Abstract
Over two decades ago, Jerold Edmondson and Li Jinfang conducted linguistic fieldwork in the China–Vietnam borderlands, and posited the existence of a migration corridor running between Guizhou province and the northern provinces of Vietnam, passing through the eastern part of Yunnan. This hypothesis was based on linguistic evidence, so they called it the language corridor, and looked at the correspondences between the speech of isolated communities in northern Vietnam and speech varieties in Guizhou, Guangxi and Yunnan. Historic migrations of populations, large and small, bring with them other traces of their original homelands. In the case of the Tai-speaking peoples of the China–Vietnam area, the evidence includes local varieties of vernacular character scripts based on Chinese. This article will present new information based on survey evidence, and propose a new enhancement of established methods in historical anthropology and historical linguistics.
Introduction
In this article I seek to answer the following questions: Is there evidence in the vernacular scripts of the Tai peoples for the language corridor postulated by Jerold Edmondson and Li Jinfang? Can analysis of vernacular scripts assist in the identification of homelands?
The answer to both these questions is yes.
I present here some of the evidence from a survey of the vernacular character scripts of Guangxi, Guizhou, Yunnan and northern Vietnam, as used by peoples such as the Zhuang, Bouyei, Tày and Nùng (Holm, 2013). Vernacular scripts are scripts that are used ‘among the people’. They are as a rule not standardised, and vary considerably from place to place. Knowledge of how to write the script and how to recite documents written in it is handed down within communities by traditional means, from master to disciple or from older family members to the younger generation. The fact that the script varies from place to place makes documents difficult for outsiders to read, but it also means that the script can be used as an indicator of group, sub-group, ethnicity, or place of origin.
The information contained in the script, I would argue, is just as rich as the information contained in the sound system of a local spoken language. In fact, because of the way in which the script is transmitted, local scripts often contain within themselves traces of past social interactions, including local interactions that happened elsewhere if the communities moved from place to place. The script thus acts as a palimpsest and, properly decoded, can be read as a record of past social contacts and interactions. In this respect, vernacular scripts may have certain advantages over spoken languages.
Edmondson and Li: Findings and proposed date
In a ground-breaking article published in 1996, Jerold Edmondson and Li Jinfang analysed evidence from fieldwork on minority languages conducted in southern China and northern Vietnam. Their main focus was on small isolated language communities with small populations. Their main finding was as follows: Close study of the population distribution and dialect geography of some of the small languages in South China and northern-most Vietnam reveals an unmistakable but subtle pattern. This pattern indicates that there was a language corridor as reflected in a string of settlements or groups such as the Gelao, Bouyei, Pa-hng, Bolyu, Bugan, Kam, Sui and possibly Yi (formerly called Lolo) along a line running from the western part of Guizhou province across Guangxi Province along the Yunnan-Guizhou Plateau following the course of the Sông Lô River into Hà Giang and Lào Cai Provinces in Vietnam (Edmondson and Li, 1996: 983).
It is worth noting in passing that most of the groups they mention are speakers of Kra-Dai (Tai-Kadai) languages, and that they identified a starting point in western Guizhou, end points in Hà Giang and Lào Cai provinces in northern Vietnam, and the route in between. We will come back to these details. They noted also that in many cases ‘the main communities of these peoples are otherwise organized’ and that the overall pattern of migration ‘can only be seen by looking at small isolated remnant groups’ (Edmondson and Li, 1996: 983). The groups they looked at included the Gelao, the Kam (Dong), the Pa-hng, the Sui (Shui), and speakers of two Mon-Khmer (Austroasiatic) languages (Bolyu and Bugan). Their primary evidence was mainly patterns of tone splits and diagnostic vocabulary lists, and in some cases they were able to identify the location of originary communities.
They went on to suggest possible dates and causes for this migration: If one considers specific causes for the corridor, then the historical events that come to mind are the Opium War of 1840–42
between Britain and the faltering Qing Empire and also the Taiping Rebellion of 1851. The 1840s were turbulent years … (p. 988)
This explanation was based on testimony from informants, who said that their communities had been in Vietnam for around 150 years. Otherwise, as an explanation, it is not very specific. The authors’ wording in this particular passage makes it clear that they were putting forward a preliminary hypothesis. This was entirely reasonable under the circumstances, since their main focus and main contribution was to document the linguistic phenomena on the ground, namely the startling patterns of correspondences in phonology across vast distances. We will need to revisit these questions further below.
Edmondson went on in a later article to flesh out this picture with more data derived from linguistic fieldwork, supplemented by the use in the field of a laryngeoscope. The phonetic and phonemic data are more precise and plentiful, and he looked again at the Kam data. Whereas in the first paper the Urheimat (original place of origin) of the single surviving Vietnamese Kam speaker had been identified as the Congjiang 从江 area in the far southeastern corner of Guizhou, this time the identification was with the Shuikou 水口 survey point in Liping 黎平 county, just to the northeast of Congjiang (Edmondson, 1998: 136). He showed also how this correspondence could be demonstrated by a quantitative analysis of shared features (p. 137). Edmondson also looked at data for Sui, Giáy (Yay or Bouyei), and Red and White Gelao. His informant for Giáy came from Lào Cai, and Edmondson showed that his speech matched up most closely with the Bouyei dialect survey locations in the extreme southwest of Guizhou (p. 145).
In all of this, the underlying assumption seems to have been that the migration corridor was unidirectional, going in a southerly and westerly direction, and that the destination was ‘the more tranquil territory of Vietnam’ (Edmondson and Li, 1996: 983). This conception needs to be reconsidered. First of all, there is good evidence that migration was two-directional rather than unidirectional, at least in some areas and under certain conditions. In other words, some people from the south of China migrated back from Vietnam as well as to Vietnam. There were times when northern Vietnam was ‘more tranquil’, but there were also times when the opposite was the case. Second, there was more than one ‘language corridor’, probably a great many routes, leading to a variety of provinces throughout the northern part of Vietnam. And finally, Edmondson and Li’s studies were concentrated on smaller outlying groups and language isolates for the most part, and one could easily get the impression that larger language groups like the Zhuang and Bouyei were not involved. They were. 1
Date and reasons for ethnic migration
Edmondson and Li postulate a date in the 19th century and possible reasons for the exodus of minority peoples from Guizhou. The dating is based on informants, but the reasons for the migration need to be re-examined. We need to bear in mind that there was more likely to have been a series of migrations, rather than a single migration.
An earlier stimulus to large-scale migration may have been the annulment of the native chieftaincies and imposition of direct rule (gaitu guiliu 改土归流) in many parts of South and Southwest China during the Ming and Qing. A large number of such conversions took place in Guizhou during the Yongzheng 雍正 reign period (1723–1736). These changes were often followed by large-scale influxes of Han Chinese settlers into lands previously occupied by non-Han minority peoples. For example, in the area of Xingyi 兴义 in the far southwestern corner of Guizhou, the Qiannan shilüe 黔南识略 recorded that the population was originally 80–90% Bouyei and 10–20% Yi, but by the late 18th century the Han migrant population was already 70–80% of the total, and the indigenous inhabitants only constituted 20–30%. 2
For that matter, we can hypothesise that the out-migration of minority peoples from Guizhou could well have started soon after the first major incursions of the Chinese state, during the military campaigns and administrative consolidation of the Southwest in the early Ming period.
Of course, when out-migration was caused by expropriation of lands originally occupied by non-Han peoples, migration would have been unidirectional, since there was no home to go back to. In the Zhuang-speaking areas in the western half of Guangxi, however, and in some other areas, the reasons for out-migration were different. Many of these areas remained under indirect rule by native chieftains (tusi 土司) during the Ming and Qing, the farming population were effectively serfs tied to the land, and the chiefly regimes kept inflows of refugees at bay. The most common reasons for out-migration were natural disasters such as droughts and famines – which were seen as life-threatening but temporary – and chaotic conditions caused by warfare among the native chieftains themselves. 3
Evidence from the vernacular scripts
We now turn to evidence from the vernacular scripts of the Tai peoples of Guangxi, Guizhou, Yunnan and northern Vietnam. Scripts travelled, as well as languages. Our evidence comes from a survey conducted by the present writer and Zhuang colleagues, with additional material drawn from reliable published editions of traditional texts. 4 The main groups included in this survey, which assembled data from traditional texts taken from 45 separate locations, were the Bouyei, the Zhuang, the Nong and Sha sub-groups of the Zhuang in eastern Yunnan, and the Nùng and Tày in Vietnam. In a survey of 60 common words, we found striking evidence of graphic continuities across dialect boundaries. Some of these continuities correlate with the boundaries of pre-modern native chieftaincies, but some of them are related to patterns of migration.
The traditional texts in the survey include ritual texts recited by a variety of religious practitioners (vernacular priests called mogong 庅公, ritual masters and Daoist priests), song texts sung antiphonally at traditional song markets, moral homilies, verse narratives and playscripts. 5 The manuscripts themselves are mostly of relatively recent dates, given the custom in Lingnan of burning all the belongings of a deceased person in order to avoid any possible consequences of death pollution. Each succeeding generation of acolytes or sons copies the texts from the master priest, song king or family elder, and usually this is done with meticulous care. The result is that the manuscripts preserve content and even graphic conventions that are much older than the manuscripts themselves. 6
I present the evidence from these manuscripts in the form of maps. The thick black lines are intended merely to draw readers’ attention to the graphic commonalities, and have no function as isoglosses.
It needs to be pointed out that these maps represent an aggregation of data. That is to say, they show only the most common graphs for common words. In vernacular texts, it is most commonly the case that a single morpheme will be represented by more than one character, and sometimes as many as five. Sometimes one representation predominates, and the others are found only in small numbers, but in other cases different graphic representations are found in more or less equal numbers.
The maps thus furnish us with a snapshot that allows us to see some of the most important relationships. For finer analysis, and the identification of the homelands or migration trajectories of textual traditions, we would need to look more closely at the raw, unaggregated data and at whole texts rather than just a sample of 60 common words. 7 So the evidence presented here must be considered preliminary.
Map 1: Here variants of the same graph are found in Tianlin 田林 in northwestern Guangxi, Xichou 西畴 in eastern Yunnan, and in the Thổ (Tày) marriage songs of Cao Bằng and Lạng Sơn. It is significant that the Xichou community were Nong 侬 speakers – that is, speakers of a Central Tai (CT) dialect – and yet exhibited many features of Northern Tai scripts in their vernacular writing, mostly connected with scripts in Guizhou. This particular character is given a Southwestern Mandarin (SWM) pronunciation in Guizhou. 8 Aen, the classifier for objects, has a short vowel, and this graph is usually used in the central part of Guangxi to represent words with a long/a/. In Vietnam, this graph in the Tày script is usually interpreted as a borrowing from Vietnamese Chữ Nôm, in which 咹 is used to represent ăn ‘to eat’. Vietnamese ăn ‘to eat’ and Tày ăn clf. for objects are homophonous, or nearly so, so this interpretation is at least plausible. However, this character is found in standard Chinese dictionaries Modern Standard Chinese (MSC) è, ‘to eat; stop’, or àn ‘to stop’ (of sound)), and is not unknown in Chinese texts. Also, the evidence from further north furnishes us with an alternative explanation.
Aen (Ɂan1) classifier for inanimate objects (map based on Holm 2013:90)
Map 2: Here the continuum runs between Zhenning 镇宁 in west-central Guizhou down as far as Xichou in eastern Yunnan, and also further eastwards and down along the Hongshui River as far as Mashan 马山 and Du’an 都安 in central Guangxi. Readers will notice that there are two other graphic systems which are continuous across the China–Vietnam border. One is 欧 MSC ōu (part of a surname Ouyang 欧阳) and variants, which are found in Nùng and Tày texts in Cao Bằng and Lạng Sơn but also throughout various parts of Guangxi, including the southwest, the south, west-central Guangxi and the northeast. The other system, based on 幼 MSC yòu ‘infant’, is found in the marriage songs of Cao Bằng and in former Yishan 宜山 county in the northwest.
9
Aeu (Ɂau1) ‘to take’ (map based on Holm 2013:100)
Map 3: The continuum here runs between Zhenning through Ceheng 册亨 and down to Xichou. In central Guangxi the graph 拜 MSC bài ‘make obeisance’ is usually used for syllables with a long /a/ vowel, and the usage here is a good indication of a reading based on SW Mandarin, which as we have noted lacked the long vowel–short vowel distinction. Readers will also see other graphic signs on both sides of the China–Vietnam border. Other examples of a Zhenning–Xichou continuum are found in the data, such as bak ‘mouth’.
Bae (pai1) ‘to go’ (map based on Holm 2013:110)
Map 4: These particular graphs are confined to areas in Guizhou and eastern Yunnan where SWM readings predominate. The graph 垻 MSC bà ‘dam’ is a simplified form of 壩 bà, and has pre-modern readings such as Early Middle Chinese (EMC) paɨʰ/pɛiʰ: that is to say, it is not an Entering Tone (入声 rusheng) character. Its use here to represent a word with final /-k/ in Zhuang and Bouyei is characteristic of Guizhou and northern Guangxi, where manuscript readings of Chinese characters in vernacular texts are based mainly on SWM. The character 埧 MSC jù is simply a graphic look-alike.
Bak (paːk7) ‘mouth’ (map based on Holm 2013:134)
Map 5: One of these characters is 觉 MSC jué ‘to perceive’, found in Tian’e 天峨 county in northern Guangxi, and the other is 掌 MSC zhǎng ‘to take hold of’, found in Xichou in eastern Yunnan. The characters here just resemble each other graphically, and are based on 掌 zhǎng, read as cam (ɕaːm1), with final /-ŋ/ representing /-m/, as is frequently found with SWM-based systems.
Cam (ɕaːm1) ‘to ask’ (map based on Holm 2013:164)
Map 6: Here the graph 的 MSC dì is found representing dwk ‘to hit’ in what seems to be a line running from Ceheng in southwestern Guizhou down through Funing in eastern Yunnan to Nùng in Cao Bằng province. Readers will note it is also found in Donglan 东兰 county in northwestern Guangxi and in Qinzhou 钦州 in the far south of Guangxi.
Dwk (tɯk7) ‘to hit’ (map based on Holm 2013:252)
Map 7: Here the graph 改 MSC gǎi ‘change’ is used for the generalising classifier gaiq in Tianlin and in Cao Bằng Tày. Readers will also notice that there is another location for this usage in Shanglin, in the central-eastern part of Guangxi. This would seem to be a very old usage. It is also found in Vietnamese Chữ Nôm, representing cái, a classifier with a range of meanings similar to those of gaiq in Zhuang.
Gaiq (kaːi5) ‘clf.; ‘this, that’ (map based on Holm 2013:276)
Map 8: Here the linkage is between Tianlin in the northwestern part of Guangxi and the Tày marriage songs of Cao Bằng. This morpheme in Zhuang is a Han loan from 讲 MSC jiǎng ‘to speak’, but it is frequently written with Chinese characters or vernacular graphs unrelated in meaning but having roughly the same pronunciation. Q is a vernacular graph composed of the ‘mouth’ radical and 巷 MSC xiàng ‘laneway’ (the latter probably a substitution for 港 MSC gǎng ‘harbour’ or some other homophonous graph in the same graphic-phonetic series).
Gangj (kaːŋ3) ‘to speak’ (map based on Holm 2013:286)
Map 9: Here the graphic continuity stretches between Tianlin in northwestern Guangxi and the Tày scripts of Cao Bằng and Lạng Sơn. In general, the Tày vernacular script has only about 30% in common with Zhuang scripts on the other side of the border, which also mostly represent CT languages, but there are still signs that the Tày script was in contact with Zhuang and Bouyei.
Gonq (kɔn5) ‘before, in front’ (map based on Holm 2013:296)
Map 10: Here the commonality is between Zhenning in west-central Guizhou and Jingxi 靖西 and Napo 那坡 counties in the far southwest of Guangxi province; the survey point in northern Vietnam is Nùng from Trà Lĩnh district in Cao Bằng province. Readers will note that a similar graph is found also in southern Guangxi. The southern locations are areas where the majority of the population is Nong-speaking areas (i.e. speakers of Southern Zhuang or CT dialects). However, it is significant here that Bouyei migrants are found in scattered districts across Jingxi, Napo, and other southern counties in Guangxi.
10
Gou (kou1) ‘I, me’ (map based on Holm 2013:304)
Map 11: Here the commonality is between two locations in central Guizhou – Anshun 安顺 and Guiyang – and Xichou in eastern Yunnan. The graphs used are different – 哥 MSC gē ‘older brother’ in the case of the Guizhou examples, and 呵 hē ‘to shout loudly’ in Yunnan – but in each case 可 MSC kĕ ‘able’ is the common graphic element.
Guh (ku6) ‘to do’ (map based on Holm 2013:312)
Readers will notice that there is another system found on both sides of the China–Vietnam border on the Figure 11 map: this is one based on 乞 MSC qĭ ‘to beg’, found in various graphic combinations in Jingxi, Napo and Debao 德保 counties in southwestern Guangxi and Tày and Nùng texts in Cao Bằng and Lạng Sơn provinces in Vietnam. These graphs represent a CT etymon hat7 ‘to do’, and are based on a pre-modern or dialect reading of this character: EMC and Late Middle Chinese (LMC) khit, or for example Tiandong Pinghua hɐt33, Hán-Việt (HV) khất. 11
Map 12: The connection here is between Zhenning in Guizhou and the Tày marriage songs of Cao Bẳng and Lạng Sơn provinces in Vietnam. However, it will be noted that 來 MSC lái ‘to come’ is found as a way of representing this morpheme that is found widely in central and east-central Guangxi, as well as in Libo 荔波 in southern Guizhou, so the equivalence here may be accidental, or may have come from elsewhere.
Lai (laːi1) ‘much, many’ (map based on Holm 2013:426)
Map 13: The connection here is between Zhenning and Xichou in eastern Yunnan. The same character is found representing lawz ‘which?’ in Libo in southernmost Guizhou and Tian’e, Huanjiang and Du’an in north-central Guangxi. In all these areas SWM readings predominate. The character 勒 MSC lè ‘bridle’ for lawz (lɐɯ2) must be a reading based on SWM, such as Liuzhou lǝ31 (Liu, 1995: 139), since Middle Chinese readings are EMC lǝk and LMC lǝǝ̆k, and Pinghua has pronunciations like lɐk33 (Tiandong), all with final /-k/.
Lawz (laա2) ‘which?’ (map based on Holm 2013:446)
Map 14: Here, there is a graphic continuum extending across the northern border of Guangxi down to Xichou, but also evidence of a large, well-entrenched graphic system based on 眉 MSC méi ‘eyebrow’ extending right across north and central Guangxi and down into Cao Bằng and Lạng Sơn. This graph is widely used in the Tày vernacular script (Chữ Nôm Tày) for mì ‘to have; exist’ in other northern provinces of Vietnam as well.
Miz (mi2) ‘to have; exist’ (map based on Holm 2013:504)
Altogether, it is surprising how many of the graphs for our list of 60 common morphemes demonstrate long-range connections between Guizhou (especially southwestern Guizhou), eastern Yunnan and northern Vietnam. We have listed here 14 out of 60, which is already close to 25%. There are other patterns of long-range connections in the data, however, between the central-western and southwestern parts of Guangxi and the scripts in northern Vietnam.
Wider issues
The migrations out of Guizhou are only part of a wider story. In Guangxi, I have discussed the dates and the reasons for migration and mobility among Tai-speaking peoples in a series of articles. These indicate that there were periodic bouts of out-migration from village communities in Guangxi and Guizhou. Out-migration took place in response to military campaigns, droughts and other life-threatening disasters, but there was also back-migration once the extreme conditions had abated. Some migration was family-based, but some was migration involving whole villages, following village-level decisions to move. 12
Our evidence from the script suggests that the language corridor documented by Edmondson and Li was only one of many. The evidence from the vernacular script suggests that there was movement also from many areas in the western part of Guangxi down into provinces in the north-eastern part of Vietnam, such as Cao Bằng and Lạng Sơn. Quite a few of the connections highlighted by the graphic continuum suggest routes from Guizhou through the eastern part of Yunnan and down into Cao Bằng and contiguous provinces.
The fact that people took their scripts with them when they migrated reminds us of a basic fact. The language corridor was a corridor not just for languages, but for entire human communities. The human beings in these communities brought their families and cultures with them, as well as their languages. Depending on whether they were refugees or migrating for other reasons, they would also bring as many of their essential possessions as they could manage, including tools, bedding, domestic animals, and so on. So, in a sense, ‘language corridor’ could be a problematic term if it gives the impression that languages could travel on their own.
There is also a fundamental question of directionality. How we understand this basic question will influence the ways in which we imagine and hypothesise the past social history of this region. Migration ‘down’ into Mainland Southeast Asia – mainly Vietnam but also Laos and Thailand – is now reasonably well- known. Was there also migration in the other direction? And if so, at what times, in what numbers, and so on. And how did the refugees know where to go?
My answer is that there is some evidence for south–north migration as well as north–south migration, and also some evidence for reiterative movement back and forth across the border. 13 These patterns of movement are connected with a cultural complex among the Tai-speaking peoples of Guangxi and Guizhou (primarily the Zhuang and Bouyei, but also other groups). One could refer to this phenomenon as the Jiaozhi shuttle, the name Jiaozhi 交趾 referring to the old Chinese name for the administrative region that is now northern Vietnam.
By the time of the Ming dynasty, if not earlier, Jiaozhi was a destination for refugees, local commoner families fleeing unrest. In the lyrics of the ‘Brigands’ Songs’ (fwencaeg) of Pingguo 平果 and Tiandong 田东 counties, Jiaozhi (Geu in Zhuang) is mentioned in parallel with Bo’ai 剝隘 (Aiq), an important market town and staging post in easternmost Yunnan. 14 The lyrics of the Brigands’ Songs devote an entire section to recounting the process of refugees leaving their homes: casting divination blocks to decide whether to flee to Jiaozhi or to Bo’ai, strapping children onto their backs, putting grain into bags, putting the chickens into bamboo cages, and abandoning their fields and villages and setting out on the road. The songs describe hardships along the way: finding places to stay overnight, sleeping in ditches and potholes, and finding shelter from the rain. They also mention a number of places along the way that served as staging posts, and recount how long the journey would take. 15 The songs talk about this as a temporary expedient, with families setting up business in Jiaozhi but planning to return to their home as soon as conditions permitted. This strategy is mentioned explicitly in the lyrics of the Brigands’ Songs: 16
Of course, the fact that the northeastern provinces of Vietnam were predominantly Tày-speaking made such sojourns less dangerous and more amenable, since there were commonalities in culture and linguistic milieu, with a substantial percentage of common vocabulary (at least 60%). Zhuang-speaking refugees were able to find ways of communicating with the local Tày population fairly quickly (Hoàng Văn Ma et al., 1974: 7). Until very recent times, the population of ethnic Vietnamese in these areas was quite small. At any rate, the annual recitation of these songs at traditional song fairs in the Zhuang- and Bouyei-speaking areas of Guangxi and Guizhou kept the memory of these possibilities alive, and thereby also rehearsed for local people what they too might have to do if circumstances required.
Tales about fleeing to Jiaozhi are also found in other narrative and liturgical texts among the Zhuang and the Bouyei. In the Hanvueng (Holm and Meng, 2015), a long narrative text recited in some areas at the funerals of people who died violent deaths, the Goose King, dispossessed of his realm, flees to Jiaozhi with his entourage, builds up a following, and prepares to return and take back his kingdom. Versions of this tale current among the Bouyei in southern and central Guizhou include the Jiaozhi episode. 17 It seems that in Guizhou, too, there was an historical memory, current among non-Han peoples, about Jiaozhi as a place of possible refuge in times of crisis. Just how widespread this was, and whether it was current also among the folklore, seasonal songs and ritual texts of peoples such as the Gelao, the Miao, the Yao and the Yi (Lolo), as well as the Bouyei, is something that would be worth exploring.
Surviving evidence suggests that there developed a culture of migration among non-Han communities in the southern provinces of China, to and from Jiaozhi. Songs and rituals not only mentioned such migration, but kept knowledge of it alive in communities far from the border. In numerous ways, the discourse current among such communities reminded people of the possibility of leaving home if the need arose, and the routes and means they should take to arrive safely at their destination.
Evidence from the script survey, as well as documentation on the diversity of dialects spoken in the southern counties of Guangxi and eastern Yunnan previously mentioned, also point to another important aspect of migration, which is that many migrants and refugees from Guizhou and other points north often settled in southern districts of Yunnan and Guangxi, rather than continuing on to Jiaozhi. This accounts for a good part of the linguistic and graphic diversity in those areas. 18
Homelands
Close analysis of the vernacular character scripts of migrant communities in the northern part of Vietnam can be used to help identify the regions from which the communities came. What I have found thus far in my current fieldwork and subsequent analysis of vernacular texts among various Nùng groups in northern Vietnam – Nùng An, Nùng Giang, Nùng Fan Slihng and Nùng Cháo – is that the vernacular scripts used by these communities are still by and large the same as those current in their original homelands: Long’an county in the case of the Nùng An, Jingxi county in the case of the Nùng Giang, Wancheng in Daxin county in the case of Nùng Fan Slihng, and Longzhou in the case of Nùng Cháo. Detailed documentation of texts from migrant communities in Vietnam and home locations in Guangxi is currently under way. The picture that is beginning to emerge is somewhat complicated, because religious practitioners or scribes in the migrant communities – the men who wrote the manuscripts we study – clearly interacted linguistically and culturally with their new surroundings, and gradually adopted some of the graphic usages current in their new environment – in the case of northern Vietnam, interacting mainly with Tày communities and adopting Tày graphic usages.
In one group we have recently investigated, the Ngạn people of the eastern districts of Cao Bằng, we have evidence that they arrived in Cao Bằng some 400 years ago, when their forebears were recruited by the Mạc 莫 royal court, then located in Cao Bằng, to serve as bodyguards for the royal family. The Ngạn are speakers of a Northern Zhuang (Northern Tai) dialect, and their ancestors came from the Tianyang 田陽 area in western Guangxi. 19 Preliminary analysis of the script employed by their vernacular priests indicates that this script is still 70% identical with the vernacular scripts used in the Tianyang area. Some of the other 30% results from borrowing graphic usages of other peoples in the area – Tày, Nùng and Vietnamese – and some of it results from local innovation.
In other cases, where migrant communities have moved down into Vietnam stage by stage, rather than all at once and straight through (as a result of military deployment), what we find when we examine the vernacular scripts they use is an even more nuanced picture, in which graphic borrowings have evidently taken place at each location in which the community has stayed for any length of time. Detailed demonstration of this point must await further studies, but meanwhile one may note that this is in line with the evidence I have already published for mixtures of dialect renderings in Zhuang texts, and graphic renderings imported from elsewhere. 20 Using the data from the survey of the Zhuang script, which is being continually supplemented with new data from new locations, we can expect in future to get a better overview of the places through which migrating groups may have travelled, which can then be tested through investigations in the target areas, including linguistic and anthropological fieldwork.
This raises the possibility in future of adding data from vernacular scripts to the existing arsenal of analytical tools and methods for documenting and exploring mobility and migration pathways of human populations through history.
Footnotes
Acknowledgements
The author gratefully acknowledges the long-term support of a good number of Zhuang colleagues in Guangxi and elsewhere. In particular, I should mention Professor Meng Yuanyao of the Guangxi University for Nationalities and the late Mr Ling Shudong, formerly Deputy Director of the Zhuang Ethnological Museum in Jingxi County, Guangxi. In Vietnam my chief research collaborator has been Professor Trần Trí Dõi of the National University for Humanities and Social Sciences Hanoi.
Declaration of conflicting interests
The author declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
1.
In fact Edmondson himself conducted an examination of the Nùng An people in Quảng Uyên district, Cao Bằng, a major sub-group of the Nùng and speakers of a Central Tai language, and postulated that Nùng An speakers had come from Long’an 隆安 county in Guangxi. See Edmondson, 2002. On the general language situation in northern Vietnam, see Edmondson and Gregerson, 2007.
2.
Qiannan shilüe juan 27, p. 223 of the 1991 edited edition. This work, compiled by Aibida 爱必达, who was then serving as the Governor (xunfu 巡抚) of Guizhou, is a source that dates from the 14th year of the Qianlong reign period (1749), but contains later additions. The same source notes that after the Miao rebellion in the 2nd year of the Jiaqing reign period (1797), ‘the indigenous Miao people grew daily fewer, and migrant guest households grew daily more numerous. The guest people already occupy 7 or 8 out of every 10, and the indigenes occupy only two or three out of ten’ (土著之苗民日耗, 流寓之客户日增。客民已占十之七八, 土民仅居十之二三). See Qiannan shilüe, 1991 edn, p. 226.
3.
On famines, see Holm, 2010: 14–16.
4.
Published in Holm, Mapping the Old Zhuang Character Script (2013). For further detail about each example discussed below, see the relevant section in this work.
5.
For details such as title, location, manuscript type, number of lines and source, see Appendix 1, ‘List of Texts Included in the Survey’, in Holm, 2013: 801–812.
6.
The manuscripts also embed many pre-modern readings of Chinese characters. Most of these, in the central part of Guangxi, are readings corresponding to Early Middle Chinese (EMC). For correspondences with Old Chinese, which are admittedly much less frequent, see Holm, 2017.
7.
However, preliminary tests were conducted on manuscripts from Donglan and Tianyang-Bama, comparing the results of calculations based on our selection of 60 common words with calculations based on the whole manuscripts. These tests indicated that for script typology, at least, or for calculations of degrees of commonality between different survey points, the calculations based on our 60 common words were very close to those based on whole manuscripts. For the Donglan and Tianyang-Bama manuscript material, see Holm, 2003 and 
.
8.
On the general characteristics of SWM readings of Zhuang vernacular characters, see Holm, 2013: 42–44. The feature relevant here is that SWM typically lacked the contrast between long and short vowels found in Middle Chinese and other southern dialects such as Pinghua and Cantonese.
9.
For discussion of pre-modern readings of these graphs, modern dialect readings, and correlations with Hán-Việt (HV), the official Vietnamese system for reading Chinese, as well as other information, see the relevant sections in Holm, 2013.
10.
Holm, 2010: 9–14, 33. For the location of the Zhong 仲 communities in Ningming, see the map on page 10; for the location of the Yui in Jingxi, see the map on page 12.
11.
Pinghua readings are taken from Li Lianjin, 2000.
12.
See Holm, 2009 and . The analysis presented in later writings was based partly on these studies.
13.
Readings from southern dialects are found in vernacular texts from the Tianyang area, where northern dialects are spoken. The evidence is reviewed in Holm, 2015: 16–19, and the historical context discussed in Holm, 2015: 20–21.
14.
West and upstream from Bose 百色 on the Youjiang 右江 river. See Luo Hantian, 2009.
15.
Luo Hantian, 2009: 43–47. Places mentioned are Runghsae (Nungxi O西, unidentified), Rungh-ndoeng (Nungdong O檂, probably present-day Xin’anzhen 新安镇 in Pingguo 平果), Rungz’an (Long’an 隆安), Giuzleih (Qiaoli 乔利 in Mashan 马山), Bingmax (Pingma 平马 in Tiandong 田东), Couloeg (Zhoulu 周鹿 in Mashan 马山), Baksaek (Bose 百色), Hawgiet (Jinjie进结 in Tiandeng天等), and Yiengfouz (Xiangfu香福, present-day Longmingzhen 龙茗镇 in SW Tiandeng 天等, not far from the border with Cao Bằng).
16.
Nong Minjian and Tan Zhibiao, 2005: 258. The original manuscript text is on the left, and a transcription into Zhuangwen 壮文, the official romanised script, is on the right. The text here is reportedly based on a manuscript in the possession of Tan Shaoming 谭绍明, of Taipingzhen 太平镇 in central Pingguo 平果. A photo-reprint of this manuscript is not included in Nong and Tan; rather, the original manuscript text has been re-transcribed. In any case it is the content of the lyrics that is at issue here, rather than the graphic details. Similar lines are found quite widely in other manuscripts. Other texts mention other goods that might be peddled, e.g. candle wax (Zh. laebcuk) and incense (Zh. yieng): see Luo Hantian, 2009: 42–43.
17.
See Holm and Meng, 2015: 25–26. There was also migration upstream and downstream between the Bouyei and Zhuang-speaking highlands of Guangxi, so another possibility might be that there are elements of this narrative which have been borrowed from areas further south.
18.
On this linguistic diversity see Holm, 2010: 32–58.
19.
Holm, fieldwork, Cao Bằng province, February 2015 and August 2015.