monosemic theories of word-world correspondence. Through these arguments, Asaṅga demonstrates the inexpressibility of essential nature (as it is commonly understood), challenges essentialist theories of meaning, and exposes the ‘self-referential nature of language’ (Tzohar 124).

I enjoyed the comparison between Nāgārjuna (c. 150–250 CE) and Asaṅga’s argumentative strategies, which contributes to Tzohar’s explanation of how Asaṅga apprehends the meaningfulness of his own discourse despite the illusory nature of language (Tzohar 110, 116). Throughout the book, Tzohar shows an awareness for the difficulty of talking about language when we cannot step outside of it. His masterly treatment of the inexpressibility paradox in Asaṅga’s work should receive recognition alongside his analysis of Sthiramati.

Though I have much praise for Tzohar’s monograph, I do have several criticisms. I sympathise that established understandings familiar to Buddhist Studies academics like myself might not be available to readers outside the field. Tzohar’s writing is relatively dense: a student might not need a dictionary to wade through technical terms, but his work requires firm familiarity with Yogācāra thought and the philosophy of language with an analytic philosophy bent. Whilst his use of specialist language is not accessible for all graduate students, I found Tzohar’s prose easier to process on a second reading. Part Three offers clearer writing. Importantly some of Tzohar’s strongest explanations and intertextual knowledge are found in footnotes throughout. The way that Tzohar builds strings of sentences and concepts is certainly very skilful but can be difficult to follow. He embarks on close textual analysis whilst simultaneously trying to weave together broad themes and questions, ultimately demanding frequent ‘summary’ or ‘recap’ paragraphs. These appear at various points in the chapters, not only at the end of subsections. Even with this signposting, it is sometimes tricky to follow Tzohar’s line of argument, and a reader can easily lose sight of the chronology of figures and works explored. This is particularly true for the diverse definitions of upacāra under which the book’s thinkers operate without sufficient or collective contrast. Though Tzohar made the decision not to include thinkers who came after Sthiramati in the history of Yogācāra thought, the book may have benefitted from recommendations for further reading (for example, on Chinese manuscripts of these early Yogācāra texts). Ultimately, I found that the monograph reads as a conceptual history of Sthiramati’s upacāra rather than upacāra in itself, and since Tzohar does not compare different thinkers to Sthiramati throughout, this concentration only becomes truly evident in those final two chapters that focus on Sthiramati.

I found the comparative points between the Mahāyāna Buddhist schools of  Madhyamaka and Yogācāra somewhat restrained. Tzohar presents Yogācāra – or more specifically Sthiramati, given my above pronouncement – as the Mahāyāna school that overcomes the problems purported to accompany the paradox of language’s delusive yet salvific power. There was potential to examine these Buddhist refutations of essentialist theories of reference through understanding both schools’ doctrines as approaches to teaching non-Self and/or expressions of non-Self (anātman). If words and their designations do not have an ‘essential nature’ as we might understand it, perhaps these Yogācāra writings on language and the nature of reality lead back to anātman (language is non-Self) in the same way that the Madhyamaka teaching on emptiness leads back to anātman (all phenomena are non-Self) – both being skill-in-means (upāya-kauśalya). Though Tzohar discusses the similar argumentative strategies of Nāgārjuna and Asaṅga when they defend their ‘self-negating’ positions (Tzohar 110), I was surprised that he did not take the opportunity to make further doctrinal connections between Yogācāra philosophy of language and keystone Buddhist concepts such as anātman, though perhaps this would not be in keeping with his subtler style of presenting material.

This being said, A Yogācāra Buddhist Theory of Metaphor is a great scholarly achievement: both methodologically and in its new emphasis on Sthiramati’s contributions to Yogācāra thought as an original thinker beyond commentary. In particular his handling of the externality and intersubjective agreement conundrum in Chapter Six is outstanding and left me with a greater appreciation for the subtleties and diversity of Yogācāra doctrines on the nature of reality. Despite my reservations about his writing, Tzohar’s commitment to tackling such intricate philosophical arguments whilst still providing close readings of individual texts – and a number of strong original translations – is impressive. His surveys of scholarly literature, such as the aforementioned discussions in Chapter Three and the well-known Yogācāra studies ‘Idealism’ debate in his Introduction, are robust outlines. Throughout the monograph, Tzohar addresses thorny debates in scholarship without falling into their abyss, acknowledging historical discourse while continuing purposefully with the task at hand. Tzohar confidently considers the provenance of both Yogācāra texts and Buddhist Studies scholarship without becoming distracted from his own project.

Tzohar’s analysis of a complex topic can be expectedly difficult to follow, though his handling of historical scholarly debates, close textual analyses, and new translations are highly commendable aids to the patient and diligent reader. A Yogācāra Buddhist Theory of Metaphor is a landmark contribution to scholarship on Yogācāra thought, offering a methodical study and fresh insight into Yogācāra theories of language and perception – particularly in Asaṅga and Sthiramati’s works.

Roy Tzohar’s A Yogācāra Buddhist Theory of Metaphor is an outstanding work on Yogācāra philosophy, especially Sthiramati’s innovative and commentarial synthesis of doctrinal elements from different Yogācāra sources. Tzohar’s analysis of Sthiramati is the focus of this review. I will begin with a brief summary of Part Three (Chapters 5-6) followed by comments and questions which require further discussion.

Chapter 5: ‘What It All Comes Down To: Sthiramati’s Pan-Metaphorical Claim and Its Implications’

In Chapter 5, Tzohar proposes that the Yogācāra stance is more concerned with the meaningfulness of Buddhist doctrine than the lack of ontological commitment in Madhyamaka philosophy. To justify and legitimize the overall meaningfulness of the Yogācāra salvific discourse, whilst at the same time admit the inexpressibility of reality, Sthiramati innovatively replaces a theory of meaning with a theory of sense. This explores the ‘causal and mental underpinnings of language’ (Tzohar 7). Sthiramati presents a ‘pan-figurative’ claim to all language use whilst at the same time establishing the meaningfulness of the Yogācāra’s own discourse (Tzohar 154-57).

According to Tzohar’s analysis, Sthiramati begins his argument with the stance that the self and all things do not substantially exist and can only be discussed in terms of mental appearances. These constructed mental appearances are the result of the transformation of consciousness (vijñāna-pariṇāma), a changing process of ‘causal mental development which is identical to the Buddhist notion of dependent-arising (pratītya-samutpāda)’ (Tzohar 160). The terms ‘self’ and ‘things’ are figurative because ‘their primary referents are absent from their locus of reference’ (Tzohar 160). Since the ‘self’ and ‘things’ can stand for all phenomena in the Buddhist view, all language use is figurative designation because all phenomena are absent from the locus of reference. 

Tzohar introduces Sthiramati’s refutation of his opponent’s views, an opponent who claims that ‘the metaphorical meaning is necessarily parasitical on another meaning that is literal and primary’ from two perspectives: the universal (jāti) and the individual particular (dravya) (Tzohar 162). Sthiramati counter-argues that there is an inherent gap between what we say and what we mean, ‘between the apparent referents of words and their alleged actual referents (Tzohar 164). He argues that language can be meaningful without referring to real entities. There is a distinction between the primary referent/constructed object (mukhyo padārthaḥ), its experienced qualities (guṇa) and its assumed essence (svarūpa) (Tzohar 165-66). Sthiramati reaches the conclusion that ‘there can be no correspondence between words and their referents’: only their qualities are accessible to language (Tzohar 166).

In this sense, metaphors are the denotation marking the presence of certain qualities based on mental appearances (Tzohar 167). Mental appearances, according to Sthiramati, are constructed as the transformation of consciousness, which is ‘a momentary causal mental process’ (Tzohar 168). The transformation of consciousness is the locus of reference of metaphors. Furthermore, this causal mental process, constantly in flux, is identical to the notion of dependent origination. At the same time, it operates as a conceptual construct (vikalpa). A word’s actual referent is not an object, but rather ‘the entire causal process’ giving rise to that object’s appearance. The sense of language is ‘roughly the description of this causal process in terms of the Yogācāra model of consciousness’ (Tzohar 169). Here, a theory of meaning is replaced by a theory of sense. 

For Tzohar, the Yogācāra school benefits greatly from this understanding of meaning. The Yogācāra established ‘different levels of discourse and posited a hierarchy of meaning within the conventional realm of language’ (Tzohar 170). Language which uses more accurate and more complete causal descriptions is more meaningful, even though these descriptions are conventional. In this way, different levels of meaningfulness of discourse are distinguished within the conventional realm of language. The referent of a term can remain fixed even when the meaning of a term changes (Tzohar 171).

Since the momentary causal mental process has qualities, it is imagined and constructed as various entities in the conventional realm. Therefore, we are called to accept that there is ‘some underlying extranominal reality’, the momentary causal mental process – the nature of dependent origination – existing within the conventional realm (Tzohar 174). This is where the Yogācāra school disagrees with the Mādhyamikas, who emphasise that selves and phenomena are merely nominal. From the Yogācāra perspective, the Madhyamaka view is a kind of ontological nihilism. 

I generally agree with the underlying logic in Tzohar’s arguments, where he skillfully discusses metaphor, mental appearances, the transformation of consciousness and dependent origination, and links all these to the meaningfulness of language without compromising on the idea that referents are absent from the locus of reference. However, there are some points requiring further attention. 

[1] Sthiramati, and Buescher, Hartmut. Sthiramati's Triṃśikāvijñaptibhāṣya, 38: atha vā vijñāvad vijñeyam api dravyata eveti kecin manyante | vijñeyavad vijñānam api saṃvṛtita eva na paramārthata ity asya dviprakārasyāpy ekāntavādasya pratiṣedhārthaḥ prakaraṇārambhaḥ |

[2]《唯識三十論頌》：「由假說我法， 有種種相轉。 彼依識所變， 此能變唯三」(CBETA 2021.Q3, T31, no. 1586, p. 60a27-28).

1. Tzohar argues that Sthiramati’s Triṃśikāvijñaptibhāṣya was not concerned with the ontological nihilism of Madhyamaka school as much as with the meaningfulness of Buddhist doctrinal discourse. I cannot find this attitude in TriṃśBh. Sometimes, Tzohar magnifies this reading of anxiety about the meaningfulness of Buddhist teaching, implying it was a concern in the entire Buddhist community. This seems an over-estimation of the situation. Contrary to Tzohar’s observation, I would argue that Sthiramati has little anxiety about the meaningfulness of Buddhist teachings given the nature of language. What concerns Sthiramati are the two extreme views of perceiving reality incorrectly. [1] This is why he composed this treatise. A Yogācāra Buddhist Theory of Metaphor lacks sufficient textual evidence to demonstrate that the motivation of Yogācāra philosophy of language is to salvage the meaningfulness of Buddhist discourse, which is much more than a defence in the face of Madhyamaka critique. 

2. For Tzohar, one viewpoint differentiating Sthiramati from Vasubandhu is that the former argues explicitly that ‘all language use is metaphorical’ (emphasis added) (Tzohar 204). But Tzohar does not reference a specific text in which Sthiramati express this claim. I cannot find such a statement in TriṃśBh. Sthiramati typically uses the term upacāra, which can be roughly translated in English as ‘metaphorical expression’ or ‘metaphorical designation’, but there is no specific discussion on the relation between the ‘metaphorical’ and ‘language’. I am sceptical as to whether the Sanskrit term ‘upacāra’ has a comparable meaning to the English term ‘metaphor’. We do not find the implication of ‘metaphor’ in Xuanzang’s translation of Sthiramati’s TriṃśBh. On most occasions, Xuanzang translates upacāra as ‘假说’, which would translate into English as something like ‘unsubstantial expression’, ‘nominal designation’ or ‘provisional expression’ [2].

   Similarly, Tzohar emphasises that many Buddhists before Sthiramati rendered language unreferential but did not go so far as to call it figurative (Tzohar 166). But still, I cannot find Sthiramati stating a marked difference between the unreferential and the figurative in the TriṃśBh, or at least I cannot find substantial textual evidence from Tzohar to support his point.

Chapter 6: ‘Conversing with a Buddha: The Yogācāra Conception’

Chapter 6 focuses on the problem of incommensurability: how an enlightened bodhisattva can participate in our conventional realm and communicate with beings within saṃsāra. The three natures teaching of the Yogācāra school is introduced by Tzohar through Alan Sponberg’s ‘pivotal model’. This model teaches that ‘dependent nature, when free of the imagined nature, is seen as the perfect nature’ (Tzohar 181). Tzohar suggests that to make the free-imagined-dependent-nature, the object of knowledge, two kinds of knowledge must be logically and chronologically established. These are the ‘nonconceptual’ or ‘nondiscriminatory knowledge’ (nirvikalpajñāna) and the ‘awareness obtained subsequent to it’ (tatpṛṣṭhalabdhajñāna) (Tzohar 180-88). On the basis of ‘nonconceptual knowledge’, ‘knowledge obtained subsequent to it’ emerges, allowing us to apprehend the dependent as the object of knowledge (Tzohar 182).

In Tzohar’s analysis, consequently, ‘awareness obtained subsequent to [nonconceptual knowledge]’ is supplemented by an understanding in the Mahāyānasaṃgraha about how an enlightened bodhisattva can actively and deliberately interact with beings within saṃsāra. This differs from the model of a ‘mindless Buddha’ who passively accomplishes all kinds of roles without ‘cognitive conceptual content’ (as seen in the Mahāyānasaṃgrahabhāṣya) (Tzohar 185). ‘Awareness obtained subsequent to [nonconceptual knowledge]’ is the extension of said ‘nonconceptual knowledge’ and allows us to perceive the dependent nature without imagination. In other words, it is identical to knowledge of causal relations (Tzohar 187). Aspirations (āśaya) are also discussed here within Yogācāra lore, interpreted as ‘the way in which volitional and intentional content affects and fashion [sic] one’s perceptions of phenomenal appearances and existents’ (Tzohar 185). For an advanced bodhisattva, it is the way in which ‘he is able to manipulate his perception of phenomena at will’ (Tzohar 186). Overall, with the ‘awareness obtained subsequent to [nonconceptual knowledge]’ and aspiration, bodhisattvas are able to perceive causality in the conventional world, and can act effectively without a substantial locus for conceptualization. 

The next problem Tzohar examines is ‘how beings at different levels of advancement on the Buddhist path can converse in a meaningful way despite their deep epistemic discrepancies (Tzohar 171). Tzohar focuses on the notion of ‘intersubjectivity’ or ‘intersubjective agreement’. Sthiramati’s opponent claims that an understanding of intersubjectivity must involve an ‘emergent notion of objectivity’ (Tzohar 191). However, Sthiramati cannot agree that beings claiming to experience the same thing simultaneously attests to the existence of external physical objects. This stance would not explain the experiential discrepancies regarding perceived objects (Tzohar 194).

Instead, the Yogācāra school holds that ‘intersubjectivity is characterized not only by agreement, but also by necessary experiential discrepancies’ and does not depend on the existence of mind-independent objects (Tzohar 194-95). For Yogācāra thinkers, intersubjectivity is ‘constituted by the shared portions of individual karmic streams’ (Tzohar 199). We can have a concurrent perception of the same object because of shared karmic seeds and impressions, but we do not perceive it in exactly the same way because that portion of mental activity (or karmic streams) is not shared. Take the example of human beings versus pretas (hungry ghosts): each group has their own particular pattern of causal mental continuums, and sometimes these causal mental continuums have shared features (bringing intersubjective agreement). Yet at the same time, differences between beings’ karmic streams produce discrepancies, bringing intersubjective disagreement. There can be both agreement and disagreement regarding perceived objects.

Returning to the previous question, this feature of intersubjectivity can explain the possibility of commensurability for ‘the discrepant perspectives of the bodhisattva’s nonconceptual experiences with ordinary ones’ (Tzohar 202). Consequently, ‘the elimination of discursive thought applies only to the bodhisattva’s private experiential realm’, not the intersubjective realm, with the latter maintained through ‘the shared karmic streams and discursive thought of unenlightened others’ (Tzohar 202). Shared karmic streams are in fact the common causal mental process. This provides the foundation for the bodhisattva’s interactions and communication with other beings and allows them to abide in saṃsāra. 

I find the argument in Chapter 6 inspiring. Tzohar conducts an exceptional analysis of the topic of commensurability and intersubjectivity. Nevertheless, in this review, I highlight several points that demand further investigation.

1. The author introduces the ‘pivotal model’ of the three natures and relates this to the two kinds of knowledge. If we accept this model, I wonder how nonconceptual knowledge and subsequent awareness can be chronologically ordered. According to the pivotal model, once freed from the imagined nature, the dependent nature is the perfected nature. This means that upon obtaining the nonconceptual knowledge, one is freed from the imagined nature, then dependent nature will appear as perfected nature. Dependent nature and perfected nature must be attained simultaneously, and they are in fact the same experience for the enlightened mind. There is no time gap between these two kinds of knowledge. I propose a revised model to align with this understanding: 

Figure: Flowchart linking the three natures with the two knowledges (hand-drawn by Shi. Xian’gui).

Image description: There are six text-objects in a complex flowchart. These are: ‘imagined nature’ (written in black), ‘dependent nature’ (black), ‘perfected nature’ (orange), ‘nonconceptual knowledge’ (orange), ‘(complete) dependent nature’ (red) and ‘subsequent attained knowledge’ (red). Two arrows lead from ‘dependent nature’: to ‘imagined nature’ and ‘perfected nature’ simultaneously. An arrow continues from ‘perfected nature’ to ‘(complete) dependent nature’, which has an ellipsis-line to ‘subsequent attained knowledge’. Similarly, ‘perfected nature’ has an ellipsis-line to ‘nonconceptual knowledge’. There is an arrow leading from ‘nonconceptual knowledge’ to ‘subsequent attained knowledge’. Through the colour coding and the relationships between the objects in the flowchart, we can understand that the perfected nature is nonconceptual knowledge, and that the (complete) dependent nature is the awareness obtained subsequent to it. The relationship between the perfected nature and the (complete) dependent nature runs parallel to the relationship between nonconceptual knowledge and the awareness obtained subsequent to it.

In this new model, if one attains nonconceptual knowledge and is freed from the imagined nature, the dependent nature appears as the perfected nature, thus one perceives the perfected nature. However, this does not mean one has a complete understanding of the dependent nature. One must observe the conventional on the foundation of the perfected nature, gradually attaining a thorough understanding of the dependent nature, which is the subsequently attained awareness. 

2. Tzohar draws concepts together without explicating the underlying relations among them. For example: aspiration, subsequent attained knowledge and the shared mental causal process, all of which are critical to the possibility of commensurability. Tzohar does not discuss their relations in detail, making his core argument unclear.

We may ask whether it is shared karma or aspiration enabling the bodhisattva to abide and communicate in the conventional realm. If it is shared karma, then how does one’s karmic stream remain unchanged following enlightenment? If shared karma still has function for the bodhisattva, how can the bodhisattva be free of cause and effect in the conventional world? However, if it is aspiration which enables the bodhisattva to abide in saṃsāra and that aspiration, as Tzohar mentions, and allows a bodhisattva to manipulate their perception of phenomena at will, then why do we need the theory of shared karma for commensurability? Furthermore, does this aspiration mean that a bodhisattva has transcended cause and effect in the conventional realm? Or that shared karma is created by aspiration itself? 

According to Tzohar, shared karma allows for intersubjective agreement and for a bodhisattva to abide in saṃsāra to communicate with other beings. Though I agree that shared karma might offer a common environment for a bodhisattva and sentient beings, I doubt it can help the bodhisattva to communicate in the conventional realm. Since the bodhisattva’s individual experiential realm has been transformed through the elimination of discursive thought, we would expect there to be great epistemic discrepancies between an enlightened bodhisattva and sentient beings, even regarding the same object. There is no common causal mental process allowing for communication between the two. 

These questions might require further consideration in this book. This being said, A Yogācāra Buddhist Theory of Metaphor is an excellent monograph on Yogācāra philosophy of language and should be recommended reading for every Yogācāra scholar today.

Bibliography

Sthiramati, and Buescher, Hartmut. Sthiramati's Triṃśikāvijñaptibhāṣya: Critical Editions of the Sanskrit Text and Its Tibetan Translation. Beiträge Zur Kultur-Und Geistesgeschichte Asiens; Nr. 57. Wien: Verlag Der Österreichischen Akademie Der Wissenachaften, ÖAW, 2007. 

Tzohar, Roy. A Yogācāra Buddhist Theory of Metaphor. Oxford: OUP, 2018.