1Within the quickly evolving linguistic scenario of the Amdo sprachbund, Dongxiang, a variety of the Shirongolic (Southeastern Mongolic) branch, spoken by a Muslim minority of Gansu Province, has acquired idiosyncratic features, and is undergoing significant sociolinguistic changes. Among the contact-induced changes currently taking place, there has appeared a strong tendency of some speakers to lenite the uvular consonants, and a geographically uneven emerging development of tonal features – a profound novelty for a traditionally atonal language.
- 1 The majority of the Southern Peripheral Mongolic languages have been adequately described, as indep (...)
2As an independent form of Mongolic, like most of the Southern Peripheral Mongolic languages Dongxiang has been fairly documented from a typological perspective. This documentation was carried out during the mapping expeditions commissioned by the URSS (Union of Soviet Socialist Republics, the 1950s) and PRC (People’s Republic of China, the 1950s and the 1980s)1, though some scholars also documented it independently (Kuribayashi 1989; Field 1997; Kim 1998). The major pertaining studies comprise a few short grammars (Liu 1981; Bu He & Čingeltei 1985; Todaeva 1997; Kim 1998), one comprehensive reference grammar (Field 1997), some dictionaries (Bu He 1983; Kuribayashi 1989; Ma &Chen 2000), and a few collections of texts (Todaeva 1959, 1961, 1997; Bu He 1987).
3In contrast, the sociolinguistic implications are still under-researched, though the sociocultural factors contributing to this scenario have been recently explored (i.e. Roche 2014; Druijven et al. 2017, 2018; Ward 2019; Zhang 2020). And while the field of sociolinguistics has seen some contributions, including noteworthy works by Julie Lefort (2012, 2015, 2017), it should be noted that much of the existing research has primarily focused on morphology, syntax, and language contacts. Consequently, sociolinguistic investigations into the phonological and phonetic dimensions are still limited. While, for instance, Saren Bao (2006) offers a comprehensive examination of phonological changes in relation to Middle Mongolian and the influence of Chinese on the Dongxiang language, and Kenneth Field (1997) has contributed to a deeper understanding of shared syllables between Linxia Chinese and Dongxiang, there remains a limited body of work specifically addressing the phonetic situation of Dongxiang language. The present research proposes to explore the sociolinguistic implications of the language contacts on Dongxiang, focusing specifically on the above-mentioned phonetic variations, and exploring their possible sociological implications. It investigates the lenition of the uvular consonants and the emergence of tonality, as proposed sociophonetic variations among the Dongxiang speakers. The present case study, therefore, draws attention to the ongoing phonetic variations, and to the existence of sociolinguistic variables that are salient among a limited number of speakers, which provides a basis for future research.
4The following sections will explain the theoretical foundation to this case study, focusing on the most relevant concepts referenced, and will outline the most salient phonetic and phonological features of Dongxiang to frame the present discussion. Thereafter is provided a description of the economic and religious factors that may trigger the modifications under discussion, and of the stratified and complex linguistic scenario of the Dongxiang varieties. This study’s adopted methodological approach and the challenges of investigating such features will then be discussed. Afterward, the dataset of this case study will be presented and given a preliminary analysis of the distribution of the variants in the speech repertoire of the participants. The final discussion outlines the speaker perception and metalinguistic awareness as the ground for the diatopic divergence and presents relevant examples to show the actual phonetic differences.
5This article investigates the partial loss of uvulars and the emerging tonal prosody in Dongxiang as sociophonetic phenomena, thus claiming geographically dependent sociocultural causation behind these variations. It also suggests that a certain degree of awareness and consequent agency are responsible for these changes. In fact, being the uvulars and the atonality both distinctive traits of Dongxiang, they are stereotypical (Labov 1963), thus subject to conscious awareness of the speakers. Hence, I argue that the variations under discussion represent indexes to specific linguistic attitudes. In particular, considering that the changes are rather region-specific, assuming a sociophonetic perspective, the diatopic variation appears to index specific local linguistic attitudes and ideologies, more predominant and more relevant among certain communities and specific areas. I suggest that the variations imply metalinguistic awareness of the indexical meanings, and that should also be linked to a degree of agency, based on cultural and social constructs (Sicoli 2011).
6The proposed research is framed by the sociolinguistic and anthropological linguistic perspective and focused on the phonetics and phonology of Dongxiang. Within the sociophonetic perspective, this study intends to address in what way Dongxiang speakers, with different geographical and dialectal backgrounds, within different sociocultural environments, exploit their stratified linguistic-phonetic repertoire to convey extra-linguistic meanings. The theoretical underpinning of this research is that, generally speaking, linguistic changes involve sociocultural factors and are associated with social meanings. In fact, on the one hand, phonological variations have long been linked with social factors (cf. Labov 1963; Trudgill 1974; Romaine 1978; Guy 1981; Van Coetsem 1988). On the other hand, phonetic changes also typically index sociocultural values, and over the past two decades, the correlation of social meanings with phonetic (Stuart-Smith 2004; Hay & Maclagan 2006) and prosodic changes (see Szakay 2006, on pitch, and Warren 2005, on intonation) has been investigated more closely. However, though associated with social values, these variations can imply a different degree of awareness and be both conscious and unconscious (Kootstra & Muysken 2019). Socially driven phonemic variations in contact-induced language change (CILC), when they happen consciously, can happen to accommodate certain linguistic attitudes or social perceptions (Berry et al. 1997). In truth, they are rarely caused directly and exclusively by borrowings alone (Kennard & Lahiri 2020). A long-lasting process of borrowings joined with the aware perception of fine phonetic details can cause phonetic and phonological adaptations through “phonetically minimal transformation” (Peperkamp 2005). And, in relation to the degree of awareness plausible for phonetic features, it should be noticed that speakers, somewhat regardless of their educational background, have access to “fine phonetic details” beyond that related to their native language (Dupoux & Kakehi 1999).
7The current case study references indexicality and metalinguistic awareness as crucial analytic tools for investigating the sociological implications under consideration. The modern concept of indexicality was originally developed in the 1930s; however its original conceptualisation has been gradually implemented and interpreted over the century. In the last decades, it has been gradually adopted and extended by several 20th-century academic traditions, including those of linguistic anthropology. Within this framework, it has been used by linguists and anthropologists to talk about the social implications of certain linguistic expressions, demonstrating that these expressions either presuppose or entail particular social relations or social situations (Fillmore 1975). The concept is also well established in the sociolinguistic tradition, used as a tool to analyse how linguistic choices may index aspects of speakers’ or interlocutors’ identity (Eckert & Labov 2017). But, in recent years, this concept has drawn increasing attention also from other branches of linguistics new to this concept, and to explain several phenomena, even those traditionally not approached with this explanation (cf. Nielsen & Sansiñena 2021 for discussion). The current study forms part of this scenario in proposing to adopt the concept within the Amdo studies.
8On the other hand, the modern conceptualisation of metalinguistic awareness has been first formalised by Benveniste (1966). In Problèmes de linguistique générale, he addressed the issue of awareness when discussing a speaker’s ability to distance himself from the “object-language”, and recognise it as a communicative device, a code. Though metalinguistic awareness is generally divided into four categories (phonological, lexical, syntactic, and pragmatic), there is nowadays general agreement that it relies on analysed knowledge and cognitive control, rather than being a distinct set of language-specific abilities (Bialystok & Ryan 1985). And in fact, in general, a stronger degree of metalinguistic awareness is associated with language acquisition, especially with literacy (Nagy & Anderson 1995) and bilingualism (Bialystok 1988; Thomas 1992).
9The current research also relies on another fundamental concept, the tonogenesis, namely the appearance of lexical tone in a toneless language. The term itself was coined by Matisoff (1970), but the modern discussion about the emergence of tonal prosody was already lively in the 1950s, in regard to the development of tones in Vietnamese (Haudricourt 1954). Tonogenesis is usually solely linked to changes in phonation and the loss of other phonemic distinctions (Ratliff 2015). There are, however, exceptions to this model within the Amdo sprachbund: a Mongolic toneless variety of this area, Monguor, shows a tendency to mimic tonality by adopting “rudimentary tonal features” in Chinese loanwords (Dwyer 2008). In particular, Monguor shows an atypical tonogenesis, which seems to have been triggered by language contact and not exclusively based on internal phonetic changes. I argue that this is also the case with Dongxiang.
10The Amdo sprachbund is well known for its vivid and long-lasting contacts, which has been widely documented especially in the last two centuries. In the late 20th century, several scholars investigated the areal context, describing it for the first time as a whole, with its peculiar internal dynamics, and proposing in turn different names for it. Arienne Dwyer, for the first time, referred to the group of language entities of the area with the term “sprachbund” (coined by Trubetzkoy in 1928): “The extended contact of Tibetan, Chinese, Turkic, and Mongolic languages over the centuries encouraged the gradual development of similar forms in these languages; hence, the term sprachbund or ‘language association’ (Dwyer 1995, p. 144). Some years later, Juha Janhunen termed it both Qinghai Linguistic Complex (1) and Amdo sprachbund (2): (1) “Altogether, some fourteen to fifteen separate linguistic entities can be distinguished today within the Qinghai Linguistic Complex. Genetically, they represent four different language families: Sinitic (Chinese), Bodic (Tibetan), Mongolic, and Turkic” (2007, p. 86); (2) “The Amdo Sprachbund […] comprises ca. 15 languages and/or principal dialects spoken in the Sino-Tibetan borderland historically known as the Amdo Province of Ethnic Tibet, and today divided between the Gansu and Qinghai Provinces of China” (Janhunen 2012, p. 72). Keith Slater termed it Gansu-Qinghai sprachbund: “The languages of this Gansu-Qinghai border region, which originate in four language families (Mongolic, Turkic, Sinitic, and Bodic), are all converging towards one common set of structural features. The region may well be considered a linguistic area, or Sprachbund […] The research on the Gansu-Qinghai Sprachbund is only beginning” (2003, pp. 307- 308). Lastly, Hans Nugteren (2011) referred to it as “Qinghai-Gansu languages”.
- 2 This language (glottolog: hezh1244) is known as Hezhou language (hezhouhua 河州话) Linxia language (li (...)
- 3 It is mainly spoken by Hui, Bao’an, and Dongxiang.
- 4 Lanyin or Lan-Yin Mandarin (lanyin guanhua 兰银官话, glottolog: xibe1241) is a northern variety of Chin (...)
11Dongxiang has been spoken for centuries within this heterogeneous linguistic area and has been influenced and shaped by the continuous contacts. Due to the geographical features of the area inhabited by the Dongxiang and their linguistic attitudes, however, the language has been in no contact with varieties spoken by non-Muslim communities (Lefort 2012). This is particularly significant in relation to the Sinitic languages. In fact, in the area there are several varieties belonging to this family, but their impact on Dongxiang greatly differs. In the capital of the Linxia 临夏 Hui Autonomous Prefecture, Linxia City (once known as Hezhou 河州, Gansu Province), there is a rather mixed variety2 predominantly spoken by Muslim communities3, which therefore has strongly impacted Dongxiang, at least since the 15th century (Lefort 2017). On the other hand, Dongxiang has had minimal direct contact, within the limits of the Dongxiang Autonomous County with the northern Chinese varieties (Lanyin)4, until the 1990s, after which it has received exposure as a consequence of the urbanisation of the area and the growing internal mobility. This is also the case with Putonghua, with which Dongxiang people has had no contact, up until a few decades ago. The exposure to Putonghua has been also favoured and above all by its promotion. The emergence of these new Sinitic varieties has influenced the language and resulted in code-mixing (ibid.; McConvell 2008; Xu 2017).
- 5 However, /q/, though indigenous to Mongolic, also occurs in borrowings from Arabic and some Turkic (...)
12Dongxiang consonant inventory consists of 29 phonemes, of which three are non-indigenous (/tsʰ/, /ts/, /ᶎ/), and are only found in words of non-Mongolic origin. The system includes three uvular consonants, which are strictly language-specific sounds5: the plosives /q/, /qʰ/, and the rhotic /ʁ/. The uvulars may be subjected to variability. However, their realisation still falls under the uvular category. In more detail, the rhotic /ʁ/, which is a voiced uvular fricative, may be realised, especially in fast speech, as a voiced uvular approximant [ʁ̞]. The phoneme /q/, a voiceless unaspirated uvular plosive is generally not subject to variability. The phoneme /qʰ/, a voiceless aspirated uvular stop, before high vowels may be affricated and realised as /qχʰ/. The affrication of aspirated plosives is an areal feature, shared with Monguor, another Mongolic variety (Field 1997).
- 6 This is an oversimplification, as some of the Mongolic varieties do not allow clusters in the onset (...)
- 7 The Sinitic variety here mentioned is the Hezhou or Linxia variety (cf. note 2). In this instance, (...)
13Phonologically speaking, Dongxiang has lost vowel harmony, a typical Mongolic feature. Such simplification is detectable to different extents in many Mongolic varieties of the sprachbund (cf. Nugteren 2011). In Dongxiang, vowel harmony is still traceable in the structure of the words, since it is not uncommon to find a regularity of the vowel distribution in words of Mongolic etymology. However, it is reminiscent of a previous but no longer productive phonological rule. In fact, except for a limited number of cases, no harmonic rule governs the suffixation. Furthermore, Dongxiang does not show a Mongolic type of syllabic structure ((C)(C)V(V)(C)(C)6) because it does not allow consonant clusters in coda position and permits them in onset position only when the second consonant is a glide. Its syllabic structure is identical to the Linxia Hui7 ((C)(G)V(G/N)).
- 8 It is not found in Dagur and Central Mongolic.
- 9 The proportion between native words and loanwords is still under debate. For further discussion, se (...)
14Concerning the prosody, Dongxiang shows predictable word-final stress in words of Mongolic origin, which is shifted when a suffix is added. This stress pattern, on the ultimate syllable, is not typical in other Mongolic varieties8, except Moghol and the Mongolic varieties of the Amdo sprachbund (Nugteren 2011, p. 35). Because of that, it has been suggested that this stress pattern is an areal feature of the Amdo Mongolic languages (Field 1997; Nugteren 2011). The word-final stress pattern applies to words of Mongolic origin, which are still most of the Dongxiang vocabulary (59,35%)9. It, however, does not hold true for loanwords, which do not have a predictable stress pattern due to their original stress, or in the case of Chinese, most probably because of their original tones (Field 1997, p. 154). That being said, in regard to tonality, so far, Dongxiang has been considered toneless.
15“Language contact has often been claimed to be a causal factor of language change” (Meisel et al. 2013) and there is no denying that Dongxiang is impacted, as all the varieties of the Amdo sprachbund, by the language contacts happening in this linguistic area. “Yet, sociolinguistic research shows that the role of contact often turns out to be overestimated once the environmental factors conditioning the use of the relevant constructions are taken into account” (ibid.). Language change is a consequence of multiple factors: “internal constraints act jointly with external constraints in shaping language contact outcome” (Sankoff 2001), and this refers not only to the linguistic internal factors, but also to the sociocultural features of the speech community, which constitutes the ground for the language contacts and therefore conditions the impact of the language contacts.
- 10 Lefort (2012) has presented a parallel observation, drawing from the economic boom that took root i (...)
16In the case of Dongxiang, two major factors need to be considered as triggers of the variations under discussion: the religious background and the economic situation of the minority. Gansu is the third most Muslim province in China (1,38 million Muslims), and more specifically, nearly all the Dongxiang people are Muslims – 99,95 percent of the population of the Dongxiang Autonomous County, of which 88 percent is Dongxiang (China’s 2010 Census Data). This has unavoidably a great impact on the identity building of the Dongxiang people, and consequently their linguistic choices. Among them, some people may be less aware of their Mongolic background but consider their identity as strictly related to Islam (Druijven et al. 2017; Wang 2017). This has been already proven to have directly affected their linguistic behaviour because of the shared religious identity with the Hui minority: the religious affinity between them have played a significant role in the spread of the Linxia dialect and bilingualism within the Dongxiang community, which over time, have induced changes in the Dongxiang language (i.e. the syllable structure, cf. Field 1997). To an equal degree, the poverty of the area and the consequent economic marginalisation of the local minority languages represent significant elements participating in the CILC. The region is one of the most impoverished of China (2020 GDP per capita according to the National Bureau of Statistics of China is 36 038 CNY, making Gansu the last in the ranking) which, given the recent economic development of the rest of the Country, on one side has encouraged urbanisation in the area (Lefort 2012), and on the other side, has induced a high number of Dongxiang people to migrate looking for better economic opportunities (Ma 2004). In addition, Dongxiang is decreasingly the sole maternal language: in fact, though the number of monolingual speakers is steadily low since the 1980s (cf. Chen 2006), the prevalence of bilinguals (Dongxiang-Linxia) and trilinguals (Dongxiang-Linxia-Putonghua) appears to be on the rise10, suggesting a growing trend of multilingualism in the region.
17The sociocultural and economic context of the area has already provided fertile ground for language exchanges in the past, causing Dongxiang to be a strongly stratified variety. In fact, diachronically, besides exhibiting Mongolic and Turkic strata, the language has quite strong Sinitic features, and an influence, primarily due to religious reasons, from Arabic and Persian (Field 1997). And because of these contacts, it has acquired peculiar features, which in some cases are common to the whole area (i.e. word-final stress, Nugteren 2011, p. 35), and in some cases, are unique to Dongxiang (i.e. syllabic structure, cf. sect. 4 “Phonetic and phonological peculiarities of Dongxiang in the Amdo Mongolic domain”).
- 11 Yao 2020.
- 12 Diglossia is the use of two varieties in fairly strict compartmentalisation by a single language co (...)
18Synchronically, the language also shows a certain degree of variability; in fact, though it does not have proper dialects (fangyan 方言), three local varieties (tuyu 土语) have been recognised: Suonanba 锁南坝 (with about half of all Dongxiang speakers), Wangjiaji 汪家集 and Sijiaji 四甲集 (respectively, with about 30% and 20% of all Dongxiang speakers)11. These varieties share the same grammar but slightly differ phonetically and in terms of vocabulary (Liu 1981, pp. 4-5). In addition, I have observed in Linxia Prefecture (November 2019), among the bilinguals, a fluctuation between the languages. The varieties repertoire, however, does not seem to be based on two polar linguistic codes (Linxia and Dongxiang) but involves diglossia and dilalia12 – one prevailing on the other depending on the specific areas. I also observed an intermediate variation in the proficiency of the speaker, both in Dongxiang and Linxia, as the linguistic variation is a continuum rather than a polarised differentiation. In other words, depending on the dialectal background of the speakers, their hometown, and their membership in certain social groups, there is a significant variation in the context where they use Linxia, Dongxiang, or a combination of both. I argue that such diversity may be accounted for on the ground of sociocultural factors.
- 13 Zhong Xiang 钟翔 (b. 1966), literary double for Ma Zhongxiang 马忠祥, is a Dongxiang writer of Kangle Co (...)
- 14 Hashengerile 2014; Bao Saren 2006; Yondrol et al. 2020.
- 15 The author herself communicated in Mandarin Chinese, and when needed, the minority author Zhong Xia (...)
- 16 The jingtang yu, literally “the language of the scripture halls” is a special language developed by (...)
- 17 This system is known by several names: in Eastern and Northeastern China is called xiao’er jing 小儿经(...)
- 18 The development of a literary language for Dongxiang is founded on pinyin, but modelling it after t (...)
- 19 Suutarinen 2013.
19The research material predominantly collected has been audio material of prompted language production and sociolinguistic interviews. However, along with it, were also collected four books written by the Dongxiang minority author Zhong Xiang13, some additional material14. Mandarin Chinese in conjunction with the local common language, Linxia, was the variety used15 to conduct the interviews and ask the participants to perform specific tasks. The written material presented to them was also in Mandarin Chinese. In those circumstances where they were asked about Dongxiang, the words/sentences were always solely uttered, never presented in written form since the participants did not have any knowledge of the official script developed for the Dongxiang (Dunxian 东乡 pinyin, see Ma & Chen 2000) nor the few existing unofficial adaptations. In fact, on one hand, the unofficial scripts, which are the jingtang yu 经堂语– a medium of instruction developed to transpose Islamic texts into Chinese characters16, and the xiao’et jing 小儿经 (xiaojin; ثِیَوْݣٍْ)17 – a transcription system of Chinese into Arabic, were widely used in the religious sphere until the 1980s, among the Muslims of China, especially the Hui, but rarely known by Dongxiang speakers. On the other hand, in recent years an official Latin script has been developed for Dongxiang of the Mongghul and Monguor alphabets (Kim 2003)18, but this one is also not used among the common population because is not taught in school though it is used in educational material. This situation may be explained considering that among Dongxiang “basic education in the native language is still rare”19, since the Dongxiang community has always focused on Chinese, not Dongxiang (Chen 2006), making it rare to find a Dongxiang speaker able to read and write in these adaptations, besides the intellectuals (Suutarinen 2013). Besides, the literacy level is still extremely low: in 2000, 63,29 percent of the population had never been to school or had a low level of literacy in Chinese.
20The audio material was recorded with the aid of the portable recorder Zoom H4N Pro and stored in a lossless format (.wav) in order to carry out acoustic analysis on PRAAT. This software was chosen for the current case study because it allows formant analysis, particularly crucial in unideal settings, and pitch analysis, both fundamental to examine the phonetic variations under consideration. I collected the material with four approaches. Firstly, I recorded natural speech. I relied primarily on this approach because, although the phonetic changes under discussion are not an object of social perception, the least invasive approach was preferable given the subject of study. Secondly, through the task of translating fairytales, I get a production with more controlled phonemes, but still semi-natural. In this case, the consultant is generally more attentive to the production but not affected by hypercorrection due to the length of the task. Thirdly, through a translation of short, standardised phrases I got a production that was as controlled as possible. I decided to include this approach to have a basis for comparison with less controlled speech and to investigate the speakers’ awareness. Fourthly, I conducted sociolinguistic interviews. Among the data collected, I chose the data points with the more favourable signal-to-noise (SNR). As for the naturalness of the utterances produced, most of the data points were chosen on the basis of this parameter, but on occasion, when it was relevant, also more controlled utterances were selected – and in such cases, it is clearly stated.
21The participants were strongly willing to cooperate because, with different sensibilities, they were proud of their identities and interested in research on their language. However, there were a few factors to consider. Firstly, the recordings were possible only with unideal SNR levels due to the economic and environmental context. To investigate sociophonetic variations, performing acoustic analysis on material recorded in noiseless rooms would have been ideal. That being said, “research on language and interaction in the ‘chaos’ of ordinary activities has grown considerably in the past years” (Xiang 2021), given the importance of naturalistic speech. And the formant analysis is increasingly refining the criteria to identify specific sounds even in unideal SNR (Stevens 1978; Parikh & Loizou 2005; De Decker 2015; Jiang et al. 2011). Secondly, it was difficult to access all ranges of age, sex, and social groups. Dongxiang are a very hospitable people but also have a conservative culture. As a woman, I had the possibility to interact with both men and women – which would have been impossible if I were a man; but, on the other hand, I had no access to several areas where only men were allowed. In addition, approaching children was nearly precluded: I have had several opportunities to meet them, especially in the school environment, but for interviewing them I did not have the required authorisation. For similar reasons, I have not had the chance to record audio material in Dongxiang Autonomous County, which has also limited access to the elder generation. For the current case study, I had still a very diversified age range, but for more extensive research, the participation of children and elders may prove essential, and its prevention is a limiting factor. Finally, though the impact of the observer’s paradox on phonetic phenomena is relatively limited, it remains a factor that should not be disregarded, especially when conducting sociolinguistic research on minority languages. This is particularly significant in the context of low-resourced languages, where the scarcity of data exacerbates the challenges posed by these limitations.
- 20 The fieldwork has been conducted within the borders of Linxia Hui Autonomous Prefecture, with the a (...)
- 21 The level of proficiency was indicated by the participant, and on several occasions, it was also po (...)
- 22 Participant A was born in Dongxiang Autonomous County but raised from an early age in Linxia City.
- 23 Participant B was born in Kangle County but grew up from an early age in Linxia City.
It is necessar (...)
- 24 Participant C was born and grew up in Guanghe County.
- 25 Participant D was born and grew up in Kangle County.
22The study is based on a small-scale dataset, collected with the aid of four participants within the area of the Linxia Prefecture20 at the end of 2019. The material has been collected with the scope of conducting a preliminary case study on the above-mentioned sociophonetic variations. It is structured to document the variations of fluent speakers of Dongxiang in the production of the uvular consonants and the prosody. All the participants were Dongxiang/Linxia bilinguals, except for one who was able to speak Linxia, but not with native speaker proficiency21. The participants were purposefully chosen of different ages and hometowns. Two participants were from the city of Linxia (21 years old, hereafter referred to as participant A22, and 53 years old, participant B23), and two from the countryside (38 years old, participant C24; 45 years old, participant D25). The dataset by itself does not represent a sufficient corpus for a quantitative investigation, and additional material and a larger sample are desirable. Nevertheless, because of its variety and focus, it allows one to draw qualitative conclusions on the sociophonetic changes and their diatopic divergence.
Table 1. Participants
|
Age
|
Origin
|
Proficiency in Linxia
|
|
Participant A
|
21
|
Linxia City
|
Fully proficient
|
|
Participant B
|
53
|
Linxia City
|
Fully proficient
|
|
Participant C
|
38
|
Guanghe County
|
Fully proficient
|
|
Participant D
|
45
|
Kangle County
|
Not fully proficient
|
23The analysis first focuses on the investigation of the prosodic variation, in particular by looking at the pitch patterns of a set of homophones, and the pitch oscillation of selected sentences/locutions. Afterward, the lenition of the uvulars is examined, by analysing different degrees of lenition in the same word in different contexts.
- 26 In Dongxiang there are three most commonly used syllables with this function: gjə, -ji, and -dzi. T (...)
24The homophonic syllable under discussion is gjə, a verb of Mongolic origin (<*ki “to do, to make”), also used as an auxiliary, and in disyllabic verbs borrowed from Chinese as a deverbal suffix26. This syllable has been chosen among others, due to its versatility and because it exhibits relevant pitch variations depending on the context and meaning.
Figure 1. ma faŋ gjə, is an example of gjə used as a deverbal suffix and shows a peaking tone ⟨231, ˧˦˩⟩
ma1 fã24 gjə231, “thank you”
It derives from the Chinese disyllabic verb mafan 麻烦, “to cause trouble”. In its Chinese context, it may be observed in idiomatic phrases such as mafan nile 麻烦你了, where it is used idiosyncratically to convey gratitude. This semantic shift remains consistent in the Dongxiang language, where this root retains a connotation of gratitude
© Giulia Orlando, image took in PRAAT from first-hand audio recorded material
25The peaking tone of gjə as a suffix in Chinese loanwords was prominent in several occurrences of this dataset, thus both when pronounced spontaneously and when elicited. In this occurrence, the phrase is pronounced by participant D (countryside) and was taken from a simulated conversation, which was a task performed by participants C and D, during a joint meeting with them. The example is not an interrogative phrase, nor at the end of the conversation, nor followed by a pause; therefore, the pitch shown is not affected by the intonation of the sentence nor by the list effect.
- 27 The participant was asked to translate the Chinese phrase dagong 打工, “to work for earning a living”
Figure 2. wilie gje, in this example, gjə is used as a verb with deverbal function, in combination with the noun wilie “job”, with the meaning “to work”27, calquing the corresponding Chinese verb, and it shows a peaking tone, with a sharp peak ⟨253, ˨˥˧⟩
wi21lje22 gje253, “to work”
The word “wilie” is a native Mongolic word (*īli > *üjile, “deed, action, work”)
© Giulia Orlando, image took in PRAAT from first-hand audio recorded material
26The word in this case is elicited. The speaker, who is participant A, was asked to talk about job-related vocabulary. To minimise the list style effect, the participant was asked to use the word in context. The tonal pattern of figure 2 resembles the one observed with direct borrowings (see fig. 1), which also applies to the function expressed by giə in both cases and the structure (N + giə → V).
Figure 3. iŋ giə, is an example of gjə as a verb, though used with the function of speech connector, and it shows a level tone ⟨44, ˦⟩
iŋ13 gjə44, “to do like that”
The verb ingiə, which literally means “to do like that” is here suffixed with the lose temporal suffix -sə, and used as a lexicalised item. See Field 1997, pp. 399-404 for discussion about the peculiarities of this verb.
© Giulia Orlando, image took in PRAAT from first-hand audio recorded material
- 28 The story translated is known as “Xiaoma guo he” 小马过河, “Little horse crossing the river”.
- 29 Here used as a noun, with the meaning “means, way”.
27In this small corpus, in several examples when gjə is used with this meaning, it occurs with a level tone. This occurrence is pronounced by participant D (countryside) and taken from a long translation task. As for figure 1, also in this case, the example chosen is a non-interrogative sentence. It was taken approximately in the middle of the translation task28, which guaranteed semi-natural speech. It is pronounced at the beginning of a sentence, after a small pause and followed by the Mongolic noun “kha”29.
Figure 4. gjə here is used as a verb, with the meaning “to be called”, and it shows a low rising tone ⟨155, ˨˦˦⟩
gje155, “to be called”
The pitch image is isolated from the sentence chi nere yan gie ne? [tɕʰi neɾɪ jan gje nə], “what’s your name?”.
© Giulia Orlando, image took in PRAAT from first-hand audio recorded material
- 30 The context in which gje is produced is neutral since all the words uttered in this phrase are of M (...)
- 31 2SG name INT call-IMP.
- 32 The participant pronounced this sentence five times: the first one was not considered because too u (...)
28In this case, the verb gjə is elicited. The speaker, who was participant A (city), was asked to produce this exact sentence (tɕʰi neɾɪ jan gjenə30, “what is your name?”) in order to obtain the production of “gjə” with the meaning “to be called”. The participant produced the sentence five times31. The occurrence selected was the least affected by the approach32.
29The phenomenon under discussion in this case study appears to be still incipient; therefore, the distinctiveness observed and described for giə is not ubiquitously detectable. It is, however, noteworthy that both participant A (city) and participant D (countryside), used a similar tone pattern for the two giə with deverbal function. Thus, from the data examined, it is possible to observe instances of contrastive pitch patterns, salient among a limited number of speakers, regardless of their place of origin. Since the considered corpus is still limited, the resulting findings are unavoidably preliminary. The present analysis intends to provide a basis for future research with a quantitative methodology and a more extended corpus.
Table 2. The homophonic syllable giə: pitch outline and word function
|
Tone
|
Participant
|
Method
|
|
Deverbal giə I (non-Mongolic)
|
231, ˧˦˩
|
D (countryside)
|
Simulated conversation
|
|
Deverbal giə II (Mongolic)
|
253, ˨˥˧
|
A (city)
|
Elicited
|
|
Verbal giə I (“to do”)27
|
44, ˦
|
D (countryside)
|
Translation task
|
|
Verbal giə II (“to be called”)
|
155, ˨˦˦
|
A (city)
|
Elicited
|
- 33 It is necessary to point out that these examples are both questions. However, Dongxiang morphology (...)
30Besides the pitch variations in homophones, it is also possible to detect rudimental tonal features in words within the sentence and locution context which, however, are not yet distinctive. The examples considered here are produced by participant A (figs 5-6)33 and by participant D (fig. 7). Figures 5 and 6 are taken from the same set of sentences elicited to analyse giə with the meaning “to be called”, and the ones selected here are the most natural produced (fig. 5: second utterance; fig. 6: fourth utterance).
Figure 5. tɕʰi neɾɪ jan gjenə, “what is your name?”, Participant A, 1
© Giulia Orlando, image took in PRAAT from first-hand audio recorded material
Figure 6. tɕʰi neɾɪ jan gjenə, “what is your name?”, Participant A, 2
© Giulia Orlando, image took in PRAAT from first-hand audio recorded material
31It is possible to observe, in figure 6, that the participant produced a tonal-like prosodic pattern. In comparing the examples, it should be noticed that they have a slightly different speech rate: in general terms, the speech rate does impact the pitch; in this specific case, a more lilting pitch would have been expected in figure 5. However, that utterance shows a predictable pitch, thus appears to be unaffected by the production speed. In fact, the first utterance shows a predictable word-final stress, while in the second one, it is possible to notice a constant pitch oscillation, which cannot be solely explained by referring to the standard stress pattern of Dongxiang.
Figure 7. nje ka moʐəχa, “a little horse”, Participant D
© Giulia Orlando, image took in PRAAT from first-hand audio recorded material
32In this example (fig. 7), uttered by participant D, a prosodic variation is observable: by looking at the pitch pattern it is possible to notice two points in which the word-final stress is not applied: nie and mori (here pronounced as moʐə). In the first case, this can be easily explained by considering nie ka as a prosodic unit. However, in the case of mori, to which is added a suffix (ha 哈, see below), a change in the stress place is expected, namely falling on the suffix itself, which does not occur in this case. Thus, the utterance produced shows a non-standard pitch pattern, not justifiable only looking at the prosodic rules of Dongxiang. It is interesting to point out that the utterance is part of the translation task mentioned in figure 3, for which the participant was instructed to translate a story using exclusively Dongxiang, and that it was uttered at the beginning of the task, meaning that it was more unlikely for the participant to unintentionally use non-Dongxiang elements. In addition, participant D was the only one not fully proficient in Linxia.
33In light of the above, it is significant to observe that the non-predictable pitch seems to coincide with the introduction of the Linxia suffix 哈 (here pronounced as χa), thus seems to be actually affected by the introduction of an alien element from a tonal language in the clause. The present corpus does not allow to speculate further on the likelihood of an impact on prosodic variations caused by code-mixing; however, it must be taken into account that the phenomenon of code-mixing has been already demonstrated to phonetically affect Dongxiang (Lefort 2012).
34It is worth mentioning that the speakers from the city showed this pitch oscillation more consistently through almost the entirety of the utterances produced, while in the speakers from the countryside, it was a lesser regular feature: i.e. in the translation task performed by participant D, across 4 minutes 46 seconds, it was clearly possible to notice only 9 instances of pitch oscillation; on the other hand, in a short story told by participant A, across 32 seconds, 4 instances of pitch oscillation were noticeable.
- 34 This contour tone has a pitch that descends and then returns.
35Another noteworthy difference encountered in the dataset is that in the individual utterances elicited, the participants from the city are more regularly producing tonal-like pitch patterns: i.e. when the vocabulary for family members was elicited, participant A produced practically every word with atypical stress. To avoid the list effect, the participant was asked to use the words in context, but regardless, all the disyllabic words were showing a dipping34 tone-like prosody on the first syllable. This was noticeable regardless of the words’ origin, as much in integrated Linxia loanwords as in words of Mongolic origin. In this case, the SNR does not allow drawing visual conclusions with the only aid of the chart. For reference, a table is reported showing the pitch extracted from the same utterance. The word considered is hatɕʰɪ, the Dongxiang word for nephew: ha (285,58-285,85) pitch is 534 ˥˧˦, tɕʰɪ (286,20-286,28) pitch is 22˨, roughly a low flat one.
Figure 8. hatɕʰɪ, “nephew”, pitch outline (timeframe: 285,58-286,28)
© Giulia Orlando, image took in PRAAT from first-hand audio recorded material
36From the data analysed, in regard to the prosodic variation under consideration, it is possible to claim that all the participants showed to different degrees a tendency to tonal-like prosody, but the adoption of tonal-like features seems geographically dependent.
- 35 Listed as “qɑla” in Bu He 1983 (p. 68) and “khala” in Ma & Chen 2001 (p. 223).
37The other phenomenon observed during the fieldwork was a geographically uneven tendency to lenite the uvular consonants. This phenomenon was completely absent with the speakers from the countryside (fig. 9), but relatively frequent with the speakers from the city. The alteration resulted in two different outcomes: /k/ or /h/. The lenition into a velar was more frequent, but in some cases, a complete loss of the consonant (debuccalisation) also occurred. The word examined is qala “where”35. Figure 9, which is an utterance produced by participant C (countryside), offers an example of uvular retention among the countryside speakers. The figures 10, 11 and 12, on the other hand, nicely illustrates the alteration on the same word.
Figure 9. qɪlə
© Giulia Orlando, image took in PRAAT from first-hand audio recorded material
- 36 This is also noticeable from the fact that the vowels are not reduced, as in figure 9.
38In figure 10, the word is pronounced clearly and slowly, and the initial voiceless unaspirated uvular is uttered distinctly. The speaker, in this case participant A (city), was asked to repeat the word; therefore, the pronunciation was more controlled than the one produced by the same speakers in figure 936.
Figure 10. qaola
© Giulia Orlando, image took in PRAAT from first-hand audio recorded material
39Figures 11 and 12 are both natural speech utterances. In figure 11, the consonant is completely lost and lenited in a /h/. In figure 12, though the consonant transcribed is still a /q/, it is possible to observe a substantial difference in the spectrogram, if compared with figure 10. In this case, the sound uttered was still slightly uvular but tending to the correspondent velar one.
Figure 11. hala
© Giulia Orlando, image took in PRAAT from first-hand audio recorded material
Figure 12. qola
© Giulia Orlando, image took in PRAAT from first-hand audio recorded material
40In addition, during the fieldwork, it was possible to explore the attitudes of the participants towards their ethnic language. From the interviews I observed the following: participant A was a Dongxiang/Linxia bilingual but showed a higher linguistic confidence in Linxia and was very willing to use Linxia loanwords. Participant B was perfectly bilingual, and open to using Linxia loanwords. Participant C shown to be a perfect Dongxiang/Linxia bilingual, but was clearly eager to keep the maternal language, and not very willing to use Linxia too much. Participant D, the only one not fluent in Linxia, was totally unwilling to become more fluent in Linxia or use Linxia loanwords. Hereafter I report an extract of the interviews conducted with participants A, C and D:
Interview with participant A
Can you speak Dongxiang? What is your opinion about this language? Tell me whatever you feel to say about it.
Interviewee: I tend to speak more often Chinese, because I can speak it quite well, but I also speak Dongxiang. Well, for example, with my grandparents I have always communicated in Dongxiang, because they cannot speak Chinese. I have spoken it since I was a child because it’s my heritage, I know it well, though I sometimes use some Chinese words while speaking it. I think the family environment affected me, because, well, everyone around me [the family] speaks Dongxiang, so I picked it from a very young age… and well also the environment in general, because I have not always lived here [in Linxia City]. When I was younger, I lived in the countryside. Then one of my elder brothers came here, and all the family gradually moved here.
Interview with participant C
What do you think of the language of your minority? Can you tell me about your opinion on the language or the culture? What is your hope for it? Tell me whatever you feel to say about it.
Interviewee: well, our language has a long history. I feel we could know it better. Here, it’s happening a sort of economic advancement, um, and cultural also. Fewer people, especially the younger generations, know how to speak it. We really hope that our, uh, language will not disappear. I hope we can pass our knowledge on to the next generation.
How is the school system here?
Interviewee: In the past, there were more bilingual classes; now there are still some, but I feel not as much. The younger generations should have more access to Dongxiang, otherwise they are going to lose it. We must keep speaking it. If the children can maintain it, Dongxiang will not disappear. That’s why I feel that it is necessary now that our schools make an effort in this direction. Unfortunately, it doesn’t seem to be the case: for example, the son of a friend of mine is going to the class next door. They use Dongxiang and Chinese, but they definitely promote Chinese, er, Mandarin I mean, more. And of course, when they write, they just write Chinese. And now, most of the students just go out of the region to work after passing the exam. I understand, this is because of the Country’s development over the last years. I hope Dongxiang will not disappear.
When or how often do you use Dongxiang?
Interviewee: Dongxiang is used on some occasions, for example, at home with the family, because it is just more convenient… many people in the countryside know a lot more Dongxiang vocabulary than Chinese. In my case, when I am at home I speak only Dongxiang, because my family members are all native Dongxiang. But when we go out of Guanghe, we may need Mandarin to talk to outsiders, or even when we get a phone call. Some people, however, are a bit scared to speak Chinese, maybe they feel they need to think longer to speak it well. Anyway, whenever possible, it is very convenient to use our dialect also out of the county, because we can talk about our own things without caring about others.
Interview with participant D
What do you think of the language of your minority? Can you tell me about your opinion on the language, or the culture, whatever you feel to say about it? What is your hope for it?
Interviewee: Well, Dongxiang has always been passed down orally: the older generation taught it to the younger, and so on, up to now. And in this way the language has not disappeared, everyone just passed it orally. Now, the country is acting to preserve the language, it is a national policy to preserve the minorities. But if people start to go out of their homeland, maybe, everyone will gradually but eventually forget their own language, even if the country tries to save it…I think it is possible even in one generation, because children have been exposed to Mandarin so much. Dongxiang is just a dialect, and many teachers cannot speak it properly. Also, it is becoming rarer to find a teacher able to speak it, because they are getting younger and younger. So, if the environment around the children is all speaking Chinese, they are going to forget it, no matter what. As for me, I try to use Mandarin as little as possible when I go out of Kangle. If I meet a fellow villager, then we start communicating in Dongxiang. Of course, Dongxiang is not always, you know, convenient to use, if people around you cannot speak it.
41The sample is still limited to further speculating on the generalised linguistic attitudes, but already as it is, it shows a varied range of them, seemingly aligned with the presumed factors described in section 4 (Phonetic and phonological peculiarities of Dongxiang in the Amdo Mongolic domain).
- 37 Acuo & Xiang 2015, pp. 483-497.
- 38 For further discussion, refer to Dwyer 2008.
42Before analysing the evidence presented, it is worth highlighting that the phenomena observed are not isolated in this area. In regard to the partial loss of uvulars, lenition phenomena in the Amdo sprachbund have already been documented (Nugteren 2011). As for the prosodic changes described in this case study, they are grounded on former evidence from the neighbouring varieties. In fact, similar phenomena are already attested in the area in the same phylum. Besides the above-mentioned case of Monguor (Dwyer 2008), in this particular area, variations in tonality have already happened and been reported in both directions. On the one hand, Wutun37 – a northwestern Mandarin tonal variety spoken in the Qinghai region has become toneless due to its contact with a toneless Tibetan variety, the Amdo Tibetan (Li 1986; Sandman 2016), though its tonoexodus is still under debate (Acuo & Xiang 2015). On the other hand, Bao’an, another Shirongolic toneless variety, has been described to be in the process of developing contrastive pitches in Linxia Hui loanwords due to its exposure to this variety (Li 1983, 1986). Compared to other examples of tonogenesis, these emerging prosodic changes in Dongxiang are unusual because they seem prompted by contact with Chinese and subject to context-dependent variability. In fact, tonogenesis is usually explained by considering internal phonetic causes (Abramson 2004). However, as mentioned above, there does exist already one documented example of atypical tonogenesis38: Southeastern Monguor. Its tonogenesis seems to have been triggered by language contacts and not exclusively by internal phonetic changes. Dwyer hypothesises that the development of contrastive prosody in Southeastern Monguor signals an incipient phonemic tonal system, which is observable in some homophones distinguishable only due to different pitch patterns and in the increasingly detectable innovative prosodic realisations of Mongolic lexemes (Dwyer 2008).
43As for Dongxiang, it is usually described as a toneless language with word-final stress in unborrowed vocabulary. As mentioned above in greater detail (cf. sect. 4, “Phonetic and phonological peculiarities of Dongxiang in the Amdo Mongolic domain”), as in other varieties of the Amdo sprachbund, Dongxiang stress predictably falls on the ultimate syllable in words of Mongolic origin. It is, however, considered a non-contrastive prosodic feature. On the other hand, in loanwords, the stress is irregular; this is because of the originating stress patterns, or in the case of Linxia Hui loanwords, possibly due to the originating tonal patterns. But besides the unpredictability, the pitch variations, so far, fall still under the stress category.
44I, however, hypothesise that the vivid contact with Linxia is causing a shift toward tonal prosody. At this stage, most of the variations in the pitch are still non-contrastive, but in some cases, it is possible to observe a prosodic differentiation in homophonic lexemes, as in the case of the syllable “giə”.
45Since the present study is based on four native speakers, it allows only preliminary conclusions. In the future, for further descriptions of the phenomenon, a larger sample is advisable. Nevertheless, these preliminary observations allow us to understand and frame the emerging tendency.
46On the basis of the evidence presented in this article, it is possible to notice the emerging of tonal-like pitch contour in the speech production. The ongoing prosodic changes are not yet systematised nor standardised, but it is already possible to capture the phenomenon’s onset: the Dongxiang speakers are producing non-distinctive tonal prosody and are also already showing in some rare instances of contrastive tone in homophones. However, while these first stages of prosodic changes suggest a developing tonogenesis, the phenomenon does not equally affect the rural and urban areas. In fact, in the rural areas in fast natural speech, the data show that the pitch oscillations are less significant than in the city. This diatopic divergence is aligned with the information shared by the participants: in the interviews conducted (cf. sect. 7 “The dataset”), rural speakers demonstrated a rather reluctant attitude towards Chinese and were keener to maintain their native language and preserve it. At this stage, it is still speculative, but the minor use of tonal features appears to coincide with the feelings shown by the countryside participants, and their desire to preserve their ethnic language and use Mandarin as little as possible. In this regard one should consider that while the number of monolinguals has been steadily low for at least four decades, the bilingualism appears to be on the rise (cf. sect. 5 “The sociocultural and economic factors”). This can lead to a more preservative attitude from the monolinguals, which may be salient among the interviewed speakers, considering that one of the two participants from the countryside had a limited proficiency in Linxia. I argue that a lower adoption of tonal-like features from the countryside speakers shows a certain degree of agency and indexes the more preservative linguistic attitude observed among the participants. On the other hand, the urban speakers were more willing to adapt to the dominant Sinitic variety and also showed a willingness to speak Chinese. I, therefore, argue that these different attitudes are the reason behind a slower and lesser adaptation of the prosody of the rural speakers, and an earlier variation in the speech of the urban speakers.
47As for the phenomenon of uvular lenition, it is possible to notice a drastic difference between the participants. The urban speakers are increasingly replacing the uvulars with less language-specific sounds. The speakers from the countryside, on the other hand, retain more uvular sounds than urban speakers. In fact, in their speech, lenition is absent. On the basis of the analysed data appears plausible to consider this difference as based on sociocultural factors, namely the uvular retention seems to index the desire to preserve a distinctive trait. However, only natural speech conversations could prove this. On the other hand, given the stronger exposure to Chinese and the willingness to use it among the participants from Linxia City, it appears reasonable to suppose that the emerging lenition shown by the urban speakers is associated with these attitudes. In regard to the degree of awareness behind the lenition, since the sounds are indeed language-specific, they are the object of a high degree of awareness. This would explain the greater divergence in the phenomenon of lenition, compared to the prosodic changes.
48The current case study, therefore, argues that Dongxiang phonetics and phonology are currently undergoing some contact-induced language changes, impacted by contact with Linxia. It claims that the CILC are triggered by sociocultural factors, such as the religious identity and the economic environment, but are not evenly observable among the urban and rural contexts. It proposes that the diatopic divergence is reasonably affected by the attitudes shown by the participants, to which it is aligned. In addition, this study offers a sociocultural explanation to the incipient tonogenesis observed, providing basis for further research on tonogenesis’ causation, besides the linguistic internal factors.
Flux RSS
