The Classification of Bedouin Arabic: Insights from Northern Jordan

Abstract

The goal of the present paper is to provide a revaluation of the classification of the Bedouin dialects of Northern Arabia and the Southern Levant, based on published or publicly available data and on first-hand data recently collected amongst some Bedouin tribes in Northern Jordan. We suggest extending previous classifications that identify three types of dialects, namely A (ʿnizi), B (šammari), and C (šāwi). Although intermediary or mixed types combining šammari features with šāwi features were already noted, our data suggest that further combinations are possible, either because they had so far been unnoticed or because recent levelling and dialect mixing have blurred the boundaries between some of the varieties.

1. Introduction

The goal of the present paper is to provide a revaluation of the classification of the Bedouin dialects of Northern Arabia and the Southern Levant, based on published or publicly available data and on first-hand data recently collected by the authors amongst some Bedouin tribes in Northern Jordan. We suggest extending Cantineau’s (1936, 1937) classification that identifies three types: A (ʿnizi), B (šammari), and C (sāwi). Although Cantineau already noted intermediary or mixed types combining šāmmari features with šāwi features, our data suggest that further combinations are possible, either because they have so far not been noticed or because recent levelling and dialect mixing have blurred the boundaries between some of the varieties. Foundational surveys include Cleveland (1963) who, much in the same way as Blanc (1964) coined the gilit–qəltu dichotomy, coined the dialectonyms biqūl, bikūl, bigūl, biʾūl and yigūl based on the 3.m.sg. of the imperfective of the verb *qāl ‘he said’. Further developments can be found in Palva (1984). Palva divides the Bedouin dialects of the Southern Levant into four groups, as below:

• The dialects of the Negev Bedouins.
• The dialects of the Arabia Petraea Bedouins such as the Ḥwēṭāt.
• The dialects of the Syro-Mesopotamian sheep-rearing tribes, which corresponds to the šāwi type (Cantineau’s type C, Younes and Herin 2016).
• The dialects of the North Arabian Bedouins (Cantineau’s types A and B).

The problem with the biqūl–yigūl appellation is that it fails to capture the difference between a major split in Jordan, namely between dialects that exhibit final /n/ in the imperfective endings -īn and -ūn and those which exhibit -ī and -ū (Herin 2019). Using the 3.m.pl. of the imperfective of qāl would partially solve this problem, which, combined with geography, yields the following classification: Southern ygūlu, Central ygūlu, and Northern ygūlūn. Central ygūlu is in many ways identical to the Northern ygūlūn šāwi C; the presence or absence of /n/ is the main difference. Only Southern ygūlu is an extension of the North-West Arabian type (Palva 2011). Our focus will be the hitherto under-studied Northern ygūlūn type with a special focus on the Misāʿīd dialect which exhibits many šammari features such as the apophonic passive (yiḏkar ‘it is remembered’) or a [d^j] reflex of */ǧ/ (d^yibal ‘moutain’), but also šāwi-like traits such as the [q] < /ġ/ (qēr ‘other’) and more surprisingly, features that are reminiscent of North-West Arabian such as the resyllabification of *ʾinC₁aC₂aC₃a… into ʾinC₁C₂v́C₃a (ʾin^ǝḥkúmat ‘it was ruled’). Consequently, the major taxonomies have to be combined to represent the overall picture more accurately. Additionally, sociolinguistic developments which have affected the classification of these dialects, such as dialect contact and koineization, need to be incorporated.

The data on which this paper draws were collected amongst members of the Misāʿīd tribe in 2019 in the municipality of Umm al-Ǧimāl in Northern Jordan, twenty kilometres East of Mafraq. With the help of Youssef Al-Sirour, a permanent resident of Umm al-Ǧimāl and an immediate member of the community under investigation, we visited local families and recorded two casual conversations. Because of the limited nature of the corpus, the present discussion should be considered provisional until more data are collected. We will first sum up Cantineau’s classification followed by those put forward by Cleveland and Palva. Based on our own observations, we suggest essential amendments to these classifications. We then present the salient features of the dialect, followed by a small sample taken from the recordings. The last part deals with the classification of the present dialect in the light of previous literature. We also highlight some methodological issues regarding data collection, levelling, and short-term accommodation.

2. Cantineau’s Classification

The first scholar to draw a comprehensive classification of the Bedouin dialects of Northern Arabia is Cantineau (1936, 1937). The first distinction relates to the occupational profile of the Bedouins located in this area, whom Cantineau called “grands nomades” (‘great nomads’) as opposed to “petits nomades” (‘little nomads’). The former designates tribes which mostly rely, at least historically, on camel rearing, and the latter designates tribes which were mostly active in sheep rearing. This bipartite separation was further divided into three broad groups to which he attributed the letters A, B, and C. The A-group designates camel-rearers from the ʿNiza confederation. The B-group refers to camel-rearers from the Sămmar confederation, whereas the C-group refers to the sheep-rearing tribes of the Syro-Mesopotamian bādya ‘steppe’. More marginally, Cantineau also talks about three smaller subgroups, the variety of ar-Rass in the Gaṣīm region in the central-northern part of Saudi Arabia, the dialect of al-Ǧōf located in the far north of Saudi Arabia, and finally the dialects of the oasis of the Syrian desert of al-Qarītēn, Palmyra and Suxne.

Some features of the A-group (ʿNiza) include the affricate [ʦ] and [ʣ] of etymological /k/ and /g/ (Standard Arabic /q/) in the vicinity of front vowels: ćalbati ‘my she-dog’ (< kalbati), ǵiddam ‘front’ (< giddām). Etymological /ǧ/ can be realized [g^j], [d^j], and [ʤ]: did^yād^ya ‘hen’(< daǧāǧa ‘hen’). The feminine ending -a exhibits no raising except in the vicinity of /i/, /ī/, or /j/ in which case it raises towards [æ]: laḥyä ‘beard’ (< liḥya). Etymological diphthongs /aw/ and /ay/ are not monophthongised although the distance between the two elements is reduced, yielding, respectively and approximately, [o^w] and [ɛ^j]: ǧowz ‘nut’ and beyt ‘tent’. An important feature is the so-called gahawa syndrome, understood as the insertion of an anaptytic /a/ vowel between /ġ/, /x/, /ḥ/, /h/, or /ʿ/ and a following consonant of the type Ø → /a/ / aX_C in which X is one of the aforementioned consonants and C is different from X: ḏ̣ahr → ḏ̣ahar ‘back’. In addition to this, *C₁aC₂aC₃v sequences are resyllabified into C₁C₂v́C₃v: xšíba < xašaba ‘piece of wood’. The gahawa syndrome is also active in the passive participle template *maC₁C₂ūC₃, in which case it also combines with the resyllabification rule: maḥṭūṭ → maḥaṭūṭ → mḥaṭūṭ ‘put’. Another important distinction introduced by Cantineau is trochaism vs. atrochaism. While these terms refer to a type of meter in Classical Greek poetry, his use of this parameter entails a particular syllabic type. Accordingly, Cantineau separates trochaic from atrochaic varieties. Trochaic varieties have the tendency to favour sequences of Cv/Cv syllables. CvC syllables are tolerated in final position or if followed by Cv or a final CvC/CvC: iḥáṣadan ‘they (f.) harvest’, yākalan ‘they (f.) eat’, rāsa-na ‘our head’, nāgat-i ‘my she-camel’. Atrochaic dialects do not restrict sequences of CvC syllables: iḥáṣdan ‘they (f.) harvest’, yāklan ~ yāčlan ‘they (f.) eat’ rās-na ‘our head’, nāgt-i ‘my she-camel’. The A group is strongly trochaic.

As far as morphology is concerned, these dialects feature the nominal suffix -n commonly called ‘nunation’ in Semitic studies, which essentially marks nouns denoting indefinite specific referents when they are complex NPs consisting of a nominal head and a modifier (Holes 2004). Another salient feature is the pronominal indexes which feature a final /n/ in the prefix conjugation: t(v)gūlīn ‘you (f.) say’, t(v)gūlūn ‘you (m.pl.) say’ and y(v)gūlūn ‘they (m.pl.) say’. As far as bound pronouns are concerned, a noticeable trait is the allomorph -ah of the 3.f.sg. after a final weak root consonant: ʿaly-ah ‘on her’ and abw-ah ‘her father’. The 2.m.sg. and 2.f.sg. in those dialects surface as -k and -ć after words ending in a short vowel: farás-k ‘your (m.) horse’. The 2.m.pl. and 2.f.pl. forms are -kam and -kin and the 3.m.pl and 3.f.pl. are -ham and -hin. Specific independent forms of free pronouns include 1.sg. āna and 1.pl. ḥinna. Another salient feature is the forms of the verbs axaḏ ‘he took’ and akal ‘he ate’, instead of kala, xaḏa.

As far as group B (šammari) is concerned, much of the phonology and morphology is shared with group A. Differences arise in the following features. As noted by Cantineau (1937, p. 130), “l’imāla de la terminaison féminine est nette et forte, a un tel point qu’elle semble résister au tafxīm d’une consonne précédente”: gargūre ‘she-lamb’, nāge ‘she-camel’. These dialects are also characterised by the lenition of the feminine plural ending -āt in pause in which case it reduces to -āⁱ: xams əbṣalāⁱ ‘five onions’. Concerning bound pronouns, šammari dialects exhibit -ak and -ić in the 2.m.sg. and 2.f.sg. with any vowel syncope. In addition to this, the 1.sg. allomorph -an surfaces in all positions: ḏ̣rub-an ‘he hit me’ (< ḏ̣arab-an → ḏ̣aráb-an → ḏ̣rub-an). Cantineau also notes the allomorph -(w)o after final long -ā: ġadā-o ‘his lunch’. Our data suggest that this allomorph is selected after any long vowel, whether plain or monophthongised.

Group C dialects, also known as šāwi dialects, are spoken by the sheep-rearing tribes of the Syro-Mesopotamian bādya ‘steppe’ and its fringes. Distinct features include the affricates [ʧ] and [ʤ] as reflexes of /k/ and /g/ in front vowel environments. The reflex of etymological /ǧ/ is always the affricate [ʤ]. A slight raising towards [æ] of final -a and -ā is heard in non-back and non-velarised contexts: šinīnä ‘butter milk’, iḥnä ‘we’. In terms of phonotactics, *maC₁C₂ūC₃ stems are not susceptible to the gahawa syndrome and hence, there is no resyllabification. Šāwi dialects are also atrochaic, in that sequences of CvC syllables are not restricted: yihárban ‘they (f.) escape’, yāklan ‘they (f.) eat’. Specific morphological forms are 1.sg. āni ‘I’ and iḥnä ‘we’ for free pronouns and the pairs -kum/-čən and -hum/-hən.

3. Cleveland’s Classification of the Dialects of Transjordan

Cleveland (1963) is an attempt to classify the dialects spoken in Jordan and Palestine, both sedentary and Bedouin. Cleveland coined new terms using the 3rd person singular of the verb qāl ‘he said’ in the imperfective in order to designate the different dialectal groups. His first cluster, which he calls yigūl, refers to all the Bedouin varieties which lack the b- prefix of the imperfective. The second group he distinguishes is bigūl, by which he refers to the sedentary populations of Jordan, including some locations on the west bank of the Jordan river. His third group is the bikūl type, which is characteristic of the sedentary rural populations of central Palestine. Lastly, the biʾūl group incorporates the sedentary urban populations of Palestine, including those which settled more recently in Jordan. Cleveland does not mention a biqūl group which would include the Druze dialect of Azraq, Northern Jordan. This dialect is as yet undocumented but research in this community is ongoing and the findings will be published in due course.1 As we will see below, Cleveland’s classification does not capture important differences found amongst the Bedouins. It also fails to capture the divergences amongst the indigenous sedentary dialects of Jordan, which, although all belong to the bigūl group, exhibit a sharp division between a southern muʾābi type and a northern-central balgāwi-ḥōrāni type.

4. Palva’s Classification

Palva (1984) delves deeper into Cleveland’s classification using a larger pool of variables. Palva mentions the urban Palestinian dialects, which correspond to Cleveland’s biʾūl. As far as rural dialects are concerned, he distinguishes between Galilean dialects (biqūl), central Palestinian dialects (biḳūl), south Palestinian dialects (bigūl), north and central Transjordanian dialects (bigūl), and south Transjordanian dialects (bigūl). His classification of the Bedouin dialects includes those of the Negev Bedouins (bigūl), the dialects of southern Jordan (yigūl), the dialects of the Syro-Mesopotamian sheep-rearing tribes (yigūl), and lastly the dialects of the North Arabian Bedouins (yigūl). Palva’s classification distinguishes well between all the subgroups of the sedentary types but lumps together sub-divisions within the Bedouin type that ought to be differentiated. In the dialects of the Syro-Mesopotamian sheep-rearing tribes, no distinction is made between the dialects of the Jordan valley and the šāwi type. As regards the dialects of the North Arabian Bedouins, no further distinction is made between Cantineau’s A and B groups.

5. Addenda to Cantineau, Cleveland, and Palva

5.1. Younes’ Subgrouping of Ca

So far, only tribes which had šāwi type dialects had been located and for some of them investigated, thus belonging to Cantineau’s C group. These are for example the Nʿēm, Lhēb, and Bani ʿAzz who, in Lebanon, are mainly located in the Northern and Eastern parts of the country. The dialects spoken by these tribes are all unmistakably of the šāwi type, exhibiting features such as the /č/ and /ǧ/ reflex of etymological /k/ and /g/, a first or second degree raising of final -a and -ā to [æ] or [ɛ], atrochaism, absence of the gahawa syndrome on the *maC₁C₂ūC₃ template, the pseudo-verb wədd ‘want’, and the lexeme əṯəm for ‘mouth’. In recent fieldwork carried out in the central part of the Bekaa valley by one of the authors of the present study, two new Bedouin tribes were investigated: the Abu ʿĪd and the ʿĪdīn. Their presence in that part of the country had been, until then, unnoticed. Indeed, the presence of Ḥsina clans, who are a big sub-section of the ʿNiza confederation and to whom the Abu ʿĪd and the ʿĪdīn are connected, was already attested in Syria. The Ḥsina are to the ʿNiza what the Ṭayy are to the Šammar in that they are the first clans who migrated northwards into the Syro-Mesopotamian steppe around a millennium ago. This resulted in a prolonged contact with Bedouin tribes who had migrated earlier into the area such as the Muwāli, Ḥadīdīn, and Nʿēm—who had dominated the Syro-Mesopotamian steppe. The linguistic outcome of this prolonged contact was convergence towards the šāwi type. After investigation, it turned out that the dialect of the Abu ʿĪd and the ʿĪdīn exhibited a similar profile, with core šāwi features alongside with ʿnizi features. For instance, these dialects exhibit no raising of -a and -ā, gahawa active in the *maC₁C₂ūC₃ template, the verb yibi ‘he wants’, and a more pervasive use of nunation. This state of affairs led us to coin a new term for this type of configuration, using Cantineau’s terminology. Consequently, it seemed opportune to use the combination of Ca letters to designate this type of dialects: upper case C for the šāwi component and lower case a for the ʿnizi component. Cantineau (1937) already used such a combination of letters for the varieties spoken in the Gaṣīm area in modern-day Saudi Arabia that combine predominantly šammari features alongside with ʿnizi features: Ba.

5.2. Herin’s ygūlu vs. ygūlūn

As noted in Herin (2020), one of the shortcomings of Cleveland’s yigūl type is that it lumps together three sub-types within the Bedouin dialects of Jordan: the dialects of the Jordan valley Bedouins such as the ʿAǧārma, ʿAdwān, and ʿAbābīd, the dialects of Bedouins of northern Jordan such as the Bani Ṣaxar, Sardiyye, Sirḥān, Āl ʿĪsa, and Misāʿīd, and finally the Bedouin varieties of Southern Jordan such as the Ḥwēṭāt, Bdūl and Zawāyda. The Jordan valley type differs from Cantineau’s C group in that they lack the final /n/ in the imperfective endings -īn and -ūn, also found in the dialects of the Bedouins of northern Jordan. It appears that it would be more conclusive to use the 3.m.pl. inflexion of the imperfective of the verb gāl to capture some of these differences. The following general classification would arise:

(I)

Sedentary bigūlu,

(II)

Southern Nomadic ygūlu,

(III)

Central Nomadic ygūlu, and

(IV)

Northern Nomadic ygūlūn.

6. Features of the Misāʿīd Dialect

In 2019, Bruno Herin, Enam Al-Wer, and Youssef Al-Sirour began fieldwork amongst the Misāʿīd tribe in Umm al-Ǧimāl, Northern Jordan. The fieldwork was facilitated by Yūsif, who is a member of the tribe, as noted above. In this exploratory phase of the research, we recorded two forty-minute sessions consisting of casual conversations and narratives. These recordings were subsequently transcribed and analysed. In the remainder of this article, we present our analysis of the salient features of this dialect based on these recordings.

6.1. Phonology

The phonetics of the feminine ending was mostly recorded as the unraised reflex [a]: šidīda ‘severe, extreme’, šāša ‘piece of fabric/muslin’, mayya ‘water’, waḥda ‘one (f.)’. A first degree raising was recorded in sāknä ‘dwelling (f.)’, ʿašīrä ‘clan’, ʿuṯmāniyyä ‘Ottoman’, lahd^yä ‘speech, accent’. A second degree raising was also recorded in a handful of items such as zġīre ‘small’ and kṯīre ‘much (f.)’, and also after an emphatic sound as in miḥmāṣe ‘coffee bean roasting pan’. The unraised reflex [a] is typical of ʿnizi type (in the Syro-Mesopotamian steppes) whereas the first-degree reflex is equally found in the šāwi varieties as in the ʿnizi dialects, although it is contextually conditioned (e.g., in front contexts). The second-degree raising found in some items most likely represents short-term accommodation, induced by the presence of speakers of other Jordanian dialects.2 It may also be indicative of the course of future developments in the dialect, viz. convergence to koineised Jordanian varieties, especially since the younger members of the tribe have frequent face-to-face contact with speakers of other Jordanian dialects through formal education and in the workplace. The raising heard in miḥmāṣe after a velarized consonant on the other hand, is typical of the šammari type. Despite some degree of variation in the realization of the feminine ending in our data, the distribution found amongst the informants overall is consistent with the ʿnizi type.

In pause, a slight aspiration occurs after the feminine ending: ʿašīrä^h# ‘clan’, gibīlä^h# ‘tribe’. This feature is found in both the A ʿnizi and B šammari groups.

The etymological diphthongs /aw/ and /ay/ are both monophthongised to /ō/ and /ē/, respectively: fōg ‘above’, yōm ‘day’, ḥōl ‘around’, dōr ‘turn/point in time’, and bēt ‘tent’, ṯnēn ‘two’, xēl ‘horses’. Diphthongised realisations occurred in Zbeyd (tribal patronym), xēyš ‘jute’. These reflexes are common in the group C šāwi dialects. Groups A and B usually have more consistent slight diphthongised reflexes.

As far as the affrication of etymological /k/ and /g/ is concerned, the recorded reflexes all pattern respectively with the šawi type /č/ and /Ǧ/: hīč ‘so’, čimä ‘desert truffle’, čiṯīr ‘much’. Only one instance of /ǧ/ < /g/ was recorded in ṭīǧ ‘endure’. Other items which were expected to be realised with /Ǧ/ were recorded with /g/: šarg ‘east’, giddām ‘in front’. This, in all likelihood, is a short-term accommodation phenomenon induced by the presence of speakers of standard Jordanian. The same observation can be made about non-affricated reflexes of /k/ in items such as kān ‘he was’, kiṯīr ‘much’ (also recorded with /č/, see above), and kibīr ‘big’ all of which are normally affricated in the vernacular.

Etymological /ǧ/ was recorded /d^y/ in d^yibal ‘mountain’, d^yaw ‘they (m.) came’, and id^yīban ‘they (f.) brought’. The affricate /Ǧ/ was also recorded: yiǧūn ‘they (m.) come’, ǧawwa ‘inside’, ǧild ‘skin’. The /d^y/ reflex is common in groups A and B whereas the affricate /ǧ/ is a hallmark of the šāwi type. The indigenous reflex is undoubtedly /d^y/. Although a short-term accommodation effect cannot be ruled out, the presence of /ǧ/ could also be due to earlier change within the dialect, as noted by Cantineau in some camel-breeder varieties.

An interesting and somehow unexpected feature that was occasionally recorded is the qalqala, understood to be the uvular realisation of etymological /ġ/: qēr ‘other’ (<ġēr), qāli ‘expensive’ (<ġāli), muqsil ‘washing area’ (<maġsil). To the best of our knowledge, this phenomenon is a hallmark of the Mesopotamian šāwi dialects.

Final /t/ in the plural feminine ending -āt interestingly drops in pause: ġuzā́# ‘raids’, šaġlā́# ‘things’, Rḏ̣aʿiyyā́# (toponym), ḥalālā́# ‘livestock heads’. This feature, as mentioned above, was already noted as commonly occurring in the B and Bc dialects.

The laryngeal stop /ʾ/ was recorded once as pharyngeal /ʿ/ in saʿalt ‘I asked’, which is a salient feature of North-West Arabian. In addition to this, /ʿ/ is often glottalised in pause: hassāʿ# [hassaːʕ^ʔ] ‘now’, māniʿ# [maːniʕ^ʔ] ‘hindrance’, bēʿ# [beːʕ^ʔ] ‘sale’.

Expectedly, *C₁aC₂aC₃v sequences are resyllabified into C₁C₂vC₃v: skánaw (<sakanaw) ‘they settled’, Šrufāt (tribal patronym < Šarafāt). Our corpus also attests the presence of resyllabification in derived templates such as form VII *ʾinC₁aC₂aC₃a: ʾinəḥkúmat ‘it was ruled’ (inḥakamat → inḥkamat → inḥkúmat → inəḥkúmat).

As far as the gahawa syndrome is concerned, it appears to be present in the dialect. Examples are nḥáṣid ‘we harvest’ (here combined resyllabification náḥṣid → náḥaṣid → naḥáṣid → nḥáṣid), baʿad ‘after’. Our data do not attest the presence of the gahawa syndrome in *taC₁C₂īC₃ and *maC₁C₂ūC₃ templates, which would suggest that it patterns in this respect with the šāwi type. Further data are needed to firmly confirm this observation.

As expected, the article receives primary stress as is normally the case in all of the Bedouin varieties of the area. To the best of our knowledge, only monosyllabic words of the type C₁v̄C₃ and disyllabic words of the type C₁vC₂v(C₃) can trigger the stress of the definite article. Attested instances in our data are: ʾál-muṭar ‘the rain’, ʾán-nifal ‘the clover’, ʾál-ʿarab ‘the Bedouins’. In addition to this and quite unexpectedly, we also encountered a stressed article with a C₁vC₂C₃v word in ʾáṣ-ṣaḥra ‘the desert’. Further data are needed to confirm whether stress assignment on the article is licenced in other words of this type and also possibly in other templates, which, as far as we know, would be a novelty.

An unexpected stress-related feature we found in the data is the second syllable stress in the plurals of C₁vC₂vC₃ type as in nigáṭ “points” which also surfaced as ngaṭ after high vowel elision in unstressed position. This is a feature found in North-West Arabian (Palva 2011).

6.2. Morphology

In the realm of verbal morphology, it appears that both the allomorphs -aw and -am in the 3.m.pl in the perfective are found: winn-o gṭaʿam kassaram min-ʿind giddām al-ǧamal ‘and there they had cut and broken into pieces (the engravings) in front of the camel’. The -aw allomorph was recorded in the following: ḥǝmaw baʿaḏ̣-ham ʿāšaw u-tikāṯaraw u-lamma tikāṯaraw, d^yaw ^ǝṯbitaw hānä ‘they protected each other, lived and multiplied and when they multiplied they came and settled here’.3 These examples suggest that -aw and -am allophones are not in complementary distribution, unlike in some šāwi tribes along the Middle-Euphrates where one of the allomorphs is used exclusively in pause.

Person prefixes in the imperfective were often recorded with /a/ vowel: yaṭlaʿ ‘he goes out’, takbar ‘it gets bigger’, yamši ‘he walks’, talga ‘you find’. This is a typical camel-rearing trait not found in the šāwi dialects.

Initial glottal stop verbs such as akal and axaḏ behave similarly to what is found in the B, Bc, and C groups: kalēt-o ‘I ate it’, unlike ʿnizi-type dialects which have akalt and axaḏt ‘I ate/have eaten’, ‘I took/have taken’.

As far as derived forms are concerned, the causative Form IV template *aC₁C₂aC₃-yiC₁C₂iC₃ is well attested in our data: nǝṭǝlʿ-o w-unǝnǝfḏ̣-o ‘we take it out and dust it’, yumṭar ‘it rains’, yiwṣil ‘he brings’. The presence of this feature is not diagnostic of any sub-group but in the context of dialect contact and levelling, it is a noticeable feature. The imperfective of Form V *taC₁aC₂C₂aC₃ was recorded as ytiC₁aC₂C₂aC₃ as in ytidarrab ‘he trains’. Given that šāwi dialects are known for having yiC₁aC₂C₂aC₃ (yidarrab), the presence of this form is another indication of the camel-rearer background of the present dialect. This, in all likelihood, should also happen in form VI *taC₁āC₂aC₃ but our data lack instances of any verb of this type.

Another typical camel-rearer feature that is found in our data is what is referred to as the apophonic passive, known to be lost in the šāwi varieties. Only two instances were recorded: yiḏkar ‘it is remembered’ and timadd ‘it is presented’. The template in the imperfective yiC₁C₂aC₃ in which the /i/ vowel contrasts with the /a/ vowel was noted above as a marker of the active forms. Further data are needed to assess the productivity of the apophonic passive in the modern-day form of the dialect.

The pronominal morphology of the dialect appears to be mixed. We recorded the first person free forms ana and iḥna, which are found in the C-šawi group. Inversely, the bound plural forms -kam and -ham were found, which are camel-rearer forms. In the feminine plural, only the third person -hin is recorded in the data, but no second person. The first person singular bound pronoun surfaced as -an after a consonant: wǦiʿat-an ‘it hurt me’, tūd^yaʿ-an ‘it hurts me’. This -an form is typical of the B and Bc groups. In the same vein, we recorded the form -wo after long vowels, which are also found amongst the B and Bc groups: ʿalē-wo ‘on him’, ^ǝnnxallī-wo ‘we let him’, šifnā-wo ‘we saw him’. Moreover, an -ah allomorph in the 3rd person feminine singular was recorded after final /w/ and /y/ stems: ʿaly-ah ‘on her’, abw-ah ‘her father’, which patterns with both the A and B camel-rearer dialects. After consonants, initial consonant bound pronouns all have initial vowel allomorphs: bilād-a-na ‘our country’, kill-a-ham ‘all of them’. This, of course, is reminiscent of the trochaic syllable type of the dialect and a distinctive feature of all the A and B camel-rearer varieties.

7. Dialect Sample

We present here a sample of the recordings to enable the reader to capture the nature of the dialect. Because much of the sessions consisted of group conversations in which turns were for the most part quick and uncontrolled, it was difficult to isolate long stretches of monologue. Another problem that quickly surfaced was the presence of several instances of mixed forms, which are due to dialect mixing and perhaps ongoing changes in the dialect itself. As explained earlier, the session involved participants with different dialect backgrounds, which as we quickly realised, prompted the informants to accommodate towards other Jordanian dialects. Nevertheless, the two short excerpts exhibit salient features that can be safely attributed to the local form of speech of the Misāʿīd tribe.

Speaker 1: Bū Ṣāliḥ:

ʾal-Misāʿīd ham ʾakbar ʿašīrä w-al-ʿašāyir hāḏōl d^yiwār-na ʿašīrtēn ḏōl… kull al-ʿašāyir hāḏi ḥōl baʿaḏ̣-ha hān sāknä b-al-manṭaga hāy d^yīrān. ʾu-sābigan gabǝl an-nās kānat ^ktǝġzi^k ʿala baʿaḏ̣-ha sābigan gabǝl-ma nḥkúmat ha-l-^ǝblād yaʿni […] ʿala dōr al-ʿuṯmāniyyä yimkin tḥakm al-^ǝblād hāḏi kānt an-nās tḥǝ́ma baʿaḏ̣-ha b-al-g̣uẉẉa. yaʿni yġázu baʿaḏ̣-ham u-hāḏōl..ḥasb g̣uẉẉt al-ʿašīrä lli giddāma-ham […] mā-ni waʿi kiṯīr ʾana ʿumr-i yimkin ʾakṯar min-sabaʿīn sinä, ḥass mā smaʿǝt min-ha-l-^ǝgdām gabǝl. g̣āḷaw al-Misāʿīd mā ʿumra-ham ʾinno xaḏaw, ʾilli yfukkūn ḥāla-ham b-ǝl-ġǝzā́, yimd^yūn min-ġazu ^kyifukko^k ḥāla-ham, dāyman manṣūrīn sibḥānaḷḷāh.

The Misāʿīd are the biggest tribe and the other tribes are our neighbors, the two other tribes… All these tribes live next to each other here in the region, they are neighbors. In the past, people used to raid each other, before the region was under control […] I think in the days the Ottomans controlled this region, people used to protect themselves in a warlike manner. I mean they used to raid each other and these… It depends on the strength of the tribe which is facing them […] I don’t remember well, I am maybe older than seventy, it comes from what I have heard before from the elders. They said that the Misāʿīd never took, those who emerge during raids, they get out of the raid they emerge, always victorious God bless.

Speaker 2: Umm Ṣāliḥ:

ʾaxabbr-o bass ʾana mā daggēt la waḷḷa ʾi šufǝt waḷḷa šuft hān daggēt ʿa-l-īd-i wǦíʿat-an u-daggēt ʿalē-(h) […] bass yaṭlaʿ ad-damm xalaṣ yarbuṭan ʿaly-a^h ʿādi yiḏ̣all yōmēn mā tgīm-o winn-a xaḏ̣ra… bass ǝn-nās mā tʿárif inno ḥaṛām gabǝl… ʿa-l-basāṭa ʾi waḷḷa ʿa-l-basāṭa, zīnä w-aʿlāǦ ʾi ʿlāǦ ḏ̣arba ʾi ḏ̣arba ḏ̣arba gabǝl l-wāḥad lama yūǦaʿ-o katf-o katf-o ^kyiduggo^k ʿalē-wo ybaṭṭal yūǦaʿ-o ʿlāǦ yaʿni […] waḷḷa madri šift waḷḷa nās wāǦid ṯalāṯ ^ǝngáṭ ʾi billa la waḷḷa mā marrat ʿalay-yä mā ḏikart-ä… šuft niswān bī-hin ṯalāṯ nigáṭ ^ǝkbāṛ ʿaǦāyiz ʾi…

I will tell him but I didn’t get tattooed, by God I saw, here, I tattooed my hand [and] it hurt, I tattooed it […] when the blood comes out, it’s finished, they (f.) tied it normally for two days until it turns into a bruise… But people before didn’t know it was ḥaṛām… Because of simpleness, by God, because of simpleness, beauty, and remedy, yes, remedy, a blow, a blow, before, when someone had a sore shoulder, they would tattoo it and the pain would stop, I mean [it’s a] remedy […] by God I don’t know, I saw a lot of people with three dots [tattooed], yes, by God, this did not happen to me, I can’t remember it… I saw women with three dots [tattooed], old women yes…

8. Discussion and Conclusions

Below (

Table 1. Features of the Misāʿīd and the Bedouin sub-groupings.

	A (ʿNiza)	B (Šammar)	C (šāwi)	Bc (Sattelite Šammar)	D (North-West Arabian)
imāla treatment	X	X		X
Aspiration of -a in pause	X	X		X
Diphthongs			X		X
Affrication			X	X
Etymological /Ǧ/	X	X		X
qalqala			X
Elision of /t/ in -āt#		X		X
Resyllabification of *C1aC2aC3v	X	X	X	X
saʿal for saʾal					X
Resyllabfication in derived verbs					X
gahawa syndrome	X	X	X	X	X
Stress on plural C1vC2v́C3					X
Stressed definite article al-	X	X	X	X	X
Trochaism	X	X		X
3.m.pl. perfective -am/-aw			X
Vowel /a/ in the imperfective	X	X		X
kala-xaḏa		X	X	X
Form IV	X	X	X	X	X
Form V et VI ytiC1aC2C2aC3/ytiC1āC2aC3	X	X		X
Apophonic passive	X	X		X
Free pronouns ana-iḥna			X	X
Bound 1.sg.-an		X		X
Bound 3.m.sg.-wo		X		X
Bound 3.f.sg.-ah	X	X		X

) is an overview of all the features discussed above and their distribution in the relevant dialectal groups. As mentioned earlier, Cantineau attributed a letter-code to the different groups he investigated. The two-way division is between the camel-rearer type which sub-divides into A (ʿNiza) and B (Šammar) and sheep-rearer C group (šāwi). In accordance with this classification, we decided to allocate the letter D to the North-West-Arabian type. From the Table 1 below, it quickly appears that the dialect of the Misāʿīd patterns with camel-rearer type. More precisely, it also appears to be closely connected to Cantineau’s Bc type. In addition to this, our sample also reveals šāwi-like features such as the realisation of etymological /ġ/ as [q] (Younes and Herin 2016) and the treatment of diphthongs. Moreover, and quite surprisingly, some features that are attested in the North-West Arabian sub-group were found in the data. These are for example the resyllabification of *C₁aC₂aC₃v in derived forms such as *inC₁aC₂aC₃a, the second syllable stress in plurals of the *C₁vC₂vC₃ pattern (which may also lead to first vowel elision), and also saʿal for saʾal. In conclusion, the dialect of the Misāʿīd matches for the most part the Bc sub-group but with šāwi-like features and also characteristics that are reminiscent of the North-West Arabian type. The question is how to account for such a pattern. There are at least two possibilities. The first one is that more complex configurations may have been unnoticed by Cantineau who indeed was not in a position to get large samples of data from all the tribes in the area. The second possibility is that recent dialect contact between speakers of all these sub-groups may have occurred, leading to dialect mixtures, as instantiated in our sample.

In terms of data collection and methodology, fieldwork in contexts that involve a fair amount of dialect contact can yield puzzling and conflicting linguistic output. This can also be exacerbated by short-term accommodation in the direction of the speech variety of the researcher(s). It is therefore paramount to secure the presence of an insider participant who can take the lead in carrying out data collection.

As far as the general classification of the dialects of Jordan and beyond is concerned, combining Herin and Younes’ amendments to Cleveland, Palva, and Cantineau’s classifications, it seems reasonable to posit the following taxonomy. We suggest that subsequent research should be framed within this canvas.

(I)

Sedentary bigūlu

• Muʾābi (southern, Karak, Ṭafīle, etc…)
• Balgāwi-Ḥōrāni (central-north, Salt, ʿAǦlūn, etc…)

(II)

Southern Bedouin ygūlu (Ḥwēṭāt, Bdūl, Zawāyda, etc…)

(III)

Southern Bedouin bigūlu (mostly Nagab and Sinai)

(IV)

Central Bedouin ygūlu (ʿAǦārma, ʿAdwān, ʿAbābīd, etc…)

(V)

Northern Bedouin ygūlūn

• ʿNizi
• Šammari
  • Bc (Misāʿīd)
• Šāwi
  • Ca (Bū ʿĪd et ʿĪdīn in Lebanon, so far unattested in Jordan)