1 – Introduction
1The Dongxiang language is spoken by approximately 300,000 people, mainly in Dongxiang Autonomous County, which is located in Gansu Province in the northwest of the People’s Republic of China (PRC). This Mongolic language is part of the “Gansu-Qinghai Sprachbund” (Slater 2003), which groups together languages of various linguistic groups (Sino-Tibetan, Turkic, and Mongolic) that have converged at various levels under the influence of surrounding Chinese dialects. The Dongxiang language has been strongly influenced by the Chinese Linxia dialect (línxiàhuà 临夏话), spoken in the neighboring town of the same name, particularly in terms of vocabulary and phonology. This dialect has itself been influenced by surrounding Mongolic languages. The isolation of the Dongxiangs, who live mostly in mountain areas, and the desire of speakers of Dongxiang to preserve themselves from the influence of non-Muslim communities have meant that, for a long time, contact with speakers of other varieties, including Mandarin (Pǔtōnghuà 普通话) and northern dialects has been absent from the linguistic landscape.
2This sociolinguistic situation has been stable for at least six centuries, but is today undergoing drastic change under the effect of the economic development China has experienced over the last two decades. Urban growth, particularly in the town of Suonanba, is increasing contacts between populations and languages, leading to the appearance of new social and linguistic behaviors in a section of the population. Mobility, growing use of new technologies, and an increase in the number of schools are all factors stemming from urban growth that have widened contact between previously almost entirely exclusive languages and the Linxia dialect (or Linxia Mandarin). There has been a gradual introduction of various northern Chinese dialects, including the Lanzhou dialect (lánzhōu huà 兰州话), and the national standard variety of Mandarin (Pǔtōnghuà 普通话), into the speech of a certain age bracket, notably the youngs. The appearance of these new practices, which have been observed and attested to by members of the linguistic community and which result from Dongxiang’s state of economic and social transition, raises the question of the possible emergence of a language variety spoken exclusively by the young and, alongside this, the appearance of a new social category.
3This paper will attempt to describe the new practices that are evolving in a context of nascent urbanization, in order to assess the degree to which they are specific to young urban speakers in the Dongxiang community in comparison to other speakers (particularly the oldest and those from rural areas). My aim is to answer the following questions:
4a) To what extent are these new behaviors and ways of speaking specific to urban dwellers and the young in particular? The emergence of these new practices is, after all, being encouraged by current changes (urban growth and increased contact between languages).
5b) Do these practices differ from those caused by previous language contacts and which led to the creation of the Dongxiang language?
6c) Given this context, is it proper to categorize this as the speech of “urban youth”?
7This is an initial, primarily descriptive, approach to understanding the possible appearance, in practices and representations, of a process of sociolinguistic categorization emerging amid continuous linguistic and social change.
2 – The Scientific Approach and Field Investigation
The Scientific Approach
8This research on the Dongxiang language is a continuation of work that began with a study of Chinese loan words in Khalkha Mongolian, spoken in Mongolia. The aim was to observe effects of contact between languages through the transmission of loan words in a situation of monolingualism (or at any rate in a situation where there is no Mongolian-Chinese bilingualism) and where there is very little contact between the two languages today. The second step was to continue this study in a context of more intense contact and widespread Mongolian-Chinese bilingualism; I therefore directed my research toward Chakhar Mongolian in Inner Mongolia, a province in the north of the PRC near the border with Mongolia. Lastly, I pursued my research further by focusing on the Dongxiang language, where a situation of very intense contact has resulted in a mixture of Mongolian and Chinese, but without the language having been completely transformed into a mixed language. Initially, the work simply involved describing linguistic changes in the Dongxiang language due to contact with Linxia Chinese, with a socio-historical perspective. It was also necessary to conduct a field investigation to collect data.
9The primary aim was to highlight the relationship between Dongxiangs and their language so as to understand their representation(s) of it and whether it is an important factor in the formation of their identity. It was necessary to understand why, after centuries of contact with speakers of Linxia Chinese, Dongxiangs have preserved their language while other communities often abandon their “ethnic” language in favor of a local variety of Chinese. The second aim was to gather data in sufficient quantity to analyze these changes and consider them in relation to various studies that have been made over the last three decades. It was particularly important to observe any recent phenomena caused by the influence of other varieties of Chinese in this contact situation. The field investigations, and subsequent analysis of the corpus, led me to the observation that the language of the Dongxiangs is in a situation of change and that these changes reflect the socio-economic transformation of their society. Social change, including linguistic change, is affecting the youngest section of the community to a greater extent than other sections. The initial approach was not, however, to observe the interactions of speakers from any age group in particular. It was later, during my involvement in the study days organized by the Mauritius Institute of Education (of the University of Mauritius) in 2010, that I turned my attention to the possible emergence of youth and youth language categories in an urban context (Lefort 2012 and forthcoming A).
Investigation and Sampling
10I conducted two field investigations, the first from March to July 2009 and the second from April to July 2010. I first performed an initial survey in the form of a questionnaire which enabled me to gather data on representations. This survey was performed throughout the county, in both rural and urban areas, and included all segments of the population. The purpose of the second investigation was to establish a linguistic corpus. First, spontaneous recordings of speech were made randomly in all parts of the county, and then data were gathered within a single network of family and friends, in the town of Suonanba and the village of Maomao two kilometers away. The purpose of establishing an initial corpus was, on the one hand, to become familiar with the language and, on the other, to ascertain whether ways of speaking differed from one area of the county to another. I focused the study on the second corpus, which allowed me to analyze a specific network in both a rural and an urban context.
11The second corpus is made up of spontaneous speech and planned situations. Interactions between family members and friends took place in the two main households, in Suonanba and Maomao. The aim was to observe and analyze contact phenomena between speakers from different age groups in a spontaneous context and to see whether they altered their way of speaking according to their interlocutor. In terms of the planned situations, a single speaker was asked to relate a story, tale or news item. This made it possible to observe whether the phenomena are the same, constant, or influenced by the context of the communication and by interaction with various interlocutors. All of the examples presented in this paper are extracts from the second corpus that were recorded only in Suonanba. All observations involved interactions between trilingual speakers (Dongxiang, Linxia Chinese and Putonghua) from the younger generations and their peers or elders. They were recorded in a household and at a restaurant. The main topics of conversation were family, work and education. A summary of participant profiles is provided in Table 1 below.
12The speakers recorded belong to the same network of family and friends and represent the various social and professional categories to be found in Suonanba (educated, illiterate, mobile young people, young immigrants, housewives, girls, etc.)
Speaker Profiles 
Speaker Profiles 
3 – Sociolinguistic Context
3.1 – Towns and Urban Areas in Dongxiang County
13As various researchers in the social sciences and geography have noted (including, for example, Topalov et al. 2010), notions of towns and urbanity are far from being universal and vary according to the place and society in question. It is thus important to describe representations of towns and urbanity in Dongxiang County in order to define the target of sociolinguistic research and address the question of linguistic practices specific to this developing environment. The Dongxiang community is organized administratively according to the size of the populations that reside there and in line with China’s official administrative subdivisions: xiāng 乡 (township), cūn 村 (village), zhèn 镇 (town) , xiàn 县 (county seat), and shì 市 (city) . The largest urban entity that exists in Dongxiang County is the zhen, while the town of Linxia is a xiàn jí shì 县级市 (county-level city). Before these subdivisions were introduced by the Communist government in 1949, i.e. before the establishment of the PRC, when the terms were borrowed from Chinese, only one term existed – the baza (town, from the Turkish bazar), which was used to refer to Linxia in particular. Since their awareness of the world was limited to this region, the Dongxiangs required only one term to refer to the only town they knew. Today, this term has become synonymous with Linxia, and other nearby towns (Lanzhou, Xining, etc.) are referred to as shi (city).
Regional Map of Dongxiang Autonomous County 
Regional Map of Dongxiang Autonomous County 
14According to these definitions , the zhen represents an intermediate stage between countryside and town; but this term contains a clear notion of urbanity, since residents of a zhen are given an urban hukou  (chéngshì hùkǒu 城市户口). However, while the term zhen is linked to a notion of urbanity from an administrative point of view, in reality, levels of development vary from zhen to zhen: while some only display limited urban development,  in others the process of urbanization is well under way. For inhabitants, the distinction between shi and zhen is related to the size of the town. In other words, they understand these two terms in accordance with the official definitions and do not associate them with a given level of development. It is in fact the notion of urbanity that is not defined in the Dongxiang community, as it is conceivable only in association with the term shi (city). Although urbanization in Linxia has reached a very advanced stage, for the inhabitants of Dongxiang County it is not a town like any other: it is still the baza. It is thus understandable that the inhabitants of a zhen in Dongxiang are unable to think of themselves as living in a town and leading an “urban” lifestyle. This perception of urbanity should also be placed into perspective in relation to two other factors: on the one hand, the notion of territoriality on the scale of China as a whole, and on the other, the way Dongxiangs perceive their own community. The immensity of China’s land area and the size of its population are factors that define, on a scale much larger than elsewhere in the world, notions of size in relation to urban areas. The city of Lanzhou, for example, which is the capital of Gansu province and has more than 2 million inhabitants, is generally considered to be only a small and moderately developed city, while in France it would be seen as a very large urban agglomeration. Moreover, even though the Dongxiangs – like many ethnic minorities in China – feel oppressed by the Han ethnic group, which composes 98 per cent of the total population,  they also admire the economic development achieved by the Hans. Finally, as social actors, many Dongxiangs have a negative view of their community, which is much less advanced in terms of economic and urban development than other parts of the country.
15Given this context, they are unable to perceive a zhen in their county as a town for two reasons. First, the population of zhens in Dongxiang is not sufficiently large in comparison with other Chinese towns. Second, given the negative view certain Dongxiangs have of their community, they are unable to think of a place in Dongxiang in positive terms, i.e. as a town, which represents social and economic success. Nevertheless, from an objective point of view, one of Dongxiang’s four zhens – Suonanba – can be defined as a town, at least according to the definition given by Western researchers. For example, Calvet (1994) describes “a place where agrarian social structures are breaking down and being replaced by a domestic economy that creates new markets and migrant flows.” This is indeed what has happened in the case of Suonanba, whose status as a county-level town draws workers and students from all over the county, as well as Han business owners and their families, thanks to the presence of political institutions and increasing numbers of businesses, residential buildings and educational facilities. Suonanba has approximately 15,000 inhabitants and its population is growing by around 4.5 per cent a year  largely because of the inflow of migrants. This is a new phenomenon that had never previously existed in these parts of the county.
Various zhens and villages in Dongxiang Autonomous County
Various zhens and villages in Dongxiang Autonomous County
3.2 – Plurilingualism, Economic Development, Sociolinguistic and Behavioral Changes
3.2.1 – Plurilingualism and the Development of the Dongxiang Language
16The Dongxiang language is regarded as a Mongolic language, based on thirteenth-century Middle Mongolian, and has been strongly influenced by the Linxia dialect, a dialect of northwestern Chinese group which is mutually intelligible with other dialects in the same group (Field 1997; Bao 2008). In lexical terms, loan words from Linxia Mandarin account for approximately 34 per cent of total vocabulary,  the remainder being divided among words of Mongolian origin (which form the main body of the lexis), words of Turkish origin, and Persian and Arabic loan words. In phonetic terms, Dongxiang shares 93 per cent of its phonemes with Linxia dialect (K. Field 1997). As with other Mongolic languages, the word order is SOV. Linxia dialect was influenced by and adopted SOV word order through contact with the Dongxiang language and other Mongolic languages in the surrounding region.  However, the Linxia dialect and Dongxiang language remain mutually unintelligible. The Linxia dialect is the lingua franca for the region as a whole, while Dongxiang is a vernacular language mainly used in Dongxiang County.
17This reciprocal influence is due to two major socioeconomic and religious factors. First, Linxia’s status as the region’s historic economic center has resulted in the regular convergence of surrounding populations on the town, mainly for reasons of commerce. Regular and recurring contacts between populations have led to an influence of Mongolic languages in particular on the local Linxia dialect. Second, the influence of the Linxia dialect on the Dongxiang language is mainly due to the desire of the Dongxiang community to move closer, on a religious and cultural level, to the neighboring Muslim community, which is essentially made up of speakers of that dialect. At the same time, geographic isolation and the desire of speakers to reduce contact with other, non-Muslim communities have enabled the Dongxiangs to preserve their language until very recently from the influence of other northern dialects, particularly Putonghua, the political and linguistic cement of China. Like other researchers (Field 2007; Bao 2008), I believe that the influence of Chinese on the Dongxiang language may result from intensive Mongolian-Linxia dialect bilingualism among a significant proportion of Dongxiang speakers, which has encouraged the appearance of loan words and code-switching in speech. Some of these phenomena later became permanently integrated into the language, meaning that they were subjected to “grammaticalization,” to adopt the term used by Auer (1999). Dongxiang-Linxia bilingualism still exists and is widespread within the community, particularly in Suonanba. Moreover, in 1949 Suonanba became the hub of the county’s administration, whose cadres were mainly trained in Linxia. Given this context, phenomena specific to multilingual communities, such as widespread switching between the Dongxiang language and Linxia dialect (Bao 2008), have frequently been observed.
18The economic boom that began elsewhere in China at the beginning of the 1980s has recently reached Dongxiang County. Development is opening the area to the rest of the world and changing the social situation, on the one hand by increasing the number of activities available and by modifying behavior, and on the other hand by creating new language contacts, in particular with Putonghua and northern dialects. The spread of these linguistic varieties is particularly affecting younger speakers, who are the most exposed to them. We can therefore say that a trilingual community is now emerging, within which the main social actors are the youngest members of the community.
3.2.2 – Diversification of Activities and New Language Contact
19For about a decade, economic and urban development has seen the appearance of new activities that are changing both the socioeconomic context and the sociolinguistic situation in Suonanba. During our field investigations,  we observed that education, mobility, urbanization and increased purchasing power are all factors that have encouraged the spread of Putonghua and the northern varieties in Dongxiang and created a new generation of speakers in Suonanba. First, an increase in the number of schools in which classes are taught in Putonghua has enabled this standardized variety to spread more intensively and regularly than in the past. Previously, teachers spoke Putonghua poorly or not at all, and used the Linxia dialect or a linguistic compromise that combined several varieties of Chinese (often believing it was Putonghua or a close variant thereof), as the main means of communicating with students (Chen 2006). Today, economic development has increased the standard of living and salaries in Suonanba, thereby attracting teachers who are better qualified, but who often use a northern dialect according to their place of origin, rather than standard Putonghua. In any case, if education can play a role in the spreading of exogenous forms, this essentially occurs among the youngest members of the community whose families have the financial resources to invest in the education of their children. Moreover, state and private investment in the construction of infrastructure such as apartment buildings and stores, as well as an increasing number of service businesses, such as hotels and restaurants, are attracting qualified workers from Linxia and other parts of the province, who mainly speak one or more northern dialects. Unskilled workers and laborers from Dongxiang, who comprise the local workforce, work side by side with these speakers and acquire the language varieties through contact with their bosses and managers. They are mostly young men and women who have not yet started a family.
20In parallel with these changes, the opening-up reforms  of recent years have created a new, mobile flow of unskilled workers who leave the county to find short-term employment in the province’s other towns. The members of this workforce acquire other varieties of Chinese and, when they express themselves in Chinese, use a kind of “linguistic compromise” that combines several varieties belonging to the northern group of dialects. This change mainly affects young men, and it should be noted that it concerns all of the youngest men in Dongxiang, not only those from Suonanba. Finally, increased purchasing power has enabled a section of the population to buy television sets that show programs broadcast in Putonghua, as well as cell phones that display only Chinese characters. These “new technologies” play a key role in the acquisition of Putonghua, particularly for housewives who have not had access to education and who are less mobile than men for social and religious reasons. 
21Thanks to this urban and economic development, the diversity of socioeconomic activities has, in recent years, led to the appearance of a trilingual generation. Younger speakers are inherently more mobile and educated than speakers of previous generations, as opportunities for travel and education were virtually nonexistent before the past decade. Until 2003, residents of Dongxiang (living in both zhens and villages) were holders of a rural hukou, which considerably limited opportunities to move to towns. Elsewhere, the illiteracy rate in Dongxiang fell from 98 per cent in 1980 to 60 per cent in 2002 (Li et al. 2007), which demonstrates the efforts made by local government in terms of education. Moreover, new technologies such as television, cell phones (SMS), and to a lesser extent the Internet have spread within the community of Suonanba. However, technologies that require the ability to write can reach only the educated and well-off, which again means the youngest members of the population, those who were privileged enough to go to school. There is therefore a nascent trilingualism, but it is not shared by the whole of the linguistic community: it chiefly affects the youngest people, who have greater access to activities that enable the acquisition of Putonghua and varieties of northern Chinese dialects.
3.2.3 – Social and Linguistic Behavioral Changes
22The new activities discussed above are leading to the emergence of new social behaviors, particularly linguistic practices. Although social actors are categorized according to their professional activities (religion, commerce, farming, teaching, etc.), there are no defined social classes within the community (Chen 2006). Similarly, there is no social categorization of “youth” in the Western sense of the term, i.e. those in a transitional stage between childhood and adulthood who make up an intermediary age category with certain social behaviors in common.
23In the Dongxiang language, there is only one Chinese loan word, nienqin (young), which refers to a person who is comparatively younger than another person, but this term does not refer to a defined social category. The transition from childhood to adulthood is above all related to marriage, which generally takes place between the ages of 16 and 18 for women and 21 and 30 for men. At the same time, professional activity may also be used to categorize somebody as an adult: a 16-year-old who has a job may be considered an adult, while a young 17-year-old married woman is equally so. Elsewhere, while a certain number of attitudes, moral qualities and dress codes are specific to and shared by the oldest segment of the population, the same cannot be said for the youngest members of the community, which do not form a homogeneous and socially categorized group in the same way as their elders. Indeed, individual attitudes within the same age group can vary greatly according to the person’s role and activity, which means that existing social categories are related first and foremost to individuals’ social roles, rather than to the age group to which they belong. Given this context, it is easier to understand why categories such as “educated,” “mobile worker,” “housewife,” and “girl”  are defined, while the category “youth” is not.
24Variations in attitudes can therefore be observed in people from the same age group, according to their gender, whether they live in a rural or urban area, whether they have worked in a town, or whether they are educated. There has been no specific research on this subject, but we can say that there is a difference between the young and old – particularly in Suonanba – which can be observed through behaviors. The behaviors to which the social actors themselves referred most often during surveys were ways of dressing and linguistic practices. We observed that the wearing of a chachia  can reveal a kind of identity claim among young mobile workers, who use this religious sign to distinguish themselves from Han Chinese and align themselves more closely with other Muslim ethnic groups (the Hui in particular) when they are outside of their community. Among women, the wearing of a toque as a religious sign instead of the traditional cowl, but also of jeans, beaded clothes, high heels, makeup, etc., is directly linked to economic development and opening up. Through their clothing, they reveal their (relative) emancipation in comparison to traditional society and express their desire to display an urban and modern identity under the influence of contact with Han Chinese. These clothing habits are characteristic of certain young women from Suonanba whose purchasing power has increased, enabling them to buy products other than basic necessities. There are thus a certain number of clothing habits which are peculiar to younger individuals in Suonanba and which are not shared by all Dongxiangs in the same age group elsewhere in the county.
25In parallel with these new clothing habits, new styles of speech can be observed which are specific to the younger generations who have had access to mobility, education, work, new technologies, etc., and live in Suonanba. A certain categorization of speakers in territorial terms is now becoming possible from a scientific point of view, as well as from the point of view of speakers themselves. In other words, a form of the Dongxiang language specific to Suonanba is beginning to appear, one which can be identified as such by both speakers from Suonanba and other parts of the county. These new ways of speaking include instances of phonic variation, loan words, code-switching, and calques, which will be presented and analyzed below.
26When listening to recordings of conversations made in the field,  both young and older speakers expressed views such as “they speak like that in Suonan”  or that “that’s not real Dongxiang.” The first comment is related to the notion of territoriality discussed earlier, while the second implies that there is a standardized form of the Dongxiang language. The Dongxiang language has not been standardized, however, and there is therefore no prescribed or prescriptive norm linked to such standardization. Furthermore, ways of speaking Dongxiang vary from one village to another,  and no one variety enjoys more prestige than another.  It should be noted that, while speakers are able to identify these manners of speech as specific to the zhen of Suonanba, they are unable to identify or describe precisely how they differ from ways of speaking in other villages (in terms of accent, vocabulary, syntax, etc.). As stated earlier, these speech styles are not shared by the entire Dongxiang community, even in Suonanba, where they are often associated with the practices of young people.
4 – Description of Practices 
27These judgments indicate that there is a way of speaking specific to the youngest speakers in Suonanba and which is identified as such by other speakers. However, our observations show that, while new social and linguistic behaviors are appearing among young speakers, they are not shared by all. New practices can be associated with categories that are related to young people’s social activities (educated, mobile worker, immigrant worker or housewife); this does not mean, however, that the ‘youth’ category is socially relevant from the point of view of either speakers or the researcher. These new ways of speaking, whether conscious or not, are for the most part linked to the influence of varieties of Mandarin recently introduced into the community. Changes in linguistic behavior appear in concrete form in phonic, lexical and syntactic terms.
4.1 – Phonic Variations
4.1.1 – Speech Rate and Phoneme Production
29One of the characteristics of these ways of speaking is a more rapid rate of speech among young speakers in comparison with older speakers.  In terms of phonetics, this results in variable production of certain phonemes. The most common phenomenon is the delabialization of labiodentals [f] into velar fricatives [h].
30The following is an extract of a conversation with Fatima, a 56-year-old illiterate housewife, and her niece, Maciyan, a 22-year-old student in the last year of high school:
31Regardless of the age of the person she is speaking to, Maciyan always pronounces the word for ‘marry’ dahhala instead of dafala. This phenomenon has nothing to do with the influence of Chinese on the Dongxiang language and is related to certain historic phonic modifications, which have subsequently been adopted into the language. However, the speech rate of the eldest, even when they speak rapidly, is never associated with such phonic modifications.
32Elsewhere, we have noted several examples of elisions related to speech rate, such as:
33Rapid speech, common among the youngest speakers, seems even more so among young speakers in Suonanba. Youths in rural areas tend to adapt their speech rate when speaking to their elders, while those in Suonanba do not.
4.1.2 – Phonic Variation and Code-Switching
34We noted that certain educated young speakers from Suonanba, when speaking Dongxiang, produced phonemes from a Linxia Chinese loan word in the same way as they did in Putonghua when that loan word already exists in the language. The most common phenomenon is the transformation of alveolar fricatives [ʤ] (written as j in pinyin) in integrated loan words from the Linxia dialect, in front of the phoneme [i], into an alveolar stop [d] (written as d in pinyin), as in Putonghua.
35In the following example, Fatima and Onguo (a 22-year-old student in the last year of high school), who are mother and daughter respectively, use different loan words for the word “telephone” (indicated in bold):
36When speaking Dongxiang, Fatima and Onguo maintain differing pronunciations of the loan word “telephone” – i.e. jianhua (integrated loan word from the Linxia dialect) for Fatima and dianhua (spontaneous loan word from Putonghua) for Onguo – regardless of whom they are addressing. Similarly, for the following terms, Fatima always uses the integrated loan word, as in the Linxia dialect. However, Onguo mainly uses Putonghua pronunciation, even though her pronunciation may vary depending on whom she is speaking to. They are words built on the same root – diàn 电 – which means “electricity.”
|j)||jianshi||dianshi or jianshi||电视||‘television’|
|k)||jianying||dianying or jianying||电影||‘film’|
37These phonic variations mainly concern the most recent loan words,  since Chinese-integrated loan words (from the Linxia dialect) that are related to religion or customs are still pronounced in the same way by all speakers; for example jingcha (tea given as a gift at engagement ceremonies) is never pronounced dingcha, as it could be if it were pronounced as it is in Putonghua. On the other hand, the integrated loan word from the Linxia dialect jian ‘electricity’ is always pronounced jian by Onguo, since electricity appeared in Dongxiang County at the end of the 1950s. All loan words from Putonghua that we observed are related to new technologies. Their appearance in the Dongxiang language should be understood in relation to economic development and the opening up of the county to the outside world. Fatima, who does not speak Putonghua well, tends to “Linxianize” these loan words, i.e. she applies the phonetic rules specific to loan words from the Linxia dialect (d [d]> j [ʤ], in front of [i]) to loan words from Putonghua. In other words, she integrates them into the language by regularizing their form as if they were loan words from the Linxia dialect.
4.2 – Code-Switching: Repetition and Substitution
38Certain youths from rural areas modify their speech when they are in towns. They use more Linxia Chinese words to indicate to their interlocutor that they are more “urban”, or rather to hide their rural origin and/or the fact that they are uneducated.
39In the following example, a 19-year-old waiter, from the zhen of Narisi,  is talking to his manager, Tongtong, an educated 21-year-old woman. The young man has never been to school, learned to speak the Linxia dialect after arriving in Suonanba (three years earlier), and understands Putonghua: 
40In this example, he uses two successive segments with identical grammatical structures. The two segments are made up of the same lexical units, both originating from Dongxiang, and only the lexical unit “food” is variable. In the first segment, he uses the word budan, of Dongxiang origin, and in the second he uses cai, the Chinese equivalent. 
41The same phenomenon of repetition/substitution was observed with Ma Hong, a 23-year-old female from Suonanba who is illiterate and learned the Linxia dialect and Putonghua with her brother and sister, who both went to school. Ma Hong is speaking to her sister Onguo, a young student in the last year of high school.
42The sentence is made up of two segments where the group [verb + gie] is used each time. In the first segment, the verb (beye jiere olune) is of Dongxiang origin, and in the second, the verb (huaiyun) is Chinese. The Dongxiang language has two homonymous verbs gie. One is a discursive linking verb and the other is an auxiliary verb that enables the introduction of Mandarin loan verbs (Field 1998). In the first segment of the sentence, the verb gie is clearly a discursive linking marker, while in the second case, its syntactic role is ambiguous: it can be understood both as an auxiliary introducing the verb huaiyun “to be pregnant,” which is of Chinese origin (huai yun 怀孕), or as a discursive linking verb. Indeed, the auxiliary verb gie enables disyllabic verbs of Chinese origin to be introduced in speech and transformed into integrated (or established) loan words where there are no equivalents of Dongxiang origin (Field 1998). In this case, there is a Dongxiang equivalent and it is used in the first segment of the sentence: beye jiere olu. The use of the verb huaiyun therefore has no justification in lexical terms. In the second segment, the use of the verb huaiyun can be understood as a spontaneous borrowing where the verb gie- that follows it is thus a discursive linking verb; in other words, as in the case of the speaker in example (m), a grammatical structure is repeated, with the verb of Dongxiang origin in the first segment replaced by an equivalent Chinese word in the second segment. The use of an expression whose lexical units are entirely of Dongxiang origin (beye jiere olune), and the subsequent replacement of that expression by its Chinese lexical equivalent (huaiyun), may indicate the speaker’s desire to demonstrate his knowledge of Chinese – to demonstrate to his interlocutor that he is “from town” and knows how to be urban and educated. The investigations and related observations show that this repetition phenomenon exists only in Suonanba; speakers of the same age who live in rural areas do not use it.
43Such changes have been observed elsewhere. For example, Thiam (1994) describes a similar phenomenon in Senegal, where he demonstrates that the town develops a specific way of speaking one or several languages. Among other things, he points to the fact that there is a way of speaking Wolof or other languages of Senegal specific to Dakar. There is a certain prestige attached to town-dwellers’ ways of speaking, and the language of the town therefore becomes a model that speakers copy and imitate to indicate their integration into urban life. Elsewhere, Auzanneau (2001) notes that the integration of English and French elements into a Wolof structure can create a positive image for youths in Senegal, who thereby claim their identity as urban dwellers. The use of Linxia Chinese and Putonghua in Suonanba is more widespread than elsewhere and speakers seem to consider that the introduction of Chinese units in discourse reflects a way of speaking typical of Suonanba. The phenomena of repetition/substitution are particularly noticeable among young illiterate women and young male workers from rural areas. In employing them, such speakers may wish to display an urban identity because they feel inferior to educated people. But such phenomena are observed only in particular communicative contexts where the speaker finds himself, consciously or otherwise, in a position of “inferiority” to his interlocutor.
4.3 – Code-Switching and Syntactic Variation
4.3.1 – Amalgamation of the Functions of Endings and Homonymous Copula -de
44In Dongxiang, the ending -de, an indication of the locative dative, is placed after the noun while the verbal ending -de (or -dene), its homonym, an indication of the immediate, is always placed after a verb and never after an adjective. These endings can never be used at the end of a sentence and always introduce another segment. In Putonghua, the enclitic particle de (的) can indicate a determiner, possessive or genitive. It is placed directly after a verb, noun or adjective. It can also be used at the end of a sentence as an indication of affirmation.
45We observed that certain young and educated Dongxiang speakers use this ending in the same way as the Chinese enclitic particle de after adjectives, nouns and verbs: Chinese grammatical functions are thus combined with a Dongxiang homonym.
46In this example, Maciyan uses -de after an adjective as a marker of relative clause. The position of the particle -de when used as the Dongxiang ending -de does not reveal the substitution in morphosyntactic terms. On the other hand, in terms of meaning, it is clearly a loan word from Chinese. The similar position of the Chinese enclitic and the Dongxiang ending enables the syntactical order to be preserved.
47In the following example, in a conversation between Fatima and her son’s friend, Ma Qiang, a 19-year-old student in the final year of high school, -de is used after a verb and an adjective placed at the end of a sentence:
48Here, -de is used as an emphatic marker of affirmation. The position of -de after the adjective and the verb does not contradict the morphology of the Dongxiang language, which is agglutinative, but it drastically changes syntax, since usually only a verb with an aspect marker can end a sentence. These ways of speaking were observed only in young and educated speakers from Suonanba. Other trilingual speakers observed (immigrant or uneducated youths) did not express themselves in this way and use the same turns of phrase as older speakers.
4.3.2 – Calques
49As the analysis of this phenomenon is complex and extensive, we will take here as an example the case of certain calques of resultative verb structures  from Putonghua (Lefort, forthcoming C) and, in particular, cases where the verb gholu (to transform) is involved.
50In Dongxiang, the verb gholu (to change or transform) is used to indicate a change in state. It is always placed at the end of sentences. In northern Chinese dialects, the verb chéng 成, which means “to change” or “transform,” as well as “to become” or “succeed,” can be placed after certain verbs and can play the role of a resultative verb, where it expresses change or success: it is therefore found in V + cheng structures. Certain young speakers from Suonanba use the verb gholu in the same way as the Chinese verb chéng, with an extension of the meaning: gholu no longer simply means “to change into,” but also “to succeed.” Furthermore, these young speakers use it in the same way as in Chinese, with a “V + gholu” structure. It is a calque of the Chinese resultative structure. For example, qianji-Ø gholugva-zhuo, which means “to succeed in signing,” is calqued on the structure of the Chinese qian cheng (签成) “to succeed in signing.” The oldest speakers would simply say qianji-gva-zhuo “to have signed.”
51More complex structures can be found, which result from both code-switching and borrowing. The verb gholu can be combined with the verb bienji “to change” (an integrated loan word from the Linxia dialect) in constructions where the structure is fixed, as follows: [S + bienji-SEM (DIFF.EVENT) (ÉVEN. DIFF) + OBJ + gholu-SEM / ASPECT]. This type of calque is used only by the youngest speakers, whether educated or otherwise, whereas the oldest speakers use only the verb bienji or the verb gholu and do not combine both in a single sentence.
52We also noted other utterances resulting from language contact that are the result of both borrowing and code-switching, as in the following example, where two sisters, Ma Liping (26 years old, illiterate) and Onguo (22 years old), are having a discussion.
53In this example, the integrated loan word bianji (to transform) is replaced by a spontaneous loan word from the northern varieties biancheng. This is borrowed and inserted as a single lexical unit, whereas in Chinese it is made up of two independent verbs, the second combined with the first to indicate its result. This is, therefore, a spontaneous loan word (borrowed from the Putonghua biancheng), as well as an example of code-switching (the integrated loan word is replaced by its equivalent in Putonghua and the word order is modified).
54Such phenomena could be found among all the young speakers from Suonanba, educated or not, whom we met; on the other hand, they were not present in the discourse of immigrant speakers.
5 – Towards the Social Categorization of a “Youth Language” or “Urban Youth Language”?
55As demonstrated above, there are ways of speaking specific to the younger generation in Suonanba. However, these do not constitute “slang” speech or a secret language shared by a group of youths seeking to recreate their own world through language, as has been observed in other urban societies. Even though certain linguistic phenomena such as substitution/repetition seem to be linked to a kind of identity-related speech, comparable to those described in other parts of the world (Bulot et al. 2002), they appear only in certain specific communicative contexts. They seem to characterize individuals rather than participate in the creation of a collective identity, even though it is conceivable that, through mutual influence, personal behavior could contribute to such a construction in a kind of formal and/or functional convergence. Further, these instances of substitution/repetition are not shared by all of the youngest speakers; they can thus be regarded as constituent parts of these new ways of speaking, but they are not specific to them.
56The diversity of phenomena encountered and the heterogeneity of these practices reflect the complexity of the Dongxiang community, which is undergoing rapid change and forming new categories of social actors (mobile labor, immigrants, educated people, etc.) that had not previously existed. With the formation of these new categories, new ways of speaking specific to Suonanba’s youngest speakers are being identified, though these ways of speaking have not really been categorized in terms of either the speakers’ representations or their language practices. In other words, despite the fact that new linguistic feature, identified by both speakers and researchers, are clearly appearing in the practices of the young generation, these feature are not contributing to the construction of a “youth” social category or a “youth language” sociolinguistic category.
57There may be several reasons for this. First, the phenomena observed, such as borrowing, code-switching, and resulting grammatical modifications, are in no sense a feature specific to “youth,” since they are found in most plurilingual societies around the world (see Kerswill 2010, De Féral, this issue), without necessarily being related to speakers’ ages. Elsewhere in Dongxiang, they are also used by older speakers, but mainly with the Linxia dialect. However, given the similarity of word order in Linxia Chinese and Dongxiang, syntactic variations are less frequent when these two languages are involved in discourse. On the other hand, young people’s ways of speaking give rise to more salient innovations, since the languages they use are less closely related in grammatical terms. The Dongxiang language has a purely oral tradition; there is no standardized norm or ideology of a standard, which leaves room for all kinds of linguistic innovations, particularly in lexical terms, and has facilitated code-switching to a certain extent. Elsewhere, the Dongxiang language in its present state is the result of intensive contact with the Linxia dialect. It therefore seems that new behaviors are a result of implied varieties rather than contact, which has always shown through in uses. The reason why these practices are seen in only a section of speakers – in this case young people – is that this group is more exposed to new varieties of Chinese, owing to the recent emergence of new social activities to which older speakers have less access.
58Elsewhere, since linguistic change seems to be occurring at the same frantic pace as economic development, the population as a whole has not yet gained the perspective necessary to identify these new language practices and associate them with a particular group. In addition, the fact that new linguistic traits are not shared by all of the speakers in a given age group in no way encourages the identification and classification of these ways of speaking. The rapidity of social and linguistic change sets the linguistic and sociolinguistic dynamics of this region apart from those of Western countries, where changes are slower and have been taking place for longer (see Le Page and Tabouret-Keller 1985).
59If we add to this the fact that there is no notion of “urbanity” in the Dongxiangs’ way of thinking and that there is no socially constructed category specific to youths, we can understand how the population as a whole does not attribute the new language practices to a category of people who could be described as “urban youths.” As this category does not already exist among the Dongxiangs, it seems difficult for the researcher to create one on the basis of language practices observed mainly among young speakers. This point raises the question of the validity or usage of pre-constructed social categories by researchers in a given situation (see Auzanneau, Juillard and Leclère-Messebel, and De Féral, this issue).
60Therefore, in the context under study, the researcher can question the relevance of constructing a “youth language” category. Can we “impose” a scientific categorization based on observations and analyses specific to Western societies when such a categorization lacks an equivalent in the society being studied? In the case of Dongxiang in particular, there is a severe lack of perspective on the situation, and we still do not know how these linguistic features will evolve. As a result, we should be cautious when drawing conclusions: it is easy to imagine that economic development will spread to other parts of the county and that these linguistic traits will be adopted by other speakers in the same age group as the urban/rural barrier is eroded. One could go further and imagine that new technologies and social mobility will become more widespread and that even the oldest speakers will adopt these new ways of speaking. In such a transitional situation, it therefore seems unnecessary to want to construct a scientific category at any cost.
6 – Conclusion
61Dongxiang County is in the throes of major economic and social change. The process of urbanization that has recently begun is increasing the number of activities available throughout the county, and these are creating new types of language contacts. These mainly involve northern Chinese varieties as well as Putonghua, and are giving rise to new linguistic features that can be observed among speakers in a certain age group – young people – living in Suonanba. While these linguistic features are not shared by all speakers in a defined category, they are distinct from those of the oldest speakers and are identifiable as such. In concrete terms, they are represented by various phonetic, lexical and syntactical phenomena, such as assimilation, code-switching, borrowing, etc. These features are not specific to young people or town dwellers, but rather reflect the state of sociolinguistic transition that is affecting a certain section of Dongxiang’s population.
62The context of rapid change in society as well as language means that these linguistic phenomena cannot yet be definitively attributed to any group in particular and indicates that further, as yet unforeseeable, changes are likely. Elsewhere, the absence of a socially constructed and stabilized youth category and of a notion of urbanity does not favor the development of a consistent categorization of these linguistic traits by Dongxiang society. A scientifically constructed “youth language” therefore seems unnecessary in this context, or even liable to create an artefact, as has occurred with many other descriptions of so-called “youth language.”
List of Abbreviations
632SG: personal pronoun, 2nd person singular
643SG: personal pronoun, 3rd person singular
75PAST NOM.: past nominalizer
76ENCL PART.: enclitic particle
81SEM: switch event marker
82REFL.SUFF.3SG: reflexive suffix, 3rd person singular
84- Ø: integrative
People who have not attended school and can neither read nor write Chinese characters, the Arabic alphabet or any other writing system.
Note on the original French: [There is no perfect equivalent for this word in French, which can be translated as bourg (large market town) or petite ville (small town)].
The transcription with an accent gives the pronunciation in Mandarin Chinese. In the rest of this paper, the words zhen, xian, etc. without accents, are lexicalized.
The maps included in this paper were produced by the author.
“Administrative division under the direct authority of the xian.” Bo, Gujin et al. (2008, 1274).
A kind of family identity card that indicates a household’s permanent residence and, depending on the type, allows certain privileges.
In this case, rural characteristics dominate: while there are buildings, stores, and official institutions, the roads are not asphalted, animals wander the streets, etc.
Official figures from the Chinese government website, consulted on October 10, 2009: www.gsjsw.gov.cn.
Official figures from the Chinese government website, consulted on October 10, 2009: www.gsjsw.gov.cn.
Figures given by researches vary from 40 to 60 per cent, but our estimates correspond to those of K. Field (1997), who mainly bases them on analysis of vocabulary of the Dongxiang-Chinese dictionary by Ma and Chen (2001).
Researchers’ opinions differ on this point: some think that Mongolic languages influenced the Linxia dialect (Luo 2004), while others think it could be the result of imperfect learning.
The first part of the investigation mentioned above, which was conducted in 2009 and 2010 and involved 107 speakers, including 52 residents of a rural area and 55 residents of an urban area, i.e. in possession of an urban hukou; I interviewed 57 women and 50 men. The results obtained and presented herein reflect only the population sample interviewed and cannot be considered to be representative of the entire Dongxiang community, even though they do throw light on the nature of the plurilingual community being studied.
Reform and opening-up policies (gǎigé kāifàng 改革开放) were introduced in the PRC in 1978 and have since been applied to various extents in different regions. It should be noted that Dongxiang Autonomous County, as a “special zone,” did not benefit from opening-up policies until the end of the 1990s.
The Dongxiang community is mainly Muslim and very conservative; the role of women is often confined to that of housewife.
This term refers to uneducated single young women who are not engaged in a professional activity and remain at home to serve their families.
A hat worn by Muslim men to express their religious belonging – equivalent to the Jewish kippah.
I played recordings from the second corpus at the request of certain speakers, after the final compilation in July 2010, without having developed a real investigation beforehand.
‘Suonan’ is an abbreviated form of ‘Suonanba.’
Speakers are generally able to recognize another speaker’s village of origin by his or her way of speaking, accent, etc.
Only Arabic enjoys a superior status in relation to other languages, due to its religious role. It is important to note that it is not a lingua franca or language of communication; it is used only by imams during services, in association with the Linxia dialect.
Transcriptions of extracts are based on the pinyin alphabet from the Dongxiang-Chinese Dictionary by Ma and Chen (2001).
The pinyin alphabet is used in this paper for the transcription of extracts.
These observations were made by playing recordings from the first corpus to other speakers; the latter identified more rapid speech rates and explained the phonetic differences themselves by pointing out that there is a reduction of syllables in certain words. No specific study of this phenomenon has been carried out.
Families from Suonanba began to own television sets only from the start of the 2000s, but this does not mean that televisions were unknown and unnamed before then.
Zhen located in the south of Dongxiang County.
Words underlined with a single line are of Dongxiang origin, while those with a double line are of Chinese origin.
The word used is the same in the two varieties: the speaker refers to both Putonghua and the Linxia dialect by associating the two varieties with “Chinese.” It is therefore impossible to determine if it is a loan word from Putonghua or Linxia Chinese.
As their name indicates, resultative verbs in Chinese express the result or consequence of the verb to which they are associated.