The Southern Moroccan Dialects and the Hil¯ali Category

: The aim of this paper is to review the classiﬁcation of the southern Moroccan dialects, advancing on the general description of these varieties. Recent descriptive studies provided us with new sources on the linguistic reality of southern Morocco, shedding light on the status of dialects commonly classiﬁed as Bedouin or ‘Hil¯ali’ within the Maghrebi context. To do so, the paper highlights conservative and innovative features which characterize the dialects of the area, focusing mainly—but not exclusively—on the updated data for two distant localities in southern Morocco: Essaouira and its rural outskirts—the Chiadma territory (Aquermoud and S¯ıdi ¯Ish.¯aq)—and Taﬁlalt, in south-eastern Morocco. The southern dialects have been situated in an intermediary zone between pre-Hil¯ali and Hil¯ali categories for a long time. Discussing their situation may contribute to understanding what distinguishes them as a dialectal group and also the validity of the ‘Hil¯ali’ category in the Moroccan context.


Introduction
Traditionally, dialectologists have divided the Maghrebi dialects into two categoriespre-Hilāli and Hilāli-within a diachronic perspective which associates linguistic features to the waves of Arabization in North Africa, from the works of W. Marçais, such as the seminal text Comment l'Afrique du Nord a été arabisée (Marçais [1938(Marçais [ ] 1961, to more recent scholarship (cf. Aguadé 2018). Based on these two types, Colin ([1937Colin ([ ] 1945Colin ([ , 1986 proposed a sub-classification to Western Maghrebi dialects, or Moroccan dialects precisely, grouping them into: parlers citadins, parlers montagnards, parlers bédouins and parlers juifs. Regarding the Hilāli-Bedouin type in Morocco, authors have attempted to tackle the problem of grouping different linguistic varieties under this category. Colin (1986Colin ( , p. 1196 proposed that the Moroccan Bedouin dialects could be divided according to their levels of conservatism. That is the case of some dialects of the Sahara area-but not exclusively (e.g., Casablanca, Kampffmeyer 1912)-which retain features such as the realization [g] of *qāf and the maintenance of interdentals (e.g., /d / and /t/). The same aspect was observed by Lévy (1998, p. 19) who points out that Hilāli and Ma Q qili dialects found in the Atlantic plains are quite different from the Ma Q qili type in the Sahara (e.g., H . assāniyya). In agreement with this view, Heath (2002, p. 8) drew a distinction between Hilāli central type dialects and the Saharan ones, which-according to him-are restricted to southern oases and parts of the Atlantic plains in Morocco.
Regarding the Bedouin category in Morocco, Taine-Cheikh (2017) points out: "la situation reste complexe à décrire pour les parlers qui ne sont ni pré-hilaliens ni du type 'saharien '" (p. 25). That is the case of the southern Moroccan dialects, for which the application of the Hilāli category remains doubtful, despite of the confirmation of the [g] realization and the loss of interdentals, both commonly associated to it. In this manner, we pose the question of whether the findings on the southern dialects, and the revision of their classification, might contribute to shedding the light on the Hilāli category within the global linguistic reality of Morocco?
More recently, the endeavor of dialectologists for classifying the so-called Hilāli-Bedouin dialects has come out again after 'Hilāli' and 'Ma Q qili' terms were called into question. Benkato (2019) criticized that dialectologists erroneously linked Medieval historical facts-originally incorporated from Ibn Khaldun by French orientalists-with the modern linguistic reality of the Maghreb. He argued that there is a lack of evidence on the direct connection between Medieval tribes, taken as a "reliable unit of sociolinguistic analysis" (sic)-such as the Ma Q qil-and the Arabic dialects spoken nowadays in the region (p. 21). In this way, understanding the condition of southern Moroccan dialects might contribute to understand the validity of categories, such as 'Hilāli'. Nevertheless, it is important to say that the link between historical factors and the current linguistic reality should never be totally discarded.
To explain the distribution of the "southern" linguistic features over this part of Morocco, I argue that they are associated not only to the process of Arabization of this area, but also to modern historical factors, including the trans-Saharan trade route-connecting the Sahel to the Atlantic-and its effects on the populational movements on this area until the nineteenth century. In this way, I also try to explain the reason why distant localities in the south share common linguistic features and how their nature impacts on the validity of Hilāli-Bedouin category for classifying southern Moroccan dialects. This may be a complementary explanation to defining the origin of common features found in southern dialects.
In this context, the aim of this paper is to review the classification of the southern Moroccan Arabic dialects, highlighting some of the features which might single out these varieties. To do so, the study relied on the recent collection of dialectological data for the Atlantic strip, in Essaouira region (Francisco 2019a(Francisco , 2019b(Francisco , 2022, and south-eastern Morocco, represented by Tafilalt (Heath 2002;Behnstedt 2004)-without ignoring previous descriptive studies on other varieties of the region.

Descriptive Studies on the Southern Varieties
According to the traditional classification of Moroccan dialects, the southern dialects belong to the Hilāli or Bedouin type, given that they spread from the Atlantic plains-in the area of Mogador-to the eastern part of the country, the Mouloya basin and the Moroccan Sahara (Colin [1937(Colin [ ] 1945. Part of these dialects is commonly distinguished from the "truly" Bedouin dialects-Saharan or Ma Q qili type-, given that their classification was thought based on maintenance and loss of conservative features (Colin 1986(Colin , pp. 1195. Most of the dialects from the Atlantic strip to Tafilalt lost traditional conservative features (e.g., the interdentals), due to different degrees of Arabization of the Berber tribes (Heath 2002, pp. 8-9). Taine-Cheikh (2017, pp. 25-26) proposes the category 'parlers "hilaliens" du Sud marocain' to set together the dialects spoken from the Atlantic coast as far as the Algerian border. Her description relied on studies for the dialects of Skoura, Sous and Essaouira, which exhibit the realization of *qāf as [g] and the loss of interdentals. In fact, the realization of *qāf in southern Moroccan dialects remains a complex issue to the classification of these varieties under the label 'Hilāli', given that both g (Bedouin) and q (sedentary) alternate-as phonemes and allophones-, varying lexically. Moreover, the voiced g continues to be very usual in these varieties (Heath 2002, p. 9).
Given the new data for the dialects of the region, this paper gives special attention to the dialects encountered in two geographical extremities of the south not considered in the study mentioned above. For Tafilalt 1 : Rissani and Erfoud (Heath 2002);Īgli, Erfoud, Ma Q did, z-Zrīgāt and z-Zāwya ž-ždīda (Behnstedt 2004) and Qr . ab S@bbāh . (Behnstedt n.d. 2 ); and Judeo-Arabic of Ksar es-Souk and Rich (Heath and Bar-Asher 1982). For Essaouira (Mogador), I considered the Muslim and Jewish 3 dialects of the city, and also the variety of Chiadma territory (Aquermoud and SīdiĪsh .ā q) 4 in the rural surroundings of the city (Francisco 2019a(Francisco , 2019b(Francisco , 2022. Essaouira data, specifically, may prove to be a valuable source to understanding the linguistic reality of southern Morocco, due to the nature of the settlement in the city, which attracted speakers from different parts of the south since its foundation in the second half of the eighteenth century, as we may see in the next session.
Apparently, the dialects of both southern regions share most of the common features indicated by Taine-Cheikh previously, with a few exceptions, as we may see in Section 3, which consequently bring implications to the classification of these varieties. That may be explained due to historical facts related to the Arab settlement in these localities and also the lasting linguistic contact between the southern dialects.

Historical and Linguistic Connections in Southern Morocco
The History of population settling in southern Morocco may explain the linguistic proximity between south-eastern Morocco-the Tafilalt area-and the Essaouira region on the Atlantic strip, which includes the Chiadma territory. Concerning the Arabization process of both areas, well known historical sources indicate that these territories were occupied by Ma Q qil tribe members at some point, after the beginning of the second wave of Arabization in the Maghreb with the Banū Hilāl invasions in the XI century. Modern sources continued to narrate the movements of these groups in southern Morocco, region which became gradually more connected by centuries-long trade routes.
It is well known that the Ma Q qil tribes entered the Maghreb accompanying the Banū Hilāl (XI-XIII) and settled mainly on the outskirts of the latter's territory, specially the Sous and the region corresponding to current Mauritania. In eastern Morocco, the D wī Mans .ū r 5 settled along the Moulouya River and the deserts of Tafilalt, from Taourirt-in northern Morocco-to the Draa Valley, as far as Sijilmassa (Ibn Khaldun 2011, p. 2361). La Chapelle (1930, p. 89) claims that they remained in Tafilalt until the nineteenth century living among other tribes under Berber rule.
In southwestern Morocco, the settlement of groups of Ma Q qil origin happened more lately, during the Saadian rule (XVI-XVII), groups such š-Š@bbānāt 6 and l@-Mnābha emigrated from the Sous and established themselves on the territories of Q Abda and Z Q ir, on the Atlantic plains in central Morocco, but also at the surroundings of Marrakesh 7 (Colin [1937(Colin [ ] 1945. Moreover, in 1765, the foundation of the port town of Essaouira (Mogador), on the limits between Chiadma (Arabic speaking) and Haha (Tachelhit speaking) territories, attracted peoples, not only from these two neighboring territories, but also from distinct parts of the Sous, and among them š-Š@bbānāt and l@-Mnābha once again were attracted to the Atlantic plains taking part in the formation of Mogador's population (al-Kānūnī 1932;ar-Ragrāgī 1935;as-Sūsī [1966as-Sūsī [ ] 2005as . -S . iddīqī 1969).
Later in the nineteenth century, the flow of the trade of the Trans-Saharan route shifted westward to the Atlantic coast, due to the important role of the port of Essaouira 8 for the international trade. The city became connected with southern Moroccan cities by routes with Akka and Guelmim 9 . In this way, Essaouira was connected indirectly to Tafilalt and West African regions 10 . In the second half of the nineteenth century (1860-70s), the greater portion of the West Africa trade comes into Morocco via Tindouf and Sous to Mogador (Dunn 1971, pp. 278-80). Caravans were moving between Essaouira, Tafilalt and sub-Saharan regions, connecting their populations who probably used Arabic as a lingua franca in commercial relations 11 . In the nineteenth, Essaouira used to receive annually one or two caravans composed of thousand camels and smaller caravans as well, trading export commodities-such as gum and ostrich feathers-but also gold and slaves for the local market (Dunn 1971, p. 271). As Lévy (1998, p. 13) points out, certainly linguistic exchanges took place due to contact of the caravans with local populations while passing by rural markets on their routes across the south.
The trans-Saharan slave trade was also very active by that time. El-Hamel (2013) shows that there was a continuous import of thousands of slaves into Morocco by wellestablished trade routes (Tindouf, Ijil and Twat). He estimates that, by the end of the nineteenth century, the total black population was of half a million people (pp. 245-46). According to him, many were sold in the markets of Fez, Mogador and Marrakesh (p. 251), and besides that, a part of the enslaved people from sub-Saharan Africa could be found in the sugar refineries near Essaouira, in Haha and Shishawa territories (p. 152).
It cannot be ignored that Essaouira and Tafilalt were connected, despite of the distance, by the caravans moving between the two regions due to the trans-Saharan trade. And this fact may be important to explain certain singular linguistic features in both localities.

Southern Moroccan Features (Results)
The following selected features may help understanding more deeply what set the southern dialects together or apart, according to the innovative or conservative nature of these traits.
Nevertheless, the retention is not attested in most of the southern dialects analyzed here. Even in the few dialects that it is attested, Chiadma and Tafilalt, the feature still occurs along with the reduction of diphthongs in /ī/ and /ū/. The maintenance of diphthongs and allophonesō andē in both localities does not appear to be a result of Trans-Saharan trade connections, given that the feature is absent from the urban Essaouira dialect. Therefore, the feature seems to be of Saharan origin, as attested by its occurrence in Saharan dialects, and might be an evidence of the nature of the Arab settlement in both localities.
The prevailing opinion is that the occurrence of -āt (3f. perf.), in Moroccan dialects, is due to the analogy with weak verbs (e.g., mšāt , "she went") (Heath 2002, p. 223;Aguadé 2008, p. 291). Regarding the diffusion of the feature, urban centers-such as Casablanca, Meknes and Marrakesh-may play an important role in it. For instance, Aguadé interprets the occurrence of -āt in Settat xădmāt , "she worked" (Aguadé 2013, p. 4) as a convergence towards the Casablanca variety. In my opinion, regarding southern Morocco, Marrakesh may also have diffused the suffix in the region, given that it is well attested in the city, e.g., Marrakesh s@mQāt "she listened" (Sánchez 2014, p. 121).
It is not clear whether -@t spread earlier than -āt. On one hand, the neighboring Saharan type varieties do not exhibit the ending -āt, as one can attest in H . assānīya: kt@bt "she wrote" (Cohen 1963, p. 91); or in Algerian Sahara in the Mzāb region: k@tb@t "she wrote" (Grand'Henry 1976, p. 43). On the other hand, the ending -āt is found in other parts of the Maghreb: Eastern Libya ik'tib-at "she wrote" (Owens 1984, p. 105). It is also found in the Maghreb neighbouring areas, like in West Sudanic 15 katabat "she wrote" (Owens and Hassan 2009, p. 713).
The fact is that -āt-in the Moroccan case-must consist of a conservative feature just like in other parts of the Arabic speaking world. The ending -at, with a short vowel, in strong verbs, is found in many eastern dialects, not only inside the Arabian Peninsula, but also outside of it in Bedouin-type dialects (Gaash 2013, p. 49).

The Clitic -ki (2s.f.)
The occurrence of the clitic -ki (2s.f.) is very common in semiverbs all over Morocco. Heath (2002, p. 242) confirms it, but he did not analyze the use of -ki in the possessive function.
Regarding the origin of this feature, I claimed previously (cf. Francisco 2019b, p. 164) that the occurrence of -ki in the possessive function-in Essaouira and Tafilalt-resulted probably from a morphological analogy with semiverbs, given that it is absent in Saharan type varieties such as H . assānīya nsāk @nti "he forgot you" (2f.) (Cohen 1963, p. 151) or the Mzāb region variety in Algeria, which do not distinguish the gender, using only -@k/-k (Grand'Henry 1976, p. 67). However, the historical links between southern Morocco and the West African region could provide us with a new hypothesis. It is not unrealistic to think that -ki entered Morocco, and remained restricted to the south, due to the slave trade connecting Tafilalt to the sub-Saharan region, given that enslaved boys and girls were brought to Morocco from parts of Western Africa, such as Nigeria and Chad (El-Hamel 2013, pp. 130-31). Moreover, the clitic is found in West Sudanic buyūt-ki "your houses (f.)" (Owens and Hassan 2009, p. 712), being clearly a retention. Such hypothesis would deserve a more in-depth discussion though.
Regarding the exclusive occurrence of -ki in Essaouira and Tafilalt, it could be explained by the linguistic link resulted from the caravans of Arabic speakers which connected both localities.
In the south, this feature appears to be evidently of Saharan origin, but restricted to Sous and Essaouira, probably due to the settlement of Saharan dialects speaking tribes as mentioned in the Section 2.
The particle is a Hilāli feature attested in other parts of the Maghreb as well. The preverb bais found in the Sahara, in Algerian southwest: Saoura ba-iQ@rr@s "he will get married" (Grand'Henry 1979, p. 224); and the particle bis also used to express the future in Bedouin Libyan dialects, e.g., Al-Khums b-yQāw@d "he will repeat" (Benmoftah and Pereira 2017, p. 317). Sánchez (2014, p. 183) mentions the occurrence of the future particle bāin some dialects of Yemen, for which he suggests a common etymology with the verb ba "to want" in Marrakesh. Despite the fact that Ma Q qil tribes are assumed to have come from Yemen, the verb bġa is employed in H . assānīya to express intention only, not expressing the future (Taine-Cheikh 2004, p. 225). However, H . assānīya does apply the same structure above with the verb idōr . "to want" as an auxiliary to express the future: idōr . it .ī h . "he is going to fall" (ibid, p. 224). This structure to express future is an innovation common to Hilāli-Bedouin dialects in southern Morocco, but also in the Sahara and other parts of the Maghreb. -Cheikh (2017, p. 26) proposed that the southern Moroccan dialects were similar to the Casablanca dialect according to a list of common features presented above. On one hand, the southern dialects exhibit traditional Hilāli features, such as the realization [g] for *q, but, on the other hand, they also exhibit the loss of interdentals, a distinctive trait of Bedouin dialects. Considering the difficulty to classify these dialects as Hilāli, the selection of features by Taine-Cheikh (cf. Section 2.1) attempted to draw a group of southern dialects, but it was not able to single these varieties out, distinguishing them from other varieties in Morocco. To give a few examples, the following features cited above (cf. Section 2.1) are quite spread all over the country: the future particleġādi; the no gender differentiation for the 2s. clitic -k; and the ending -īw for defective verbs (3pl.imperf).

Taine
More recent data, especially from Essaouira and Tafilalt, shed a new light on the reality of southern dialects. Comparing them with the well-known dialects of the region-Sous and Skoura-and also with the dialects of the Saharan neighbouring areas-H . assānīya and the dialects of Algerian regions of Saoura and Mzāb-demonstrated that the varieties of the southern region are not so homogenous as we thought previously. The findings revealed a Bedouin color for the southern area due to the occurrence of retentions and innovations, some of them comprehending Saharan variants.
The Saharan traits attested in the southern varieties co-occur with the variants presented by Taine-Cheikh (cf. Section 2.1). They are the following retentions: maintenance of diphthongs /ăw/, /ăy/ in pharyngealized and plain contexts, sometimes realized as [ō] and [ē]; and the suffix -u (pl. imperf.) for defective verbs. Both features are spread in Bedouin dialects beyond the Moroccan borders. Nevertheless, in the south, the dialect of Skoura is an exception, not exhibiting these features. Curiously, regarding the suffix -u, there is no register of it for Tafilalt.
Concerning the conservative trait -āt (3.f. perf.), the variant is well spread all over Morocco along with -@t, however, it proved to be dominant in the south, where the latter is seldomly registered nowadays, except for Skoura. Despite of the absence of -āt, in H . assānīya and other Saharan varieties of the region, it does seem to have a Bedouin origin, as the suffix is attested in other parts of the Maghreb (e.g., Eastern Libya) and also in neighbouring varieties, like West Sudanic. And even though the feature occurs in other parts of Morocco, it can be considered characteristic of southern dialects.
Another representative case is the verb ba~bġa "to want" and the particles derived from it (b-, ba-) attached to imperfective verbs to express future. This feature consists of the single variant connecting all the southern dialects apparently, distinguishing them from the dialects of northern Morocco. Since the feature has reflexes on H . assānīya and is attested in other varieties across the Maghreb, being registered even in Yemen, this innovation may be an evidence of a common Bedouin or Hilāli origin for the southern dialects, along with traditional traits such as [g] for *qāf and the occurrence of interdentals in a previous stage.
Within southern dialects, the clitic -ki (2f.), in the possessive and object functions, builds a bridge between Essaouira and Tafilalt, not being attested in any other dialects across the whole country until now. The retention of the clitic restricted to these southern extremities, West and East, could be explained by the contact performed by caravans in the Trans-Saharan trade routes linking distant parts of the south with the Sahara and the Sub-Saharan Africa. This shared past in the south must have played an important role to the diffusion of the other features mentioned here as well.

Conclusions
Southern Moroccan varieties proved to exhibit more Bedouin features than previously thought. Colin (1986) tackled the Bedouin-Hilāli issue, pointing out that some dialects could not be classified fully as Bedouin-type for not maintaining conservative features. That was the case of the southern Moroccan dialects, which were thought not to exhibit Saharan-type features. Here, linguistic findings revealed the opposite though.
In agreement with the traditional dialectology scholarship, the occurrence of Saharantype retentions in southern sites could be explained by the settlement of the Ma Q qil in the area as confirmed in historical sources. Nevertheless, it is important to note that the Ma Q qil tribe settled in Tafilalt and in the Atlantic strip with a gap of centuries. Furthermore, there is no linguistic evidence that the Ma Q qil tribe presented a dialectal unity-as the tribe was subdivided into distinct groups (but .ū n)-and, if so, that it preserved this unity through centuries until modern times. Therefore, populational movements caused by multiple factors, such as the Trans Saharan trade, famine crisis and epidemics which swept southern Morocco, should not be ignored when trying to explain the spread of certain variants.
The only innovation feature which seems to gather the bunch of southern dialects, linking them with other Bedouin varieties in Algerian Sahara and Libya, is the future construction ba~bġa "to want" > b-/ ba-+ imperf. Apparently, the origin of this feature goes back to the Yemeni particle bā-, nevertheless, it is not attested in H . assānīya, though the dialect is associated to the Ma Q qil tribe, supposedly from Yemen. Despite of that, H . assānīya attests a similar structure with the same function (e.g., idōr . it .ī h . , cf. Section 3.5), which may have developed after the former with ba~bġa > b-/ba-. Moreover, the occurrence of the structure (b-/ ba-+ imperf.) in other parts of the Maghreb may warn us that not all the features in current Saharan varieties, especially H . assānīya, may be representative of the "purest" Bedouin-type, or Hilāli, neither in Morocco nor in the Maghreb.
The occurrence of certain conservative features in certain parts of Morocco and the Maghreb also corroborates the previous argument. That is the case of the retention of the ending -āt (3f. perf.) and the conservative clitic -ki (2f.) both absent from Saharan-type varieties.
The current linguistic situation of southern Moroccan dialects highlights the limitation of the 'Hilāli' category, including the 'Ma Q qili' label, to deal with dialectal layers within Moroccan varieties. The difficulty in applying this category is due to the co-occurrence of features, or variants, of distinct origins in the current dialects. Trying to determine dialectal groups based on the Arabization waves does not prove to be sufficient anymore, given that speakers of distinct Arabic dialects have been in movement and contact for centuries in the area.  I collected the data for the rural area of Essaouira in Aquermoud and SīdiĪsh .ā q during linguistic fieldwork in the city of Essaouira in the years of 2016, 2017 and 2018. Some inedit data for Essaouira and the Chiadma territory are published in this paper. 5 Ibn Khaldun (2011) claims that the Ma Q qil tribe were Bedouin from Yemen, being divided into three groups (but .ū n): D wi Q Ubīdullāh, at-T a Qā liba, D wi Mans .ū r e D wi H . assān (pp. 2363-70). 6 Dialectal pronunciation of the name. In Kitāb al-Q ibar (Ibn Khaldun 2011) the name is registered aš-Šubbānāt.

7
On the reasons these tribes moved to this area, see: az-Zayyānī (1886); ad-D . u Q ayyif (1986) and al-P Ufrānī (1998). 8 See Schroeter (1988) on the role of Essaouira for international trade in the nineteenth century Morocco. 9 Guelmim was one of the redistribution and places of concentration of caravans such as Aboudam, Ghadames and Assiout (Miège 1981, p. 96). 10 In pre-modern times, the Sahel, or the sub-Saharan West Africa was designated in Arabic under the term bilād as-Sūdān. The recent Historical publications on the Trans-Saharan trade, cited in this paper, maintained the use of the term "Sudan". 11 For the possible use of Arabic as a lingua franca in the trade with West Africa, see Levtzion (2000, p. 64) and Bouwman (2008, p. 135). 12 The Chiadma realization of this vowel is very similar to H . assānīya as described by Cohen (1963, p. 53) who sees it as a long vowel followed by an "appendice labial ou palatal", which he represents byē y ,ō w .

13
-īt is the Jewish variant for -āt of strong regular verbs, in Essaouira (Francisco 2022, in press). Heath (2002, p. 224) posed two hypothesis on the origin of the suffix -īt: (i) a mutation of *-at or (ii) a lengthening of *-@t to keep the 3. perf. f. suffix distinct from the first and second persons suffixes.