The Jebli speech between the media and the city: exploring linguistic stereotypes on a rural accent in Northern Morocco

1.1 Theoretical framework

Two key theoretical concepts underlie our analysis, namely salience and stereotype. In linking them together, we follow the model of Hachimi (2018), who tries to understand the role of salience in linguistic change or maintenance.

The concept of salience was first set out by the founding scholars of variationist sociolinguistics, most notably Labov (1972) and Trudgill (1984). Basing herself on these authors, Hachimi (2018: 63–64) defines salience as “the prominence or the degree of availability to speakers’ awareness that a linguistic form has in relation to other forms”. She argues “for a socioevaluative interpretation of salience, framing the latter as susceptibility to what Silverstein calls ‘indexical order’ (2003: 193), that is, the multiple presupposing and entailing semiotic associations between linguistic forms and social meanings”.

Rácz (2013) makes a distinction between cognitive salience, which pertains to a segment that “has a large surprisal value when compared to an array of language input” (2013: 37), and social salience, which arises when a particular variable, in addition to being especially prominent, unexpected or surprising in the normal scale of variation, provokes certain attitudes or cultural stereotypes (2013: 2). Rácz argues that sociolinguistic salience can be interpreted in one of two ways: as a label that is applicable to variables that carry social indexation, a concept akin to that of ‘markedness’; or as something resulting from external sources, namely the dynamics arising among speakers and the basic cognitive skills of the individual. If salience is rooted in speaker dynamics, “any linguistic variable could theoretically be chosen by the community to mark social indexation” (Rácz 2013: 30), a framing which is consistent with that proposed by Labov (1972). However, only those forms which entail a particular indexation, social meaning or identity, as afforded by the social dynamics of the context, may be socially salient. If the notion of awareness which appears in the first definition is also taken into consideration, the greater the awareness of the salient feature, the more likely it will become a stereotype.

Thus, the concept of stereotype is linked to the notion of salience. Silverstein (1998: 127) defines stereotypes as “beliefs about denotata of a word or expression that are not derived from strictly linguistically mapped sense categories but that, notwithstanding, give descriptive backing to uses of the term. Stereotypes are associable with words and expressions as these are used by specific, historically located groups of users in the division of linguistic labor”. In this regard, Hachimi (2018: 64) affirms that “an indicator (a first-order index) differs from a marker (second-order index) or a stereotype (third-order index) based on their degree of ideological transparency”.

Labov (1972) proposes three criteria to identify stereotypes: besides being associated with a particular dialect or sociolect, a stereotype must generate a high degree of awareness among speakers and it must also be the object of overt commentary. The speaker’s “awareness” plays a significant role here, since sociolinguistic variables may be categorized as either indicators, markers or stereotypes, depending on the level of awareness that speakers have of that particular feature. In other words, certain linguistic features may pass completely unperceived by speakers even though they index social stratification (indicators), while other features are laden with attitudes or prejudices and can signal stylistic variation (markers). Among the latter, “variables which have an especially high level of awareness associated with them are called stereotypes” (Trudgill 1984: 10). Trudgill (1984) notes that stigmatization, involvement in linguistic change, phonetic distance, and phonological contrast are all factors which can intensify speakers’ awareness of a specific feature and cause them to modify their pronunciation under certain circumstances, reflecting a sort of ‘linguistic insecurity’. According to Labov (1972: 117), such linguistic insecurity is illustrated “by the very wide range of stylistic variation used by lower-middle-class speakers; by their great fluctuation within a given stylistic context; by their conscious striving for correctness; and by their strongly negative attitudes towards their native speech pattern”.

Bennis (2001a, 2001b, 2002, 2003 frames linguistic insecurity in the Moroccan context as a quest for linguistic legitimacy. His research focuses on epilinguistic speech and attitudes toward the different languages – Berber, Arabic and French – present in the Beni Mellal area of Central Morocco. Linguistic insecurity, i.e. the fact that speakers affirm not to use features while they actually do^[4] – such as the use of the second person feminine imperfective verb form when addressing a man, e.g.: aži tgəlsi ‘come sit down’ (instead of the masculine form aži tgləs (cf. Bennis 2003: 355) – leads to the emergence of various linguistic strategies, ranging from the avoidance or rejection of stigmatized forms to linguistic accommodation or even the ascription of a positive value to stigmatized forms. Such strategies for dealing with linguistic stigma can be encountered in other Arabic-speaking communities such as that of Upper Egypt.^[5]

1.2 Literature review

Extensive research has been carried out on linguistic variation and change in many parts of the Maghreb. However, we will limit ourselves here to noting a few key works centred on linguistic stereotyping and stigmatization which, even though they focus on varieties outside Morocco, are relevant to the case under study. Walters (1989: 297) conducts a study on three phonological variables in the Tunisian context. He demonstrates that one of the variants of the variable /ɛ:/ is regarded as a stereotype and, therefore, avoided by young and better educated people. Thus, the variable is found to be a marker of age, class, sex and provenance, and is on its way to becoming a marker of style. Gibson (1998: 107, 258) reports that the non-monophthongized reflex of the Old Arabic short diphthong *ay is generally associated by Tunisian speakers with the dialect of Sfax, a coastal city whose dwellers are the subject of mainly negative stereotypes (arrogance), but also positive ones (efficiency). In his study of the Arabic variety spoken in Tlemcen, Algeria, Dendane (2013) describes how male speakers accommodate their speech when interacting with people from other locations. Accordingly, focussing on */q/, he concludes that Tlemceni males avoid the glottal stop [ʔ] variant, adopting instead the velar [g] characteristic of rural speech, whereas Tlemceni females tend to be linguistically more conservative and maintain the [ʔ] in all communicative situations.

Even more pertinent – because of our focus on linguistic representation in vernacular varieties in Morocco – are the various studies on linguistic variation and change triggered by stereotyping and stigmatization. In her series of studies analysing the linguistic practises of Casablancan speakers who are originally from or have a family background in Hachimi (2007, 2011, 2012, 2018), Berrechid and Tafilalt (Hachimi 2018), Hachimi discusses the social meanings that have led to variation and/or accommodation of different phonological and morphological features. Similarly, Falchetta (2019b) examines how the sociolinguistic status of three variable features have changed in an urban Moroccan Arabic variety, namely the alternation between [g] and [q], affrication of /t/ and alternation between the verbs /dwa/ and /hdərˁ/.

Two other studies centre on variation in the realisation of */q/. While Moumine (1995) looks at variation between [q] and [g] in Casablanca and finds that “variability in linguistic behaviour is also socioculturally constrained”, Vicente (2021) explores the variants [Ɂ] and [q] of the variable /q/ in Tetouan, concluding that the value of this trait varies according to social group, with the elderly regarding it as a prestige form and the younger generations stigmatizing it. Also in connection with stigmatization, Messaoudi (1999) reports that the stigmatization attached to [ǧ], which is characteristic of Jebli speech, tends to be stronger among educated speakers. Finally, Barontini and Ziamari (2009) explore the different motivations underlying the use of features characteristic of ‘male’ speech by women of different ages in Meknes.

The vernacular varieties spoken in Larache and Ouezzane, however, have attracted relatively little scholarly attention. Concerning Larache, the last century saw the publication of three short story collections in the Larachi vernacular (Alarcón y Santón 1913; Klingenheben 1927; Marchand 1905) and the stories contained in the first of these have been dialectologically analysed (Moscoso García 2003b). More recently, Guerrero (2015) described the current synchronic situation of this variety. As for the Ouezzane vernacular, most studies in this area are either only partial or reflect ongoing research, such as Jaouhari’s (1986) master’s thesis focussing on the verbal system, Khoukh’s (1993) undergraduate thesis describing the main particularities of the variety spoken in Ouezzane and its hinterland, Ech-Charfi’s (2016) analysis of the notion of urbanization by comparing the Ouezzane and Tangiers varieties and El Khomssi’s (2017) preliminary study of the speech of a weekly market seller living in the countryside and working in the city. More recently, in her doctoral dissertation, El Khomssi (2020) describes the linguistic particularities and social representations of two tribes from the Ouezzane province, the Rhouna and the Beni Messara [sic]. Her informants were either living in the tribes’ original rural locations or had migrated to the large urban centres outside the Ouezzane province. Finally, in a series of articles, Benítez Fernández (2016, 2019, 2022) analyzes variation related to age, gender and rural versus urban speech styles in the Ouezzani vernacular. As can be seen, most of these studies are descriptive analyses of the diatopic variation. Ech-Charfi, El Khomsi and Benítez Fernández have a more sociolinguistic approach, but none of them has pointed out the issues of stigmatization and linguistic insecurity so far.

3.1 Phonetics

Dental stops /t/ and /d/ may undergo spirantization when in intervocalic or postvocalic position, as exemplified in (1).

The spirantization of dental stops /t/, /d/, /ḍ/, the bilabial /b/, the velar stop /k/ and to a lesser extent the uvular /q/ is a common feature in most varieties of Jebli Arabic.^[8]

Another interesting phonetic feature found in our corpus is the unconditioned devoicing of the pharyngealized voiced dental stop /ḍ/ (2). However, it is worth mentioning that this sound change occurs sporadically and does not affect all lexical items displaying an etymological or non-etymological pharyngealized /ḍ/ (3).

Pharyngealized /ḍ/ may devoice to /ṭ/ in most, but not all, Jebli dialects.^[9] This phonetic phenomenon also occurs in the speech of the inhabitants of Tangiers and Tetouan, two Northern Moroccan urban centres whose dialects have been strongly influenced by the Jebli varieties spoken in their surroundings.

As is often the case in several Northern Moroccan varieties, Old Arabic */ǧ/ is realised as a voiced palato-alveolar affricate [ǧ] when geminate or in the vicinity of /r/ (4). This affricate realization has been seen by some scholars as an allophone of fricative /ž/, the main reflex of */ǧ/ in Moroccan Arabic. Nevertheless, its unconditioned occurrence in a number of Jebli dialects suggests that affricate /ǧ/ is a phoneme rather than an allophone of /ž/ (Messaoudi 1996: 173).

Voiceless affricate /č/ occurs instead of voiceless fricative /š/ in the verbs šāf ‘to see, to look’ and šədd ‘to catch, to take’ (5). Such instances of affricated /š/ are generally explained as resulting from a reanalysis of the cluster tš as a single element č in forms with the t- prefix (e.g. tšədd ‘you take’), a pattern which is then generalized to all forms in the stem (Heath 2002: 139).

A further phonetic feature in our corpus is the general treatment of voiced bilabials /m/ and /b/ as “sun letters”, i.e. as sounds that trigger the assimilation of the definite article l- (6). This trait has been reported for several Jebli varieties.^[10]

3.2 Verbal morphology

In Moroccan Arabic, verbal predication is negated by means of the discontinuous marker ma___-š. The second element is an enclitic which always surfaces as –ši in the speech of the Jebli character in the Jebli & Beldi comedy sketches (7). This pattern is consistent with the available data on Jebli Arabic.^[11]

Another morphological feature which might be ascribed to Jebli Arabic and which we have found in the corpus is the lack of paradigm reconstruction in the imperfective plural forms of defective verbs^[12] (8).

Like many other Arabic dialects, Moroccan Arabic makes use of certain particles which precede the imperfective in order to express temporal and aspectual nuances. These so-called “imperfective auxiliary particles” are either ka- or ta- in most Moroccan varieties.^[13] In our corpus, though ka- predominates, we found a few instances of la-^[14] (9), a variant which may be regarded as typically Jebli as it has not been recorded outside the Jbala region.

On the basis of its alternation with ka-, as well as its meagre presence in the corpus, it could be argued that la- is slowly losing ground to the more prestigious ka-, which prevails in the “mainstream dialects of Moroccan Arabic”.

3.3 Nominal morphology

Another typical Jebli feature which has also spread among most urban varieties in northwestern Morocco and which is widely attested in our corpus is the use of the 2nd person singular gender-independent pronoun ntīna ‘you’ (10), as opposed to the 2nd person singular masculine nta and feminine nti variants which predominate in mainstream Moroccan Arabic varieties. Although the lack of gender distinction in 2nd person singular pronouns is common to many pre-Hilali varieties throughout the Maghreb, the form ntīna (with a final -na) might be seen as typically Jebli as it also occurs in the genetically related village dialects of the Msīrda and the Trāra in northwestern Algeria.^[15]

An odd morphological feature of several Northern Moroccan dialects is the 2nd person and 3rd feminine singular imperfective prefix d-.^[16] In the Jebli & Beldi corpus analysed here, the Jebli character alternates the use of this prefix with the most common variant t- (11).

As has been shown above, most of the features that depict the speech of the Jebli character in the series of sketches under study are actually attested in Jebli dialects, a fact that may suggest speakers perceive them as typically Jebli, in their stereotypical portrayal of Jeblis. The conscious choice of these features when writing the sketch script indicates that they are perceived as salient. This fact may allow us to regard them as markers.

4.1 Phonetics

The voiced postalveolar fricative /ž/ belongs to the repertoire of most Moroccan varieties. However, alongside this form in our corpus of recordings from Larache and Ouezzane we encountered on rare occasions another variant, the voiced postalveolar affricate /ǧ/, which is characteristic of Jebli varieties.^[17] These rare instances were produced by informants with marked Jbala origins, as documented by Guerrero (2015: 47) and Benítez Fernández (2022) in their respective studies. It is worth noting that the affricate variant was explicitly stigmatized by Larachi informants, who perceived it as being “too rural” and as betraying a Jebli accent.

A further distinguishing phonetic trait involved the realisation of */q/ as a voiceless uvular stop. Beside this realization, two others characterizing different varieties in Morocco could be also found in our corpus. For instance, the voiced velar plosive [g] identifies some Bedouin dialects, and the glottal stop [Ɂ] characterizes some sedentary varieties – both in the old centres (medinas) of cities like Tetouan or Fez and in several Jbala villages.^[18] The [q] is the most generalized variant all over Northern Morocco and our corpus was not exceptional in this regard. The voiced velar plosive [g] was restricted to certain lexical items and was used by younger informants with a high level of education who had experienced mobility. As for the glottal stop, this trait could be found, particularly in Ouezzane, either among men of different ages who had only basic education and worked as craftsmen, or among women who exhibited a marked Jebli style, not only in their speech, but also in their attire.

With regard to informants’ attitudes to this feature, it became clear that the glottal stop [Ɂ] variant is overtly censured by the Ouezzani population, with some of our informants regarding it as characterizing the speech of “country folk” (n-nās dyāl l-bādya).^[19] This is a very interesting finding in that [Ɂ] has been shown to hold a positive value within the speech communities of some Northern Moroccan old medinas such as that of Tetouan. Conversely, the fact that Ouezzanis seem to link [Ɂ] to a rural background could be due to its occurrence in several Jebli dialects. This association with “ruralness” is reinforced by the findings reported in El Khomssi (2017, 2020.^[20]

4.2 Nominal morphology

The only salient morphological feature regarded as Jebli in our spontaneous speech corpus was the use of the 2nd person singular gender-neutral pronoun ntīna ‘you’. Guerrero (2018) has identified this pronoun as a hallmark of Jebli Arabic, though it is also attested in some Northern Moroccan urban varieties (mainly Tangiers and Tetouan), probably owing to considerable rural-to-urban migration within this region. As far as Ouezzane is concerned, while the use of ntīna seems to have been a very common feature in the past (Khoukh 1993),^[21] data elicited by Benítez Fernández (2014–2018) suggest that this variant is slowly losing ground to other forms with gender distinction, given that it only obtains in the speech of informants with the lower prestige Jebli background. As regards Larache, speakers could be separated into two groups according to the presence or absence of gender distinction in the 2nd person singular pronoun they used. One group comprised those speakers who used gender-neutral ntīna, while the other was made up of those speakers who distinguished masculine nta-ntāya from feminine nti∼ntīna∼ntiyya. Interestingly enough, the use of gender-neutral ntīna was disapproved of by some of our informants, who perceived it to be a feminine form, and hence inappropriate for talking to a male interlocutor. They argued that a man addressed as ntīna might take offence because his manhood was being called into question. Surprisingly, some of these same informants denied using ntīna for both genders in their own speech, yet the audio recordings clearly disproved this. This denial would seem to be a clear-cut case of linguistic insecurity, as already pointed out in Bennis (2001a, 2001b, 2002, 2003. The indexing of the ntīna variant as “effeminate” has already been reported in works on Moroccan varieties since it has already led to attrition in the use of the pronoun for both masculine and feminine among various heritage communities living in Casablanca, and even the reconfiguration of the verbal paradigm (Hachimi 2012, 2018). It is very conceivable that the same attrition process is underway in the dialects of Larache and Ouezzane, especially bearing in mind the fact that the masculine nta (you) and feminine nti (you) variants are not ideologically marked.