Al-Jallad. A Manual of the Historical Grammar of Arabic

MHGA, v. 2020-1 A Manual of the Historical Grammar of Arabic Notes on key issues in phonology, morphology, and syntax Ahmad Al-Jallad Version 2020-1 Draft for classroom use; check back for regular updates 1 MHGA, v. 2020-1 © Ahmad Al-Jallad contact: [email protected] Cover Photo: A boulder bearing an early Islamic Arabic inscription and a Safaitic inscription from northeastern Jordan, by A. Al-Jallad (June 2019). 2 MHGA, v. 2020-1 Prefaces Preface to version 2020-1 The second iteration of this handbook now includes a discussion of the broken plurals, focused on comparative Semitic cognates, their distribution, and some remarks on reconstructable patterns. The section is still preliminary and will be deepened in the next update. In terms of chronological divisions, I have suggested a new category, PreHistoric Old Arabic, to encompass the epigraphic fragments of the first half of the first millennium BCE as well as reconstructions of the language of this period based on them. Section 1.2 includes a discussion of sound changes operating in Old Ḥigāz (QCT) and the Tamīmī dialects of the Arabic Grammarians. A preliminary discussion on agreement, drawing on the work of S. Bettega, has been added. The discussion of the reflexes of case in the modern dialects elaborates in more detail on the development of the nominative and wawation. A detailed presentation of the mythologies of Arabic and some remarks regarding their composition has been added to section 0.2.1. Finally, part V now includes a chrestomathy of Safaitic with fully vocalized and glossed texts. New references are now linked to online documents but the bibliography remains very incomplete and will be unified and updated in a future version. Ahmad Al-Jallad January, 2020 Preface to version 2019-1 I first compiled this manual in 2014 to teach the Historical Grammar of Arabic at the Leiden Linguistics Summer School. I have since continued to update it with new material and insights, and have used various iterations to teach my classes at Leiden University and again at the Leiden Linguistics Summer School, the second time with Dr. Marijn van Putten. The book as it stands now is incomplete; future iterations will cover subjects not treated here, such as the plurals, the morphology of the infinitives and participles, and syntax. The bibliography is not fully formatted and the appendix of texts contains mostly Old Arabic inscriptions but will soon be expanded to include texts from all periods. This text has not been copy edited so please forgive any typos and other infelicities. It is my intention to keep this book open access and free for all to use for research purposes and instruction. Please feel free to cite this text but be sure to include the version number. I will archive the versions at H-Commons so that previous versions are available even though the main text will continue to be updated. 3 MHGA, v. 2020-1 Visit my academida.edu (https://leidenuniv.academia.edu/AhmadAlJallad) page to comment a permanent “session”. Users are encouraged to send me suggestions and improvements to better the overall text; I will acknowledge these contributions in the notes. I would like to thank Marijn van Putten for his corrections on this draft while using this manual in his courses and privately. Ahmad Al-Jallad Columbus, January, 2019 4 MHGA, v. 2020-1 Table of Contents 0. Arabic defined and its subgroupings 0.1 Arabic, linguistically defined 8 9 0.2 Arabic’s earliest history based on the epigraphic and archaeological evidence 12 0.2.1 Mythologies of Arabic’s past 0.3 The Arabic language family 20 23 0.3.1 Divisions of Old Arabic 23 0.3.2 Pre-Modern Islamic period 30 0.3.3 Literary Varieties 32 0.3.4 Modern Vernaculars 33 I Phonology 35 1.1 Proto-Arabic consonants and vowels 35 1.2 Proto-Arabic sound changes 36 1.2.1 Possible sound changes 37 1.2.1 Northern Old Arabic 38 1.2.2 “conventional” Classical Arabic sound changes 38 1.2.3 Old Ḥigāzi (QCT) 40 1.2.4 The Tamīmī dialects of the Grammarians 40 1.4 Excursus on some key consonants 41 1.4.1 Notes on the Sibilants 41 1.4.2 Notes on the Emphatics 45 1.4.2.1 Qāf 46 1.4.2.2 Ṣād 48 1.4.2.3 Ḍād 49 1.4.2.4 Ẓāʾ 52 1.4.2.5 *g = ‫ج‬ 52 1.4.2.6 The merger of ‫ ض‬and ‫ظ‬ 53 1.2.3 Common Sound changes in medieval and modern Arabic 1.5 Proto-Arabic – Semitic Sound Correspondences II Morphology 2.1 Independent Pronouns 53 54 56 56 2.1.2 The duals 61 2.2 Clitic Pronouns 61 5 MHGA, v. 2020-1 2.3 Nominal Inflection 65 2.3.1 State 65 2.3.2 Case Inflection in Proto-Arabic 66 2.3.2.1 Development of the case system in Classical Arabic 69 2.3.2.2 Development of the case system in Nabataean Arabic 70 2.3.2.3 Development of case in the QCT 71 2.3.2.4 Development of case in Tihāmah Arabic 72 2.3.2.5 Development of case in Najdi Arabic 72 2.3.2.5 The Development of case in the early Islamic period 72 2.3.2.6 Development of case in most modern Arabic languages 73 2.3.3 The adverbial endings 74 2.3.4 Gender 74 2.3.5 Number and agreement (in progress) 75 2.3.5.1 Singulative 75 2.3.5.2 Adjectival plurals and the external endings -ūna/īna and -āt-. 76 2.3.5.3 Broken Plural System 76 2.3.5.3.1 Common strategies for the pluralization of CVCC and CVCVC nouns 78 2.3.5.3.2 CuCuC 79 2.3.5.3.3 CaCīC 80 2.3.5.3.4 CiCaC/CuCaC 80 2.3.5.3.5 CiCCān/CuCCān 80 2.3.5.3.6 CaCaCat 81 2.3.5.3.7 CuCCāC 81 2.3.5.3.8 CuCaCāG 82 2.3.5.3.9 Plurals of paucity 82 2.3.5.3.10 Augmented plurals (preliminary) 83 2.3.5.3.11 Plurals of quadriradical nouns/nouns with long vowels CāCiC, CvCvvC(at) 84 2.3.5.3.12 a-insertion external plurals 86 2.3.5.4 A Note on Agreement 87 2.3.5.5 The Dual (noun) 88 2.3.6 Definite Marking 2.3.6.1 Assimilatory Patterns 88 91 6 MHGA, v. 2020-1 2.4 Morphology of the demonstratives and relative pronouns 2.4.1 Demonstrative particles and pronouns 92 92 2.4.1.1 Proximal demonstratives 93 2.4.2.2 Distal demonstratives 94 2.4.2 Relative Pronouns III The Verbal System 3.1 Prefix Conjugation 95 99 99 3.1.1 The vowel of the prefix 100 3.1.2 Irrealis Mood inflection 101 3.1.3 Mood in Old Arabic 102 3.1.4 Mood in the QCT 102 3.1.5 Mood in Classical Arabic 103 3.1.6 Modal alignment in the modern vernaculars 103 3.2 Suffix conjugation 105 3.3 Verb classes 107 3.4 Derived Stems 109 IV Notes on Syntax 115 4.1 Infinitive 115 4.2 Negation 116 4.3 Interrogative and conditional particles 118 V Appendix of early Arabic texts 119 5.1 Safaitic Chrestomathy 119 5.2 Old Arabic poetry 132 5.3 Funerary Inscriptions 134 5.4 Prayers 136 5.5 Dedicatory and Narrative 137 5.6 Votive 138 5.6 Arabic texts in Greek letters 139 7 MHGA, v. 2020-1 0. Arabic defined and its subgroupings The Arabic languages are a branch of the Semitic language family, today spoken by more than 300 million people. They include extinct epigraphic varieties, such as Safaitic, Hismaic, and Nabataean Arabic, as well as Classical Arabic, medieval literary varieties, often termed Middle Arabic, the myriad of modern vernaculars, and Maltese. In the past, most scholars regarded Classical Arabic, the literary language of AraboIslamic civilization, as the ancestor of all other members of this family. Yet in the wake of epigraphic research beginning in the 19th century and the serious study of the modern vernaculars on their own terms, it is clear that Classical Arabic is a sister language to other forms of Arabic rather than their direct ancestor. Classical Arabic and all of the other varieties mentioned above developed from an unattested common ancestor conventionally called Proto-Arabic. Proto-Arabic: This term refers to the reconstructed, common ancestor of all varieties of Arabic, from the ancient epigraphic forms to the modern dialects. It is unclear when Proto-Arabic split off from Central Semitic, its immediate ancestor. Northwest Semitic was already distinct in the 2nd millennium BCE, and Ancient South Arabian is first attested in the late 2nd millennium BCE. It is therefore possible that the grammatical and lexical features characteristic of Arabic emerged in this period. In terms of attestation, the examples of the Arabic language date to the early 1st millennium BCE, which provides a terminus ante quem for the branching off of Arabic and its diversification. Based on the epigraphic evidence and early features of contact with Northwest Semitic, Proto-Arabic was likely spoken in northwest Arabia and the southern Levant. By the second half of the 1st millennium BCE, the language began to spread throughout the Arabian Peninsula (see below). 8 MHGA, v. 2020-1 Map 1: Hypothesized Urheimat of Proto-Arabic; Northwest Semitic = Black; ProtoArabic = Red 0.1 Arabic, linguistically defined The Arabic languages are defined by an array of grammatical innovations distinguishing them from other Semitic languages. These innovations emerged in Proto-Arabic and were subsequently inherited by its offspring. Not all forms of Arabic will display all of these developments, but if a particular language exhibits most of these, then it can be reasonably suggested that the missing features were lost or absent by reason of gaps in documentation. The isoglosses characteristic of Arabic were first laid out by J. Huehnergard (2017) and modified by Al-Jallad (2018). Innovations of Huehnergard (2017), abridged: 1) the deaffrication of *s3 [ts] and its merger with *s1 [s] 2) the loss of the 1st person singular pronoun ʔanāku 3) the replacement of mimation with nunation (tanwīn) 4) the levelling of the -at allomorph of the feminine ending to nouns terminating in -t, compare Classical Arabic qātilatun to Hebrew qōṭɛlɛt < *qāṭilt; relics survive in words like bint- ‘daughter’ and ʔuḫt- ‘sister’. 9 MHGA, v. 2020-1 5) the levelling of the -na ending of the 3rd feminine plural prefix conjugation to the suffix conjugation, producing qatalna (Modern Arabic qatalin) from earlier *qatalā. 6) the mafʕūl pattern as a paradigmatic passive participle of the G-stem: ProtoArabic *maktūbun ‘written’. 7) the vowel melody u-i for the passive: Proto-Arabic *kutiba ‘it was written’. 8) the preposition fī ‘in’, grammaticalized from the word ‘mouth’ 9) the replacement of the anaphoric use of the 3rd person pronouns with demonstratives based on the proximal base: compare Proto-Central Semitic *suʔa ‘that’ with Classical Arabic ḏālika; Psalm Fragment ḏēlik; Najdi ḏāk; Levantine hadāk, etc. 10) the presence of nunation on nominal heads of indefinite asyndetic relative clauses: Najdi kilmatin rimyat ‘a word which was thrown’; Classical Arabic raǧulun raʔaytu-hū ʔamsi ‘a man whom I saw yesterday’. To these innovations, I (2018) would add: 11) The complex and asymmetrical system of negation, mā + suffix conjugation; lā + prefix conjugation, indicative, lam + prefix conjugation, jussive, and lan (<*lā-ʔan) + prefix conjugation subjunctive. 12) pre-verbal tense and aspect marking, Classical Arabic qad faʕala ‘he has done’, sawfa yafʕalu ‘he will do’; Safaitic s-yʕwr [sa-yoʕawwer] ‘he will efface’; Levantine b-yiktob ‘he is writing’, etc. 13) the use of ʔan(na) as a complementizer. 14) the independent object pronoun base *(ʔiy)yā. 15) the use of the a-marked prefix conjugation (yafʕala) as a subjunctive. 16) quasi-suppletive imperative for the verb ‘to give’, based on the h-causative of ʾatawa ‘to come’, hāt, hātī, etc. from *haʔti, etc. Eg. Levantine Arabic hāt ‘give’; Hismaic ht [hāt] idem. 17) a unique set of prepositions, including *ʕinda ‘at, with’, *ladun/*laday ‘at with’; ʕan ‘away, about’, etc. 18) a special vocative suffix in *mma: Classical Arabic allāhumma ‘O Allāh’; Hismaic hltm [hāllātomma] ‘O Allāt’. 10 MHGA, v. 2020-1 Arabic is classified as a Central Semitic language (Huehnergard 1995; Huehnergard and Rubin 2011; Al-Jallad 2018a), a sub-grouping of West Semitic. Its closest linguistic relatives are the Northwest Semitic languages (Ugaritic, Hebrew, Aramaic) and Sabaic in South Arabia. This classification is based primarily on the realignment of the verbal system, as will be discussed in section III. Figure 1: Classification of Arabic within Semitic (Huehnergard and Rubin 2011). In former times, Arabic was regarded as a South Semitic language (see, for example, Moscati 1964), based on some affinities with Modern South Arabian and Geʿez, but these seem to be due to areal diffusion either in a part of the Proto-West Semitic dialect continuum or in the historical period. These features include the L-stem, the broken plurals, and the *p > f sound change. The first two features are likely reconstructable to Proto-Semitic and are therefore not valid for sub-classification. The *p > f sound change perhaps did not operate in Proto-Arabic and only affected dialects that moved into the Arabian Peninsula in the historical period. Most scholars today reject the South Semitic subgrouping on the basis that it is not supported by any innovations. 11 MHGA, v. 2020-1 0.2 Arabic’s earliest history based on the epigraphic and archaeological evidence The earliest documented Arabic speakers inhabited North Arabia and the southern Levant, perhaps centered on and around the Ḥawrān, in the early 1st millennium BCE.1 Little about this stage of the language is known; nearly all surviving fragments consist of personal names and, perhaps, a single proper noun. One inscription from this period and region -- from Bāyir, Jordan at the upper end of the Wādī Sirḥān -- has been discovered: a short prayer in an undetermined Ancient North Arabian alphabet (Hayajneh, Ababneh, and Khraysheh 2015). The text invokes in the Arabic language the gods of ancient Edom, Moab, and Ammon, suggesting a degree of cultural interaction between the Arabic-speakers of the eastern steppe and the Canaanitespeaking kingdoms east of the Jordan. The linguistic features attested in the epigraphic record suggest that Old Arabic constituted a dialect continuum, which can be divided into two zones: a northern continuum and the upper Ḥigāz (Old Ḥigāzī). Map 2: The Old Arabic dialect continuum; mid-1st millennium BCE (?) 1 See Ephʿal 1982, 1974; Macdonald 2009; Al-Jallad 2018a. 12 MHGA, v. 2020-1 By the second half of the 1st millennium BCE, Arabic-speaking peoples had moved west, giving rise to the Nabataean kingdom on what was previously ancient Edom. The Nabataeans expanded north and south, spreading their language with them. By the 1st c. CE, Nabataean writing culture had reached the northern Ḥigāz, where, before this period, another Semitic language known as Dadanitic held sway.2 A large number of Nabataean texts, including one in the Nabataean Arabic vernacular, were carved at ancient Ḥegrā (modern-day Madāʾin Ṣāliḥ),3 and Nabataean trading colonies extended as far south as the Yemeni frontier. The Nabataeans also expanded to Taymāʾ and Dūmah, perhaps introducing Arabic to these oases and, eventually, replacing the local, non-Arabic Semitic languages, Taymanitic and Dumaitic, respectively. 4 At Qaryat alFāw, where there is archaeological evidence for a significant Nabataean colony, the influence of Arabic can be seen in a small number of local inscriptions produced in Ancient South Arabian languages, such as Minaic and Sabaic.5 2 On the linguistic features of Dadanitic, see Al-Jallad 2018b. 3 On these texts, see Healey 1993. 4 On Taymanitic, see Kootstra 2016. 5 The most famous of these is the Rbbl bn Hf ʿm epitaph; see Al-Jallad 2014; Beeston 1979. 13 MHGA, v. 2020-1 Map 3: Hypothesized expansion of Arabic by the end of the 1st Millennium BCE or early 1st millennium CE. At the same time that Nabataean trade, and consequently writing, flourished, the Arabic-speaking nomads east of the Ḥawrān, stretching from southern Syria to Dūmat al-Jandal, developed a sophisticated writing culture. While Arabic-language texts in this region date as early as the 1st millennium BCE, by the 1st c. BCE, a huge number of inscriptions in the Safaitic script, the northern-most variety of the South Semitic script, were produced, documenting in detail the local dialects of Arabic. Over 40,000 Safaitic inscriptions are so far known, and it is possible that more than twice this number remain undiscovered in the Syro-Jordanian Ḥarrah (basalt desert).6 In the same period, Arabic-speakers, stretching from Madaba to Tabūk, produce a large number of texts in another Ancient North Arabian alphabet called Hismaic.7 6 On the Safaitic inscriptions, see Al-Jallad 2015. 7 On Hismaic, see King 1990; Zwettler and Graf 2004. 14 MHGA, v. 2020-1 The competition of Arabics, so to speak, continued for the first few centuries CE, but by the 4th c. CE, one script and writing tradition had prevailed -- Nabataean. Indeed, in this century, the Namarah epitaph (328 CE) of the self-proclaimed malk ʔal-ʕarab koll-ah ‘king of all the Arabs’, Marʔalqays BAR ʕamro, set in stone the first truly monumental Arabic-language text in the Nabataean script.8 The events recorded in this document -- Marʔalqays’ battles against Asad, Nizār, Maʕadd, and Maḏḥiǧ -- mark the first appearance of the legendary tribal groups documented in Islamic-period sources. The Namarah Inscription (wiki commons) In northwest Arabia, the Nabataean script began to exhibit innovative letter shapes, leading towards the Arabic script proper. This phase of the script, spanning from the 3rd to the 5th centuries CE, is called by scholars Nabataeo-Arabic. By this period the Ancient North Arabian scripts seem to have disappeared and Nabataeo-Arabic is the exclusive epigraphic witness to the Arabic language, save for transcriptions of anthroponyms in Greek and Aramaic. 8 For the latest edition of this text, see Macdonald in Fiema et al. 2015. 15 MHGA, v. 2020-1 Mape 3: Geographical distribution of transitional script (Nehmé 2010) Nabataeo-Arabic inscription, 428 CE, Sakaka = S1 (Nehmé 2010) By the late 5th c. CE, the Arabic script, as we know it, appears for the first time in the epigraphic record. Inscriptions on a trade route north of Nagrān (Bīr Ḥimà), likely produced by travellers from the north, attest a number of Arabic anthroponyms in the fully evolved Arabic script. In the 6th c. CE, the script is also attested in the northern Ḥigāz, Dūmat al-Jandal, and Syria, indicating that Arabic, by this period, had spread widely across the Arabian Peninsula, replacing, at least in writing, the pre-Arabic Semitic languages of the Ḥigāz and North Arabia. It is unclear when Arabic first penetrates south-west Arabia (modern-day Yemen). By the end of the 1st millennium BCE, inscriptions from the northern Yemeni frontier, the 16 MHGA, v. 2020-1 so-called Haram region, exhibit a mix of Sabaic and non-Sabaic features, which could suggest a non-Sabaic, and possibly Arabic, substrate.9 However, so far, no pre-Islamic texts in the Arabic language have yet been discovered in Yemen nor is there compelling evidence for the influence of Arabic on Sabaic, or other Ancient South Arabian languages, in Yemen proper. So while it stands to reason that Arabic vernaculars, perhaps moving south along the Ḥigāz, entered Yemen in the pre-Islamic period, evidence in support of this is lacking. It is very possible that Yemen was not Arabicized in a significant way until the Islamic period. There is even less evidence as regards the spread of Arabic to eastern Yemen (Ḥaḍramawt), Oman and East Arabia in the pre-Islamic period. There are no preIslamic Arabic texts from these regions and, at least in the case of Oman/eastern Yemen, non-Arabic Semitic languages continue to be spoken there till this day. While no pre-Arabic languages survive in East Arabia today, the epigraphic record attests a shadowy language termed Ḥasaitic, stretching from the Ḥasā in the north to the Oman Peninsula in the south. The nomads of the Najd, Ḥigāz, and south-central Arabia produced a large number of inscriptions in varieties of the South Semitic script which scholars term “Thamudic”. While most of these texts consist simply of signatures, the ones that do contain more often attest languages quite distinct from Arabic, and most of the longer texts remain undeciphered. 9 See Stein (2004) on the features of these texts. 17 MHGA, v. 2020-1 Thamudic C text, #80, Najd (Winnett and Reed 1973)10 h dgn l-yd h-ʾlht mlt-s */hā dagan la-yad haʔilāhat millatu-su/ 'O Dagon, may his people be in the company of the gods' Thamudic C (Eskoubi 1999), Taymāʾ region wdd f sw | tʾlʿsswʾ | wdd (undeciphered) 10 This is my reading and translations. Winnet and Reed give the following translation: O Dṯn, I have a disease (?). By Hutaim for Tais. 18 MHGA, v. 2020-1 Map 4: Epigraphic Map of Pre-Islamic Arabia (Al-Jallad 2018b) It is unclear when and under which circumstances Arabic replaced these languages as a vernacular. Since Arabic seems to have taken root first in urban centers across the Peninsula, it is possible that the language diffused outwards from oases and towns, replacing the non-Arabic Semitic languages of the nomads, or that the language was spread by migrations of nomadic populations from the north, who assimilated the preexisting tribes of these areas. In the early 7th c. CE, Arabic, and more precisely the Arabic of the Ḥigāz, was catapulted onto the world stage. The once triumphant Nabataean Arabic yielded in the face of the Conquest’s momentum. At the town of Nessana, in the Negev, we can witness the reunion of the old Nabataean dialect with its forgotten sibling in the Greek transcriptions of the 7th c. CE. By the end of the 7th c. CE, no trace of the older Nabataean vernacular is to be found.11 11 On the Arabic of Nessana, see Isserlin 1969; Al-Jallad 2017b, 2017a. 19 MHGA, v. 2020-1 These new forms of Arabic were the vernaculars of the elites of the Arab Conquests and the language of Islam’s scriptures. Indeed, the Qur’an proclaims itself to be in ʿarabī ‘Arabic’, in order for its audience to understand.12 Much like the spread of Arabic across the Peninsula in Nabataean times, following the Conquests, Arabic was established in urban centers across the Umayyad state, and slowly diffused outwards to rural areas. Waves of later migrations over the centuries, both local and long distance, spread Arabic far beyond the urban enclaves of Islam’s first century. At the same time, a new kind of linguistic competition emerged. Different Peninsular Arabic dialects vied for prestige -- the Ḥigāzī vernacular of the Umayyad elites, as attested in early Islamic papyri, Greek transcriptions from this period, and indeed the Qur’anic Consonantal Text itself, was confronted by the artistic dialect of the pre-Islamic odes, the language of which seems to have had roots in the dialects of south-central Arabia. The prestige of the Qaṣīdah, which had become the medium of royal panegyrics in Umayyad times, seems to have given it an edge, and by the 8th century, even Qur’anic reading traditions inclined towards this register. In this period, a robust grammatical and lexicographical tradition evolved to document Arabics that were in-line with the norms of the Qaṣīdah, canonizing forever prescriptive notions of what ‘correct’ Arabic should be. This linguistic melting pot produced the Arabophone word we know today -- the myriad of vernaculars and the literary language of Islamicate culture, Classical Arabic. 0.2.1 Mythologies of Arabic’s past There are numerous mythological accounts of Arabic’s pre-history in Islamic-period sources. By the Abbasid period, Arabic had achieved supernatural status; the practical sense of ʕarabiyyun mubīnun - a message in the everyday language that could be understood - had been replaced by the notion that the Arabic language itself was sacrosanct and proof of the Qur’an divine message. And as such Arabic acquired a sacred history. It became the ur-language of mankind, a divine mode of communication revealed to prophets by angels, an antediluvian tongue first spoken by Adam and preserved by ancient giants following the confusion of languages at Babel. Such 12 Qur’an 12:2: ʔinnā ʔanzalnā-h qurānā ʕarabiyyā laʕallakum taʕqilūn (in reconstructed Old Ḥigāzī). 20 MHGA, v. 2020-1 accounts of Arabic’s origins are diverse and have not been comprehensively studied. Retsö (2003: 34-53) does an excellent job of assembling many of these tales and it is worth mentioning a few here: ʔaḫbār al-Yaman, attributed to ʕubayd b. Shariyah, the tutor of Muʕāwiyah b. ʔAbī Sufyān. Muʕāwiyah said: 'Who are the ʕarab ʕāribah (primordial Arabs) and who are the ʕarab mustaʕribah (new Arabs).' He (ʕubayd) said: 'O Muʕāwiyah do you and other learned men not know that they are ʕād, Ṯamūd, Ṭasm, Ǧadīs, ʔiram, al-ʕamāliqah, Ǧurhum, and Qaḥṭān son of Hūd? They were the first peoples and among them was Yaʕrub who spoke Arabic. Everyone took it from Yaʕrub son of Qaḥṭān son of Hūd, and Arabic is traced back to him. It is said [to be] ʕarab because Yaʕrub was the first who spoke it and no one else spoke it before him.' ʔibrāhīm carried his son ʔismāʕīl from his land and made him settle in Makka, and we, Ǧurhum, were the people of the Holy Land. Then ʔismāʕīl grew up among us and spoke Arabic and married among us. All the sons of ʔismāʕīl are from the daughter of Muḍāḍ, son of ʕamr, the Ǧurhumite, and the father of ʔismāʕīl is from us, and you, O Qurayš, are from us and the ʕarab are [descended] from each other. (Trans. Retsö 2003: 34)13 Retsö convincingly argues that the ʔaḫbār was probably written in the early Abbasid period and reflects Yemeni claims to Arab identity and authenticity in that period. The antiquarian Hišām b. al-Kalbī (d. 819 CE) provides us with a very different account of Arabic origins. He claims that the first to speak Arabic were the ʕamālīq, a race of ancient giants. It is said that the ʕamālīq were the first to speak Arabic when they travelled from Babylon (following the flood) and they and Ǧurhum were called the ʕarab ʕāribah ... God made ʕād, ʕābil, Ṯamūd, Ǧadīs, ʕimlīq, Ṭasm, ʔumaym and the sons of Yaqṭan understand Arabic. (Trans. Retsö 2003: 35) Other accounts directly connect Arabic with revelation and prophecy. Wahb b. Munabbih, the 7th-8th c. Yemenite storyteller, claimed that god had sent down Arabic and all 29 letters of its script to the prophet Hūd (Retsö 2003: 40). Retsö does not translate the term ʕarabiyyah based on his views of its ritualistic role, but I have given it as Arabic as I see it simply as the name the Arabic language - in all of its forms - in the literary register. Words in parenthesis are my additions. 13 21 MHGA, v. 2020-1 Kaʕb al-ʔaḥbār, the 7th c. Yemenite storyteller, firmly places Arabic’s origin in the heavens and at the beginning of history. He claimed that the first to speak Arabic was Gabriel (the angel) and he was the one who taught it to Noah, and Noah taught it to Shem. In another report, Kaʕb claims that Adam was the first to establish Arabic and Syriac writing and that he spoke all languages. A popular theme in Arabic origin stories is to situate the language either before or in the direct aftermath of the Great Flood. Ibn Al-Kalbī stated on the authority of his father and others that Arabic was spoken in the days that the tongues were confused in Babylon in the time of Nimrod son of Canaan son of Kush son of Ham son of Noah. Ibn ʕabbās claimed there were 80 people with Noah on the Ark, one of them being Ǧurhum, the offspring of Shem. The antediluvian language survived on the Ark and Ǧurhum took it to the Ḥigāz when he and his tribe settled in Mecca. There they encountered Ishmael and his mother, and this sets the stage for the transmission of Arabic to the Abrahamic line, and ultimately to Muhammad. Another stream of traditions places the origin of Arabic with Ishmael. In several sayings attributed to Muhammad, it is Ishmael who was the first to speak Arabic fluently (naṭaqa), and sometimes “clear Arabic” (ʕarabiyyah mubīnah) at the age of 14. Accounts differ in terms of how Ishmael acquired the tongue, whether from the Ǧurhumites or from God himself. With so many conflicting accounts and so many “first Arabics”, naturally there were traditions to “rank” these competing primordial varieties. Ibn al-Kalbī and al-Šuʔaqī son of Quṭāmā claim: “the first to speak in Arabic was Yaʕrub son of Qaḥṭān and his Arabic was purer (ʔafṣaḥ) than the older language, the Arabic of ʕād and Ṯamūd and the ʕamālīq and Ṭasm and Ǧadīs and the sons of Yaqṭan son of ʕeber and Ǧurhum son of ʕeber son of Sabaʔ son of Yaqṭan. Others claim that the Arabic of Ishmael and Maʕadd son of ʕadnān was purer -- this view is in line with the aforementioned prophetic traditions. The foregone accounts are just a small sample of the varied traditions and opinions that exist about Arabic’s origins in medieval Islamic works. The major categories are clear. The first encompasses traditions that place Arabic’s origin with Yaʕrub. These reflect the construction of a Yemeni-centric past by the Yemeni factions of the Abbasid 22 MHGA, v. 2020-1 period. Claims to the original Arabic, and as such, to being the primordial Arabs (ʕarab ʕāribah) no doubt reflect an attempt to bolster their political and cultural claims to Arab leadership, which were not widely accepted in this period. The second trend uses Arabic as a bridge to connect Arabian legends with the epic biblical past -- the presence of the eponymous ancestors of Arabian tribes at events like the Flood and Babel explain the transmission of this antediluvian tongue to the historical period, and carves out a place for Arabian folklore within the larger biblical frame of storytelling. Finally, the third trend seeks to underscore the sacred nature of Arabic by connecting it important patriarchs like Ishmael. The fact that the language was ‘revealed’ rather than naturally acquired further highlights the uniqueness of the Qur’an and its inimitability, and the central role played by the Arabs in the unfolding of sacred history. 0.3 The Arabic language family 0.3.1 Divisions of Old Arabic Old Arabic: This term refers to the sum of evidence attested before the rise of Islam in documentary sources such as epigraphy and papyri, terminating with the Qur’anic Consonantal Text. It does not encompass the material gathered by the Arab grammarians in the 8th and 9th century, nor does it cover the language of the Arabic odes (Qasidah) attributed to pre-Islamic times. By focusing on documentary evidence from the pre-Islamic period, we can be sure that the language was not filtered by later, prescriptive grammatical norms. Indeed, the Arabic recorded in these sources is rather distinct from the materials found in later Arabic grammatical writings, attesting to the importance of an evidence-based Old Arabic. Pre-Historic Old Arabic The Arabic of the early first millennium BCE has few literary witnesses. It is known from a handful of transcriptions of anthroponyms in cuneiform documents and a small number of Ancient North Arabian epigraphic texts. This stage of the language was spoken by Arab tribes of North Arabia and Syria known primarily from external sources. Its character is mostly reconstructed based on later stages of the language and the fragmentary evidence mentioned above. 23 MHGA, v. 2020-1 Early 1st millennium BCE Ancient North Arabian inscription from Bāyir, Jordan (Hayajneh et al. 2015) Northern Old Arabic dialect continuum By the middle of the 1st millennium BCE, a dialect continuum of Old Arabic stretched from the southern Levant to the northern Higaz, and perhaps as far east as Dumah. The sources for this continuum are uneven and fragmentary. There are no linguistic features that suggest these forms of Arabic constitute a genetic sub-grouping. Rather, the continuum appears to develop directly from Proto-Arabic without any clear branching. The following paragraphs will briefly outline their documentation. Safaitic: These texts span the Syro-Arabian basalt desert, the Ḥarrah. Some fortythousand inscriptions are known so far, a number that continues to grow each year. The chronological limits of this material is unclear. The earliest datable texts perhaps go back to the 3rd c. BCE while the latest are dated to the 3rd c. CE, but the vast majority of texts are undatable and so may stretch back much further in time. The Safaitic texts are highly formulaic, and while the majority comprise only personal names, several thousands of texts contain narrative prose and ritualistic language, which, when taken together, shed clear light on the dialects of Arabic of this region. 24 MHGA, v. 2020-1 Safaitic inscription from NE Jordan (Al-Jallad 2017b) Hismaic: The Hismaic inscriptions range from the area of Madaba in central Jordan to northwest Arabia, around Tabuk. The published corpus consists of around 3700 texts, most of which contain only personal names and short phrases. A few longer inscriptions are known from the Madaba region and these reveal a language strikingly similar to Classical Arabic, both in terms of grammar and stylistics. Hismaic Votive inscription from Madaba region (Zwettler and Graf 2004) 25 MHGA, v. 2020-1 Nabataean: The Nabataean dialect of Arabic is known primarily through the personal names attested in Nabataean Aramaic, but in the Classical Nabataean period, only one text in the Arabic language is carved in this script, the En Avdat inscription (see appendix). The Nabataean inscriptions are concentrated in the Nabataean kingdom, in northwest Arabia and the southern Levant. Stray texts can be found elsewhere, as far south as Yemen. After the fall of Nabataea, more Arabic elements appear in Nabataean Aramaic inscriptions, and two more near complete Arabic texts are known, JSNab 17 and the Namarah inscription. These texts provide our clearest glimpse of the western dialects of Arabic. The latest witness to Nabataean Arabic is the Petra Papryi and the Nessana Papyri. These 6th c. CE Greek-language documents contain the final attestations of Nabataean Arabic in the form of transcriptions of toponyms, oikonyms, and personal names. The En Avdat inscription (Kropp 2017) - Nabataeo-Arabic inscriptions: Between the 3rd and 5th centuries CE, the Nabataean script begins to exhibit “evolved” letter shapes on the path towards the Arabic script. The language of these inscriptions is mixed: the formulaic components are in Aramaic while other elements are in Arabic. The short texts, however, do not provide the opportunity to diagnose fully their language, but they appear to agree with Nabataean Arabic in all respects. - Late 5th and 6th c. Arabic-script inscriptions: By the late 400s, the Nabataean script had given rise to the Arabic script we know today. The language of the earliest texts 26 MHGA, v. 2020-1 in this script, however, remains similar to its Nabataeo-Arabic predecessor. The inscriptions are essentially composed in Arabic with Aramaic formularies. These texts exhibit a degree of linguistic heterogeneity, suggesting that there was no unified tradition of writing Arabic. I provisionally place these under the ‘northern Old Arabic dialect continuum’ assuming that they continue Nabataean Arabic, until further evidence suggests otherwise. Ḥarrān inscription, southern Syria (Fiema et al. 2015) Graeco-Arabica: A major source for northern Old Arabic is the copious amounts of Arabic personal names and vocabulary in Greek transcription. The onomastic materially is studied comprehensively in Al-Jallad (2017a). A small number of SafaiticGreek bilingual inscriptions are known (Al-Jallad and al-Manaser 2016; Al-Jallad forthcoming) and one completely Arabic text composed in Greek letters has been published (Al-Jallad and al-Manaser 2015). 27 MHGA, v. 2020-1 Graeco-Arabic inscription (=A1, Al-Jallad and al-Manaser 2015) Old Ḥigāzī The first clear branch of Arabic is Old Ḥigāzī, a term referring to the ancient dialects of the northern and perhaps central Ḥigāzī. This group is characterized by a few linguistic innovations, including the use of a new relative pronoun series based on the Central Semitic portmanteau demonstrative *hallaḏī (Huehnergard 1995), producing Arabic ʔallaḏī, etc. Another innovation is the replacement of the infinitive as a verbal complement with a subordinated verb, usually introduced by ʔan. In the Dadanitic script: The earliest attestations of Old Ḥigāzī occur in the inscriptions in the area of ancient Dadān (present-day Al-Ula), an oasis near Hegra (Madain Saleh). While these texts are written in the Dadanitic script and language, distinct from Arabic, elements of Old Ḥigāzī appear in some inscriptions, suggesting that some of the population spoke this variety of Arabic. The most salient features are the relative pronoun ʾlt /ʔallatī/ and the ʔan yafʕala construction. 28 MHGA, v. 2020-1 JSLih 384, courtesy OCIANA Qur’anic Consonantal Text: The earliest Qur’anic manuscripts are dated to the latter half of the 7th century, and as such they are not strictly pre-Islamic. Nevertheless, their language and orthography differs in important ways from later norms, indicating that they continue a pre-Islamic tradition. The QCT signifies the language of the Qur’anic text itself and not the reading traditions imposed upon it. Several studies of the rasm, the textual skeleton, have shed important light on its linguistic character, revealing a dialect rather distinct from Classical Arabic. The presence of the relative pronoun allaḏī along with the ʔan yafʕal construction indicate that the language of the QCT belongs to the same linguistic stratum as the Old Ḥigāzī of the Dadanitic inscriptions, both distinct from the northern Old Arabic dialect continuum, in which these features are unattested. Marginal Arabic Elements of Old Arabic can be found on the periphery of Yemen in pre-Islamic times. At Qaryat al-Fāw and Nagrān, a small number of texts exhibiting Arabic features embedded within Ancient South Arabian are known. These could reflect peripheral, transitional dialects between Arabic and Sabaic or, perhaps, texts commissioned by Nabataean colonialists, whose presence is supported by ever-increasing archaeological evidence, and whose vernacular likely colored the inscriptions. A number of penitential Middle Sabaic inscriptions from the Haram region of the Yemeni Jawf also exhibit an admixture from a non-Sabaic/Ancient South Arabian 29 MHGA, v. 2020-1 language and some features, such as past tense negation with lam (lm yġts¹l ‘he did not wash’) are isoglosses of Arabic. The exact linguistic environment responsible for the emergence of these inscriptional hybrids is unclear for the moment but several hypotheses have been put forward - see Macdonald (2000) and Al-Jallad (2018b: 2829) for a detailed discussion. Haram 34, an example of a mixed register inscription with elements of Sabaic and North Arabian (Image, courtsey DASI https://dasi.cnr.it/index.php?id=30&prjId=1&corId=0&colId=0&navId=683582184&recId =4013&mark=04013%2C005%2C001) 0.3.2 Pre-Modern Islamic period The Psalm Fragment: This text, an Arabic translation/gloss of Psalm 78 in Greek letters, is perhaps the earliest fully vocalized Arabic document from the Islamic period. I have argued that its language reflects the latest stage of Old Ḥigāzī. While the text is 30 MHGA, v. 2020-1 undated, I would suggest placing its production somewhere in the 9th century, possibly as early as the late 8th. The editio princeps is Violet (1901); see a forthcoming monograph on the document by Al-Jallad (forthcoming). Damascus Psalm Fragment (Mss simulata orientalia 6; STAATSBIBLIOTHEK ZU BERLIN - Preußischer Kulturbesitz, Orient-abteilung). Papyri of the 1st Islamic Century: These documents pre-date the prescriptive specter of Classical Arabic, although they are often edited as if that register was intended. The texts, I would suggest, basically reflect the same language as the Psalm Fragment, and attest the latest stage of Old Ḥigāzī. One, however, must caution against treating the entire corpus as a homogenous unit, as linguistic features from other registers of Arabic permeate these documents in later periods. On these documents, including examples of Greek transcriptions, see Hopkins 1984; Al-Jallad 2017c; Isserlin 1969; Kaplony 2015. The language(s) of the Qaṣīdah: One of the common linguistic features uniting the Old Arabic sources is the absence of nunation, tanwīn. This feature, so characteristic of Classical Arabic, is attested first in the corpus of rhymed and metered poems attributed to the pre-Islamic period by Muslim scholars. Tanwīn is an ancient feature (see 2.3.1), cognate with mimation in Akkadian and Ancient South Arabian, although its realization with a n seems to be unique to Arabic. Its absence in the northern dialect 31 MHGA, v. 2020-1 continuum should therefore be understood as a loss, perhaps an areal development. So then, how did an archaic dialect of Arabic, preserving this ancient grammatical ending, survive until the Islamic period, all the while bypassing attestation in the epigraphic record? While the language of the pre-Islamic odes is not uniform, and poets were certainly free to draw on forms foreign to their vernacular for metrical purposes, these texts do exhibit the same innovations that characterize Old Ḥigāzī. I would therefore suggest that the language of the Odes is a descendent of Old Ḥigāzī, but splitting off in the prehistoric period, following the innovation of its characteristic features but before the loss of nunation. Since the tradition of composing the ancient Odes seems to have been localized to South Central Arabia, a place where non-Arabic languages are attested in ancient times. If Arabic was introduced into this region around the turn of the Era, then the linguistic evidence suggests that it was from the southern Ḥigāzī. 0.3.3 Literary Varieties Classical Arabic: Classical Arabic is a vague umbrella term used to cover a wide variety of sources, most often the language documented by the Arabic Grammarians, the reading traditions of the Qur’an, the pre-Islamic Odes, and texts written in the Islamic period. These sources are not homogenous and can vary significantly over time and place. As such Classical Arabic is not a single variety of the Arabic language but should rather be construed as a blanket definition covering what is prescriptively possible in written Arabic in pre-modern times. Middle Arabic: Middle Arabic is a scholarly term covering texts produced in premodern times that contain deviations from the perspective norms of Classical Arabic. This term covers what is clearly register mixing, as one encounters in manuscripts of the 1001 nights, to true dialectal texts, as one often finds in the vocalized and unvocalized Judaeo-Arabic documents from the Cairo Geniza. An honest examination of the written documents from pre-modern times suggests that a far greater amount of texts than what is usually assumed fall into these categories. For an excellent description of the state of the art in Middle Arabic studies, see Khan (2011) and the references there. 32 MHGA, v. 2020-1 0.3.4 Modern Vernaculars Depending on how one counts, there are dozens of distinct dialects of vernacular Arabic spoken today across the Middle East and North Africa. Since early Islamic times, vernacular Arabic has always been seen through the lens of the written register, the Classical Arabic varieties. Pre-modern scholars and many modern ones as well understood the vernaculars to be corrupted forms of Classical Arabic. The differences between the two were usually explained through the process of imperfect language acquisition or the corrosive effects of language contact (see the classical discussion in Versteegh 1997). More than a century of research on the modern dialects has soundly disproven this line of development. The modern vernaculars do not constitute a homogeneous mass, descending monogenetically from Classical Arabic, but nor do they reflect, as a whole, a linear development from pre-Islamic varieties implanted across the Middle East and North Africa following the conquests. The story of the modern dialects is one of contact and convergence. The spread of Arabic did not happen only one time during the initial Arab Conquests of the 7th century. The first dialects implanted during this period lie buried under waves of later Arabics, all converging in different ways with each other. Ancient forms of Arabic, such as those attested in the northern Old Arabic dialect continuum and Old Ḥigāzī mix with later innovations that emerged in the medieval period. In addition to this, Classical Arabic casts its distinct shadow over this process for over a millennium, and influenced the development of the dialects just as much as it did other Islamicate languages. While the effects of Classical Arabic on, say, Persian are rather obvious, it is sometimes more difficult to distinguish intrusions from the literary language in the modern dialects, except for the latest phase of contact where such loans tend to have distinct phonological characteristics. While most of the familiar modern dialects (i.e. Rabat, Cairo, Damascus, etc.) are sedimentary structures, containing layers of Arabics that must be teased out on a caseby-case basis, the dialects of the periphery, i.e. rural areas (rural Palestinian) and Arabic islands in non-Arabic speaking areas (Anatolian Arabic, Maltese, etc.), preserve snapshots of older linguistic situations. 33 MHGA, v. 2020-1 Many dialects of the Arabian Peninsula have avoided the momentum of convergence that has affected dialects of urban centers and those spread after the conquests. The dialects of the Najd, for example, appear to reflect an independent strand of Arabic, closely related to the language of the Qasidah. While certainly in contact and influenced by Classical Arabic and other varieties, there does not seem to be evidence for the introduction of new varieties of Arabic to this region en mass following the Arab Conquests. In southwest Arabia, some varieties appear to descend directly from Proto-Arabic rather than through the medium of Old Ḥigāzī and have, overall, not converged with major strands of modern Arabic, such as the Rigāl Almaʿ or the Fayfi vernacular. Some of the vernaculars of this region have also converged with Ancient South Arabian, most likely Late Sabaic. The influence of the latter can be heard in major points of grammar such as the pronominal suffixes of the past tense verb, negation, basic vocabulary, and more. For example: some dialects have taken over grammatical features characteristic of those languages, such as a -k endings in the suffix conjugation: katabk ‘I wrote’ rather than katabt(u), cf. Gəʿəz katabku, Sabaic sṭrk /saṭarku/. Modern Arabic is most often classified based on geography according to five general zones: Mesopotamia, Arabia, Levant, Egypt/Sudan, and the Maghreb. For an excellent overview of the features of the modern vernaculars, see Holes (ed. 2018) and the classic handbook edited by W. Fischer and O. Jastrow (1980). For a brief outline of the key features of the modern vernaculars, see the chapters on Arabic in Weninger, ed. 2011. See this link for a rough outline on the distribution of the main Arabic dialect clusters: https://en.wikipedia.org/wiki/Varieties_of_Arabic#/media/File:Arabic_Dialects.svg. 34 MHGA, v. 2020-1 I Phonology 1.1 Proto-Arabic consonants and vowels Bila Labiovelar Interdental Dental/ Palatal Velar Bial Alveolar Voiceless p [ph] ~ [f] t [th] k [kh] Voiced b [b] d [d] g [g] ṭ [tˁ] q [q] Pharyngeal Glottal Stop Emphatic ʾ [ʔ] Fricative Voiceless ṯ [θ] ḫ [x] ḥ [ħ] Voiced ḏ [ð] ġ [ɣ] ʿ [ʕ] Emphatic ẓ = ṯ̣ [tθˁ] h [h] Sibilant Voiceless s1 = s [s] Voiced z [z] Emphatic ṣ [tsˁ] Approx. w [w] Trill y [j] r [r] Lateral Voiceless s² = ś [ɬ] Voiced l [l] Emphatic ḍ = ṣ́ [ɬˁ] Nasal m [m] n [n] The reconstruction of these values is justified in the discussion in 1.4. It is important to note here that the reconstruction of pharyngealization for the Proto-Arabic emphatics is uncertain. 35 MHGA, v. 2020-1 Vowels short vowels: *a, *u, *i It is very likely, but impossible to prove, that the short vowels had phonetic allophones at the proto-Arabic stage. The realization of *a may have ranged from [ɔ] to [æ], as in many forms of Arabic. *i may have been realized as [i] and [e] and *u as [u] and [o]. long vowels: *ā [aː], ū [uː], ī [iː] There is no evidence to suggest that *ā had conditioned allophones at the Proto-Arabic stage. The northern Old Arabic dialects realize this phoneme as [aː] in all environments. diphthongs: *aw [au], *ay [ai] 1.2 Proto-Arabic sound changes Proto-Arabic phonology is considerably conservative, and only a few sound changes distinguish the language from Proto-Semitic: 0) *s > h at word boundaries: *suʔa > huwa; this rule is blocked in most verbal and nominal roots on account of leveling one value across the paradigm; *samiʕa > hamiʕa but *yasmaʕ > yasmaʕ (no change), and so the s is restored to the suffix conjugation. 1) Merger of *s³ [ts] and *s¹ [s] to [s]; deaffrication of *z [zd] > [z] Proto-Semitic Classical Arabic Sabaic self *napsum nafsun nfs¹m ten *ʕaɬarum ʕašarun ʿs²rm garment *kitswatum kiswatun ks³wtm 2) *ah > ā / _# The scope of this rule is relatively small because the case endings followed most nominal III-h stems, and the jussive of III-h roots would have been paradigmatically restored based on other members of the paradigm. It applies mainly to the interrogative mā < *mah, cf. Ug mh and it in non-word final position, mahmā ‘whatever’ and perhaps the terminative ending, *ah > ā. 3) *w > y / i_ *raṣ́iwa > *raṣ́iya, but *riṣ́wānu ‘to be satisfied’ 36 MHGA, v. 2020-1 4) collapse of triphthongs in some environments14 *iGi/u > ī *yaśkiyu > yaśkī *uwu > ū *yadʕuwu > yadʕū *aGū > aw *daʕawū > daʕaw *aGī > ay *tarṣ́awī > tarṣ́ay 5) *h > ʔ #_vCC ́ *hapʕála > *ʔaphʕála *hinna > ʔinna *han- > *ʔan1.2.1 Possible sound changes *p > f This change is found in all of the modern dialects and is described by Sibawayh for classical Arabic. However, there is some evidence to suggest that the phoneme remained [ph] in Proto-Arabic. In Safaitic, both Greek [p] and [ph] are represented by the f glyph and never b, suggesting that f likely signified a stop rather than a fricative. Transcriptions of Old Arabic names in Greek sometimes represent the reflex of Arabic *p with Greek Pi: Χαλιπος = /ḫalīp/, Classical Arabic ḫalīf-.15 It is also possible that [f] was already an allophone of *p at the Proto-Arabic stage, before being levelled to all environments in later varieties. Glottalization > pharyngealization The emphatic correlate of Proto-Arabic is unclear. Nearly all of the modern dialects exhibit pharyngealization, but this does not imply that the feature is Proto-Arabic. The ancient evidence is ambiguous and two features could suggest that glottalization remained in the earliest stages of Arabic: (1) the emphatic series is unvoiced and (2) the emphatics do not affect vowel quality. This evidence is, however, circumstantial and it is equally possible that pharyngealization set it at the Proto-Arabic stage without affecting other features of pronunciation. 14 G = glide, w/y; on the history of the triphthongs in Arabic, see Van Putten 2017. 15 For a more detailed discussion, see Al-Jallad 2017a 37 MHGA, v. 2020-1 1.2.1 Northern Old Arabic 1) Nunation is lost 2) The high vowels are realized slightly lower, *i as [e] and *u as [o]. 3) In Safaitic, final short high vowels, *u and *i, are eventually lost. 4) In the Nabataean dialect, it seems that word final *ayV has shifted to [æː] or [aː]: dwšrʾ = Δουσαρη(ς) [ḏū-śaræ] and ṣfʾ [ṣafā] ‘clear’ < *ṣafaw. 5) Unstressed *u becomes /i/ when contiguous with y, *tuyaym > tiyaym.16 6) Irregular assimilation of n to a following consonant, especially in unstressed environments - Safaitic tẓr [ettaṯara] ‘to await’ <* intaṯara. ̣ ̣ 7) Pharyngealization: the rendering of Proto-Arabic *ṣ́ and possibly *ṯ ̣ with Greek Zeta in the Petra Papyri suggests that the emphatic correlate was pharyngealization in 6th c. Petra. Pharyngealization may have set in earlier. In a Palmyrene graffito from the Ḥarrah (unpublished), the Arabic word saṭar- ‘line, writing’ is rendered in Palmyrene as ṣṭrʾ, suggesting emphatic spreading in the Arabic original. If the ṭ were realized as an ejective, then the initial sibilant would not have been affected and the author of this graffito could have rendered it faithfully with Samek or Šin. 1.2.2 “Conventional” Classical Arabic sound changes From Proto-Arabic, the following sound changes are required to produce the standard pronunciation of Classical Arabic 1) eventual deaffrication of ṣ and possible pharyngealization [tsˁ] > [sˁ] 2) *p > [f] 3) Deaffrication and voicing of *ṯ ̣ [tθˁ] to [ðˁ] and *ṣ́ [ɬˁ] to [ɮˤ]; palatalization of *g [g] > [ɟ] (voiced palatal stop) and ultimately to palato-alveolar affricate [ʤ]; shift of *ś [ɬ] > [ç] and eventually š [ʃ]. 4) Spread of emphasis: *iṣ́taraba > iḍṭaraba 5) Collapse of triphthongs17 16 ibid. See Van Putten 2017. Also, note that Sibawayh describes varieties where áGi yields ē, *ḫawifa > ḫēfa; mawita > mēta but *qawula > qāla. 17 38 MHGA, v. 2020-1 *aya and *awa > ā *banaya > banā *daʕawa > daʕā *suphlayu > suflā *áGi/u > ā *qáwuma > qāma *aGí/ú > i/u *qawúmtu > qumtu *nawímtu > nimtu 6) y/w > ʔ / ā_18 *samāyun > samaʔun ‘sky’, ‘rain’ 7) Emergence of front/back allophones of the vowels, *a becomes [æ] but [ɒ] in backed environments, [ɟæmiːlun] vs. [tˁɒriːqun] *i becomes [i] and [e] and *u [u] and [o]. 7) ʔaʔ. > ʔā *ʔaʔkulu > ʔākulu, against Safaitic ʾʾmr [ʔaʔmar] 8) Emergence of CC clusters from some biradical roots19 *binun > (i)bnun; *ṯinun > (i)ṯnun 9) Vowel Harmony across h in unstressed clitic: This very specific sound rule becomes part of conventional Classical Arabic but was not universal in the sources described by the grammarians. Basically, it states that the vowel of the 3rd person pronouns -hu, hum and hunna harmonizes with a preceding i-class vowel or y, so fī kitābi-him ‘in their book’ from *fī kitābi-hum and ʕalay-hi from *ʕalay-hu. Pausal Rules 9) Movement of stress to the penultimate syllable of an utterance 10) Loss of un/in syllable after the sentential stress (perhaps first becoming a nasalized vowel): ḏahaba ʔilā miṣra záyd < *ḏahaba ʾilā miṣra zaydun 11) an > ā after the sentential stress ḍaraba ʕamrun záydā 12) at > ah in utterance final position raʔaytu fāṭimah < *raʔaytu fāṭimat < *raʔaytu fāṭimata 18 The glide of the L-stem, qāwala, yuqāwilu is restored analogically. 19 For the reconstruction of these forms with a syllabic resonant, e.g. *bṇum, see Testen 2017. 39 MHGA, v. 2020-1 1.2.3 Old Ḥigāzi (QCT) The following rules describe sound changes that distinguish this dialect from conventional classical Arabic. Rules 2, 4, most of 5 (exceptions in 3 below), 6, 7, and 8 appear to apply to the language of the QCT as well, in so far as we can detect from the orthography. 1) penultimate (historically antepenultimate) stress (see below; Al-Jallad 2017) 2) loss of tanwīn and final short vowels - but perhaps these survive in liaison positions, e.g. before the article. 3) Collapse of triphthongs *aya > ā : *banaya > banā *awa/awá > ā : dáʕawa > daʕā; ṣalawátuk > ṣalātuk (orth. ṣlʾtk) *áwa > ō : nagáwat > naǵōh (cf. Classical Arabic naǧāh, orth. ngwh) 4) Partial loss of the glottal stop; see Van Putten 2018., resulting in a. homo-organic glide: ruʔūs > ruwūs; ḫāṭiʔah > ḫāṭiyah b. loss post-consonantally: yasʔal > yasal c. lengthen preceding vowel: śaʔn > śān; biʔs > bīs; muʔmin > mūmin 5) loss of first ta syllable in a word-initial tata sequence *tataraḥḥamu > taraḥḥamu (cf. Hismaic trḥm = taraḥḥamu) 1.2.4 The Tamīmī dialects of the Grammarians This provides a summary of sound changes attributed to the eastern Tamīmī dialects by the Arabic Grammarians. 1) a > i / _L (=larynageal) i or ī: baʕīrun > biʕīrun; ḍaḥiqa > ḍiḥiqa > ḍiḥqa 2) u > i / ih_ : dāri-hum > dāri-him; dāri-hu > dāri-hi (this affects mainly the pronominal suffixes) 3) i > 0 / unstressed open syllable: munṭaliqun > munṭalqun; faʕila > ḟaʕla (fiʕla! see above). 4) *úwi > ǖ : *qúwila > qǖla ‘it was said’, cf. QCT qyl /qīl/ 40 MHGA, v. 2020-1 1.4 Excursus on some key consonants 1.4.1 Notes on the Sibilants The status of the Old Arabic sibilants was first subjected to close examination in A.F.L. Beeston’s 1962 paper, “Arabian Sibilants”. His reading of Sibawayh suggested to him that the value of ‫ س‬in the Arabic of the 8th c. CE and earlier was [∫]: “The other sibilant, present in the “garment” and “soul” words, [reflexes of *s³ and *s¹, respectively (my insertion)], is described by Sibawaihi as having its point of closure between the tongue-tip and the hard palate a little behind the teeth; while this description may be regarded as not wholly inconsistent with some variety of [s] sound, it is far more probable that what he is here describing is a [∫].” (Beeston 1962: 244) Before discussing Beeston’s position let us first examine Sibawayh’s exact statement: َ ‫وم َّما بين‬ ‫ط َرف اللسان وفُ َويْقَ الثَنا َيا ُم ْخ َر ُج الزاى والسين والصاد‬ “And between the tip of the tongue and a little bit above the incisors is the point of articulation of the ‫ س‬,‫ز‬, and ‫”ص‬ While Sibawayh’s “a little above the incisors” could in theory describe a palato-alveolar articulation, here it is important to consider which other sounds occupy the same point of articulation. If Sibawayh intended a [∫] for ‫س‬, then it would also follow that his ‫ ز‬was a [ʒ] and his ‫ ص‬was a [∫ˁ]. There is no evidence for such realizations at any period in the history of Arabic, or in other Semitic languages. Thus, we must accept Sibawayh’s description as referring to an alveolar sibilant as it regards the reflexes of ‫ ز‬and ‫ص‬, and so it is unclear as to why the same phrase must describe a palato-alveolar sibilant in the case of ‫س‬. The obvious answer is that it does not. Since 1951, our picture of the Proto-Semitic sibilants has sharpened and it is now generally held that the three non-emphatic “sibilants” were actually realized as follows:20 20 On the reconstruction of the sibilants, see Kogan 2011 and the references there. 41 MHGA, v. 2020-1 Proto- Semitic [s] = *s¹ [ɬ] = *s² [t͡s] *s³ Based on this reconstruction, the plain [s] of Arabic does not represent a shift from [∫] > [s] but rather the preservation of the original value of the phoneme. This of course begs the question as to why these sounds were sometimes confused in Nabataean and Palmyrene inscriptions and why early loans from NWS containing š were borrowed into Arabic with ‫س‬. The answer is complex and must be dealt with following a discussion of *s². There is little doubt that the phoneme signified by the glyph ‫ ش‬goes back to a voiceless lateral fricative in Proto-Semitic, [ɬ]. This value, however, was unknown to Sibawayh. The Ḍād was considered unique in terms of its lateral point of articulation, which suggests that the ‫ ش‬was no longer its unemphatic counterpart. Sibawayh’s description of the point of articulation of the ‫ ش‬along with the other palatals strongly suggests that it was realized as a voiceless palatal fricative, [ç].21 This realization, however, seems to have been unique to Sibawayh’s Arabic, and is certainly not attested in the preIslamic material or even contemporary transcriptions of Arabic into other languages. There is a chain of evidence which suggest that the true lateral value of this sound obtained in Old Arabic. The first is the name of the Nabataean deity, Dusares. The name is written in several forms across several scripts, but the etymological form appears to be ḏū-śaray, meaning ‘he of the Śaray mountains’, and may in fact be an epithet of the Edomite deity Qōs. In any case, the relative-determinative pronoun is clearly Arabic,22 and the second term, whether of Edomite origin or Arabic itself, reflects In Beeston’s terms, the ‫“ ش‬cannot be interpreted as indicating anything else than an approximation to the German “Ich-Laut” (1962:224). 22 In fact, it is identical to its Proto-West Semitic value, but considering that the etymological interdental was long lost in the NWS, the most likely candidate for the production of this epithet is in fact Arabic. 21 42 MHGA, v. 2020-1 an etymological lateral. The term is consistently written in the Nabataean script as ‫דושרא‬, which conceals the etymological value of the sound as the etymological lateral and alveolar sibilant were written with Nabataean ‫ש‬. Two important pieces of evidence, however, suggest that the value of this letter was not a sibilant, neither [s] nor [∫]. Macdonald pointed out in several places that the value of s² in Safaitic could not have been the same as modern Arabic [∫], as the glyph was never used to transcribe Aramaic š = [∫] (Macdonald 2000, 2004). For this, Safaitic always uses its s¹. At the same time, Safaitic uses the s² glyph to transcribe the name ḏū-śaray. Now one could still argue that the value of both Safaitic and Nabataean s² was in fact [ç], which would be distinct enough from Aramaic [∫] to preclude its use for the transcription of this sound. The argument against this view is that the reflex of the lateral is always given with σ in transcriptions of Arabic names. This contrasts with the representation of etymological *ḫ, which is more often than not represented with the spiritus asper (ᴓ in transcription). The value of *ḫ was a front velar fricative, [x]. Had the reflex of *s² been a palatal fricative, which is just one point further forward, we would expect that at least in some cases it would have been given with zero or perhaps on occasion χ. The fact that this is not the case combined with its non-use for NWS [∫] strongly suggests that the sound remained a lateral. Given this, it is curious why the sound is never represented by a digraph λσ as found later in the transcription of Hebrew sīn (NWS *baśam > Eng balsam). It would seem that the voiceless alveolar lateral fricative sounded close enough to Greek [s] to the ear of Near Eastern scribes to not warrant the use of a digraph. In general, there appears to be an aversion to the use of digraphs in the transcriptions of Semitic names in Near Eastern Greek, where as the practice is rather common in Egyptian documents. With this established, we are brought full circle back to the realization of *s¹. I have argued in many places that the use of s¹ for Northwest Semitic šin simply indicates that s¹ was its closest approximation. With the establishment of s² as [ɬ] it becomes clear that s¹ was the only true, plain sibilant in the language. This, however, tells us nothing about its phonetic realization. If Old Arabic *s¹ were in fact [∫], then that would mean 43 MHGA, v. 2020-1 the plain alveolar sibilant [s] did not exist in the language. This is uneconomical since all later stages of Arabic preserve the [s] value of this sound. Such a reconstruction would therefore posit the following chain [s] > [∫] > [s]. However, were the sound realized as a simple [s], it would then be difficult to explain its rendering in Aramaic with both ‫ ש‬and ‫ס‬. Two possible explanations come to mind. The first is that Arabic *s¹ was not quite a plain alveolar sibilant [s] but rather an apical [s̺], similar to Modern Greek or Amsterdam Dutch. This pronunciation is typical of languages with only a single sibilant, and so would be expected of an Arabic where *s² was a lateral. While such an explanation would work, there is perhaps another aspect of “transcription” that has been overlooked by previous scholars. The Aramaic of the Nabataean and Palmyrene inscriptions is a form of Official Aramaic, the administrative variety of the Achaemenid Empire. While Nabataean betrays the influence of substrate from both Arabic and Western Aramaic, Nabataean Aramaic, as it was written, was certainly not the mother tongue of anybody in the Nabataean realm. On the occasion that the language was actually spoken, an artificial learned pronunciation must have accompanied it. If the authors of the Nabataean inscriptions were in fact speakers of Arabic, as it is now clear, the question is - would those who used Official Aramaic as a written language have pronounced ‫ ש‬as [∫], a non-existent sound in their vernacular, when they read the language aloud? The answer I think, based on analogy with the use of Arabic as a literary language in Turkey and Iran, for example, is no. Scribes of those languages pronounced Arabic‫ ض‬,‫ ظ‬,‫ ذ‬and ‫ ز‬all as /z/, and used them with some variation to spell Iranian or Turkish words with /z/. In this case, it is probable that Arabic-speaking scribes pronounced Aramaic ‫ ש‬and ‫ ס‬as [s], and so both were used with some variation in the rendering of Arabic names. The higher distribution of ‫ ש‬may be due to the sound’s overall higher frequency in the language and perhaps assisted by the etymological correspondences. This same explanation can also account for why why the abecedaries place Arabic ‫ س‬in the place of Aramaic ‫ש‬. The plain affricate [ts] = s³ merged with [s] = s¹ in all varieties of Arabic, and so Huehnergard is right to reconstruct this shift for Proto-Arabic. This shift was probably part of a larger process of deaffrication, affecting the reflex of *z [dz] as well. While the emphatic stops would have had a phonetic motivation to resist deaffrication, there is 44 MHGA, v. 2020-1 no reason to assume that deaffrication would have applied only to s³ and bypass other non-emphatic affricates. The reconstruction of the Arabic sibilants is as follows: Proto- Semitic Old Arabic Sibawayh Conventional Classical Arabic pronunciation and (most) modern vernaculars s1 = [s] [s] [s] [s] s2 = [ɬ] [ɬ] [ç] [∫] s3 = [t͡s] [s] [s] [s] 1.4.2 Notes on the Emphatics As stated earlier, it is unclear whether the emphatics of Proto-Arabic remained glottalized or if they had already become pharyngealized, and if this process affected all the emphatics at the same time. We will assume for the sake of economy that they were pharyngealized, but all possibilities will be discussed below. We can, however, be sure that they were voiceless and did not affect the quality of adjacent vowels. ProtoSemitic Old Arabic *[tθ’] *[tθˁ] or* [tθ’] *[t’] *ṭ [tˁ] or [t’] *[ts’] *ṣ [ts’]/[tsˁ] or [s’]/[sˁ] *[tɬ’] *ṣ́ [ɬˁ] or [ɬ’] *[k’] *q [q] or [k’] 45 MHGA, v. 2020-1 1.4.2.1 Qāf The reflex of the glottalized velar stop *q [k’] is transcribed with the glyph for the emphatic velar or post-velar stop in all of the Semitic scripts. Palmyrene23 ‫* מקימו‬/moqīmo/ Gk Μοκιμος Nabataean24 ‫* אלקימו‬/ʾal-qayyimo/ Gk Καιμος This indicates quite clearly that the sound change *q > [g] was unknown in these early periods. Moreover, we can be sure that *q was not realized as a /g/ in the North Arabian alphabets, as this sign is never used to transcribe foreign /g/: grmnqṣ (LP 653) = GERMANICUS, and not **qrmnqṣ. Moreover, the q is transcribed consistently with Greek κ in the Graeco-Arabica, indicating that it was both unaspirated and voiceless. We cannot, however, know from transcriptions whether or not the sound was realized as a uvular stop once pharyngealization set in or if it remained a glottalized velar stop. Sibawayh states the following about the *q: ‫ومن أقصى اللسان وما فوقه من ال َحنَك األعلى ُم ْخ َر ُج القاف‬ ِ ‘And from the furthest back of the tongue and that which is above it of the hard palate is the point of articulation of the ‫’ق‬ This description is clearly one of a post-velar rather than velar stop, as Sibawayh describes the velar ‫ ك‬as originating ‫ أسفل‬, that is, ‘in front’ of the ‫ق‬. Sibawayh, however, is much less clear when it comes to voice. Two categories appear in the Kitāb which seem to intersect with properties of voice and aspiration, maǧhūr and mahmūs. Carter correctly points out that a simple binary interpretation of voiced – voiceless does not explain the facts, but other solutions are equally unsatisfying.25 23 Stark 1971:96 24 Negev 1991:58 25 For a summary of previous views, see Carter (2004:126) and Al-Nassir (1993: 36). 46 MHGA, v. 2020-1 Sibawayh’s maǧhūr and mahmūs sounds ‫فاما المجهورة فالهمزة وااللف والعين والغين والقاف والجيم والياء والضاد والالم والنون والراء والطاء والدال والزاى‬ ‫والظاء والذال والباء والميم والواو فذلك تسعة عشر حرفا‬ ‫وا َّما المهموسة فالهاء والحاء والخاء والكاف والشين والسين والتاء والصاد والثاء والفاء فذلك عشرة ُ احرف‬ Watson et al. argue that mahmūs and maǧhūr signify turbulent airflow and nonturbulent airflow, respectively. If this understanding is correct, then the classification of [q] as a maǧhūr sound does not imply that it was voiced, but simply unaspirated. This interpretation is corroborated by transcriptions from the Umayyad period in which this sound is consistently transcribed with the Greek unaspirated stop κ, and never γ. *q > ʔ In many modern dialects of Arabic, *q is realized as [ʔ]. Sibawayh makes no mention of this realization, but there are two curious cases in Safaitic where etymological *q is written with the ʾ-glyph, both in the word qyẓ > ʾyḍ. The significance of the use of ʾglyph here for etymological *q is unclear. In one of the inscriptions, ʾyḍ occurs next the word qbll “reunion”. This could suggest that q > ʾ was perhaps originally a conditioned sound change or that the spelling of qbll was traditional while ʾyḍ reflects a contemporary pronunciation.26 *q = <γ> Only one clear case of *q written with γ is known to me – the word Αλγασαγες in P.Petra 17. There are two possible interpretations of this term (Al-Jallad et al. 2013:37), of which only one requires a connection with the Arabic root √qṣṣ. The relevant one for our discussion is a connection with the term qaṣqaṣ in CAr.27 If this is correct, then it would suggest, at the very least, the sound was fronted to a uvular position, which the scribe heard as voiced in this particular case. This explanation is much more likely 26 On problems with assuming a writing tradition in the context of Safaitic, see (Al-Jallad 2015, §1.2). CAr qaṣqaṣ “the breast of anything”(Lane, 2527b). The term is assumed to refer to a feature of the toponymy, like a hill. 27 47 MHGA, v. 2020-1 than arguing for a full *q > g shift since the remaining cases of *q in this corpus are written with κ.28 1.4.2.2 Ṣād *ṣ = [t͡s’] The Nabataean town of Nessana in the Negev was the meeting point of two types of Arabic during the Conquests, which we are witness to through the Greek transcription of personal names, beginning in the early 6th c. CE and ending in the late 7th. One of the most pronounced differences is the transcription of the emphatic affricate. Steiner (Steiner 1982: 81) noticed an interesting development in the spelling of the name of town following its fall to the Muslims in the early 7th century. Before the Conquests, the town was spelled in Greek as Νεσσανα, while by the late 7th century, the name was occasionally spelled as Νεστανα, corresponding to ‫ نصان‬in the Arabic documents. AlJallad (2014c) configured this evidence with the spelling of ‫ العصر‬in a 9th c. CE translation of the Qurʾān into Greek as αλεξαρ and a close reading of Sibawayh’s description of the sound to reconstruct an early [t͡sʕ] pronunciation of this phoneme in the Arabic of the Conquests. At the same time, the spelling Nεσσανα suggests that the *ṣ was already de-affricated in pre-Islamic Arabic of the Negev. I have also argued elsewhere (Al-Jallad 2014a, 2017a) that the evidence from the Graeco-Arabica suggests a similar development throughout the northern Old Arabic dialects, as we find no clear instances of *ṣ represented by Greek digraphs στ or τς, or simply τ, in contrast with Greek transcription of Punic, where the affricate is sometimes represented as other than σ (Steiner 1982: 60-65). While Sibawayh’s ṣ was clearly pharyngealized, it is also likely that the ṣ of northern Old Arabic was as well on account of the fact that it was deaffricated. I will return to this point below. The matter of voice is much clearer. Reflexes of ṣ are virtually always transcribed with σ, suggesting that the sound was voiceless, regardless of its other features. Only one example — in a damaged context — of a voiced realization of this sound is attested: the author of C 2823-4 (+ Greek) transcribes the name ḫlṣ written in the Safaitic script as Αλ̣ιζ̣ου, suggesting that Greek [z] was the closest sound to his ṣ. With only one For example, the family name αλ-Κουαβελ /al-qowābel/ or the toponym αλ-Κεσεβ /al-qeseb/. 28 48 MHGA, v. 2020-1 attestation, however, it is difficult to determine how widespread this phenomenon was and, moveover, since this transcription is only known from a poor handcopy, it may simply be an error of the copyist. In another Safaitic-Greek inscription, this time with a proper photograph, the name nṣrʾl is written as Nασρηλος, pointing towards a voiceless pronunciation. So what are we to make of this evidence? Transcriptions from the Islamic period and Sibawayh’s preferred pronunciation suggest affrication and pharyngealization while the northern Old Arabic dialects suggest deaffrication. Here we should note that we are not forced to choose between pharyngealization and glottalization. In fact, the Modern South Arabian languages indicate that these two co-articulations could have a complementary distribution.29 Perhaps in the northern dialects, deaffrication preceded the shift from glottalization to pharyngealization, producing an ejective sibilant [s’]. The instability of this sound, which is exceedingly rare in the world’s languages, motivated the fronting of the secondary articulation, producing [s ʕ] < *[s’] < *[t͡s’].30 The development of pharyngealization in this phoneme could have catalyzed the eventual shift to pharyngealization in the rest of the emphatic series. 1.4.2.3 Ḍād Sibawayh’s phonetic description of the ‫ ض‬glyph leaves little doubt that the Arabic which interested him preserved a lateral realization of this phoneme, most likely [ɮˁ]: ‫ومن بين َّأو ِل حافة اللسان وما َيليها من األضراس ُم ْخ َر ُج الضاد‬ ِ And from between the front edge of the tongue and the adjacent molars is the point of articulation of the ‫ض‬ Two other forms of evidence are usually summoned to support the idea of an ancient lateral in Arabic. The first is the spelling of the name of the Arabian deity Rḍw as Ruul-da-a-a-u in the Esarhaddon Prism, which dates to 673-672 BCE. This pronunciation seems to have originated in the northern oasis of Dūmah, which the Assyrians termed âl dan-nu-tu lúA-ri-bi ‘the strong city of the Arabians’. Such a description, however, 29 For the situation in Mehri, see (Watson 2012, §1.1.1.2) 30 I have suggested a similar development in (Al-Jallad 2014a, §3.7.2). 49 MHGA, v. 2020-1 does not tell us anything about the language spoken at this oasis. Only three inscriptions from Dūmah (WTI 21-23), composed in a unique local variant of the South Semitic script, are known, and they are relatively uninformative from a linguistic point of view. Incidentally, all three attest the divine name Rḍw. The equation of Dumaitic Rḍw with neo-Assyrian Ru-ul-da-a-a-u indicates that the sound was a lateral but the use of the da syllable unfortunately cannot tell us about voice. The Neo-Assyrian d could represent both the voiced stop d and the emphatic ṭ. The choice to use it for the representation of the lateral here may simply have stemmed from its emphatic quality. The ta sign is used to represent the unemphatic lateral: Neo-Babylonian ba-al-tam-mu, cf. Hebrew or Phoenician bōśam or, more likely, bāśām ‘Commiphora opobalsamum (a tree)’.31 The second commonly cited example comes from an account of Herodotus (mid-5th c. BCE) regarding the deities worshiped by the Arabs of eastern Egypt. He states: Herodotus, Historia 3.8 ὀνομάζουσι δὲ τὸν μὲν Διόνυσον Ὀροτάλτ, τὴν δὲ Οὐρανίην Ἀλιλάτ Now they [the Arabs] call Dionysos Orotalt and Urania they call Alilat Many scholars have considered this name a garbled form of Rḍw or perhaps even Palmyrene ʾrṣw = */ʔar-roṣ́aw?/, wherein the reflex of the emphatic lateral was represented by λτ, similar to the neo-Babylonian spelling listed above. While it is probably pointless to attempt to elucidate phonological realities from such a corrupted form of the name, if - and this is a big if - the λτ sequence does reflect an original representation of the phoneme *ṣ́, it would also seem to suggest the presence of affrication in light of the Greek transcription of the plain voiceless lateral of Semitic *baśām is βάλσαμον. The use of τ must then signal affrication, as it did in transcriptions of Phoenician ṣ as στ. Thus, the ancient Arabic of the Sinai could have preserved its voiceless configuration, and possibly its original affricate/ejective quality as well, [t͡ɬ’]. The NWS languages consistently transcribe this phoneme with the emphatic affricate, ṣ. This, in and of itself, only proves that it had not merged with *ṯ,̣ which was transcribed 31 See Steiner (1977: 129); see also Kogan (2011:78) for discussion and further bibliography on this word. 50 MHGA, v. 2020-1 separately with ṭ. Indeed, there is no evidence for the merger of these two sounds throughout the Nabataean corpus. A single exception to this seems to be the name Hatra, which his rendered as ḥṭrʾ in the local Official Aramaic inscriptions. The Arabic name of the town from the Islamic period is al-ḥaḍr, and, on this basis, several scholars have tried to derive the Aramaic form from the Arabic root √ḥḍr ‘to reside, dwelt, or abode, in a region, district, or tract of cities, towns, or villages, and of cultivated land’ (Lane, 589a). This would assume that the Arabic lateral fricative had shifted to a stop or interdental fricative, perhaps merging with *ṯ,̣ which was also voiceless (see above). Before positing such an important shift, one should disqualify the possibility of an Aramaic origin. In fact, the name has a perfectly good Aramaic etymology, namely, an ‘enclosure, hedge, or fence’, a reflex of the root √ḥṭr, cognate with Ar ḥaẓara ‘to forbid, prohibit’ (Lane, 595).32 Note that had the name been drawn from Arabic originally, but from the root √ḥẓr rather than √ḥḍr, it would have appeared identical to its Aramaic cognate in the Aramaic script, and indeed in Greek and Latin, Ατρα and HATRA, respectively. Thus, the base ḥṭr could reflect either Aramaic or Arabic, but neither case requires the association with the root ḥḍr. The form from the Islamic period, al-ḥaḍr, must simply reflect the late confusion of ḍ and ẓ or perhaps folk-etymologization. The dialects expressed in the Safaitic and Hismaic scripts likewise reflect a preservation of *ṣ́ as a distinct phoneme. The glyph for *ṣ́ in Safaitic and some of the other Ancient North Arabian scripts is identical to the glyph for *ḏ in ASA. One should, however, not read too much into this as the history of these alphabets is far from clear and their similarities may be accidental. The same phoneme is represented by two concentric circles in Hismaic. This fact has been the subject of extensive speculation, none of which stands scrutiny. Our only clue into the phonetic realization of these sounds is through Greek transcription. In all cases, this phoneme is transcribed with Greek σ. This tells us two things: the sound was voiceless and not an interdental or a stop. These parameters agree with the 32 Klaus Beyer, Die aramäischen Inschriften aus Assur: Hatra und dem übrigen Ostmesopotamien, Göttingen, 1998. 51 MHGA, v. 2020-1 original value of this phoneme, namely, an emphatic lateral fricative or affricate, [t͡ɬ’]. This sound is attested in transcription far less frequently than the reflex of *ṣ, but nevertheless, no overt representation of affrication is found. This could suggest deaffrication to [ɬ’] and then the natural shift to [ɬˁ]. Limited evidence for the voiced realization of *ṣ́ comes from 6th century Petra, Elusa, and Nessana, where the phoneme is given with Greek Zeta, indicating that it had not merged to the value of the emphatic interdental. 1.4.2.4 Ẓāʾ As mentioned earlier, all of the ancient evidence points towards a realization of *ẓ distinct from *ḍ. This phoneme is always given in Greek transcription with Tau, even in bilingual Safaitic-Greek texts. This minimally indicates that the sound was voiceless, but the consistent use of the unaspirated stop contrasts with the representation of the plain interdental, which fluctuates, even in bilingual texts, between Tau [t] and Theta [th]. Ιαιθεου = yṯʿ = /yayṯeʕ/ Γαυτος = ġṯ = /ġawṯ/ This suggests that the onset of the emphatic interdental was an affricate, [ tθˁ] or perhaps [tθ’]. The sound described by Sibawayh is clearly the pharyngealized counterpart of ḏ [ð] and this is how it is realized in the contemporary pronunciation of Classical Arabic, as well as in most modern vernaculars that have not lost the interdentals. In southwest Arabia, however, a voiceless realization of this consonant survives, [θˁ], and a reflex of this sound is found in some modern vernaculars of the Maghreb, [tˁ] < *θˁ. 1.4.2.5 *g = ‫ج‬ There can be no doubt that this phoneme was realized as voiced velar stop in ProtoArabic, [g], and this reflex is attested widely in the modern vernaculars (Egypt, Yemen) and in Old Arabic, the phoneme is only represented by Greek γ [g]. Sibawayh was 52 MHGA, v. 2020-1 certainly aware of this pronunciation, which he describes as the ǧīm which is like the kāf, but he does not deem it appropriate for the performance register. The pronunciation he does endorse, however, seems to have been a palatal stop rather than a palato-alveolar affricate [d͡ʒ], which is used in the standard pronunciation of Classical Arabic today. 1.4.2.6 The merger of ‫ ض‬and ‫ظ‬ Perhaps the most ubiquitous sound change in Arabic today is the merger of the emphatic lateral and interdental to the value of the interdental, which in most forms of Arabic was [ðˁ]. These two phonemes are consistently kept apart in Nabataean Arabic, Safaitic, Hismaic, the QCT, and remain distinct in some vernaculars of southwest Arabia. The earliest evidence of their merger occurs in the 6th c. transcriptions of Arabic from the Negev (P.Ness) and Petra (P.Petra) where both phonemes are transcribed with Greek Zeta. This would suggest a merger, not towards the value of the interdental, but rather to a voiced reflex of the emphatic lateral, [ɮˁ], something perhaps found in Andalusi Arabic as well. It is possible that the spelling of ẓ with ḍ in Safaitic ʾyḍ /ʔayāṣ́/ reflects a merger to the lateral value as well. In Islamic-period transcriptions, both sounds are given with Delta, maybe suggesting that they had already merged towards the emphatic interdental. In the earliest Arabic documentary texts, the two sounds are confounded as they are in the earliest Christian Arabic texts as well. The merger of *ẓ and *ḍ sometimes occurs in Late Sabaic, perhaps suggesting that the source of this merger was southern Arabia, whence it diffused at a rather late period. 1.2.2.1 Sound changes in non-conventional Classical Arabic 1.2.3 Common Sound changes in medieval and modern Arabic 53 MHGA, v. 2020-1 1.5 Proto-Arabic – Semitic Sound Correspondences ProtoArabic Transcri ption CAr Ugaritic Biblical Hebrew Official Aramaic Geʿez Akkadia n ProtoSemitic ʾ ‫ا ى و‬/ [ʔ] ʾ ‫א‬ ‫א‬ ʾ [ʔ] b ‫ ب‬/ [b] b ‫ב‬ ‫ב‬ b ʾ/∅ b [b] g [g]/ ‫ج‬ g ‫ג‬ ‫ג‬ g g [g] d [d] / ‫د‬ d ‫ד‬ ‫ד‬ d d [d] h [h] / ‫ه‬ h ‫ה‬ ‫ה‬ h [h] w [w] / ‫و‬ w ‫ו‬ ‫ו‬ w ʾ/∅ w [w] z [z] / ‫ر‬ z ‫ז‬ ‫ז‬ z z [dz] ḥ / ‫[ح‬ħ] ḥ ‫ח‬ ‫ח‬ ḥ [ħ] ṭ ‫ ط‬/ [tˤ] ṭ ‫ט‬ ‫ט‬ ṭ ʾ/∅ ṭ [t’] y [y] / ‫ى‬ y ‫י‬ ‫י‬ y y [j] k [k] / ‫ك‬ k ‫כ‬ ‫כ‬ k k [k] l [l] / ‫ل‬ l ‫ל‬ ‫ל‬ l l [l] m [m] / ‫م‬ m ‫מ‬ ‫מ‬ m m [m] n [n] / ‫ن‬ n ‫נ‬ ‫נ‬ n n [n] s [s] / ‫س‬ s¹ ‫ס‬ ‫ס‬ s š [ts] ʿ / ‫ع‬/ [ʕ] ʿ ‫ע‬ ‫ע‬ ʿ [ʕ] f [f] / ‫ف‬ f ‫פ‬ ‫פ‬ f ʾ/∅ p [p] ṣ ‫ ص‬/ [sˤ[ ṣ ‫צ‬ ‫צ‬ ṣ ṣ [ts’] ṣ́ = ḍ ‫ ض‬/ [ɮˤ] ḍ ‫צ‬ ‫ע‬ ḍ ṣ [tɬ’] q ‫ق‬/ [q] q ‫ק‬ ‫ק‬ q q [k’] r [r] / ‫ر‬ r ‫ר‬ ‫ר‬ r r [r] s ‫ س‬/ [s] s¹ ‫ש‬ ‫ש‬ s š [s] t [t] / ‫ت‬ t ‫ת‬ ‫ת‬ t t [t] 54 MHGA, v. 2020-1 ṯ ‫ث‬/ [θ] ṯ ‫ש‬ ‫ת‬ s š [θ] ḫ [x] / ‫خ‬ ẖ ‫ח‬ ‫ח‬ ḫ ḫ [x] ḏ [ð] / ‫د‬ ḏ ‫ז‬ ‫ד‬ z z [ð] - ṯ̣ = ẓ ‫ ظ‬/ [ðˤ[ ẓ ‫צ‬ ‫ט‬ ṣ ṣ [tθ’] ġ ‫ ع‬/ [ɣ] ġ ‫ע‬ ‫ע‬ ġ [ɣ] ś */ɬ/ ‫ش‬/ [ʃ] s² ‫ש‬ ‫ס‬ ś ẖ/∅ š [ɬ] 55 MHGA, v. 2020-1 II Morphology 2.1 Independent Pronouns 1st person common singular ProtoSemitic ProtoArabic Safaitic/ Hismaic Nabtaeo -Arabic QCT Classical Arabic Levantine Emirati Moroccan Baghdadi *ʔanā *ʔanā ʾn ʾnh ʾnʾ ʔana ʔana < *ʔanā ʔāna ʔana/ʔāna /ʔanaya ʔānī *ʔanāku LOST There is no evidence for the long form ʔanāku in any form of Arabic and therefore Huehnergard (2017) posits its loss in Proto-Arabic. Moroccan Arabic has innovated a new long form with a suffixed ya, which is of uncertain origin. The final vowel of the Proto-Arabic pronoun was probably long and the first vowel short. Forms with the opposite order, such as Emiratī Arabic ʔāna, are likely due to metathesis. The Classical Arabic form ʔana, with a final short vowel, is perhaps due to contamination with the second person series, which has short final vowels. Baghdadi (and elsewhere) ʔānī appears to be derived from the methathesized form ʔāna, with the levelling of the vowel of the accusative and genitive forms of this pronouns, which are nī and ī, respectively. In the Nabataeo-Arabic script and a few 6th c. CE Arabic-script inscriptions, the pronoun is spelled ʾnh, which is best interpreted as an Aramaeogram, that is, a spelling frozen from the Nabataean script’s Aramaic past. The Ḥarrān inscription attests ʾnʾ which must represent /ʔanā/. 56 MHGA, v. 2020-1 2nd masculine Singular ProtoSemitic ProtoArabic QCT Classic Levantine al Arabic Ṣanʕā nī ʕAsir /Ḥigāz Najdi Moroccan *ʔanta *ʔanta ʾnt ʔanta ʔent ʔant ʔant ʔant NA *ʔantah *ʔanta h NA ʔantah (pause ) ʔente /ʔenta NA ʔantah ʔanta nta/ntaya 2nd Feminine Singular ProtoSemitic ProtoArabic QCT Classic Levantine al Arabic Ṣanʕā nī ʕAsir Najdi Moroccan *ʔanti NA ʾnt ʔanti NA NA NA NA NA *ʔantih( ?) *ʔantī/ h (?) NA NA ʔentī ʔantī ʔantī ʔantī nti/ntiya The comparative evidence requires the reconstruction of two forms of the 2nd person pronouns, a short form and perhaps a longer, topicalized or emphatic form, terminating with an h. This is because in the modern Arabic dialects, as well as in other Semitic languages, the loss of the final vowels on these pronouns is irregular. Some dialects exhibit by-forms, one reflecting an original form with a final short vowel: ʔent < *ʔenta < ʔanta and ʔenta/e < *ʔenta/eh < *ʔantah. The e-reflex of the final vowel of this pronoun resembles the reflex of the feminine ending in many Levantine dialects, pointing towards a form terminating in *ah.33 The feminine singular form only exhibits a reflex with a final long vowel in the modern dialects of Arabic. The QCT spelling, however, seems to reflect an original short vowel, unless the long vowel was shortened as often happens to final ī in its language, e.g. yā rabbi ‘O my lord’ < *yā rabbī. It is logical to posit an emphatic form *ʔantih from which stems from the emphatic form ʔantih, through perhaps a marginal sound change of ih# > ī or contamination with the feminine ending on the 2nd person prefix conjugated 33 This idea is developed in Al-Jallad 2014c. 57 MHGA, v. 2020-1 verb, e.g. taktubī. If the QCT form is indeed secondary, then it is possible that this change occurred at the Proto-Arabic stage, and only one pronoun may be reconstructed for the 2nd person feminine, namely, *ʔantī. This, however, requires an explanation for the Classical Arabic form -- provisionally, I would suggest that it is taken from the QCT. These pronouns have not yet been attested in the pre-Islamic epigraphic record. 3rd person singular ProtoSemitic ProtoArabic Classical Arabic QCT Safaitic Levantine Egyptian *suʔa *huwa huwa hw hw [howa] hū hū *suʔati *huwati huwah NA huwwe/hūti howwa(t) Proto-Semitic Proto-Arabic Classical Arabic QCT Levantine Egyptian *siʔa *hiya hiya hy hī hī *siʔati *hiyati hiyah hyh (?) hiyye/hīta heyya(t) Proto-Semitic made a distinction between nominative and oblique independent 3rd person pronouns, the latter terminating in the syllable *ti. While it appears that the functional difference between the two forms was lost at the Proto-Arabic stage, they nevertheless survived in usage. Reflexes of the oblique forms might be found in Classical Arabic huwah and hiyah, where the Grammarians interpret the final h as ‘protecting’ the vowel in pause. There is only one possible case in which an oblique form may be attested in the QCT, in 101:10, which gives the pronoun as hyh [hiyah] < *hiyat < *hiyati (but other explanations are possible).34 Most modern dialects show reflexes of the oblique form (Zaborski 1996), mostly without the t but some preserve it. 34 This suggestion was first made by Adam Strich, whom I thank. 58 MHGA, v. 2020-1 These in general have replaced the old nominative forms. Their phonological development follows the same path as the word ‘one hundred’. *miʔatu > *miyatu > *miyat > *miyah > *miyyah *hiʔati > *hiyati > *hiyat > *hiyah > *hiyyah 1st person plural Proto-Semitic ProtoArabic Classical Arabic QCT Levantine Najdi Egyptian niḥnu *naḥnu naḥnu nḥn neḥna /neḥen /eḥnā ḥinnā iḥnā The plural is unattested in the ancient material, but QCT nḥn must reflect either /naḥn/ or /naḥnu/. A common analogical change in the modern dialects levelled the vowel of the oblique ending, -nā, to the independent pronoun, producing naḥnā, which, in some dialects, resulted in the dissimilation of the first vowel to i, neḥnā. Reflexes of the original form persist in Syria and the Gulf, e.g., neḥen < naḥnu. An innovative form *ḥin+ā/na is found in several dialects, producing iḥnā and in the Peninsula dialects, ḥinn. The origin of this form is unclear. 2nd person plural ProtoSemitic ProtoArabic Classical Arabic QCT Najdi (readings) Baghdadi 2mp *ʔantum(ū) *ʔantum(u) ʔantum ʔantum /ʔantumū ʔantum ʔentū 2fp *ʔantin(ā) *ʔantin(nah) ʔantunna ʔantunna ʔantin ʔenten The second person plurals have two forms -- a base form and one modified by verbal morphology. Several Ḥigāzī Qur’anic reading traditions attest the form ʔantumū, which results from the addition of the masculine plural ending ū from the verb to the pronoun, 59 MHGA, v. 2020-1 e.g. taktubū ‘you mp. write’. The existence of such forms throughout Semitic may suggest that such by-forms go back to Proto-Semitic. The feminine form ʔantinna results from the same process, but does not continue the Proto-Semitic form terminating with an ā (which is originally from the suffix conjugation). Instead, it uses the termination from the prefix conjugation, e.g. taktubna. The base form *ʔantin, while unknown in Classical Arabic, is attested in some modern vernaculars, e.g. Najdi ʔantin and not ʔantínn. The vowel of the masculine was originally u and the feminine i, based on the comparative evidence and the modern vernaculars. The u vowel in both pronouns in Classical Arabic is the result of secondary leveling. The innovative dialectal form ʔentū results from the expansion of the verbal ending -ū to the second person base * ʔant-. 3rd person plural ProtoSemitic ProtoArabic Classical Arabic QCT Najdi (readings) Baghdadi 3mp *sum(ū)ti *hum(ū) hum humū hum humma 3fp *sin(ā)ti *hin(na) hunna hunna hin henn Proto-Arabic appears to have lost the original oblique forms, sunūti, sināti. No oblique forms are attested in the ancient evidence. Like the second person plural series, the 3rd plurals can be augmented by verbal morphology -- *hum by the masculine plural ū and *hin by the feminine plural na.The original feminine *hin is preserved in some dialects, e.g. Najdi, while the augmented form is the only one Classical Arabic knows. The 3mp form humma, with the doubling of the medial m, seems to result from contamination with hinna, although the preservation of the final /a/ requires an explanation. It may result from the spread of the /a/ of the 3rd singular series to this form. 60 MHGA, v. 2020-1 Masculine and feminine have collapsed to one form in most modern dialects. In many parts of the Levant, the pronoun is hinne. This may bee the result of convergence with Aramaic or perhaps the levelling of the feminine form, which must have been *hinnah. 2.1.2 The duals There is some debate as to whether the dual pronouns can be reconstructed to ProtoSemitic (e.g. Weninger 2011). Since each branch attests dual pronouns, their reconstruction seems rather uncontroversial. What is unclear, however, is their realization. The dual pronouns in Sabaic (and Ancient South Arabian) as well as Dadanitic terminates in a y, which likely points towards a diphthong /ay/. In Classical Arabic and the QCT, these pronouns terminate in ā, spelled hmʾ in the latter and never with an alif-maqṣūrah. Classical Arabic and Sabaic show the same endings on the verb and pronouns, while Dadanitic exhibits a heterogeneous situation. The dual paradigm in Dadanitic Verb hẓlh /haẓallā/ Suffix Pronoun -hmy /humay/ The Dadanitic situation may reflect the original alignment, where the ending ā indicated the subject while the ending -ay is found on oblique usages, paralleling, and perhaps ultimately derived from, the nominal system. Thus, Classical Arabic must have levelled the -ā ending for all situations: katabā and kitābu-humā, from *kitābu-humay. This was not the case in Proto-Arabic, however. Safaitic exhibits a -y ending on the dual verb, suggesting leveling in the opposite direction, ḍlly /ṣ́allalay/ ‘they both were lost’, indicating that the Proto-Arabic situation was heterogeneous. In Safaitic the pronoun hm occurs with a dual antecedent, but the writing of word final diphthongs is not consistent, and so this spelling can equally reflect /homay/ and /homā/. 2.2 Clitic Pronouns Genitive and accusative pronouns are clitics, related in form to the nominative ones discussed above. Problematic forms will be discussed below. 61 MHGA, v. 2020-1 1cs 1cs/genitive Proto-Semitic Proto-Arabic *ī *ya *ī (after nom?) *ya (after certain vowels) Two allomorphs of this pronoun must be posited for Proto-Semitic. The consonantal ya occurs most often after long vowels, but its exact distribution in Proto-Arabic is unclear. The -ya form is used following the genitive in some reading traditions of the Qur’an. This usage is also found in a Dumaitic inscription, which reflects a language otherwise indistinguishable from Arabic: sʿdn ʿl-wddy /sāʕidū-nī ʿal-wadadi-ya/ ‘help me in the matter of my love’. In the modern dialects it is found after long vowels, e.g. Levantine ʔabūy(a) ‘my father’, Classical Arabic riǧlay-ya ‘my two feet’. The accusative form *nī can be reconstructed to Proto-Arabic, e.g. *ṣ́araba-nī ‘he struck me’. The short form -n /-ni/ , attested in the QCT, likely results from the widespread, and mostly pausal, shortening of final *ī in that dialect. 2ms Proto-Semitic Proto-Arabic 2ms *-ka *-ka 2fs *-ki *-ki / *-kī (?) These pronouns have the shape k in Safaitic and Hismaic, and <k> in the QCT, and the Classical Arabic forms terminate in short vowels. In most modern dialects, the pronouns have shifted to ak and ik, suggesting harmonization with the vowel preceding the suffix before its loss. While the masculine form is almost universally realized as ak in the modern dialects when in word-final position, the feminine has two forms, ik and kī. In dialects which exhibit both reflexes, the latter form appears after long vowels, which could be interpreted as follows 1) In this position, the masculine and feminine form would no longer be distinguished following the loss of short vowels. Since these were distinguished 62 MHGA, v. 2020-1 everywhere else in the language, speakers may have extended the suffix ī from the nominative pronoun to the clitic. Masc. *ʔabū-ka > ʔabū-k Fem. *ʔabū-ki 2) ʔabū-k > ʔabū-kī, extension of ī from ʔantī. Also possible is the operation of a marginal metathesis rule affecting high vowels in this position. *ʔabū-ik > ʔabū-ki. Since vowel length in the high vowels is no longer distinguished in word-final position, the metathesized short i merged with ī. 3ms/3fs ProtoSemitic Proto-Arabic 3ms *-su -hu / -Vnnahu 3fs *-sā -hā / -Vnnahā The masculine singular form must be reconstructed as -hu, with a short vowel. In Classical Arabic, the vowel harmonizes with a preceding /i/, so *kalbu-hū ‘his dog’ (nom) vs. *kalbi-hī ‘his dog (gen). This appears to be a particular development of Classical Arabic, and perhaps of some eastern dialects, but cannot be reconstructed for Proto-Arabic. Indeed, Old Ḥigāzī maintained the u vowel in all environments, and this is indeed what we find in the Damascus Psalm Fragment and in many modern dialects, e.g. Egyptian ʕalē-hum; Psalm Fragment γαλειὑμ /ʕalei-hum/, etc. Another particularity of Classical Arabic is contrastive length harmony: the vowel of this pronoun is short after long vowels but long after short vowels: banā-hu ‘he built it’ kalbu-hū ‘his dog’ Some modern dialects exhibit the opposite distribution: the vowel is long after long vowels, e.g. Levantine ʔabū-hu ‘his father’. This may be the result of a metathesis rule suggested above: ʔabū-uh > ʔabū-hu. 63 MHGA, v. 2020-1 The 3fs is much more difficult to reconstruct and seems to exhibit reflexes of both a long *hā and short *ha. The latter form is encountered in Old Arabic, for example, in the Namarah inscription: mlk ʾl-ʿrb kl-h */malk ʾal-ʿarab kollah/ or Safaitic w lh rgm */wa lah-har-rogm/ ‘and the cairn is hers’ < */wa la-ha har-rugmu/. The suffix -ah is also quite widespread in Najdi Arabic. The most reasonable explanation to my mind is the leveling of length across the paradigm, thus asymmetric hu – hā was changed to -hu – -ha. Proto-Arabic -hā is reflected in the QCT <hʾ> */hā/ and the modern dialects, e.g. Levantine (West Bank) sayyārit-hā ‘her car’; Levantine (Damascus) binta < bintha < bintuhā. Old Arabic and some modern dialects attest 3rd person clitic with a prefixed n. Such forms are known in other Semitic languages (e.g. Ugaritic –nh) and therefore appear to be retentions from Proto-Arabic. These forms are attested in modern East Arabian dialects, those of Central Asia, etc. and in Safaitic. In the modern vernaculars they are restricted to the participle, while in Safaitic they occur after almost all verb forms. East Arabian Safaitic Participle ḍarbinno < *ḍāribannuh < NA *ṣ́āribannahu Prefix Conjugation NA yʿwr-nh /yoʕawwer-annoh ‘he will efface it’ Suffix Conjugation NA ʾgʿ-nh /ʔawgaʕa-nnoh/ ‘he caused him pain’ Imperative NA śʕ-nh /śīʕ-annoh/ ‘follow him’ 2p ProtoSemitic Proto-Arabic 3ms *-kum(ū) *-kum(ū) 3fs *-kin(ā) *-kin(na) 64 MHGA, v. 2020-1 The ending km is attested with two female antecedents in Safaitic, which could reflect either the loss of gender or a dual form, perhaps /komay/. The QCT has km and kmwin junction, going back to PS kumū. The feminine form is unattested in the ancient material, but the QCT has kn, which could be either original <kin> or Classical Arabic <kunna>. The modern dialects reflect an original *kin. As in the independent forms, Classical Arabic levelled the vowel of the masculine form to the feminine. The Classical Arabic feminine form is augmented by the feminine plural verbal ending -na. Dialectal mp form -ku < kū is the result of the same analogy that produced intu. 3p Proto-Semitic Proto-Arabic 3ms *-sum(ū) *-hum(ū) 3fs *-sin(ā) *-hin(na) The Namara inscription attests hm */hom/ rather than hmw */homū/. The modern dialects point back to an original 3fp *hin, while Classical Arabic *hunna reflects the leveling of the vowel from the masculine form. 2.3 Nominal Inflection 2.3.1 State Proto-Semitic nouns have two states: unbound (the default state) and bound (construct) forms. Construct forms were used in possessive constructions, namely, when a noun was followed by another noun in the genitive cases or clitic pronouns, or when the noun headed an asyndetic relative clause.35 Definiteness was not morphologically marked in Proto-Semitic nor was it in ProtoArabic, as we shall see below. Unbound forms terminated in the nasal -m in the singular (mimation), and feminine sound and broken plurals, and -na in duals and masculine sound plurals: 35 For the reconstruction of Proto-Semitic nominal morphology, see Huehnergard 2004. 65 MHGA, v. 2020-1 Unbound Bound (construct) Unbound Bound (construct) Nominative wāridum wāridu wāridūna wāridū Genitive wāridim wāridi wāridīna wāridī Accusative wāridam wārida The only change Proto-Arabic experienced here is the leveling of the n-endings to the singular/broken plural forms, producing nunation (tanwīn), thus, *wāridum > wāridun. 2.3.2 Case Inflection in Proto-Arabic Proto-Semitic inflected its nouns for three cases (see above) in most singulars and broken plurals. Two cases, nominative and oblique, are distinguished in other situations, but their distribution differs. Proto-Arabic had the following declensions: Unaugmented nominal stems (singular and broken plurals) Proto-Arabic Classical Arabic Akkadian Ugaritic Nom *kalbun kalbun kalbum kalbu Gen *kalbin kalbin kalbim kalbi Acc *kalban kalban kalbam kalba Five Nouns, Construct (unbound ʔabun) Proto-Arabic Classical Arabic Akkadian Geʿez Nom *ʔabū-ka ʔabū-ka abū-ka ʔabū-ka Gen *ʔabī-ka ʔabī-ka abī-ka ʔabū-ka Acc *ʔabā-ka ʔabā-ka abā-ka ʔabā-ka 66 MHGA, v. 2020-1 III-y declension (1) Proto-Semitic Proto-Arabic Classical Arabic (indef) QCT Nom *bāniyum *bānin bānin wd /wād/ Gen *bāniyim *bānin bānin wd /wād/ Acc *bāniyam *bāniyan bāniyan wdyʾ /wādiyā/ This declension results from the loss of i/uGV triphthongs. Diptotes Proto-Arabic Classical Arabic Ancient South Arabian Ugaritic Nom *ʔaḥmadu ʔaḥmadu thmt */tihāmatu/ ʔugaritu Gen *ʔaḥmada ʔaḥmada thmt */tihāmata/ ʔugarita Proto-Arabic Tihāmah Dialects Nabataean Arabic Sabaic *marwatu *bayṣ́āyu marwah ʿbdt */ʕobodat/, from *ʕubudatu/a, rather than ʕubudatun which would yield ʕbdtw. kdt /kiddatu/ , rather than kdtm Acc Diptotes Feminine Nouns36 Nom Gen Acc bayṯạ̄ y *marwata *bayṣ́āya kdt /kiddata/, rather than kdtm Feminine proper nouns are diptotic in Classical Arabic, Nabataean (lacking wawation), and Sabaic (lacking mimation). In the Tihāmah dialects, all nouns terminating with the feminine *at are diptotic, on account of the absence of wawation/nunation. This 36 This reconstruction is based on Van Putten 2017b. 67 MHGA, v. 2020-1 distribution can also explain the fact that in the QCT why the indefinite accusative of feminine nouns does not terminate in ʾ = /ā/: these forms never carried nunation and so the sound change an# > ā did not operate. The diptotic feminine is most likely a Proto-Arabic feature and perhaps even ProtoSemitic. It is easier to spread triptosy to all nouns, preserving an archaic situation in a closed class of nouns like personal names, rather than to spread diptosy from personal names to encompass all feminine nouns. Dual Proto-Arabic Classical Arabic Akkadian Ugaritic Nom *kalbāni kalbāni kalbān kalbāma Gen *kalbayni kalbayni kalbīn kalbēma Acc The oblique dual has been generalized in all modern vernaculars. The only ProtoArabic innovation in this paradigm appears to be the dissimilation of the final /a/ to /i/, perhaps first in the nominative form and then generalized to the genitive. Masculine plural Proto-Arabic Classical Arabic Akkadian Ugaritic Nom *mālikūna mālikūna šarrū malakūma Gen *mālikīna mālikīna šarrī malakīma Acc The oblique masculine plural has been generalized in all modern vernaculars. 68 MHGA, v. 2020-1 Feminine plural Proto-Arabic Classical Arabic Akkadian Ugaritic Nom *malikātun mālikātun šarrātum malakātu Gen *malikātin mālikātin šarrātim malakāti Acc 2.3.2.1 Development of the case system in Classical Arabic 1) A definite declension develops which is triptotic in singular/broken plurals and lacks nunation, and diptotic in the feminine plural, lacking nunation. This declension overrides diptosy in singular/broken plural nouns. Proto-Arabic Classical (def+indef) Arabic (indef) Classical Arabic (def) Nom *makātibu makātibu al-makātibu Gen *makātiba makātiba al-makātibi al-makātiba Acc 2) Triptotic declension is levelled to nouns terminating with the feminine ending -at. Proto-Arabic Classical Arabic Nom *kalbatu kalbatun Gen *kalbata kalbatin Acc kalbatan 3) The nunated accusative /-an/ is realized as ā in pausal position. 4) Development of a new III-w/y declension: nouns terminating in -ayV, following the collapse of triphthongs, produces a non-inflecting declension 69 MHGA, v. 2020-1 III-w/y declension (2) ProtoArabic Classical Arabic (indef) Classical Arabic (def) Nom *hudayun hudan < *hudān al-hudā Gen *hudayin hudan al-hudā Acc *hudayan hudan al-hudā 2.3.2.2 Development of the case system in Nabataean Arabic37 Evidence for Nabataean case is fragmentary and must be pieced together from a variety of sources. An active case system seems to be present in the ʿĒn ʿAvdat inscription (see appendix). Nabataean names, both in consonantal writing and in Greek transcription, preserve vestiges of original case marking, e.g., Αβδοβαλος /ʿabdo-baʿl/ (nom.) and ‫ עבדאלבעלי‬/ʿbdʾlbʿly/, probably ʿabdo-albaʿle (gen). The following developments explain the attested evidence. The nominative case is moreover attested in a Hismaic inscription from Wādī Ram, well within the Nabataean realm (Macdonald 2018, and appendix). 1) final short vowels are lost, resulting in the elimination of case on diptotes: Proto-Arabic Nabataean Arabic Nom *ʕubudatu Gen *ʕubudata ʕobodat, and later ʕobodah Acc 2) Loss of nunation produces new set of word final vowels. The reconstructed Nabataean Arabic case system in its earliest stages was as follows. Gray cells indicate 37 This reconstruction is based on Al-Jallad forthcoming. 70 MHGA, v. 2020-1 purely reconstructed forms based on phonological changes and white cells indicate attested forms. Triptote Diptote III-y/w 1 III-y/w 2 Dual MPL FPL Nom kalbo ʕobodat wādī phatē kalbān ʔasadūn banāto Gen kalbe kalbayn ʔasadīn banāte Acc kalba wādeya 3) In Late Nabataean Arabic (1st c. CE onwards), the nominative is generalized to all situations, producing ‘wawation’. 2.3.2.3 Development of case in the QCT38 1) *an# > ā 2) nunation is lost 3) final short vowels are lost 4) no analogies operate to element case in other environments Triptote Triptote (indef) (def) Nom kitāb ʔal-kitāb Diptote III-y/w 1 IIIy/w 2 Dual MPL FPL madīnah wād hudē gamalān mūmnūn ʔāyāt gamalayn mūmnīn Gen Acc kitābā wādiyā Triptote Diptote III-y/w 1 Nom baql madīnat Gen baql madīnata Acc baqla madīnata dāmeya 38 III-y/w 2 dām(ī)(?) phatay Dual MPL FPL gamalān maqtūlūna mośreqāt gamalayn maqtūlīna phataya This reconstruction is based on van Putten and Stokes forthcoming. 71 MHGA, v. 2020-1 2.3.2.4 Development of case in Tihāmah Arabic Tihāmah Arabic shares with Classical Arabic the definite declension. 1) Final short vowels are lost, eliminating case in definite nouns and diptotes. 2) In some varieties, nunation is lost, producing a new set of final short vowels. 3) Analogy with non-case inflecting forms generalizes the nominative to nouns, producing something similar to Nabataean wawation. In the dual and masculine plurals, the oblique is generalized. 4) The feminine plural does not exhibit wawation, indicating that it was inflected as a diptote, perhaps in analogy with the feminine singular. Triptote (indef) Nom Feminine Triptote IIIIII(def) y/w 1 y/w 2 kalbu/kalbun marwah im-kalb wādī fatā Dual MPL FPL gamalēn maqtūlīn banāt Gen Acc 2.3.2.5 Development of case in Najdi Arabic 1) final short vowels are lost, eliminating case in the definite declension and in diptotes. 2) vowel quality is neutralized before nunation, obscuring the inflection of case there. 3) analogy with non-declining singular/broken plurals eliminates case in duals and masculine plurals, preserving the oblique form. Nom Triptote (indef) Triptote (def) III-y/w 1 III-y/w 2 Dual MPL FPL kablən al-kalb wādī fatā kalbēn magtūlīn banātən Gen Acc 2.3.2.5 The Development of case in the early Islamic period In transcriptions of Arabic names from the 7th c. CE in Greek, a regular opposition between Αβου /abū/ (nom) and Αβι /abī/ (gen) is observed. The Damascus Psalm 72 MHGA, v. 2020-1 Fragment occasionally preserves the genitive case with pronominal suffixes (see appendix), a feature also found in Phoenician. 2.3.2.6 Development of case in most modern Arabic languages In most modern Arabic vernaculars, case and nunation have disappeared entirely, save for loans from Classical Arabic or other dialects. These languages, nevertheless, appear to descend from a system like the QCT, where only the accusative case of the indefinite declension survived in singular/broken plural nouns. This case was reanalyzed as an adverbial marker, one of the functions of the accusative, e.g. barrā ‘outside’ and ḥadā ‘anyone’ <*ʔaḥadā. The inflection of the dual and masculine plurals was lost in analogy with the absence of inflection elsewhere. The accusative is moreover preserved in some marginal vocative usages, e.g. Levantine yā-bā ‘O father’ < *yā-ʔabā < *yā-ʔaban; yā-mmā ‘O mother’ < *yā-ʔimmā < *yā-ʔimman. Ancient wawation, vestiges of the nominative case, survives in vocative kinship terms in Levantine (and other) vernaculars: ʕammo ‘paternal kinsman’ < cf. Nabataean Arabic ʕammo, Classical Arabic ʕammun; ḫālo ‘maternal kinsman’; sīdo ‘grandfather’ < *sīdun; sitto ‘grandmother’ <*sīdtun < *sīdatun. The use of wawation on feminine nouns contradicts the Nabataean situation and suggests that the feature was extended to feminine kinship terms following the collapse of the case system. In fact, the original distribution is preserved in vocative nouns in Mesopotamia, where masculine forms terminate in u/o while feminine forms simply terminate in the feminine ending a. These facts suggest that the “vocative” forms simply continue the older nominative nouns with the expected distribution of triptotic and diptotic declensions. Proto-Arabic Nabataean Mesopotamia (vocative) Syria *ʕamrun ʕamro ḫālo ‘uncle!’ ḫālo *ʕubudatu ʕobodah ḫāla ‘aunt!’ ḫālto 73 MHGA, v. 2020-1 2.3.3 The adverbial endings In addition to the case endings, two “adverbial” endings are reconstructable for ProtoSemitic. Proto-Semitic Akk Ug Hebrew Locative *baytum ‘at home’ bītum bētu N/A Directive *baytis ‘to home’ bītiš bētah hab-baytâ Reflexes of the adverbial endings in Arabic There is no evidence for the terminative ending in Arabic; this is perhaps due to the fact that the Proto-Arabic sound change ah# > ā would have caused it to merge with the accusative in several forms of the language. Indeed, in the QCT one would not be able to distinguish between the accusative and the terminative, both being realized as ā. Perhaps the occasional appearance of -ā on diptotes in the QCT reflects an original terminative ending, e.g. Q 2:61 ʾhbṭwʾ mṣrʾ ‘go down to Egypt’, where mṣr is usually a diptote. In this case, however, other explanations are possibe, including uncertainty as to whether this foreign noun was a triptote or diptote - in fact, several reading traditions do treat it as a diptote and read it as miṣra, against the consonantal text. . It has long been recognized that the ending -u on adverbs such as qablu and baʕdu is the reflex of the Proto-Semitic locative ending. The form with final nasalization is found in the preposition ladun ‘at’, lit. ‘by the hand’ < *la-yad-un (Grande 2017). This form is attested in the QCT, Classical Arabic, and in Safaitic. The dual construct form laday is found in Dadanitic, in the Sabaic inscriptions from Nagrān, and in Classical Arabic as well. 2.3.4 Gender As Huehnergard observed, the primary innovation with regard to the feminine ending in Proto-Arabic is the levelling of the allomorph with the vowel /a/ to all nouns, save for some high frequency terms (bintun, ʔuḫtun, etc.). 74 MHGA, v. 2020-1 The feminine ending at was never in word final position in Proto-Arabic and therefore the sound change of at > ah (and later > a) cannot be posited for the earliest ancestor of the Arabic languages. This change did not occur in Safaitic or Hismaic and seems to have affected the later stage of Nabataean Arabic. This sound change is very likely the result of contact with Aramaic and, as such, tends to affect urban dialects of Old Arabic, and is only rarely found in the inscriptions of the nomads. Since the reflex of Proto-Semitic *t was heavily aspirated in Old Arabic [th], the lenition of the stop component left aspiration: *th > h / _#. The at > ah sound change operates in the QCT and 6th c. CE Nabataean Arabic, as evidenced by the Petra Papyri, as well as in all the Nabataeo-Arabic and 6th c. CE Arabic-script inscriptions. In Classical Arabic, the sound change only affects utterance-final feminine nouns, which have in this position lost nunation and final short vowels. Most modern dialects agree with the QCT in that the sound change affects all nonconstruct feminine endings. Nevertheless, some dialects do not descend from such a situation, and indeed preserve the final t in nearly all circumstances (Van Putten 2017). 2.3.5 Number and agreement (in progress) a. The basic, unmarked form of the noun reflects the singular or collective. b. Proto-Arabic continues the Proto-Semitic method of pattern replacement for pluralization, although many of the patterns may reflect secondary developments, see Ratcliffe (1998). External plural endings, ūna/īna (m) and āt- (f) are common on verbal adjectives. The feminine ending āt-, however, can pluralize feminine nouns with or without internal stem changes: ḥamlatun > ḥamlātun ‘military expedition’; ẓulmun > ẓulumātun ‘darkness’. The masculine external plurals very rarely occur on nouns, ʔarḍun > ʔaraḍūna ‘land’. All of these issues are discussed in more detail below. 2.3.5.1 Singulative The ending -at is used to form a singulative from collective nouns. Individuative plurals can be formed from both collectives and singulatives: Classical Arabic baqarun ‘cattle’, 75 MHGA, v. 2020-1 sing. baqaratun ‘a cow’, individuative baqarātun ‘a number of cows’. This system remains active in most Arabic languages and has cognates elsewhere in Semitic, e.g. Mehri: beqerēt ‘a cow’ and beqār ‘cattle, cows’. 2.3.5.2 Adjectival plurals and the external endings -ūna/īna and -āt-. Verbal adjectives in particular originally constituted a separate declension which formed its plural by means of suffixes, ūna/īna (m) and ātun (f). These remain largely intact in Proto-Arabic, although substantivized verbal adjectives will tend to form broken plurals. In Northwest Semitic, Akkadian, and many modern Ethiopian languages, these endings were leveled for all nominal forms. In some modern dialects of Arabic - especially on the periphery such as Cypriot and Central Asian Arabic - the sound plural endings have expanded and replaced broken plural strategies, Cypriot Arabic kalp > kalpát ‘heart’ cf. Old Arabic qulūbun; 2.3.5.3 Broken Plural System Nominal plurals are formed primarily through pattern replacement. While this strategy is not completely predictable there are clear tendencies towards the pairing of certain singular and plural patterns. In some cases, especially with quadriradical nouns, pluralization is generally predictable. The broken plural system is a feature of Arabic, the Modern South Arabian languages, the Ancient South Arabian languages, and Ethio-Semitic, and has been used in the past to argue for a genetic subgrouping comprising these four groupings. The strategy, however, is most likely a Proto-Semitic feature, with clear structural parallels in other Afro-Asiatic languages. Isolated remnants have been identified in Akkadian and Northwest Semitic but their interpretation and significance remains debated, and the reconstruction of the entire plural system of any node of Semitic has not yet been carried out. A key issue is not only the reconstruction of plural patterns as such but cognate singular-plural pairs. Huehnergard and Rubin (2011: 272) point out at least one common plural between Arabic and Akkadian suggesting that it should be reconstructed for Proto-Semitic: Akkadian ṣuḫarû ‘<*ṣuḫarāʾu ‘lads’ and Arabic ṣuġarāʾu (pl. of ṣaġīrun) ‘small, young’. Other, idiosyncratic plurals such as the plural 76 MHGA, v. 2020-1 of the word night, Classical Arabic layālin < *layāliyun, plural of layl (< *laylay) and Gəʿəz layāləy, sg. lelit (<*laylīt) suggest that it is possible reconstruct with confidence the original form plural of some isolated nouns, despite changes in the singular form across languages. Vestiges of broken plural patterns in Northwest Semitic may also support the reconstruction of a particular plurals. Since the system had disappeared in Northwest Semitic, there was less time/opportunity for analogy to operate and produce pattern redundancy. For example, the plural of Hebrew zāḵār ‘male’ is zəḵūr, a cognate of Arabic ḏukūrun (sg. ḏakarun) ‘male’, perhaps suggesting alternative Arabic plural forms like ḏukūratun and ḏukrānun are secondary extensions. Analogical extension is a powerful force in the broken plural system; patterns spread across semantic and morphological groups, and original plural patterns can lend themselves to secondary pluralization. It is not uncommon to find a singular noun with several plural forms - in such cases, it is unclear if a single pattern for can be reconstructed or if such variation existed in the earliest stages. To illustrate, let us consider the word for tooth, Proto-Semitic *sinnum, which gives Classical Arabic sinnun and Gəʿəz sənn. In both languages, the noun has multiple plural forms. plurals of *sinnum Gəʿəz sənan ʔäsnan Arabic ʔasnānun ʔasunnun ʔasinnatun Is it that *ʔansānum is the original plural? And the word was attracted to other pluralization patterns at a later stage? Or is the common ʔaCCāC pattern simply due to chance? In both languages, this pattern is one of the more productive strategies of pluralization. A comprehensive study of this issue across cognate core vocabulary remains a desideratum. The most in-depth study of the broken plural system is Ratcliffe (1998). He demonstrates that singular - plural pairs are not entirely randomly assigned and there are some predictable tendencies across languages families, while at the same time the study highlights the issue I have mentioned in the previous paragraph: a unified broken plural system is not reconstructable for any common ancestor of the languages 77 MHGA, v. 2020-1 of Arabia and Ethiopia. He does however show certain tendencies that unite the Modern South Arabian languages and Ethiopic (p. 208). 2.3.5.3.1 Common strategies for the pluralization of CVCC and CVCVC nouns The general method of forming plurals of these singular types is the use of a pattern with a long vowel, either ū or ā, between the second and third radical. The most common patterns are ʔaCCāC, CuCūC, and CiCāC. A few illustrative examples from forms of Arabic follow as well as a discussion of cognate patterns in other Semitic languages. ʔaCCāC: This pattern is common with CVCC- and CVCVC-type nouns, e.g. Classical Arabic: ḍirsun >> ʔaḍrāsun ‘molar’ waznun >> ʔawzānun ‘scale’ ḥukmun >> ʔaḥkāmun ‘decision’ Safaitic: sfr [sepʰr] >> ʾsfr [ʔaspʰār] ‘writing’ Levantine: ʔalam >> ʔeʔlām ‘pen’ žesem >> ažsām ‘bodies’ This pattern appears to be cognate with Ancient South Arabian ʾCCC plurals although the vowels of this pattern are not recoverable. Gəʿəz ʔäCCaC is also cognate with the Arabic pattern, and on this basis, it seems reconstructable to at least Proto-West Semitic. CiCāC: In addition to CVCC type nouns, this pattern often pluralizes CaCīC adjectives. In Gəʿəz, the cognate CəCaC pattern is used as the feminine singular counterpart of CaCīC rather than a plural. Since the feminine singular is often used in concord with non-human plurals in Arabic, one may speculate on a connection between these two patterns. Classical Arabic: ǧabalun >> ǧibālun ‘mountain’ kalbun >> kilābun ‘dog’ 78 MHGA, v. 2020-1 kabīrun >> kibārun ‘large’ Levantine: kalb >> klāb < *kilāb ‘dog’ kbīr >> kbār ’ < *kibār ‘large In some dialects, however, the plural of the CaCīC adjective is CuCāl - Egyptian Arabic kubār ‘elders’. This pattern is also attested in Classical Arabic, rāʕin >> ruʕāʔun ‘shephard’. Van Putten 2019 has argued that the CuCāC pattern is triggered by emphatic environments. This pattern is impossible to detect in the ancient epigraphy, but likely underlies spellings such as Safaitic: rgl ‘foot soldiers’, where context demands a plural interpretation, e.g. ḫrṣ ʿl-rgl-h ‘he watched over his foot soldiers’. CuCūC: This plural pattern is impossible to detect in the North and South Arabian epigraphic record. It is absent in Gəʿəz - the closest parallel is ʔaCCuC, hagär >> ʔahgur < *ʔahgūr. But vestiges of this pattern are found in Hebrew, with at least one cognate pair mentioned above: Classical Arabic ḏukūrun and Hebrew zəḵûr ‘males’. Classical Arabic: naǧmun >> nuǧūmun ‘star’ qalbun >> qulūbun ‘heart’ quflun >> qufūlun ‘lock’ Psalm Fragment: *ṭayr >> ṭiyūr ‘bird’ < *ṭuyūr Levantine: dars >> drūs ‘lesson’ < *durūs 2.3.5.3.2 CuCuC Classical Arabic kitābun >> kutubun ‘writing’; Levantine safīne >> sufun ‘boat’. The pattern is possibly found in Sabaic CCC plurals but impossible to prove. The pattern seems to be absent in Gəʿəz but may be found in Mehri II-w/y roots, e.g. ḫawl ‘maternal uncles’ from *ḫūl < *ḫuwul,39 cf. Classical Arabic nūqun ‘she-camels’ < *nuwuqun. 39 *ū́ becomes aw in Mehri. 79 MHGA, v. 2020-1 2.3.5.3.3 CaCīC This plural pattern is the opposite of the CaCīC >> CiCāC pair (above) as it pluralizes CiCāC singulars, Classical Arabic ḥimārun >> ḥamīrun ‘ass’. It is also rarely used for CvCC nouns, ʕabdun >> ʕabīdun ‘slave’. As with CuCuC, it is impossible to detect this pattern in the epigraphy. It does not seem to be present in Gəʿəz; however, vestiges of the pattern, sometimes doubly marked with external plural endings, is found in Hebrew, e.g. pesel >> pəsîlîm ‘idol’. As Ratcliffe points out (1998: 157), the pairing of a CvCC singular (pesel < *pasl) with a CaCīC plural is also vestigial in Arabic, isolated to a few common nouns, ʕabdun > ʕabīdun ‘slave’ (above). 2.3.5.3.4 CiCaC/CuCaC Gəʿəz exhibits a plural with a high vowel in the first syllable and a short /a/ in the next which could reflect a merger of both Arabic CiCaC and CuCaC plurals. In Gəʿəz this pattern is used to pluralize CVCC nouns, so ʔəgr >> ʔəgär ‘foot’; ʔəzn >> ʔəzän ‘ear’, while in Arabic it is primarily used for CVCCat- singulars. The quality of the high vowel in the first syllable seems to be determined by the singular in Classical Arabic or the quality of the glide in the second syllable of II-w/y nouns, rukbatun >> rukabun ‘knee’; ḥikmatun >> ḥikamun ‘wisdom’; ḫaymatun >> ḫiyamun ‘tent’. 2.3.5.3.5 CiCCān/CuCCān The pattern with an -ān ending seems to have been originally productive for individuative plurals of higher sentients. A similar plural ending is found in Gəʾəz, where it is restricted to mostly masculine human plurals and does not trigger stem ablaut -ṣadəq >> ṣadəqan ‘just’; liq >> liqan ‘elders’. An -ān termination occurs in Akkadian but there it has a particularizing function unrelated to plurality - šarrāqum ‘thief’ vs. šarrāqānum ‘that particular thief. The suffix can be pluralized as well to denote a specific group, again not necessarily connected to plurality šarrū ‘kings’ vs šarrānū ‘a number of individual kings’. Now, perhaps the Arabic plural patterns CvCC-ān reflect original plural stems - fiʕlun and fuʕlun - with the individuative/particularizing suffix ān. At least the CuCC pattern functions as a broken plural for ʔaCCaC and CaCCāʔ singulars and perhaps the CiCC pattern is an ablaut variant. This individuating/particularizing plural was naturally more often used with higher sentients, like animals and humans. From here the pattern could have spread to other nouns belonging to the same pattern. 80 MHGA, v. 2020-1 Classical Arabic: fārisun >> fursānun ‘horseman’ ġazālun >> ġizlānun ‘gazelle’ fatan >> fityānun ‘youth’ non-animates include: ḥāʔiṭun >> ḥīṭānun ‘wall’ Safaitic: ẓby [ṯabeyy] >> ẓbyn [ṯobyān] ̣ ̣ Levantine: wādi >> widyān ‘valley’ šabb >> šubbān ‘youth’ 2.3.5.3.6 CaCaCat This plural pattern is used mainly for nominalized participles and adjectives referring to persons, so Classical Arabic ṭālibun >> ṭalabatun ‘student’; ḍaʕīfun >> ḍaʕafatun ‘weakling’. This pattern is cognate with Gəʿəz CaCaCt, which is the normal plural of its active participle pattern CäCaCi: säraqi >> säräqt ‘thief’ and of adjectives of the CäCiC pattern (cognate with Arabic CaCīC), ṭäbib >> ṭäbäbt ‘wise’. So, in this case, while the plural patterns are cognate and pluralize the same semantic class of nouns, they do not have cognate singular patterns. The cognate of the Arabic CāCiC participle pattern in Gəʿəz forms adjectives and not participles, e.g. ṣadəq 'righteous’, and pluralize with the suffix -an (see above). Mehri also makes use of this pattern in a similar way to Arabic, ḥōkəm (ḥākim) >> ḥkomət (< *ḥVkamt) ‘ruler’, but it is unclear whether or not such pairs are loans from Arabic. The pattern CuCaCat can form plurals of III-w/y active participles - it is unclear if this is the same pattern with a conditioned shift of a > u in the first syllable or simple another pattern, nāḥin (< *nāḥiyun) >> nuḥātun (< *nuḥawatun). 2.3.5.3.7 CuCCāC This pattern is used to pluralize substantivized active participles of the pattern CāCiC, Classical Arabic: kātibun >> kuttābun ‘scribe’. A related pattern with a short second vowel is also attested CuCCaC, sāǧidun >> suǧǧadun ‘bowers in prayer’. It is unclear 81 MHGA, v. 2020-1 what determines their distribution. These patterns are not clearly attested elsewhere in the Semitic language family and therefore seem to be Arabic innovations (Ratcliffe 1998: 110). 2.3.5.3.8 CuCaCāG If Huehnergard is correct in connecting Akkadian ṣuḫarû ‘lads’ to Classical Arabic ṣuġarāʔu, then this pattern would appear to be reconstructable to Proto-Semitic, and would appear to pluralize singulars of the CaCīC(at) pattern, ḫalīfatun >> ḫulafāʔu ‘deputy’, ḍaʕīfun >> ḍuʕafāʔu ‘weak’, and perhaps also CāCiC, šāʕirun >> šuʕarāʔu, if this does not reflect a secondary attraction to this pattern on semantic grounds. This pattern is not found in Gəʿəz, but Ancient South Arabian provides a parallel, CCCw, hśr >> hśrw */husarāwu/ ‘steersman’. ʔaCāCiC: This pattern is very rare in Arabic and is usually interpreted as a “plural of a plural” or “hyperextended”; it is often applied to biradical nouns: Classical Arabic yadun > ʔayādin (<* ʔayādiyun) and ismun > ʔasāmin ( *ʔasāmiyun). This pattern is rarely used for triradical stems, biʔrun > ʔabāʔiru(n), which gives rise to the place named bāyir in Jordan, attested in the Safaitic inscriptions as ʾbʾr likely /ʔabāʔer/. This plural is found in Gəʿəz with triradical nouns, bagʕ >> ʔabāgəʕ ‘sheep’ but does not form a plural of biradical nouns like ‘hand’, ʔəd >> ʔədaw, ʔäʔdaw. 2.3.5.3.9 Plurals of paucity According to the grammatical descriptions of Classical Arabic, plurals between 3 and 10 are indicated by a prefixed ʔa-syllable and stem changes. It is unclear if this system is an artificial product of Arabic grammar, an innovation in a small subset of Arabic dialects, or a genuine Proto-Arabic (or Proto-Semitic) feature. Plurals of paucity are not productive in the modern Arabic dialects although their patterns are present, and it is impossible to detect such a system in the epigraphic record. Some ʔa-prefixed patterns are found in other Semitic languages where they do not carry this function, e.g. *ʔaCCāC and *ʔaCCiCt in Gəʿəz. This can be interpreted in two ways: either Arabic reappropriated these patterns as paucals or Gəʿəz, as well as the non-Classical forms of the language, have lost this function. 82 MHGA, v. 2020-1 Vestiges of the paucal use of the ʔaCCuC in the modern Arabic dialects can be found in special ‘count’ plurals of a few high frequency nouns, Levantine tišhur ‘months’ used after the numbers 3-10: ḫamis-tišhur ‘five months’ from earlier *ḫamsat-ʔašhur, contrasting with the alternative plural šhūr. ʔaCCuCun: This pattern is supposedly the paucal counterpart to CuCuC plurals: nahrun >> ʔanhurun ‘river’; Levantine šahr >> ušhur < *ʔaśhurun ‘month’. But note that not all ʔaCCuC forms have alternative CuCuC plurals - there is no nuhurun for example - nor do CuCuC plurals form ʔaCCuC plurals of paucity, e.g. sittatu kutubin ‘six books’ and not sittatu ʔaktubin. This plural pattern does not have a cognate form in Gəʿəz and it is impossible to distinguish in the epigraphic record. ʔaCCiCatun: This pattern is supposedly the paucal counterpart of CaCaCatun/CiCaCatun patterns: Levantine rġīf >> irəġfi ‘loaf’ <*ʔaġrifah; Safaitic ʾlh [ʔelāh] >> ʾʾlht [ʔaʔlehat]; Classical Arabic ǧanāḥun >> ʔaǧniḥatun ‘wings’. This pattern finds cognates in other Semitic languages, such as Sabaic ʾCCCt and Gəʿəz ʔaCCəCät. ʔaCCāCun: This pattern, discussed above, is said to be the paucal counterpart to CiCāCun. The pattern is cognate with Gəʿəz ʔaCCaC, ləbs >> ʔalbas ‘clothing’, and is rather common in the modern dialects where no paucal function is apparent. ʔaCCiCāyun: The supposed paucal of CuCaCāyun, this pattern is found in Sabaic ʾCCCw, probably ʔaCCiCāw. Classical Arabic qarībun >> ʔaqribāʔu ‘relative’; Safaitic ʿrḍ [ʕarīṣ́] >> ʾʿrḍy [ʔaʕeṣ́āy] ‘valley’. This pattern is preferred for III-w/y and geminated CaCīC singulars: šadīdun >> ʔašiddāʔu ‘harsh’; ġaniyyun >> ʔaġniyāʔu ‘wealthy’. 2.3.5.3.10 Augmented plurals (preliminary) -at: Both the CiCāC and CuCūC patterns can be augmented with the feminine ending. Classical Arabic: faḥlun >> fuḥūlatun ‘stallion’ ḥaǧarun >> ḥiǧāratun ‘stone’ Levantine: ʕamm >> ʕmūmi < *ʕumūmat- ‘uncle’, ‘kinsman’. 83 MHGA, v. 2020-1 This augmented form in turn can be pluralized, producing CuCūCāt, Classical Arabic futūḥātun ‘conquests’; Levantine bzūrāt ‘seeds’ (sg. bizir). The -at ending can moreover form the plural collective of substantives referring to persons, Classical Arabic: ḥammālun >> ḥammālatun ‘porter’; muslimun >> muslimatun ‘muslims (collective)’. In modern Arabic, the feminine ending added to the genitilic adjectives serves the same function, Levantine: il-libnāniyye ‘the Lebanese people’ (sg. libnāni). Root augmentation 2.3.5.3.11 Plurals of quadriradical nouns/nouns with long vowels CāCiC, CvCvvC(at) The main strategy of pluralization for nouns of these classes is a pattern with a long ā and this seems reconstructable to Proto-West Semitic. Arabic and Gəʿəz, however, differ in the deployment of cognate patterns. The plural patterns CaCāCiC and CaCāCiC(a)t are common to Gəʿəz, Arabic, and Ancient South Arabian, and Modern South Arabian. The third radical is a glide (which becomes a glottal stop in Classical Arabic unless the root vowel is a y) when this pattern is used to pluralize triradical nouns with a long vowel in the final syllable, Classical Arabic ǧazīratun >> ǧazāʔiru; Mehri ḫəlēq > ḫəlōweq ‘dress, cloth’, and Sabaic fʿwl and fʿyl plurals; Gəʿəz gäraht >> gärawəh ‘field’; wəḥiz >> wäḥayəzt ‘river’. In Arabic, CaCāCiC pluralizes singulars of the CvCCvC type, ʕaqrabun >> ʕaqāribu ‘scorpion’, while the t-augmented pattern is mainly used as a collective plural of people, perhaps owing to the feminine gender: ǧabbārun >> ǧabābiratun ‘despot’; mulḥidun >> malāḥidatun ‘heretic’. Gəʿəz, however, has a polarity rule -- the t-augmented plural is used for singular nouns without the feminine ending and the CaCāCiC pattern is used for singular nouns with the feminine ending (Dillmann 1903:244): ʕäqrab >> ʕäqarəbt ‘scorpion’; ʔäṣbaʕt >> ʔäṣabəʕ ‘finger’. A similar tendency seems apparent in Sabaic, where mfʿl is the plural of both singular patterns mfʿl and mfʿlt while the t-augmented plural mfʿlt is used only for singular mfʿl. 84 MHGA, v. 2020-1 Mehri seems to show signs of an original polarity rule. CvCCvC nouns pluralize as in Arabic, məḫbāṭ >> məḫáwbəṭ ‘cartridge’; ḫəlēq > ḫəlōweq ‘dress, cloth’. The taugmented form is found as well, most often with masculine singular nouns (Rubin 2018: 95): hērək >> həráwkət < *harāwikt (?) ‘thief’; skayn >> skáwnət <*sakāwint (?) ‘knife’. The pattern CaCāCīC is used primarily for quadriradical nouns with a long vowel between C3 and C4, Classical Arabic miftāḥun >> mafātīḥu ‘key’. This pattern is not found in Gəʿəz and is not distinguishable in the epigraphy. The pattern CawāCiC in Arabic usually pluralizes CāCiC singulars with the insertion of the glide w in between C1 and C2. In Gəʿəz, this pattern is mainly used with quadriradical nouns with a glide as C2, and as such it should not be construed as a separate pattern but simply the CaCāCiC(t) pattern applied to such root: kokäb (< *kawkab) >> käwakəb ‘star’. In Arabic, this pattern however is used in contexts without an etymological glide: fārisun >> fawārisu ‘horseman’. It is possible that this is an extension of the CaGāCiC pattern and constitutes an innovation. A parallel is perhaps found in Harsusi ġōrəb > ġəwōrəb ‘base of neck’, but more often, the plural of CvvCvC nouns geminates the second root radical: ġābər >> ġəwabber ‘pregnant she-camel’. CaCāCiCu plurals of III-w/y nouns: While the CaCāCiC pattern is usually a diptote, when it is applied to III-w/y roots, it is triptotic, reflecting the III-w/y declension: layālin ‘nights’ < *layāliyun and not the expected layāliyu, which would produce layālī; marṯiyatun >> marāṯin ‘elegy’; ʔafʕā >> ʔafāʕin ‘viper’. A related pattern CaCāCaW applies to CvCCay and CvCCāy, as well as III-w/y CaCaC and CaCīCat nouns, and could be cognate with Sabaic CCCw plurals. The final awV is realized as /ā/ in Classical Arabic, thus: Classical Arabic: manan (< *manayun) >> manāyā ‘fate’ (<*manāyawu), from the root mnw hadiyyatun >> hadāyā ‘gift’ fatwā (< *fatway) ‘legal opinion’ > fatāwā ʕaḏrāʔu (* ʕaḏrāyu) ‘virgin’ >> ʕaḏārā 85 MHGA, v. 2020-1 This pattern is extended to the isolated noun yatīmun ‘orphan’, producing yatāmā < *yatāmaw. The reasons for this are unclear. 2.3.5.3.12 a-insertion external plurals The normal mode of pluralizing CvCCat nouns is with the external plural ending āt and the insertion of /a/ between C2 and C3. The inserted vowel may harmonize with a preceding high vowel. The noun *ʔarṣ́un ‘earth’ forms its plural with the external masculine plural endings - a polarization reminiscent of the t-augmented plurals of masculine bases in quadriradical nouns. The same is witnessed in the plural of sanatun >> sinūna ‘year’, cf. Hebrew šānîm. Classical Arabic: niʕmatun >> niʕamātun / niʕimātun ‘favor’ ẓulmun >> ẓulamātun / ẓulumātun ‘darkness’ ʔarḍun >> ʔaraḍūna ‘land’ This strategy of pluralization is cognate with the normal mode of pluralizing CVCC nouns in Northwest Semitic, Ugaritic kalbu > kalabūma; Hebrew ʕébeḏ > ʕăbāḏīm and is therefore reconstructable to at least Proto-West Semitic (and likely Proto-Semitic despite its absence in Akkadian). External plurals, however, can be found on nouns without a-insertion: ʔahl-ūna ‘inhabitants’ and rarely ʕabdūna ‘slaves’, in this case it is said to refer to an “individual plural” of living beings. In many modern dialects, the external masculine ending can be found on broken plurals as well: Levantine ḏa ̣ ʕīf >> ḏʕ̣ āf >> ḏʕ̣ āfīn ‘weak’; ṣāḥeb >> ṣḥāb >> ṣḥābīn ‘friend’. 86 MHGA, v. 2020-1 2.3.5.4 A Note on Agreement Agreement in Arabic will be treated in more detail in a future iteration of this book in the Syntax section but a few remarks here are in order as the issue is directly relevant to number. Most Semitic languages exhibit natural agreement between a head and modifier - masculine plural nouns, regardless of animacy, take masculine plural agreement and the same for feminine nouns. Agreement in Arabic, on the other hand, depends on whether or not the noun is human or non-human and/or singular or collective. No comprehensive study of agreement in Semitic has been carried out so it is unclear if this system is an Arabic innovation or if it reflects a much older alignment easily lost in the other Semitic languages. In modern written Arabic, the rule is inanimate plurals take feminine singular agreement: kutubun kaṯiratun ‘many books’; ʔinna l-kilāba ǧāʔat ‘the dogs have indeed come’. In Safaitic, both natural and feminine singular/plural agreement is attested for inanimate plurals but there are enough enough examples to divine a system. Both methods of agreement are also permitted in many modern dialects, Levantine kitob ktār and kitob ktīre. The poetry and QCT as well as many modern dialects reveal a finely tuned system of agreement that reflects distinctions of individuation. This system was uncovered and described in excellent detail in the important study of Bettega (2019). His results are summarized in the following table. Bettega (2019, table 6) Adjective Morphology (in progress) 87 MHGA, v. 2020-1 2.3.5.5 The Dual (noun) The dual ending in Proto-Semitic was -āna (nominative) and -ayna (oblique), cf. Akkadian -ān, īn; Ugaritic -āma, -ēma. The main development in Arabic is the dissimilation of the final vowel to i, producing āni and āyni. Like the external masucline plural, the final n is a marker of the unbound state and drops when duals are placed in construct. The dual was likely applied to any noun, although it has been greatly restricted in its distribution in most languages, remaining mostly on natural pairs in Hebrew, for example. The dual in the modern dialects remains productive, although case inflection is lost and the oblique is generalized, īdēn ‘two hands’, ktābēn ‘two books’. In some cases the dual form has taken over a plural, ʕēnēn ‘eyes’, where then a new dual suffix, apparently clipped from feminine nouns, is used to form the true dual: ʕēntēn ‘two eyes’. The former has been labelled by Blanc (1970) the ‘pseudodual’. 2.3.6 Definite Marking Proto-Semitic and Proto-Central Semitic lacked a definite article and this situation was inherited by Proto-Arabic. The definite article is lacking in the Hismaic inscriptions and marginally in Safaitic, indicating that this feature cannot be reconstructed to the protolanguage (Ahmad Al-Jallad 2018b). Safaitic: ḥl dr snt ... /ḥalla dawra sanata.../ ‘he camped in this place the year...’ w lm yʿwr sfr /wa-lam yoʕawwar sephra/ ‘and may the writing not be effaced’ Hismaic: w ḫṭṭ gml /wa-ḫaṭṭaṭa gamala/ ‘and he carved the camel’ (next to a rock drawing of a camel) Instead, both these languages attest an h element with a demonstrative force (see below). The definite article appears to have spread to Arabic through contact with Canaanite in the southern Levant. The earliest article form is ha, with gemination of the first consonant of the following word. A prefixed article of this type is attested in 88 MHGA, v. 2020-1 cuneiform transcription from the 8th c. BCE from ancient Dūmah, were the word ‘shecamels’ is spelled AN-NA-QA-A-TE, perhaps transcribing the form ʔan-nāqat- or han- naqāt- (Livingstone 1997). By the middle of the 1st millennium BCE, the ha- demonstrative, perhaps motivated by contact with Canaanite, had developed into a full-fledged definite article, with the agreement patterns found elsewhere in Central Semitic. Towards the end of the 1st millennium BCE, the ʔal- article is attested in the Nabataean dialect. This form of the article is also marginally attested in the Ḥigāz, in the substrate of the Dadānitic inscriptions, although the chronology of those texts is unclear. From a geographic perspective, then, the ʔal-article seems to be a western form, first emerging in the NW Arabia. In addition to this, the Nabataean ʔal-article does not appear to exhibit assimilation of the l to following coronals. A few personal names, however, indicate that other article forms existed in the Nabataean realm, for example, ʿbdʾbʿly /ʕabdo-ʔab-baʕle/. While the ʾ article is attested in Safaitic as well (see below) and found in modern vernaculars, we must be careful not to draw far reaching conclusions from these marginal Nabataean examples. In the case of ʿbdʾbʿly, the scribe may have simply omitted the l by mistake. The dialects of the Ḥarrah exhibit other article forms. The definite article ʾ-, that is a prefixed glottal stop, is not infrequently attested. This seems to reflect a form ʔan- with assimilation of the n to the following consonant. The ʔal article is also attested, but rather infrequently. It is possible that some examples of the ʾ-article reflect the ʔal article with assimilation of the coda, but other times this interpretation is impossible, for example Safaitic ʾbkrt = /ʔab-bekrat/ ‘the young she-camel’. The etymology of the ʔal-article is disputed. The main ideas are that it either 1) results from a dissimilated form of the ʔan (<*han) article or 2) derives from the hal presentative, which is attested as an article in the Thamudic F inscriptions, for example. The ʔam-article is attested only once, in an unpublished pre-Islamic Arabic-script inscription from the Tabūk region, ʾmʿm /ʔam-ʕām/ ‘the year’. This article form is no doubt the result of the assimilation of the ʔan-article to labial consonants. 89 MHGA, v. 2020-1 Article form in the QCT The QCT exhibits the non-assimilating article, similar to Nabataean, but it is unclear if this is simply an orthographic practice or if it in fact reflects a phonetic reality in the Qur’anic dialect. At least some evidence from later sources suggest that the nonassimilating article survived well into the Islamic period and may therefore have been part of the original dialect of the QCT. Al-Sīrāfī reports in his commentary on Sibawayh the following from Al-Kisāʾī: :‫ قال‬.‫ أو الراء والنون‬،‫حكى الكسائي أنه سمع العرب تبين الالم يعني الم المعرفة عند كل الحروف إال عند الالم مثلها‬ ‫ ولم أسمعها من العرب‬،‫ الصامت‬:‫يقول بعضهم‬ ‘Al-Kisāʾī stated that he heard the Arabs preserving the lām, meaning the lām of the article, before every phoneme except the lām like it, or the r and n. He said: some of them say al-ṣāmitu, but I have not heard this from the Arabs’.40 Article form in early Islamic Arabic Greek transcriptions from the first Islamic century indicate that the ʔal-article assimilated to coronals. Article form in the Psalm Fragment This document is perhaps the latest written example of the non-assimilating ʔal-article. Since Arabic orthography does not seem to influence the transcription system of this document in other cases, it is very likely that its spelling reflects a phonetic reality. Article forms in Modern Arabic Most modern Arabic dialects exhibit a definite article strikingly similar to Classical Arabic, but there are notable exceptions. In Egypt, for example, the coda of the article assimilates to velar consonants, so ik-kalb ‘the dog’ <*il-kalb. The variety of ancient article forms witnessed in the pre-Islamic southern Levant survives in southwest Arabia. There one may still hear the am-article, and less frequently the an- and aarticle, with gemination of all following consonants. While it is common to regard these 40 I thank H. Sidky for this reference. See: https://books.google.com/books?id=schwDwAAQBAJ&pg=PT451&lpg=PT451#v=onepage&q&f=false. 90 MHGA, v. 2020-1 forms as loans from Ḥimyaritic, we must stress here that there is no epigraphic evidence from South Arabia to suggest the existence of a prefixed nasal article. These article forms are true Arabic variants, having nothing to do with Sabaic or any other Ancient South Arabian language. Vestiges of this diversity are frozen in certain lexical items elsewhere. For example, the am-article is encountered in the widely attested word for ’yesterday ’imbāreḥ cf. Classical Arabic al-ḅāriḥah. Loans into languages that were in contact with early Arabic sometimes show variant article forms. Awjila Berber for instance has borrowed the word for ‘needle’ as tanəbret; the first t is part of Berber noun morphology – thus the word for ‘the needle’ must have originally been an-ʔibrat (Van Putten and Benkato 2017). 2.3.6.1 Assimilatory Patterns The *han article: The h-definite article exhibits consistent assimilation of its n-code to the following consonant in northern Old Arabic. The few exceptions occur in a handful of inscriptions written by men from North Arabia, in particular, from the Ḥwlt tribe. Thus, it would appear that the non-assimilated form was native to that region in pre-Islamic times. This would accord with the situation attested in the Dadanitic inscriptions, which attested the form hn before words beginning with a laryngeal, e.g. hn-ʔʕly ‘the highest’. There are so far no examples of the non-assimilated ʔan-article. The northern dialects of Old Arabic did not assimilate the coda of the ʔal-article to coronals, thus we have in transcription in the Petra Papryri αλνααρ /alnahar/, αλσουφλη /alsuflē/, αλσιρα /alṣīrah/, αλσουλλαμ /alsullam/, etc. The same is found in Nabataean and Safaitic inscriptions, e.g. Namarah ʾltg /ʔal-tāg/ ‘the crown’; Safaitic ʾlnbṭy /ʔal-nabaṭeyy/, ‘the Nabataean’. The coda, however, is assimilated in the Graeco-Arabic inscription A1 αδαυρα /ʔad-dawra/ ‘this place’ and in the Dadanitic inscriptions of the northern Ḥigāz, ʾs¹fr = /ʔas-sifr/. The Rbbl epitaph of Qaryat al-Fāw also exhibits an assimilating article, ʾs¹my = /ʔas-samāy/. There are several ways to understand this distribution. It is possible that the assimilation of the coda is archaic, and reflects the levelling of the ʔan (<*han) article to words beginning with coronals while the ʔal (< *hal) allomorph was generalized in 91 MHGA, v. 2020-1 other situations. In this case, the Nabataean distribution would be innovative, resulting from the levelling of one form to all environments. Likewise, those dialects exhibiting the ʔa-article in all environments would reflect the opposite, innovative development. Otherwise, one could take the non-assimilating ʔal-article as original and understand its assimilation to coronals as innovative. The latter solution however relies on an adhoc change, namely, the assimilation of l. The onset of the article was originally a true consonant, h and then ʔ. The loss of the glottal stop in this position is not as frequently attested as its preservation. In the Nabataean inscriptions, one sometimes encounters the loss of the alif of the article when it is preceded by a construct noun. The same is true in Safaitic, for example, whblh, which is given in Greek transcription as Ουαβαλλας /wahb-allāh/, and is found in the Rbbl bn Hfʿm epitaph of Qaryat al-Fāw, wlʾrḍ /wal-ʔarḍ/. In most modern dialects, the definite article is an underlying l, which takes its vowel, either before it or after, from its context: Levantine il-walad vs. li-wlād. In the Najdi dialects, and elsewhere, the vowel of the article remains /a/, although it can be elided when contiguous with another vowel, Najdi al-bēt ‘the house’ vs. fī l-bēt ‘in the house’. 2.4 Morphology of the demonstratives and relative pronouns 2.4.1 Demonstrative particles and pronouns In Old Arabic, the most common demonstrative element is a prefixed h-, attested in Safaitic and Hismaic and which is recorded by the Arabic Grammarians and is common in the modern vernaculars, e.g. Levantine ha-l-walad ‘this boy’. The h- prefix does not inflect for gender and number and so following Pat-El (2009), it is probably wrong to classify it as a pronoun. There can be no doubt that the ha- demonstrative is related to the article; however, the two have a different syntax. At least in Safaitic, the hdemonstrative can precede the first term of a genitive construction, thus: HCH 79: h- dmyt zmrt DEM- image.CNST flute-playing girl ‘this image of a flute-playing girl’ 92 MHGA, v. 2020-1 The demonstrative pronominal series exhibits a reduced inflectional paradigm, originally expressing only three categories, masculine and feminine singular, and common plural. There is no evidence for case inflection in the demonstratives. 2.4.1.1 Proximal demonstratives Masculine singular Proto-Arabic Safaitic Hismaic 6th c. Arabic-script inscriptions *ḏā ḏ */ḏā/ ḏh, ḏʾ */ḏā-h(a)/ or */ḏāʔ(a)/ dʾ */ḏā/ Developments: Only the forms lacking the hā prefix are attested in the pre-Islamic period, at least until the QCT. The hā-forms may have been a southern variant, perhaps beginning in the Ḥigāz. Support for this possibility may be found in Dadānitic, which attests a dual demonstrative hḏh ‘these two’ perhaps */hāḏ-ā/. It should be said though that the singular forms lack the hā-prefix. While many vernaculars today only exhibit the form with a hā prefix, the direct reflex of Proto-Arabic *ḏa is attested across Arabia and in Egypt, where it is realized as ḏā and da, respectively. The Hismaic form terminating with a h may be the masculine equivalent of the QCT feminine form hdh */hāḏīh/, Classical Arabic hāḏīhi. Feminine Singular ProtoArabic Safaitic Namarah JSLih 384 QCT Classical Southwest Arabic Arabia *tī t */tī/ ty /tī/ ʾlt */ʔallatī/ tlk /tilka/ (distal) tīka (distal) tā/ tīh etc. Developments: The principle of archaic heterogeneity motivates us to reconstruct the t-forms for the Proto-Arabic feminine singular series, even though most Arabic languages have levelled the ḏ-onset from the masculine to the feminine. The Namarah inscription, the Classical Arabic distal, relative pronoun (ʔallatī), all support the 93 MHGA, v. 2020-1 reconstruction of the vowel as ī, while the ā reflexes stem from the levelling of the vowel of the masculine singular to the feminine. Most modern vernaculars exhibit forms that go back to the element *ḏī, often with the prefixed hā-demonstrative, which results from the leveling of the masculine onset to the feminine, e.g. Najdi (hā-)ḏī; Lebanese haydi; Egyptian di; etc. The addition of the ī suffix, signifying the feminine singular, to the demonstrative prefix hā produces hāy < *hā-ī in many modern dialects, Levantine hāy il-binit ‘this girl’. Common Plural ProtoArabic Safaitic QCT Classical Arabic Rigāl Almaʿ Lebanese *ʔulāy ʾly */ʔolāy/ hwlʾ /hāw(o)lā/ hā-ʔulāʔi wula hawle Developments: The plural base does not inflect for gender and, at the proto-Arabic stage, lacked the hā-prefix. The final -i of the Classical Arabic form is likely a euphonic vowel, meant to prevent the shortening of the ā in a closed syllable. Many modern dialects have created new plural demonstratives by combining what was analyzed as the singular base, hāḏa and the plural demonstratie ula < *ʔulāy, Levantine hāḏōla < *hāḏa-ulā; Egyptian dōl < *ḏā-ula; Najdi hāḏōl. 2.4.2.2 Distal demonstratives The distal/anaphoric demonstrative use of the 3rd person pronouns has disappeared, replaced by the modification of the proximal demonstratives with the element -ka. At the Proto-Arabic stage, the distal bases were simply modified by this element, producing: MS ḏāka FS tīka 94 MHGA, v. 2020-1 CPL ʔulayka/ʔulāyika The differences between the Classical Arabic by-forms ʔulā and ʔulāʔi may stem from different ways of resolving the closed super-heavy syllable produced by the addition of the distal ka to this form. Old Ḥigāzī: The QCT uniquely exhibits a distal form with the particle li intervening between the demonstrative pronoun and the deictic ka, producing forms like ḏālika, tilka < *tīlika, and ʔulāyika. These forms could be Old Ḥigāzī innovations, attested also in the Damascus Psalm Fragment and the early Islamic papyri. They become, perhaps on account of such documents, the main forms employed in Classical Arabic, although the grammatical tradition provides many more options. The QCT and some modern dialects in Southwest Arabia also reanalyze the deictic element -ka as a pronominal suffix, giving rise to addressee agreement, producing forms like ḏālikum when addressing a group. Such forms are not found in other forms of Arabic and do not seem to be reconstructable to Proto-Semitic. It is impossible to prove if these are innovations of Old Ḥigāzī or simply an areal feature of West Arabia. 2.4.2 Relative Pronouns Proto-Arabic had several strategies of subordination, including the use of a relative particle/pronoun. The relative pronoun is derived from the demonstrative, but with one key difference – the feminine singular form was based on the masculine, thus reducing the paradigmatic asymmetry. Proto-Arabic Sabaic Ugaritic Masculine singular *ḏū (nom) *ḏī (gen) *ḏā (acc) ḏ d Feminine singular *ḏātu (nom) *ḏāti (gen) *ḏāta (acc) ḏt dt Masculine Plural *ʔulū (nom) *ʔulī (obl) ʾlw (nom) ʾly (obl) dt 95 MHGA, v. 2020-1 Feminine Plural *ʔulātu (nom) *ʔulāti (obl) ʾlt dt (?) The Proto-Arabic relative pronoun series is most faithfully preserved as a relativedeterminative pronoun (i.e. ḏū l-qarnayni ‘he of the two horns’) in Classical Arabic and the QCT. The former naturally exhibits some allomorphy. Classical Arabic relative-determinative pronoun Masculine singular *ḏū (nom) *ḏī (gen) *ḏā (acc) Feminine singular *ḏātu (nom) *ḏāti (gen) *ḏāta (acc) Plural *ʔulū (nom)/ḏawū *ʔulī (obl)/ḏawī From these forms, we may understand the development of the relative pronoun series in later Arabic languages. Safaitic: Safaitic derives a new plural form based on the onset of the singulars, producing ḏawū/ (nom), /ḏawī/ (gen). This is similar to the by-form ḏawū/ḏawī attested in the Classical Arabic relative determinative series. Based on word-boundary spellings, the singular continued to inflect for case, attesting a ḏū (nom) and ḏī (gen). Nabataean: Case inflection in the Nabataean relative disappeared, resulting in dw for all situations, e.g. ʿbddšrʾ /ʕabdo-ḏū-śarē/ ‘servant of Dusares’. The other forms are not attested. Modern Vernaculars: A number of modern Yemeni dialects as well as those of the Maghreb exhibit a non-inflecting ḏ-relative pronoun, ḏī in Yemen and simply d- in the Maghreb. These go back to the generalization of the masculine singular form. The ḏū of Ṭayyiʔ: The generalized ḏū is ancient. The Arabic grammarians were aware of such a form, usually placing it in Yemen and in the dialect of Ṭayyiʾ, whose territory was in the Najd, in the area of Ḥāʾil. 96 MHGA, v. 2020-1 Definite-marked relative pronoun In some modern dialects of the Maghreb, we find iddi < *ildī, which appears to be the relative base *ḏī preceded by the definite article, *ʔalḏī and *ʔaḏḏī. A similar form is attested in Safaitic, e.g. hḏ */haḏḏī/. Old Ḥigāzī Proto-Central Semitic had a portmanteau demonstrative pronoun comprising three elements, han + la + demonstrative (Huehnergard 1995). Ugaritic Hebrew Masculine singular hnd hallazê Feminine singular hndt hallazû Plural NA Old Ḥigāzī grammaticalized this demonstrative into a relative pronoun, replacing the older relative series (although the older forms survive as relative-determinatives). The oldest attestation of this feature occurs in the Dadanitic inscription JSLih 384, which attests the feminine singular ʾlt = ʔallatī. The plural form is difficult to reconstruct. Rabin (1951) suggests that this form, which is usually pointed ʔallāʔi, may reflect the original common plural of this series. If this is correct, then it is possible that the original plural was ʔallay, which would produce the QCT form ʾly. The plural was eventually given adjectival endings, producing the familiar forms ʔallaḏīna and ʔallawāti/ʔallāti. Some dialects, it is said, even extended case inflection to these forms (the demonstratives originally did not inflect for case), producing, for example, ʔallaḏūna in the masculine plural. This process gives the familiar Classical Arabic/QCT paradigm. QCT Classical Arabic JSLih Psalm Fragment Masculine singular ʾldy */ʾallaḏī/ ʔallaḏī NA ελλεδι /elleḏī/ Feminine singular ʾlty */ʾallatī/ ʔallatī ʾlt */ʾallatī/ NA Masculine plural ʾldyn */ʾallaḏīn/ ʔallaḏīna NA NA 97 MHGA, v. 2020-1 Feminine plural ʾly */ʾallay/ - ʾlt */ʾallāt/ ʔallāti/ʔallawāti NA NA Modern Vernaculars: Most modern vernaculars use a relative pronoun that goes back to the ʔalla-series, mostly ʔilli/ʔalli. The etymology of this form is uncertain. It may be the result of the generalized common plural form *ʔallay (Stokes 2018) or it may be the result of the loss of the final syllable of a generalized ʔallaḏī. The masculine singular form ʔallaḏī is generalized in many modern dialects in Yemen and, in former times, across the Arabic-speaking world; it is common in the so-called Middle Arabic texts, where it does not inflect for gender or number. These forms likely reflect a dialectal reality rather than some artificial medial form, between dialectal ʔilla and the fully inflecting Classical Arabic ʔallaḏī, etc. Dual forms: It is difficult to know whether or not the dual relative pronouns are reconstructible to Proto-Arabic. Their forms clearly draw on nominal morphology, and would appear to be a rather late extension of the dual ending of nouns to the demonstrative. 98 MHGA, v. 2020-1 III The Verbal System 3.1 Prefix Conjugation Proto-Semitic had two finite verb stems, yaqtul, which expresses the preterite, and yaqattal, a non-past durative/imperfective.41 Person-number-gender is indicated by prefixes and suffixes. The paradigm is as follows: Preterite Imperfective 1 ʔaqtul naqtul ʔaqattal naqattal 2m taqtul taqtulū taqattal taqattalū 2f taqtulī taqtulna taqattalī taqattalna 3m yaqtul yaqtulū yaqattal yaqattalū 3f taqtul taqtulna taqattal taqattalna Proto-Semitic verbs in subordinate clauses could take two suffixes, *-u and *-na > Assyrian ni. The *-na ending also occurs on verbless clauses, indicating that it was a clitic. Proto-Central Semitic seems to have grammaticalized these endings on the preterite to form a new, non-paste tense, yaqtulu. Retsö has argued that the final -u should be identified with the locative adverbial ending. The use of locative constructions to form the durative aspect is widely attested in the world’s language, and, in a way, foreshadows modern Arabic forms with the prefixed bi- (on this, see below). The Proto-Central Semitic non-past continues into Arabic unchanged: Proto-Central Semitic Proto-Arabic Ugaritic 1s *ʔaqtulu *ʔaqtulu ʔaqtulu 2ms *taqtulu *taqtulu taqtulu 2fs *taqtulīna *taqtulīna taqtulīna 41 For a reconstruction of the Proto-Semitic verbal system, see for example, Huehnergard 2004; Stephan Weninger 2011, and the references there. 99 MHGA, v. 2020-1 3ms *yaqtulu *yaqtulu yaqtulu 3fs *taqtulu *taqtulu taqtulu 1p *naqtulu *naqtulu naqtulu 2mp *taqtulūna *taqtulūna taqtulūna 2fp *taqtulna *taqtulna taqtulna 3mp *yaqtulūna *yaqtulūna yaqtulūna 3fp *taqtulna *yaqtulna taqtulna The original Proto-Semitic preterite survives in a few frozen constructions, in negation following lam, *lam yaphʕal ‘he did not do’ and *lamma yaphʕal ‘he has not yet done’, and in the conditional construction *ʔin yaphʕal ‘if he had done’. 3.1.1 The vowel of the prefix The vowel of the prefix conjugation is determined by the thematic vowel of the stem (Barth-Ginsberg Law). If the theme vowel is high, the prefix vowel is /a/, and if the theme vowel is /a/, the prefix vowel is /i/. This distinction was lost in Classical Arabic, where the /a/ vowel was leveled in all circumstances, e.g. yaqtul, yasmaʕ; however, in some modern dialects of Arabic the original distribution obtains, e.g. Najdi yaktib, yismaʕ and was apparently active in the eastern Tamīmī dialects, termed taltalah by the Arabic Grammarians. The alternation seems active in Old Arabic as well, in so far as one can tell from Greek transcriptions, A1 ειραυ /yirʕaw/ ‘they pastured’ vs. Ιαμλιχος, a personal name from the prefix conjugation /yamlik/. The first person singular of the modern vernaculars that continue to exhibit BarthGinsberg’s law do not exhibit any vowel alternation in the 1st singular prefix. The Classical Arabic form ʔiḫālu ‘me thinks’ may, therefore, in fact be a loan expression, perhaps from some other Arabian language. Thus, it is possible that Proto-Arabic lowered the original *i vowel to /a/ on account of the initial glottal stop of the prefix. This would be similar to the lowering of the theme vowel in verbs with gutturals, including ʔ. Proto-Arabic indicative prefix conjugation CCuC CCiC CCaC 100 MHGA, v. 2020-1 1s *ʔaqtulu *ʔakbisu *ʔasmaʕu 2ms *taqtulu *takbisu *tismaʕu 2fs *taqtulīna *takbisīna *tismaʕīna 3ms *yaqtulu *yakbisu *yismaʕu 3fs *taqtulu *takbisu *tismaʕu 1p *naqtulu *nakbisu *nismaʕu 2mp *taqtulūna *takbisūna *tismaʕūna 2fp *taqtulna *takbisna *tismaʕna 3mp *yaqtulūna *yakbisūna *yismaʕūna 3fp *taqtulna *yakbisna *yismaʕna 3.1.2 Irrealis Mood inflection Volitive/Jussive: The volitive, the so-called Jussive, continues in form the ProtoSemitic preterite. It is usually preceded by the asseverative li- in the QCT and Classical Arabic. Exceptions occur when it is the second member of a chain of modal verbs, as in the famous opening line of the Muʕallaqah of Imriʾi l-qays, qifā nabki ‘stop you both, let us weep’. The volitive can occur without the asseverative in Old Arabic (Safaitic) and the modern dialects as well. Volitive with asseverative Classical Arabic: fal-yafʕal ‘let him do’ Safaitic: f-l-yʿwr m-ʿwr /phal-yoʕawwar maʕ-ʕawwara/ ‘may whosoever effaces (this writing) be made blind’ Without asseverative Safaitic: h lt yslm /hā-llāt yeslam/ ‘O Allāt, may he be secure’ Levantine: yəftaḥ il-bāb ‘let him open the door’ 101 MHGA, v. 2020-1 While the volitive use of the prefix conjugation remains intact in the modern vernaculars, in most cases the ancient form has disappeared. This is clearly seen in medial weak verbs. The volitive of these contains a medial short vowel, e.g. Classical Arabic yaqul ‘let him say’ vs. yaqūlu ‘he says’. Had the modern vernacular volitive come from the ancient form, we’d expect in, say Damascus Arabic, **yəʔol rather than the attested yəʔūl. The latter form, in light of other members of the paradigm, must come from the subjunctive form (see below), *yaqūla. Subjunctive: The subjunctive, at least its exact distribution, appears to be an innovation of Arabic. It is restricted to subordinate clauses, either complements introduced by *ʔan or result clauses following *pha-. The etymology of this termination is unclear; a final -a is attested in subordinate clauses in Old Assyrian and may be cognate with the West Semitic form. Most scholars have connected it with the cohortative of Hebrew, ʔal ʔēbûšâ ‘let me not be ashamed’; ʔezrəʕâ ‘let me sow’. While the shift from volitive > subjunctive is not too problematic, there remains the problem of connecting Hebrew â, which must go back to *ah, to Arabic -a. It would instead seem that the cohortative in Hebrew should be connected with the directive ending and, hence, be equated with the sequence li-+volitive. 3.1.3 Mood in Old Arabic The volitive must be inferred syntactically, e.g. with verbs following lam. No morphologically distinct forms have yet been attested. The subjunctive is morphologically distinct in Old Arabic, but the orthography only permits its detection in III-w/y verbs. Indicative Subjunctive Safaitic ydʿ /yadʕī/ reads’ ‘he nngy /nangeya/ ‘that we may be saved’ Hismaic ybk /yabkī/ weeps’ ‘he ygzy /yagzeya/ ‘that he may fulfill’ 3.1.4 Mood in the QCT 102 MHGA, v. 2020-1 The loss of final short vowels in the QCT wreaked havoc on the mood system, setting the stage for its eventual collapse. Based on the consonantal text, the following system seemed active (3rd person): Strong verbs Indicative Subjunctive 3ms yaqtul 3fs taqtul 3mp yaqtulūn Jussive yaqtulū 3fp yaqtul(i)n II-w/y Indicative Subjunctive Jussive 3ms yaqūl yaqul 3fs taqūl taqul 3mp yaqūlūn yaqūlū 3fp yaqul(i)n III-w/y Indicative Subjunctive Jussive 3ms yabnī yabn 3fs tabnī tabn 3mp yabnūn 3fp yabnū yabnīna 3.1.5 Mood in Classical Arabic The modal system of Classical Arabic continues virtually unchanged the system reconstructed for Proto-Central Semitic and hence Proto-Arabic. 3.1.6 Modal alignment in the modern vernaculars 103 MHGA, v. 2020-1 The modern modal system emerges from a situation similar to that attested in the QCT. The subjunctive and volitive merge in form and function to either the subjunctive or indicative. This can be detected in the conjugation of medial and final weak roots as well as with the masculine plurals and 2nd feminine singular. Merger to: strong II-w/y Subjunctive, Levantine yiktibū < *yaktubū yiqūl < *yaqūla Indicative, Qəltu yəktəbūn < *yaktubūna yəqūl < *yaqūlu III-w/y yibnī < *yabniy < *yabniya yəbnī < *yabnī A new way of marking the indicative/durative emerges: modal prefixes. The indicative continues to be the marked form. The following prefixes and their etymologies are common: Mesopotamian: qa and da < *qāʕidā, active participle ‘sitting’ Levantine: bi < preposition bi- ‘in’, ‘at’, ‘with’ Maghrebine: ka < active participle, *kāyin ‘being’ The modal use of the unmarked form continues, although it can optionally be modified by modal verbs, most often the imperative ḫallī ‘let’. The energic: A final mood of the prefix conjugation is attested, the so-called energic, which consists of two forms, a short form with the termination -an and a long form with -anna. These forms are not yet attested in Old Arabic nor are they known in the modern vernaculars. They do, however, seem to be archaic with cognates in other Semitic languages. A connection with the Akkadian ventive am has been suggested (Hasselbach 2006). 104 MHGA, v. 2020-1 3.2 Suffix conjugation The West Semitic suffix conjugation derives from a predicative adjective construction in Proto-Semitic with a clitic nominative pronoun (Huehnergard 1987). In West Semitic, a fientive class developed with an a-theme vowel in place of the i/u of the stative adjective. Proto-West Semitic Fientive: *qatalku ‘I have killed’ Proto-West Semitic Stative: *kabidku ‘I am heavy’ *kaburku ‘I am grown’ Proto-Arabic levelled the feminine plural termination from the prefix conjugation to the suffix in the 3FP and 2FP. In addition to this, it leveled the t-onset of the 2nd person pronominal suffixes to the first. Finally, the vowel of the pronominal suffix of the 1cp was leveled with the possessive suffix, changing *nū to nā. Innovative forms are in bold. Proto-Central Semitic Proto-Aabic 1CS *waṯabku waṯabtu 2M *waṯabta waṯabta 2F *waṯabti waṯabti 3M *waṯaba waṯaba 3F *waṯabat waṯabat 1CP *waṯabnū waṯabnā 2MP *waṯabtum(ū) waṯabtum(ū) 2FP *waṯabtin(ā) waṯabtin(na) 3MP *waṯabū waṯabū 3FP *waṯabā waṯabna 2CD *waṯabtumā waṯabtumā 3MD *waṯabā waṯabā 3FD *waṯabatā waṯabatā 105 MHGA, v. 2020-1 a. In South Arabia and Ethiopia, the /k/ of the first person pronominal suffix was leveled to the second persons, producing Gəʿəz bähälku, bähälkä, bähälki and Sabaic qtlk3. The Arabic languages of Yemen have taken over this distribution, producing forms like kunk ‘I was’, kunki ‘you were’ (fs). b. The suffix conjugation often has an optative force, which is a continuation of the PS semantics of the old yaqtul preterite, e.g. Old Arabic (Hismaic) ḏakarat allāto ʔaśyāʕa-nā kelāla-hom ‘May Allāt be mindful of all our companions’. Thematic classes in Arabic Proto-Semitic Proto-Arabic (a ~ u) (a ~ u) (a ~ i) (a ~ i) (a ~ a) (a ~ a) (i ~ a) (i ~ a) (u ~ u) (u ~ u) kataba - yaktubu wasina - yasinu fataḥa – yiftaḥu (II, III gutturals) ʕalima – yiʕlamu kabura – yakburu 106 MHGA, v. 2020-1 3.3 Verb classes Geminate Proto-West Semitic ProtoArabic Hismaic Safaitic Classical Arabic Levanti ne Maghr ebine 1cs *radadku *ḥalaltu NA NA radadtu raddayt raddt 3cs *radada *ḥalala ḫṭ /ḫaṭṭa/ ḥl /ḥalla/ ḥṭṭ /ḫaṭaṭa/ ḥll /ḥalala/ radda radd radd Already at the Proto-Central Semitic stage, geminate sequences of CxvCxv shifted to CxCxv (Huehnergard 1995). This change seems to have been optional, as uncontracted forms obtain in Sabaic, Ugaritic, and indeed in Old Arabic. Safaitic and Hismaic exhibit both contracted and uncontracted suffix conjugation forms, perhaps suggesting that the former are from a chronologically shallower stage of the language. Classical Arabic and the QCT only know the collapsed form. In the QCT, the verb ẓalla is spelled ẓlt in the 1st person, suggesting a pronunciation /ḏa ̣ lt/. Nearly all modern dialects have merged the geminate class with the III-w/y class, producing a hybrid form in the suffix conjugation *radday- in the 1st and 2nd persons and the collapsed geminate forms in the 3rd person, radd and raddat. This is identical in form with the suffix conjugation of the D-stem of III-w/y verbs. The confusion probably originated in the 3rd feminine singular, where both classes are identical, e.g. raddat ‘she responded’ (G-stem, geminate) and ṣallat ‘she prayed’ < *ṣallayat. The Maghrebi form raddt does not continue the ancient uncollapsed form, *radadtu, which would surface as **rdədt, but is rather an innovation that results from the addition of the pronominal suffixes to the 3rd masculine singular form radd. In some Sudanese dialects and in Rāziḥit, the geminate verbs have fully merged with III-w/y, resulting in 3rd person forms terminating in a vowel, Sudanese radda < *raddā and Raẓiḥit raddē. II-w/y 107 MHGA, v. 2020-1 Proto-West ProtoSemitic Arabic Hismaic Safaitic Classical Arabic Levantine 1cs *qawumku *qawúmtu NA NA qumtu ʔimit 3cs *qawuma *qáwuma mt /māta/ mt /māta/ myt /mayeta/ qāma ʔām Medial-weak verbs can be reconstructed as triradical for Proto-Arabic, a fact supported by the Geʿez forms, 3ms kona < *kawna < *kawəna. Tri-radical forms are preserved in Safaitic, beside by-radical ones suggesting that the collapse of the triphthong in these circumstances had already begun to spread. The allomorphy of the paradigm in Classical Arabic, the QCT, and the modern dialects can only be understood from a triradical starting point. The collapse of the triphthongs to different qualities based on the placement of stress produced the following patterns: *qáwuma > qāma but *qawúmtu > *qūmtu > qumtu and *nawíma > *nīmtu > nimtu.42 Based on these patterns, and the Geʿez distribution, II-w/y verbs must have only had a high theme vowel in the suffix conjugation, either /i/ or /u/. III-w/y ProtoCentral Semitic ProtoArabic Hismaic 1cs *banayku *ʔatawku *banaytu NA *ʔatawtu 3cs *banaya *ʔatawa *banaya *ʔatawa Safaitic QCT Classical Arabic Levantine NA bnyt /banayt/ dʿwt /daʕawt/ banaytu daʕawtu banayt daʕayt bny /banē/ dʿʾ /daʕā/ banā daʕā banā daʕā bny bny/s²ty /banaya/ rare: s²tw; dʿ /daʕā/ ʾtw; A1 αθαοα Final weak roots were triradical in the suffix conjugation as well and both triphthongs were preserved at the Proto-Arabic stage, as evidenced by the Safaitic and Hismaic inscriptions. However, already in Safaitic, there was a tendency to merger III-w with III- 42 On this sound change, see Bauer 1912. 108 MHGA, v. 2020-1 y, perhaps triggered by the sound change *iwV > iyV. This would result in all active participles of III-w/y roots having a /y/ as a third consonant as well as verbs with an itheme vowel: *raṣ́iwa > raṣ́iya ‘to be pleased’, Classical Arabic raḍiya *ʔātiwatun > *ʔātiyatun ‘coming’ fs., Classical Arabic ʔātiyatun The introduction of a y into the paradigm of III-w verbs catalyzed the merger between the two classes. In Hismaic, the triphthong /awa/ collapsed to ā while the aya triphthong remained intact, resulting in a situation comparable to the QCT, where III-y and III-w are distinguished orthographically. Hismaic QCT Proto Arabic bny = banaya ‫ بنى‬/banē/ *banaya dʿ = daʕā ‫ دعا‬/daʕā/ *daʕawa In Classical Arabic, the triphthongs of both verbs collapse, merging them in the 3ms and 3fs, while they remain distinct in the 1st and 2nd persons. In all modern vernaculars, III-w and III-y complete merge to III-y, completing a change witnessed already in Safaitic. In most cases this vowel is ā, but in Raziḥit the vowel is ē. 3.4 Derived Stems This section will provide a reconstruction of the Arabic verb stems with some remarks on their semantic dimension. Stem Arabic Form Stem Arabic Form G I Gt VIII D II tD V C IV Ct X L III tL VI 109 MHGA, v. 2020-1 cD N/A N VII D-stem Proto-Arabic Safaitic Classical Arabic Levantine *qattala ʿwr /ʕawwara/ qattala rawwaḥ *yuqattilu yʿwr /yoʕawwer/ yuqattilu yərawweḥ Causative or factitive of the G-stem, and can sometimes express pluractionality. There is considerable overlap between the D and the C. The u-vowel of the prefix is reconstructable based on the comparative Semitic evidence, vocalizations of the Old Arabic participle, e.g. Μογαιερος /moġayyer/, and Classical Arabic. tD-stem Proto-Arabic Safaitic/Hismai c Classical Arabic QCT Najdi Cairo *taqattala ts²wq /taśawwaqa/ tafaʕʕala tnzl /tanazzal/ tifaʕʕal itfaʕʕal *yatqattalu trḥm /taraḥḥam/ < *tataraḥḥam yatafaʕʕalu ydkr /yaḏḏakkar/ < *yatḏakkar ytafaʕʕal yitfaʕʕal /yitfaʕʕal This forms the medio-passives of the D. The form yatafaʕʕalu seems to be post-ProtoArabic innovation. As Diem (1982) argues, the other Semitic languages point towards an original yatfaʕʕalu vocalization. Classical Arabic leveled the suffix conjugation stem to the prefix; other Arabic languages, such as Cairene, have clipped the prefix stem, producing a new suffix conjugation form with an it prefix (Van Putten, pc.). The sequence tatafaʕʕalu loses its first ta in some forms of Arabic (as early as Hismaic and the QCT). 110 MHGA, v. 2020-1 L Proto-Arabic Classical Arabic Levantine *qātala qātala sāfar *yuqātilu yuqātilu ysāfer This form has become a reciprocative in Classical Arabic, but it is difficult to determine whether or not this was its original function. In other Semitic languages, it is purely lexical. tL Proto-Arabic Classical Arabic Najdi *taqātala taqātala tuwāǧah *yatqātalu yataqātalu yitwāǧah Medio-passive of the L. The same developments of the tD apply to the tL. C Proto-Arabic Safaitic Classical Arabic Najdi *ʔaqtala ʔs2rq /ʔaśraqa/ ʔaqtala ašmal *yu(ʔa)qtilu ys2rq /yośreq/ yuqtilu yišmil This stem, which goes back to Proto-Semitic *sapris and *yusapris, is affected by the sound change s > h > ʔ in Arabic. The penultimate vowel of the suffix conjugation was leveled to /a/ in all attested forms of Arabic. Lexicalized h-stems exist in all forms of Arabic and appear to be frozen from a pre-Proto-Arabic period (e.g. hāt ‘give!’) or reflect borrowings from other languages, e.g. muhayminun. Š-causatives are also attested in the modern vernaculars and the ancient dictionaries. These are most certainly ancient loans, for example, šašqala ‘to exchange money’, compare with Hebrew šeqel, the cognate of which in Arabic is ṯaqlun. The verb šaqlab, yišaqlib, šaqlūb ‘to turn upside down, is common in the modern vernacular. Ct Proto-Arabic Classical Arabic QCT *(ʔ)(v)staphaʕ ala istfaʕala ʾstfʿl /ʔastafʕala/ (?) 111 MHGA, v. 2020-1 *yastaphʕilu yastafʕilu ystfʿ lyastafʕil/ The medio-passive of the C, where the original *s¹ is preserved by virtue of its nonword boundary position. The QCT and some modern Arabic dialects have a true ʔasyllable before the s-morpheme while Classical Arabic is a prothetic vowel that can e elided in certain contexts. It is unclear which form should be reconstructed for ProtoArabic. Gt Proto-Arabic Safaitic Classical Arabic *tanẓara *intaẓara tnẓr /tanẓara/ iftaʕala tẓr /taẓẓara/ or ettaẓara/ s2tky /eśtakaya/ itfaʕal *yantaẓiru ytẓr /yattaẓer/ yitfaʕal yaftaʕilu Cairene (passive) This is the medio-passive of the G-stem, but in most cases the stems containing this afformative have become lexicalized (with the exception of Egyptian Arabic). The original vocalization of the suffix conjugation is unclear. Egyptian Arabic exhibits a prefixed t and such a form is possibly attested in Safaitic. Other forms of Arabic exhibit an infix. The interpretation of this distribution follows that of the tD stem – namely, that Proto-Arabic had a prefix in the suffix conjugation and an infix in the prefix conjugation and that these were levelled in different ways in the subsequent languages. Such a distribution is attested in Sabaic. N Proto-Arabic Safaitic Classical Arabic *naqtala nġḍb /naġṣ́aba/ inqatala ingiṭaʕ *yanqatilu yqʾ /yaqqaʔ/ < *yanqaʔ, jussive from root qyʔ yingaṭiʕ yanqatilu Najdi 112 MHGA, v. 2020-1 This is the passive of the G. Safaitic suggests that the n-morpheme of the suffix conjugation in Proto-Arabic was originally nV-, compare to Akkadian naprus and Hebrew nipʕal. Other forms of Arabic produced a new suffix conjugation clipped from the prefix conjugation, with a prothetic syllable. Classical Arabic does not form N-stems of I-w/y verbs, but these are formed normally in Old Arabic and the modern vernaculars, thus ngʿ /nawgaʕa/ and Levantine inwažaʕ. L2-Stems Related to the L-stem are verb forms with diphthongs in between C1 and C2 of the root, e.g. Levantine sawlaf, yisawlif ‘to converse’ or Najdi dēwar <*daywara ‘to go in circles’. Such forms remain productive, for example, Lebanese yikawriz ‘to go on a cruise’. Reduplication and n-insertion Reduplication is used to from the so-called form IX, which indicates colors and defects. The basic stem of the suffix conjugation is ifʕalla < *ifʕalala in Classical Arabic, perhaps clipped from an original *yiphʕalilu. Reduplication in the L2-stem also produces verbs of color and defect, e.g. iḫḍawḍara ‘to be green’, Safaitic ḥwwt /eḥwawat/ ‘to become dark’. Rare infixed an forms are also known in Classical Arabic ifʕanlā, and may be related to the Akkadian tan iterative. Such forms have not yet been attested in Old Arabic or the modern vernaculars. Imperatives The imperatives are clipped from the stem of the jussive prefix conjugation and are renewed frequently. For example, the Proto-Arabic imperative of III-w/y verbs terminates in a short vowel: *(i)bni (build!, 2ms). This form should yield ibin in Levantine, following the loss of final short vowels. Yet the imperative is ibni, formed from the synchronic jussive, which is tibni ‘may you build’. The ancient imperative survives in some Peninsular dialects, e.g. Najdi ibn ‘build’!. 113 MHGA, v. 2020-1 Internal Passive The internal passive must be reconstructed for Proto-West Semitic, but its vowel melody, namely u-i, seems to be unique to Arabic. The vowels are only known from Classical Arabic. The modern vernaculars exhibit internal passives that can be derived from this melody, e.g. najdi sriǧ < *suriqa ‘he was robbed’; Levantine ḫliqt < *ḫuliqtu ‘I was born’. Internal passives are attested in Safaitic but their vowels are unclear: ṣlb ḥbb-h ‘his beloved was crucified’ /ṣoleba ḥabīb-oh/. 114 MHGA, v. 2020-1 IV Notes on Syntax 4.1 Infinitive While later forms of Arabic employ a subordinated finite verb where other Semitic languages use a nominal form (the infinitive), Old Arabic seems to have had both options. The infinitive had many functions: The infinitive in a sequence of verbs rʿy h-rmḫ bql w kmʾt pasture.SC.3MS DEF- camel.COL herbage CONJ gather truffles.INF ‘he pastured the camels on spring herbage and gathered truffles’ The infinitive with a nominal subject ngm bn ẓʿn bn rgl bn ṣʿd w s²tt-h nwy LA Gn05 CONJ winter.INF-3MS pastureland ‘by Ngm son of Ẓʿn son of Rgl son of Ṣʿd and he spent the winter on pastureland’ l ʾ{l}{t} s¹fʾ-hm {ʾlt} feed.INF-3MP ‘O Lt, may they provide sustenance’ h VOC A similar construction is attested in the QCT. For example: Q 91:31 fa-qāla la-hum rasūlu llāhi nāqata llāhi wa suqyā-hā ‘and the messenger of Allah said: [do not harm] the camel of Allah or [prevent her from] drink’ Perhaps better: ‘here is a she-camel of Allah so let her drink (infinitive with pronominal subject)’ The infinitive as a command bʿls¹mn trwḥ b- mṭr Bʿls¹mn send the winds.INF with-rain ‘O Bʿls¹mn, send the winds with rain’ h VOC The infinitive to express purpose rgʿ return.SC.3MS ht low-lying land b- ʾbl with-camel.COL rʿy pasture.INF l-ḥrt towards- Ḥrt ʿf ʿawf.CNST 115 MHGA, v. 2020-1 ‘he returned towards the Ḥarrah with camels to pasture on ʿawf (plants) of low-lying land’ In Old Ḥigāzī, the infinitive complement of a finite verb was largely replaced by a subordinating construction introduced by the ʔan element and a subjunctive prefix conjugation. Dadanitic: ʾn ykn l-h wld ‘that he may have offspring’ QCT/Classical Arabic: ʔan yafʕala Most modern dialects have replaced the infinitive with a serial verb construction with a modal verb in second position: Qəltu qa-yərīd yəftaḥ ‘he wants to open’; Levantine b-yərūḥ yədros ‘he is going to study’. 4.2 Negation Negative Adverbs *ʔin: A negator common in the QCT, usually used in constructions followed by ʔillā. It is perhaps related to the Geʿez negator ʔi. *lam: Negates the past with the volitive/jussive (old preterite prefix conjugation). It is likely a contraction or clipped form of the negative adverb lamma ‘not yet’ <*lā + ma with junctural doubling. The construction lamma yafʕal ‘he has not yet done’ is attested in Classical Arabic. The lam yafʕal construction is an important Arabic innovation, attested widely in early forms of the language, in Safaitic, the QCT, the substrate of the Haram Sabaic inscriptions, and in early Middle Arabic texts. The construction was eventually marginalized by the spread of mā + suffix conjugation (see below). *lā: The negator was originally restricted to indicative forms, but it has spread in Arabic to the negation of the volitive, lā tafʕal ‘do not do!’, replacing the older negator ʔal-. The opposite is found in Sabaic, where ʔal is used in indicative contexts *lā-ʔan: The negation of the explicit future. The form lan is only attested in the QCT and in Classical Arabic, reflecting a contraction. The form lʾn is attested once in Safaitic, lʾn yqtl /lā-ʔan yoqtala/ ‘may he never be killed’. *mā: This negative adverb, grammaticalized from the interrogative mā ‘what’, originates in rhetorical negative constructions such as mā bi-yadī šayʔun ‘what thing 116 MHGA, v. 2020-1 is in my hand’ > ‘nothing is in my hand’. This negator applied to the suffix conjugation creates the negative preterite, a construction that competes with the older lam yafʕal syntagm. In later forms of Arabic, the mā + suffix conjugation completely replaces lam yafʕal. The mā negator can be applied to verbless sentences. Three syntagms are known, all of which are tolerated in Classical Arabic. *mā huwa ragulun: Classical Arabic; modern dialects *mā huwa ragulan: Old Ḥigāzī (QCT), the so-called mā al-Ḥigāziyyah. The accusative predicate likely stems from an analogy with sentences containing kāna. *mā huwa bi-ragulin: QCT, Najdi dialects (= *mā hū b-raǧil): the use of the locative pronoun to mark the predicate finds parallels cross-linguistically and may have been motivated to distinguish this construction from interrogative sentences. *laysa: The verb laysa negates equational and existential sentences. Its etymology is unclear but most likely has a non-Arabic origin. For hypotheses, see Al-Jallad, Lasya. The word was likely borrowed into Proto-Arabic and is already attested in Safaitic, ls /laysa/. mū, etc.: Some modern vernaculars, such as Iraqi, have grammaticalized a new negator based on the fusion of mā and the nominative pronouns, mā hū > mū, mā hī : mī, etc. manno, etc.: A similar construction, based on oblique pronouns introduced by the particle ʔanna, is common in the Levant, a construction perhaps related to the mā alḤigāziyyah: manno < *mā ʔannoh; mannak < *mā ʔannak, etc. miš/muš: The sentential negator derives from the construction mā + pronoun + the word ‘thing’ šī: *mā-ẖū-šī > *mā-hū-š > *mūš > *muš; *mā-hī-šī > *mā-hī-š > *mīš > *miš. Variation in the middle vowel suggests that the form conjugated for gender in its earliest stages before being generalized. 117 MHGA, v. 2020-1 4.3 Interrogative and conditional particles *mā: This derives from Proto-Central Semitic *mah (cf. Ugaritic mh), the original form of which is preserved in the adverb mahmā ‘whatever’. Once this adverb is grammaticalized as a negator a new interrogative emerges (below). *ʔayy-śayʔin-hū: This phrase grammaticalizes into a new adverb once mā is lost. The full phrase is frequently attested in Hadīth, suggesting it was a part of the spoken language once these materials were collected. Various shorten forms emerge in the modern vernaculars, Levantine ʔayš and šū; Gulf Arabic šinu, šinhu ‘with agreement of final pronominal element’. *man/mī: Proto-Arabic may have had both *man ‘who’ and *mī (cf. Hebrew mî). Safaitic attests mn and m, which may be interpreted as reflexes of these forms or perhaps the assimilation of the n in the latter to the following consonant. In the modern vernaculars, the form mīn is common, which may reflect a hybrid of the two forms. 118 MHGA, v. 2020-1 V Appendix of early Arabic texts 5.1 Safaitic Chrestomathy This presents a small selection of Safaitic inscriptions fully glossed with reconstructed vocalizations and translations. These texts can be read with the grammatical sketch published in the Routledge volume, The Semitic Languages, 2nd edition (below). The interpretation of the vocabulary follows The Dictionary of the Safaitic Inscriptions (AlJallad and Jaworska 2019). Grammatical sketch: Al-Jallad, A. 2019. “Safaitic”. In: J. Huehnergard and N. Pat-El, The Semitic Languages, 2nd edition, pp. 342-366. New York: Routledge. https://www.academia.edu/35134178/AlJallad._2019._Safaitic_The_Semitic_Languages_2nd_edition_ Abbreviations: prep = preposition; cnst = construct state; sc = suffix conjugation; pc = prefix conjugation; imp = imperative; cpro = clitic pronoun; art = definite/demonstrative article; conj = conjunction; comp = complementizer; m = masculine; f = feminine; c = common; s = singular; d = dual; p = plural; 1 = first person; 2 = second person; 3 = third person; voc = vocative; asv = asseverative particle; rel = relative particle; n = nunation; quant = quantifier. Editorial symbols: {} = damaged glyph; [] = restored glyph; <> = erased/extra glyph; () = uncertain vocalization/gapped term; ---- = damaged section. Note on vocalization: The vocalization of the vocabulary is derived from Al-Jallad and Jaworska (2019), based ultimately on Al-Jallad 2015 and 2017 (https://www.academia.edu/7583140/AlJallad._2017._Graeco-Arabica_I_the_southern_Levant). Personal names are vocalized according to their closest equivalents in Greek transcription. For some names, several vocalizations were possible - I simply settled for the most common. I have not vocalized group names that have not appeared in Greek transcription. Sigla: See Al-Jallad 2019 for a list of references and identifications of the sigla used here. 119 MHGA, v. 2020-1 HaNSB 307 lsʿdlh prep- Saʿdallāh Ḥyn Ḥayyān bn son.cnst ḏ rel.ms ʾhl -h family-cpro.3ms ʾs ʾAws ʾl lineage f-h-lt conj-voc-Allāt bn son.cnst frṯ Frṯ ẓnʾl Ṯạ nnʾel w- tśwq conj-to long.sc.3ms qbll to reunite.inf w-bny conj-to build.sc.3ms ʾ-nfs art-funerary monument w-dʿy conj-to invoke.sc.3ms ʾl-[l]t ʿl-mn prep- Allāt prep-rel bn son.cnst ʾlprep- w-ġnmt conj-spoil yḫbl-h efface.pc.3ms-cpro.3ms reconstructed vocalization le-Saʿdallāh ben ʾAws ben Ṯa ̣ nnʾel ben Ḥayyān ḏī ʾāl Frṯ wa-taśawwaqa ʾel-ʾahloh pʰa-hā-llāt qeblāl wa-ġanīmat wa-banaya ʾan-napʰsa wa-daʿaya ʾel-allāt ʿalman yoḫabbel-oh Translation By Saʿdallāh son of Aws son of Ṯạ nnʾel son of Ḥayyān of the lineage of Frṯ and he longed for his family so, O Allāt, may there be a reunion and spoil and he built the funerary monument and invoked Allāt against whosoever effaces it. C 2446 l-sʿd prep-Saʿd bn son.cnst mrʾ bn Marʾ son.cnst ʿ[l-]ʾḫ-h nr prep-brother-cpro.3ms Nūr w-gd[ʿ]{w}ḏ w-wgm conj-to grieve.sc.3ms qtl[-h ] ʾl-{n}bṭy [ ] to kill.sc.3ms-cpro.3ms art-Nabataeans {r}ʿy nʿm ʿwḏ to pasture.prtcp.ms livestock.cnst ʿAwīḏ w-ʾlt-dṯn nr Nūr w-ḍf conj-Ṣ́ aypʰ f-h-lt mʿmn conj-voc-Allāt-Mnʿm w-gdḍf 120 MHGA, v. 2020-1 conj-ʾelat-Dṯn conj-Gadd-ʿAwīḏ conj-Gadd-Ṣ́ aypʰ ṯʾr m-ḏ ʾslf w-wlh to avenge.sc.3ms prep-rel.ms to commit.sc.3ms conj-to be distraught.sc.3ms k{b}{r} greatly sḥr broken heart ʿl-ʾḫ-h prep-brother-cpro.3ms ḥbb-h beloved-cpro.3ms l-ʾbd prep-eternity Reconstructed vocalization le-Saʿd ben Marʾ ben Nūr wa-wagama ʿal-ʾaḫī-h Nūr qatala-h ʾal-nabaṭeyy rāʿeya naʿāma ʿAwīḏ wa-Ṣ́ aypʰ pʰa-hā-llāt Mʿmn wa-ʾelata-Dṯn wa-Gadda-ʿAwīḏ wa-Gadda-Ṣ́ aypʰ ṯaʾera meḏ-ḏī ʾaslapʰa wa-waleha kabīra sāḥera ʿal-ʾaḫī-h ḥabīb-oh le-ʾabad Translation By Saʿd son of Marʾ son of Nūr and he grieved {for} his brother Nūr, {whom} the Nabataeans killed while pasturing the livestock of (the tribes) ʿAwīḏ and Ṣ́ ayf; so, O Allāt-Mʿmn and goddess of Dṯn and Gadd-ʿAwīḏ and Gadd-Ṣ́ ayf, may he have vengeance against him who committed (this act); and he was greatly distraught with a broken heart for his brother, his beloved forever. Alays 1 = AMSI 41 l-msk bn prep-Māsek son.cnst again ʾsd bn slm ʾAsad son.cnst Sālem w-qʿd conj-to halt.sc.3ms ʿd wrd f-ḏkr h-mt f-qṣf to go to water.sc.3ms conj-to remember.sc.3ms art-dead conj-to grieve.sc.3ms f-h-lt ʿmr ṣdq-k conj-voc-Allāt to grant life.imp.fs righteous one-cpro.2fs w-gnn conj-to protect.imp.fs w-m-mt conj-prep-death ls neg fṣy deliverance Reconstructed vocalization le-Māsek ben ʾAsad ben Sālem wa-qaʿada ʿawda wāreda pʰa-ḏakara ham-mawtē pʰa-qaṣapʰa pʰa-hā-llāt ʿammerī ṣadīq-ek wa-gannenī wa-mem-mawt laysa pʰaṣāy 121 MHGA, v. 2020-1 Translation By Māsek son of ʾAsad son of Sālem and he halted again while going to water and remembered the dead so, O Allāt, grant long life to your righteous worshipper and protect (him) but from death, there is no deliverance. MSNS 2 l-grmʾl prep-Garmʾel bn son.cnst sʿd bn Saʿd son.cnst qḥś Qaḥaś ʾḫ-h f-lm brother-cpro.3ms conj-neg w-bġy conj-to seek.sc.3ms yʿd to return.pc.3ms f-qṣf conj-to grieve.sc.3ms Reconstructed vocalization le-Garmʾel ben Saʿd ben Qaḥaś wa-baġaya ʾaḫā-h pʰa-lam yaʿod pʰa-qaṣapʰa Translation By Garmʾel son of Saʿd son of Qaḥaś and he sought his brother but he did not return so he grieved. AWS 237 l-ḫzmʾ bn prep-Ḫazmāʾ son.cnst ʾm m cond rel m rel kn h-gml Kawn art-camel ʿwr to efface.sc.3ms ʿwr to efface.sc.3ms w-qṣy-h conj-to dedicate.sc.3ms-cpro.3ms h-rḍw f-l-yʿwr voc-Roṣ́aw conj-asv.be blinded.pc.passive.3ms w-l-yqʾ conj-asv.to be thrown out.pc.3ms b-ṣdq prep-friend Reconstructed vocalization le-Ḫazmāʾ ben Kawn hag-gamal wa-qaṣaya-h ʾem(mā) maʿ-ʿawwara hā-roṣ́aw pʰa-le-yoʿawwar maʿ-ʿawwara wa-le-yeqqaʾ be-ṣadīq Translation 122 MHGA, v. 2020-1 By Ḫazmāʾ son of Kawn is this camel and he dedicated it (to a deity); if one should efface (it), O Roṣ́aw, may whosoever effaces be made blind and let him (then) be thrown out (of the grave) by a friend. HaNSB 304 l-ḏl bn prep-Ḏayl son.cnst qmr Qamar śrk bn Śārek son.cnst w-mṭy conj-to journey.sc.3ms ḏ ʾl rel.ms lineage f-h-śʿhqm ġnmt conj-voc-Śayʿ-haqqawm spoil w-rmy b-rmḥ-h conj-to cast.sc.3ms prep-spear-cpro.3ms b-sf-h prep-sword-cpro.3ms rbḥ Rebḥ w-ḫzr conj-to strike.sc.3ms f-mrq kll slsl-h conj-to throw off.sc.3ms quant chains-cpro.3ms f-gdʿwḏ ġnmt w-slm w-ḫlf l-slḥ-h conj-Gadd-ʿAwīḏ spoil conj-security conj-recompense prep-weaponscpro.3ms m-ʾl prep-lineage nbṭ Nabaṭo w-ʿwr ḏ conj-blind.imp.ms rel ḫbl to efface.sc.3ms Reconstructed vocalization le-ḏayl ben Śārek ben Rebḥ ḏī ʾāl Qamar wa-maṭaya pʰa-hā-Śayʿa-haq-qawm ġanīmat wa-ramaya be-romḥ-oh wa-ḫazara be-saypʰ-oh pʰa-maraqa kelāla selsāl-oh pʰa-Gadda-ʿAwīḏ ġanīmata wa-salāma wa-ḫalpʰa le-selāḥ-oh meʾ-ʾāl nabaṭo wa-ʿawwer ḏa ḫabbala Translation By Ḏayl son of Śārek son of Rebḥ of the lineage of Qamar and he set off on a journey so, O Śayʿhaqqawm, let there be spoil! And he cast his spear and struck with his sword and threw off all his chains (of bondage) so O Gadd-ʿAwīḏ, [grant] spoil, security, and recompense for his weapons from the people of Nabataea and blind him who would efface (this writing). KRS 1023 l-ʿlm bn prep-ʿālem son.cnst ṣʿb bn Ṣaʿb son.cnst grmʾl Garmʾel bn son.cnst ḏʾb Ḏeʾb 123 MHGA, v. 2020-1 w-mrd conj-to rebel.sc.3ms ʿl- h-mlk grfṣ prep-art-king Agrippa ksr {h-}sl{s}[lt] to break.inf art-chains Reconstructed vocalization le-ʿālem ben Ṣaʿb ben Garmʾel ben Ḏeʾb wa-marada ʿal-ham-malk Agrippoṣ kasra has-selselāt Translation By ʿālem son of Ṣaʿb son of Garmʾel son of Ḏeʾb and he rebelled against king Agrippa to break the chains (of bondage). KRS 78 l-ʾdm bn prep-ʾĀdam son.cnst Ṣaʿd ysmʿ{l} Yesmaʿīl w-wgm conj-to grieve.sc.3ms ʿl-ʾḫ-h prep-brother-cpro.3ms w-ʿl- ḥd conj-prep-Ḥadd bn son.cnst ysmʿl Yesmaʿīl bn son.cnst ṣʿd ʿl-ʾs prep-ʾAws ʾḫ-h f-h-lt brother-cpro.3ms conj-voc-Allāt ryḥ grant relief.imp.2fs w-qyt conj-to protect.inf Reconstructed vocalization le-ʾĀdam ben Yesmaʿīl ben Yesmaʿīl ben Ṣaʿd wa-wagama ʿal-ʾaḫī-h ʿal-ʾAws wa-ʿal-Ḥadd ʾaḫī-h fa-hā-llāt rayyeḥī wa-qeyat Translation By Ādam son of Yesmaʿīl son of Yesmaʿīl son of Ṣaʿd and he grieved for his brother, for Aws, and for Ḥadd his brother so, O Allāt, grant relief and protect! ASWS 73 l- rbʾl bn ḥnn bn ẓʿn prep-Rabbʾel son.cnst Ḥanūn son.cnst Ṯạ̄ ʿen bn ḫyḏ bn son.cnst Ḫāyeḏ son.cnst 124 MHGA, v. 2020-1 ʿḏr w-wrd ʿOḏayr conj-to go to water.sc.3ms f-mlḥ conj-Aquarius ḥḏr cautious of drought.prtcp.ms f-ḏkr f-ʾmt f-ʾmt conj-Aries conj-Libra conj-Libra w-ngʿ conj-to grieve in pain.sc.3ms ʿl-ḥbb prep-beloved rʿy-h hgr to pasture.sc.3ms-cpro.3fs to migrate.prtcp.3ms snt year.cnst myt to die.sc.3ms w-ʿl-h-ʾbl conj-prep-art-camels m-mdbr prep-inner desert bnt Banat Reconstructed vocalization le-Rabbʾel ben Ḥanūn ben Ṯạ̄ ʿen ben Ḫāyeḏ ben ʿOḏayr wa-warada ḥāḏera pʰamelḥ pʰa-dakar pʰa-ʾāmet pʰa-ʾāmet wa-nawgaʿa ʿal-ḥabīb wa-ʿal-haʾ-ʾebel raʿaya-hā hāgera mem-madbar sanata mayeta Banat Translation By Rabbʾel son of Ḥanūn son of Ṯạ̄ ʿen son of Ḫāyeḏ ben ʿOḏayr and he went to water cautious of drought, then in Aquarius, then in Aries, then in Libra, (and) then in Libra (again) while he grieved for a loved one and for the camels which he had pastured migrating from the inner desert the year Banat died. KRS 3074 l-sqn bn prep-Sīqān son.cnst wtr ḏ ʾl Watr rel.ms lineage ʿmrt w-ʾlt ʿAmarat conj-goddess.cnst ʾ-{ġ}b <s>slm art-unseen to be secure.sc.3ms w-ʾgʿ-nh conj-to harm.sc.3ms-cpro.3ms ʿn ʿAyn {ʾ}-śnʾ art-enemy bn son.cnst ʾns ʾAnas Reconstructed vocalization le-Sīqān ben Wetr ḏī ʾāl ʿAmarat wa-ʾelata-ʾaġ-ġayb salema wa-ʾawgaʿa-nnoh ʿAyn ben ʾAnas ʾaś-śāneʾ Translation 125 MHGA, v. 2020-1 By Sīqān son of Watr of the lineage of ʿAmarat and, O goddess of the unseen, may he be secure as ʿAyn son of ʾAnas, the enemy, has caused him harm. HH 1 l-ḥyn prep-Ḥayyām bn son.cnst ʿq{d}t ʿOqaydat ḏ rel.ms ʾl lineage kmy Komayy w-n{ṣ}b w-ḏbḥ w-ḥll conj-to erect a cult stone.sc.3ms conj-to sacrifice.sc.3ms conj-to camp.sc.3ms w-ḫrṣ conj-to keep watch.sc.3ms {ʾ}śyʿ-h companions-cpro.3ms ḍbʾn to raid.prtcp.3mp f-h-lt w-dśr conj-voc-Allāt conj-Diśar [s¹][l]m to make secure.imp.d w-qb{l}{l} conj-reunion {f}-{h}-{l}t conj-voc-Allāt {r}w[ḥ] to grant relief.imp.fs w-{ġ}nmt conj-spoil Reconstructed vocalization le-Ḥayyān ben ʿOqaydat ḏī ʾāl Komayy wa-naṣṣaba wa-ḏabaḥa wa-ḥallala waḫaraṣa ʾaśyāʿ-oh ṣ́ābeʾīna pʰa-hā-llāt wa-diśar sallemā wa-qeblāl pʰa-hā-llāt rawweḥī wa-ġanīmat Translation By Ḥayyān son of ʿOqaydat of the lineage of Komayy and he erected a cult stone and made a sacrifice and camped; and he kept watch for his companions who were on a raid so, O Allāt and Diśar, grant security and a reunion; O Allāt grant relief and spoil. Ms 44 l-zd bn Prep-Zayd son.cnst rgl Rāgel h-ʿrḍ art-valley myt bn to die.sc.3ms son.cnst snt year.cnst w-smʿ conj-to hear.sc.3ms ʾn comp f-sḫr w-h-bkrt w-rʿy conj-to pasture.sc.3ms h-ʾbl art-camels qṣr Caesar myt to die.sc.3ms flfṣ Philipp w-h-gdḍf 126 MHGA, v. 2020-1 conj-to be fooled.sc.3ms conj-art-she camel conj-voc-Gadd-Ś ̣ayf lʿn to curse.imp.ms pleasure.pc.3ms ḏ rel.ms yʿwr to efface.pc.3ms w-ġnmt conj-spoil l-ḏ dʿy prep-rel.ms to read.sc.3ms m rel yhnʾ to bring h-tll art-writing Reconstructed vocalization le-Zayd ben Rāgel wa-raʿaya haʾ-ʾebela haʿ-ʿerś ̣a sanata mayeta ben qayṣar wasameʿa ʾan mayeta pʰeleppʰoṣ pʰa-soḫera wa-hab-bakrat wa-hā-Gadda-Ṣ́ aypʰ (e)l(e)ʿan ḏā yoʿawwer mā yohanneʾ wa-ġanīmat le-dī daʿaya hat-telāla Translation By Zayd son of Rāgel and he pastured the camels in the valley the year the son of Caesar died but he had heard that Philipp had died and was fooled and this (drawing of a) she-camel (is by him) so, O Gadd-Ṣ́ aypʰ curse him who would efface what brings pleasure and may he who would read this writing have spoil. LP 495 h- rḍw flṭ-n m-bʾs w-nḥyy voc-Roṣ́aw to deliver.imp.ms-cpro.1c prep-misfortune conj-to live.pc.1c Reconstructed vocalization hā-roṣ́aw pʰalleṭ-nā meb-boʾs wa-neḥyaya Translation O Roṣ́aw, deliver us from misfortunate that we may live (long). LP 180 l-msk bn prep-Māsek son.cnst ẓnnʾl Ṯọ naynʾel w-trwḥ conj-to set off.sc.3ms l-yśrq prep-to migrate.pc.3ms f-h-lt conj-voc-Allāt mʿdt return.inf bn nr bn yʿmr son.cnst Nūr son.cnst Yaʿmūr l-mdbr prep-desert w-slm m-śnʾ conj-to be secure.sc.3ms prep-enemies 127 MHGA, v. 2020-1 Reconstructed vocalization le-Māsek ben Ṯọ naynʾel ben Nūr ben Yaʿmūr wa-tarawwaḥa le-yośreqa lemadbar pʰa-hā-llāt maʿādat wa-salema meś-śonnāʾ Translation By Māsek son of Ṯọ naynʾel son of Nūr son of Yaʿmūr and he set off (at night) to migrate towards the inner desert so, O Allāt, may there be a (safe) return and may he be secure against enemies. KRS 941 l-nʿmn bn ṣʿd bn prep-Noʿmān son.cnst Ṣaʿd son.cnst trace.cnst ysmʿl w-wgd ʾṯr Yesmaʿīl conj-to find.sc.3ms ṣʿd f-ngʿ w-bʾs m ẓll Ṣaʿd conj-to grieve in pain.sc.3ms conj-to despair.sc.3mp rel to remain.sc.3mp w-rġm conj-to be stuck down.sc.3ms say.sc.3ms m{n}{y} fate {ʿ}{n}{y} {w}-{q}l to suffer.prtcp.ms conj-to ḫbl-h trḥ w-h-lt ʿwr to drive mad.sc.3ms-cpro.3ms grief conj-voc-Allāt blind.imp.fs yʿwr to efface.pc.3ms ḏ rel.ms h-s{f}r art-writing Reconstructed vocalization le-Māsek ben Ṣaʿd ben Yesmaʿīl wa-wagada ʾaṯra Ṣaʿd pʰa-nawgaʿa wa-baʾesū maṯ-ṯ̣ allalū wa-roġema manaya ʿāneya wa-qāla ḫabbala-h taraḥ wa-hā-llāt ̣ ʿawwerī ḏā yoʿawwer has-sepʰra Translation By Māsek son of Ṣaʿd son of Yesmaʿīl and he found the traces of Ṣaʿd and grieved in pain - for those who remain (alive) despair - because he (= Ṣaʿd) was struck down by Fate while suffering; and he (= Māsek) said: grief has driven him mad; and O Allāt blind him who would efface this writing. C 4443 l-śmt bn lʿṯmn bn śmt bn śrk bn ʾnʿm 128 MHGA, v. 2020-1 prep-Śāmet son.cnst Leʿoṯmān son.cnst Śāmet son.cnst Śārek son.cnst ʾAnʿam bn son.cnst lʿṯmn w-wgm ʿl-ʾm-h Leʿoṯmān conj-to grieve.sc.3ms prep-mother-cpro.3ms w-ʿl- dd -h w-ʿl-ʿm conj-prep-paternal uncle-cpro.3ms conj-prep-ʿAmm w-ʿl-ʾnʿm conj-prep-ʾAnʿam qtl-h {ʾ}l ṣbḥ f-wlh ʿl-bn to kill.sc.3m lineage.cnst Ṣabāḥ conj-to be distraught.3ms prep-son.cnst ḫl-h trḥ w-rʿy maternal uncle-cpro.3ms perished.prtcp.ms conj-to pasture.sc.3ms h-ḍʾn art-sheep w-rḥḍ b-ṯbr w-ḫl h-ś[n]ʾ conj-to wash.sc.3ms prep-Sagittarius conj-to keep watch.sc.3ms art-enemy f-h-lt slm conj-voc-Allāt to be secure.sc.3ms w-wgd ʾṯr conj-to find.sc.3ms trace.cnst ʾḫ-h f-ndm brother-cpro.3ms conj-to be devastated.sc.3ms Reconstructed vocalization le-Śāmet ben Leʿoṯmān ben Śāmet ben Śārek ben ʾAnʿam ben Leʿoṯmān wawagama ʿal-ʾomm-oh wa-ʿal-dād-oh wa-ʿal-ʿAmm wa-ʿal-ʾAnʿam qatal(a/ū)-h ʾāl Ṣabāḥ pʰa waleha ʿal-ben ḫāl-oh tarīḥ wa-raʿaya haś ̣-ś ̣aʾna wa-raḥaś ̣a beṯāber wa-ḫalla haś-śāneʾa pʰa-hā-llāt salema wa-wagada ʾaṯra ʾaḫī-h pʰanadamma Translation By Śāmet son of Leʿoṯmān son of Śāmet son of ʾAnʿam son of Leʿoṯmān and he grieved for his mother and his paternal uncle and for ʿAmm and for ʾAnʿam whom the people/lineage of Ṣabāḥ had killed so he was distraught for the son of his maternal uncle, who had perished, and he pastured the sheep and (ritually) cleansed during Saggitarius and kept watch for the enemy so, O Allāt, may he be secure; and he found the traces of his brother and was devastated (by grief). C 2947 l-{ś}krʾ bn prep-Śakrāʾ son.cnst rmyn Ramyān bn mġṯ ḏ ʾl ʿmrt-n son.cnst Moġīṯ rel.ms lineageʿAmarat-n 129 MHGA, v. 2020-1 w-ndm conj-to be devastated ʿl-ʾḫ-h mlṯ mqtl prep-brother-cpro.3ms Malṯ killed.prtcp f-h-lt w-dśr conj-voc-Allāt conj-Diśar b-hld prep-Hld nqmt mn-mn mṣr-h retribution prep-rel-to attack.sc.3ms-cpro.3ms Reconstructed vocalization le-Śakrāʾ ben Ramyān ben Moġīṯ ḏī ʾāl ʿamaraten wa-nadamma ʿal-ʾaḫī-h Malṯ maqtūl be-Hld pʰa-hā-llāt wa-Diśar neqmata men-man maṣara-h Translation By Śakrāʾ son of Ramyān son of Moġīṯ of the lineage of ʿAmarat and he was devastated (by grief) on account of his brother Malṯ who was killed at Hld so O Allāt and Diśar [grant] retribution against whosoever attacked him. Is.K 205 l-khl le-Kahl bn son.cnst śʿʾ Śayʿāʾ w-ḥll ---conj.to camp.sc.3ms ʾḥrb to f-rsy conj-to remain.sc.3ms plunder.sc.3ms m-ḥrn mn-ṯlg s[n]t prep-Ḥawrān prep-snow year.cnst ġlḍ h-nʿm Ġāleṣ́ art-livestock bḥr w-h-ʾns mlṣ early summer conj-art-people to escape.sc.m w-qṣf ʿl-ḍʾn-h conj-to grieve.sc.m prep-sheep-cpro.3fs mḥr[b][t] plundered.prtcp.f Reconstructed vocalization le-Kahl ben Śayʿāʾ wa-ḥalla ---- pʰa-rasaya meḥ-ḥawrān men-ṯalg sanata ʾaḥraba Ġāleṣ́ han-naʿāma bāḥūra wa-haʾ-ʾonās maleṣ(a/ū) wa-qaṣapʰ(a/ū) ʿalṣ́aʾn-ah maḥrūbat (maḥrūbāt) Translation By Kahl son of Śayʿāʾ and he camped ---- and remained in place, (away) from the Ḥawrān on account of snow the year Ǵāleṣ́ plundered the livestock during the early summer and the people escaped but grieved for their sheep, which were taken as plunder. 130 MHGA, v. 2020-1 AMSP 1 l-ġyrʾl prep-Ġayyārʾel bn son.cnst m-ʾhl-h prep-family-cpro.3ms f-h-ym conj-art-day ġṯ ḏ ʾl ḥẓy w-rḥl Ġawṯ rel lineage Ḥaṯāy ̣ conj-to depart.sc.3ms f-ḥll-h m-ḥrb conj-camping-cpro.3ms prep-war ʾḫr last.cnst hn here rʾs ḏkrt f-h-ym to be first.sc.3ms fame conj-art-day ʿny hn here mn ḫṣf to suffer.sc.3m rel to be tracked.sc.3ms ʾḫ[r] final.cnst meadow ḥll camping ʾḫr final.cnst ḥll camping f-h-ym hn conj-art-day here ḥll ḥdd w-ṯwy b-h-rḍt camping go to border.sc.3ms conj-to alight.sc.3ms prep-art- w-ḫrṣ conj-to watch.sc.3ms f-h mouth-cpro.3ms ḫl-h skrn maternal uncle-cpro.3ms Sakrān b-q{l} prep-to say.inf fz-h good fortune-cpro.3ms yr{b} to exalt.pc.3ms f-h-lt conj-art-Allāt slm to be secure.sc.3ms Reconstructed vocalization le-Ġayyārʾel ben Ġawṯ ḏī ʾāl Ḥaṯāy ̣ wa-raḥala meʾ-ʾahl-oh pʰa-ḥolūl-oh meḥ-ḥarb raʾesa ḏekrata ʿoneya man ḫoṣepʰa pʰa-hay-yawma honā ʾāḫer ḥolūl pʰa-hay-yawma honā ʾāḫer ḥolūl pʰa-hay-yawma honā ʾāḫer ḥolūl ḥaddada wa-ṯawaya be-har-rawṣ́at wa-ḫaraṣa ḫāl-oh sakrāna yarobb pʰū-h beqawl pʰawz-oh pʰa-hā-llāt salema Translation 131 MHGA, v. 2020-1 By Ġayyārʾel son of Ġawṯ of the lineage of Ḥaṯāy ̣ and he left his family so may his halting be (only) for war encampment foremost fame! encampment those who are tracked suffer encampment so let here this day be the final so let here this day be the final so let here this day be the final He went to the boundary of the land and alighted in the meadow and kept watch for his maternal uncle, Sakrān, his mouth exalting him saying ‘may he have good fortune’; so O Allāt may he be secure. KRS 818 l-wdmʾl prep-Wadamʾel Ġarb bn son.cnst h-ṣmd art-high place w-ḏbḥ gml ʿl-h conj-to sacrifice.sc.3ms camel prep-cpro.3ms f-slm conj-to secure.sc.3ms m rel ʿwr efface.sc.3ms grmʾl Garmʾel yṯʿ Yayṯaʿ bn son.cnst nḫr bn Naḫr son.cnst ġrb m-śnʾ w-ʿwr prep-enemy conj-blind.imp.ms h-ʾsfr art-writings Reconstructed vocalization le-Wadamʾel ben Garmʾel ben Naḫr ben Ġarb haṣ-ṣamda wa-ḏabaḥa gamala ʿalay-h pʰa-sallama Yayṯaʿ meś-śāneʾ wa-ʿawwer maʿ-ʿawwara haʾ-ʾaspʰāra Translation By Wadamʾel son of Garmʾel son of Naḫr son of Ġarb, at this high place, and he sacrificed a camel upon it so may Yayṯaʿ provide security against the enemy and blind whosoever effaces these writings. 5.2 Old Arabic poetry 132 MHGA, v. 2020-1 (1) ʿĒn ʿAvdat (pre 150 CE), Nabataean Arabic (Kropp 2017; Fiema et al. 2015; Ahmad Al-Jallad forthcoming) p-ypfʿl lʾ pdʾ w lʾ ʾtrʾ pha-yapʕal lā pedā wa-lā ʔaṯarā p-kn hnʾ ybʿnʾ ʾlmwtw lʾ ʾbʿh pha-kān honā yabġi-nā ʔal-mawto lā ʔebġā-h p-kn hnʾ ʾrd grḥw lʾ yrdnʾ pha kān honā ʾarād gorḥo lā yorednā And he (the god Obodas) acts; there will be no ransom nor scar And if death should seek us, may he not allow it to obtain (its goal) and if a wound should desire (a victim), let it not desire us (2) KRS 2453, Safaito-Hismaic, undated (Al-Jallad 2015)43 l ḥg mt w lẓ ṯrm la-ḥagga mōt wa-lāṯṯ̣ ̣ ṯarām f-mykn ḫlf lyly-h w-ʾwm-h pha-moyakān ḫalph layālayoh wa-ʔaywām-oh w-hʾ bʿl ybt w l-h bt w m nm wa-hāʔ baʕl yabīt wa-lā-hu bāta wa mā nām Mōt has held a feast; the scorner eats established is the alternation of his nights and days and, behold, Baʿl sleeps; he indeed slumbers but is not dead (3) Marabb al-Shurafāʾ War Song, undated but probably 1st c. BCE-1st c. CE (Al-Jallad 2017b) (see 5.1) l ġyrʾl bn ġṯ ḏ ʾl ḥẓy w rḥl m-ʾhl-h le-Ġayyār-el ben Ġawṯ ḏī ʔāl Ḥaṯāy ̣ wa-raḥala meʔ-ʔahl-oh f ḥll-h m-ḥrb pha-ḥolūl-oh meḥ-ḥarb f h-ym hn ʾḫr ḥll pha-hay-yawma honā ʔāḫer ḥolūl rʾs ḏkrt raʔosa ḏekrata f h-ym hn ʾḫr ḥll pha-hay-yawma honā ʔāḫer ḥolūl ʿny mn ḫṣf ʕoneya man ḫoṣepa f h-ym hnʾ ʾḫ[r] ḥll pha-hay-yawma honā ʔāḫer ḥolūl ḥdd w ṯwy b-h-rḍt w ḫ{r}ṣ ḫl-h skrn yr{b} f-h b-q{l} fz-h f h lt slm 43 Vocalization is hypothetical based on Safaitic but this text reflects an entirely different register and perhaps is much older than the rest of the Safaitic corpus. 133 MHGA, v. 2020-1 ḥaddada wa-ṯaweya be-har-rawṣ́at wa-ḫaraṣa ḫāl-oh sakrāna yarobb phū-h be-qawl phawz-oh pha-hā-llāt salema By Ġayyār-el son of Ġawṯ of the lineage of Ḥaṯāy ̣ and he left his family And may his halting be (only) for war so let here this day be the final encampment Foremost fame! so let here this day be the final encampment Those who return suffer so let here this day be the final encampment He went to the boundary of the land and alighted in the meadow and kept watch for his maternal uncle Sakrān, his mouth exalting (him) saying ‘may good fortune be his’; So O Allāt may he be secure 5.3 Funerary Inscriptions (1) Namārah inscription NAB (328 CE; southern Syria); ty nfš mrʾlqyš br ʿmrw mlk ʾl-ʿrb kl-h dw ʾšr ʾl-tg tī naps marʔal-qays BAR ʕamro malk ʔal-ʕarab koll-ah ḏū ʔasar ʔal-tāg w mlk ʾl-ʾsryn w nzrw w mlwk-hm w hrb mdḥgw ʿkdy w- gʾ wa-malk ʔal-ʔasurayn wa-nizāro wa-molūk-hom wa-harraba maḏḥigo ʕakdāy wa-gāʔ b-zg-h py rtg ngrn mdynt šmrw mlk mʿd w nḥl b-bny-h be-zagg-oh phī rotog nagrān madīnat śammaro malk maʕadd wa-naḥḥal be-banī-h ʾl-šʿwb w wkl-hm p ršw l-rwm f lm yblʿ mlk mblʿ -h ʔal-śoʕūb _____ pha lam yabloġ malk mablaġ-oh ʿkdy hlk šnt 223 ywm 7 b-kšlwl blsʿd dw wldh ʕakdāy halaka ŠNT 223 yawm 7 be-kaslūl be-l-saʕd ḏū walada-h This is the funerary monument of Marʾalqays son of ʿamrō king of all the Aras, he who bound on the diadem, and king of the two Syrias and of Nizār and their masters and he put Maḏḥig to flight thereafter and brought his standard into the gates of Nagrān, the city of Šammar, king of Maʿadd; and he divided among his children the peoples 134 MHGA, v. 2020-1 and appointed them (for discussion); thus, no king has achieved his rank; thereafter, he died the year 223, on the 7th day of Kaslūl...(perhaps, in happiness, and with heirs). JSNab 17 Nab (267 CE, Madāʾin Ṣāliḥ; latest edition Fiema et al. 2015) JSNab 17 (Aramaic is bolded) dnh qbrw ṣnʿ-h kʿbw br DNH qabro ṣanaʕa-h kaʕbo BR ḥrtt l-rqwš brt ḥāreṯat le-raqōš BRT ʿbdmnwtw ʾm-h w hy ʕabdo-manōto ʔemm-oh wa-hī hlkt py ʾl-ḥgrw halakat fī ʔal-ḥegro šnt mʾh w štyn sanat MʾḤ W ŠTYN w tryn b-yrḥ tmwz w lʿn W TRYN B-YRḤ TMWZ wa-laʕan mry ʿlmʾ mn yšnʿ ʾl-qbrw MRYʿlmʾ man yośanneʕ ʔal-qabro d[ʾ] w mn yptḥ -h ḥšy (w) ḏā wa-man yaftaḥ-oh ḥaśay wld -h w lʿn mn yqbr w {y}ʿly mn -h wold-oh wa-laʕan man yaqbor wa-yaʕlay men-noh “(1) This is the tomb which Kaʿbō son of Ḥāreṯah built (2) for Rqwš daughter (3) of ʿbdmnwtw his mother, and she (4) died in ʾal-Ḥegrō (= Ḥegrā) (5) in the year one hundred and sixty (6) two in the month of Tammūz so may (7) Mry-ʿlmʾ (lit. lord of eternity) curse whosoever alters44 this tomb (8) or opens it except (9) his children and may he curse whosoever buries or removes from it [a body].” Vogue 404.1, Safaitic l ksṭ … w wlh ʿl-bn-h zʾm w bny l-bn-h h-nfs le-kāseṭ wa-waleha ʕal-ben-oh zāʔem wa-banaya le-ben-oh han-naphsa The sense of the root šnʿ ‘alter’ is found in Aramaic but is not known in Classical Arabic, but it is uncertain if the word had this sense in Old Arabic as well, so I have not bolded it. 44 135 MHGA, v. 2020-1 By Ks¹ṭ … t and he was distraught for his son, who had died, and he built for his son this funerary monument. HaNSB 307, Safaitic l sʿdlh bn ʾs bn ẓnʾl bn ḥyn ḏ-ʾl mʿyr w ḏ-ʾl frṯ w tśwq ʾl-ʾhl-h f h lt s¹lm w qbll w ġnmt w bny ʾ-nfs w dʿy ʾl- [l]t ʿl- mn yḫbl-h le-saʕdallāh ben ʔaws ben ṯann-el ben Ḥayyān ḏī ʔāl moʕayyer wa-dī ʔāl pharaṯ wạ taśawwaqa ʾel-ʾahl-oh pha-hā-llāt salāma wa qeblāla wa-ǵanīmata wa-banaya ʔannapsa wa daʕaya ʾel-llāt ʕal-man yoḫabbel-oh By Saʕdallāh son of ʔaws son of Ẓannʾel son of Ḥayyān of the lineage of Mʿyr and of the lineage of Frṯ: and he longed for his family and so, O Allāt, may there be security, reunion with loved ones, and spoil; and he built the funerary monument and called upon Allāt against anyone who would damage it [the funerary monument]. JSLih 384 Dad nfs ʿbdsmn bn zdḫrm ʾlt bnh slmh bnt ʾsʾrśn nafs ʕabd-samīn bin zayd-ḫarm ʔallatī banah salmah bint ʔaws-ʔarśān The funerary monument of ʿbdsmn son of Zdḫrm which Slmh daughter of ʾsʾrśn has built. 5.4 Prayers KRS 68, Safaitic h śʿhqm {ṣ}my nqt f {ʾ}{n}k bġy-h w qf{y}t-h {w} b-ḫfrt-k fltn m-mt hā-śayʕ-haqqawm ṣammaya nāqata pha-ʔennak boġy-oh wa-qaphyat-oh wa-beḫaphrat-ak pholtān mem-mawt O Shayʕ-haqqawm, he sacrificed a she-camel; for you are indeed whom he seeks and whom he follows and through your guidance comes deliverance from death. RWQ 73, Safaitic ḥḍr b-ʾẓmy h lt w h ḏs²r lʿn ḥwlt hḏ ʾṯm w wgm ʿl-ṯrm f h lt w y ḏs²r f h ds²r m ẓlm ms¹k f bqr ḥaṣ́ara be-ʾẓmy hā-llāt wa-hā-ḏū-śarē laʕʕenū ḥawalata haḏḏū ʔaṯamū wa-wagama ʕal-ṯaram pha-hā-llāt wa-yā ḏū-śarē pha-hā-diśar maṯ-ṯ̣ alama māseka pha-baqqerū ̣ he camped by permanent water near ʾẓmy; O Allāt and O Ḏu-śarē, curse the Ḥawalit (tribe) who acted wrongfully and he grieved for Ṯrm, so, O Allāt and O Ḏū-śarē, then O Diśar, whosoever would/has oppress(ed) Māsek, split him in two. 136 MHGA, v. 2020-1 KJC 46 Hismaic w m ḥll ḍyr-h wa-maḥ-ḥallala ṣ́eyār-oh ht ʿśw w rsl hāt ʕeśāwa wa-resla smʿt ḏśry w ktby sameʕat ḏū-śaray wa-kotbay And whosoever encamps (in the desert) on account of his sins Give [an offering] of an evening meal and milk that Ḏūśaray and Kutbay may hear Wādī Ram Hismaic (Macdonald 2018) l ʾbs¹lm bn qymy d ʾl gśm w dkrt-n lt w dkrt lt wśyʿ-n kll-hmle-ʔab-salām ben qaymay dī ʔāl gośam wa-dakarat-nā llāto wa-dakarat llāto aśyāʕa-nā kelāla-hom By ʾbslm son of Qymy of the lineage of Gśm. And may Allāt be mindful of me [or us] and may Allāt be mindful of all our companions. AWS 237 Safaitic l ḫzmʾ bn kn h-gml w qṣy-h ʾm m ʿwr h rḍw f l yʿwr m ʿwr w l yqʾ b ṣdq le-ḫazmāʔ ben kawn hag-gamal wa-qaṣaya-h ʔemmā maʕ-ʕawwara hā roṣ́aw phalyoʕawwar maʕ-ʕawwara wa-le-yeqqaʔ be-ṣadīq By Ḫazmāʔ son of Kawn is this camel and he carved it; if one would efface (it), O Roḍaw let the one who would efface it be made blind and let him be thrown out (of his grave) by a friend 5.5 Dedicatory and Narrative Ḥarrān, Arabic script 568 CE (Fiema et al. 2015) ʾnʾ srḥyl br ṭlmw bnyt dʾ ʾlmrṭwl snt 463 bʿd mqsd [mqds?] ḥybr nʿm ʔanā śaraḥīl BR Ṯā ̣ lemo banayt ḏā (ʔa)l-marṭūl sanat 463 beʕad maqsad (=maqdas?) ḫaybar naʕām I, Śaraḥēl son of Ṯā ̣ lemō, built this martyrium the year 463 on behalf of [the priest (?)] of Ḫaybar in grace. 137 MHGA, v. 2020-1 Narrative HaNSB 304, Safaitic l ḏl bn śrk bn rbḥ ḏ-ʾl qmr w mṭy f h śʿhqm ġnmt w rmy b-rmḥ-h w ḫzr b-sf-h f mrq kll slsl-h f w gdʿwḏ ġnmt w slm w ḫlf l-slḥ-h m-ʾl nbṭ w ʿwr ḏ ḫbl le-ḏayl ben śarīk ben rebḥ ḏī ʔāl qamar wa-maṭaya pha-hā-śayʕ-haqqawm ġanīmat wa-ramaya be-romḥ-oh wa-ḫazara be-sayph-oh pha-marraqa kelāla selsāl-oh wagaddo-ʕawīḏ ġanīmat wa-salām wa-ḫalph le-selāḥ-oh meʔ-ʔāl nabaṭ wa-ʕawwer ḏā ḫabbala By Ḏl son of S²rk son of Rbḥ of the lineage of Qamar and he journeyed in haste so, O S²ʿhqm, grant spoils; and he cast his lance and struck with his sword, then threw off all his chains of bondage, so O Gaddo-ʕawīḏ, grant spoil and security and compensation for his weapons from the Nabataeans, and blind him who would obscure [this inscription]. C 2446 l sʿd bn mrʾ bn nr w wgm ʿ[l-] ʾẖ -h nr qtl[-h ] ʾl {n}bṭy [ ] {r}ʿy nʿm ʿwḏ w ḍf f h lt mʿmn w ʾlt dṯn w gd[ʿ]{w}ḏ w gdḍf ṯʾr m- ḏ ʾslf w wlh k{b}{r} sḥr ʿl-ʾḫ-h ḥbb-h l-ʾbd le-saʕd ben marʔ ben nūr wa-wagama ʕal-ʔaḫī-h nūr qatal-oh ʔal-nabaṭeyy rāʕeya naʕām ʕawīḏ wa-ṣ́ayph pha-hā-llāt maʕmān wa-ʔelat-daṯan wa gaddo-ʕawīḏ wagaddo-ṣ́ayph ṯaʔr meḏ-ḏī ʔaslapha wa-waleha kabīra sāḥera ʕal-ʔaḫī-h ḥabīb-oh leʔabad By Saʕd son of Marʔ son of Nūr and he grieved {for} his brother Nr, {whom} the Nabataean killed while pasturing the livestock of (the tribe of) ʕawīḏ and Ṣ́ayf; so, O Lt-Mʿmn and ʾlt-Dṯn and Gd-ʿwḏ and Gd-ḍf, he will have vengeance against him who committed this act; and he was constantly distraught with a broken heart over his brother, his beloved forever. 5.6 Votive Madaba Inscription, Hismaic (Graf and Zwettler 2004)45 l flhn bn ḥnn bn ʾtm ḏʾl [nt](g) w sqm l-ʾlh Ṣʿb f tḍrʿ w tʿny w tś[d](d) l-h b-kll m fʿl li-falhān bin ḥonayni bin ʔatmi ḏī ʔāli natgi wa-saquma li-ʔelāhi Ṣaʕbi pha-taṣ́arraʕa wa-taʕānaya wa-taśaddada la-hu bi-kilāli mā phaʕala w nḏr ʾrbʿt ʾs¹lt m-nrt w ʿfnt w ytḥl b-ṣḥry w llk trḥm ʿly w ḏkrt lt ʾs²yʿ-n kll-h(m) wa-naḏara ʔarbaʕa ʔasliʕat min-nīrata wa-ʕaphanata wa-yatḥalla bi-ṣaḥrāya walawlā-ka taraḥḥama ʕalayya wa-ḏakarat allātu ʔaśyāʕa-nā kilāla-hum 45 I have vocalized this text based on the En Avdat inscription and transcriptions of Nabataean Arabic vowels. 138 MHGA, v. 2020-1 .... w lʿnt lt mn yḫ[r]bs² wqʿ-n ḏ wa-laʕanat allātu man yuḫarbiś waqʕa-nā ḏā By Flhn son of Ḥnn son of ʾtm of the lineage of Ntg and he became for the sake of the god Ṣaʿb and he has been reduced to abject supplication and became afflicted despite having exerted himself on his behalf through all that he has done and he vowed four commodity lots of indigo and verdigris pigments ... and these so that you might show mercy upon him; and may Allāt be mindful of all of his companions...and may Lt curse whosoever would obscure this inscription of ours. 5.6 Arabic texts in Greek letters Graeco-Arabic inscription A1 (Al-Jallad and al-Manaser 2015) 139 MHGA, v. 2020-1 ΑΥΣΟΣΟΥΔΟΥ ΒΑΝΑΟΥΧΑΖΙΜ ΜΟΥΑΛΙΔΑΜΙΑΘΑ ΟΥΑΜΙΣΕΙΑΖΑΘΑΟΕΩ ΑΒΑΝΑΑΑΔΑΥΡΑ ΑΟΥΑΕΙΡΑΥΒΑΚΛΑ ΒΙΧΑΝΟΥ 1Αυσος Ουδου 2Βαναου Χαζιμ3μου αλ-Ιδαμι αθα4οα μι- Σεια ζαθαοε ω̣5 α Βαναα α-δαυρα6 αουα ειραυ βακλα7 βι-Χανου[ν]8 ʾAws (bin) ʿūḏ (?) (bin) Bannāʾ (bin) Kazim ʾalʾidāmiyy ʾatawa mis-seʿīʿ śatāw wa Bannāʾa ʾad-dawra wa yirʿaw baqla bi-kānūn Translation: ʾAws son of ʿūḏ (?) son of Bannāʾ son of Kazim, the ʾIdāmite, came from Sīʿ to spend the winter with Bannāʾ in this place and they pastured on fresh herbage during Kānūn 140 MHGA, v. 2020-1 Edition of The Damascus Psalm Fragment46 v.20 ---σαχρ(α)ὐ •φασέ λετ•μαϳάὑ1 • οελευδιεὑ•φά• δατ• λεγαλ•οαχουβζ ϳεκ•διρ•ϳουγ•τι2 έυ•ϳου•ὑεϳει• μάϳδεὑ•λιχ3 χειγ•βὑϳ ---- ṣaḫr(a)h fa-sēlet mayyah wel-ewdiyeh fāḍat leʕal wa-ḫubz yeqdir yuʕtī eu yuheyyī māy(i)deh li-šiʕb-hu(hi) [sic] [*li-siʕbi-h(?)] [Forasmuch as he smote] the rock, and water flowed, and the valleys emptied; perhaps he will be able also to give bread or prepare a table for his people? Notes: 1) The other comparable manuscripts have in Arabic ‫[ االمياه‬al-ʔamyāh] and ‫[ امياه‬ʔamyāh], and while there may be space at the beginning of the word for a few letters, the Alpha following the Mu suggests a different pronunciation, akin to Levantine Arabic mayya and possibly Safaitic myt [mayyat]. 2) Corriente remarks that the syntax of this line calques the Greek.47 3) The facsimile of Violet gives an extra Chi here, while it is not apparent on the photograph. 46 47 From Al-Jallad (forthcoming) Corriente, “Psalter Fragment,” p. 304. 141 MHGA, v. 2020-1 v.21 λιδέλικ•σεμιγ ελραβ•φααμ τεναγ• οελναρ•εχτεγα λετ•φη•ϳαγκουβ οα•ρυγζ1•σαγ(αδ) γαλα•ϳσραηλ li-ðēlik semiʕ el-rab fa-ʔamtenaʕ wel-nār ʔešteʕalet fī yaʕqūb wa ruǧz saʕ(ad) ʕalā Israel Therefore the Lord heard, and he was provoked. Fire was kindled in Jacob, and wrath went up against Israel. Notes: 1) Corriente identifies ruǧz as a loanword from Aramaic rugzā.48 The other manuscripts have this form with the article. v.22 λιεν(ναὑ)μ (λαμ) ϳουμι(νου) βιλλαυ οα•λ(αμ) (ταοα)κκελου1 γαλα χαλασυ•2 li-ʔen(nahum) (la)m yūmi(nū) billāh wa-lam (yuwa)kkelū ʕalā ḫalāṣ-h Because they had no faith in God, and did not trust in his deliverance. 48 Corriente, “Psalter Fragment,” p. 306. 142 MHGA, v. 2020-1 Notes: 1) Violet renders this line as wa lā tawakkalū,49 Kahle as wa lā ittakalū,50 Blau follows Violet.51 The other manuscripts, however, give two variants: ‫( ال توكلوا‬Sinai Ms. Gr. 34 and 36) and ‫لم‬ ‫( يرجون‬Sinai, Ms. Gr. 35). The surviving letters can only be the former, yet the six lacunae are best restored with the negator lam rather than lā. 2) Kahle and Blau read χαλασυι (v.22), but on the tracing of Violet, the final Iota is barely visible, represented only by a small dot.52 The photographs show that this small dot is nothing but a word divider, and therefore the reading must be amended to χαλασυ. v.23 οα αμαρ ελσιχεβ μιν•φαυκ οα αβοαβ ελσε1 σαμα•φατεχ• wa ʔamarel-siḥāb min fawq wa ʔabwāb el-se…samā fateḥ And he commanded the clouds from above, and opened the doors of heaven Notes 1) The scribe runs out of space to complete the word [semā] and so begins writing it anew on the following line. Curiously, he uses the [a] allophone of *a in his second attempt. v.24 οα•αμ•ταρ•λεὑμ• 49 Violet, “Psalmfragment,” p. 390. 50 Kahle, Die Arabischen Bibelübersetzungen, p. 32. 51 Blau, Handbook, p. 71. 52 Kahle, Die Arabischen Bibelübersetzungen; Blau, Handbook. 143 MHGA, v. 2020-1 μ(ανν)α•λια (κυλο)υ•1 (οα)(χουβ)ζ2•μιν•ελ (σεμα)αγ•τάὑμ wa ʔamṭar lehum m(ann)a liyā(kul)ū (wa) (ḫub)z min el-(semā) ʔaʕṭā-hum And he rained Manna upon them to eat, and gave them the bread from heaven. Notes 1) The lacunae permit the restoration of four letters, which implies that short [u] was written here with Ypsilon. The letter after the Iota is heavily damaged in the photograph, and could plausibly be an Alpha or a Lambda. If one restores it as λιλ, then it would suggest a reading similar to Sinai, Mss. Gr. 35 and 36 ‫لالكل‬. However, in Violet’s copy, but not in the surviving photograph, the word terminates in an Ypsilon, favoring ‫ لياكلون‬as in Sinai, Ms. Gr. 34, but with a true subjunctive form lacking the nūn. 2) The lacunae permit the restoration of six letters, four for the word ‘bread’ and two for the conjunction οα /wa/, rendering Greek και ἄρτον. v.25 (χουβ)ζ ελμηελεικε1 (ακ)ελ•ινσέν2 (χα)βα(γ)3 βάγαθ λα•ὑμ•λεϳτεμέλ•λευ-4 (ḫub)z el-melēyke (ʔak)el ʔinsēn (ša)ba(ʕ) baʕaṯ la-hum ley(i)temellew Man ate angels’ bread; he sent them provisions that they may be filled. 1) The scribe forgot to write the Mu then added a superscript μη. The diphthong is spelled without the elongated Iota and the feminine ending lacks the Hypsilon. It would appear that the scribe was careless in the writing of this word, transcribing it according to normal Greek 144 MHGA, v. 2020-1 orthography and leaving out the conventional use of Elongated Iota and Hypsilon to represent consonantal [y] and [h], respectively. 2) The indefinite form here disagrees with all other manuscripts, which have ‫االنسان‬, cf. mayyah (v. 20). 3) Corriente takes šabaʕ as an adverbial complement of the verb ʔakal, rendering “the men ate the angels’ bread until being satiated.”53 In fact, šabaʕ begins a new clause and is the object of baʕaṯ “he sent”, the entire clause being: šabaʕ baʕaṯ la-hum lay(i)teméllew “he sent to them provisions in order that they be filled”. This renders accurately the Greek: ἐπισιτισμὸν ἀπεστειλεν αὐτοῖς εἰς πλισμονὴν. 4) On the spelling and rendering of this word, see §. v.26 α•ὑάγ•ελ•τεϳμ(αν)1 μιν•ελ•σεμα οα•ατε•βη κου ετὑ•ελ•γα σιφ2 ʔahāǧ el-teym(an) min el-semā wa ʔatē bi-quwwet-uh el-ʕāṣif He removed the south wind from heaven; and by his might he brought in the south-west wind. Notes: 1) The name of the South Wind in Classical Arabic is al-ǧanūb. The use of Teym[an] here might be an Aramaicism, tayman ‘south’. An identical term is used in the Hebrew Bible, têmān. 2) This term for the southwest wind is unknown in Classical Arabic. The term ʕāṣif is applied to rīḥ to denote a wind that blows violently (Lane, 2064b). The term is attested in the QCT (Q 10:22). 53 Corriente, “Psalter Fragment,” p. 309. 145 MHGA, v. 2020-1 v.27 οα•αμ•ταρ•γαλεϳ ὑμ•μίθλ•ελτυ ράβ•λυχουμ οαμίθλ•ραμλ ελ βου•χουρ•τη ουρ•μυγνεχαὑ wa ʔamṭar ʕaley-hum miṯl el-turāb luḥūm wa miṯl raml el-buḥūr ṭiyūr muǧneḥah And he rained upon them flesh like dust, and like the sand of the seas feathered birds. v.28 φα•οα•καγ•ατ φη•οασατ•γασ κερ•ὑμ χαυλ χη•έμ•ὑμ fa-waqaʕat fī wasaṭ ʕasker-hum ḥawl ḫiyēm-hum And they fell into the midst of their camp, surrounding their tents. v.29 φα•ακελου•οα• χεβιγου•γεδ δα• οα•χε•ὑοετ•ὑμ γεβ•λαὑμ1 146 MHGA, v. 2020-1 fa-ʔakelūwa šebiʕū ǧeddā wa šehwet-hum ǧēb la-hum So they ate, and were greatly filled; and he brought to them their desire. Notes: 1) The verb ǧēb “bring” is typical of the modern dialects of Arabic, derived from ǧāʔa bi- ‘to come with’. The verb translates Greek ἤνεγκεν ‘he brought’. This phrasing agrees with Sinai, Ms. Gr. 35, against ʔatā-hum bi-šahwat-hum in 34 and 36, and more closely matches the syntax of the Greek. v.30 (λα)μ ϳουγ•δεμου• (χ)ευοετὑμ• οα•γινδ•μα•κεν ελ•ταγαμ•φη φα•ὐ•ὐμ2 (la)m yuʕdemū (š)ehwet-hǔmwa ʕindmā kēn el-ṭaʕām fī fāh-hum They were not denied their desire; but when their food was in their mouth Notes: 1) Blau (2002: 70) transcribes this word incorrectly as φαὑμ.54 The plural ‫ افواه‬is used in 34 and 36. v.31 (o)α•ρυγζ•αλλάὐ (o)a rǔǧz allāh 54 Blau, Handbook, p. 70. 147 MHGA, v. 2020-1 then the wrath of God [rose up against them, and slew the fattest of them, and overthrew the choice men of Israel]. v.51 τεγ•β1 μεσε2 teʕb mese [and smote every first-born in the land of Egypt; the first-fruits of their] labors [in the] tents [of Cham]. Notes: 1) Ms.Gr. 34 and 36 have ‫ تبعهم‬suggesting teʕb-hum. 2) This fragment most likely reflects μεσεκεν/mesēken/, the plural of μεσκεν /mesken/ attested in v.55, which is found in Ms.Gr. 34 and 35. v.52 οα•σακ• γανεμ οα•ασ•γ1 μιθλ φιλ•β2 wa sāq ġanem wa aṣʕ miθl 148 MHGA, v. 2020-1 fil-b And he drove (his people like) sheep; he led (them) as (a flock) in the wi(lderness). 1) Violet restores this word as the causative ‫أصعد‬, a suitable rendition of Greek ἀνήγαγεν ‘he led up’, and this is found in Ms.Gr. 34 and 36. 2) Violet restores this as ‫في البرية‬. v.53 οα•αϳα•δ1 βερρί2 ϳεγζαγ(ου) οα•αγ•δ γαττα βάχρ• wa ʔahād---berrī---yeǧza(ʕū)---wa aʕd---ġaṭṭā baḥr And he guided [them with] hope, [and] they [did not] feel fear; [and the] sea covered [their enemies]. Notes: 1) Violet renders this ‫هداهم‬in Arabic, and this is found in Ms.Gr. 34 and 36, but the PF clearly attests an Alpha before the verb.This would seem to be a mixed form, with a causative prefix α and then the G-stem had(ā). If this were a true causative it would have been spelled αjδα 149 MHGA, v. 2020-1 /ʔahdā/. Less likely is the possibility that this reflects the gahawa-syndrome, i.e. the insertion of an [a] after a guttural. 2) On this word, see the discussion in §. All other manuscripts differ from the PF in having ‫عىل الرجا‬. v.54 οα•αδ•χ(αλὑμ) ϳλέ•γεβ(ελ)1 καδ•σὁ (ελ)2 γέβελ•ἁ(δα) ελλεδι•α(χα) δετ•ϳεμ(ινὑ) wa ʔadḫ(al-hum) ʔilē ǧeb(el) qads-oh (el) ǧebel hā(ðā)---ʔelleðī ʔa(stafā)det yemīn-uh3 And he brought (them) in to the mountain of his sanctuary, this mountain which his right hand had purchased. Notes: 1) The PF literally renders the Greek; the other manuscripts do not use a preposition, ‫وادخلهم‬ ‫( جبل‬Ms.Gr. 34, 36) and ‫( وادخلهم طور‬Ms.Gr. 35). 2) On the rendition of καδσὁ, see the discussion in ###. 3) Violet restored this verb as ‫ اخذت‬but Vollandt (Appendix I) restores ‫( استفادت‬Ms.Gr. 34 and 35) from a majority reading. v.55 150 MHGA, v. 2020-1 οα•αχ•ραγ ---οε1 (ε)λουμε(μ) Οα αυραθ ελ•μιρε(θ) βιλ-οαασ•κ με•σε κα•β(εjλ)2 (ϳσ)ραι(λ)3 wa ʔaḫraǧ ----oe (e)l-ʔume(m) wa ʔawraṯ el-mirē(ṯ) bil--wa ʔask--mese--qab(ēyil) (is)rāi(l) (And he cast out) the nations (from before them, and) caused (them) to inherit by a line of inheritance, (and) made the tribes of Israel to dwell in (their) tents. 1) Vollandt (Appendix I) restores ‫ جوههم‬instead of Violet’s ‫وجههم‬. This would be the first use of Omicron-Epsilon to spell ū. 2) The restoration of the elongated Iota is conjectural based on the spelling of ābāy(i)hum as αβαjὑμ. 151 MHGA, v. 2020-1 3) Violet restores this verse as ‫واسكن في مساكنهم قبائل اسرائيل‬. The vocalization of μεσε(κεν) has been discussed above (v. 51, n.2) This use of Iota in the spelling of the final syllable of Israel here rather than Eta as earlier reflects Iotacism. v.56 οα•αβ•τε•λεῦ•οα μαρ•μαροῦ• ελ•ϳ•λέὑ•ελγαλη οα•χε•ὑα•δ(α)τὑ1 λαμ•ϳεχ•φα•δοῦ• wa ʔabtelew wa marmarū el-ʔilēh el-ʕālī wa šehād(ā)t-uh lam yeḥfaḏ̣ū Yet they tempted and provoked the highest God, and kept not his testimonies. 1) Corriente (2007) reads this word as “šahādtu”, a singular, against the plural Greek μαρτύρια which it translates.55 It is possible that the scribe omitted the Alpha by mistake, as there are no examples of the syncope of *a in this dialect. In Violet’s facsimile, there is a lacuna between the Delta and Tau, where the remnants of an Alpha can be restored. The photograph is unclear in this area. All other manuscripts have ‫شهاداته‬. v.57 φα•ανκα•λε•β(ο)υ•1 οα•γα•δα•ρου• μιθλ•α•βα• ϳ•ὑμ αν•κα•λε•βου μιθλ•ελ•καυ•σ•ελ γαυγέ 55 Corriente, “Psalter Fragment.” 152 MHGA, v. 2020-1 fa ʔanqalebū wa ġadarū miṯl ābāy(i)-hum ʔanqalebū miṯl el-qaws el-ʕawǧē And they turned back and acted treacherously, like their fathers, they turned back, like a crooked bow. 1) All other manuscripts have ‫ورجعوا‬. v.58 οα[α]σ•χα•τοῦ•ὑ β•αυθάν•ϳ•ὑμ1 οα•βη•μεν•χου•τέ•τη•ὑμ•α• γα•ροῦ•υ wa (ʔa)sḫaṭū-h bi-ʔawθāni-hum wa bi-menḥūtēti-hum ʔaġārū-h And they provoked him with their high places, and moved him to jealousy with their graven images. Notes: 1) The author chose to translate βουνοίς αυτών “their hills” with Arabic ʔawṯān, the plural of waṯan, an ‘idol’, and may have been confused by the following word, γλυπτοίς. Only Ms.Gr. 36 has ‫وثانهم‬. v.59 σεμιγ•αλλάὑ• οα•τεγάφελ• (οα)αφ•σέλ•1γεδ• (δα) λι•ϳσρα(ιλ) -λ- 153 MHGA, v. 2020-1 semiʕ allāh wa teġāfel (wa) ʔafsel ǧed(dā) – li-isra[il] God heard and lightly regarded them, and greatly despised Israel. Notes: 1) On the rendering of the verb αφ•σέλ, see note #. v.60 οα•ακ•σα•χαϳμετ• σεϳλουμ• ελ-μεσ•κεν•ελ• λεδι•εσ•κεν1•φιλ• βαχερ waaqṣā ḫaymet seylūm el-mesken elleðī ʔesken fil-bašer and he rejected the tabernacle of Shiloh, his tent where he dwelt among men. 1) The C-stem (form IV) matches Ms.Gr. 35, 36. v.61 οα•ασ•-ε- λιλ• σεβ• οευ- wa ʔas(l)e(m) lilseb(ī)• (q)oe(t-hum) And he gave their strength into captivity. 154 MHGA, v. 2020-1 Works Cited Al-Jallad, A. 2015. “̀Echoes of the Baal Cycle in a Safaito-Hismaic Inscription.” Journal of Ancient Near Eastern Religions 15: 5–19. Al-Jallad, Ahmad. forthcoming. “One Wāw to Rule Them All: The Origins and Fate of Wawation in Arabic.” In Scripts and Scripture, edited by Fred M. Donner and Rebecca Hasselbach. Chicago: Oriental Institute. ———. forthcoming. The Damascus Psalm Fragment: Middle Arabic and the Legacy of Old Ḥigāzī. LAMINE 2. Chicago: OI. ———. 2014a. “Aṣ-Ṣādu Llatī Ka-S-Sīn – Evidence for an Affricated Ṣād in Sibawayh?” Folia Orientalia 51. ———. 2014b. “On the Genetic Background of the Rbbl Bn Hfʿm Grave Inscription at Qaryat Al-Fāw.” Bulletin of the School of Oriental and African Studies 77 (03): 445– 465. https://doi.org/10.1017/S0041977X14000524. ———. 2014c. “Final Short Vowels in Gə‘əz, Hebrew ’attâ, and the Anceps Paradox.” Journal of Semitic Studies 59 (2): 315–27. https://doi.org/10.1093/jss/fgu003. ———. 2015. An Outline of the Grammar of the Safaitic Inscriptions. SSLL 80. Leiden: Brill. ———. 2017a. “Graeco-Arabica I: The Southern Levant.” In Arabic in Context: Celebrating 400 Years of Arabic at Leiden University, edited by Ahmad Al-Jallad, 99–186. Leiden: Brill. ———. 2017b. “Pre-Islamic ”Ḥamāsah” Verses from Northeastern Jordan: A New Safaitic Poetic Text from Marabb Al-Shurafāʾ, with Further Remarks on the ʿĒn ʿAvdat Inscription and KRS 2453.” Proceedings of the Seminar for Arabian Studies 47: 117–28. ———. 2017c. “The Arabic of the Islamic Conquests: Notes on Phonology and Morphology Based on the Greek Transcriptions from the First Islamic Century.” Bulletin of the School of Oriental and African Studies 80 (3): 419–39. ———. 2018a. “The Earliest Stages of Arabic and Its Linguistic Classification.” In Routledge Handbook of Arabic Linguistics, edited by Elabbas Benmamoun and Reem Bassiouney, 315–31. New York: Routledge. ———. 2018b. “What Is Ancient North Arabian?” In Re-Engaging Comparative Semitic and Arabic Studies, edited by Na ’ama Pat-El and Daniel Birnstiel, 1–43. Wiesbaden: Harrassowitz. Al-Jallad, Ahmad, and Ali al-Manaser. 2015. “New Epigraphica from Jordan I: A PreIslamic Arabic Inscription in Greek Letters and a Greek Inscription from NorthEastern Jordan.” Arabian Epigraphic Notes 1: 51–70. ———. 2016. “New Epigraphica from Jordan II: Three Safaitic-Greek Partial Bilingual Inscriptions.” Arabian Epigraphic Notes 2: 55–66. Al-Nassir, A. A. 1993. Sibawayh the Phonologist: A Critical Study of the Phonetic and Phonological Theory of Sibawayh as Presented in His Treatise Al-Kitāb. London; New York: Kegan Paul International. Bauer, H. 1912. “Mitteilungen zur semitischen Grammatik.” ZDMG 66: 103–14. Beeston, A. F. L. 1962. “Arabian Sibilants.” Journal of Semitic Studies 7 (2): 222–33. https://doi.org/10.1093/jss/7.2.222. Beeston, A.F.L. 1979. “Nemara and Faw.” Bulletin of the School of Oriental and African Studies 42 (1): 1–6. Blau, Joshua. 2002. A Handbook of Early Middle Arabic. Jerusalem: JSAI. Carter, M. G. 2004. Sibawayhi. London; New York: I.B. Tauris. 155 MHGA, v. 2020-1 Corriente, Federico. 2007. “The Psalter Fragment from the Umayyad Mosque of Damascus: A Birth Certificate of Nabaṭī Arabic.” In Eastern Crossroads: Essays on Medieval Christian Legacy, edited by Juan Pedro Monferrer-Sala, 303–20. Piscataway, NJ: Gorgias Press. Diem, Werner. 1982. “Die Entwicklung des Derivationsmorphems der t-Stämme im Semitischen.” Zeitschrift der Deutschen Morgenländischen Gesellschaft, 29-84, 132. Ephʿal, Israel. 1974. “‘Arabs’ in Babylonia in the 8th Century B. C.” Journal of the American Oriental Society 94 (1): 108–15. https://doi.org/10.2307/599734. ———. 1982. The Ancient Arabs : Nomads on the Borders of the Fertile Crescent, 9th-5th Centuries B.C. Jerusalem: Magnes Press. Eskoubi, Khaled. 1999. An Analytical and Comparative Study of Inscriptions from “Rum” Region, South West of Tayma (in Arabic). Riyadh: Ministry of Education, Deputy Ministry of Antiquities and Museums. Fiema, Zbigniew T., Ahmad Al-Jallad, Michael C.A. Macdonald, and Laïla Nehmé. 2015. “Provincia Arabia: Nabataea, the Emergence of Arabic as a Written Language, and Graeco-Arabica.” In Arabs and Empires before Islam, edited by Greg Fisher, 373– 433. Oxford: Oxford University Press. Hayajneh, Hani, Mohammad I. Ababneh, and Fawzi Khraysheh. 2015. “Die Götter von Ammon, Moab und Edom in einer neuen frühnordarabischen Inschrift aus Südost-Jordanien.” In Fünftes Treffen der Arbeitsgemeinschaft Semitistik in der Deutschen Morgenländischen Gesellschaft vom 15.–17. Februar 2012 an der Universität Basel, edited by V. Golinets, H.-P. Mathys, and S. Sarasin, 79–105. Münster: Ugarit-Verlag. Healey, J. 1993. The Nabataean Tomb Inscriptions of Mada’in Saleh. Journale of Semitic Studies Supplement 1. Oxford: Oxford University Press. Holes, Clive, ed. 2018. Arabic Historical Dialectology: Linguistic and Sociolinguistic Approaches. Oxford Studies in Diachronic and Historical Linguistics 34. Oxford: Oxford University Press. Hopkins, Simon. 1984. Studies in the Grammar of Early Arabic: Based upon Papyri Datable to before 300 A.H./912 A.D. London Oriental Series 37. Oxford; New York: Oxford University Press. Huehnergard, John. 1987. “‘Stative,’ Predicative Form, Pseudo-Verb.” Journal of Near Eastern Studies 46 (3): 215–32. ———. 1995. “Features of Central Semitic.” In Biblical and Oriental Essays in Memory of William L. Moran, edited by A. Gianto, 155–203. Rome: Pontificio Istituto Biblico. ———. 2004. “Afro-Asiatic.” In The Cambridge Encyclopedia of the World’s Ancient Languages, edited by R.D. Woodard, 138–59. Cambridge: Cambridge University Press. ———. 2017. “Arabic in Its Semitic Context.” In Arabic in Context: Celebrating 400 Years of Arabic at Leiden, edited by Ahmad Al-Jallad, 3–34. Studies in Semitic Languages and Linguistics 89. Leiden: Brill. Huehnergard, John, and Aaron D. Rubin. 2011. “Phyla and Waves: Models of Classification of the Semitic Languages.” In The Semitic Languages: An International Handbook, edited by Stephan Weninger, Geoffrey Khan, Michael Streck, and Janet Watson, 259–78. Handbücher Zur Sprach- Und Kommunikationswissenschaft 36. Boston-Berlin: De Gruyter Mouton. Isserlin, B. 1969. “The Nessana Papyri. The Greek Transcriptions of Arabic.” The Annual of Leeds University 7: 17–31. 156 MHGA, v. 2020-1 Kahle, Paul. 1904. Die Arabischen Bibelübersetzungen: Texte Mit Glossar Und Literaturübersicht. Leipzig: J. C. Hinrichs. Kaplony, Andreas. 2015. “On the Orthography and Pronunciation of Arabic Names and Terms in the Greek Petra, Nessana, Qurra, and Senouthios Letters.” Mediterranean Language Review 22: 1–81. Khan, Geoffrey. 2011. “Middle Arabic.” In The Semitic Languages: An International Handbook, edited by Stephan Weninger, Geoffrey Khan, Michael Streck, and Janet Watson, 817–35. Handbücher Zur Sprach- Und Kommunikationswissenschaft 36. Boston-Berlin: De Gruyter Mouton. King, Geraldine M. H. 1990. Early North Arabian Hismaic. PhD Dissertation. London: University of London, School of Oriental and African Studies. Kogan, Leonid. 2011. “Proto-Semitic Phonetics and Phonology.” In The Semitic Languages: An International Handbook, edited by Stephan Weninger, Geoffrey Khan, Michael Streck, and Janet Watson, 54–151. Handbücher Zur Sprach- Und Kommunikationswissenschaft 36. Boston-Berlin: De Gruyter Mouton. Kootstra, Fokelien. 2016. “The Language of the Taymanitic Inscriptions and Its Classification.” Arabian Epigraphic Notes 2: 67–140. Kropp, Manfred. 2017. “The ʿAyn ʿAbada Inscription Thirty Years Later: A Reassessment.” In Arabic in Context, edited by Ahmad Al-Jallad, 53–74. SSLL. Leiden: Brill. Livingstone, Alasdair. 1997. “An Early Attestation of the Arabic Definite Article.” Journal of Semitic Studies 42 (2): 259–61. Macdonald, M.C.A. 2000. “Reflections on the Linguistic Map of Pre-Islamic Arabia.” Arabian Archaeology and Epigraphy 11: 28–79. ———. 2004. “Ancient North Arabian.” In The Cambridge Encyclopedia of the World’s Ancient Languages, edited by R.D. Woodard, 488–533. Cambridge: Cambridge University Press. ———. 2009. “Arabs, Arabias, and Arabic before Late Antiquity.” Topoi 16: 277–332. ———. 2018. “Clues to How a Nabataean May Have Spoken, from a Hismaic Inscription.” In Near Eastern and Arabian Essays Studies in Honour of John F. Healey, edited by A.H.W. Curtis, M. Al-Hamad, and G.R. Smith, 231–39. Journal of Semitic Studies 41. Oxford: Oxford University Press. Moscati, Sabatino. 1964. An Introduction to the Comparative Grammar of the Semitic Languages : Phonolgy and Morphology. Wiesbaden: Otto Harrassowitz. Negev, Avraham. 1991. Personal Names in the Nabatean Realm. Qedem 32. Jerusalem: Institute of Archaeology, Hebrew University of Jerusalem. Pat-El, Na ’ama. 2009. “The Development of the Semitic Definite Article: A Syntactic Approach.” Journal of Semitic Studies 54 (1): 19–50. Putten, Marijn van. 2017. “The Archaic Feminine Ending -at in Shammari Arabic.” Journal of Semitic Studies 62 (2): 357–69. https://doi.org/10.1093/jss/fgx026. Putten, Marijn van, and Phillip W. Stokes. forthcoming. “Case in the Quranic Consonantal Text.” Rabin, Chaim. 1951. Ancient West-Arabian. London: Taylor’s Foreign Press. Rebecca Hasselbach. 2006. “The Ventive/Energic in Semitic — A Morphological Study.” Zeitschrift Der Deutschen Morgenländischen Gesellschaft 156 (2): 309–28. Stein, Peter. 2004. “Zur Dialektgeographie Des Sabäischen.” Journal of Semitic Studies 49 (2): 225–45. Steiner, Richard C. 1982. Affricated Ṣade in the Semitic Languages. New York: American Academy for Jewish Research. 157 MHGA, v. 2020-1 Testen. 2017. Parallels in Semitic Linguistics: The Development of Arabic La- and Related Semitic Particles. BRILL. https://brill.com/view/title/1331. Van Putten, Marijn. 2017a. “The Development of the Triphthongs in Quranic and Classical Arabic.” Arabian Epigraphic Notes 3: 47–74. ———. 2017b. “The Feminine Ending as a Diptote in the Qur’anic Consonantal Text and Its Implications for Proto-Arabic and Proto-Semitic.” Arabica 64: 695–705. Versteegh, C. H. M. 1997. The Arabic language. New York: Columbia University Press. Violet, Bruno. 1901. “Ein zweispachiges Psalmfragment aus Damascus.” Orientalistische Litterature-Zeitung 10; 11: 384-423; 425-488. Weninger, Stefan, ed. 2011. The Semitic Languages: An International Handbook. Berlin, Boston: De Gruyter Mouton. https://doi.org/10.1515/9783110251586. Weninger, Stephan. 2011. “Reconstructive Morphology.” In The Semitic Languages: An International Handbook, edited by Stephan Weninger, Geoffrey Khan, Michael Streck, and Janet Watson, 151–76. Handbücher Zur Sprach- Und Kommunikationswissenschaft 36. Boston-Berlin: De Gruyter Mouton. Winnett, F.V., and W.L. Reed. 1973. “An Archaeological Epigraphical Survey of the Haʾil Area of Northern Saudi Arabia.” Berytus 22: 53–100. Zaborski, Andrzej. 1996. “Archaisms and Innovations in Arabic Independent Pronouns.” In Proceedings of the 2nd International Conference of l’Association Internationale Pour La Dialectologie Arabe Held at Trinity Hall in the University of Cambridge, 1014 September 1995., edited by Joe Cremona, Clive Holes, and Geoffrey Khan, 269– 74. Cambridge: Cambridge University Press. Zwettler, Michael J., and David F. Graf. 2004. “The North Arabian ‘Thamudic E’ Inscription from Uraynibah West.” Bulletin of the American Schools of Oriental Research 335: 53–89. 158