Mahdia Dialect: An Urban Vernacular in the Tunisian Sahel Context

Abstract

This paper aims to present some preliminary results of the linguistic analysis of the dialect of the Wilāya of Mahdia on which few studies exist, focused mainly on phonology. My analysis, here extended to the morpho-syntactic level, is based on a corpus of interviews taken from some social media pages. The sample will be composed of respondents of different geographical origin (from Mahdia and some nearby towns), gender, age and social background. A deeper knowledge of the Arabic of Mahdia region, which is a bundle of urban, Bedouin and “villageois” varieties, would contribute to throw new light on the features of the Saḥlī dialects and would add a small piece to the complex mosaic of Tunisian and Maghrebi dialects, whose traditional categories of classification should be reconsidered.

1. Introduction

During the recent years, the need to more precisely examine and describe the varieties of Arabic used in the Tunisian Sahel has been central to the scientific dialectological debate. In fact, there are few systematic studies available on Sahel varieties and the data offered need to be partially reinterpreted. In 1950, William Marçais, with regard to “les parlers villageois” of Tunisia, claimed: “on a surtout en vue ici ceux des bourgs et des petites villes du Sahel […] n’ont fait encore l’objet d’aucune enquête” (Marçais 1950, pp. 210–11).1 The scholar classified the non-coastal dialects of Sahel as “villageois” and introducted a third category of dialects sharing both sedentary and Bedouin features (see also (Marçais and Guiga 1925, p. XXV)).

The first systematic study on a Sahel variety is Textes arabes de Takroûna by William Marçais. According to him (Marçais and Guiga 1925, p. XIX), “Le parler arabe de Takroûna concorde dans l’ensemble avec ceux des centres agricoles, bourgs et villages, qui parsèment la region côtière de la Tunisie centrale, communément appelée Sâḥel […] depuis le moyen âge. Séparés les uns les autres par des différences de detail, ces parlers relèvent, quant à la phonétique et à la grammaire, d’un même type general dont le takroûni n’est qu’une variété particulière”.2 Marçais and Guiga offered a linguistic study and also a monumental glossary of the variety considered (Marçais and Guiga 1958).

In 1980, Talmoudi classified the Tunisian dialects in four groups: The varieties of Sahel, which are urban and semi-urban, these latter “spoken in small villages as Ksibet Sūsa and Khnis display features of both nomadic and sedentary dialects”. The Northern dialects are also “divided into two types: urban and rural. The villagers on the left side of Oued Medjerda speak so-called ʒba:li dialects (mountain dialects) which have features in common with North East Algerian vernaculars”. The Central Western dialects are rural and nomadic. The rural ones “resemble in several respects the East Arabic dialects”. The Southern dialects are divided in three groups: “urban dialects in Sfax, rural in the oasis and nomadic. The latter is spoken by semi-pastoral people in Sahara” (Talmoudi 1980, pp. 10–11). “The genuine dialect of Sūsa”, according to the scholar, is spoken in the Medina by the older generation, but the author also took into account the innovations of the younger generation for the composition of his study on the Arabic of Sousse (Talmoudi 1980, p. 13).

Lajmi, in 2009, Zammit, in 2014, and Sellami, in 2019, conducted some investigation on some features of Sfaxi Arabic, providing new elements for the knowledge of the dialects of Sahel.

Mion (2015, 2018)3 offered some reflections on the origin of this “third category” of dialects, that is the “villageois”, and on the phenomena that characterise Tunisian “villageois” dialects, whose ‘mixed’ features are the product of a long process of interdialectal contact between a sedentary and a Bedouin variety of Arabic. The latter was introduced by the Banū Hilāl who invaded the Maghreb around the 11th century.

Mion and Luca D’Anna during the “Prima giornata di dialettologia maghrebina” (Cagliari May 16 2019), launched the research project “The Tunisian Sahel: Dialectological, Historical and Sociolinguistic Perspectives”, which aims at shedding new light on the features of the Arabic varieties of Tunisian Sahel, which, in fact, should be better described regardless of the existing rigid classification criteria (Bedouin/urban/rural), which do not highlight the richness characterising the varieties spoken in the region (Mion 2015). A century after Marçais’ studies, however, no systematic research on the Bedouin and urban varieties of the Sahel region has been conducted, yet.

My paper intends to be a modest contribution to the knowledge of the varieties of the area, starting from the analysis of the dialect(s) of Wilāyat al-Mahdiyya (henceforth Mahdia).

Marçais and later scholars have included Mahdia Arabic among the varieties used in the coastal towns of Sahel, such as Monastir, Sousse and Sfax. Saada (1984, p. 17) included the urban dialect of Mahdia among “les parlers arabes des capitales”, but she also stated that she did not have elements to classify the varieties used in the neighbouring villages.

At present, there are few studies entirely dedicated to the Arabic of Mahdia, and the dialects of the surrounding towns and villages have never been described. Attia (1969) offered a phonological analysis of the variety used by the fishing community in the 1960s. What emerges from his paper is a quick description of the phonological inventory of the Mahdia dialect, accompanied by few examples. In his paper, the scholar highlighted some well-known features of Mahdia Arabic: The passage of the interdental /ṯ/ and /ḏ/ to /d/ and /t/, the reduction of the diphthongs /ay/ and /aw/ to /ē/ and /ō/ in the middle of a word and the voiced articulation of /q/ in some terms such as garʿa “irrigated land” and gamra “moon” (Attia 1969, p. 125). He also quickly focused on combinatory phonetics, vowels and vocalic phenomena such as final imāla, syllable patterns and prosody. Based on these phenomena, Attia defined Mahdia Arabic an urban variety.4 Yoda’s study (2008) added some important information to the knowledge of the vocal system of Mahdia dialect. In fact, the scholar focused on the special status of Mahdia Arabic having /ē/ and /ō/ phonemes, unlike the other sedentary dialects, and attributed the presence of these phonemes to the influence of the nearby village dialects. Besides, in 2019, Yoda published some texts in Mahdia Arabic, accompanied by a simple grammatical sketch concerning mainly phonology. In his paper, he writes that Mahdia Arabic: “is an eastern Maghribi sedentary dialect showing some features of the village dialects of the region, the most conspicuous being the word-final imāla ē. Among the sedentary dialects of Tunisia, Mahdāwī dialect […] is characterized by a five-long-vowel system […] and the correspondence of the interdentals of Old Arabic […] to plosives, as attested in most of the Jewish dialects of Tunisia. In this respect the dialect in question is worthy of more detailed descriptions”. (Yoda 2019, p. 55).

It is clear, however, that the few existing studies have focused mainly on the vowel and consonant system. Therefore, the linguistic analysis of Mahdia variety should be extended to all linguistic levels. It would also be fruitful to extend the research to the varieties used in the territory near the town, which remain to be investigated (except for D’Anna’s study about Chebba: see (D’Anna 2020)). In fact, Mahdia Arabic is worthy of study because it is surrounded by village varieties and, since no language can live isolated and separated from the adjacent varieties, it could have some peculiarities owing to contact phenomena. That is, as Yoda already showed, it is an urban variety containing some village features.

In addition to this, a deeper knowledge of the Arabic of Mahdia region, which is a bundle of urban, Bedouin and “villageois” varieties, would contribute to shed new light on the features of the Sahel dialects and would add a small piece to the complex mosaic of Tunisian and Maghrebi dialects. In fact, a deeper knowledge of the varieties spoken in this area would also be useful to understand through which historical and socio-linguistic dynamics these dialects originated, since we have scant data on the history of Tunisian dialect(s). More generally, this would also help provide a more detailed and specific classification of the dialects of the area (Taine-Cheikh 2017; Guerrero 2018; Benkato 2019), which share many common features, but present many differences that deserve to be highlited.5

Mahdia and the surrounding towns

Mahdia is a town whose importance and splendour are rooted centuries ago. The geographer al-Idrīsī (d. 1175–1176) describes Mahdia as a beautiful town, two days from Sfax and Kairouan. Mahdia, not long before his arrival, had a harbour visited by merchant ships from everywhere: Maghrib, Mashriq, al-Andalus and Christian countries. al-Idrīsī informs us that, during those times, Mahdia was already famous throughout the world because of its goods and renowned clothes that were exported to all the other countries, but he also adds that, since the Norman conquest, its trade has been strongly reduced. Moreover, the geographer defines the wall surrounding the town a wonder worthy of mention (Bresc and Nef 1999, pp. 183–84). Mahdia, during al-Idrīsī’s times, was composed of two towns: Mahdia, the seat of the power and the residence of the sovereign, and al-Zawīla. The latter was beautiful and densely populated with merchants. The geographer writes of Mahdia with a sort of nostalgia because the invasion of the “Arabs”, that is the Banū Hilāl,6 and of the later Norman conquest, destroyed many important and emblematic aspects of the town. Mahdia, however, remained the capital of Ifrīqiyā (Bresc and Nef 1999, pp. 183–86).

Nowadays, the vestiges of a flourishing medieval city can be retraced in its old madīna. Mahdia was built on a peninsula (see Figure 1), situated on the Eastern coast in the centre of the Republic of Tunisia. It is 200 km from the capital, Tunis, has a mild climate, which is usually affected by the Mediterranean air currents, and its economy is based on agriculture, especially on oil production, fishing and craft industries specialised in producing silk, leather clothing and mosaics. Thanks to its position on a 75-km-long coast, close to Sousse and El Jem, tourism also plays an important role in its economic activities.7 Mahdia is a town of pre-Hilalian foundation that later underwent the Hilali invasion. Founded in 909 A.D. by the Fatimid Caliph ʿUbayd Allāh al-Mahdī as the new capital of the Reign, replacing Kairouan, the Aġlabid capital, Mahdia was the first urban settlement in the peninsula. The history of the town is quite complex. The ancient nucleus was the already mentioned quarter of al-Zawīla, which was also the commercial core. During the Hilalian invasion, Mahdia regained its role of capital, after a short time in which the capital was al-Manṣūriyya. In 1087, Mahdia was conquered by Pisans and Genoans, then by Normans in 1123, in 1134 by the Hammādids and, in 1140, the Normans of Sicily imposed harsh conditions on the town. Then, Roger II caused the end of the Zīrid Dinasty Until the French protectorate of 1884, and the town was attacked and conquered by numerous dynasties and conquerors and was even destroyed and plundered (Talbi 1986, pp. 1246–947). The suburb of al-Zawīla was completely destroyed during the Hilalian conquest and rebuilt in 1200. As a consequence of centuries of riots, pillages and plagues, in the 16th and 17th centuries, the composition of the population changed above all because of two elemets: The arrival of the Muslim refugees coming from al-Andalus and the introduction of the Turkish garrisons. According to Talbi, in 1986, 60% of the population was composed of descendants of the Kouloughlis (Bearman et al. 1986, p. 366),8 affecting onomastics and customs (Talbi 1986, p. 1247).

Goitein (2010, p. 311) claimed that Mahdia and some coastal towns of Tunisia resisted the perpetual chaos in which the Hilalian invasion threw the country,9 but their hinterland was lost and was exposed to the subsequent attacks from the Normans and Italians. In the Geniza documents, the Christian and the Almohad conquests are well documented as being a catastrophe and a cause of economic decline of the region. Despite this, the economic exchanges between Tunisia, Italy, Spain and Syria continued, but some merchants with their families were obliged to move to Sicily and Egypt. Moreover, in the 11th century, the repeated pillaging by the Banū Hilāl obliged the inhabitants of the villages to seek safety within the walls of the surrounding towns; besides, some nomadic populations have returned to the steppes (Decret 2003). These population movements, only in part caused by the Hilalian invasion, must have had an important influence on the Tunisian Arabic language, regarding which we currently have scanty or no data. Mion (2015, pp. 275–76) distinguishes the two phases of the Arabization of Sahel and underlines some relevant historical events which had linguistic consequences: In the pre-Hilali period, the region was probably inhabited by some sedentary Arabized people, and during this phase, the main urban features of Sahel Arabic developed. The Hilalian invasions of the 11th century deeply troubled the region to the extent that many urban centers and villages were threatened and disappeared, so that Ibn Ḫaldūn (d. 1406), in the 14th century, writes that only some weak traces of sedentary culture could be found in some families from Kairouan or Mahdia. A relevant piece of information, linked to the population movement in Mahdia region and its consequences on the Arabic of the region, is that “en 945–46 Ismāʿīl al-Manṣūr […] quitta la ville de Mahdia pour établir sa résidence à Sabra, en provoquant la ruine de l’ancien siège de l’empire fatimide et la perte des habitants et ses faubourgs, ce qui nous incite aujourd’hui à voir en cela un évenement qui laisse le champ libre, plus tard, à un repeuplement de la part de gens beaucoup moins urbanisés” (Mion 2015, p. 275).

If we continue our imaginary journey, guided by al-Idrīsī, we find Monastir (al-Munastīr) at 30 miles of navigation from Mahdia. The geographer repports only that the town has some castles in which some fruits are produced and then exported to Mahdia and that the inhabitants of the latter bury their dead in the cemetery of Monastir (Bresc and Nef 1999, pp. 184–85). Actually, after the Arab conquest of the mid 8th century, Monastir became renowned for religious reasons because of its ribāṭ and its cemetery in which important personalities, such as the last members of the Zirid dinasty, were buried.

As recounted by the Imām al-Māzarī (d. 1141), Monastir and Mahdia appear to be very connected and seem to have been spared from the Hilalian invasions. Monastir was described as a prosperous town whose religious importance was proved by the piligrimage of numerous people from the nearby regions, similarly to what happened in the holy city of Kairouan (Soucek 1993, pp. 227–29).

Monastir is a town located on the southern end of the Gulf of Hammamet, about 160 km south of Tunis, and today is part of the Wilāyat al-Munastīr (see Figure 2). Its main commercial activities are tourism, the textile industry for wool processing, the production of salt, which is an ancient activity already described by al-Bakrī (d. 1094) due to the fact that the town was built near a salt pan, soap, olive oil and fishing. Monastir is also a university town (for further details on the history of Monastir, see (Soucek 1993, pp. 227–29)). 11

Msaken (Arabic: M’sākin) is a small town of the Tunisian Sahel located at a dozen kilometres to the south of Sousse. Administratively dependent on the governorate of Sousse, it has an increasing number of inhabitants in the summer due to the return of expatriates, most of whom work in France.12 For this reason, its inhabitants call Msaken “la Petite Paris”. Its economy is based on olive oil production. There are, however, a number of handicraft and industrial enterprises in the surrounding area, set up by old emigrants who returned to Msaken in the 1980s (Ma Mung 1984).

Msaken was founded during the Hafisd dynasty in the 14th century (Bouhlel 2009, p. 125). A brief description of Msaken as a holy place in Tunisia together with Kairouan is contained in Tunisie et tunisiens by François Bournand (Bournand 1893, pp. 311–13) who actually quotes the information offered in Promenades d’une Française dans la régence de Tunis by Voisins d’Ambre (1884, pp. 171–74). Bournand states that Msaken is a small town of about 9000 inhabitants, 9 km from Sousse, built in a slightly mountainous area planted with olive trees, famous as a religious place because of the madrasa of Sīdī ʿAlī b. Ḫalīfa, renowned in North Africa because of the high number of students and the high level of teaching of its university. Msaken is considered by the author to be like Seville, Padua, Oxford or Cologne. The town occupied a large area with houses surrounded by greenery, while the centre of the town developed along the main road and had several schools and mosques. According to M.me de Voisins, entry to the holy city was forbidden to Christians and especially Europeans. For this reason, the author claims to have seen Msaken from a nearby hill and that she found it fascinating to the the point that she was very impressed by it. Then, she gives a rather exotic and Eurocentric description of the town (Voisins d’Ambre 1884, pp. 171–74; Bournand 1893, pp. 311–14).

Ma Mung (1984, p. 163), stated that in 1984, Msaken had 41,219 inhabitants and that it was among the largest 15 or 16 towns of Tunisia, together with Mahdia. In his study, he focused on the importance of the migratory movement that affected a large part of the population from the 1960s onwards, which had important effects on the economy of the country.Today, Msaken is the second town of Sahel in terms of size; its inhabitants constantly move back and forth to the rural and village hinterland as well as to Sousse. The growing integration of this town into the economic space of the regional metropolis, Sousse, means that it has become a basin of intense activity (industrial, construction and services) and, above all, of employment for the populations of the surrounding areas. The main categories of commuters are those employed in industry and government, especially teachers (Boubakri and Lamine 1992).13 According to Bouhlel (2009, p. 126), Msaken is close to several small towns that show different linguistic features from each other as well as some common features. The scholar pointed out that Msaken Arabic has some peculiarities that distinguish it from the other Tunsian varieties. Moreover, this variety is in continuous evolution because of its inhabitants’ migration movements, and therefore the “original” Msaken variety would be spoken only by aged people, children, emigrated and housewives.

2. Materials and Methods

This paper offers some preliminary results of the linguistic analysis of the dialect(s) of the Wilāya of Mahdia, based on a corpus of interviews taken from some social media pages, which will be presented below. The sample is composed of speakers of different geographical origin, gender, age and social background. The videos analysed have been collected between 2019 and 2021 and the speakers speak varieties of Tunisian Arabic that are more or less dialectal in the sense that they can be more or less influenced by Modern Standard Arabic, and may have mixed features, depending on the the speakers’ levels of education and their role (i.e., politicians and teachers tend to use a higher register).14 Moreover, their speech can be more or less influenced by the medium used, i.e., radio or video. The role of the Arabic of the capital Tunis and of the main city of Sahel, Sousse, will be highlighted too.15

Starting from the previous studies on Arabic dialectology, and particularly of the Maghribi area and on Tunisian Arabic, I will observe and analyse some of the urban, Bedouin and village isoglosses indicated by Marçais in his studies (Marçais 1950, pp. 207–14) and I will then add some selected morpho-syntactic elements attested in the Arabic of Mahdia.

Even if the focus of this study is the urban vernacular of Mahdia, some linguistic elements of some nearby towns will be underlined in order to follow the suggestions of earlier scholars. For this reason, some video-recordings of speakers from some small villages near Mahdia, such as Tlelsa and Teboulba, have been analysed as well as audios of respondents from Monastir and Msaken, since, according to Marçais, the two towns have village features (Marçais 1950, p. 207). Similarly, Saada’s statement “On a recueilli en outre des informations concernant le parler de Bqalta16 qui ne suffisent pas à le classer (parler de Musulmans)” attracted my attention.17

Therefore, several dialogues and interviews contained in numerous online radio and tv programs and social media information pages have been analysed. Due to the difficulty of determining the real background and origins of the speakers involved in the videos available online, deriving from the analysis of online material, I have selected mainly the materials in which the origin of the speakers has been specified by the speaker or the interviewer.

Souces for Mahdia and Monastir Arabic

For linguistic data on Mahdia and Monastir, I consulted the following online radios that broadcast on their relative Facebook pages:

∗

Menara Fm (http://www.menarafm.net/ accessed on 19 August 2021) is a radio station based in Mahdia since 2019 whose editorial line is based on independence and freedom of expression. It offers many interviews with Tunisian artists and craftsmen and those from other countries in the Arab world, as well as reports, news and live radio broadcasts focused on debates on topical issues chosen by the radio speakers. It also broadcasts some fixed radio programs, such as Nahārek zīn, i.e., “Have a nice day”, in which every morning some presenters give the weather forecasts and the main news of the day.

∗

Radio MFM Tunisie (http://www.mfmtunisie.net/ accessed on 19 August 2021) is a generalist radio station that broadcasts in the regions of Mahdia, Sousse and Sfax and offers programs on sport, news, music and society. It also has daily news programs such as Hadret l-youm, i.e., “Conversation of the day”, and Mid Mag, broadcasted at midday.

∗

Radio Mahdia (https://www.radio-mahdia.info/ accessed on 19 August 2021) broadcasts almost exclusively news programs.

∗

Radio Mahdia 1 (http://www.radiomahdia1.com/ accessed on 19 August 2021) is specialised in tourism and cultural issues. It offers news programs, interviews of the citizens made on the road and interviews of local artists taken in the radio studio, and has a program devoted to children, called Matinal children.

∗

Dār al-šabāb Ražīš (https://tinyurl.com/2fmb88fx accessed on 19 August 2021) is a web radio based in Mahdia, and particularly in Rajiche, a small coastal town in the Mahdia Governorate, created under the patronage of the Ministry of Youth and Sports of the Tunisian Republic. Among the programs broadcast, Mahdia taḥkī, that is “Mahdia speaks”, is interesting since it often brings together people from different generations who discuss a cultural or social topic. This helps analyse the language of speakers of different ages and sex.

∗

Radio Monastir (http://www.radiomonastir.tn/ accessed on 19 August 2021) not only provides news from Wilāyat al-Monastir, but also from all Tunisia, and was very useful for analysing the speech of natives of Monastir and Mahdia. It offers news programs, reportages and topical talk shows.

∗

I also analysed videos taken from the Facebook pages:

∗

Revolution Mahdia, which is a page open to all the inhabitants of Mahdia region whose mission is to foster and promote the freedom of expression and circulation of information in the zone. This is often done through the broadcasting of interviews of the citizen taken on the road.

∗

News Mahdia and Ville de Mahdia, both broadcasting news, social and entertainment programs.

Sources for Msaken Arabic

News on Msaken is provided by the online tv channel Msaken Tv and by Radio RM FM, which is a radio station based in the town broadcasting entertainment programs and news. Both broadcast in their Facebook pages.

Sources for Bekalta Arabic

As regards Bekalta, I analysed some videos from the Facebook pages 100% ba9louti and Bekalta Today, which provide daily news, reportages and interviews about the town.

Except for the programs involving children and young people mentioned earlier, the speakers involved in the radio programs analysed are men and women of all ages.

When analysing an oral corpus on media or on social media, the researcher should be aware of many challenges of such a work. In fact, as Van-Mol (2010, pp. 67–68) states, “people of all layers of society appear in these media to participate, by means of oral expression, which reflects a wide variety of language capabilities and layers”. Van-Mol also distinguishes two main classes of speakers: The first is made of professional workers, such in the case of radio presenters in my corpus, or intellectuals who use a top-down strategy of communication by using a higher register with some dialectal features in order to also be understood by less educated people. The other is composed of less educated people and non-professional workers, i.e., the majority of the speakers in my corpus. However, even if, as Van-Mol states, oral media Arabic gives us the opportunity to observe how people with different linguistic competences communicate with each other, many elements that can influence the linguistic variety and register used by speakers should be taken into account. First of all, the kind of media and of program in which they are involved influences their linguistic choices as far as regards the register and the spontaneity of their speech. Furthermore, the different linguistic competences of young and aged people and of women and men (i.e., linked to age and sex) have a role in their linguistic production. In addition, the topic dealt with and kind of oral text produced by them, such as dialogues, interviews, monologues and multilogues, are important. Moreover, the audience who is going to listen to the speakers has to be taken into account because it has an influence on the speaker’s linguistic choices.18

In spite of these challenging aspects, conducting a linguistic analysis through social media resources has some advantages. Firstly, it offers the opportunity to observe many examples of informal communication among several individuals. In fact, it is possible to reach many speakers in different locations, such as small towns and villages thanks to their social media pages. Secondly, it allows research to continue despite the sanitary crisis by remotely studying a linguistic variety and formulating hypotheses that will then be verified during the fieldwork.

In fact, even if using social media resources for linguistic analysis has some advantages, it cannot substitute fieldwork. For this reason, the first results presented below will be compared and integrated with those obtained from a period of field research in which I will record some interviews. This was the very first aim of my research, but due to COVID-19, reaching Tunisia has become impossible, at the moment. The data provided by the in-field interviews will show further elements useful in verifying the hypothesis formulated and will provide a more precise description of the varieties used in the Wilāya. During the field work, eventual phenomena linked to diatopic and diastratic variation will be highlighted in order to point out the possible local variants belonging to any social group such as older and younger people, men and women. Moreover, any possible diachronic element that might have contributed to the formation of Mahdia Arabic will be identified and described. This will be done by choosing a sample of speakers that is as varied as possible and by consulting some Arab historical and geographical works.

Some selected phonological, morphological and syntactic features will be presented below.

3. Results

3.1. Phonological Remarks

3.1.1. Interdentals

It is well known (among others, (Marçais 1950, p. 201; Attia 1969, p. 22; Yoda 2008, 2019)) that in Mahdia, Arabic interdentals ḏ/, /ṯ/ and /ḏ̣/ are substituted by the respective dental consonants /d/, /t/ and /ḍ/. The data collected through the online social media pages analysed are not so unequivocal and show a certain variation.19

/ṯ/

There were several cases of dental realisation of /ṯ/ in /t/ in initial and median positions: istitnā՚iyy, istitnā՚iyya “exceptional”, aktar “more”, al-tēnī20 “the second”, ymattǝl “represents/plays”, ātār “antiquities, ruins” and the plural ātārīn, el-ǧimʿa l-tēnī wella l-tlēta “the second or the third week”. The interdental phoneme is also attested, such as in ṯawra “revolution”, iṯnīn “two (m.)” ṯmāniya “eight”, al-ṯēniya “the second”, ṯlēta or tlēṯa “three (f.)”, akṯar “more”, miṯēl “example”, ṯnāš “twelve” and kāriṯiyya “catastrophic”. In some cases, the pronunciation of dental/interdental is present in the same informant: miṯēl “example”, ṯnāš “twelve”, but ymattel “represents/plays” and tlātamiya “three hundreds”, ṯmāniya and tmāniya “eight (f.)”. D’Anna (2020, p. 88) states that in the word tlāṯa, only the second interdental is preserved, but in my corpus, there is a certain oscillation in the realisation of /ṯ/ in this word.21

The general tendency seems to show that initial /ṯ/ is more likely to be preserved. In my opinion, the trend in the preservation of the interdental /ṯ/ in the Arabic of Mahdia may be due to the influence of Tunis Arabic, which has it. In fact, the Arabic of the capital applies a centripetal force to the other “peripheral” varieties of Tunisia because of its increasing prestige, due to the fostering of travel and education during the recent years (Gibson 2002, pp. 29–30). Media and social media, on which several programs are broadcast in dāriǧa, have a relevant role in this process too. In general, as Gibson showed, Arabic dialects do not follow the general trend affecting many other languages of the world, that is they do not tend towards a levelling to Modern Standard Arabic, but they are inclined to be closer to the standard dialect variety of the region, such as in the case of Tunis Arabic.22

/ ḏ /

The phenomenon is less evident in the voiced interdental consonant, which, however, sometimes oscillates with /d/: asātida “professors”, mādabiyya “I want/I would like to”, hādāka “that (m.)”, hāda (Yoda 2019, p. 59), which alternates with hāḏa “this”, hēdī “this m”., hēdiyya “this (f.)” (attested in Tlelsa; see also Yoda 2019, p. 65; Attia 1969, p. 123). The same phenomenon is extended to all the demonstratives containing a /ḏ/. In the same speaker, we find the two realisations of the consonant in the same utterance: kadē w-kadē “so and so”, haḏeya but also hādiya “this (f.)” hāḏa “this (m.)”, naḫḏāw “we take”, but mādabiya “I want/I would like to”. This feature is attested in a wide audience, different by age and sex and has also been attested by Yoda, for instance, ḫdē “he took” (Yoda 2019, p. 63).

In Sousse, the verb hḏā loses its final /ḏ/ in the imperfect conjugation (Talmoudi 1980, p. 93), and some examples are also available in Mahdia Arabic: yāḫu “he takes”. The phenomenon is also attested by Yoda (2019, p. 63) and in Bekalta too, where we find, for instance, nāḫu aḥna “we take it”. The weakening of /ḏ/ is an ancient phenomenon, attested in Sicilian Arabic where it sometimes even passed to /l/ and the consonant has not been pronounced in Maltese Arabic since the 14th century (La Rosa 2019, pp. 116–17; Avram 2012, p. 102).

In Monastir and in Bekalta, interdentals are generally maintained, but in the latter, a certain oscillation, depending on free variants and the speakers’ levels of education, is attested in the following cases: aktar min talātīna “more than thirty”, nāḫdu “we take”, yāḫduh “they take it”, nāḫduh “we take it”, but in the same speaker we find hāḏī “this (m.)” and hāḏāya “this (f.)”, hādiya “this (f.)”, hādāka “that (m.)”, which alternates with hāḏāka, and ḫdīna “we took”.

3.1.2. /q/

In Mahdia, Monastir, Msaken and Bekalta, /q/ is generally pronounced as voiceless, but it might be realised as /g/ in the following examples: gālū lī “they told me”, gutlak “I told you”, silʿa mangūla “transported goods”, bagra “cow”, mungēla “watch”, ynaggǝz “he jumps”, gāʿdīn “[the ones] staying”. The phenomenon was noticed also by Attia (1969, pp. 125–26) who offered few examples that have been already mentioned above. As regards Msaken, Bouhlel observed that the inhabitants of the town generally pronounce the uvular stop /q/ as voiceless because they consider themselves as “citizens”, such as in the following words: iqallaʿ “eradicate”, yuqʿud “he sits down”, and qalb “heart”. According to Bouhlel, the consonant is pronounced as /g/ in some words, such as mungēla “watch” or bilgdə “well”, and in his opinion this would be the “real” pronunciation of the consonant. Bouhlel also shows a certain variation in the pronunciation of some terms, such as zqūqu and zgūgu “pine nuts”, baqrī and bagrī “veal” (Bouhlel 2009, p. 127). Actually, none of the Tunisian dialects know the exclusive use of /q/ and/or /g/. If compared to the spread of /g/, the realisation /q/ is in fact limited and restricted only to the cities of Bizerte, Tunis, Sousse, Monastir, Mahdia, Sfax, and Kairouan. Mahdia is, therefore, “entouré de g” ((Mion 2015, p. 271); see the maps in (Skik 2000)).23 Moreover, Yoda (2008, p. 484) stated that the appearance of /g/ instead of /q/ in some words in Mahdia Arabic, as well as in other Tunisian dialects, is very common.24 In addition to this, already in 1984, Saada (1984, pp. 27–28) talking about the Arabic of Tozeur, stated that the existence of a /g/ sound is not a valid criterion for establishing the “non citadinité” of a dialect in Tunisia. She also added that throughout rural Tunisia, the phoneme /g/ was used for /q/, but that it was also present in some coastal zones and towns, in Judeo-Arabic and Muslim varieties, even if the Judeo-Arabic varieties of Tunisia are usually an exception and all have /q/, except Tataouine.25

Free Variants

/k/ as /q/

In some aged inhabitants of Mahdia, among which are some fishermen, natives of the quarter Burž al-rās, /k/ is articulated as /q/ in few words or expressions, such as kif kif “the same”, pronounced qif qif, and flūqa “felucca, rowing boat”, generally attested as flūka in Tunisian Arabic.26 The term flūka is pronounced with /k/ three times in the same video by two other speakers, that is a Mahdāwī poet, while reciting a poem dedicated to Mahdia, and the speaker. Both used a higher register: The poet because of the nature of the text that he was reciting and the speaker because of his role of presenter of the documentary. Therefore, their roles may have influenced their linguistic choices. Regarding previous studies on Mahdia Arabic, Attia (1969, p. 125) did not highlight this feature, but identified an emphatic variant of /k/ indicated with /ḳ/, such as in ḳānī “I am”.

3.1.3. /ṛ/

The phenomenon is attested in some aged fishermen natives of Mahdia, such as in the words buṛž, of the toponym Burž al-rās, bḥaṛ “sea”, ṛaḥma “mercy”, aktaṛ “more”, baṛṛa “out” and dāṛ “house”. The same informant may also pronounce some words in which /r/ is not emphasised, such as ǧārek “your neighbour”. The phenomenon has also been attested in some other mid-aged speakers, women and men, but not in all the speakers of the video analysed. The realisations of /r/ as /r/ and /ṛ/ may also oscillate in the same informant, a forty-year-old female nurse from Mahdia, such as in the word aṛbaʿ and arbaʿ “four”. The emphatic /ṛ/ is also attested in Mahdia Arabic by Yoda (2008, 2019). The presence of /ṛ/ does not seem to allow fronting or raising of /a/ and /ā/. In the speakers involved in the videos analysed who do not pronounce /ṛ/, /a/ is maintained as well and is not raised nor fronted, such as in arbʿa “four”, dinār “dinar” and barra “out”. However, further investigation is required to verify the presence and the distribution of the phenomenon since it has been attested in few speakers, at the moment.27

Regarding Msaken Arabic, Bouhlel (2009, p. 127) stated that, differently from some other regions of Tunisia, the rā՚ is generally not “amplifié”, and is even pronounced re, such as in the words ureq “leaves”, krēheb “cars”. Bouhlel added that some exceptions are available, such as mrā“woman” and ḥrābiš “pills”, but he does not explain in which conditions the phenomenon occurs. In my corpus, some cases of emphasisation of /ṛ/ in Msaken Arabic have been attested, as well as the raising of /ā/ after /r/, such as in” krēheb “cars”, as indicated by Bouhlel.

3.1.4. De-Emphasisation

Some cases of loss of emphasis have been attested in Mahdia and Bekalta in speakers of different sex and age, such as in the words atfāl “children”, tbīb “doctor”, musāba “infected” durūfa “conditions”. In the latter, we notice the possible shift /ḏ̣/>/ ḍ />/d/ because of the loss of the interdental articulation. Some other examples are tufultī “my childhood”, bi-sifa “in quality of”, abyad “white”, tul tul “directly, straight”, muwātin “citizen”. The phenomenon involves all the emphatics /ḏ̣/, /ḍ/, /ṭ/ and /ṣ/ and does not seem to affect the vowel quality, which is maintained. In the corpus of Bekalta, we also find some examples of de-emphasisation and sonorisation such as mazdar “source” and mazraḥī “theatral”, in which there is a shift /ṣ/ > /z/ with a possible intermediate step of /ṣ/ > /s/. The phenomenon of de-emphasisation is also attested in Kairouan, where we find, among other words, matālib “requests”.28 In a woman of Monastir, we find a de-emphasisation of /ṭ/ realised as lightly affricated /ts/ in the exclamation yā latīf! “oh my God!”. A light affrication of /t/, already highlighted by Cantineau (1960, p. 37) as a feature attested in Algerian and Moroccan Arabic, generally also characterises Tunisian Arabic, but here involves emphatic consonants too.29 Saada (1984, p. 24) identified a /ts/ sound in the speakers of some tribes of Tozeur and Dallaji (2017, pp. 153–57) reported that in Nabeul there is a tendency to affricate /t/ that people define through the verb taštaš and the terms taštīš and taštaša. Furthermore, Maamouri (1967) described the variety of Nabeul as being characterised by a strong affrication of /t/.30 As regards the loss of emphasis, it is also typical of Tunisian Judaeo-Arabic (Taïeb and Sayah 2003). Cohen (1975, vol. 2, p. 14 and n. 7) states that some phenomena in Tunis Judeo-Arabic lead us to think that there must have been a period in which the articulation of emphatic consonants was stronger and “sans doute plus forte qu’à Tunis musulman, ce qui expliquerait le fait que les musulmans qui veulent imiter le parler des juifs, exagèrent l’emphase en même temps que les modulations expressives de la phrase”.31 Cohen’s statement also shows that in Tunisia the weakening of the articulatory strength in pronouncing the emphatics has long been attested. However, the prononciation of emphatic consonants with more or less articulatory strength from different confessional groups, i.e., Jews/Muslims, is a phenomenon of old, present in different Arab contexts because of community or identity reasons.32 Saada (1984, pp. 83–84) noticed the phenomenon in the Arabic of Tozeur and indicated a series of conditions for its occurrence. Further studies will be necessary to be able to indicate which are the causes in Mahdia Arabic and if any regularity can be found.

3.1.5. Dropping of Final /n/

In Monastir and Mahdia, the weakening or dropping of the final /n/ of numerals in pausal position, that is not followed by the ism al-maʿdūd, should be noted: ʿišrīn milyū “twenty millions”, sabʿat w-ʿišrī “twenty-seven”, ṯmānī “eighty”, ḫamsī “fifty”. The dropping of the final /n/ is an old feature, attested in Andalusi Arabic only in dual nouns (see (Ferrando 1995, pp. 50–51; Corriente et al. 2015, pp. 110, 125–26). It also should be noticed that in Yoda’s transcriptions, this feature does not emerge (see, for instance (Yoda 2019, p. 63)). In Saada’s study on the Arabic of Tozeur, some consonantal dropping involving final /n/ have been attested, such as the cases of ṯnē “two”, mnē “from where”, lē “until when”, gaʿdī “standing”, takū “you will be”. The scholar defines this phenomenon, which in Tozeur is wider spread than in Mahdia and involves some other consonants, tarḫīm, i.e., “softening” or the process of shortening a word (for a more complete definition, see (Carter 2007, p. 17), and states that only one informant uses it and therefore judges this feature as in decline/regression (Saada 1984, pp. 39–42).

3.1.6. Vowels

According to Attia (1969, pp. 126–27), Mahdia Arabic has five short vowels /a/, /e/, /i/, /o/, /u/ and five long vowels /ā/, /ē/, /ī/, /ō/ and /ū/.33 Yoda (2008, 2019) stated that Mahdia Arabic has three short vowels, /a/, /i/, /u/, and five long vowels, and assumed that Mahdia dialect, as an urban variety, acquired /ē/ and /ō/ as phonemes because of the influence of the surrounding village dialects. He also added that “Because of the existence of /ē/ and /ō/, Mahdāwī dialect dialect is unique among the Tunisian sedentary dialects” (Yoda 2008, p. 489).34 Therefore, Mahdia Arabic vowel system would be made of five long vowels and five short ones for Attia, but only of three short vowels for Yoda.35 Starting from this vowel scheme, Mahdia Arabic long vowels can undergo some qualitative changes; i.e., some cases of opening of the close vowels /i/ and /u/ are present in the varieties of the region, such as in Bekalta, were we find both zīt “oil” and the variant zēt, or hōnī “here” in Monastir (Attia 1969, p. 129) and a certain variation is also attested in Msaken, such as in the cases of rōḥ “soul” and fūl “broad beans”. Bouhlel (2009, p. 128) also provides some minimal pairs: qūm “stand up”/ qōm “people” pejorative, kūn “be”/kōn “world”.

In this section, I shall focus exclusively on the treatment of /ī/ when it is followed by the third person singular suffix pronouns. What emerges is that, in general, /ī/ + hā and /ū/ + hā are maintained, such as in mā fī-hāš “there is not” (as well as in Takrouna Arabic see (Marçais and Guiga 1925, p. XII, n. 2) and in Sousse, (Talmoudi 1980, p. 152)). However, some cases of fē-hā “in it” have been attested (Yoda 2008, p. 489; 2019, pp. 61, 65, 66). In the videos analysed in my corpus, I found only rare cases of fē-h “in it (m.)” and of fē-ha “in it (f.)” in a speaker from Knaies, a small village 11 km near Msaken, and in a small group of people allegedly from Mahdia. The majority of the people speaking in the videos analysed, coming from Mahdia or Monastir, uttered fī-h or fī-ha. The reduced presence of the phenomenon may be due, once again, to the linguistic levelling towards the Arabic of Tunis or of the other coastal towns of Sahel such as Sousse (see (Mion 2015, p. 272)) or to the conditioning of the medium used, which may push the speaker to use a higher register.36

Therefore, the vowel system of Mahdia seems similar to that of Monastir, which, according to Gibson (1998, p. 276), has villageois features since it has five long vowels as in Figure 3.

According to Marçais, Takrouna Arabic has“une particularité de détail propre aux dialectes sâḥli: l’ouverture en ē et en ō de ī et de ū” that we find in Mahdia too (Marçais and Guiga 1925, pp. XXI–XXII).

3.1.7. Diphthongs

According to Yoda (2008) and Mion (2015, p. 271), the vowel system of Sahel Arabic is characterised by five long vowels, ā, ē, ī, ō, ū, obtained from the monophthongisation of the diphthongs /ay/ and /aw/ in ē and ō. According to the traditional categories of classification, which contrast some allegedly Hilali and pre-Hilali features, in the Hilali system, /ay/ and /aw/ would be reduced to ē and ō, while the passage of /ay/ > /ī/ and /aw/ > /ū/ would be typical of the pre-Hilali dialects. Actually, the situation is more complex than this, because, as Mion (2015, p. 271) showed, the reduction of the diphthongs to ē and ō “is shared by both /q/ and /g/ varieties”. As already dealt above, the pronunciation of /q/ has been the main criterion for establishing whether a dialect had Bedouin or sedentary features, but Mion and some other scholars have raised doubts about the validity of this principle.37 Based on what emerges from my corpus, the realisation of etymological diphthongs is usually sedentary in Mahdia, in Monastir and in Bekalta, such as in the following cases: al-yūm “today” (Tlelsa), ḫīr “better”, ḍīf “guest”. Even if Yoda (2019, p. 59, n. 9) states that the diphthongs in Mahdia are usually reduced to /ē/ and /ō/, the cases identified in my corpus are similar to those of Sousse, where the diphthongs /ay/ and /aw/ are reduced to /ī/ and /ū/ (Talmoudi 1980, p. 55). Some other existing studies on Sahel varieties all go in the same direction; for instance, to recount it with Marçais’ words regarding the Arabic of Takrouna, “Comme les dialectes congénères du Sâḥel, le takroûni y suit sur quelques points des voies propres qui, l’éloignant des parlers citadins de la Régence, le rapprochent de certains dialectes bédouins du Maghréb oriental […]: les diphtongues anciennes ai et au accentuées et non en finale absolue s’y réduisent généralement à ē et ō” (Marçais and Guiga 1925, pp. XXI–XXII).38 As D’Anna (2020, pp. 88–89) showed, in Chebba, a small village 35 km south of Mahdia, etymological diphthongs are reduced to /ē/ and /ō/ and are occasionally partially diphthongised. According to Bouhlel (2009, p. 128), in Msaken Arabic, the diphthongs /aw/ can be reduced to /o/ in many terms, i.e., lōn “color”, mōz “bananas”, dōra “tour”. The situation seems to be quite complex since, once more, it is clear that Mahdia Arabic is an urban dialect surrounded by mixed or contact varieties, which can influence it in different ways and at different levels.39

3.1.8. Imāla

Several cases of final imāla, etymological or not, have been found in all positions in Mahdia and the nearby towns: hnē “here”, ʿalē “on/upon”, mšē “he went”, ḫdē “he took” ((Yoda 2019, pp. 59, 63); see also (Attia 1969, p. 130)), At the moment, the pronoun ēni “I” is the only case of imāla šadīda found in Mahdia and Monastir. (Bouhlel 2009, p. 127) added some other examples attested in Msaken Arabic, such as mī “water” and smī “sky”.

Imāla is a phenomenon shared by many Tunisian vernaculars. Gibson (1998, p. 279), for instance, analysed the pronunciation of the verbal forms mšā “he went” and mšāt “she went” and stated that in Sousse the average vowel of the feminine form is /ɛ/ and those of the masculine form are /ɛ/ and /e/. He added that among his Sousse informants, nobody uttered a long /e/, while in Monastir, there was only one speaker with /ē/ as the tense vowel. Marçais affirmed that in Takrouna Arabic “ā accentué en finale absolue y passe à peu près constamment a ē” (Marçais and Guiga 1925, pp. XXI–XXII). The feature of imāla is also shared by Tunisian Judeo-Arabic, the /ā/ is affected by imāla in all positions (Cohen 1975, vol. 2, p. 56).40

3.1.9. Raising of Final—a

Mahdia Arabic is also characterised by the raising of the final feminine singular ending –a such as in the cases of maḥdūde “limited”, ḫamse “five”, insaniyye “human”, bnayye “child (f.)”. There is a role played by syntactic elements since we also find ḫamsa ṣġār “five children”, with the numeral in iḍāfa. The presence of emphatic and pharyngal consonants usually blocks this raising.

3.2. Nominal Morpho-Syntax

3.2.1. Personal Pronouns

The existing studies on the personal pronouns of Tunisian Arabic show a complex situation, which can be summarised in Figure 4:

Mion, in his paper Réfléxions sur la catégorie des «parlers villageois» en arabe tunisien, offered a brief sketch of the first singular and plural personal pronouns “I” and “we” and of the second singular person “you” and focused on gender opposition in the pronominal system. If the village and Bedouin dialects share the presence of gender opposition in the second person singular, according to Marçais’ classification, the urban varieties do not (Marçais 1950, p. 208). With respect to this element, in Mahdia Arabic, there is no gender distinction in pronouns and verbs, as well as in Monastir, Msaken and Bekalta. So, the situation could be represented as follows in

Table 1. Personal pronouns in Mahdia, Monastir, Msaken and Bekalta.

Pronoun	Mahdia	Monastir	Msaken	Bekalta
1 p. s.	ǟni	ǟni	ǟni	ǟni/āni
1 p. p.	aḥna/naḥna	aḥna	aḥna/naḥna	naḥna
2 p. s.	ǝnti/inti41	ǝnti	ǝnti	ǝnti/inti

.

As for the first-person plural “we”, aḥna is the main pronoun used and naḥna seldom alternates. Therefore, further studies are needed to check whether this is due to the origin of the speakers interviewed in the videos analysed. In Yoda (2019, p. 65), ḥna is also attested when preceded by a vowel such as in the expression tawwa-ḥna “now we”.

The first person singular “I” generally shows a final imāla šadīdā.42 Moreover, Takrouna Arabic shows the pronoun ǟni “I”, which according to Marçais is original to the variety, and also presents the variants nǟya and nǟy, which have been imported later (Marçais and Guiga 1925, pp. XXII–XXIII).43 As regards Msaken, Bouhlel (2009, pp. 130–31) indicates āni with the meaning of “I” and “me” and the second person singular pronoun inti “you”.

3.2.2. Relative Pronouns

In my corpus, in all the towns and villages considered in this study, the only relative pronoun used is ellī and the variants illī ((Yoda 2019, p. 61) has illi) and lī when the preceding word ends in a vowel. The relative pronoun ella, defined as “villageois” by Marçais, has not been attested at the moment ((Marçais 1950, p. 211); for Chebba, see (D’Anna 2020, p. 90)). Other varieties of Tunisian Sahel, such as the dialect of Sousse, also have the variants ellī and llī, used when the relative is preceded by a word ending in a vowel (Talmoudi 1980, pp. 146–47).44

3.2.3. Numerals

In numbers 3 to 10, the ism al-maʿdūd is plural: ʿašra yyām (Yoda 2019, p. 66), xams iyyām “five days” and tlāṯa yyām “three days” (Yoda 2019, pp. 60, 63), ṯlāṯa ʿaskar “three soldiers” and arbʿa ʿaskar “four soldiers” ((Yoda 2019, p. 59); for Tunis Arabic, see (Biţuna 2011)).45 As already noticed by Biţuna for Tunis Arabic, the noun dīnār “dinars” is always used as a singular in Mahdia Arabic too (see (Biţuna 2011) and (Yoda 2019, p. 63)).46

In Mahdia Arabic, numerals from 11 to 19 have the same form that they have in Tunisian Arabic in general, that is ḥdāš “eleven”, tnāš “twelve”, tluṭṭāš “thirteen”, arbaʿaṭāš “fourteen”, ḫumṣṭāš “fifteen”, səṭṭāš “sixteen”, sabʿaṭāš “seventeen”, tmūnṭāš “eighteen” and tsaʿaṭāš “nineteen” (some numerals are present in (Yoda 2019): passim). One of the main evident characteristics is the disappearance of the pharyngeal /ʿ/ or its assimilation to the /t/ (Biţuna 2011, p. 32). These numbers from 11 to 19 are usually followed by a singular, such as in tnāš ʿām “eleven years” and ḫamsṭāšən ʿām “fifteen years”. It is worth noting the form tluṭṭāšən el-snā “eighteen years”, followed by a determined singular noun. Some other examples, taken from Yoda’s texts, are xumṣṭāš-in yōm and xmuṣṭāš-in yōm “fifteen days” (Yoda 2019, pp. 65, 66), attested in the same speaker. It is well known that in Maghribi dialects, numerals 11 to 19 show an -n form when in direct annexation with the maʿdūd ((Marçais 1977, p. 178); for Tunis Arabic, see (Biţuna 2011, p. 32)). In fact, in Sousse, numerals from 11 to 19 also have an –n final shape when in iḍāfa (see (Talmoudi 1980, p. 169)). Concord from 20, in Mahdia Arabic, reflects the rules of fuṣḥā too, such as in the cases of ʿišrīn yōm “twenty days” ((Yoda 2019, p. 65); for Tunis Arabic, (Biţuna 2011, p. 34)) and mya ʿaskar “one hundred soldiers” ((Yoda 2019, p. 59); for Maghribi Arabic, see (Marçais 1977, pp. 173–80), for Tunis Judeo-Arabic, (Cohen 1975, p. 232); for Takrouna Arabic see also Marçais; (Guiga 1925): passim).47

The tendency towards weakening or to let the final /n/ drop in tens, in the towns considered in this analysis, has already been dealt with above; some other cases involving dozens are ḫamsī “fifty” and arbaʿī “fourty”.

3.3. Verbal Morpho-Syntax

3.3.1. An Urban Conjugation

Verbs in the Arabic variety of Mahdia and the surrounding towns, that is Monastir, Msaken and Bekalta, are generally conjugated according to the stem yimši/yimšīw and yaqra/yaqrāw, traditionally defined as pre-Hilali (Marçais and Guiga 1925, p. 49 and passim; Marçais 1950, p. 209)48 as well as in other urban varieties of Sahel, such as Sousse and Sfax (Talmoudi 1980, pp. 86–87, Sellami 2019; Herin and Zammit 2017, p. 143). Some examples are yemšīw “they go”, naqrāw “we read”, nalqāw “we find”, netmannēw “we hope”, nistennēw “we wait for” (see also Yoda 2019, pp. 59, 66)49.

Even if data for Mahdia Arabic seem to be clear, as Mion (2018, pp. 117–18) pointed out, for Sahel, the issue cannot be reduced to the simple opposition pre-Hilali/Hilali conjugation, since a large part of Tunisian varieties have mixed features with plural perfect conjugation in -āw and imperfect in –ū, i.e., mšāw/yamšu (see also (Mion 2015, pp. 272–73)).50

3.3.2. Feminine Third Person Singular of Weak-Final Verbs

In Mahdia, Monastir, Bekalta and Msaken, the feminine third person singular of the perfect tense has a long vowel and –ā-t ending, also showing imāla, such as in the following cases: mšēt “she went”; taʿaddēt “she passed”. The long vowel, preserved as in the so-called pre-Hilali dialects, has been also attested by Yoda (2019, p. 65) in xdāt “she took” and ʿṭāt “she gave”. In addition, the surrounding urban varieties have similar features (see Talmoudi 1980, pp. 86–88).51

3.3.3. Perfect Tense Pattern

Similar to the other urban dialects, Mahdia Arabic does not show gender opposition at the feminine second person singular ənti of the perfect tense, just like the other urban varieties of Sahel such as Msaken (Bouhlel 2009, p. 131), Sousse (Talmoudi 1980, pp. 77–78) and Sfax (Lajmi 2009, p. 138; see also (Mion 2015, p. 273)).

Gender opposition at the second person singular “you”, in fact, is maintained only in conservative52 Bedouin or village dialects in which the feminine person has the desinence –ti (Marçais 1977, p. 36).

3.3.4. Use of the Verb rā

In the Arabic of Mahdia, the verb to see is expressed through the verb šāf “to see, to look at, to watch”, common to a wide variety of Arabic dialects,53 and rā “to see”.

Additionally, in Sfaxi Arabic, the two verbs alternate with a certain predominance of rā (Lajmi 2009, p. 140; Zammit 2014, p. 34; Sellami 2019) and, in Sousse, the verb rā is used, but with a peculiarity, that is the first person singular of the perfect tense rūt (Talmoudi 1980: 76, 90) and not rīt, as in Kairouan and Mahdia. Among the examples, we find rītuš? “have not you seen it?” (Yoda 2019, p. 66). In Mahdia Arabic, šāf is very used, as in the cases of šūf, šāf (Yoda 2019, pp. 60, 63, 66). In Tunis Judeo-Arabic, the verb rā is used and it has maintained the dipthongation /ay/ (Cohen 1975, p. 106). From an historical perspective, the verb rā replaced šāf in Sicilian Arabic, in Sicilian Judeo-Arabic and in Maltese. Moreover, it was very productive in Andalusi Arabic too (La Rosa 2019, p. 259; Burgaretta 2016, p. 89; Corriente et al. 2017, pp. 508–9, for šāf, p. 742).

3.3.5. Passive

The passive form of verbs is usually expressed through an initial t- pattern, used in Tunisia and in other Maghribi varieties (see, i.e., (Cohen 1975, pp. 123–25)), “né, sans doute, sous l’influence analogique des réfléchis-passifs à t- initial des thèmes V et VI, procédant respectivement des thèmes II et V, ce thème procède de verbes du thème fondamental généralement en usage. Il en constitue le réfléchi-passif” (Marçais 1977, p. 66).54

The following examples have been attested in Mahdia, Bekalta, Monastir and Msaken: tǝbnat “it was built”; yitnaḥḥa “it is removed”, yitfahem “it is understood”, tǝtḥet fuq al-rās “it is placed on the head”, tuṭfī “it switches off”, mā tuṭfiš “it does not switch off” (for the phenomenon in Sfaxi Arabic, see (Sellami 2019)).55

4. Discussion and Conclusions

Marçais (1950, p. 214) stated that the Tunisian Sahel is characterised by sedentary varieties, which break the continuum of Sulaymi dialects and divided Bedouin Tunisian dialects into two main groups: Hilali and Sulaymi. According to this kind of classification, the features of Mahdia Arabic would have some characteristics attributable to the Banū Hilāl, such as the imāla of internal vowels, the masculine singular third person pronoun suffix –u, opposed to the suffix –a(h) of the Sulaymi group and the passive in t-. Some Sulaymi features would be the weaker articulation of emphasis and the final imāla (see (Marçais 1950, p. 217; Ritt-Benmimoun 2014, p. 354; Taine-Cheikh 2017, pp. 20–21). However, Ritt-Benmimoun’s studies on the Bedouin Arabic of South Tunisia have showed that the categorisation Hilali/Sulaymi is not always definite and obvious and led us to wonder whether “there is a real zone of transition between areas where the S and the H dialects are spoken or if these areas are separated by a more or less well-defined boundary, perhaps corresponding exactly to the settlement area of the different tribes” (Ritt-Benmimoun 2014, p. 358).

In 2017, Taine-Cheikh pointed out that Marçais’ criteria of dialect classification had to be revisited and stated:

Reste, me semble-t-il, une question, celle de la valeur de la distinction entre parlers hilaliens vs sulaymites vs maʿqiliens. S’il existe bien de groupes de parlers plus ou moins différenciés et s’il est nécessaire de leur attribuer un nom, je ne suis pas sûre que ces trois désignations d’origine socio-historique soient d’une réelle précision et donc, d’une véritable secours (Taine-Cheikh 2017, p. 38).56

Guerrero (2018) revisited the idea of the villageois category in which the rural Tunisian dialects are generally included, and based on 20 phonological, morpho-syntactic and lexical features, demonstrated that they show important differences from the Algerian and Moroccan varieties, which, instead, constitute a group with consistent features. Benkato (2019, pp. 11–12) pointed out that the classifications of Bedouin dialects, and particularly Sulaymi dialects, derived from W. Marçais’ statements, arose in turn from de Slane and G. Marçais’ assumptions, were “taken as fact”, even if the scholar did not offer any genealogy of these tribes. Benkato added that “these categories were hardly based on sound linguistic argumentation and instead more on the personal experience and reputation for their creators” (Benkato 2019, p. 14).

For all these reasons, Marçais’ traditional criteria of classification therefore cannot be applied tout court to the analysis of Mahdia Arabic. That is, it is not possible nor methodologically appropriate to attribute a precise and certain origin to all of its features. In spite of this, the first data presented in this contribution confirm that Mahdia Arabic is a sedentary variety showing some “contact” features.

According to the traditional classification, typical sedentary traits are the voiceless articulation of /q/, the relative pronoun ellī and the conjugation pattern mšēw/yemšīw. As regards the latter, it is useful to stress what Mion (2018, p. 118) claimed:

Alors que -īw du Maroc à la Tunisie doit être vue comme sédentaire et citadin, un système transversal avec un parfait –/āw/ et imparfait –/ū/ est en réalité bien plus fréquent que ce que la simple opposition méthodologique pré-hilalien/hilalien laisserait entendre. Si ce système doit être conçu comme transitoire au sein d’un continuum don’t les deux pȏles seraient justement les typologies pré-hilalien/hilalien, alors une bonne partie de la Tunisie (à l’exception de ses métropoles et des variétés Marāzīg) devrait être considérée paradoxalement comme une zone de transition.57

Some “contact” features attested in Mahdia Arabic are the final etymological imāla, and the opening of /ī/ towards /ē/ in the particle fī followed by the personal suffix pronoun –hā.

However, the situation does not always seem to be clear and definite since sedentary dialects also share many elements with Bedouin Hilali dialects, such as the following used in Mahdia region: The pronoun for the third person masculine singular –u, and the realisation of final /ī/ and /ū/ as /e/ and /o/, when followed by the pronoun –hā, which “is not considered as an isogloss marking either Bedouin or sedentary varieties. However, it is attested in dialects not generally considered as ‘rural’” (D’Anna 2020, p. 93). According to Ritt-Benmimoun:

The affinity of the H [Hilali] dialects with Tunisia’s sedentary dialects can be found in Philippe Maarçais’ statement that in the regions with a prevalence of H dialects sedentary dialects had originally been spoken which were subsequently overlaid by Bedouin dialects (Ritt-Benmimoun 2014, p. 355).

As regards the reduction of diphthongs to /ē/ and /ō/, which is generally considered to be a Bedouin feature, is attested by Yoda (2019) in Mahdia Arabic, but is not much attested in my corpus where, instead, a strong presence of /ī/ and /ū/ is noticed. Therefore, further investigation will shed new light on the real attestation and distribution of this feature.

Besides, some of the main sedentary features of Mahdia Arabic are also shared by the so-called villageois varieties, such as the voiceless realisation of /q/,58 the final etymological imāla šadīda, the lack of gender opposition in verbs and pronouns and the long vowel –ā in the desinence –ā-t of the feminine third person singular of final-weak verbs (Marçais 1950, pp. 207–12).

From a diachronic point of view, some elements pointed out in previous studies offer a sketch of a slightly different variety of Mahdia Arabic. For instance, in 1969, Attia observes that /d/ “remplace l’interdentale /ḏ/ qui a disparu du parler de Mahdia” Attia (1969, p. 123) and in 2008, Yoda identifies a series of phenomena, which in my corpus are not so obvious, such as the presence of fē-ha or the reduction of diphthongs to /ē/ and /ō/. These and other traits suggest a possible explication, that is that the Arabic of Mahdia is undergoing a change, a sort of linguistic levelling or standardisation towards the Arabic of the capital Tunis and, probably, also through the influence of the main city of Sahel, together with Sfax, Sousse, which is only 60 km away. As already shown above, Gibson’s research goes in this direction59 as well as Mion’s observations about a series of phenomena that, in a certain way, contribute to create this situation of standardisation not only in Sahel Arabic, but more generally in Tunisian Arabic. Among these factors, I find particularly interesting the role of the media, and above all, television, that Mion considers a facilitator for the diffusion of Modern Standard Arabic to all the social strata. Television is therefore responsible for the reintroduction of some phonemes in people’s speech, such as the hamza and the voiceless articulation of /q/,60 and also for the new diffusion of the –āw/īw perfect tense endings. Moreover, it helped the spread of the prestigious dialectal Arabic of the capitals, such as the Tunisian of Tunis (Mion 2018, p. 120).61 The fact that the aged fishmen natives of Burž al-rās, interviewed in the already-mentioned documentary on this quarter, all pronounce the interdental /ḏ/ as /d/ confirm Attia’s observations and lead us to consider this feature as a confirmation of the linguistic levelling in progress, above all among the youngest and more educated inhabitants of Mahdia. According to Sayahi (2019, p. 237), the fact that in recent years the institutions often choose to utter their official speeches in Tunisian Arabic fostered the use of the dialect in the public space for public communications and not only for private or semi-private occasions (on this subject, see also (La Rosa 2018)).

Moreover, I believe that the diffusion of the use the smartphone has been crucial, above all after the 2011 Arab Spring(s). In fact, thanks to it, the internet has become more accessible to a large part of the population. Easy and cheap access to the internet has allowed the acquisition and the circulation of information from far-off places within the country and abroad. Thanks to the use of social media, such as Facebook and WhatsApp, or the use of YouTube, the spread of several forms of oral and written artistic forms and cultures has become more and more easy.62 What is most important here is that the linguistic contact between people from different regions of Tunisia, speaking different varieties of Arabic, has become easier too. I think that this has helped the linguistic levelling process in progress by bringing people together, in spite of the geographical distance.

Furthermore, we should also take into consideration that the role of the migration movements inside Tunisia and abroad (see (D’Anna 2017) regarding the Sicilian community of Mazara, mainly consisting of Tunisians from Mahdia and Chebba) for study and work reasons may have had particular relevance in the levelling of some linguistic features contained in the varieties of Sahel. In fact, “migration from the cuntryside to cities has constituted the most significant demographic change in the last two centuries. Linguistic contact in cities is more intense and involves faster changes than in rural areas. […] In recent decades, a significant increase in the frequency and variety of types of inter-dialectal and intra-dialectal contact occurred causing faster and greater degrees of levelling” (Vicente 2019, p. 106). Linguistic levelling is also evident in the town, which is probably most affected by migration movements among those mentioned in this study, i.e., Msaken. As Bouhlel (2009, p. 133) already pointed out, in fact, Msaken Arabic underwent a deep change during the last 50 or 70 years, affecting mainly, but not exclusively, lexicon.

In addition to this, as shown above, the medium used for the interviews in my corpus has a crucial role since, to a certain extent and in many cases, it conditions the speaker who knows that he/she is being filmed or recorded and, therefore, tends to use linguistic structures and/or a lexicon belonging to a higher register, that is the standard variety, be it Modern Standard Arabic or Standard Dialect.63