This essay is in memory of Zhang Chunde 张纯德 (1941-2021), a great scholar of Nasu and other Yi peoples. Careful and generous readings of earlier versions by Aurelie Névot, Bruce Mannheim, and two anonymous readers for EMSCAT helped transform a bloated and digressive initial offering into the present, relatively compact form.
- 1 The phrase “united and equal” is from the title of a book intended to introduce China’s “minority n (...)
1China’s intellectuals rediscovered the indigenous peoples of the southwest during China’s war of resistance against Japan (1937-1945). In 1937, the nation’s elite institutions of higher education moved from Beijing southwest to Hunan, Yunnan, and Sichuan, where students and faculty began to invest scholarly attention in the indigenous peoples of the borderlands (Israel 1999; Mullaney 2010). Though the Communist Party had already famously built alliances with some of these peoples during the Long March, this new encounter had a more scholarly character. Take the exodus of the Southwestern Consolidated Universities to Kunming, Yunnan, in the spring of 1938, only a year after many of the institutions consolidated in the University had arrived in Changsha, Hunan. The students and professors of the Literature Department walked together more than 1 000 miles through the rough mountains of south Hunan, Guizhou, and east Yunnan. The Department’s faculty gave lectures; the students conversed while they walked (Ma 2000). The scholars knew the ethnic others they met along the way as “Miao bandits” (Israel 1999). Though most were afraid of them, others were intrigued by their difference and fascinated in by the forms of oral and written textuality they displayed in songs, ritual chants, and written ritual texts (Ma 2000). After the University set up in Kunming, direct encounters between scholars and indigenous farmers and traders multiplied, generating a variety of linguistic and ethnographic projects. Some focused on the several forms of writing employed by Tibeto-Burman speaking peoples. Scholars construed texts written in these scripts as literature and made them the objects of extensive translation projects. Such endeavors grew enormously in number and scope after the socialist revolution, when the state supported translation as part of its effort to forge China’s indigenous peoples into “minority nationalities” and full participants in the Chinese nation: “united and equal”1.
- 2 The alliances of writing with other kinds of graphism are often explored by those who study forms o (...)
2In extracting these many forms of written text from their living contexts, revaluing them as literature, and transforming them into the master language of Chinese, these projects invested in what we might call a modernist image of writing, grounded in enlightenment theories of language and meaning (Taylor 1985, p. 248; Watson 2013). The modernist image of writing, which also guided projects to translate exotic scripts in other parts of the world, located the value of any written text in the semantic content of its voiced equivalent. Projects shaped by this image approached written texts as strings of semantic values that could be fully excavated and fully represented in translation. Such projects severed texts from their inherent powers of graphism2. If a writing system used pictures, as did Maya and the Dongba script of northwest Yunnan, this was taken as a complication that may have made translation more difficult but did not change writing’s central task: to represent words of spoken language (Watson 2012, 2013; Poupard 2018, 2020). The first efforts to create full translations of texts in the indigenous scripts of southwest China had laudable intent: to counter the sense, ubiquitous in the 1940s and still common today, that these were not mature, fully developed writing systems. Translation in the modernist form worked against this idea by displaying these texts’ capacity to fully and unambiguously represent a semantic content.
- 3 In texts, Né (...)
- 4 Unless otherwise noted, Né graphs and Nasu-language transcriptions are drawn from Ma Xueliang’s dic (...)
This essay explores one of the indigenous scripts of southwest China, the Nasu or Eastern orthography of what is now called Yi 彝 script, used in Nasu-ethnicity communities in north Yunnan and west Guizhou. Texts written in this script refer to it as Né
writing3,4. I argue that ubiquitous modernist assumptions about writing deeply misconstrue what this writing system was and what it could do. When recited in its ritualized contexts, Né script repeatedly manifested and traversed fissures between language and image. The power of Né writing was not limited to encoding and expressing language; it extended to shaping relations among bodies – graphic, sculptural, and enfleshed. Put simply, the argument of this essay is that the force of Né writing did not lie merely in its faculty of representing a sequence of semantic values but in its capacity to manifest a distinctive form of life for humans, ancestors and spirits.
3A writing system is never a single topic; it is always a bundle of many. While this essay begins with writing, it does not confine itself to writing, nor to language narrowly construed. Instead, it works out associations between textual practice, graphic practice and ritual practice in an anthropological mode, if anthropology is understood to explore relations among domains of human experience conventionally viewed as separate. I show Né textuality crossing language and image on three levels: graph, verse, and page. While each level can be analyzed far more extensively than I do here, my effort is to show how they work together, especially in ritual, which is the context for most Né textual practice. At the level of graphs, I briefly demonstrate that the script was composed of networks of graphs, some associated through phonetic similarity, others through semantic similarity or indexical suggestiveness. These networks coalesced around central forms, manipulated by adding and subtracting marks. When abstracted from the glottographic function of the script, these forms could appear as discrete but mutating things, or bodies, which can be seen moving through any page. I argue that this aesthetic feature of the script is commensurate with the central use of Né script in ritual: to materialize the bodies of nonhuman meta-persons such as ghosts, spirits, and ancestors, by drawing them out of the world’s substrates.
4The second step of my argument looks at the way divination, used in every ritual, creates an implicit theorization of writing. In divination, writing emerges as textual marks created by nonhuman beings, which are then organized with graphs that interface with the system of spoken language. I make this point with the help of a pig’s scapula used for divination and inscribed with Né graphs that assign meaning to the cracks on the scapula’s surface. Moving next to verses, I argue that Né graphs tend to carry so many different meanings that semantic values can often only be specified by embedding graphs in parallel structures across verses. At the same time, Né verses both inscribe oral poetic language, where ambiguity is often a positive value, and transform oral poetry in patterned ways. Poetic verses in Né script tear ordinary words apart and distribute them across parallel structures. Written verses build graphs into multiple relational networks that often allow for several different readings. Such operations open up spaces for indeterminate meaning and nonmeaning to circulate between the graphs and their enunciation. This too is commensurate with the sense that writing participates in manifesting the bodies and voices of nonhuman beings, sometimes decipherable, sometimes not, on the world’s surfaces.
5Finally, verses were typically written on pages and assembled into books. I show how the ordered, bounded, rectilinear form of the written page was iterated and extended in the form of a ritual field, populated with graph-like sculptural bodies of nonhuman beings. Ritual fields were diagrammed in verses and in images attached to texts. Texts were recited and recopied in ritual fields, and ritual fields were folded into the pages of books. Ritual action was about moving ghosts, spirits, and ancestors through a layered world, as graphs on a page, effigies in a field, and abstract bodies in a cosmos.
- 5 Névot (2019) makes a more extended effort along similar lines, using different methods.
6The modernist image of writing severs writing from its graphic and sculptural alliances, assumes that writing merely encodes a semantic sequence, and asserts that meaning can be fully excavated from any text and fully represented in another writing system. In showing some ways that Né script far exceeds this image, this essay hopes to dismantle the common assumption that as soon as we encounter writing, we already know what it is5.
7Before working through this sequence of graph, verse, and page in relation to Né script, I briefly describe the ethnographic encounter of the early 1940s that provided the material for this essay and introduce the regional ecology of languages, writing systems, and texts in which Né script participated.
8Among the Consolidated Universities Literature Department students who trekked from Changsha to Kunming in 1938 was Ma Xueliang. Ma was a student in Chinese Literature at Peking University, studying with the eminent linguist Luo Changpei, and interested in the Shuowen, the first dictionary of the Chinese language. After graduating in wartime Kunming, Ma began a PhD in linguistics directed by Li Fang-Kuei, an expert in Athabascan languages trained at the University of Chicago with Edward Sapir and Leonard Bloomfield. The war had inspired Li Fang-kuei to return to his native China to study the Tibeto-Burman languages of the southwest. Li and Ma spent a summer together southeast of Kunming researching the Sani (or Ni) language, one of the many languages that Ma would later help classify as Yi 彝. Ma wrote his dissertation on Sani language, then became Li Fang-Kuei’s research assistant. He studied the Cuanwen congke (1936), a three-volume compilation and translation by Ding Wenjiang of texts from rural Guizhou in a script that Ding and others called Cuanwen 爨文 after the Cuan-clan rulers of central Yunnan in the 4th and 5th centuries. It was probably this work that inspired Li Fang-Kuei to send Ma Xueliang to Luquan County in northern Yunnan to look for books in the script. A famous 16th-century cliff inscription in the script was to be found there, and the county was known as a center for writing (Ma 2000).
- 6 On the Lv̩hu native prefects, also known by the Chinese surname Feng 鳳, adopted in the 16th century (...)
- 7 Nuó (in Hanyu pinyin) was the local pronunciation of the surname 那 (Zhang 1995).
- 8 Ma published these translations under his own name while acknowledging his collaboration with Zhang (...)
Ma Xueliang found many books in the script, which he learned to call Luowen 罗文 after the Luowu 罗婺/Lv̩hu 
native prefects of Wuding, who had dominated northern Yunnan from the 10th century until they were deposed by the Ming armies in 15676. He was determined to learn to read, and he searched for a teacher among the literate ritualists called pemɒ́ who kept, copied, and recited the books. Eventually, he met Zhang Wenyuan, whose family home had been, for six generations, a node in a network of readers and copyists of the script that extended through north Yunnan and west Guizhou. Zhang Wenyuan brought Ma Xueliang to the village of Wande in Wuding County, just west of Luquan. There, they visited the hereditary home of the Nuo 那 native chieftains of Mulian 慕連 or Maolian 茂連 native domain, who claimed descent from the Lv̩hu native prefects7. Their host was Lady Nuo An Heqing, who was to be the last of the Nuo chieftains, presiding over a much diminished domain. In house’s document room, Nuo An Heqing showed the two scholars a “stone chamber” piled with manuscripts: the administrative, legal, genealogical, and ritual archive of the Nuo native chieftains (Ma 2020; Meng 2015). Some 500 of these manuscripts were Né-script books. Ma and Zhang spent about a year in the house together in 1942-1943, cataloguing these books and translating two, recited at funerals and at a feast for the ancestors called némv̩̏8. In 1943, Ma negotiated the sale of the entire archive to the Peip’ing National Library for 100 000 yuan and a certificate of appreciation from the Ministry of Education (Meng 2015).
9The essay uses books from this collection, others collected in Luquan and Wuding around the same period, and the ethnographic and linguistic work done in 1942-43 by Ma Xueliang and Zhang Wenyuan. Inspired by Ding Wenjiang’s (1935) translations, Ma and Zhang refined Ding’s “four-way comparison” (si duizhao) system of translation, which became the gold standard for translations of texts from the indigenous scripts of the southwest into Chinese. Four-way comparison uses four lines of text: 1) the original script, usually reoriented to conform with the reformed Chinese orientation of lines of text running left to right, top to bottom; 2) a graph-by-graph transcription using International Phonetic Alphabet; 3) a gloss that matches each original graph with a corresponding Chinese graph; 4) a translation into Chinese, using Chinese grammar and word order, rigorously matching the number of original graphs in each line to the same number in Chinese. This system, displayed in many multivolume sets of translations produced by local governments and research institutes, institutionalizes the modernist image of writing. It displays translatability in its most triumphant form, insisting that each graph correspond to stable phonetic and semantic values, that a definite meaning can be fixed for each verse independently, that this meaning can be fully translated, and that translation can exhaust the significance of any text. At the same time, in its relative transparency, the system leaves traces that can undermine each of these principles.
- 9 There is now an extensive body of work on the Sani or Ni variety of the script, by Aurélie Névot. F (...)
In 1954, Ma Xueliang participated in the nationwide “ethnic identification project” that created official ethnic categories for all of China’s indigenous peoples. Confronted with enormous diversity in the southwest, ethnic identification teams folded peoples who spoke some sixty distinct languages into a single nationality, which they called Yi 彝. Mullaney (2010) has shown that this project was oriented towards the future: a hope that these groups, however disparate, could eventually be convinced to adopt a unified identity. Though in China Yi remains the term applied to all these groups and their languages, Western linguists have adopted other taxonomies for the complex linguistic situation. David Bradley (2009) has proposed the language group Ngwi, based on a variety of autonyms, including Né
, used by Nasu speakers. Ngwi comprises about a hundred languages, some sixty of which are spoken by peoples officially grouped into the Yi nationality. Four distinct varieties of script are associated with some of these languages: 1) an Eastern variety, associated with Nasu languages spoken in north Yunnan and west Guizhou; 2) a Northern variety, associated with Nuosu languages spoken in southwest Sichuan and northwest Yunnan; 3) a Southern variety, associated with Nisu, spoken in south and south-central Yunnan; and 4) a Southeastern variety, associated with Sani or Ni languages, spoken in south-central Yunnan9. In the 20th and early 21st centuries, all four orthographies have been subjected to processes of reform and standardization, producing several new, standardized scripts (Bradley 2001; Wasilewska 2012; Névot 2019). My examples here are all from the unreformed Eastern or Nasu variety and the dialect of Nasu spoken in Wuding and Luquan counties, as recorded in Ma Xueliang’s (1998) dictionary.
- 10 See Névot (2019) for a comprehensive summary of the many inconclusive hypotheses about the origins (...)
- 11 In some Né-script histories, such as the genre titled the “Book of Yi and Han Teachings”, Né ʂà̠ (...)
- 12 Sanders was commenting on an essay by John Kelly (2007) about writing in China and India. For some (...)
10Though useful, identifying the four Yi orthographies with specific languages gives the mistaken impression that ethnicities, languages, and writing systems can be matched up. Even within each of the four distinct orthographies, there was much variation, both regional and local. Regionally, for instance, the version of the Eastern or Nasu orthography used in north Yunnan displayed systematic differences from versions of the Eastern orthography used in Guizhou (Bradley 2011). Locally, writing was guarded as secret by lineages of literate ritualists who developed many variations on graphs within their own traditions. Moreover, the entire area where these scripts were used was vibrantly multi-ethnic; many languages were spoken, and most speakers in majority indigenous places were multilingual. Though Nasu was the language of the former ruling lineages in the indigenous parts of Wuding and Luquan, other languages were spoken by significant portions of the population there. These included seven other Ngwi languages (Aluo, Naisu, Lipo, Lolo, Miqie, Geipo and Hani), two Hmong languages (Hmong and Ahmao), one Tai language (Dai), and Yunnan Chinese (Gao 2017; Lama 2012). The sources show only Nasu being used to recite Né texts, but it is possible that some were recited in other Ngwi languages and certain that audiences were multilingual10. Since the morphology of Ngwi languages is not affixal, a logographic script worked well and was easily adoptable (Bradley 2009). While the origin of the script remains obscure, it is likely that, like other writing systems, this one was created in one location and then moved through the region colonizing languages and being modified by its users11. “The early history of writing suggests that writing is originally something that encounters language from the outside rather than flowing directly out of language”, suggests Seth Sanders (2006)12.
11Né and other Yi scripts inscribed poetic language. These scripts must be understood as embedded in an “areal network” of languages, united “by what people do with their languages,” as Jane Hill put it (Hill 1978, p. 1; Michalowski 1987, p. 4). In this region, speakers invest in a rich variety of formal poetic forms, including ritual speech to ghosts, spirits and ancestors, laments, creation epics, and dialogical courting songs. These forms used poetic formalization and innovation to create performative breaks with ordinary speech. They were employed to communicate across divides perceived as ontological: between the living and the recently dead, between people and ghosts, spirits, and ancestors; and between young men and young women. Not just to communicate, but to influence and manipulate: to grip bodies and compel them to act in desired ways. Across many languages, these textual forms use similar patterning constructs, including consistent patterns of meter and verse, parallelism, and other poetic devices. These textual forms often connected to sculptural and graphic bodily forms, appearing sometimes as corpses, sometimes as drawings, and often as effigies built of materials gathered from forest and field (Mueggler 2001, 2017).
- 13 Such a revolution would be suggested by old “literacy thesis” promoted by Goody & Watt (1963), Have (...)
12Writing, adopted in a minority of the Ngwi-speaking communities in the region, did not revolutionize these poetic practices13. Writing used the same poetic devices as these spoken forms, often reducing the length, variety, and complexity of poetic practices that moved across the narrow defile from oral texts to graph, verse, and page. But writing also added layers. It added other ways to connect language to graphic bodies; other ways to force meaning to spring out of the polysemies and ambiguities of verses; other ways to organize ritual action using the squared-off margins of a written page as a guide. Writing added to ritual practice an ocular-manual channel in parallel with the oral-aural channel of spoken text forms. In this laminated semiotic context, one of the foremost tasks of writing was to theorize itself as writing. After briefly describing how one central feature of Né graphs creates an immediate connection to graphic bodily forms below, I move to one way that Né writing used in divination practice creates an implicit theorization of how writing draws the bodies and speech of nonhuman beings out of the surface of a scapula, an image of the cosmos.
13Like every organized system of inscription, Né script packs into its graphs the history of their formation. This history shapes the experiences of those in contact with the script, both as they decipher its phonetic and semantic content and as they sense and interact with its aesthetic affordances.
- 14 Substantially logographic examples include Tangut, used by the rulers of the Western Xia dynasty (G (...)
14Né script was certainly inspired by Chinese, which is much older. Né borrows the characteristic feature of Chinese writing: its way of matching graphs with words, shared only by scripts that were also inspired by Chinese14. In Chinese and Né, the phonetic value of the graph is at the level of the syllable rather than the phoneme. Though Chinese is often contrasted with so-called phonetic writing (meaning writing in which the phonetic value of graphs is at a level below the syllable), both Chinese and Né are obviously phonetic: all graphs have a specific, recognized pronunciations within a particular dialect (though we shall see that Né frequently bends this rule). In modern Chinese and Ngwi languages, many words are formed of two or more syllables, which require two or more graphs to write. When these words are broken down into syllables, however, each syllable retains a sematic value of its own.
15Né script and Chinese script are the only living writing systems that match graphs to words in this way. Is Né script, merely a version of Chinese script, constructed on the same principles while using different graphs? Not at all. I want to briefly point out one central feature of Né script that makes it very different from Chinese and lends it specific aesthetic qualities. The point of this demonstration is to show that Né script might be perceived by its readers as many series of unstable body-like forms that move through every text, changing shapes but retaining recognizable identities.
- 15 See Boltz (2000, 2011) for descriptions of this recursive process.
Unlike Chinese, which forms graphs as compounds of other graphs in a recursive process that produces graphs of up to five or more parts, nearly all graphs in Né are “unit characters”15. That is, they are all simple graphs, not compounded out of other graphs. The very few exceptions include graphs that are simply doubled, such as the frequent
má, “soldier, to salute, to adopt a relative”, which doubles the even more frequent
mà, “no, not.” Or
tɤ̠̀, “layer, to grow”, which doubles
vɤ́, “bird, to hide”. Aside from rare instances like these, Né graphs cannot be broken down into smaller units that are graphs themselves. Yet Né graphs do follow identifiable principles of development. Unit characters are made into different unit characters by adding, subtracting, or modifying elements, along two paths.
16On the first path, graphs are modified to make new graphs with different semantic values but similar phonetic values, creating what we can call, using a vocabulary developed to analyse Chinese script development, phonophoric series (Boltz 2011). Figure 1 shows three examples of phonophoric series in Né. In the first column, elements are added to a graph to create different graphs; in the second, elements are added and subtracted; in the third, graphs transform along several axes with no single element remaining stable.
Figure 1

© Erik Mueggler
17Because phonophoric series in Né do not usually hold any single graphic element stable, their boundaries are often fuzzy. They are families of graphs that resemble each other and have similar phonetic values, rather than rigorously determinable bounded series.
The second path of graph construction in Né can be called semantic extension or polyphony (Boltz 2011). In this procedure, graphs are modified to create new graphs with different phonetic values but suggestive indexical connections. This is an extensive feature of Né script. We can take the graph
p’ɔ̏, “scapula, to prognosticate with scapulamancy” as an example. A sizeable fraction of Né graphs are treated by the script’s readers as icons – pictures of an idea expressed by the words they represent. These are not pictographs or ideographs in any meaningful sense: they are not realistic pictures that stand for concepts. For instance, only after you know that
ɤù̠, means “saddle” does it begin to look like a pair of stirrups. In the same way, once a reader knows
p’ɔ̏ means “scapula” it begins to look like the triangular shape of a scapula, with a crack on its upper part. The shape of the graph
p’ɔ̏ can then be modified to stand for a closely related concept, such as
ɖȕ, “to burn”.
p’ɔ̏, “scapula” is related to
ɖȕ, “to burn” through a suggestive indexical connection: the common procedure for scapulimancy, familiar to every copyist and reciter of Né script, in which a diviner piled dry grass on a scapula’s surface and burnt it to create cracks (Zhang et al. 2007, p. 359).
- 16 Mannheim demonstrated similar patterns of indexical suggestiveness, involving sound rather than wri (...)
Brought together, these two paths of graph creation form extensive networks of graphs, where a single unstable but recognizable form appears both in graphs with similar phonetic values and graphs with suggestive indexical connections. For instance, to look at the network that includes
p’ɔ̏, “scapula”, and
ɖȕ, “to burn”,
ɖȕ is forms a phonophoric series with
dɯ, “short, stump, to emit (steam)”;
dɯ̏, “bad, weak, incapable, twisted”; and
tȕ̠, “to awake, to stand”. For its part,
p’ɔ̏, is also related to its near homonym
p’ɔ, “flag”. The cloth
p’ɔ, “flag”, creates a suggestive indexical link to
p’ȕ, “cloth”.
p’ȕ, “cloth”, is connected through indexical suggestion to
ɬɯ̠́, “pocket, to shed a (snake) skin”.
ɬɯ̠́ is associated in a phonophoric series with
lɯ̏, “building” and
ɬɯ̠́, “to herd, to frighten, to observe”. On another indexical path,
p’ȕ, “cloth” transforms into
tɤ́, “to embroider, to decorate”, which appears to have the iconic association of pulling a thread on a needle through the cloth. The clear associations of
p’ȕ with cloth and thread may also motivate
ɲɑ̏, “spider”, which is related in a phonophoric series to
ɲɑ, “sticky, to adhere”16. These associations might be presented as a diagram (see fig. 2).
Figure 2. Horizontal rows indicate phonophoric series; lines represent indexical connection

© Erik Mueggler
The form
is used in many other graphs as well, but even this limited demonstration shows that Né script can be perceived as a network of indexical connections, often employing a degree of iconicity (such as the crack in the scapula of
p’ɔ̏, “scapula”, or the needle and thread in
tɤ́, “to embroider”), out of which emerge brief phonophoric series.
A reader is thus led to the impression that the foundational forms of Né graphs are physical objects: human or nonhuman bodies or body parts. Perhaps the most obviously anthropomorphic body in Né script is the non-human form
p’v̩́, ‘male ancestor,’ a frontal figure with limbs, a male organ, and a prominent mouth. And the most vivid icon for a body part might be
pə, “vagina”. But
p’i, “buttock” is vivid enough, especially when paired with
ɬɤ̏, “cave” to become 
p’i ɬɤ̏, “anus”.
p’i, “buttock” retains much of this iconicity as it transforms along a phonophoric axis to become
p’i, “female ancestor”. When both
p’v̩́, “male ancestor” and
p’i, “female ancestor” are seen as (foundational) bodies, then other graphs might also begin to take on embodied form. For instance, the crack in the scapula of
p’ɔ̏ might appear as the mouth of a speaking being. And other graphs in the various
series that we have already examined also begin to reappear as bodies:
tȕ̠ “to awake” as a prone body whose only moving part is a mouth;
dɯ̏, “bad, weak, incapable” as an erect body with a mouth;
ɬɯ̠́, “pocket” as an erect body with hands by its sides;
tɤ́, “to embroider, to decorate” as a body with a fancy sash or belt;
ɲɑ̏, “spider” as a crawling body emitting thread. This is not to say that the
form always retains its integrity. It opens at the top in
ɲi, “to come”, perhaps forming a pair of welcoming arms; it bursts at the bottom in
dʐɒ, “to be in a place, to have, the bark of a hunting dog”.
18These are not necessary intuitions for readers, merely possible ones. What I wish to point out is that the tendency of Né graphs to grow up around foundational forms peoples a text with embodied beings and part-beings, each appearing in many different guises. The threads of suggestive indexicality that dance through every page of Né script tend to animate it.
- 17 See Névot 2019 for several examples from the Sani part of the Ngwi-speaking world.
19That Né script is built of bodies and body parts that sometimes cut themselves loose from graphs is not surprising when we consider that most Né texts were used in practice to give spirits, ancestors, ghosts, and demons bodily substance so they might receive gifts, negotiations, and commands. In this section I want to follow this idea further by tracing an implicit theorization of writing which emphasizes the idea that writing draws nonhuman metapersons and their parts and capacities out of the world’s varied surfaces. Né scriptive and ritual practice is full of such implicit attempts to explain the nature of writing in graphic and gestural practices and origin stories17. Indeed, it seems to be one of the tasks of Né ritual to think about the nature of writing and to work out hypotheses about how it works and what it can do.
In the spirit of the graph
p’ɔ̏, which preoccupied us in the last section, I draw my example from the context of scapular divination. Prognostication using scapulae was an essential aspect of every engagement with nonhuman beings in the Nasu-speaking communities of northern Yunnan, as in most Ngwi-speaking regions. In these communities, divination was performed with many materials, but chicken bones and pigs’ scapulae were the most prominent. Pemɒ́ recited long texts devoted to scapular divination at funerals and ancestral feasts – we will look at one such text in the next section. To briefly prefigure my argument, scapular divination was about reading a text,
mí, of cracks on the surface of a scapula. As in much Né textual practice, reading was combined with copying. Copying the cracks on a scapula, diviners reorganized them into written graphs, 

sv sú mù, which could then be recited as language. As implicitly theorized in divination, writing was fundamentally an effort to coordinate marks that stood for the speech of nonhuman beings, emerging from the world’s surfaces, with the system of spoken human language.
20As the ideas underlying scapular divination were widespread across the Ngwi-speaking world, we can begin with an example slightly removed from the Nasu context: a conversation with Nuosu ritualists. I say slightly removed because the Nasu communities of the former domain of Mulian in northern Wuding county were closely involved with Nuosu communities in Pulong domain, immediately across the Jinsha river. There was much intercourse across the river, encouraged by marriage alliances between the Nuo chieftains of Mulian and the Nuosu-ethnicity Sha lineage of chieftains of Pulong. The example I draw on here emerged from a conversation between the ethnographic linguist Zang Chunde, who descended from a pemɒ́ lineage in the former Mulian domain, and two Nuosu ritualists name Li Zhifa and Sha Shifu, not from Pulong but from Ninglang county, Yunnan.
21Li and Sha showed Zhang Chunde how to prepare a sheep’s scapula for divination by piling fire-grass on its surface and lighting it to produce cracks. Then, as the fire was burning, they recited an invocation to the scapula, which described the wanderings of the sheep over the mountains. The sheep followed the clouds and learned their secrets, followed the fogs and mists and learned their affairs. It learned “the speech of the sun, the language of the moonlight”. “After three years, you became a clever sheep bone; after three months, you became a prognosticating bone; a sheep bone that can foretell ten thousand things. Not the hind leg but the front leg; a sheep scapula with a snow-white surface; a sheep scapula with a blade bright as gold” (Zhang et al. 2007, pp. 359-360).
22Having absorbed all the wisdom of the universe, the scapula became a miniature analogue of the cosmos. If the bone was oriented with the broad edge upward (as in fig. 3), the top part of the surface reflected the will of heaven; the narrower bottom part reflected the earth and the underworld; the left side bore on matters of the host household; the right side was about the affairs of guests. While most ordinary people could discern the meanings of simple cracks using this schema, more detailed diagnosis required the services of a specialist. Li and Sha listed thirty-three specific configurations of cracks and their meanings (Zhang et al. 2007, pp. 360-361). This list summarized a systematic graphism, with a lexicon of crack-graphs with discernible semantic values but no stable phonetic values.
Figure 3. Pig scapulae from Ma Xueliang’s collection

© Courtesy of the Institute of History and Philology, Academia Sinica
Another exhibit brings us back to the Nasu-speaking communities of Wuding and Luquan. Among the artifacts that Ma Xueliang collected in those communities are several pigs’ scapulae. (See fig. 3). Five graphs have been written on one scapula with a bamboo writing stick, on an uncracked surface that has been prepared for divination. The text reads 



vá̠ p’ɔ̏ dù̠ dze su, “pig’s scapula for the eagle’s descent”, indicating that this bone is to be used in a funeral ritual or ancestral feast, where the eagle’s descent, a metaphor for death diving from the sky to capture a loved one, was to be described in detail. Another scapula, possibly even the matched pair of the first, has been burned to create cracks and pits and then inscribed with two colors of ink. Black marks have been added to highlight certain cracks and pits. Other marks have also been added, such as the spiral near the center of the scapula and the constellations of dots near the right edge. The graph
p’ɔ̏ has been written at least five times with lines and arrows extending from the graph to point out specific p’ɔ̏, or divination cracks. In at least three places
p’ɔ̏ begins a phrase or verse – likely compressed formulae for the meaning of certain cracks. All this is written over a layer of marks in red ink, which are faded and unreadable.
- 18 Névot writes of a similar ritual use of a scapula in the context of the yearly Mizhi sacrifice in a (...)
23This scapula was probably a kind of teaching text on which a master labelled significant cracks and pits for the benefit of an apprentice diviner18. At the same time – and this too may have been a pedagogical demonstration – the scapula performs a simple theorization of writing. We might translate this demonstration into our own analytical language in a fashion that, while admittedly speculative, nevertheless hews closely to the evidence I have presented. We have learned that a scapula is an icon of the universe. But this icon is more than merely a representation. It is an iteration of the earth and sky, concentrated and ordered, absorbed by the wandering sheep or pig and transformed through sacrifice and cleaning into a snow-white surface open to inscription. The marks that emerge from this surface express the influence of nonhuman animate powers on a household – also present in the scapula as hosts (on the left side) and guests (on the right). These marks are signs, singly articulated with possible semantic values but no definite phonetic values – the “speech of the sun, the language of the moonlight”. To read these marks, one first goes over them in black ink. In effect, one copies them, as copying is the first step that any apprentice pemɒ́ undertakes when learning to read a new text.
- 19 This parallels the idea, proposed by some linguists who study the origin of Chinese writing, that t (...)
Just as in copying a text written in graphs, copying these marks emphasizes some, overlooks others, and reorganizes still others (like the pits in the scapula’s center that become three evenly spaced dots). In this way, copying initiates a reformulation of the graphism that emerges from the world’s surface into a register intelligible to the human eye. The next step is to link graph with sound. The scapula itself makes a sound, p’ɔ̏, as each crack/mouth opens on its surface19. The icon
drawn on this surface depicts the opening of a mouth/crack on the scapula, coordinating this sound with the crack on the one hand and the system of human language on the other. An indexical sign completes and reinforces this relay: the line or arrow that links the icon
to the crack that emits the sound p’ɔ̏, which the icon now stands for as the symbol
p’ɔ̏. Having made this leap with
p’ɔ̏, one then extends it in the compact phrases written on the scapula. One then elaborates it with more graphs, each of which takes advantage of the double articulation of semantic and phonetic values with graphic signs already established.
24This implicit theory of writing begins with the world, concentrated in the surface of a scapula, which has become an image of the entire universe: sky, earth, underworld, host, guest. The world emits its own graphism, expressing its own languages. Writing reformulates this graphism and creates the articulation between these reconstructed marks and human language. Writing is always copying, and copying is always a reformulation and transformation. One gets the sense in reading Né-script books that writing remains close to these origins. Just as the world’s surfaces are strewn with creatures, the books seem populated with bodies – those unit forms, always changing but aways identifiable, that move through every verse of every text. It is as though when the crack/mouth of the scapula opens, the language it emits, voicing the intentions of the unseen beings of earth and sky, is a string of bodies. We are reminded again that the central purpose of Né script in ritual is not so much to represent or express (though it does all this too), but to draw out the nonhuman beings from the world’s surfaces, to negotiate with those beings, and to move them through the world.
25Most Né-script texts are written not on bones but on paper. Originally, ritual texts were written by spirits in the sky, the pemɒ́ Zhang Wenyuan told his collaborator Ma Xueliang in 1942. Pemɒ́ brought the texts to earth, tied to the horns of oxen. As the oxen crossed the seas, the books got wet, and the pemɒ́ disassembled them and lay the pages out on trees along the bank to dry. Leaves adhered to the wet pages and obscured half the writing. The task of a pemɒ́ copyist and reciter is to fill in the missing text. Yet no recitation is complete, and much that was written remains uncopied and unsaid (Ma 1987a, pp. 64-65).
26This little story efficiently expresses the sense that Né-script verses are strewn with openings or gaps through which other possible meanings might shine – a theme I follow in this section. Arranging graphs into verses and creating parallel structures that linked verse to verse were necessary for creating texts that could impose definite semantic values on the string of graphic forms. But I want to show that while verses made Né script intelligible, they also disarticulated and redistributed ordinary language, disturbing simple correspondences between word and meaning, creating new possibilities for polysemy, and making ample space for sounds to which meaning was only weakly attached.
Aurélie Névot (2019) has made a thorough analyses of verse parallelism in texts in the Sani version of Yi script, which I cannot match here. For my purposes, however, I need only a couple of simple examples. In keeping with the theme of scapular divination, I draw my samples from a Né-script text about pig-scapula divination read at an ancestral feast called némv̩̏, performed in and around Mulian domain. The archive of the Nuo chieftains of Mulian contained some 41 texts about divination, including 20 about pig scapulae, 9 about cattle scapulae, and 12 others, mostly about chicken-bone divination (Yang 2010). This one is titled 





Vá̠ p’ɔ̏ dɯ ɬí xə̀ bə̀ su, “Book for exorcising evil and welcoming good with a pig’s scapula” (CYZRZ 2007, pp. 94-182).
- 20 This collection of 1 223 manuscripts includes the Né script texts from the Nuo family archive and a (...)
- 21 Here and in the excerpts below I follow the Nasu-language transcription given in CYZRC 2007, which, (...)
27Like most Né-script texts, this one is arranged in five-syllable verses interspersed with occasional verses of three, seven, nine, or eleven syllables. This is a very familiar form in the region, common in all the four orthographies of Yi script and also used in oral texts with no written equivalents. Like 74 percent of the Né-script manuscripts in the National Library collection, this text was written top to bottom, left to right (Yang 2010). The other 26 percent are written top to bottom, right to left like traditional Chinese texts20. For legibility, I have rearranged my examples left to right, top to bottom, following the conventions of western languages and modern Chinese, with verses indented to align with their parallel verses and a phonetic transcription placed below the graphs (CYZRZ 2007, p. 115)21.
1.
one day a chieftain arrived
t’à nì ndz’ɯ dzù tɕ’ì
2.
came to harass with power
fə ʂɯ́ t’ì lè t’o
3.
one day a chieftain arrived
t’à nì mɔ́ dzù tɕ’i
4.
came to harass with punishment
k’o ʂì̠ t’ì lè t’o
- 22 Ma 1998, p. 340. The first definition of (...)
Lines 1 and 3 are identical but for the third graph in each. This position is taken by the graph
ndz’ɯ in line 1 and
mɔ́ in line 3. In ordinary speech, these two lexemes are usually used together as ndz’ɯ mɔ́ to mean “chieftain”. In this excerpt, the word is split up and distributed between two lines. By itself, each syllable has several meanings. As a noun,
ndz’ɯ (or nts’ ɯ) refers to a king, a lord, or a chieftain; as an adjective, it means “male”, when applied to birds, or “gorgeous”22. By itself,
mɔ́ (or mɒ́) is a minister or agent of a chieftain, but it also means “old”, and is commonly made part of a word as a mark of respect, as in 
pemɒ́, “ritualist”. When reunited through the parallelism of these two lines, each can unambiguously mean “chieftain”. But standing alone in its line, each might also be read to retain traces of its other meanings, emphasizing the maleness, splendor, and worthiness of the chieftain. In this very common strategy, the formalization of speech rips apart words of ordinary language and redistributes their parts, allowing a fullness of signification to shine through the rifts it creates.
These four lines contain more instances of parallelism, of course. In lines 2 and 4, the phrase 
k’o ʂì̠ is made parallel to 
fə ʂɯ́, “administrative power”. This makes it possible to read 
k’o ʂì̠ as “punishment” or even “law”, while by itself it would mean, more generally, “to bind” or “to warn”. And finally, these verses are embedded with others in a typical series that moves from “chieftain” to “ritualist” to “craftsman” and then to “farmer”. Read through this series, which occurs frequently in this text and many others, these four lines appear not so much as a protest against chieftains who harass their subjects than as a formal classification that matches persons to actions that define them: chieftains to governance and punishment, ritualists to speaking, craftsmen to displaying their skills, farmers to herding and planting. The text makes its central themes – harassment, obstruction, evil luck, and bad fortune – circulate through this series, hooking on to and letting go of each category of person in sequence.
28These devices are typical of formal texts in the region, whether written or not. But Né-script texts display other strategies that make use of the specific capacities of writing – or rather of this form of writing. In brief, graphs are read relationally, and are often built into multiple relational networks at the same time, allowing for distinct interpretations of individual graphs to emerge from any reading. Here is another example from the same text (CYZRZ 2007, p. 115):
1.
Send away a flock of evil obstructions
tʂ’ì̠ ve du ɤa̠ xo
2.
And after you do this
t’ì ɬí ʑà̠ ve du
The phrase that makes these lines echo each other is 
ve du, in the center of line 1 and at the end of line 2. In line 1,
ve, “evil” modifies tʂ’ì̠, “obstructions” (used repeatedly in this sense in prior verses), and
du is made to modify tʂ’ì̠ ve, “evil obstructions” to mean “a crowd or collection”. A reader knows that
du must mean this because the verse in which it is embedded is strictly parallel to many other verses that sum up sections of the text by telling the ritualist to send away a crowd or collection of evils of specific kinds. The ordinary meaning of
du is actually “feather”, so I have glossed it “flock” rather than “crowd”. How did du, “feather” come to occupy this position in the line, a position which stands for “crowd” or “collection”? 
ve du, using a different graph for du, means “evil ghost”. Written 

, tʂ’ì̠ ve du might mean “ghostly obstructions”. The writer or copyist seized the opportunity to make “evil obstructions” more vivid by inserting
du after ve in the position where a word meaning “crowd or collection” goes, thus allowing the phrase to vibrate between two possible readings: “ghostly obstructions” and a “crowd [or flock] of evil obstructions”.
But 
ve du also appears at the end of line 2, where it means “after”. Again, a reader knows it must mean “after” because line verse parallels many other lines in the text that also instruct the ritualist to proceed to the next step. In most cases, these instructions are written exactly the same except for the last graph, which stands for “after”. When
dù, “suffering” is the topic, for instance, ve du, “after” is written 
. In each case, the writer has found a way to create a parallelism in graphism as well as sound by substituting a homophonic graph from the previous line to write du at the end of the second line in a position that has to stand for “after” no matter what the ordinary meaning of the graph. The phrase t’ì ɬí ʑà̠ ve du, “after doing this”, is written about 25 times in this text, with the graphs
,
,
,
, and
occupying the place of the final du. Finally, the word rendered ve du in all these ways in the text is usually ve ɤɯ̠́ in ordinary speech, which would be written 
.
- 23 I borrow the sense of “shimmer” used here from Roland Barthes’ lectures gathered in The Neutral: “t (...)
29“All media of transmission require a material channel, and the characteristic of every material channel is that beyond – and, as it were, against – the information it carries, it produces noise and nonsense”, wrote the media theorist Friedrich Kittler. All textuality “stands in an essential […] relation to a non-meaning, which it must exclude. It is defined not by what it means but by the difference between meaning and non-meaning, information and noise, that its medial possibilities set into place” (1990, p. xiv). At the level of the verse, Né ritual texts open up an extensive space for meaning to shimmer into and out of existence23. In this space, graphs are often rendered into sounds without a requirement that a determinate meaning be fixed; meanings of some verses are deferred until they can be completed a few lines later; other meanings distribute themselves across widely separated parallel series of verses; and possible meanings accumulate in a repeated verse until it begins to collapse into non-meaning.
The “medial possibilities” of Né texts, in other words, bring the boundary with non-meaning right up to the text’s surface. I have argued that one model for Né writing is the surface of the scapula. On this model, writing first emerges as cracks emitted by nonhuman beings. These traces of nonhuman bodies and voices (the image of which is the graph
p’ɔ̏, an icon of a scapula, with a crack/mouth) are gradually consolidated into written graphs that carry phonetic and semantic values. The scapula’s white surface, an image of the cosmos, is the porous membrane between the sounds and marks of ghosts, spirits and ancestors and their human translations. At the level of the verse, ritual writing in Né script preserves space for the inarticulate and nonhuman noise on which the written graph is founded to pass through this membrane.
When Né-script graphs were arranged into verses – which was most of the time – they were almost always inscribed on the form of a page. The page form was as fundamental to the practice of Né script as the verse form, for the page echoed the form of a ritual field, creating a resonance between graphs and the graph-like sculptural forms for ghosts, spirits and ancestors, called
ŋk’v̩̏ that populated every ritual field. Many Né ritual texts are diagrams of ritual fields in verse; others contain drawn images of ritual fields in diagram form; and all are intended to be recited only within the squared-off margins of a ritual field (see fig. 4). The patients of ritual action were moved along paths recited from the page, built into the field, and inscribed into the world at the same time. This interfolding of page, field, and world; and of graph, sculptural form, and being, was the substance of ritual power. In this context, Né texts continued to elaborate understandings of what writing was and what it could do. I have argued that scapulimancy performed an implicit theorization of writing at the level of the graph. In this section, we look to texts about divination to see how a similar self-description was elaborated at the level of the page.
Figure 4. Diagram of a divination field, from a text on pig-scapula divination, with
ŋk’v̩̏ marked as straight and branched lines and labeled with graphs

© Yang 2010, doc. no. 470
Most Né-script texts are in the form of books, bound with string on the left margin. The Né-script books in the Nuo family archive ranged from just a few pages to 144. The longest works were sometimes divided into two books, and often several different works were bound together into a single volume. The books were in many sizes, from about 18 cm to 30 cm tall and from about 15 cm to 30 cm wide. The most common layout was taller than wide and about 25 cm by 17 cm. Most books had no cover or title page. Titles were inscribed on the first page, often circled with a jagged or decorative line. Pages had slim margins, rarely marked or decorated. One work (which excited Ma Xueliang very much) was printed with the woodblock technology with which books in Chinese and Tibetan had been printed for centuries. This turned out to be thirteen separate editions of the
ɲȅ ɲo̠ ɲȅ mv̩̏ su, or “Book of good deeds”, a heavily annotated translation of the popular Chinese-language morality book, Taishang ganying pian (Yang 2010; Meng 2015; Ma 1986).
The textual ideologies that literate ritualists attached to Né writing emphasized the dependence of writing on paper bound into books, as the story I have related about the origin of Né ritual texts illustrates. Ritual texts were already books when they were brought from the sky; their pages were disassembled and reassembled into books on earth after they got wet crossing the seas. Pemɒ́ learned and recopied these books in their homes or the homes of their teachers, but they only activated the texts by reciting them during a ritual in the context of ritual fields,
dʑɔ̏, populated with
ŋk’v̩̏, sculptural spirit bodies. In the field, the most important spirit bodies often sat in house-like canopies, 
tʂ’ə̀ k’ə, built of leaved tree branches. Some species of spirit were embodied not in one ŋk’v̩̏ but in a cluster of twelve or more. Many ritual texts begin by constructing the ritual field in verse. Here is the opening of a text titled 



vá̠ p’ɔ̏ p’v̩́ nu ɲè, “Offering the ancestors a pig’s scapula to hear the genealogy”, recited at the némv̩̏ feast for the ancestors (CYZRZ 2007, p. 1):
1.
to the canopies come a crowd of ŋk’v̩̏
tʂ’è k’ə lè ŋk’v̩̏ tsò̠
2.
to the canopies come a field of ŋk’v̩̏
tʂ’è k’ə lè ŋk’v̩̏ dʑɔ̀
30The field for the ancestral feast featured five spacious canopies, built to house ancestral bodies – a form of ŋk’v̩̏ – consisting of paired human figurines nestled in hollow peach logs. “A pemɒ́ comes to sacrifice; a pemɒ́ arrives to speak”, the text continues. Then:
3.
put up a fence around our pasture
ɬó̠ ɤa̠ də̀ ŋo té̠
4.
build a dike around our grain field
ŋo te də̀ və̠́ nd’ù
5.
frame a house on our house lot
dzù lè lɯ́ hè tsò̠
31The text constructs the boundaries and margins of the field – a fence, a dike, a house lot – and repeats the common metaphor that canopies are houses. The verses that end this section reaffirm the clear parallel between the circumscribed page with its text and the fenced and diked field with its ŋk’v̩̏:
6.
the ɲɯ́nè lineage comes to order the genealogy
ɲɯ́ nè lè ɲè xɔ́
7.
the craftsman pemɒ́ comes to build ŋk’v̩̏
kɯ̠́ pe lè tʂó ŋk’v̩̏
- 24 For examples of prominent Nasu and other Yunnan Yi genealogies, see Zhang (1999).
The ɲɯ́nè lineage would have been the lineage of the affine of most recently deceased ancestor being celebrated at this feast, responsible for hiring a pemɒ́ to recite the host’s lineage from a written text. The némv̩̏ feast was the occasion to add names to any genealogy and to reorder it if necessary. Even more than most ritual texts, genealogies were chains of linked and named bodies (Ma 1987a)24. These two verses are in strictly parallel form, ordered around the verb
lè, “to come”, reinforcing the parallel between the written names of the genealogy and ordered rows of the ŋk’v̩̏ of ancestral bodies seated in the five canopies.
When a ritual field was prepared, green leaves were scattered on the ground to substitute for the verses obscured when leaves stuck to the pages of the original texts after they were soaked crossing the seas. On this ground, already conceptualized as a page of text, were built the canopies and ŋk’v̩̏ (Ma 1987a, p. 65). Figure 5 is a diagram of a field for the ancestral feast, appended to a funeral text. The five canopies are rendered as a great mansion with a forecourt, three courtyards, and a rear pavilion (echoing the form of the Nuo chieftains’ house). Sculptural ŋk’v̩̏ line the courtyards. Positions for different categories of living kin are marked out with Né-script graphs in the first circle around the mansion. Diagrams like this were often supplemented by textual diagrams. The central text performed at the ancestral feast, 




Né mv̩̏ mɑ́ nu tɕ’ì su, or “Book for sacrificing animals at némv̩̏”, devotes its final fifty-five verses to working through the ritual field, canopy by canopy. These verses “read” the field as though walking it, describing the spirit each ŋk’v̩̏ embodies and placing each in spatial relation to the other with prepositions such as
dʑí, “below” and 
ɤɯ́bu, “beside” (Ma 1987a, pp. 70-86). Such passages, common in Né-script ritual texts, might also be read as instructions for setting up a ritual field. Texts were copied onto fields as canopies and ŋk’v̩̏; fields were copied into texts as diagrams and graphs.
Figure 5. Diagram appended to a funeral text, outlining the ritual field

The field is surrounded by a perimeter of ŋk’v̩̏ made of paired branches, each pair embodying an unspecified textual verse and an unspecified ghost. A procession of agnates walks this perimeter, each holding an emblem: a book, sacrificial animals, flags, musical instruments, rifles, and ancestral bodies
© Ma 1987b, p. 200
32The diagram in figure 5 depicts the fifth day of the ancestral feast, devoted to divination. Like the scapula discussed above, this diagram too is an image of an implicit theorization of writing. On this day, pemɒ́ planted a row of ŋk’v̩̏ in a circle around the ritual field. Each ŋk’v̩̏ was a pair of foot-long branches stuck upright into the earth. Leaves and bark were stripped from one branch of each pair and left intact on the other. The leaved branch of each pair, Zhang Wenyuan told Ma Xueliang, was a verse of text while the bare branch was a ghost (Ma 1987a, p. 64). That is, one branch was a verse, unread or unreadable, the other was that verse’s meaning or semantic content, unspecified or unspecifiable. In the context of divination, this fundamental pairing was writing and “ghost” rather than writing and “word” since textual marks were encountered first as unspecified forms (the cracks in a scapula) and only then matched with words (the names, intentions, and histories or characters of the ghosts or spirits found to emit those cracks). It was the task of divination to organize the empty form of “ghost” into words.
33This embodied theorization of writing included readers. A pemɒ́ led a procession of male lineage members around the perimeter of paired branches while reciting a divination text. This ritualist can be seen, book in hand, at the bottom left of the diagram in figure 5. Two men follow leading sacrificial animals. A long line of figures trails behind them. At the rear of the procession are agnates bearing on their backs ancestral bodies being honored at the feast. This line of agnates moves along the line of paired branches like finger and eye moving along a line of graphs. If these are readers they too are composite. Yet unlike the ghost/text forms at their feet, these composites are deliberately specified. Each ancestral body is named on a piece of white silk tucked into its peach; the name of the agnate who carries the ancestral body is written on another piece of silk wrapped around the log; the other agnates are specified by their unique emblems – flags, rifles, musical instruments, and sacrificial animals. If reading gives sense and specificity to the abstract pairing of text (graphs) and ghost (bodily/semantic content), then its first and most fundamental specification is the composite reader: body and emblem, body and name.
34Pages and fields were not metaphors of each other but iterations and extensions of each other. We might understand textuality in general as a rough and contingent correlation among interfolded surfaces of different kinds. In the némv̩̏ feast, graphs on a page were iterated and extended as ŋk’v̩̏ on a field. As text was folded into field, bodies emerged. Bodies of humans, ancestors, ghosts and spirits took material form in flesh, stone, paper, and branches. The page/field form contained these beings and located them in relation to each other: genealogical, hierarchical, temporal, and spatial. The lost verses in the form of leaves scattered on the ground suggest that the interleaving of page and field produced uncertainty and ambiguity as central forces in this iterative ordering of social relations among humans and nonhumans. If reading created rough correlations between populous page and populated field, it also produced rifts in sense or wells of disembodiment out of which other forms and relations might emerge.
35The translation projects that have brought the texts in the indigenous scripts of China’s southwest to a scholarly public are a beautifully consistent example of the modernist image of writing. These projects, which have become a scholarly industry in ethnically diverse regions, sever texts from their local, ritualized, recitational contexts and press them into the service of representation. This display of the texts’ capacity to represent is then crafted into the power to represent the cultural legacies of the southwestern indigenous peoples, on the national model of fifty-five local “minority nationalities” and one encompassing national majority. Each translation into Chinese makes it clear that one script is local, embedded, and vernacular, while the other is an instrument of global knowledge and power and demonstrates that anything done in the former can be represented equally well in the latter.
36In contrast, I have shown here that at the levels of graph, verse, and page Né script forges connections that reach beyond correspondences with language. I have argued that the graphic lexicon of Né script is built around a limited number of foundational forms that move through networks of intersecting series, connected by phonetic similarity or indexical suggestion. As aesthetic forms, these figures can appear as bodies with mutable appendages traveling through any text and connecting it to the core intention of all Né ritual texts: to fix inchoate nonhuman beings in bodies that can be put into relation with human bodies. In verse, Né script used parallelism to craft sense from sequences of exuberantly polysemic graphs. At the same time, verses tore apart the words of ordinary language, bended phrases to meanings they could not acquire outside parallel structures, and created series of substitutions, one graph morphing into several others, each importing traces of other meanings into the verse it temporarily inhabited.
37At the level of the page, Né ritual texts interleaved with the fields on which rituals were performed. Page and field each borrowed and contemplated the form of the other. Né textuality formed a world of surfaces and bodies. Surfaces of different material forms folded into one another; bodies and relays emerged as relays among interfolded surfaces. These correlations were contingent and polyvalent; they did not fully code or fully decipher any surface. Between specifiable bodies and relations were graphs without semantic values, bodies without names, cracks without divined sense, all awaiting willed creative acts to provisionally complete a relay and incise that relay into specifiable, iterable, readable form. At all levels, Né textual practice produced implicit theorizations of what writing was and what it could do.
- 25 One such effort, scholarly rather than popular, is Zhu (2009), written by a Nasu resident of Wuding (...)
38The scholarly industry of translation and the many forms of “culture work” it has spawned in southwest China have done ordinary indigenous people there little good. On the contrary the transformation of textuality into “literature” for the consumption of the nation’s majority have reinforced the widely held sense that the cultural legacies of “minority nationalities” are vastly inferior to that of the Han majority. Even the most extensive collections of Né-script “literature” in translation make it appear poverty stricken in comparison to the rich, varied, and ancient literatures of the language of translation. A better approach, more aware of the actual histories of the colonized peoples of the Yunnan-Guizhou plateau, and more useful to indigenous efforts to reconnect with those histories25, is to understand indigenous textuality in this region as a form of life. Né script is more than graphs on a page. When considered in the context of its creation, recitation, and reiteration, it can be seen as fashioning a distinctive form of life in which human persons and nonhuman meta-persons emerge as textual marks in the interfolded surfaces of a written world.