1/ The Hakka or “Guest People”
Dialect as a Sociocultural Variable in Southeast China
This essay is concerned with developing the proposition that the linguistic diversity of Southeast China was a variable influencing the social organization of the region. That differences in language were used to differentiate social groupings will be demonstrated through a study of the interaction of Hakka-speakers in Guangdong and adjoining areas of Guangxi with speakers of Cantonese.
An anthropologist may point to several features of Southeast China which marked off that region from the remaining territory occupied by the Han Chinese. In so doing, he may choose to maintain a synchronic perspective, but yet insist upon a flexible baseline so as to enable a sufficient documentation of any points developed. This approach has been adopted by the one work attempting an anthropological analysis of Southeast China as a whole. Maurice Freedman’s Lineage Organization in Southeastern China is based upon material some of which is “avowedly sociological,” but of which “a great part . . . derives from a very heterogeneous collection of books and articles,” involving “jumps in space and time” (1958:vi). Drawing upon nineteenth-century and pre-Communist twentieth-century data, Freedman presents a model for the social organization of Guangdong and Fujian. The present essay takes Freedman’s monograph as a starting point, and seeks to add another dimension to the image of southeastern Chinese society which has already been presented.
Freedman (1958:2–9) was interested in establishing for Southeast China the significance of the localized and stratified patrilineage in the ordering of interpersonal relations and the constituting of social groupings. The localized lineage certainly was one of the elements of the Han Chinese sociocultural repertory that saw a great development in the Southeast, and a manifestation of this was the common occurrence of villages inhabited by members of one lineage only.
In regards to regional differentiation, Southeast China possesses yet another distinctive feature. This is the very great number of mutually unintelligible tongues to be found in the area. Guanhua (Mandarin), says R. A. D. Forrest:
has largely replaced the original local dialects all over China north of the Yangtze River. But the provinces of the South and Southeast, from Kiangsu [Jiangsu] to Kwangtung [Guangdong], have maintained intact their local forms of Chinese, sometimes called the “dialects” par excellence. (1948:10)
Apart from the Wu dialects, spoken in the vicinity of the mouth of the Yangtze, Forrest distinguishes “three main groups of dialects: Cantonese, Hakka, and the Fukien [Fujian] dialects” (1951:673). Hakka and Cantonese are primarily confined to Guangdong, although there are extensions of both into Guangxi. Of the Fujianese group, the Shantou (Swatow) and Chaozhou dialects are also spoken in Guangdong, in the extreme North (ibid.:673–74).
We may thus note the interesting fact that Fujian and Guangdong, the very two provinces where “the coincidence of agnatic and local community . . . appears to have been most pronounced” (Freedman 1958:1), was also a region displaying considerable linguistic heterogeneity. But the presence of several dialects in an area of considerable size in itself does not indicate that interaction between different linguistic populations was significant in terms of group formation and constitution, i.e., in terms of social organization. Additional bits of information, however, bring out the point that the linguistic diversity of Fujian and Guangdong was carried down to the local level, and that many of the inhabitants of the two provinces may have found themselves frequently dealing with people speaking dialects other than their own. Forrest finds that the Fujian “dialects,” though “sharply defined against the two great dialects of Kwangtung [Guangdong],” nevertheless do “differ between themselves.” In addition to the Xiamen (Amoy), Shantou, Chaozhou, Jianyang, and Jianning dialects of the Fujianese group, there are others, including the dialect of Fuzhou City, which is ‘said not to be understood further than some forty miles from Fuchow [Fuzhou] itself” (Forrest 1951:675–76). And when Forrest tells us that in Guangdong “there is no clear boundary between Cantonese and Hakka, as the two peoples tend to occupy separate villages in the same areas” (1951:674), a linguistic fact takes on sociological significance.
The scanty Western-language material relevant to a study of the interaction between the different linguistic populations of Southeast China has, for a variety of historical reasons, almost exclusively dealt with the settlement of the Hakka among the Cantonese, and the subsequent conflicts that developed between the speakers of these two dialects. Indeed, it was the proportions which these conflicts reached during the mid-nineteenth century that initially stimulated Westerners to note and publish facts on the Hakka and Cantonese, their differences, and their disputes. Undoubtedly, a careful examination of Chinese-language material will yield much data on the significance of linguistic differences in the social configuration of Fujian, and in those parts of northern Guangdong where the interaction was among speakers of different Fujianese dialects or between them and Hakka and Cantonese. But the utilization of Chinese sources dealing with local conditions in all of Guangdong and Fujian would require a research effort beyond that demanded in an essay of this sort. The Western material relating to Hakka-Cantonese interaction has within it information sufficient to suggest the major outlines of a more restricted study. I have therefore utilized data drawn from Chinese sources only to amplify points which otherwise would be inadequately covered, and to provide a historical introduction. While this essay is thus a case study of the interaction between two linguistic populations only, it should be noted that Hakkaspeakers were intermixed with Cantonese over wide areas of Guangdong (as well as adjacent parts of Guangxi). A significant part of Southeast China is thereby involved. The full areal extent of Hakka-Cantonese interpenetration will emerge in the following section.
It would be best to conclude these introductory statements with a brief summary of those parts of Freedman’s work that relate to the present essay.
Freedman (1958) considers that southeastern Chinese society in the main was given its configuration by the interplay of two modes of organizing relationships. Individuals would be members of landed and stratified patrilineages where there would be access to corporate resources, and where the lineage would operate as a unit for either the defense or the expansion of its corporate holdings and perhaps also the private holdings of lineage members. Given the high degree of localization and nucleation, often resulting in the lineage being congruent with the village, political and economic differentiation would take place within the context of organized communities of agnates, and the advantages offered by membership in a corporate group would be unequally distributed. This indeed was a reflection of the fact that the kin-based lineage was immersed in a larger society, one with strong socioeconomic class differentiation.
All of China was characterized by a gentry-peasantry dichotomy, in which the possession of wealth, mainly or ultimately in the form of land, would enable one to prepare himself or his children for eligibility in the state bureaucracy. And between those eligible for participation and the local representatives of state power there was a mutual bond involving membership in the same socioeconomic class. This bond was manifested in the monopoly which the gentry of any region would possess over access to the local representatives of the state. In regard to the southeastern Chinese lineage, class differentiation thus had a functional value, for it meant that there would be individuals who could represent the corporation in dealings with the state. But the gentry also represented the state in dealing with the lineage, and this, plus the position of the gentry as landlords and as manipulators of corporate wealth, led to a continuing tension between the gentry and the peasantry. Antistate activity took the form of participation in secret societies, the membership of which crossed lineage lines. This indicated both the inability of the lineage to fully defend its corporate interests against the state and the limitations of kinship bonds in an economically differentiated society.
It can be seen from the above that the model Freedman presents for Southeast China involves a balancing of the principle of group affiliation through agnatic kinship against that of group affiliation through participation in statewide class alignments. In this essay I will strive to validate the assertion that associated with the linguistic heterogeneity of Southeast China was the fact that dialect was a third structural variable, a third means of group affiliation.
I
Evidence relating to Hakka-Cantonese conflicts should be viewed in conjunction with the historical record of the Hakka migration to Southeast China from the North and with certain historical developments which were concurrent with it. In this way it will be seen that there were many specific factors involved in the process by which antagonistic linguistic populations emerged in Guangdong and Guangxi.
The most comprehensive discussion of the sequence of Hakka migration is contained in the works of Luo Xianglin (1933; 1950). Both Luo and other writers on the subject (Campbell 1912; Eitel 1873–74; Hsieh 1929; Piton 1873–74) rely heavily upon data drawn from the lineage genealogies of groups currently recognizable as Hakka-speakers. The use of such data presents two problems of interpretation: (1) whether such sources are uniformly reliable for the entire period with which they deal; and (2) at what point do these sources relate to the Hakka as a linguistically differentiated population?
Hsien-chin Hu (1948:12; cf. Freedman 1958:7; Chinese and Japanese Repository 1865:282) has indicated that not until the Song dynasty (960–1279) did the lineage begin to assume its present form; not until then did the keeping of genealogies as we know them today commence. The unreliability of pre-Song family records has also been attested to by writers of the Hakka. Luo Xianglin’s (1933:41) plea that genealogies must be used for research on Hakka migrations even prior to the Song “movement to revise the genealogies” is based upon their importance as practically the only source of earlier data, but involves an acknowledgment of their deficiencies. Hsieh T’ing-yu (1929:211) and Charles Piton (1873–74:225) are disturbed by the similar descriptions of origins and migrations which are to be found in the pre-Song portions of both Hakka and Cantonese lineage records. These similarities may in part be accounted for by the fact that the intensified absorption of non-Han groups in Southeast China, following the removal of the Song dynasty to the south of the Yangtze (Wiens 1954:183), occurred during the very period in which there was a general reconstruction of genealogies. The genealogical records produced by newly arriving Chinese might have served as models for the older inhabitants of the region.
As to the original home of the Hakka, there is only the genealogical evidence as a guide, and this indicates that previous to the Eastern Jin (317–420) the ancestors of the Hakka were primarily to be found in Henan (Luo Xianglin 1933:63).
Luo interprets the present distribution of the Hakka as the result of five successive “migratory movements” southward. The first of these began in 317 C.E. in conjunction with the southward flight of the Jin, and extended as far as central Jiangxi (1933:43, 45). Towards the end of Tang (618–907), the campaigns of Huang Chao gave stimulus to a second migration (ibid.:50). Henan was again affected, from which province there were movements to Jiangxi and southern Fujian, while some of those previously settled in northern and central Jiangxi also entered southern Fujian as well as immediately adjoining parts of Guangdong (ibid.:46). Huang Chao was not active in north-central Jiangxi so that the earliest immigrants there remained unmolested (ibid.:46). It is for this region that George Campbell reports that “the t’u t’an [tutan, “local speech”] which we readily identify with the Hakka at first becomes less intelligible than the Mandarin, as we go northward, so we are at a loss how to classify the people” (1912:474). The contrast between this northwestern sector of linguistic gradation and the sharp dialect differentiation which elsewhere marks off the Hakka from their neighbors leads us to suspect that during the migrations at the end of Tang, Hakka was not yet strongly differentiated from the speech of contiguous Chinese. The ancestors of the Hakka at this point were in the forefront of Chinese overland migration, for Xu Songxi has shown that the groups being encountered, and at least to an extent absorbed, by the Hakka in southern Fujian were non-Han (1946:162).
Certainly the genealogical records do attest to the continuity between the late-Tang settlement of southern Fujian and the present Hakka population in that area and in Guangdong. There is even a high degree of specificity: a great many, if not the majority, of Guangdong Hakka appear to trace their ancestry directly to the Fujianese village of Shibi (Shakpiak) in Ninghua County (Nenfa), Tingzhou fu (prefecture) (Campbell 1912:474; Hsieh 1929:216; Luo Xianglin 1933:55–57; Piton 1873–74:222). The material relating to Shibi would seem to indicate a high degree of nucleated settlement involving speakers of the Hakka dialect. This concentration and relative isolation of Hakka-speakers in southwestern Fujian and the concurrent absorption of neighboring non-Han groups extended from the close of the Tang through the Southern Song (1127-1279), and lasted perhaps as long as four hundred years (Hsieh 1929:211; cf. Oehler 1922:351).
Although the main impact of Hakka migration was to be felt later, the Hakka “incubation period” in southwestern Fujian already had begun to show signs of ending during the Southern Song. Tingzhou, in the center of the Hakka district, was described as follows in a memorandum of 1171 recorded in the Fujian tongzhi (Fujian provincial gazetteer): “Ting-chou [Tingzhou] has the most bandit troubles in Fukien [Fujian]. In a period often years, we have been forced to take arms thrice. . . . Many people are out of employment and become bandits” (quoted in Hsieh 1929:212).
Across the border, in that region of northern Guangdong today comprising Meixian and its vicinity, the situation was quite different. There the countryside was comparatively underpopulated, but the inhabitants seemed to have title to all the cultivable lands. Meizhou, the destination of many of the immigrants from southwestern Fujian, was described by the Southern Song scholar Wang Xiangzhi in his Yudi jisheng (The wonders of the world). In Meizhou, Wang writes: “The land is neglected and the people are lazy; farmers are few and the cultivators are all sojourners from Tingzhou and Ganzhou [in southern Jiangxi]; therefore no one suffers from a lack of fields” (quoted in Luo Xianglin 1933:57).
These earlier inhabitants of northern Guangdong were soon displaced or absorbed by a growing stream of Hakka from Fujian. During the changeover from Song to Yuan rule Hakka began to enter Guangdong in great numbers, deriving from the adjoining xian1 of Ninghua, Changting, and Shanghang, with the majority of persons still departing from Shibi (Luo Xianglin 1933:54). Concurrent with the filling up of the entire northern and eastern part of Guangdong with Hakka-speakers was a steady decrease in the population of Ninghua xian (county). The Ninghua xianzhi (Ninghua County gazetteer) records a drop in the xian’s population from 35,000 “families” in 1253, during the Southern Song, to 12,588 “families” in 1391, at the beginning of the Ming (1368–1644), and to 5,279 “families” at its close (Hsieh 1929:219). These figures, even though their accuracy may be questionable, suggest that while the fierce fighting attendant to the Mongol occupation of the region in 1277 (Luo Xianglin 1933:50–51) might initially have contributed to the depletion of the population, the presence of emptier lands to the south continued the process.
The census figures also suggest that there was not a steady process of population outflow in conjunction with population growth. What we see, rather, is the result of a prior imbalance: overpopulation within a restricted area and underpopulation elsewhere, presuming equivalent ecological carrying capacity. In addition to the isolating factors posed by mountainous terrain, we may assume that a relatively predictable if diminishing food supply might for a time offset any pioneering tendencies within an area of high population density. In addition, the possibility that the inhabitants of a comparatively underpopulated region might hold and have the means of asserting prior rights to some or all potentially usable lands would to an extent limit the quantity and quality of land available to newcomers. But if the population of a given locale would begin to exceed the carrying capacity of the land, a surplus population would be created, and forced to seek new means of subsistence. The new means might take the form of banditry, or would be sought through emigration. In the latter case pioneers would not only be searching for lands for themselves, but might transmit information back to the home region. (George Campbell notes that from the establishment of a school in Meizhou by a Ninghua scholar during the Southern Song, “both places were known to each other” [1912:477].) The involvement of many individuals in the pioneering activities of a few meant that information about new settlement possibilities could lead to large-scale movements from the overpopulated regions, so that within a short period of time people of diverse origin might be living interspersed with the “natives” of previously sparsely settled areas.
During the very period of the buildup of a large Hakka population in southwestern Fujian, the province of Guangdong was undergoing intensive sinicization. The growing penetration of the Chinese administrative apparatus during the Tang and Northern Song (960-1127) was not initially accompanied by an appreciable increase in the influx of Han Chinese into the area (Xu Songxi 1939:177–78). It was with the transfer of the Song capital south of the Yangtze that the intensive absorption and settlement of the Southeast commenced (Wiens 1954:183).
Now the sinicization at this time of Southeast China, especially Guangdong, set the stage for the encounter of linguistically heterogeneous but Chinese populations. We have already noted that during the Song the development of lineage organization was associated with the spread of Chinese culture. When, at the end of the Southern Song, large numbers of people began to enter Guangdong from the north, they were confronted by a population that had come to possess a cultural repertory quite similar to their own. There were differences, chief among them being dialect. Most of Guangdong was probably now populated by Cantonese-speakers, and it was at this juncture that the inhabitants of the mountainous regions of southwestern Fujian entered Guangdong and became known as the “Guest People”—Kejia or, in its Cantonese pronunciation, Hakka (cf. Oehler 1931:8). In contradistinction, the older inhabitants assumed for themselves the title of Punti (Bendi, “Natives”) (cf. Laai 1950:92). In the emergence of Guangdong’s cultural landscape it can be seen that, at least in the historical sense, the development of localized agnatic groupings and the establishment of different linguistic populations were associated phenomena.
The movement we have discussed has been classified by Luo Xianglin as the third Hakka migration, and its general direction was from southern Jiangxi and southwestern Fujian to the eastern and northern sectors of Guangdong. It was associated with the flight of the Song across the Yangtze and the subsequent conquest of all China by the Mongols (1933:50). By the end of the Yuan, northern and eastern Guangdong were exclusively Hakka regions (ibid.: 58).
The fourth period of Hakka migration is considered to have begun on an intensive scale with the Manchu conquest in 1644. The main route was from northern and eastern Guangdong to the middle portion of the province as well as to its central coastal areas. Hakka-speakers also reached eastern and central Sichuan, eastern Guangxi, and Taiwan. Hakka from southern Jiangxi and southern Fujian as well as from the Meixian region of Guangdong moved into western Jiangxi and southern and central Hunan (Luo Xianglin 1950:34).
The resumption of the southward push was conditioned by a process of population nucleation in northern Guangdong similar to that which had occurred in Fujian. Throughout the Ming and early Qing the Hakka population in northern Guangdong had been steadily expanding (Luo Xianglin 1933:59). There is evidence that prior to the Manchu invasion, population pressure in the Jiaying zhou (prefecture) region (including Meixian) was already forcing considerable emigration and creating a bandit problem. Government officials might expedite the former to control the latter. The Tianxia chunguo libing shu (Merits and faults of all the countries in the world) says that Boluo, in central Guangdong, had
heretofore . . . consisted of the native inhabitants only. The fields were extensive and the people few. . . . During the reign of Kia-Lung [1522–73], there were increasing mountain bandits in the northeastern part of the province, but the district was the least disturbed by the disorders. . . . In the year 1548–49, the people from Hsingning and Ch’anglo [Xingning and Changlou, districts in Jiaying zhou] carried their belongings and came. . . . The local inhabitants objected, but the officials said that these districts were crowded and poor . . . and that it was better to let them stay with the natives. After this . . . others from Ting Chou [Tingzhou] and Chang Chou [Changzhou] in Fukien [Fujian] also came. . . . The natives were weaker and the newcomers stronger. Hence disputes and quarrels arose. (Cited in Hsieh 1929:220)
The middle of the Qing saw Hakka-speakers already spread throughout many parts of Guangdong, and by 1730 large numbers of Hakka were in the vicinity of Guangzhou (Eitel 1873:162). There was now a wide area in which speakers of the Hakka and Cantonese dialects lived interspersed, and the subsequent “history” of the Hakka strongly features violence between these two populations.
During the second quarter of the nineteenth century, the continuing expansion of the Hakka population in the region stretching from the northeast to the southeast of Guangzhou, especially in the vicinity of Sihui, Kaiping, Enping, and Taishan, led to increasingly frequent conflicts with the Cantonese; the most violent battles occurred after 1856 and lasted until 1867, when the governor of the province decreed the establishment of the subprefecture of Chiqi as a sort of reservation where Hakka displaced by the fighting could reestablish themselves.
Agricultural Land and Population Density of Hakka Xian and of Xian which Were the Sites of the 1850–67 Hakka-Cantonese Conflicts
1. Lou Xianglin (1950:53–54).
2. Chinese Economic Yearbook, vol. 1, (1935:F50–F53). A mu equals 0.0667 hectares.
3. Administrative Districts (1947:73–79).
4. Lou Xianglin (1933:3, 62).
Since Chiqi ting, formed out of a portion of Taishan xian, was too small to hold all of the refugees, many went on to the extreme southwestern tip of mainland Guangdong as well as to Hainan Island and overseas. Luo Xianglin sees in the order of 1867 the commencement of the fifth period of Hakka migration, a migration still in progress at the time of his writing (1933:3, 63; 1950:32–33).
The nineteenth-century Hakka migration into the interior of Guangdong was a regional manifestation of a growing pressure of population upon land resources throughout China. This pressure was felt more consistently in Guangdong than in areas such as the Yangtze Delta, where a brief respite was later obtained due to population losses incurred during the Taiping Rebellion (Ho 1959:167–68).
Surveys and estimates made during the 1930s and 40s indicate that differential population density still prevailed in Guangdong, with the northern Hakka zone having less cultivated land per person than the xian surrounding and comprising the Pearl River estuary, the region of the most violent Hakka-Cantonese conflict. If we view internal migrations as a process of adjusting population distribution to the carrying capacity of the land, the attraction offered by the central Guangdong xian becomes clear. (See table 1.1.)
Though Hakka migrations may be said to continue to this day, by the late nineteenth century the spread of Hakka-speakers throughout Guangdong was an accomplished fact. In addition, speakers of southern Fujianese dialects had been settling in parts of the province. The heterogeneous nature of the countryside did not go unnoticed at the time. E. J. Eitel remarks that
the population of the Canton Province is at the present time as mixed as the population of England was some time after the Norman conquest. The first invaders, who now-a-days style themselves Puntis . . . have assumed ownership of the Province, and, on the whole, successfully struggled with the other invaders, two different races . . . who are now-a-days distinguished by the names Hak-ka and Hok-lo (or Ch’aou-chow-men) [Chaozhou]. . . . The Hok-los kept for the most part near the seacoast and the banks of the larger rivers, and did not spread far over the interior of the Province; whilst the Hak-kas . . . spread all over the country. (1867:47)
The net result of the successive southward expansions of the Hakka was that by 1950, of the ninety-seven Guangdong xian, fifteen were entirely Hakka-speaking in composition while an additional fifty were inhabited by Hakka living in some degree of dispersion among speakers of other dialects. In twenty-eight of the latter xian, Cantonese comprised all or part of the non-Hakka population (Luo Xianglin 1950:53–54).
II
The foregoing historical sketch gives an indication of the extent, both temporal and spatial, of Hakka settlement among speakers of other dialects. Available data also allow for an examination of the ways this settlement was carried out among Cantonese speakers in Guangdong and adjoining parts of Guangxi.
Note initially that departures from Hakka districts could be either on an individual or group basis, and that different emigrating groups would be differently constituted. From what informants born in Meixian have told me, it appears that people leaving Hakka districts for other parts of Guangdong would often go by themselves. They might go together with their families, which could be extended to varying degrees. Business associates or friends might accompany each other, and departing groups could also include remote kinsmen, affines as well as agnates. But whether such groups would all settle in the same locale was problematical, for the dispersion involved in emigration might be carried even further by limited settlement possibilities.
The amount and type of land available for use by newly arrived Hakka immigrants could be conditioned by several factors. The most important of these involved the system of land tenure in Guangdong, which was such that, barring state interference, all land, either cultivated, uncultivated, or waste, would already be part of an individual or corporate estate. When the New Territories were still “new,” the British recorded that
all hills and waste lands are claimed by the nearest villages or most powerful clans in the neighbourhood, or even at a distance. Even portions of the sea and the bed of the sea, foreshore, sand beaches and any land whatever which may be turned into use and profit are claimed. (Hong Kong 1900a:17)
According to Chen Han-seng, lands in the making were also claimed by localized lineages in the Pearl River Delta region:
River current and sea tide work daily and incessantly to bring sediments and to form sandy but exceedingly fertile lands; and by protection and the planting of a certain kind of grass thereon, they may be converted into cultivable fields within three years. For such new lands the wealthy clans compete among themselves to pay the taxes, in order that they might claim their ownership. (1936:29)
Eroded hillsides, from which only grass could be procured, might be controlled by either agnatic or territorial groupings. C. K. Yang notes that the poor peasants of Nanching, a multisurname village near Canton, “could not cut grass that was within the boundaries of neighboring villages” (1959:67). In the Hakka areas of northern Guangdong, localized lineages would have claim to all available grasslands. According to W. Bernard Paton, it was “not uncommon for trouble to arise with a neighboring clan over this grass cutting business. Sometimes the cutters are accused of encroaching beyond their rightful boundaries” (n.d.:67).
Thus, as a result of the historical sequence of settlement in Guangdong, Hakka arriving in a region new to them would find all land already under claim. But to hold title to land is not necessarily to obtain agricultural wealth from it. Hakka entrance into Cantonese-held regions was expedited by the fact that much of the land claimed was not under cultivation. Yet the lands available for lease were of a poorer variety, being less accessible to water for irrigation. The pattern of Hakka settlement would in part be determined by the location of the more valuable lands, for these lands would already be farmed by Cantonese-speakers. In hillier zones, highly productive lowlands would of course be interspersed among poorer hillside plots, so that initial Hakka settlement would involve a high degree of dispersion among Cantonese-speakers. This seems to have been the case over a wide area of Guangdong and adjoining parts of Guangxi during the nineteenth century. In the New Territories, the process of Hakka settlement has been described as follows: “They first settled in the narrow mountain valleys, and there high up in the smallest valleys they laid out their rice fields, where there was to be found only a small stream” (Oehler 1931:15).
A similar situation was reported by a late-nineteenth-century traveler among the “hills in central Canton” who said that “through all the region traversed on this journey I found that the Puntis, or native Cantonese, occupy most of the plains and level arable land, while the Hakkas are taking possession of the upper and less accessible valleys” (Henry 1886:273).
Shortly afterwards, H. Compte Meyners d’Estray noted that in six of the xian of Guangzhou Prefecture the Hakka were still outnumbered by the Punti:
Here the Punti . . . have remained masters of the soil. They inhabit generally the fertile valleys and force the Hakka . . . to content themselves with the mountains, where the land is less fertile; there, where the immigrants are less numerous, they pay, ordinarily, the rent of the lands to the first inhabitants of the country. (1890:31)
This situation prevailed in other regions where the Hakka were recent arrivals among Cantonese-speakers. A local observer said that during the early-eighteenth-century settlement of Guangxi by the Hakka, “there was much mountain valley and waste land in [Guangxi]. The Hakka people have been hired to cultivate the lands” (quoted in Laai 1950:94).
The limited lands available for Hakka settlement were not conducive to the immediate clustering of large groups of people at one locale. Statements relating to the initial appearance of Hakka on Cantonese-owned lands indicate that the new arrivals would first settle on a very small group or even individual basis. Meyners d’Estray characterizes the Hakka migration as proceeding “by innumerable little branches who spread out over Kwangtung [Guangdong].” He adds that the Hakka slip into the midst of the Punti “as little families” (1890:32). Later he remarks that they enter Cantonese regions “one by one or by several families at a time” (ibid.:32). That the Guangdong Hakka “live in scattered hamlets or houses” prompted Wilhelm Oehler to conclude that “for this reason they . . . are not as clannish as other tribes” (1922:353).
The association between tenant status and a dispersed settlement pattern was noted in Guangxi. Laai Yi-faai cites the Xunzhou fuzhi (Xunzhou prefectural gazetteer): “Those wealthy people who own large areas of land are quite willing to let [the Hakka] cultivate the land in return for rent. [The Hakka’s] houses are built in the fields far from neighbors” (1950:96).
The “wealthy people” in the above statement presumably were private owners, and not the representatives of larger corporate bodies; yet there is more specific evidence which indicates that Hakka would settle on lands belonging to individual as well as collective landlords. Johann Heinrich Vömel says of the Hakka that while they show a propensity toward becoming roving peddlers and specialized craftsmen in non-Hakka regions of Guangdong, and in this capacity do not as a rule maintain long periods of residence “in a foreign region,” yet,
on the other hand in many places they are in a tenant relationship with the Punti, in whose fields they have settled, so that through the payment of a small sum of rent a kind of dependency-relationship is still expressed, even though they have always been working the field. Or the dependency-relationship will be maintained through the continuing payment by the Punti of the land tax, even though the field belongs to the Hakka, from whom he still collects the [requisite] sum. (1914:598)
Permanent tenure, with the landlord responsible for tax payments, was also found in the New Territories, where it was recorded that there were
disputes between individual landlords, as distinct from clans, and tenants as to whether rent is to be paid to the landlord or to the Government. These disputes having arisen owing to the tenants having confounded the rent due to their landlords and the Crown rent due to the Government. (Hong Kong 1900a:8)
The confusion which the British saw between rents and taxes was due to the permanent tenure which certain farmers held toward the lands they were working. We would want to know how Hakka newcomers could arrive at such a favorable position on Cantonese-held lands. The tenancy system we are dealing with here was similar or identical to the “joint ownership of land” which Chen Han-seng says was to be found in northern Guangdong. There
the landlord . . . often possesses only the so-called liang-tien [liangtian] meaning the land for which tax or liang must be paid; and the tax responsibility is supposed to justify the rent collection. The tenant often owns what is chi-tien [jitian], meaning the soil itself, or the surface. (1936:52)
Chen adds that two theories attempt to account for the origin of joint ownership. One would have the original peasant owners placing themselves under the protection of powerful families to obtain tax reductions. Peasant “gifts” to their benefactors eventually developed into a regular rent, which was then fixed through the landlord’s assumption of permanent rights to the subsoil of the land. The other theory was that “the claim to permanent use of the surface of the land derived from a long period of permanent tenancy.” People who would assume ownership of “cultivable but uncultivated government lands” would lease them out to tenants: “Much cost of development and primary cultivation had to be borne by these tenants; a lease contract therefore became customary under which the landlord could not transfer the tenancy” (ibid.: 52-53).
Of the two theories, it is obviously the latter which would apply in the case of Hakka settlement. For corporately owned lands there is much more direct evidence that newly arrived Hakka stimulated the activation of dormant holdings for which they received permanent guarantees of tenure. The Report on the New Territories at Hong Kong states that
it is noteworthy that the majority of those who bring out-of-the-way plots of land under cultivation are Hakkas. . . . This due to the industry of the Hakkas . . . and also to the fact that the best and most available land had been appropriated by the Puntis before the Hakkas settled in the district. The Hakkas have . . . reclaimed large tracts from the sea, and made many a hillside, hitherto barren, yield good crops. (Hong Kong 1900a:16)
The same source describes very clearly the relationship between the initial cultivation of corporately claimed lands and permanent tenure:
When land is brought under cultivation for the first time, the cultivator . . . applies in the first instance to the clan or village which has taken the land under its protection. Generally, the arrangement with the clan or village results in a lease in perpetuity being made out stating the situation of the land and the amount of rent in grain or local money that has to be paid by the cultivator. (Ibid.:19)
Also note: “The clans have . . . claimed large tracts of land, which they have never occupied, but which they have leased in perpetuity to others, who undertake to bring the land under cultivation” (ibid.:20).
The evidence I have presented allows a reconstruction of the first stage of Hakka settlement of parts of Guangdong and Guangxi already occupied by Cantonese. Initial Hakka settlement was conditioned by the availability of land of a restricted type, which in turn was a manifestation of the differential population density between fairly contiguous areas, which in the past had influenced Hakka emigration. The arrival of Hakka-speakers could lead to the formation of a relationship of benefit both to themselves and their Cantonese landlords. Indeed, if we grant the term “dialect group” sociological significance, the Cantonese-Hakka relationship at this stage may be described as symbiotic. But I have no evidence to justify calling the Hakka-speakers a dialect group. Nevertheless over wide areas Hakka would be obtaining subsistence in return for which they would increase the amount of wealth derived by the Cantonese from already existing holdings. The Hakka tenants lived interspersed among Cantonese-speakers, some of whom were their landlords. We have seen that some of the landlords would be villages or kin-based corporate bodies. The two linguistic populations at this stage had different settlement patterns; this was reflected in a strict linguistic exclusiveness in Cantonese villages, with the absence of Hakka settlement at village sites already developed. Charles Piton notes that
where the Hakka are mixed up with the Punti, they are considered as intruders, and have very often no share in the local idol and must therefore satisfy themselves with worshipping their ancestors, which they do . . . in want of a proper ancestral hall. (1870:219)
The lack of an ancestral hall here reflects tenant status and the consequent absence of individuals of sufficient wealth to endow such a structure (cf. Freedman 1958:47). But Hakka nonparticipation in territorial-based forms of institutional worship indicated their physical exclusion from Cantonese villages (cf. Freedman 1958:81).
III
Though initial Hakka settlement was of advantage to the speakers of both dialects, there is evidence that landowning individuals or corporate bodies might display a degree of hesitation before granting the permanent tenure necessary to induce the arrival of Hakka cultivators. In the early eighteenth century, Ho Ping-ti tells us,
Some conscientious provincial officials repeatedly memorialized . . . that unless security of tenure was guaranteed by the government there was no hope of attracting the surplus population of eastern Kwangtung [Guangdong] to develop the fertile land that was plentifully available in the west-central section of the province. (1959:219)
The concern of the provincial officials was due to an ambivalence on the part of landowners, indicating that the latter perhaps realized the ultimate consequences of permanent Hakka settlement might be very different from the immediate ones. For in any given region, increasing Hakka immigration coupled with an overall increase in population would soon place the hitherto symbiotic Cantonese-Hakka relationship within a very different framework. Numerical increase in the Hakka population was accompanied by an effort to expand and secure, rent-free, the lands which they were farming. Such activities on the part of a population definable in linguistic terms led to the formation by both Hakka-speakers and Cantonese-speakers of groupings the members of which took common dialect as a sign of common interests relating to the securing of agricultural holdings. Moreover, relationships involving mobilization for the defense of aggrandizement of such holdings could be defined on the basis of common or different dialect. Laai Yi-faai recapitulates a development of this sort in Guangxi:
At first the Punti landowners were quite satisfied to allow the Hakkas to cultivate the land in return for rents. Gradually more lands were cultivated by Hakkas who were tenants, and more Punti landowners treated their lands as an investment from which they obtained interest in the form of rents. This process had the effect of consolidating the Hakkas more firmly in the area dominated by the Puntis, for . . . they had to hang together in order to check the exploitation of the Punti landowners. Furthermore, the infiltration of the Hakka population in the Punti area created a situation of constant competition between the Hakka tenants and the Punti tenants. The latter, like the Hakkas, also wanted more land for cultivation. The hatred on the part of the Punti tenants was gradually intensified; and feuds were fought in 1851 between the two groups, as, for example, in the district of Yung-ch’un [Yongchun]. Under these circumstances the Hakka were further drawn together to face opposition of not only the Punti landowners but also of the Punti peasantry. The differences in dialect . . . made the two groups even more hostile to one another. (1950:94-95)
If Hakka-Cantonese conflicts were in fact aligned along dialect lines, it follows that if one or both of the opposing sides consisted of an economically stratified population, resources differentially distributed but relevant to defense would in some fashion be made available to all the speakers of the dialect concerned. Unless this were the case, we would be hard put to affirm the equivalence of common dialect with common economic interest. This in fact did happen in the Guangxi struggles, for we learn from Laai that the Punti landlords soon began to avail themselves of a group of professional fighters, sometimes known as the tuanlian (local militia) (1950:96). Inasmuch as the fighting involved land for rental as well as land for permanent ownership, the actions of the Cantonese landlords aided both their own cause and that of the Cantonese tenants. Piton may have noted the same phenomena in Guangdong, where, he says, “on the outbreak of any local feuding the Hakka takes the field in person” while the Puntis often “hire mercenary troops to do the fighting for them” (1873–74:224).
Since the dispersed settlement pattern of the Hakka precluded structural localized groupings of a large size, the conflicts with the Cantonese in Guangxi necessitated the mobilization of individuals from over a wide area. The Xunzhou fuzhi, after relating how the Hakka live in houses “built in the fields far from neighbors,” adds that “the relationships among them” are
yet . . . extremely intimate. When there are enemies or feuds against the [non-Hakkas], a call for help is answered by the hundreds who, with spears on their shoulders and shovels on their backs fight so furiously that they have a contempt for death. (Laai 1950:96)
If the violent end of a Cantonese-Hakka symbiotic relationship resulted in the expansion of Hakka holdings, the establishment of Hakka villages soon followed. It is for good reason that the dispersed pattern of Hakka settlement does not tally with the way population distribution in general has been described for Guangdong. It appears that the agricultural tasks which engaged the bulk of the people of Guangdong favored residence in nucleated villages. The strategic value of their sites gave villages greater durability than did the agnatic groupings which might at any given juncture comprise then. I will briefly digress from the main theme of this essay to develop this point.
It is commonly recognized that the population of most of the Chinese countryside, including the Southeast, was to be found settled in discrete villages (Freedman 1958:1, 8; Hsiao Kung-chuan 1960:14, 560). This is clearly the case in the present-day New Territories, where “the general pattern of rural settlement is that of nucleated villages” (Tregear 1958:53). In terms of population constitution, however, there emerges no consistent pattern of association between a given village and one or more agnatic groupings. A village, to be sure, might still be inhabited by the agnatic descendants of the village founder. This was the case in the Guangdong village that Wang Xingrui briefly investigated. A single-surname settlement, the village had been founded during the Southern Song by the lineage ancestor (1935:43). Another example is the village of “Pu-lu-wei,” founded by a branch of the “Chin” lineage during the Qianlong period (1736–95) (Chinese and Japanese Repository 1865:283). It was the nearby village of “Ho-au,” however, that was first settled by the Chin lineage, during the Song, and we are told in the same source that
Ho-au was at that time inhabited by two clans of the K’ung and Liang names, but who disappeared in the degree as our number [the Chin lineage] increased. These clans built the Yung-fuh temple, the only public building in and near Ho-au. (Ibid.)
Moreover, other villages were inhabited at different times by different lineages. In the early part of the twentieth century, “Phenix [sic] village,” in northern Guangdong, consisted in the main of individuals bearing the same surname. When the ancestor of this group first moved to the village during the later Ming, towards the end of the sixteenth century, it was already occupied by people from other lineages. These latter managed to persist in the village until as late as the Qianlong period (Kulp 1925:68–69, 292). In the village of “Nanching,” near Guangzhou, “there seems to have been an ecological succession of clans,” and “old villagers said that in the dim past there was a Hua clan and a Fang clan” (C. K. Yang 1959:12).
We may also note that the variations over time in patterns of occupancy were matched by a lack of consistent congruence between lineage and village. While Chen Han-seng initially states that in Guangdong “usually one village is inhabited by one clan” (1936:37), his more specific examples indicate that there was a great deal of variation in village composition. In “Chao-an,” he reports, “nearly one-half of the villages of the entire district are inhabited by people bearing the same clan name,” and in Huiyang “more than one-half of the villages . . . are so inhabited.” Yet while it would certainly appear that single-lineage villages were “clearly very common” (Freedman 1958:3), we will later see how internal population movements did make for the appearance of multisurname villages in Guangdong.
Descriptions of farming village sites in Guangdong indicate that their location was determined, or at least influenced, by the southeastern mode of agricultural production. The quality of agricultural land mainly depended upon the availability of water for the inundation irrigation necessitated by wet-rice farming, and so there was a strong negative correlation between land elevation and productive value, with steadily diminishing returns precluding any use for rice-growing purposes of land beyond a certain elevation (C. K. Yang 1959:24–25). We would expect that the more productive lands would be left for farming, and that a village would be situated within reach of them; in less level terrain it would have to be located on the side of a hill between the lands farmed and the lands left uncultivated. Such was the case in the hilly sectors of the New Territories (Tregear 1958:84), as exemplified by “Chong Village,” a Hakka settlement there. The location of Chong Village is such that “a low tree-covered hill rises behind the village and in front is a pond and terraced fields dropping away to the valley floor” (Pratt 1960:148). In Guangdong, Eitel noted that Hakka villages “are invariably built each on the brow of a hill or at any rate backed by a small grove of trees” (1867:39).
In the level rice-growing plains, the location of a farming village might also be determined by certain features of the physical environment. In the New Territories, again, villages “situated on the flat paddy plains are almost invariably sited on patches of slightly higher land and stand as islands in the midst of their fields” (Tregear 1958:53). Considerations of flood control as well as the preservation of the most productive lands for agricultural purposes presumably would enter into such an arrangement.
It is clear that the village mode of settlement did not correspond to the kind of Hakka settlement initially required by the poor and scattered quality of the residual lands left to them. Yet village settlement was the best adjustment to the agricultural and environmental demands of the Southeast, and Hakka would soon begin to group themselves into new villages and take over old ones. With the possible exception of the type of settlement discussed immediately below, either peaceful or violent village formation would involve a readjustment of holdings which would engender much hostility along dialect lines.
In certain cases, the peaceful establishment of Hakka villages could come about through state interference in local affairs. Very widespread in its effect was the series of decrees commencing in 1660, which called for the evacuation inland of the entire coastal population of Guangdong, Fujian, Zhejiang, Jiangsu, and Shandong (Luo Xianglin 1950:27). Of these provinces, Fujian and Guangdong were the hardest hit (Xie 1930:812). The Chiqi xianzhi (Chiqi County gazetteer) says that between 80 and 90 percent of the coastal population of Guangdong perished when they were driven inland (Chiqi 1920:8/2a). In 1684, when the order to return was given, “of the people of various xian who had been driven inland, not more than two or three could return to each village” (Luo Xianglin 1950:27). Guangdong officials then petitioned that colonists be settled there, and among those arriving were many from the Hakka zone in northern Guangdong and adjoining parts of Fujian and Jiangxi. The gazetteer relates that many of these Hakka settled in the xian of Taishan, Heshan, Gaoming, Kaiping, Enping, Yangchun, and Yangjiang, where “many lived interspersed with the natives” (Chiqi 1920:8/3a). When the Hakka arrived in the coastal regions, the level and most fertile lands had already been secured by the Cantonese, for the latter had lived in areas adjoining the evacuated zone and so were able to enter first (ibid.:8/4b). This was the case over a large part of the Guangdong coast extending at least as far north as Haifeng and Lufeng. There, it appears that as late as the 1920s and 1930s Hakka-speakers were still primarily tenants on Cantonese-owned lands (Wales 1952:199; cf. Eto 1961:163–64).
But whether the Hakka were tenants or owner-cultivators, coastal lands available to them were not conducive to the settlement at one locale of newly arriving groups of kinsmen. This is clearly borne out by the following description of the fragmentation of Hakka agnatic groupings entering into the Xiangshan xian coast sectors.
Part of the Chen lineage is to be found in the Xiangshan village known as Zhanyong. The lineage founder’s descendant in the ninth generation moved from his ancestral home in Xingning (in the northern Hakka region) to Zengcheng, in central Guangdong. He had four sons, the eldest being Yinglue. Then “Yinglue’s sons divided into six large fang.”2 In 1732, the fang of the three oldest moved to Zhanyong, where they “again divided and dwelt” in Shenwan, Guyou, Shagang, Ziding, and Gouhuan, all of which were in the same xian. In 1737 the founder of the Liu lineage also moved from Xingning to Shenwan, where his descendants resided at the time the Xiangshan xianzhi (Xiangshan County gazetteer) was written. This village and Zhanyong in addition were two of the four settled by the descendants of Mao Yuanfeng, who arrived at Xiangshan in 1725, having moved from Wuhua (Xiangshan xianzhi, cited in Luo Xianglin 1933:262).
Thus we find that in Zhanyong we have at the minimum members of the Chen and Mao lineages, in Shenwan there are Chens, Lius, and Maos, and that both the Chens and Maos were dispersed among other villages as well. It seems safe to say that the limited settlement possibilities the region offered for arriving Hakka meant that available village sites could not completely absorb extended agnatic groupings. To secure the best of what land was obtainable, agnates would have to scatter across several villages, and this resulted in the formation of multisurname communities.
A recent account from the New Territories contains another example of the formation of a multisurname Hakka village. Harold Ingrams describes how a Hakka named Kong Tai Kuen arrived at the “Land of the Jumping Dragon,” a region consisting of several villages all inhabited by Cantonese belonging to the Tang lineage:
Kong rented a house and became a tenant farmer. He recommended two of his relatives to come along also, but they stayed only three years and then returned to the Kong ancestral village . . . while Kong gave up farming . . . and moved to Fan Ling. (1952:162)
But Kong also recommended another Hakka, Chan Lok Chom, a convert who was a “Reverend of the Basel Evangelical Missionary Society.” Chan arrived in 1897 and, “intending to settle down as a farmer, he rented a house” in one of the Tang villages, “but in 1898 he decided that his Punti farmhands were too much for him and his neighbors too difficult, so . . . he made up his mind to return to mission work and rented his land to kinsmen, Chan Kiu and his father” (Ingrams 1952:163).
Chan Kiu arrived in 1900, and “meanwhile other Christian Hakkas, of the family of Lin, had come in . . . to farm.” Two of them
built a row of eight houses for themselves and the Chans, forming the nucleus of a Hakka village. The Tangs never did like the intruding Hakkas, and one of them said that the houses would interfere with the Fung Shui [fengshui, lit. “wind and water,” i.e. geomantic configuration]. (Ibid.)
But the matter was settled by the district officer, and “the village was founded in 1902” (ibid.).3
The village grew larger and larger. . . . There are about 20 houses or terraces in the village now. . . . Today the village . . . consists of a collection of multiple families. There are Tsuis and Lins and Pangs and Chans and Cheungs . . . and Tangs—about ten different families not related by blood. Herein lies the difference between them and the Punti Tangs, none of whom are Christians. Not all of the villagers are in the Church or on the land, though there is scarce a family not represented in one or the other. (Ibid.)
The heavy Christian influence in this village probably played a part in attracting certain nonagnates to one settlement. But the common denominator of all the villagers was dialect, not religion. Both examples indicate that the “fragmentary and heterogeneous nature of the lineage elements” which made up the Chinese populations migrating to certain parts of the Nanyang or Southeast Asia (Freedman 1957:24) was duplicated by the Hakka in Guangdong itself, and was accentuated during settlement.
It is of some consequence that the incident regarding fengshui was recorded in the New Territories, a region in which peace and social order is enforced by a powerful state apparatus. I suspect that had the dispute occurred in those areas of Guangdong or Guangxi where state authority was tenuous, recourse might have been had to violent means of settlement. For it is clear that while both peaceful and warlike measures were employed by the Hakka as they developed their villages, resultant pressures reinforced at the village level the strict linguistic dichotomization of the countryside. Wilhelm Oehler depicts the changeover from Cantonese to Hakka occupancy of villages in quite absolute terms:
It is an impressive sight, that of the old Punti villages, enclosed by ivy-covered brick walls and towers half in ruins and surrounded by ditches, and all about the rows of white houses of the Hakka, into whose hands the greater part of the land has fallen. Soon after the appearance of a Hakka house inside the walls of a Punti village, the Puntis disappeared completely. (1922:35)
That the shift in population was as complete as described by Oehler is borne out by a British survey of the New Territories conducted at the end of the nineteenth century. The investigators reported that Hakka villages “are inhabited solely by members of their own race, just as the Punti villages are inhabited only by Cantonese, though there are a few villages in which both races are represented” (Hong Kong 1900b:188–89).
Simple arithmetic shows that the mixed villages were few indeed, for of the 423 villages in the New Territories, 161 were Cantonese-speaking and 255 were inhabited by Hakka (Hong Kong 1900b:188–89), leaving only seven where Cantonese and Hakka lived side-by-side. Fuson, writing in 1926 mentions a village near the New Territories border which thirty years before had been inhabited solely by Cantonese: “Today,” he says, “only Hakka live in that village” (1929:12).
Even where Hakka villages had formed, Cantonese villages tended to be larger and wealthier. It must be remembered that an initial condition of Hakka settlement was the fact that the holdings of Cantonese villages were large enough for them to have land to spare. In addition, the poorer lands, which the Hakka farmed, would not allow the maintenance of nucleated populations as large as those of the Cantonese. The 225 Hakka villages in the New Territories were inhabited by 36,070 people, averaging out to 160.3 persons per village; the 161 Cantonese villages contained a population of 64,140, or 398.3 persons per village (Hong Kong 1900b:189). “Since the Hakka in general are poorer than . . . the Punti,” notes Hubrig, “their houses and villages are not built or arranged as well, so that even from a distance one can distinguish a Hakka village from a Punti village” (1879:102).
He adds that “Punti villages are closed, surrounded by walls and moats, whereas the villages of the Hakka for the most part are open” (ibid.:103). Cantonese holdings could support larger village populations than those of the Hakka, and they might also be capitalized in the form of fortifications. Indeed, at the village level Hakka-speakers could still be predominantly tenants working Cantonese lands. Hsiao Kung-chuan describes the conflict that could emerge out of such a situation. In 1852 Enping was the scene of a series of fights where “in a number of villages ‘all the tenants who cultivated the land of the Puntis refused by force to pay rent. . . .’ In some instances Hakka tenants slaughtered their Punti landlords and put the torch to their houses” (1960:431).
Thus the formation of Hakka villages might not detract from the need for large-scale mobilizations when strife would develop with the Cantonese. The Chinese and Japanese Repository describes a Hakka-Cantonese conflict in which many Hakka villages, containing an unspecified number of lineages, engaged in a long and largely successful campaign to expand their holdings at the expense of a single Cantonese lineage, the “Chin” (1865:283). In what is now Huiyang, members of the Chin lineage had come to be the sole inhabitants of the town of Ho-au, and in the early part of the eighteenth century had built in addition, Pu-lu-wei. Pu-lu-wei,
as early as 1737 . . . was surrounded by a wall of about 20 feet high, with sixteen parapets, which gave it the air of an old castle. The length of the wall is about half an English mile, and is surrounded by a ditch from 10 to 20 feet deep, and from 50 to 60 wide. (Ibid.)
The remainder of the description indicates that the Chin holdings had extended considerably beyond the two villages:
After many changes of fortune, Ho-au was in 1843 involved in a conflict with the Hakka, which nearly ruined the entire clan. About three English miles south-west of Ho-au there is a marketplace, which was built by the people of Ho-au. The Hakka, to whom the market was let, refused in that year to pay their taxes, and . . . appeal was had to arms. Both parties fought for about six years, when another market-place, also belonging to Ho-au, gave occasion to a conflict with even more powerful clans.
In 1850 more than 90 villages united for the extermination of the Chin clan. Pu-lu-wei fell by treachery, and the people were stripped of the last piece of dress they had on their person; but though more than 5,000 men surrounded Ho-au, where only from 300 to 500 fighting men were, the Hakka had not the courage to enter the village, but moved off without any booty. Both parties being at last exhausted, peace was concluded, one of the market places lost to the Chin clan, the other submitted, and promised to pay further rent.
In 1856 war broke out anew, and frightful murders would have been committed, had not the writer of these lines induced the parties to come once more to terms. (Ibid.)
It is interesting to note from the above that arrangements could be made for an orderly transfer of property as a result of victories or defeats sustained in the course of armed conflict. The same held true in the New Territories, for there, we are told, “some land in the Kowloon district is held under a title by capture, where the clans fought and the losing clan gave up a field as the price of the cessation of hostilities” (Hong Kong 1900a:23).
It seems obvious that even in the most disturbed times some security of possession was necessary so that agricultural work might continue.
Fighting at the level of contesting villages of course also could involve more than one Cantonese village or lineage. Meyners d’Estray makes the general statement that
several communities inhabited by members of one single tribe [Hakka-speakers or Cantonese-speakers] often form alliances against the intruders or against the original inhabitants with the goal of dispersing or exterminating the opposing party, and taking possession of its goods. (1890:99-100)
A specific instance of such an “alliance” is taken from the Enping gazetteer by Hsiao Kung-chuan:
During the reign of Hsien-feng [Xianfeng] (1851–1861) the Hakkas caused disturbances. . . . The various clans that dwelt within a radius of ten li4 organized themselves into a group . . . calling itself Wu-fu Pao. Funds were raised and a building erected at Sa-hu-hsü . . . where meetings were called to discuss matters as they arose. (1960:343)
We should note that this last reference to Enping as well as the conflict at Ho-au seems to have involved the participation, at least on one of the opposing sides, of defined local groupings which united on the basis of common dialect. There seems to have been at Ho-au a steady growth in the number of villages engaging the Chin lineage, and the references to the conclusion of the hostilities indicate that there was a structured leadership on both sides which could make decisions affecting the combatants involved. To have an effect such decisions would have to apply to territorially bounded units and at Enping the dimensions of these units and their links are clearly described.
Another instance where the speakers of one dialect all were members of a grouping definable on other than linguistic grounds is to be found in the Chiqi xianzhi. By 1855, the gazetteer relates, fighting, which was then current between the Hakka and the Cantonese, had spread to Yangjiang ting (now a xian), which was on the coast and south of the main area of conflict. Here the Hakka were comparatively few in number. During 1855 and 1856, most of the Hakka living in the Wubao region of the xian fled north to certain Hakka areas in Xinning. There was one exception, however, and this exception is instructive. “The various villages of the Lin lineage,” we are told, “were quite wealthy and did not desire to flee.” They therefore agreed to split their fields with the Cantonese, and the two groups (Lin and the Cantonese) “jointly requested the ting office to record the agreement” so that “at that time a calamity was avoided” (Chiqi 1920:8/8b).
In Yangjiang, then, the flight of the bulk of the Hakka and the isolation of the Lin lineage resulted in the immediate localization of the conflict. There was no longer a population intermixed with the Cantonese that was definable in terms of dialect only, and which might include persons or groups whose holdings would in varying degrees be vulnerable to Cantonese attack. There was rather an integrated Hakka lineage, one which obviously contained individuals able to represent the group and make the concessions necessary to preserve at least some corporate holdings. This points to the problem with which we must later deal; for we will see that in the areas where Hakka and Cantonese continued to live interspersed, there would be no convenient congruence of a dialect population with a localized grouping, and thus no readily available means by which conflicts involving alignments along dialect lines might be contained or negotiated.
IV
The material I have presented seems to indicate that in rural Guangdong and adjoining Guangxi, village settlement was strictly along dialect lines. One would also assume that the network of affinal ties would by-and-large also be restricted to people speaking the same dialect. But I do not know of any studies of mainland Southeast China which have devoted attention to this problem, and any statements must remain on the level of conjecture. There is, however, some evidence which indicates that certain features of southeastern Chinese social organization would minimize the effect on dialect distribution of whatever intermarriage between speakers of different dialects did occur. Phenix village, located in northern Guangdong, was inhabited by a group of agnates speaking the Chaozhou dialect of southern Fujianese. The Hakka term for speakers of this dialect is usually rendered in English as Hoklo, though Daniel Kulp writes Holo (1925:79). The situation of Phenix village is described as follows: “In seven [xian] round about Phenix village, Hakka is spoken. The [xian] in which the village is located is the only one in which no Hakka is spoken.” There was thus an entire xian from which the Phenix village lineage could presumably draw Chaozhou-speaking wives. Yet, “when exogamous marriages with Hakka girls bring these new brides into the village, they are faced with the necessity of learning a new language” (ibid.:79).
Now Kulp gives us no idea as to the extent, if any, that “exogamous marriages with Hakka girls” were necessitated by a lack of Chaozhou women from other lineages. What he does tell us of significance is that the virilocal residence demanded by localized patrilineages could act as a force maintaining dialect differences even when speakers of one dialect might be outnumbered by and intermarry with a population speaking another dialect. A group of male agnates could maintain their nucleation and enforce dialect conformity on the part of newly arrived wives. This fact emerges in a stronger way from Lin Yueh-hwa’s description of the Hwang [Huang] lineage-village in northern Fujian. The founding ancestor, we are told,
had migrated out of south Fukien [Fujian] up along the Min River and had settled down . . . long ago in this little village of the Hwang lineage. It was his misfortune that at the time of his arrival all the land surrounding the village was already occupied by the earlier settlers. Yet by dint of hard labour, he had gained a foothold in the village. (1948:60)
Ninety-nine percent of the village population now belongs to the Hwang lineage, and only the village inn is run by outsiders (ibid.).
Today the Hwang villagers, all of the same descent, are closely bound together in so fierce an allegiance to each other against outsiders that their settlement is named “the barbaric village.” Moreover, the clan still retains a special brogue of the dialect of South Fukien which the neighboring villages do not understand. When the Hwang villagers do try to communicate with outsiders, they must use the dialect of the whole district, the Kutien dialect. . . . To this very day the first ancestor’s descendants, in their hearts as well as in their language, are truly a clan. (Ibid.)
While it is to be doubted that Hakka is the dialect involved in this case, it is nevertheless clear that here an alien dialect was in fact preserved even though to all women marrying in, it would initially be unintelligible. Lin says that a reason for its maintenance was the mutual reinforcement exercised by lineage solidarity and dialect differentiation. In the context of an isolated lineage this is certainly true. But I will later discuss the possibility that kin-based solidarity might be extended to nonkinsmen speaking the same dialect. For our purposes, the most important aspect of Lin’s description is that it illustrates what may be called the “dialect-preserving” possibilities afforded by southeastern Chinese lineage organization.
V
We have seen that in Guangdong and adjoining parts of Guangxi kinsmen or coresidents in the same village might unite with outsiders speaking the same dialect for purposes of defense and offense. Hakka-speakers not localized into villages also might so unite. It is thus obvious that in regions of Hakka-Cantonese intermixture the boundaries setting off localized groupings might be rendered irrelevant in terms of alignments based upon common or different dialect. In an effort to discern structural coherence in large-scale mobilizations based upon common dialect we should examine the one sociological model which takes into account the fact that not all conflict in Southeast China proceeded along kin-based or village lines.
Maurice Freedman has said that in Southeast China, “there would appear to have been two alignments of conflict, which cut across each other. In some conflicts lineages were ranged against lineages; in other conflicts, lineages, or class sections of them, were united in their common hostility to the state” (1958:124).
I have attempted to demonstrate that the first alignment, with the important qualification that “village” might in certain cases be substituted for “lineage,” could in mixed Hakka-Cantonese areas be involved with alignments based upon dialect. “Common hostility to the state” might mean membership in secret societies cutting across kin lines, and if members were found in highly differentiated lineages we may suspect that they gave expression to the latent opposition of the common people to the representatives within their communities of the centralized government.
If we accept the dichotomization of Southeast as well as the rest of China along class lines, we must add that there is evidence that differences in dialect might qualify the common front presented against the state. I derive this evidence entirely from William Stanton, who, when speaking of the Triad Society in Hong Kong, notes that
in some lodges the members are chiefly Puntis, and as a whole, these are the most lawless. . . . The members of both the Man On and the Fuk I Hing lodges are Hoklos, and these, like the Fukienese and others speaking a similar dialect, combine principally for mutual assistance in sickness or distress, or their quarrels. Some of the Hakkas combine to commit unlawful acts such as piracy, armed robberies and blackmailing, but most of them seem to carry out the precepts handed down and keep the aim in view for which the Society was founded. (1900:28)
A Triad uprising in Huizhou, where Hakka live interspersed among Cantonese, is described by Stanton:
In 1886 there was an insurrection of about three thousand Triad men . . . at Nimshan in the Waichau [Huizhou] prefecture of the Kwangtung [Guangdong] province. This was chiefly amongst the Hakka population and was caused by the oppressive acts of Government officials, several of whom were slain at the first outbreak. On this occasion four hundred men from Hong Kong, principally stonecutters, assembled at Kowloon, armed with swords and revolvers and commanded by two chiefs . . . whose aim was to reinforce the Nimshan army. (Ibid.:23)
It is quite probable that these stonecutters were also Hakka. Occupational specialization in urban centers seems to a large extent to have been divided along dialect lines, and of the many descriptions of this I will choose from Rudolf Lechler’s, who quite categorically tells us that “on the island of Hongkong, all the stone cutters are Hakkas” (1878:359).
We see, then, that dialect alignments might be expressed in non-kin based groupings, which as sociological entities were well established in China. But I do not know if membership in such groupings would completely satisfy the organizational requirements of a large Hakka population, which might be socially differentiated. One could deduce from the material I have presented so far that wherever the Hakka population contained individuals of gentry rank, these might unite with commoners in times of great duress. This has been reported by Hsiao Kung-chuan for the fighting in Enping (1960:425), and gentry figure prominently in the Chiqi xianzhi’s description of the mid-nineteenth-century Hakka-Cantonese conflicts throughout central Guangdong. In the gazetteer it is claimed that “native gentry” (tu shen) caused the spread of the Hakka-Cantonese fighting to Xingning in 1856, by publicizing a false report of an imminent Hakka attack (Chiqi 1920:8/9a). In June of the same year, the following “Hakka gentry” (Ke shen) were said to have organized the defense against an expected Punti attack in a region of Xingning xian: the shengyuan (lowest-level degree holder) Chen Guangzhi, and the jiansheng (next-to-the-lowest-level degree holder) Jiang Beijun, Li Zhaolong, Chen Hongdai, and Zhong Hongsi (ibid.:8/13b). In the same source it is protested that gazetteers written by Cantonese invariably give distorted descriptions of Hakka-Cantonese conflicts. In the Xingning xianzhi (Xingning County gazetteer),
in reference to the Hakka people [Ke min], they are always called “Hakka bandits” [Ke fei]. The Hakka gentry [Ke shen] are called “rebel chieftains” [ni shou] or are called “bandit chieftains” [zei shou],”
but,
in this work [Chiqi xianzhi] a fair balance is maintained in everything recorded concerning the native populace and gentry [tu zhong shen min]. They are always referred to as “natives” [tu ren] or “native gentry” [tu shen]. (Ibid.:8/10a-10/b)
In regard to secret societies, it can be added that, even if placed in different “lodges,” members of the same society speaking different dialects presumably would have gone through the same initiation rites (cf. Freedman 1958:123), and in many ways feel bound together by common obligations and aims. It may be that the establishment of “lodges” on a dialect basis prevented tensions which otherwise would have arisen in an organization whose hostility was toward the state and the persons representing it, rather than toward different but equivalent sectors of the peasant population.
To return to the question I have posed, if recognition on the basis of common dialect did indeed involve more than kin-based and territorial loyalties during times of Hakka-Cantonese strife, then, barring the interposition of peace from outside sources, it can be expected that conflicts between different dialect populations might in an uncontrolled fashion spread over wide areas where people speaking the different dialects lived interspersed. This is indeed what happened in nineteenth-century Guangdong and Guangxi, and an examination of these events will show us that the structural problems posed were in one fashion or another solved.
At about the middle of the last century, there seem to have been two regions where Hakka-Cantonese conflicts were especially widespread and violent. One was centered in the valley of the Yu River in southern Guangxi, extending from Yongchun through Gui to Guiping. The struggles were most severe in Gui, where one authority, probably in 1848, reported that Hakka-Punti feuding had been going on for five years (Laai 1950:179). By 1850, Hakka-Cantonese conflicts had spread throughout the southeastern part of Guangxi and into adjoining western Guangdong (ibid.). B. C. Henry was referring to this fighting when he observed that “the stockaded villages, forts, and barricades in the mountains [near Hepu, in western Guangdong] bear witness to their [the Hakka’s] struggles in the past” (ibid.:1886:144). The disturbances in Guangxi ultimately would have much more than a regional significance, for it was here that the displaced Hakka became adherents to the doctrine advocated by Hong Xiuquan and Feng Yunshan (Boardman 1952:14). But we must for the moment direct our attention to Guangdong proper.
The other zone of Hakka-Cantonese feuding was in central Guangdong, and although the conflict here had no great repercussions on a national scale, it was the cause of much destruction within the province. By 1851, feuding was already endemic in Enping, Kaiping, and Xinning (Taishan), to the southwest of Canton, as well as in Zengcheng, which was to the northeast of the same city (Luo Xianglin 1933:3). A Hakka “informant” of Elisabeth Oehler-Heimerdinger (n.d.:63) told her that the troubles in his region began as early as 1851, when his village was burnt by Cantonese.
Bandits were also active in the countryside, and some of them were associated with the Red Turbans. E. J. Eitel supplies a report by W. F. Mayers, whom he says had a position in the English “consular service.” According to Mayers, as given in Eitel, Cantonese hostility towards the Hakka reached a climax when the latter assisted the government in suppressing the “Hung-t’ow rebellion.”
For very many years clan fights have been common between the two groups [Hakka and Punti], but the mutual antipathy grew to a climax after 1854, in which year the Hakka clans located there remained for the most part faithful to the government, during the rebellion which was largely joined by the Puntis until suppressed by the Viceroy Yeh. At this time, Hakka and Punti clans dwelt interspersed over all the S. W. districts notably [Xinning, Xinhui, Enping, Kaiping, Gaoming, and Heshan]. After the suppression of the Hung-t’ow rebellion, the ill feeling between the races took the shape of an internecine warfare, in which the authorities were powerless to interfere. (Eitel 1873:162-63)
Luo Xianglin adds that Yangchun xian was also involved. He notes that Hakka actively assisted in the suppression of the original rebellion, and that in doing this (and here he is quoting from a joint report submitted by the viceroy of Liang-Guang5 and the governor of Guangdong), “their motivations were very deep.” The fighting became generalized when the Hakka killed several members of the “native gentry,” whereupon the Hakka-Cantonese conflict “spread like wildfire” (1933:3).
By 1860, Mayers says, the fighting was so severe that there were “shipments of arms and even the dispatch of armed steamers from Hongkong to assist one or other of the belligerent parties” (Eitel 1873:163). Hakka speakers were gradually driven out of increasingly large areas of the Western Circuit, an administrative district in Guangdong, and formed many “wandering bands.” Towards the end of 1862, many of these attacked and occupied the coastal town of Guanghai, “until driven out by an Imperial force cooperating with the Punti clans.” During 1864, the remnants of the displaced Hakka, “who had become half banditti, half refugees,” gathered in the mountains of the Western Circuit, where
at various points . . . such as [Nafu, Jinji, Wuhang, and Chishui] . . . they established little Republics, in which they tilled the ground, built habitations, and defended themselves as best as they could against the incursions of . . . [the] Punti. (Eitel 1873:163)
In 1866 several measures were taken to end the fighting and resettle the Hakka, and in the following year the Governor of Guangdong ordered a portion of Taishan (Xinning) xian set aside exclusively for Hakka settlement. Cantonese in the area, which was constituted as Chiqi ting, were ordered to give up their lands to the Hakka, in return for which they received vacated Hakka farms in the interior (Luo Xianglin 1933:62). There are various estimates as to the number of casualties which resulted from the fighting, and Luo says that the total losses on both sides in dead, wounded, and missing were at least five-hundred-thousand (ibid.:3). For our purposes, it is safe to say that many lives were lost, and that the fighting reached very large proportions indeed.
The fighting in Guangxi and Guangdong followed logically from the conditions of Hakka settlement. A discussion of this involves a restatement of some points already covered. That in China kinship was the most obvious force structuring social relationships and binding people into groupings hardly needs to be dwelt upon. For the moment we are concerned with the larger kin-based grouping, the patrilineage. We have seen that Hakka-speakers might be in a situation where, due to recency of arrival and constrictions in living room, they simply would not have localized lineages. At a certain point in their settlement they might not even have villages, so that the defense or expansion of their agricultural holdings could not involve rigidly defined localized groupings. In places where Hakka had been settled for a longer period of time, or where there was more land available, lineages and villages could develop. Both types of groupings, of course, are sociological entities, and in Southeast China there was a good chance that they would be convergent. In general, we would expect that decisions regarding conflicts with one’s neighbors would go through recognized channels, and that a resolution, once made, would involve a number of people, all of whom would be units in a recognizable social structure. But the circumstances of Hakka settlement were such that to limit behavior involving defense and offense to individuals encompassed within structured relationships would, in a hostile Cantonese-speaking environment, be highly maladaptive. Indeed, it might even be impossible, for the issue involves a two-sided discrimination of friends and enemies. If, over a wide area, Hakka-speakers would be engaged in expanding their holdings, then Cantonese could regard them all as a common foe. During the fighting in Guangdong a common Cantonese slogan was “Hate the Hakka and differentiate dialects” (Chou Ke fen sheng) (ibid.:3, 116; Oehler-Heimerdinger n.d.:64).
Though hostile attitudes and actions need not alter the structure of the group expressing them, under certain circumstances the behavior usually associated with structured relationships may lose its clarity of definition. Thus, alignments on the basis of common dialect could result in ever expanding aggregates of people being pitted against each other. And this is what happened in Guangxi and in Guangdong. An informant has told me that a common saying among the Hakka living in Huiyang, a xian shared with the Cantonese, was “Acknowledge one’s dialect, not one’s surname” (Ren sheng, bu ren xing).
Now, although I do not think that Oehler-Heimerdinger’s account of the Hakka Cantonese fighting in Guangdong is proper material for analysis, it certainly does depict widespread destruction of Hakka and Cantonese people and villages (n.d.:61-126). The resultant fragmentation of even nuclear families did for a time bring chaos to wide areas. It was under such circumstances that common or different dialect could be the only criterion remaining by which an individual’s behavior towards another could be defined. From Mayers’s description we may gather, however, that the turmoil did resolve itself, insofar as discernible groupings developed within the Hakka population. And information in the Chiqi xianzhi gives us a fairly good idea as to what the “republics” of which Mayers speaks actually were. The gazetteer does not supply much data about the specific locales mentioned by Mayers, except to refer to them frequently as refuges to which many Hakka fled. Other settlements, however, also places of refuge, are described in greater detail.
We have already noted the involvement of Hakka gentry in the conflicts with the Cantonese. The Hakka gentry, whose prestige and power within their own lineages and villages derived from participation in a nonlocalized nationwide upper class, were in a position to direct military activities involving Hakka-speakers from many villages and lineages. Having a tradition of managerial responsibility, they played a key role in the establishment of fortified bases within which Hakka from over wide areas would seek protection. The gentry would organize the manpower thus assembled into military units. Under these circumstances previously localized agnatic groupings, forced to come together at restricted and strategic sites, were absorbed into larger defense units whose only common characteristic was that of dialect. The gazetteer tells us, for example, that in June 1856, Cantonese in what was then the Chiqi region of Xingning killed a Hakka. The fighting spread, with the Hakka villages in the rural subdistricts (dong) of Chonglou, Sijiu, and Wushi undergoing attack. Additional attacks were feared, and two Hakka gentry of the village of Chonglou, the gongsheng (holder of the third-highest degree) Yang Zizhao and the jiansheng Yang Yuanfeng were concerned “that since the Hakka were dispersed, their situation was such that it would be difficult [for them] to defend themselves.” They then issued a call, and sixteen localized lineages, which had been distributed among eight villages, arrived at Chonglou: the Huang, Wu, and Zhu lineages of Shishan; the Yang, Li, and Zhou lineages of Daxiaoma; the Ye, Yang, Wei, and Zhou lineages of Huangshuihang; the Li lineage of Tonggu; the Zhang lineage of Shijiao; the Tang and Yang lineages of Henglong; the Lin lineage of Shangjing; and the Jiang lineage of Pangbo. The Hakka gathered at Chonglou then “divided into eastern and southern camps, prepared dwelling places, and considered means of defense” (Chiqi 1920:8/15b). Later, after additional Cantonese attacks, the military arrangements were formalized by the “establishment of a bureau” (sheju), directed by the two gentry mentioned above and one other man. It was their task “to manage all affairs relating to the mobilization of able-bodied men and preparations for defense” (Chiqi 1920:8/16a).
Another Hakka defense position was established in a coastal region of the Western Circuit. In this area there were over four-hundred Hakka villages, of which the gazetteer specifically names twenty-six: three of these were single-surname villages inhabited by the Zheng, Chuan, and Ye lineages; there were in addition six two-surname villages, distributed across which were the Lin, Xie, Tang, Huang, Chen, Zeng, Long, and Deng lineages or lineage aggregates. At Shenjingxu (Zheng lineage), the following Hakka gentry “established a bureau”: the wuju (military juren, holder of the second-highest military degree), Zhong Dayong; the shengyuan Zheng Rong, Huang Tengfang, and Chuan Tenghui; and the wusheng (military shengyuan, holder of the lowest military degree) You Qigao. They then “selected able-bodied men to defend it” (ibid.:8/12a).
Many more examples of the “establishment of bureaus” could be drawn from the gazetteer, and these involved Cantonese as well as Hakka. Indeed, on one page of the gazetteer seven locales are cited at which Cantonese were said to have “established bureaus” (ibid.:8/9a). The two examples I have given, however, suffice to show that in the absence of existing groupings capable of coping with the problems presented by conflicts aligned along dialect lines, new groupings could emerge. Of course, I do not know if the “establishment of bureaus” was the only way by which the Hakka-speakers of central Guangdong would gather for combat. Nevertheless it is clear that the very phenomena of defense and offense which we are discussing must at a certain point involve either a formal or an informal organization. Even an ad hoc arrangement is, after all, an arrangement.
That discrimination along dialect lines involved something less than complete confusion also emerges from work that has been done by Laai (1950) on the formation of the Society of God Worshipers during the Hakka-Cantonese feuding in Guangxi. I have noted that the fighting there probably had been going on since the early 1840s, and that it had spread to adjoining parts of Guangdong. In 1844 Hong Xiuquan and Feng Yunshan arrived in Guangxi after having attained very little success in propagating their ideas among the Hakka of their native Hua xian, near Canton. In Zigu Village, Gui xian, where a “cousin” of Hong’s with the surname Wang lived, they set about preaching an iconoclastic, monotheistic doctrine with heavy Christian and anti-Manchu overtones. Hong returned to Guangdong the same year, and it was not until 1847 that he rejoined Feng in Guangxi. By that time Feng had succeeded in attracting over three thousand converts to a religious society now called the Society of God Worshipers (Boardman 1952:12-14). In this presentation I am not concerned with an analysis of the factors which led to the original conversions in Guangxi or with the transformation of the Society into a revolutionary army. The fact remains that much of the recruitment of its members revolved upon the Society’s providing an organizational framework to which a large population of Hakka-speakers, embroiled in a conflict with the Cantonese, might attach themselves on the basis of common dialect.
Laai notes that there were several factors making for confusion in Guangxi. A series of droughts afflicting the area
had the most serious effect on the land of high elevation; this was the land cultivated by the Hakkas. In the lands which the Hakkas rented from the Puntis, the result of the drought was to cut productivity, so that Hakka tenants had less income and couldn’t pay their rent. (1950:184)
During the same period (1846–50), pirate activities in the Guangxi riverine system increased greatly (ibid.:107), and by 1848 “the domination of the Yü River and the Hsün [Xun] River by the . . . pirates . . . virtually isolated the . . . territories south of the rivers from the Kwangsi [Guangxi] provincial administration” (ibid.:143–44). Associated with this and the unemployment resulting from the exhaustion of several large silver mines in and about Gui xian was the appearance of many land-based bandits (ibid.:103–104). These bandits, however, formed only one element of what was essentially a rural response to the decline of centralized rule. Larger villages and towns in Guangxi organized gentry-directed tuanlian (ibid.:185). Secret societies were also prominent in the area (ibid.:178).
In the midst of this turmoil there remained the basic duality in the countryside between populations speaking different dialects, and so Hakka-Cantonese feuds began to increase (ibid.:172). Most of the larger villages were inhabited by Cantonese-speakers, who were able to organize and finance the tuanlian on an independent local basis (ibid.:185). For the Hakka, who were more dispersed, such an arrangement was impossible. Yet throughout the late 1840s an organization which seemed to have been designed to meet the needs of the Hakka had been developing. Although it had bases, it was not localized with respect to membership. Its various branches, indeed, were united by a common ideology which sharply distinguished them from other groupings. But the Society of God Worshipers, which had been introduced into Guangxi by two Hakka-speakers, was also notable for its linguistic exclusiveness.
By 1850 the Society had a large membership organized into units spread over at least fifteen xian in southeastern Guangxi and western Guangdong (ibid.:176–77), amounting to one-fifth of the area of the former province, and one-sixth of that of the latter (ibid.:184). The membership of the Society was quite heterogeneous, and included representatives of many specialized occupational classes in addition to “landowners and members of the lower intellectual class such as clerks and graduates of public examinations” (ibid.:167). Although no “high officials” or “persons of the high intellectual class” joined the Society, it was unusual in that its leadership contained “wealthy persons” (ibid.:168). Laai sees in the diverse membership of the Society many persons “whose interests collided.” He believes, however, “that the great driving force which welded them together despite conflicts among themselves was the dialect, namely the Hakka dialect, spoken among the groups joining the Society” (ibid.:171).
It was, of course, in the context of Hakka-Cantonese fighting that an organization with such a variegated membership could arise. One instance of a Hakka-Cantonese feud was related to the missionary Theodore Hamberg by Hung Jen-kan (Hong Rengan), a relative and early convert of Hong Xiuquan. After noting that “the . . . Hakka villages are very numerous in Kwang-si [Guangxi], though in general not so large and opulent as those of the Punti,” Hamberg tells how
a feeling of enmity has long existed between the two classes [Hakka and Punti], and every new incident only served to augment the hatred. At that time a very rich Hakka of the surname Wun had taken a girl as his concubine, who had been promised to a Punti man, and having agreed to settle the marriage with her parents by paying a large sum of money, he peremptorily refused to give her up to the Punti claimant. . . . Soon after, a civil war commenced between the Puntis and Hakkas of the Kwei [Gui] District, in which gradually a number of villages were involved. The fighting began on the 28th of the eighth month [September 1850], and during the first days the Hakkas had the advantage. . . . Gradually, however, the Puntis grew bolder and . . . as their number was considerably larger, they defeated the Hakkas and burnt their houses, so that these had no resting-place to which they could resort. In this distress they sought refuge among the Worshippers of God, who at this time lived dispersed in several districts, in congregations counting from one to three hundred individuals. They willingly submitted to any form of worship to escape from their enemies, and received the necessary supplies, which they were now destitute of. (1854:48-49)
While the distinctive ideology of the Society made possible its maintenance over the area, it was through its provision of a structure to which embattled Hakka could align themselves that its membership was obtained. But in October 1850, the Society was ordered to mobilize, and became an anti-Manchu revolutionary army. Many non-Hakka joined in what was now the Taiping Rebellion (Laai 1950:197), the subsequent development of which is beyond the scope of this essay.
CONCLUSION
I have tried to show that, through specific sequences of migration and settlement, differences in dialect could have an important influence in the alignment and formation of social groupings in Guangdong and Guangxi. I have made only the most qualified attempt to develop a concept of “dialect group.” For, if we mean by this term all speakers of one dialect, it clearly lacks sociological validity. If, however, we take dialect to be a sociocultural variable, in the same sense as kinship or territoriality, I think it can be shown that the broad range of social relationships which dialect discrimination could influence makes it a primary force in shaping group formation. The manifestation of this variable in sociologically significant ways depends, I believe, on the prior existence of factors conducive to the formation of groups of many kinds. Overpopulation and weakening imperial control certainly were such factors in nineteenth-century China (Hsiao Kung-chuan 1960:503), and throughout southern China growing rural autonomy led to the establishment of localized groupings of many sorts, and for all of them defense was a basic problem (Michael 1949). Hakka-Cantonese conflicts were part of a larger picture, and I do not think it can be shown that the disorders in Guangdong and Guangxi would not have occurred, albeit in different form, if these two provinces had been occupied by Cantonese-speakers or Hakka-speakers only. The significance of difference in dialect is not universal, and in certain urban contexts there is indeed a tendency for Hakka to give way to Cantonese (Forrest 1951:674). But many specific forms of social activity did directly relate to differences in dialect, and no statements as to the social organization of the area can be complete without taking this into account.
The analysis made in this essay has been limited to a geographically restricted area of China. It would be well to discuss briefly its bearing on Chinese society as a whole. We may mention that in diverse parts of China, and at different times, the juxtaposition of different linguistic populations apparently could lead to primarily hostile patterns of interaction (cf. Hsiao Kung-chuan 1960:421-23) perhaps similar to those which existed between the Hakka and the Cantonese. But whether such interaction and the resultant group formation was of sufficient frequency to warrant consideration as an important theme in Chinese society outside the Southeast is, to my knowledge, a question still unanswered. We can approach the matter from another angle, however, and note that groupings and interpersonal relationships based upon any one of a number of possible points of reference shared in common have played an important part in giving Chinese society its configuration. Morton H. Fried has spoken of this as “something which might be called t’ungism [“tongism,” from tong, “the same; together”] . . . and which might be translated as “togetherness” (1962:20). Tongism
applies to people drawn together because they speak the same Chinese sublanguage, come from the same province, come from the same county, come from the same town, or have graduated from the same middle school or college, or are members of a particular class year or either of these. (Ibid.:25)
The bonds of tongism thus might cut across ties of kin, class, and residence. It is a concept broad enough to subsume groupings resulting from different historical processes, and based upon a variety of functional requirements. Indeed, the presence of such groupings contributed to the complexity of Chinese society. In this essay I have attempted to indicate some of the developments which led the Hakka, as well as the Cantonese, to form groupings differentiated on the basis of dialect. In a general analysis of Chinese society these groupings may be considered examples of tongism writ large.
Author’s note, 1996: I originally wrote this article for a conference in 1962, and a year later it served, slightly expanded, as my master’s essay for the Department of Anthropology at Columbia University. In 1968 this later version was published in Ethnohistory (vol. 15, no. 3., pp. 237–292) with only a few minor editorial changes. With the kind permission of Duke University Press, on behalf of that journal, it is reprinted here. In preparing this article for its reappearance in this volume I have limited myself to correcting a small number of typographical errors and revising my translation of one Chinese term.
Author’s note, 1968: This paper was submitted in 1963 in partial fulfillment of the requirements for the degree of Master of Arts in the Department of Anthropology, Columbia University. An earlier version, entitled “Dialectical Differences and Social Relations: Cantonese-Hakka Interaction in the 19th century,” was presented at the Seminar on Micro-Social Organization in China, held at Cornell University, October 11–13, 1962, and sponsored by the Subcommittee on Chinese Society of the Joint Committee on Contemporary China of the American Council of Learned Societies and the Social Science Research Council.
Editor’s note, 1996: The body of this article remains in very much the same form as when it was first published. Minor editorial changes have been made in order to maintain consistency with the rest of this volume. The most significant change is that, as far as possible, Chinese terms in the original article have been changed to pinyin romanization. I am most grateful to Howard J. Martin for voluntarily taking on the painstaking and time-consuming task of converting Wade Giles and idiosyncratic systems of romanization to standard pinyin. In cases where original sources did not provide Chinese characters for personal or place names, rather than guessing at the intended character, the original spelling has been retained. Within English quotations pinyin romanization of Chinese terms is provided in brackets whenever possible.
More has been written on Hakka migrations since the original version of this article was written (see, for example, Kiang 1991, Lo Wan [1965], and S. T. Leong [1980, 1985]). On Hakka language, dialect groupings, and linguistic analyses of the origins of Hakka language, see Jerry Norman (1988), Moser (1985), Paul Yang (1967), and Hashimoto (1973). On the relevance of Hakka dialect to the Taiping Rebellion, see Kuhn (1977), Erbaugh (this volume), and Constable (1994: chap. 2). Additional sources are provided in the Introduction to and References in this volume.
1. A xian is an administrative unit below the provincial level. It is usually translated as “county” or “district.”
2. The fang is a subdivision of family, lineage, or other agnatic unit.
3. Editor’s note: The village described here is Shung Him Tong, the community of Hakka Christians described by Constable in this volume.
4. A li is a measure of distance approximately equivalent to one-half kilometer.
5. The Liang-Guang (lit., “two Guang”) refers to the provinces of Guangdong and Guangxi.