Language attitudes in Northwestern Tunisia and their implication for speech patterns

1 Introduction

This article deals with speakers’ attitudes in the Tunisian governorates of Siliana (North Central Tunisia) and Jendouba (Northwestern Tunisia) towards their own dialect, and towards the urban dialects of the capital Tunis and of the Eastern coast. The dialect of Tunis, an urban variety, is without doubt Tunisia’s most prestigious variety, and the lifestyle in Tunis and other major cities on the Eastern coast is perceived as urban and advanced. The inhabitants of Tunis are considered baldīya ‘urbanites’. In the central and northwestern regions of Tunisia, on the other hand, rural dialects are spoken (Marçais 1950; Skik 2003),^[1] the region itself is considered as rural, culturally backward and economically, socially, and politically disadvantaged. This division into urbanites and those living in rural areas is again addressed in Section 4, as it is strongly connected with the perception the speakers have of their rural dialects and that of the urban dialect spoken in Tunis. Unemployment pushes the youth to migrate both within Tunisia as well as abroad – three quarters of the country’s unemployed are youth between the age of 15 and 30 years. More than 30% of Tunisia’s youth is unemployed and this rate is even higher for women and in poorer regions, especially in the west. For example, the general unemployment rate in Tunisia is 13%, while for Jendouba and Siliana – the regions we focus on here – it is 24%. The western regions of the country have poorer infrastructure, less access to international harbors and ports, and human capital availability, and therefore remain highly dependent on agriculture, which provides mostly seasonal, low-skill, and low-wage employment and is not attractive for youth. Hence, per capita income in the Tunis region has been around twice that of the poorest regions (the Northwest and Midwest), and the poverty rate is three times higher in these regions (Boughzala 2016). The majority of the rural population is either landless or owns micro farms (less than 10 ha of rather arid land or less than 2 ha of irrigated land) and has a limited formal education. The Midwest can be considered the poorest region of Tunisia, as the northwestern and southwestern regions have already lost a large part of their population to migration and are therefore a little better off (Boughzala and Tlili Hamdi 2016). Although governmental programs have been implemented over the last decades to improve the economic situation of the poorest regions, they have been less effective in the west and could not compensate the historical bias of public investment being concentrated on the coastal regions. Unemployment and poverty strengthened the sentiment of exclusion and discrimination among the populations of the poorer regions and played an important role in the revolution 2011 (Boughzala 2016).

The situation of women is even more complex: in general, these regions have a higher percentage of women due to the emigration of a large part of the male population. Still, as shown by Bouhdiba, the number of women looking for a job abroad or pursuing their academic education is increasing. However, the lack of reliable infrastructure and their status in society still makes it harder for girls to leave their families (Bouhdiba 2019: 66, 71–72). Fewer opportunities are open to women because their mobility is more restricted. They are more likely to accept lower-wage jobs to be able to stay near home. This includes mostly seasonal work in agriculture, but such employment is often unpaid family work – or an employment in the informal sector working under very hard conditions (Boughzala and Tlili Hamdi 2016). Although more than 60% of university graduates are women, the rate of female participation in the labor force (27% in 2011) remains much lower than men’s (70%). The unemployment rate of young women in poorer regions, especially in the west is higher than the average youth unemployment rate (30%) (Boughzala 2016). This places young rural women among the social groups with the lowest status in society and potentially creates a lot of pressure, to level up in any possible way.

This economic situation is crucial for understanding the social stigma of the northwestern regions as well as their self-perception and accordingly, the perception of their dialects. We will show how certain linguistic features of these dialects are charged with stereotypes and investigate which social meanings are attached to them. These perceptions play a crucial role in processes of accommodation and levelling. To illustrate this we analyse the implications these attitudes have on the actual speech patterns of Siliani and Jendoubi speakers by concentrating on four linguistic variables. We show that the intersection of age, gender and level of education is not only crucial for the emergence of language ideologies and attitudes, but also for the actual language use. Three of these variables have been analysed in Ritt-Benmimoun (2021), with a then much smaller corpus. It was stated there that “the governorate of Siliana with the towns of Siliana, Sidi Hmada, Lakhouet, Drija, and Garia is the other region where numerous ‘urban’ features can be found. In Siliana mostly young women promote this ‘urbanisation’”. The analysis of our linguistic data will show that in fact young, educated women with a rural background are among the first to adopt urban features. Our findings can be embedded in Miller’s (2004: 935), who shows that it is mainly the phonological level where gender specific features become obvious, and that it is the awareness of a very salient or a socially marked feature that smoothes the way for accommodation.

To understand the main argument of our study it is necessary to explain the main differences between the dialects spoken in the urban centres of the northern and eastern coast (for example Tunis and Sousse) and the rural dialects spoken in Jendouba and Siliana. The four most salient characteristics are at the same time our linguistic variables used in Section 5 because they play a crucial role in linguistic accommodation:^[2]

In Tunisia the uvular [q] is representative of sedentary dialects spoken mainly in urban centres on the northern and eastern coast and in the city of Kairouan, whereas the velar [g] is predominant above all in rural Tunisia. The use of either [q] or [g] is the most important difference between urban and Bedouin-type or rural dialects of relevance to the speakers themselves who call the use of the uvular [q] qalqīl (or qalqala). [q] as “the shibboleth of all sedentary dialects in Tunisia” (Procházka and Ritt-Benmimoun 2008: 89) is thus a kind of marker to determine how urbanized a given Bedouin-type or rural dialect has become.

We will now turn to the theoretical frameworks we use for the concepts of ‘language ideologies’ and ‘language attitudes’, and the concepts of ‘accommodation’ and ‘levelling’ for the analysis of our linguistic data.

Attitudes in social psychology have been defined as a “disposition to react favourably or unfavourably to a class of objects” (Sarnoff 1970: 279). When this class of objects is or contains a language or language variety, we speak of language attitudes. Accordingly, “attitudes are a heuristic summary of experience; thus language attitudes are invoked every time interlocutors encounter a variety of speech that they have heard (or heard of) before” (Cargile and Bradac 2001: 347). It is widely acknowledged that attitudes have a tripartite structure (Edwards 1982): First, an object can be evaluated on a cognitive level, reflecting an individual’s thoughts on a certain dialect; second, on an affective level, reflecting emotions related to a certain dialect; and third, on a behavioural level, which results from these two, causing a person to behave in a certain way toward a dialect, like accommodating to it when they perceive it as more beautiful than their native dialect. Thereby, attitudes towards language varieties constantly affect and determine interpersonal communication.

While attitudes are discrete and isolated, language ideology functions as a system linking aspects of language with aspects of social organisation and other interests. Language ideology can be defined as “the cultural (or subcultural) system of ideas about social and linguistic relationships, together with their loading of moral and political interests” (Irvine 1989: 255). They are not chosen individually but determined by an individual’s position in the social order, life experience and value commitments (Kroskrity 2004). It is ideology, which leads individuals to the conviction that a certain language variety is suitable for a certain context (like an academic, religious or family context) and which makes them perceive speakers of certain varieties as intelligent or not, educated or uneducated, lazy or ambitious. The study of language ideologies is concerned with how thought patterns about language structure society and vice versa – ranging from personal interaction to concepts like nationalism (Walters 2007). Eventually, a group’s beliefs about language are crucial for its identity (Kroskrity 2004).

Accommodation within the Communication Accommodation Theory (Giles 1973 and further developed and challenged by Trudgill (1986) whose works were essential and fruitful for the study of dialect contact and change) is either convergent or divergent (Shi’ri 2009). It is convergent accommodation that is of relevance for our purpose, as it is defined as “a strategy whereby individuals adapt to each other’s communicative behaviors” (Giles et al. 1991: 7). This adaption happens by reducing linguistic differences which are often the most salient characteristics of a variety, not only to guarantee a more fluent and more effective way of communication but also to attain social acceptance (Giles et al. 1987: 14–15). Indeed, accommodating to the perceived speech style of another (mostly socially more dominant) group is a “reflection (often unconscious) of a speaker’s or group’s need for social integration or identification with another” (Giles et al. 1987: 16).

We use the term ‘levelling’ following Holes (1995b: 39), who describes it as the “elimination of very localized dialectal features in favour of more regionally general ones”. Thereby typically rural features may be replaced by urban ones (Bassiouney 2008 based on Blanc 1960: 62). Stigmatization and identity are crucial factors in this process (Bassiouney 2008): the stigmatization of local dialectal features can be a driving factor for levelling, whereas identity can be crucial in sticking to exactly these features.

2 Previous studies on language attitudes and dialect levelling in Tunisia

Though education, migration, urbanisation and globalisation are having an immense impact on dialects, accelerating linguistic accommodation and language change, not much has been published on Tunisian dialects in this respect so far: Jabeur (1987, 2000 investigated the modification of speech of rural migrants within the urban society of Rades, a suburb of Tunis, analysing various phonological variables (/q/ vs. /g/, diphthongs, the variable (R) in French words) and morpho-syntactic variables (e.g., defective verbs, relative pronoun, verbal negation). Jabeur found out that the use of urban forms is tightly bound to speakers’ integration into urban social networks (Jabeur 1987: 170). Additionally, accommodation is also a question of saliency: salient phonological features are levelled out but not non-salient morphosyntactic features (Jabeur 1987: 222). For the town of Korba, Walters (1989) examined the three phonological variables /ɛ:/, /ṣ/ and /u:/, and the variable /u:/ in his (1992) article (Walters 1992). The results from his 1989 publication are very interesting insofar as “no single sociodemographic group appears to be the innovators or the conservatives. Rather, these patternings of the behavior of these individuals are intimately tied to their life experiences, to possibilities offered them by history, by society, and by individual choice.” (Walters 1989: 298). Walters’ findings are to a large extent confirmed by Gibson in his 1998 dissertation, in which he treats the nomadic-sedentary split with three variables: Reflexes of /q/, third-weak verbs and gender marking in the 2nd p. sg., variables also to be analysed for our data in Section 5.

The realisation of /q/ is probably the most mentioned and most commented variable, and [g] is, among others, called a “stigmatized” sound within Tunisia (Sayahi 2021: 109), “the key marker of social identification” for Arabic in general (Miller 2005: 936), and is “highly salient and ideologically loaded” (Hachimi 2012: 333). As Moumine (1990) showed, [g] in Casablanca marks the lower working class, whereas [q] marks the middle and upper middle class. However, the use of [g] is attested for the male population of Tunis in a positive light: Trabelsi (1988) found out that men in Tunis prefer the use of [g] in certain contexts to give the impression of virility.

Investigating the 2nd person gender-merger in Fassi-speakers of Casablanca, Hachimi (2011: 37) showed that this trait is considered sophisticated – whereas the Casablancan speakers’ gender-differentiation with -i for distinguishing female from male addressants is considered ‘ʕrobi’ (rural). The Fassi dialect is therefore seen as ‘délicat’ (fine), while the Casablancan dialect is seen as ‘cru’ (harsh).

Gibson (1998, 2002, as well as Procházka and Ritt-Benmimoun (2008), showed that changes in Tunisian Bedouin-type dialects definitely tend toward the urban dialects, in particular that of the capital Tunis, rather than toward other Bedouin-type dialects or Modern Standard Arabic (MSA).^[3] But whereas Gibson (1996, 1998, 2002, 2012 investigated linguistic variables mainly among Tunisian students in France and Great Britain, and Procházka and Ritt-Benmimoun (2008) concentrated on language attitudes of the Bedouins in Tunisia’s far south, the regions of Siliana and Jendouba have not yet been the focus of (socio)linguistic research. While the direction of the levelling is quite clear – accommodation happens mainly towards the dialect spoken in Tunis – our focus will be on investigating who of the speakers of our sample is most prone to accommodate to it.

Associations with urban dialects of importance for our study are found in Gibson (2002: 30): “… we may assume that the speech of Tunis for other Tunisians embodies some values along the line of modernity, refinement, and education.” The dichotomy between the prestigious dialect of Tunis and the rural dialects from Tunisia’s central and northwestern parts, reflects the Egyptian situation studied by Miller, with the prestigious Cairene dialect on the one hand and the ṣaʕīdi (Upper Egyptian) dialects on the other hand. Both “low-varieties” are described as ǧaff ‘dry’, tagīl ‘heavy’ and xašīn ‘rough’, while the qualities naʕīm ‘soft’, rāʔi ‘elegant’ and mutaqaddim ‘modern’ are used to describe Cairene Arabic (Miller 2005: 917).

3 Methodology

An individual’s attitudes are generally based on their mere assumptions of a certain language or dialect, which do not necessarily correspond to actual dialectal features. People might claim to dislike the dialect of a region or country and evaluate it accordingly, without being actually able to identify it (Herbolich 1979; Walters 2007). The above-mentioned cognitive, affective and behavioural components of attitudes can diverge significantly, for example when an individual states to dislike a certain dialect but still accommodates to it, when exposed to it. Studies on language attitudes have investigated both overt attitudes, asking interviewees directly about their stance towards a language (variety), and covert attitudes, in particular by matched-guise tests (e.g. Chakrani 2013). In order to compare speakers’ cognitive and affective attitudes to their actual behaviour, we matched the results of our qualitative interviews with those of the data collected by means of standardised questionnaires, casting light on the speakers’ actual language use.

The data that underlie our analysis of language attitudes (Section 4) was gathered in two fieldwork-visits with 43 qualitative interviews and participant observation conducted by Ilona Abdelfattah between June and August 2019 in the two Tunisian governorates of Siliana and Jendouba (in the locations shown in Table 1 marked with QI). In order to obtain a sample as diverse as possible, we aimed to interview at least one local person of older (above 40 years) and one of younger age as well as one man and one woman at each location. Within our qualitative interviews we concentrated on the following themes:

1. People’s attitudes towards their own dialect or the dialect of their birthplace and its distinctive features

2. Awareness of other dialects, their distinctive features and attitudes towards them

3. Attitudes towards the qalqīl (or qalqala) ‘talking with [q]’: Is it positively or negatively connotated? Who uses [q], how and why? Do young people, especially women, talk more with [q] and why? Are there also boys who take up [q] in their speech and how is it perceived when they do so? Do people in the two governorates generally accept the qalqīl and what is their reaction towards its adoption? Would they like their children to adopt it when studying in Tunis?

4. Linguistic patterns of students who study in Tunis: Do they change their dialect and adopt e.g. the [q]? And if so, do they keep talking with [q] when returning home?

5. Difference in age and gender: Does the dialect of the elder generation differ from that spoken by younger people? Does the speech of men and women differ?

The data for the analysis of the actual speech patterns (Section 5) was gathered by eight collaborators in interviews conducted in the governorates of Jendouba and Siliana between April 2019 and September 2020, during which people were asked to repeat a short, standardised text with linguistically significant words and phrases (sample text) and a list with salient linguistic features (feature list) in their dialect.^[4] In most cases, the sample text and the feature list were given in MSA, the interviewee then repeating the sentences in his or her dialect. While we are aware that the MSA template makes influence from the High Variety unavoidable, using a standardised questionnaire makes it easy to compare a fixed set of features from speakers of different locations. We elicited this linguistic information from 43 individuals in 16 different locations (Table 1, Figure 1), resulting in 31 sample texts and 29 feature lists (Table 2).^[5]

Figure 1:

Map with the locations in Jendouba and Siliana where fieldwork was undertaken.

While each of the described methods has its limitations, combining them can give important insights into the linguistic and sociolinguistic landscape of Northwestern Tunisia.

In the following section we will elaborate on the Siliana and Jendoubi speakers’ attitudes towards their own varieties and the prestigious variety of Tunis by analysing the data provided in the qualitative interviews.

4 Language attitudes: awareness – associations – oppositions

Interviewed people were familiar with Tunisia’s linguistic landscape and the most striking features of its dialects. They know where [g] or [q] is used and are aware of the stigmatization of [g]. When asked about their dialect’s distinctive features, the most salient answer given was:

As examples, people usually gave us words like gutt-lik ‘I said to you’, in contrast to the urban qutt-lik, or ugʕud ‘stay!’, as opposed to urban uqʕud.

It was stated by several interviewed people that young women in Siliana use a lot of [q] in their speech, as in (2) uttered by a 25-year-old woman from Lakhouet:

A 46-year-old man from Balta in Jendouba even stated that they spoke with [q] (yitkallmu b-il-qāla) referring to all Silianis. However, among Silianis themselves it is commonly stated that this linguistic behaviour is restricted mainly to young women and girls. In some families girls use more [q] than, for instance, their male family members, as becomes clear from a statement by a 23-year-old man about his 22-year-old female cousin, both from Sidi Hmada:

A 21-year-old woman from Lakhouet stressed that someone who used [g] in Tunis or in the urban centres of the Eastern coast was laughed at and even despised (yiḥ ^ǝ grūk ‘they despise you’), and that such linguistic behaviour was not accepted by the baldīya ‘urbanites’. As shown by Procházka and Ritt-Benmimoun (2008: 88) also the Bedouin-type dialects of the South that are so appreciated by the speakers themselves are generally disliked in the urban centres and often an object of ridicule. During a conference we attended in Tunis in February 2020, a student told us that people of the capital looked down on [g]-speakers but that someone who talked with [q] was considered fīnu, a loan from Italian fino.

A 40-year-old woman in Balta in the governorate of Jendouba tells us about her son:

A 28-year-old male student from Rouhia who studies in Tunis relates an incident in which a female professor in Tunis said to him:

Obviously, the professor did not feel properly respected and interpreted the use of [g] instead of [q] as inappropiate within the academic context.

People complain that they, i.e. the [g]-speakers, are treated differently from those who use [q] (tafrīq ‘differentiation’) and are always at a disadvantage compared to those who use [q], even in terms of salary and marriage.^[6] For instance, a 21-year-old female interviewee from Lakhouet stated sarcastically that in order to be considered urban and refined in Siliani society it is necessary to speak with [q]:

This shows that young women are under social pressure to adopt features which are considered fīnu. If, through further research, this tendency proves to be true, then we would have a situation similar to some other Arab countries where women are, in certain respects, the initiators of language change: In Amman, for instance, the use of the glottal stop is typical of young women, for them it symbolizes modern life (Al-Wer and Herin 2011: 70) and is one of the “symbols of power and influence” in a society where women are economically and politically marginalized. Young Ammani men, instead, use [g] and to a certain extent also the glottal stop [ʔ], depending on the context, “to maximise benefit and minimise loss in social interaction” (Al-Wer and Herin 2011: 72).

The use of urban ǟna ‘I’ for the respective local variant of the 1st p.sg., which is nǟya, is another very salient feature, as becomes clear from the following statement by a 32-year-old woman from Tbainia that stresses university education as an important factor to consider in the discussion of accommodation and language change:

Others, like a 40-year-old woman from Tbainia, are very clear about not changing their dialect:

Though many people especially in Siliana may strive to talk and behave like the twǟnsa l-aḥṛāṛ (the original citizens of the capital) it is well known that anyone who tries to talk like them will not be successful. This is exemplified by a 55-year-old female speaker from Lakhouet:

Coming back from Tunis people would usually re-switch to their original dialect, a behaviour also known from other communities.^[7] A 40-year-old male interviewee from Rouhia stated that people from Rouhia using [q] would be laughed at and that such a behaviour was considered ʕāṛ ‘disgrace’.^[8]

The stereotypical images mentioned above show that linguistic accommodation is important for speakers of stigmatized dialects in order not to be exposed to prejudices or ridicule, a development shown in many sociolinguistic studies (to cite only a few: Hachimi [2011: 28] for male speakers of the Fassi-dialect in Casablanca, Procházka and Ritt-Benmimoun [2008: 88] for South Tunisians in Tunis, Miller [2005: 917–918] for Upper Egyptians in Cairo).

For Siliani speakers, the urban regions have the image of ṯaqāfa ‘culture and education’ and tahḏīb ‘a refined manner (to speak, to behave)’, whereas rural regions are associated with a certain conservatism and a traditional way of life. A 55-year-old woman from Lakhouet expressed her favour for the dialect of Tunis, calling it a lūġa maqbūla ‘an acceptable language’, whereas she characterised her own dialect as mišši maqbūla ‘not acceptable’. A 16-year-old girl from Siliana complained that all other Arabs had beautiful dialects except for them. A 68-year-old man from Garia characterized his dialect as xšīn ‘heavy’ and said nitkallmu b-il-quwwa ‘we talk rudely, loudly’. Consequently, for many people the change in dialect of young Siliani people who study in Tunis is not considered tabdīl ‘exchange’ but tahḏīb and tažmīl ‘refinement’ and ‘embellishment’.

Language attitudes also reflect what Bourdieu and Wacquant call ‘legitimate’ and ‘illegitimate’ language: While the standard or official language is considered ‘legitimate’ in the cultural sphere, local dialects and the language of the lower class are regarded as ‘illegitimate’. As speakers of local dialects usually have less access to ‘legitimate’ language, they might feel powerless in situations where standard language is expected (Bourdieu and Wacquant 1992: 145–147). In this sense, Tunis is admired for its prestigious dialect, which ascribes to its speakers a certain legitimacy in every area of life – even table manners were mentioned, with knife and fork on the urban side and eating with one’s hand on the rural and Bedouin side. According to Bourdieu and Wacquant, whether an expression or a cultural practice (such as table manners) is considered appropriate, is determined by whether the speakers’ and receivers’ linguistic habitus (determined by their class, age, sex and education) conform to the rules of the linguistic market, “a system of relations of force” (Bourdieu and Wacquant 1992: 145). The mentioning of table manners by an interviewed speaker of an ‘illegitimate’ language brings out what Bourdieu and Wacquant call symbolic violence, namely “the violence which is exercised upon a social agent with his or her complicity” (Bourdieu and Wacquant 1992: 167).

The “modern” way of speaking, i.e. a leveled dialect which drops the most discriminative features of rural “illegitimate” dialects in favour of the dialect of Tunis, is characteristically attributed to young people. A 40-year-old woman from Balta ascribed a better way of speaking and living to the young ones.

According to her, the ḥišma-concept is not as dominant as it used to be. ḥišma, meaning literally ‘shyness, shame’, can be described as a complex set of rules that regulates inter-personal relationships in traditional Arab societies, especially the behaviour between two socially unequals, i.e. a woman and a man, young and old people (see Abu-Lughod 1986: 234–238 who calls it ḥašam). The loosening of the conservative and traditional gender roles makes it possible for men and women to eat together and to talk freely with each other, a behaviour that used to be socially unacceptable in the past. Some rural women (especially of the older generations), however, are still rather tightly bound to these customs defined as ḥišma. With regard to the above mentioned concept of legitimacy, which goes along with the use of a prestigious dialect, our interviewee from Balta associates the young generation’s life with more comfort and ease:

From these statements we can clearly see the association of young people with urbanity, modernity and openmindedness whereas the older generation is frequently associated with a traditional and conservative life style, characteristcs also attributed to their respective dialects.

5 Implications for speech patterns: evidence from our data

The attitudes in Section 4 mainly concern the realisation of /q/ as voiceless or voiced and the respective local variant of the urban personal pronoun ǟna ‘I’. To get a fuller picture of the linguistic practices we also include gender distinction in the 2nd p.sg. with verbs, and the inflection of III-weak verbs. The sentences and words chosen from our corpus of sample texts and feature lists (see Section 3) will show how these variables are used by speakers in Siliana and Jendouba. As such they should exemplify the urbanisation tendencies found among the speakers in these two regions. Before the linguistic variables are presented in detail, Table 3 gives an overview of the data to which the following tables (4–7) refer to. It shows how many speakers (in percentage terms and absolute numbers)^[9] use the urban variant of the respective variables instead of the local rural form that would actually be expected for these regions. That, for example, three speakers out of 15 (in Table 3 given as “3/15”) use the urban variant taʕmil in Siliana implies that 12 use the local form taʕmli with retention of gender differentiation. Tables 4, 5, 6, and 7 provide the details of location, age and gender of those speakers who use the urban forms instead of the expected local rural forms.

6 Discussion

Most people interviewed seem to have a rather indifferent attitude towards their dialect, simply stating that this is the way they speak and they can’t change it, whereas others, especially younger women, have a more normative approach, stating that some local expressions are māṣṭa ‘boring, tasteless’, not fīnu ‘refined’, or altogether wrong. The dialects of the urban centres are well-known and loaded with stereotypes. The picture of the urban and rural dialects (and Bedouin-type dialects for Southern Tunisia) presented here corresponds to Miller’s concise description of the sociolinguistic background of the different varieties of Arabic: “urban dialects, associated with modernity and urban cultural models but also sometimes with effeminacy and decadence; and Bedouin dialects, associated with ’aṣāla ‘purity of origin’ and Arab tradition but also sometimes with backwardness and toughness.” (Miller 2004: 181).

That their dialect changes along with the modernization of society and lifestyle seems to be a natural development for many people in Siliana and Jendouba. They stated that the past was characterized by ʕṛūšīya,^[12] by a sort of tribalism that let people live within their own tribal family groups and did not allow nor enable contact to others. In the last decades the mingling of families and individuals of different tribal origins in new neighbourhoods, universities and student residences has thus naturally led to levelling and linguistic accommodation. This fits well with Miller’s findings for Egypt: “Long-term accommodation and dialect shift are thus perceived as natural phenomena and […] are often not considered to be an indication of identity change or loss of aṣāla (purity of origin)” (Miller 2005: 918).

Our linguistic data presented in Section 5 correlates to a high degree with people’s language attitudes described in Section 4. Especially people’s judgements concerning the use of [q] versus [g] are well reflected in our data.

Altogether, it can be stated that urban features are more frequent in our Siliana-sample. The urban use of three variables scores very high in Siliana: the two [q]-words; the realisation of žū as žǟw; and the use of ǟna. Siliana is the governorate where the intersection of age, gender and level of education manifests itself very clearly, with a high percentage of young, educated women among those who use the urban features. Taking into account that the monitored interview situation might have affected people in their speech, it can nevertheless be stated that especially young, educated women in Siliana were more ready to use urban ‘prestigious’ variants than older or uneducated speakers. This leading role by young women is corroborated by Miller who states: “… in contexts of dialectal contact and change, young women tend to acquire the urban variables faster than their male counterparts” (Miller 2004: 196). It is well known that also cross-linguistically, innovation mainly proceeds from women who are “out in front” (Ringe and Eska 2013: 50–51). Eight Jendoubi and nine Siliani speakers showed no variation in the investigated variables and used only the local forms. Most of them are between the age of 40 and 65, only four between 16 and 29. There are some speakers, however, whose use of urban features has to be examined in more detail.

The use of [q] by our 69-year-old speaker from Ain Draham can surely be ascribed to his high degree of education as university professor. The speech of our 55-year-old speaker from Drija, a muezzin who spent some years in Tunis, shows three urban variables but no [q]. The female speakers from Houaidia (50) and Siliana (65) use the urban form ǟna besides the local form. The 65-year-old woman might be influenced by her daughters and granddaughters who mostly adapted ǟna in their speech. In this family it could be easily observed that urban dialectal features like [q] and ǟna become more dominant over generations with grandmothers using them rarely, daughters mostly and granddaughters almost exclusively. The 50-year-old female speaker from Siliana uses sūq and qālu and the urban inflection of III-weak verbs. Though her parents are from Siliana, she herself grew up in Tunis which explains her affinity to the dialect of the capital. An 18-year-old male speaker from Siliana used qālu and sūq in his realisation of the given sample text, but in the interview about language attitudes he states that he has a low opinion of those who adopt the qalqala: šaxṣīyithum ḏ̣ʕīfa ‘they have a weak personality’ and fǟqdīn il-hūwīya mtǟḥḥum ‘they have lost their identity’. This example shows that the cognitive and behavioural components of attitudes can diverge significantly. More than that, however, it can be assumed that the interview situation and the written text influenced his realisation.

The orientation of young speakers towards the dialect of Tunis is also an orientation towards the privileged economic and social position of its speakers. Using a certain linguistic code implies the hope for access to power and more social acceptance, a fact stressed in several (socio-)linguistic studies (e.g. Bassiouney 2018: 226, 228; Holes 1995a: 285). People who are not in need of social ‘upgrading’ or, using the concept of Bourdieu and Wacquant (1992), ‘legitimacy’ or who do not believe in the possibility of achieving power this way may not care about changing their language, for example:

1. the older generations (for many of whom their dialect is a marker of identity) who are either unwilling or incapable of changing their dialect;^[13]

2. those people whose lives and jobs do not demand a lot of contact with Tunis or other urban centres;

3. those who do already have enough power and social acceptance and are not in need of social “acknowledgement” by the urban elite.

Our interviews show that parents are in an ambivalent position: Though they might dislike it somehow that their children use [q] they nevertheless find it better in terms of future job prospects and a better life in general – including the appearance and reputation of their daughters as educated and genteel.^[14] This primarily concerns the mobile families whose lives are strongly linked to urban centres for professional or educational reasons. However, the adoption of urban features does not only spread due to mobility but also within the families – younger sisters are especially ready to take up a way of speaking which is considered more fīnu, and even mothers report to have started using ǟna instead of nǟya after their daughters had adopted it.

Although some Silianis made fun of the qalqīl ‘pronunciation with [q]’ (by imitating chickens’ clucking while saying qaqaqaqaqa), many young women are prone to taking up these features in order to be considered more refined (mhaḏḏba) and urban (baldīya). This seems to hold true mainly for women, as one speaker reported that she finds it amusing to find a boy talking with [q]. Our data also shows that women promote the use of [q], whereas men only reluctantly adopt it. Informal conversation during a conference in Tunis in February 2020 confirmed that the use of [g] has not only become acceptable but is used within the urban milieu of the capital Tunis, I quote, bǟš ywaṛṛi innahu ṯawri ‘for someone to show that he is a revolutionary’.^[15] This can be compared to Hachimi’s findings for Fessis in Casablanca, where the use of [g] especially in the verb gāl, though both for men and for women, “encodes the quality of being ‘street smart’” (Hachimi 2012: 336, 334). So we seem to have the association of [g] with virility and toughness on the one hand, and the willingness of young women to adopt [q] on the other hand, which results from an association of urban dialects with femininity (see Miller 2004: 198).

7 Conclusion

The orientation of young female Siliani speakers towards the prestigious dialect of Tunis is quite similar to other rural communities in the Arab World that have a low opinion of their own dialect, which they feel to be somehow old-fashioned. However, the dialects of regional urban centres like the towns of Siliana and Ain Draham are considered more modern than those of small villages in the countryside. In these regional centres, where speakers see themselves as more advanced in life-style and dialect than the surrounding rural areas, the adoption and acceptance of urban linguistic phenomena is stronger.

Our data presented in Section 5 shows that young women in the governorate of Siliana are most likely to adopt urban features, not only as students in Tunis but also within the family. When the family is the location where linguistic accommodation takes place, mothers, following their daughters, also tend to adopt urban features. Old and middle-aged people who are less educated, and whose children are not in contact with urban varieties either, are most prone to retaining their original dialect.

The use of [g] is highly stereotyped and bears many social meanings. It is this category which is most strongly related to gendering. Our data shows that [q] among speakers in Siliana and Jendouba is primarily used by young, educated women. This role of women regarding the use of [q] is neither stressed in Gibson (1998) nor in Jabeur (1987),^[16] though both analyse this variable. Also, in the distribution of the other variables discussed above, though not equally loaded with stereotypes and social meanings, it is obvious that young women in Siliana (to a lesser extent in Jendouba) play the most important role. They can be considered the trendsetters in language change, a tendency observed in many (socio)linguistic studies (for example Miller 2004; Ringe and Eska 2013). Contrary to Walters (1989), our contribution shows very clearly that for the regions under investigation here there is a strong sociodemographic group standing out from the other groups.