Filling in the Blanks: The Material Semiotics of Visual Meditation in the Buddhist Qizil Yoga Manual

Abstract

This paper analyses the ‘grammar’, or rules, of meditation practice in the c. 5th-century Qizil Yoga Manual (QYM). The QYM was found in Kucha, a significant location in the Silk Road trade network. Meditation in this text means visual recollection, and the practices are largely structured around Buddhist Sarvāstivāda sequences with some hints of Mahāyāna or Tantric themes). As with certain Mahāyāna ontologies, the Qizil Yoga Manual meditations emphasize an expansive spatiality, employing a densely populated mental landscape in which images abound. Since visualization requires blank space for projection/construction (whether mental space or the space of the cave), one result is the spatialisation of time. Nonetheless, the QYM is not a text in which temporality is dispensed with, and the meditation formulas also retain a more restrained Sarvāstivāda manual emphasis on visual elements arising and disappearing in temporal sequence—such as we find in the Abhidharmakośabhāṣya or early Sarvāstivāda layers of the Yogācārabhūmiśāstra, such as the ‘Śrāvakabhūmi’. To address the interplay of image proliferation and temporal sequence in the text, this paper engages semiotics, or the study of sign systems. Drawing on both linguistic and visual approaches, a ‘semiotic triangle’ is proposed to acknowledge how ‘mental imagery’ is produced triadically by text, cave and body. Since the primary surviving manuscript of the Qizil Yoga Manual is fragmentary, its meaning can be hard to discern, made doubly so by the distinct cultural character of this Central Asian Buddhist Sanskrit text. Moreover, only when the material context of the manuscript is highlighted, taking into account embodied cave meditation, can the ‘readerly blanks’ be filled in. The goals of this paper are to deepen an ontological understanding of the practices in the Qizil Yoga Manual, and to show how historical textual research in religious studies can be underpinned by methods that are bolder in their theoretical scope, moving beyond the remit of translation projects.

1. Introduction

This paper1 analyses the ‘grammar’, or rules, of meditation practice in the c. 5th-century Buddhist Qizil Yoga Manual (QYM), which was found in Kucha, a significant location in the Silk Road trade network.2 Meditation in this text means visual recollection, and the practices are largely structured around Sarvāstivāda sequences3 with some hints of Mahāyāna or Tantric themes (Schlingloff 2006; Seyfort Ruegg 1967; Yamabe 1999a, p. 2, fn 1; 1999b; Bretfeld 2003; Dhammajoti 2021; O’Brien-Kop 2022; Abe 2024).4 Yamabe has identified that the South or Central Asian Buddhists texts translated5 into Chinese around the 5th century CE fell into two groups: visualization sūtras that are aligned with Mahāyāna, and meditation manuals that follow Mainstream meditative practice, the latter including Sarvāstivāda (Yamabe 1999a, p. 2, fn 1).6 According to that framework, the QYM is a meditation manual and, hence, aligned with Mainstream traditions, but it is not without some specific elements of Mahāyāna visualization.

As with certain Mahāyāna ontologies (McMahan 2002, p. 79),7 the Qizil Yoga Manual meditations emphasize an expansive spatiality, a densely populated mental landscape in which images abound. This imagistic style concords with McMahan’s analysis of the Indian Mahāyāna sūtra genre: ‘The literary style of many Mahāyāna sūtras, contrary to that of the earlier Pāli corpus, uses a great deal of visionary material depicting fantastic apparitions and otherworldly scenes’ (McMahan 2002, p. 2). In such Mahāyāna ‘world construction’ the value of language can be downplayed in favour of visual metaphors: ‘Visual and spatial symbolism provided a way for Buddhists to conceive of non-conceptual awareness that is free from the fabrications of the linguistically constituted world’ (McMahan 2002, pp. 5–6). Even a cursory and fragmentary reading of the QYM demonstrates that visuality is at the forefront of the text and, as I have argued elsewhere (O’Brien-Kop 2022), visualization inherently requires blank space for projection/construction (whether mental space or the space of the cave). McMahan argues that in texts like the Aṣṭasāhasrikā Prajñāpāramitā, we see ‘how the Mahāyāna drew upon spatial experience to formulate an understanding of wisdom and awakening’ (McMahan 2002, p. 79). In the words of the Buddha: ‘[t]he immeasurability of all-knowledge is the immeasurability of space’ (Aṣṭa page 150 R303; cited in McMahan 2002, p. 79), and, in an effort to convey transcendence, time itself is ‘de-temporalised’ and made subordinate to space (McMahan 2002, p. 79). This is evident as ‘the spatialisation of time—assimilating temporality to the always present dimension of space’ (McMahan 2002, p. 80). Nonetheless, the QYM is not a text in which temporality is dispensed with, and the meditation formulas also retain a more restrained Sarvāstivāda manual emphasis on visual elements arising and disappearing in familiar temporal sequence without such grand image proliferation—such as we find in the Abhidharmakośabhāṣya or early Sarvāstivāda layers of the Yogācārabhūmiśāstra, such as the ‘Śrāvakabhūmi’ (O’Brien-Kop 2021).

To address the interplay of image proliferation and temporal sequence in the Qizil Yoga Manual, this paper engages semiotics, or the study of sign systems. Drawing on both linguistic and visual approaches, a ‘semiotic triangle’ (Semetsky 2019) is proposed to acknowledge how ‘mental imagery’ (Nanay 2023)8 is produced triadically by text, cave and body. As the primary surviving manuscript of the Qizil Yoga Manual is fragmented (Schlingloff 1964) (and see Figure 1), there are many physical blanks that lead to ‘readerly blanks’, and so meaning can be hard to discern. This is made doubly so by the distinct cultural character of this Central Asian Buddhist Sanskrit text (in comparison to its South Asian counterparts). The method of semiotics applied here permits a capacious and innovative approach that acknowledges the materially situated nature of the QYM. In the period of its historical devising, the QYM encompassed not just words on a manuscript page; rather, it existed in a dispersed agential context that required the textuality of the manuscript and the spatio-visuality of cave imagery and the embodied awareness of the practitioner. All three domains constitute sites of signs—and the advancement of the meditation practice is indicated by interlinked communication between the three domains. Archaeologists such as Swenson have referred to the religious temple as a ‘semiotic machine’, an ‘amplified semiosphere of sacred space’ combined with ‘the distinctive material frame of ritual practice’ (Swenson 2024, p. 15). I will argue that the meditation caves in Qizil fulfil an equivalent function, albeit in a more minimal and constrained way than a temple might. An approach that draws on both linguistic and visual concepts, then, supports the goal of this paper, which is to deepen an ontological understanding of the embodied practices in the Qizil Yoga Manual. A secondary objective is to demonstrate the value of applying literary and material semiotic perspectives to ancient texts to augment a standard ‘history of religions’ emphasis on translation of meditation treatises.9 We can learn much more from such texts beyond how to meditate.

2. The Sarvāstivāda and Mahāyāna Influences and the Role of Caves

The cue for this paper is taken from Yamabe’s analysis of the historic Turfan cave images, regionally close to Qizil on the northern Silk Road in modern-day Xinjiang Province in China. The cave images show both meditating figures and objects of meditation and were likely to be directly related to practices and manuals of meditative visualization, even if the murals are more scattered and disorganized than the texts (Yamabe 2004, p. 405). Indeed, Yamabe ventures that the cave murals in the Turfan region as a whole represent a set of interconnected motifs of visual meditation (Yamabe 1999a; see also Howard and Vignato 2015) which forms part of a combined system between Chinese apocryphal texts and the QYM, a system that shares some ‘visionary images’ (Yamabe 1999a, p. 60). Yamabe speculates that due to the similarity of certain images with Tibetan Buddhist practices, these elements may be derived from ‘the esotericization of Indian Buddhism’ (Yamabe 1999a, p. 60). If such a visual language of meditation could be said to have existed, what were its principles of operation and how did it aid the goal of praxis, which was to achieve liberation? Undoubtedly, there is a symbiotic relationship between such a language of meditation and the meditation practice itself.

More recently, Yamabe and Zhao have examined the regionally related Qumtura Cave 75 murals (10th–11th century CE) and discussed the similarity of a tree image there to the tree discussed at QYM 127 V2-3. Although not corroborated as such (due to the continued deterioration of the cave inscription) (Yamabe and Zhao 2023, p. 35), Teiser (2006) and Mori (2012) had made links between the cosmology of these cave images (also described in the inscriptions) and the cosmology of the Abhidharmakośa, presupposing a Sarvāstivāda background for the QYM, which shares certain resonances with both Qumtura Cave 75 and the Abhidharmakośa. Following reconstruction of the deteriorating inscriptional evidence, Yamabe and Zhao (2023, pp. 36–41) posit a tentative argument that the cave’s image and inscriptions not only hold connections to the QYM, to the Chinese translation of the Abhidharmakośa, and to other meditation manuals such as the Damoduoluo chan jing (*Yogācārabhūmi of Buddhasena, c. 5th century CE),10 but that the cave is also related to the ‘visualization of the five cosmic disks in Esoteric Buddhism’ (Yamabe and Zhao 2023, p. 41). As Deleanu (2006) pointed out, the nascent Yogācāra meditation tradition in South Asia conjoined a basis of Sarvāstivāda with Mahāyāna influences (see also O’Brien-Kop 2021), and since this is a pattern that we also see in the QYM, it can also be argued that early Yogācāra may have been one possible thread in the QYM.

Further progress has been made towards understanding the visual meditation language in a spatio-material context by Vignato, who analysed historical photographs of the rock-cut monasteries at Kucha (dated 4th–9th centuries CE) near to the Qizil Caves (Vignato 2023, p. 903 fn2). Vignato gleaned further details about the structure, features and murals in the Qizil Caves11, including the individual meditation cells (Vignato 2023). Moreover, extensive research by Vignato and Hiyama (2022) and Vignato and Li (2024) has illuminated several new material features of meditation:12 that the meditation ‘districts’ in a monastery complex were situated away from the temple (vihāra) itself, and, most importantly, that some meditation caves took the form of corridors or antechambers designed for ‘mindful pacing’ (Vignato and Li 2024, p. 1003).13 Vignato and Hiyama (2022) divided the visual culture of Kucha into two distinct traditions known as Tradition A and Tradition B. The Kuchean cave complex (where the QYM was found) is assigned to two main periods:

Most of [the caves] belong to the first period, a crucial and creative moment characterized by the construction of caves with unique architectural features and arts with distinctive regional characters. The decorated caves display themes and décor closely related to the early Buddhist art of India and Central Asia, generally affiliated to the Sarvāstivāda school. The second period, more limited in scope, is defined by decorated caves carved for the most part in yet unexploited sections of the sites used in previous centuries. Iconography and style show notable similarity to the Buddhist art inspired by Mahāyāna ideals of the Tang.

(Vignato and Hiyama 2022, p. 2)

Structurally, Group A monasteries are the earlier group and are identified with ‘one or more monastic cells and one or more square caves’ and also include undecorated caves (Vignato and Hiyama 2022, pp. 8, 13). Group B monasteries consist ‘of one or more central pillar caves with or without accompanying monastic cells and square caves’ (Vignato and Hiyama 2022, p. 8). The authors extrapolate from these cave groups to describe two distinct traditions of Buddhism, simply called Tradition A and Tradition B. A concentration of Tradition A caves is found at Qizil14 and some of these caves may have been associated with the QYM. Moreover, through analysis of narrative motifs in the cave illustrations, Vignato and Hiyama conclude that Tradition A is closely tied to the Sarvāstivāda Vinaya tradition (Vignato and Hiyama 2022, p. 10) while Tradition B cave art more closely reflects the Mūlasarvāstivāda Vinaya, with the 5th century as an approximate watershed dating between the two traditions (Vignato and Hiyama 2022, pp. 217–18, 254). How the bare meditation caves of Tradition A were connected to the larger, decorated and more collectively oriented caves at Qizil remains unclear, although Vignato and Hiyama suggest that ‘Tradition A’ meditation may have been ‘mainly a collective activity’ while ‘Tradition B’ had a propensity for more isolated meditation in individual cells (Vignato and Hiyama 2022, p. 246). Perhaps the grand caves were a space for a prelude to individual meditation in the cells or reminders of the general milieu of meditation.15 As Vignato and Hiyama show, not all mural images were designed to be viewed, and the restricted movement afforded in some caves indicate that these images were a background to other activities such as ritual worship of a huge Buddha statue that dominated the space of the cave (Vignato and Hiyama 2022, p. 145).

Depending on how we date the QYM, it is not, then, a radical idea that it might blend Sarvāstivāda and Mahāyāna approaches to meditation. As Vignato and Hiyama (2022, p. 4, fn15) point out, The Memoirs of Eminent Monks (Gaoseng zhuan) describes how Mahāyāna monk and meditation master Dharmamitra was received (‘welcomed and revered’) by the Kuchean king in 424 CE. Moreover, The Memoirs of Eminent Monks of the Tang Dynasty (Xu gaoseng zhang) records that ‘the King of Kucha reacted positively to the Mahāyāna teachings of the Indian monk Dharmagupta […] during his stay in Kucha around 585 CE’ (Vignato and Hiyama 2022, p. 4, fn 15). Hence: ‘The co-existence of the Sarvāstivāda-vinaya and the Mahāyāna doctrine within a Buddhist community is not unusual’ (Vignato and Hiyama 2022, p. 4, fn 15). If the Qizil caves and the Qizil Yoga Manual are dated to the 5th–7th centuries, this situates the tradition in a time of conceptual overlap and transition between Sarvāstivāda and Mahāyāna ideas. Nonetheless, only a few Sanskrit manuscripts with Mahāyāna contents have been found in Qizil (Vignato and Hiyama 2022, p. 4, fn 15), while the majority were Sarvāstivādin.16

3. Sequence and Temporality: Aśubha, from the Abject to the Sublime

The Qizil Yoga Manual is a densely imagistic text, and, at first read, the meditations might seem only loosely structured. Indeed, in some sections, the expansive visualisations appear to eschew any formal grammar of meditation, such as we expect from early Buddhist meditation treatises, especially Sarvāstivādin. The array of images projected onto the āśraya (an image of bodily form, discussed below) is wide-ranging, e.g., streams of people pulsing forth from the hairs on one’s head (YL 129 V3); the mind entering into the mouth of a water creature (makara) who emerges from between one’s big toes (YL 128 R4); a man made of the essence of the āyatanas (sense bases) has four gates on his body and, as they open, skeletons emerge holding lamps and bundles of thread (YL 135 R6–136 V2); rivers like melted butter brimming with jewels and stars of different colours fill the melting āśraya of the yogin (YL 136 V5-6).17 Moreover, the visualisations often feature the endless hyperbolic imagery found in Mahāyāna texts; Collins uses the term ‘infinite extension’ (Collins 2010, p. 25) to describe the way in which the immeasurable scope of the Buddha is portrayed in Mahāyāna texts. Such layering of images can overwhelm our readerly capacity to cognitively process what we are discerning, if we read the text without due instruction. However, the overall scope and sequence of topics also reflect more staid formulas from meditation manuals (such as the Mainstream Satipaṭṭhānasutta), with an emphasis on visual elements arising and disappearing in particular sequences (Abe 2024).18

To understand the centrality of temporal sequences, let us turn to the opening section of the Qizil Yoga Manual, which describes the exercises on the repulsive constituents (aśubhaprayoga).19 The aśubha meditation forms part of a longer set of meditations on the body (kāya), which is the first stage of mindfulness (smṛtyupasthāna). Aśubha20 typically includes meditation on internal organs and parts (internal meditation) as well as meditation on an external body, especially a corpse.21 The purpose of the aśubha meditations is to gradually cultivate detachment from the state of clinging to life and to understand the unavoidable nature of impermanence, non-self and suffering. Meditation on the impure thus leads to states of purity (Dhammajoti 2021, pp. 9–10), and, as Dhammajoti shows, aśubha meditation in early Buddhism was foundational for higher states of meditation practice across different schema (Dhammajoti 2021, p. 17). Kāya meditation typically includes other bodily practices such as mindfulness of breathing, mindfulness of bodily impurities, and contemplation on the elements (e.g., Dhammajoti 2021, p. 2). The corpse meditation describes the sequence of a decomposing corpse and is often called the nine cemetery contemplations or nine charnel-ground meditations in Buddhist texts. As Schlingloff discusses, this section of the QYM’s manuscript is so fragmented that very little has survived of the original text and it must be pieced together through inference. However, there are enough words to see that the formula follows the standard progression of the Satipaṭṭhānasutta and other foundational texts (Schlingloff 2006, pp. 59–60).22 The QYM itself describes the observations of the nine categories of the corpse (Schlingloff 2006, pp. 60–62) as if in the cemetery, (ś)m(a)śānam iva (YL 115V4; Schlingloff 2006, p. 60). In many meditation texts, such as the Visuddhimagga, the imagery of the cemetery meditations is especially horrifying so as to ‘induce disgust’ (Kritzer 2020, p. 2). As Kritzer shows, however, there is at least one text in which a more streamlined and somewhat neutral version appears: chapter seven of the incipient Mahāyāna Saddharmasmṛtyupasthānasūtra (2nd to 4th century CE),23 in which the descriptions seem to ‘appeal more to the intellect’ (Kritzer 2020, p. 2) and are geared towards dispassion rather than disgust.24 Without being able to access more of the original manuscript of the QYM, we cannot assert which version of the cemetery contemplations the QYM elaborated. However, we can formulaically predict the sequence of images and make some comments on that basis.

The formula found in the Abhidharma *Mahāvibhāṣā25 sets the standard in the Sarvāstivāda texts (Dhammajoti 2009, pp. 255, 258).26 One should visualise a corpse and imagine it as one’s own future body:

• A corpse that has turned blue (vinīlaka);
• A swollen corpse (vyādhmātaka);
• A festering corpse with rank odour (vipūyaka);
• Disintegrating (vipaḍumaka; vipaṭumaka);
• Turning reddish/bloody (vilohitaka);
• Eaten by worms (vikhādita);
• Bones strewn around (vikṣiptaka);
• Bones bleached as white as shells (śvetāsthi);
• A dust-pile of gradually disintegrating bones27.

The practice sequence is integrally related to doctrine and, as Dhammajoti states, in Abhidharma and early Buddhism in general ‘the progressive spiritual attainments in Buddhist praxis are characterised by increasingly subtler states of blissfulness and peace—the most profound of which is the bliss of Nirvāṇa’ (Dhammajoti 2021, p. 7).

Up until this point, it seems as though the QYM preserves the Sarvāstivāda privileging of sequence faithfully, meaning that linear temporality is more important here than a more spatially oriented presentation of images. We have only fragments, but we can identify key words that reflect the above sequence:

• (ś)m(a)śānam iva (as if in a cemetery) (YL 115 V4);
• (u)d[dha]tāni ca marmān[y] (bloated organs/soft tissue) (YL 115 R4);
• vipītakam (yellowing) (YL 116 V5);
• paripava)navac chidrīkṛtaṃ (perforated like a sieve) (YL 116 R3-R4);
• cūrṇīkṛtaṃ bhasma (pulverized to ashes) (YL 116 R4).

It is certainly interesting that the Qizil Yoga Manual commences with aśubha meditations and, from the fragments we have, quickly gets into the charnel ground contemplations. This opening is somewhat unusual because in Buddhist mindfulness practice aśubha meditation would typically come at the end of the first stage of mindfulness (contemplation on the body, kāya) where it would follow other bodily practices such as mindfulness of breathing, mindfulness of bodily impurities or contemplation on the elements (e.g., Dhammajoti 2021, p. 2). So, why do we jump into aśubha meditation straight away in the Qizil Yoga Manual? Dhammajoti argues that starting with aśubha reflects an older and less standard form of Sarvāstivāda meditation practice. He notes that generally the ‘Abhidharma tradition promotes both the aśubhā and the ānāpānasati as the two “amṛta-dvāra” (‘gateway to immortality’) leading to Nirvāṇa’ and that ānāpānasati (mindfulness of breathing) was often presented in biographical accounts of the Buddha as the pre-eminent meditative practice in spiritual struggle (Dhammajoti 2021, p. 5). However, this was not always the case, and one passage of the *Abhidharma Mahāvibhāṣā28 asserts that aśubhā is

the initial contemplation. That is: the aśubhā precedes all contemplations. The meditators mostly enter into the Noble Path with the aśubhā as the supporting base, not with ānāpānasati or dhātu-bheda contemplation’.

(*Abhidharma-mahā-vibhāṣā, 205a21–b9, T27, no. 1545; trans. Dhammajoti 2021, p. 7) (underlining added)

This basic observation of how the QYM begins in terms of its sequencing is already revealing. It shows us that QYM is more aligned with the Abhidharma *Mahāvibhāṣā, which starts with meditation on horrors and impurities rather than other options, such as breathing or practice on the elements (which respectively follow the aśubha meditation in the QYM). The way in which the QYM reflects the early layers of Sarvāstivāda Abhidharma, as shown above, concords with Abe’s (2024, pp. 3–5) detailed discussion of ānāpanasati (ānāpānasmṛti) in section two of the Qizil Yoga Manual. Abe identifies the Vibhāṣā as one of the texts that is evidenced in the QYM (Abe 2024) although she argues that the formula for ānāpānasmṛti in the QYM does not follow the ‘highly systematized’ theory found in the Vibhāṣā (Abe 2024, p. 14). It is ultimately difficult to trace the QYM to just one textual tradition because it seems to combine a range of older and younger systems, which produces variations in the sequencing of meditation steps. The fact that the QYM opens with the aśubha sequence shows an early Sarvāstivāda influence, but the vividness of the visualisation technique elsewhere in the QYM reflects Mahāyāna religious developments. Dhammajoti characterises the 5th-century Chinese meditation manuals as being distinct repositories of this kind of fusion:

In the milieu of the 5th century CE, we witness in Chinese translations of the dhyāna-sūtras, proto-Mahāyāna and Mahāyāna texts, an interfusion of Śrāvakayāna doctrinal and meditative concerns with the majestic Mahāyāna buddha-visualization. This was the milieu of increasing popularisation of Mahāyānistic doctrines and practice of buddha-visualization, much of which being centred around the aśubhā praxis.

(Dhammajoti 2021, p. 29)29

Dhammajoti is inclined to date the QYM somewhat later than the 5th century (Dhammajoti 2021, p. 29), the date that Abe argues for (2024, p. 15).30 Given that the Sarvāstivāda elements are still dominant in the QYM (such as the aśubha meditation and other meditation sequences; see note 18), coinciding with Sarvāstivāda dominance in the region until the 7th century CE (Vignato and Li 2024), and given that the QYM has the hallmark exuberant imagery of texts like the Avataṃsakasūtra,31 the 5th to 7th centuries CE continue to be a reasonable timespan for its composition, with a leaning towards the earlier part of this range.32

We have begun with this first example of the aśubha sequence because (a) it is how the Qizil Yoga Manual opens and is, hence, the first step of meditative practice; (b) the example shows clearly how meditation sequences work through logical progression; (c) it sets up a basic function of the images in the QYM beyond aśubhaprayoga, which is to temporally move from images of abject horror to the sublime beauty of existential liberation; and (d) it demonstrates the construction of mental imagery through manuscript text, affective body, and cave environment.33 The sequential and imagistic features of the aśubha opening sequence guides the text long after we have moved on from the specifics of the aśubha section. Many of the subsequent sections of the QYM contain transitions from disturbing images at the heart of a visualisation to more tranquil images at the end of the sequence (jewels, flowers or a beautiful woman), which soothe the practitioner back towards a peaceful state. This sequential structure is repeated throughout the QYM because it contains an affective purpose: for the practitioner to hold steady during images of horror or extravagance in order to arrive at the neutrality of detachment, in the understanding that all lurid images (signalling ‘experience’) are impermanent and ultimately empty. This meditative formula has served as an example of how the ‘temporal’ grammar of meditation works. Let us turn now to consider a more theoretical approach to spatiality.

4. Spatiality: Semiosis as a ‘Global Framework’ of Meaning-Making Through Signs

A semiotic system is a text or a socio-cultural practice, and, as Yelle states, ‘the task of semiotics is to elucidate the techniques of communication that have been employed in different cultures, and attempt, so far as possible, to specify the general principles that underlie such techniques’ (Yelle 2013, p. 17). The QYM is both a text and a socio-cultural practice, since the text and the embodied practice are understood to be ontologically co-arising (whether the text is read line by line or heard aurally, it is immediately generative of mental imagery, which already creates some kind of experience, whether one is meditating or simply reading). In our analysis of the QYM’s distinctly visual language of meditation, then, it is instructive to draw some insights from semiotics, which can encompass both linguistic and visual meaning-making and which can contribute to intercultural philosophy (e.g., O’Brien-Kop 2023).

Since the QYM contains a flow of seemingly exhaustive imagery, a structuralist analysis offers a starting point for discerning the basic principles through which the text logically hangs together.34 Therefore, I will here consider structuralist semiotics. As Yelle states, the method of structuralism starts with ‘a general segmentation and notation of patterns’ in ‘sequences of text’ that are ‘broken down into their phonetic, syntactic and semantic components which are then correlated with events and patterns in their extra-linguistic context’ (Yelle 2013, p. 9). This approach helps to identify the principles of meditation which can be obscured by the complex array of images from visualization techniques. The analysis of aśubha above permits discussion of how, triadically, the text, cave and body invoke and instantiate the material ritual practice of the QYM. As Yelle explains in his analysis of the relation between semiotics and religion, such iterated communicative relations set up a particular ‘semiotic structure’, a ‘certain reflexivity’ that he calls ‘semiotic recognition’ (Yelle 2013, p. 13).35 This recognition of signs is what allows the meditator’s spiritual journey to advance. Semiotics acknowledges the circulation of signs between inner and outer worlds, between imaginative and material iterations of meaning:

Semiotic configurations crystallize only by virtue of habits, but their inner nature is not solid but liquid: chains of signs compose the human universe, and humans themselves are a ring of these chains, struggling to separate themselves from the ocean of signs they are immersed into through the construction of interpretative habits and categorical discourse.

(Leone 2014, p. 95)

Since the QYM is constructed according to such habitual configurations of signs, semiotics offers a valuable lens through which we can sharpen our analysis: triadically the text, embodiment and cave environment constitute a relational and encoded structure of experience that provides the conditions for attaining nirvāṇa.

I acknowledge Yelle’s point that modern semiotics must not eclipse the existence of historical semiotics (Yelle 2013, pp. 4–6). Through translation and close reading, this paper attempts to make sense of the QYM’s own semiotics, but due to the fragmented nature of the text, I also employ modern semiotics, focusing more on Saussure’s and Jakobsen’s work than Peirce’s, making this approach more literary than cognitive.36 Recent scholarship includes some groundbreaking work on Peircean semiotics in relation to Buddhist meditation. Arning (2009) takes a Peircean approach to the phenomenology of modern Vipassana meditation, arguing that ‘the inland empire of bodily sensations and cognitive processes encountered while meditating vindicate the heuristic and explanatory force of the Peircean categories’ (Arning 2009, p. 96). The work of Lettner (2021) also employs Peircean semiotics to make sense of the Buddhist theory of dependent co-arising. She argues that Abhidharma philosophy is ‘equivalent to Peirce’s conception of semiosis’ (Lettner 2021, p. 192), in particular, working towards a practical experience of nirvāṇa as ‘signlessness’ (also understood as emptiness or wishlessness) (Lettner 2021, pp. 192–93). From a phenomenological standpoint, Lettner connects the Buddhist problematic of an ‘ever renewed unfolding of thought’ (Lettner 2021, p. 193) with Peirce’s argument that ‘every thought sign is translated or interpreted in a subsequent one unless it be that all thought comes to an abrupt and final end in death’ (Peirce 1974, p. 284). Divino’s (2025) paper also discusses the Buddhist goal of signlessness from a Peircean perspective, foregrounding the concepts of nāmarūpa and nimitta in Pāli sources (Divino 2025). Using a cognitive semiotic approach, Divino focuses on the relation between internal intentional process, external perceptual markers or forms (dyadically constituting nāmarūpa) and the nimitta as the representation of a given sign, mental image or mental representation (Divino 2025, p. 4). Directly addressing the QYM, Albery (2025) draws on Peircean notions of iconic resemblance to account for the affective state of joy (as one of the four immeasurable meditative attainments). In his argument, the personification of joy as a ‘joyful woman of golden light’, or other images in the QYM, indicate a non-arbitrary and ‘analogical relation’ between ‘sign and signified’ (Albery 2025, p. 35). In contrast, using Saussurean perspectives, I will argue for an arbitrary relation in the specific mental imagery that I analyse further along.

As these research papers demonstrate, the ground for application of Peircean semiotics and Buddhist meditation treatises is ripe for development. Such papers draw variously on Peircean triadic concepts of representamen/object/interpretant, icon/index/symbol or firstness/secondness/thirdness. However, those are not the triadic concerns of this paper. Moreover, while those analyses tend to view a tradition of Buddhism more broadly across multiple texts or corpora, I am focusing only on close reading of one text and moreover I take a more literary approach to semiotics. Divino’s paper is concerned with how cognition works as correct perception, but my study examines the technique of skilled manufacturing of non-real mental imagery—a kind of controlled solipsistic idealism.37 Hence, my paper’s conceptual recourse to the semiotics of religion only touches briefly on Peircean ideas, taking in a broader understanding of semiotics, including visual semiotics. Whereas Arning, Lettner and Divino use Peircean semiotics to focus the phenomenology of experience or cognition, such as ‘semiosis in the internal world’ through ‘self-scanning’ (Arning 2009, p. 114), this present paper argues for the inextricability of such semiosis from a material external world, making the case for the entextualised body in a matrix of mentality and materiality. As Comeau states, ‘New material culture studies provide an approach to religion that centres the material world as the primary condition from which people are motivated to act and a method by which the implied dichotomy of mind and material in South Asian religious studies can be dismantled’ (Comeau 2020, p. 4). By taking an approach of material semiotics,38 analysing the mind–body as just one sign in a larger material system of communication, this present study is more ontological than epistemological. Indeed, as a semiotic approach, this paper is more aligned with Yelle’s work, since he stresses the materiality of semiosis (in his account of ritual).39

The practitioner’s body in the QYM is a site of meaning-making that is agentially shaped by text and cave as much as it is exerting its own subjective agency; the meditator is part of a triadic material system of text, cave, body, all connected through visual semiosis. As an analytic method, visual semiotics can help address the supra-textual (or para-textual)40 dimensions of the QYM from a structuralist perspective.

Supra-textual design encompasses the global visual language of a document and operates in three modes: textual, spatial, and graphic. The rhetoric of supra-textual design includes structural functions that provide global organization and cohesion and stylistic functions that affect credibility, tone, emphasis, interest and usability.

(Kostelnick 1996, p. 9)

Kostelnick further qualifies what this means for reading a text: ‘Attention to the visual rhetoric of the document, then, must extend to the whole document—its global framework—not only its internal workings’ (Kostelnick 1996, p. 10).41 My argument is that we should not only be concerned with the conceptual and textual design of the QYM as a meditation manual and a manuscript but also its ‘global framework’, the extended system of signification through the monastic body and via the decorated cave. Only in this expanded and interdependent supra-textual framework does meaning-making about meditation occur. Ontologically, one function of the ‘global framework’ of the system is to re-envisage the subject by dissolving boundaries between self and environment, emphasising interdependence and non-essentialism. I therefore reject any notions of ‘parallelism’ between the cave imagery and mental imagery; rather, they are dynamically interwoven and co-existent via the body.

5. The Two Axes of Meaning-Making and Representation

Before we move into combined linguistic–visual analysis, let us evaluate one useful lens from linguistic semiotics, which helps to strengthen the dual examination of temporality and spatiality in the QYM’s practice. In Jakobsen’s linguistic semiotic analysis (Jakobsen 1956; Jakobson 1985, pp. 28–33), following Saussure, there are two axes on which language functions, or, indeed, any system of meaning-making. The first axis is a horizontal (or linear) plane of combination of components in sequence (according to rules). This is called the syntagmatic axis because it shows how a unit of meaning, a syntagm, is constructed in time, i.e., we speak one word after another in a grammatical sequence. The second is a vertical axis of substitution in which words or other components can be replaced by differing options (according to rules) and which also have a representative function. This is a paradigmatic axis because it shows how we make informed or expert choices on which words to use in an utterance (why we sonically choose ‘bat’, not ‘bad’; or why we semantically choose ‘lamp’ not ‘light’). The syntagmatic and paradigmatic can be understood as, respectively, causal and contiguous modes of meaning-making.42

In linguistic semiotic terms, I argue that the imagistic meanings in QYM correlate to Jakobsen’s paradigmatic (vertical) axis, with the ocular projections onto an āśraya43 (an empty image of bodily self) flashing in constant change as one vivid image supersedes another. The āśraya of the yogin acts as a ‘blank’ or empty space, like a canvas or screen, which can be filled with images, powered by the projection of mental technique. However, to employ Jakobsen’s other axis, the structure of the QYM itself is also deeply syntagmatic (horizontal), holding all of these images in a temporal sequence in which it matters ‘what comes next’ in the grammar of meditation practice. Next, I will focus on the relationship between syntactic modes of image visualization and temporal experience—arguing that there are both radical departures in the Qizil Yoga Manual from South Asian textual traditions as well as important Buddhist-Yogic elaborations on how to gain ontological freedom. One tentative conclusion is that Sarvāstivāda provides the formal grammar or syntax, the strict sequences of meditation, while the Mahāyāna influence is more evident in the rich, florid and exciting vocabulary of images.

6. The Meditation Syntagm

Let us think further, and more technically, about the role of sequence in this meditation manual. Syntax is a sentence-like structure that is held together by rules called grammar. A unit of syntax is a syntagm, and so we can think of a unit of sequenced meditation as a ‘meditation syntagm’. The nine stages of the corpse meditation, for example, constitute a complete syntactic utterance. It is only meaningful when the components are strung together according to the rules that make a whole—if the meditator stops part way through the stages, or swaps them around, the meaning (and effect) becomes altered. The corpse meditation is a quintessential example of strict meditation syntax. The sequence of the syntactic constituents has to be preserved for overall coherence: we cannot visualise the body as dust before the corpse as turning blue. The decomposition sequence is fixed because it is naturalistically linked to time–space referents in the material world, describing the inevitable decomposition of the corpse of the future self. As an embodied praxis, the meditation also produces an ontological effect on the material mind–body of the living practitioner44—that of disgust, and a range of other affects, before reaching detachment. Due to an ontologically real dimension to this visual syntax,45 the reader/practitioner cannot substitute images or move them around in the sequence because (a) the sequence is integral to the meaning and (b) logical constraints follow the naturalist quality of this meditation syntagm, i.e., external referentiality. The syntagm enacts a kind of mathematical reduction over the nine stages, in which one or more features are stripped away at each stage to produce ontological degradation. It is like counting down to nothing, until only dust remains. Yet, the integrity of the meditation sequence is also ontologically productive—the effect of ontological degradation in the practitioner is the affirmation of the de-essentialized self.

As discussed above, this meditation syntagm follows the standard example found in Mainstream and Sarvāstivāda texts, and the QYM acknowledges this through its refrain eṣa eva kramaḥ (QYM 146 R3), (‘That alone is the sequence’)46 or compounds like kramabhedāt (‘according to the divisions of the sequence [of dependent co-arising]’) (136 R3). The text also often mentions these terms: (a) kramaśa (gradually, in due order, step by step, or by degrees), e.g., entering the sequence of visualising blue, yellow, red and white objects in order (131 V1); visualising sequentially an enemy of bare, medium and strong status during a meditation on compassion (147 R6); the gradual arising of equanimity in stages (158 R4)47; and (b) krameṇa to indicate visual emanations in stages from the yogin’s āśraya (133 R4).48 Many of these instances are familiar and formulaic Buddhist meditation sequences. However, some are more uniquely linked to the order of visual images in the QYM, such as this passage in the section on maitrī meditation at 144 V6–144 R2:

He sees the earth made from beryl and furnished with diamond thrones made from beryl. In the midst arises the āśraya of the yogin made from diamond and sitting on a diamond throne. Over him pour all beings [bearing all weapons], but they fall [and] turn into flowers below, because that diamond throne is inaccessible to them.

On the other diamond thrones appear Buddhas, who emanate in sequence, according to the previous [emergent] visibility of their uṣṇīṣas (protuberances). Through the rays (of their protuberances) the whole world is illuminated in sapphire blue, through the rays of the ūrṇā (circle of hair between the eyebrows) [emanates] moon-white [light], through the rays of limbs [emanates] gold-shining [light] and through the rays of the eight finger-nails [emanates] ruby red light. The Exalted Ones encompass the āśraya of the yogin with … on their heads.

(YL 144 V6–144 R2: trs author; Schlingloff 2006, p. 125)49

Due sequence is also indicated through other terms such as the standing phrase aṃte sarvam nirudhyate (130R6) or aṃte ca sarvaṃ jṇeyaṃ nābhyāṃ nirudhyate iti vyutthānam (e.g., 139 R3; 147 V4; 152 R3; 155 V2; 165 R5; 166 R5; 167 R3)50, which often appears at the end of a meditation practice to signify that all images are finally dissolved. These meditation syntagms show us that the basis of the Qizil Yoga Manual is to employ a standard meditation formula (especially those found in Mainstream accounts) and to follow its grammar quite strictly. However, we have also examined instances where the standard sequencing has been adjusted or in which the components of the sequence are embellished. Let us move on to the next syntagm to be analysed.

In the manual’s overall structure, the second section is ānāpānasmṛtibhāvanā (cultivation of mindfulness of breathing). In this case, we will see that imagistic embellishment occurs to a great degree. Again, the vehicle of the meditation is familiar as one component of the four foundations of mindfulness. As Abe (2024) has asserted, the meditation components are similar to the ‘classic’ account found in the Satipaṭṭhānasutta, but there are some highly graphic variations to the stages by which the yogin proceeds. In the Satipaṭṭhānasutta, the instructions centre on simple statements of awareness, e.g., ‘I am breathing in a long breath’ or ‘I am breathing in a short breath’. However, in the QYM the details are more exuberant. Peppered amid the directions to inhale and exhale are visualisations of the following: the breath being like a great coiling stream that encompasses the earth (118 V2-3); the yogin’s body filled with winds through its apertures and then sunken in a pond of oil, and made stiff or frozen in space, ākāśa (118 V5);51 blue, yellow, red and white light rays emanating from the heart (118 R2); a young boy (bāladāraka) as a symbol of the mind (cittanimitam) (118 R3); a man like a glowing coal or falling from a star (aṅgārābhatāraka) (118R4-5); a burning pond along with a burning in and out breath; and then one’s own āśraya filled with foam bubbles (phenapiṇḍākulaṃ). The interspersion of such images, some bizarre, may be considered to disrupt the basic or original meditation syntax that we find in other texts, such as the Satipaṭṭhānasutta. And the suffusive imagery in the QYM continues: as the yogin passes through visualisations of diamond thrones, beryl columns, sapphire blue lights, parasols made from dripping gems, trees of jewels emerging from the yogin’s head, silk scarves, and rains of flowers, jewels and lotuses from the mouths of countless buddhas (120R2-121V5), he knows he is on the path.

Some scholars have speculated that this distinct vividness is a result of South Asian Buddhist meditation being mixed with local systems in Kucha such as shamanic use of herbs (e.g., Bretfeld 2003).52 This is an important line of investigation which deserves further research. However, my main argument here is that the hyper-real saturation of images that characterises the QYM only seems to be disorderly or spontaneous to the untrained eye. As we read on, we are equipped to decode the symbolic significance of these constituent images in the meditation syntagm. It is not a collage of random moonbeams and sun rays, boys and foam balls, but, rather, is a precise vocabulary of meditation, somewhat unique to this text, that is repeated regularly throughout the manual and combined according to predictable rules. As the yogin moves through the breathing sequences, he progresses in the dhyānas or levels of meditative absorption:

As the yogin breathes in, a silk scarf radiating blue light (appears); on this sits the above mentioned symbols of mindfulness (smṛtinimittāni): the young boy fades into the gold disc with the beryl roots, etc., and as the yogin breathes out, a white radiating silk scarf (appears), on which sits the symbol of mindfulness: the young boy comes forth with all groups and creates a silk canopy53 over all buddhas. This is repeated several times.

Then the yogin breathes in, and an ascendant blue-lit string of pearls (appears).54 These also fade out.

The yogin breathes out and an ascendant white-lit string of pearls (appears); with all groups he makes a canopy of pearl strings over the buddhas.

(127 V4–V6; trs. author; Schlingloff 2006, p. 80)55

Even with the dense imagery, each breathing sequence is carefully punctuated by a clarification of the purpose: for the joy of the mind (cittābhipramodanāyām) (127 R1); for the concentration of the mind (cittasamādānāyām) (127 R3); for liberation (vimuktipratisaṃvedanāyām) (127 R6); for sight of impermanence (anityānupaśyanāyām) (128 V1); for abandoning (prahānānupaśyanāyām) (128 V1), for detachment (virāgānupaśyanāyām) (128 V2), for cessation (nirodhānupaśyanāyām) (128 V2).56 What is most interesting here, however, is the explicit use of ‘symbol’ (nimittam which could also be understood as a mark, sign or token) in cases such as the image of the young boy in order to inform the reader/listener that this stands in for ‘the mind’ (citta). If we adhere to the standard distinction between sensory and propositional imagination as ‘imagining perceiving x’ versus ‘imagining that x is F’ (Nanay 2023, p. 165), then we might consider the image of the boy to be part of both sensory and propositional imagination.

7. Visual Punctuation of the Meditation Syntagm

What is common to many, if not most, of the visualisations in the QYM is that at the end of the sequence of projected images, there is reabsorption into the navel. The navel point is like an end punctuation, or a full stop—repeatedly, whichever images are generated are absorbed back into the navel of the āśraya and then dissolved.57 There is a blurring, then, between the syntax of the text and the mind–body as the site and conveyor of that syntax.

Most meditations follow the same syntactical sequence of arising and dissolving in the āśraya and, especially in the dhātupragyoga section, the world realms are penetrated by different currents or streams of energy (pravāha) which travel out from the body and then return for dissolution.58 This pattern can be said to form a grammatical rule for visualisation in the QYM that governs the formation of a meditation syntagm. But the syntagm is not without variation in form: different substances are generated in the yogin’s āśraya, exit through the brows, penetrate the whole world, shoot up to the Aghaniṣṭha heaven and return back to navel—with substance currents sometimes splitting this way or that.59

Here is a parsing of a dominant syntagm that recurs throughout the QYM. Each phrase [in brackets] is like a ‘word’ in a macro-syntagm:

[elemental current exits from brow] + [as far as the air circle] + [turns back at ether] + [in through navel] + [out through head] + [to the Aghaniṣṭha gods] + [back in through head] + [out through brow and navel]60

At any point, the components in italics could be substituted (e.g., air could be substituted with water or fire; or the current may exit only through the navel and not through the brow and navel), but there are basic rules to the sequence that remain unvaried. This example suggests a striking connection between punctuation and embodiment in which the syntax of meditation itself highlights the central role of āśraya. The navel acts as embodied punctuation, it is where each visual utterance begins and ends, and it acts as a visual marker that helps us identify the start and close of each syntagm (which is not always easy, given the stream of mental imagery). The punctuation is not just about designating the end of the practice sequence, but the designation itself, dissolution, also signals spiritual awakening, since the ontological dissolution of the self as āśraya is the active acceptance of emptiness (the emptiness of form). Hence, there is an inextricability between ontology and sign, be is a linguistic or visual signifier.

In practice, what this syntax builds up is technique: there is a rhythm to this meditation syntax that is predictable, always concluding with images reabsorbed into the navel, followed by the dissolution of the āśraya. Just when the images seem too wild to tolerate, they are restored to the navel and released. How long were these images to be held in reality? For a few moments? For a long period? Perhaps they were flashing sequences. It is difficult to determine, and we return to the point that there is an unclear relationship between the cave images, the text’s contents, the material manuscript and the embodied practice. But what the visual punctuation does offer is clear cognitive markers and rests in a vividly visualised landscape. This visual punctuation therefore offers a reassuring opportunity to regroup and reorient before the next sequence (in the darkness of the cave).

8. Image Substitution and the Meditation Paradigm

As we move through these strings of meditation syntagms, we are also passing through the dhyānas or stages of absorption. Certain images serve to signal that one has progressed to a particular stage, like a signpost in the meditation cartography.61 Sometimes the key to the meaning of the images is provided for us.

Then he sees his own form like a basket (mūta) filled with different sorts of rubbish. Then … he sees … The element of ether is represented by eye sockets, etc., the element of consciousness (vijñāna) by a monkey (markaṭa), a young boy (bāladāraka), a dappled deer (binducit), a torch (ulkā) and so forth (YL 128 R5).

(trs. author; Schlingloff 2006, p. 86)62

This key, which is given to us, helps to understand that in the preceding passage the young boy represents consciousness (vijñāna)63 as does the image of torches in the sea. Without the author’s semiotic key, it is unlikely that we, as modern readers, would be able to distil such clear meanings from the suffusion of imagery. In contrast to the necessity of understanding temporal sequence in the rules of the syntax (i.e., what comes next in the meditation syntagm), the priority here is to grasp the paradigmatic substitution of the image (the young boy) for something intangible like consciousness. In fact, the young boy is a recurrent symbol of consciousness throughout the manual, but the basis of the selection of this image to represent consciousness is not easy to understand. Modern semiotics asserts that the relation of the sign to its meaning is arbitrary,64 and even if we were to try to retrieve the ‘semiotic ideology’ of 5th-century Sanskrit in which signs were posited as non-arbitrary and ‘natural’ (Yelle 2013, p. 5), any such natural connection (of boy to consciousness) is more or less lost on the contemporary reader.

Another standard example of a visual semiotic key is provided in the ‘observation of the skandhas’ in the third section of the QYM. For each of the five skandhas, or attributes of anātman, we are provided with the key of image-substitution:

form (rūpa) is represented by: foam (pheṇapiṇḍa) (132 V3)

feeling (vedanā) is represented by: bubble (budbuda) (132 V4)

wisdom (samjñā) is represented by: ray of light or mirage (marīci) (132 V5)

psychological habit (saṃskāra) is represented by: plantain leaf (kadalīpatra) (132 V5)

consciousness (vijñāna) is represented by: optical illusion (māyā) (132 V6)65

Such an explicitly descriptive key embeds the strategy to decode images of foam balls and water bubbles as representing specific skandhas whenever they appear elsewhere in the text without explication. These hermeneutic strategies are, of course, replete in the genre of Buddhist meditation manuals. A third example is found in the list of unique attributes of the Buddha (āveṇikadharmas) in the visual recollection of the Buddha. This standard image key is listed at both 162 V1–V4 on upekṣā and at 165 V4-V6 on buddhānusmṛti.

ten powers (bala) are represented by: Buddhas seated on elephants

four intelligences (vaiśāradya): four preaching buddhas on lion thrones

three kinds of foundations of mindfulness (aveṇikāni smṛtyupasthānāni)66: three

arms-bearing men with oil bowls

great compassion (mahākaruṇā): a woman of etheric hue

As Lamotte points out, this list of the attributes of the Buddha conforms to a standard Sarvāstivādin method (as opposed to the variant Sarvāstivādin list or to a quite different Mahāyāna list) (Lamotte 1970, pp. 1331–33). In all of the above instances, we see how image-substitution keys are provided as standard in meditation manuals and canonical sources, in order to prompt accurate meaning-making.

9. Filling in the Readerly Blanks

The concept of the ‘blank’ is significant in both the design and the reception of the QYM. The meditation method itself foregrounds the ‘blank’ form of the āśraya (onto which images are projected) in order to cognize the emptiness of self. The cave offers blank space in which visual meditation and material representation thereof (murals) can occur. Moreover, the physical blanks in the fragmentary, disintegrated manuscript leads to readerly blanks in interpreting the text. For us as contemporary scholars, then, the text’s own semantic keys become crucial in interpretation. As readers, we are constantly forced to fill in the literal gaps in the critical edition on a predictive basis using the syntagmatic and paradigmatic axes: what might logically or typically fill a paradigmatic gap which would also fit with the syntagm? In this section I discuss the readerly endeavour of making sense of the QYM in the light of the incomplete syntax, by partially using the lens of reader reception theory from literary studies. Reader reception theory (e.g., Fish 1982, Barthes 1995) presupposes that texts have real yet evolving effects on readers over historical time, which is separate from considerations of what the writer intended or what the objective truth of meaning in a text is.

In his impressive reconstruction of the manuscript (see Figure 1), Schlingloff frequently borrowed or recreated likely probabilities based on what he found in other parts of the manuscript (or, indeed, in other relevant texts). For these reasons, it is worth observing how the two axes (syntagmatic and paradigmatic) operate in different reception frames of the text. The axes were important for the original reader-meditators in the c. 5th century CE for memorizing and recognizing the stages of practice. And the axes are arguably even more important today to contemporary scholars because we are left guessing as to the many missing (and, hence, forgotten) words, sentences and passages. This intellectual labour also applies to any text that appears on the cave walls. In a fascinating textual reconstruction of faded inscriptions on the rear wall of Qumtura Cave 75 (near Kucha, 10th–11th century), depicting a meditating monk, Yamabe and Zhao engage digital image restoration (based on high-resolution visible light photography) to identify fragmentary characters (Yamabe and Zhao 2023). What is particularly illuminating about this study is the way the reconstructive techniques are applied equally to the visual imagery and the visual linguistic characters depicted, treating them both as a seamless single visual–linguistic language. The educated ‘guesswork’ is a constant shuttling between what the semiotician Saussure called parole (concrete instances of language) and langue (the abstract system of language)—trying to read the images before one’s eyes in Cave 75 with reference to larger textual and visual corpora in the region. Comparing other paintings and texts, Yamabe and Zhao use the axis of substitution to examine potential viable options: ‘the whole depiction might allude to asuras attacking Indra, even though we cannot see asuras in the painting; alternatively, this might be another depiction of devas’ (Yamabe and Zhao 2023, p. 32). Fortunately, we can use the formal or underlying rules of the Qizil Yoga Manual’s composition as a prop by which to make informed suggestions. Here is one such simple example of filling in the blanks:

Then from the body of the blessed one [is emitted] [a current] … descends from the sky [a rain] of precious stones and flowers.

(154 R2; trs. author; Schlingloff 2006, p. 154)67

Here, Schlingloff fills in the blanks on a predictive basis (shown by words in square brackets), based on other standing phrases in the text. Indeed, contextual knowledge that these lines appear in the sequence of the abhiṣeka,68 the rain of blessings, enables any reader with an awareness of Buddhist texts to reason that a missing word would be ‘rain’, and that it carries meanings of ‘purification, transformation, or vivification’ (Davidson 2023, p. 30). However, the knowledge that the body emits elemental currents in a certain order is more distinct to the QYM, and, hence, filling in this blank in the tattered manuscript is challenging unless one can recognise the paradigmatic and syntagmatic rules in operation for the QYM’s meditation sequences. Such is the instructive quality of this text as a manual, that we as readers are employed to be especially active. Alongside the historical meditator, we, too, have to visualise the images as they are each described, and to follow or work out the visual logic.69

10. The Visual Semiotics of Meditation

To further deepen our analysis, we may now return to visual semiotics. The very contents of the QYM present a cartography of the spiritual journey—a system of visual signification, a map to nirvāṇa that is filled with information, icons, symbols and a host of ‘visual variables, the basic building blocks of a graphic representation’ (MacEachren et al. 2012, p. 2496). In visual semiotics, icons are a specific type of symbol; they visually represent objects, concepts or ideas, and they are designed to be recognisable. This understanding is shared in the founding works of linguistic semiotics by both Saussure and Peirce, in that an icon physically resembles the thing it represents70 (it is of interest that one of Peirce’s early careers was as a mapmaker (Arning 2009, p. 104)). Hence, the young boy of the QYM is not an icon because iconic symbols are closely tied to the referent, and a young boy’s form is nothing like consciousness physically (if, indeed, any form could be said to resemble consciousness). However, we may say that the young boy is an abstract symbol of consciousness because abstract symbols are much less closely tied to any physical referent. From Peircean semiotics, another helpful concept is the ‘index’, which lies in between an icon and a symbol, in that there is a causal relation of some kind, but it is not physically definitive.71

However, if paradigmatic substitution rests on a relation of similarity and difference, then the question arises as to whether the ‘boy’ image can be considered part of a paradigm because it is arbitrary as a symbol. If the range of substitution in a paradigm is open-ended, then a symbol might bear an analogous relation to the thing it substitutes (to be phenomenally similar), but if the range is fixed (like musical notation for a particular instrument) then it is rather a code and not a paradigm of substitution. In any coded system of symbols, the motivation (Barthes 1993), or desire to represent faithfully, is not high—indeed, there is a dialectic between motivation and arbitrariness (Leone 2014, p. 96). Without such embedded ‘motivation’, the reader needs to be instructed in how to decode abstracted symbols—much like the semiotic keys that the composer of the Qizil Manual provides for the reader (discussed above).

Yet, even accepting the argument for deep structural coherence in the QYM, to what extent can the meditation language of the Kucha region be described as ‘a formal language’ in linguistic–visual terms? It is clear that the monastic meditation practitioners in and around Kucha constituted a ‘linguistic community’, a group of people who use the same system of linguistic signs (Bloomfield 1958) because they are engaging with what Yamabe terms a ‘textual circle’ (Yamabe 1999a, p. 2).72 Although I am making the argument for a regular syntax of meditation imagery, we nonetheless have to accept that this is only a language in a loose cultural sense. There is a degree to which the meanings and purpose of the Qizil Yoga Manual, the meditation practice of historical subjects, and the decorated cave walls are concordant—however, the degree of agreement is unclear. Some scholars are confident in making definitive statements about the relation between textual and cave imagery; for example, Yamabe asserts that the image of a ‘young boy’ (bāladāraka) is so common in the QYM that regarding ‘Figures 5 and 6’ in Grünwedel’s 1912 sketches of Baixihar Cave 3 in the Turfan region, ‘it should be possible to interpret these paintings as the kind of mystical visions that are described in the Yogalehrbuch’ (Yamabe 2010, pp. 261–62). However, even if such dyadic links can be argued for, the triadic factor of embodied meditation itself is historically irretrievable. The material and spatial dimensions of the visualisation practice are difficult to reconstruct with so many missing or degraded features of the caves.

The Qizil Yoga Manual itself, with its linear narrative form, is the best record of any agreement within this linguistic community about the role and meaning of meditation imagery. Yet, the connections between different cave images in Kucha and the wider region (Yamabe 2004; Vignato and Li 2024) also suggest that the meditation images form part of a larger local system that we can think of as Saussure’s langue. Nonetheless, by themselves, the text and cave images constitute a dormant language: the linguistic and visual messages are only enlivened with meaning when cognized and expressed in the body–mind of the meditating practitioner. Moreover, there are relative differences between these three parts of langue: text, body, art. The text is a long and precisely detailed linguistic message with a grammar designed to support the retrieval of meaning. In contrast, the cave images are free-floating signifiers, condensed messages that are much more open to interpretation. The functions of the linguistic and material imagery are therefore distinct: the QYM is instructive, but the cave images are reminders, visual signposts to prompt memory (or narrative). If taken as a standalone visual language, the cave images cannot be easily navigated to sequence meditation experience. As Semetsky argues, ‘pictures are remembered better than words, while an attempt to describe life events often defies words’ (Semetsky 2019, p. 3). Although the visual language of Kucha regional caves serves to preserve or reproduce information for/about/from meditation acts, it is unproven that those murals have a generative capacity in their own right to induce new meditative experiences.

As with any manual, the format of the Qizil Yoga Manual proposes a standardised design. In this case, it is the experience of meditation that is idealised, and the text highlights this with the clear visuo-spatial placement of symbols (markers) on the meditation cartography (both in the mental imagery and on the material walls of the cave). Vignato and Hiyama (2022) argue that the cave paintings are tightly bound by narrative sequence: ‘The main characteristic of wall paintings is that their position is fixed once and for all, and that each section of the narrative needs to be read in context’ (Vignato and Hiyama 2022, p. 169). The cave layout ‘leads the eye of the devotee to the intended focus and helps to correctly link the different elements in the hallowed space’ (Vignato and Hiyama 2022, p. 169). Yet, we also know that human experience (mental or visceral, psychological or religious) is variable, unpredictable, and unique—and this is especially the case for what cognitive scientists and philosophers refer to as interpersonal variation of mental imagery (Nanay 2023, p. 15). Therefore, any successful standardisation of Buddhist meditative visualisation was as unlikely in the past as it would be today for any diverse group of meditating practitioners. Which came first: spontaneous visualisation, the imagistic experience of the words in the text, or the cave images for the experience? Nanay’s definition of mental imagery is helpful here: ‘perceptual representation that is not directly triggered by sensory input’ (Nanay 2023, p. 4). We might consider the meditation formulas in the QYM as first requiring sensory stimulation to activate, such as the words of the manual (read/heard) or the cave murals (seen), but after some practice the mental imagery does not require any sensory input and becomes a matter of memory. Again, the darkened and secluded environment of the cave is a material factor in such embodied yet non-sensory production of mental imagery.

11. Tri-Relative Logic: Filling in the Bodily Blanks

At a certain point, the framework of linguistic–visual semiotics reaches its limit. In our 21st-century attempts to reconstruct meaning from the northern Silk Road in the 5th–7th century, what is missing or ‘blanked’ in our purview is the living bodies of those early practitioners or any extant material artefacts and accessories—their letters, journals, ritual objects, clothes, sandals, implements, oil lamps, robes, etc.—that could provide a clue to the individuals’ identities, practices and experiences.73 Hence, this local meditation language is what Kelso (Kelso 2000, 2008) terms a tri-relative process-structure of ‘signs as dynamic patterns … enriched with meaning’, (Semetsky 2019, p. 3). We may miss the ‘dynamic’ aspects of this meditation method, but we can look for the ‘semiotic bridges between the human mind, cultural artefacts, and events in real life’, a ‘semiotic triangle’ (Semetsky 2019, p. 3) held together by ‘a tri-relative logic’ (Semetsky 2019, p. 4). Interestingly, in terms of the materiality of the body, Greene points to a relevant passage in the Meditation Scripture of Dharmatrāta (Damoduoluo chan jing), 5th century CE. In relation to ethical restraints (saṃvāras): ‘The practitioner must contemplate the signs of these restraints in three places, within his body (āśraya?), within his meditative object (ālambana?) and within his mind (citta?)’74 (T.618:15.322a20–b12; trs. Greene 2016, p. 154). Such practice, within a ‘Sarvāstivāda or Sarvāstivāda-inspired framework’ (Greene 2016, p. 156), draws attention to the triadic relation of body, external image and mental image. A recent groundbreaking contribution has been made to building such triadic semiotic links in the work of Vignato and Li (2024) in reconstructing the daily routines of meditators in the caves. Yet, much work remains to be done to fill in the material blanks in our readerly understanding.

12. Conclusions

My broad aims in this paper have been to further highlight the Qizil Yoga Manual on the academic map for Buddhist and Yogic Studies (building on my 2022 paper) and to show how historical textual research in religious studies can expand to close readings underpinned by methods that are bolder in their theoretical scope, moving beyond the remit of translation projects. Moreover, the hope is that reading historical religious texts though material semiotics might yield analytic strategies that could be extrapolated to other types of literary texts. This paper has sought to advance my argument for meditation as a material practice (O’Brien-Kop 2021, 2023) and here I would like to connect the term meditation to mediation. Appadurai notes of mediation: ‘Mediation and materiality cannot be usefully defined except in relationship to each other. Mediation as an operation or embodied practice, produces materiality as the effects of its operation’ (Appadurai 2015, p. 225). He defines mediation as a practice that operates through specific technologies such as speech or ritual—and we can add meditation here as an embodied form of mediation, one that clearly demonstrates the reconciliation of the material and the semiotic.

More specifically, this paper has sought to identify a logical frame within a Buddhist-Yogic meditation manual that at first reading can appear to be describing a disorderly or chaotic array of images for meditation. The first insight of this paper was to point out that the visual vocabulary of the QYM is limited and rule-bound. Although it might seem as if ‘anything goes’ in this text, the elements of the meditation syntax are restricted according to two ‘axes’ of linguistic meaning-making and representation: the syntagmatic and the paradigmatic. On the syntagmatic axis, the sequences in which images appear are tightly rule-bound with predictable orders of signs and experiential events in meditation. On the paradigmatic axis, certain images reappear frequently (jewels, parasols, milk, etc.) or infrequently (knife-wielding strangers, sea monsters, shooting stars), constituting a wide-ranging but nonetheless constrained image vocabulary. A conclusion is put forward that the regularity of the temporal sequences is informed by Mainstream, especially Sarvāstivāda, meditation texts, while the unboundedness of images and image substitution75 draws on Mahāyāna (and perhaps also local) influences—representing two semiotic axes of the syntagmatic (temporality) and the paradigmatic (spatiality). My aim has been to make the Mahāyāna aspects of the imagery more explicit by focusing on the imagery’s ‘spatial’ qualities, a feature of Mahāyāna visualisation. To do this, I have tentatively engaged a methodological mapping of the Buddhist meditation sequence–image relation to the linguistic axes of syntax-paradigm and the ontological domains of temporality-spatiality. This yields a visual semiosphere of sequence-syntax-temporality (primarily Sarvāstivāda) and image-paradigm-spatiality (primarily Mahāyāna) entailing the tri-logical material system of text-cave-body that is necessary to interpret the Qizil Yoga Manual.

The second insight of this paper has been to analyse the purpose of the visual syntax and its distinctively ‘vivid’ quality. The function of these busy meditations is to excite the mind, to disturb it with horrifying and beautiful images. The detachment that arises when the mental imagery fades produces an embodied understanding of impermanence, that nothing lasts, even the most compellingly real or overwhelming image. The Qizil Yoga Manual provides a strong disciplinary training in which the goal is to be equanimous regarding whatever spectacle lies before one’s eyes (or before the eyes of the mind, i.e., mental imagery). The role of images, then, is not to induce affect in order to produce transformation, but, rather, to watch the spectacle of affect in the āśraya without engaging with it. Such hyper-real imagery holds what Semetsky calls an ‘imaginal’ quality (Semetsky 2019, p. 12)—not to be confused with imagination or fantasy but rather indicating a cognitive mode associated with the unconscious. ‘The imaginal world of archetypes constitutes a distinct order or reality corresponding to a distinct mode of perception, in contrast to the purely imaginary as simply unreal’ (Semetsky 2019, p. 12).76 Hence, the linguistic and visual language of Kucha is underpinned by its own ‘imaginal’ logic. The images can convey more than just their semantic range; the visual vocabulary and refrains also punctuate the complex sequences and return the practitioner to familiar territory in order to (a) relocate experience in the grammatical sequence, (b) prepare to start over again with a fresh sequence, and (c) end the sequence or, ultimately, the path of practice. Hence, the grammar of the syntagm serves a ‘reassuring’ purpose of holding together the uncertainty of experience. Yelle states that a ‘basic function of signs is to produce certainty in the existence of the state of affairs of which they stand as a guarantee or certificate’ (Yelle 2013, p. 17). However, this paper argues that the signs by themselves (image proliferation) are not enough to guarantee certainty in the meditation journey; while the signs often generate uncertainty, it is the grammar of the stable sequences that grants certainty.

Thirdly, there are also wider issues to be addressed here in relation to the unwieldy question of ‘what is meditation for?’ In a tradition like early Buddhism, or, indeed, early yoga, in which meditation is a structured mental process designed to cultivate control and quietude, the grammar of meditation itself is the meditation. One might argue that meditation is grammar and that if there is no grammar, then it is not meditation but some sort of other mental or spiritual activity. And why is the imagery so vivid? The continuous accenting of imagery is a strategy for what in visual semiotics would be classed as ‘optimal signification’ (MacEachren et al. 2012, p. 2496). What is optimised is attention—such ultra-vivid images entrain attention and do not easily allow it to slip out of the visual pathways being signalled.

A future direction for research would be to integrate textual study of the QYM with the spatial cave analysis of Vignato and Hiyama (2022) and their hypothesis of two sets of Buddhist traditions in Kucha and surrounding areas (Tradition A and Tradition B, discussed above). Whereas from a textual perspective, the cave art and the material environment are supra-textual modes of meaning, from the perspective of the art historian, archaeologist or human geographer, the text is an extra-visual or extra-material mode of communication of meaning. Considered together, the caves, the space within them, and the rare materials found there (text, art and artefact) constitute a semiosphere, but not as a ‘world of signification’ (Lotman 1990), rather, as a ‘locality of signification’, confined to Qizil, Kucha and connected sites on the northern Silk Road. There are limits to the degree to which we can ever be competent interpretants of the ‘meditation language’ of the region in the way that the historic community practised it, but the Qizil Yoga Manual still holds significance for us today as we try to reconfigure the remnants of the language. As Semetsky asserts:

A sign doesn’t belong to a solely synchronic slice but […] cuts across it, thereby creating a past-future connection, thus demonstrating its own diachronic dimension as an unorthodox memory including the paradoxical memory of the future.

(Semetsky 2019, p. 10)

Since the Qizil Yoga Manual signals its ‘paradoxical memory’ of the future reader, the scholar, we can be grateful for the grammar of meditation that helps us to fill in the blanks or empty spaces.