On the Voiceless Re%ex of *ṣ́ and *ṯ ̣ in pre-Hilalian Maghrebian Arabic1 Ahmad Al-Jallad* Abstract: This paper argues that the voiceless re%ex of the emphatics *ṣ́ (ḍād) and *ṯ ̣ (ẓāʾ) in some pre‑Hilalian Maghrebian Arabic dialects is in fact an archaism. These phonemes were voiceless in Old Arabic, as proven by Greek transcriptions from the pre-Islamic period, and so pre‑Hilalian Maghrebian Arabic may continue the original situation. The voiced re%exes, more common in other modern Arabic dialects and in the conventional pronunciation of Classical Arabic, are then interpreted as a later development. Keywords: Arabic dialectology, Maghrebian Arabic, emphatic consonants, pre‑Hilalian, Old Arabic, phonology 1 Introduction Advances in the comparative grammar of the Semitic languages in the 20th century have helped dispel the view of Classical Arabic (CAr) as the most archaic Semitic language.2 At almost every linguistic layer, CAr exhibits important innovations distinguishing it from Proto-Semitic (PS), and even from its more direct ancestor, Proto‑Arabic.3 There is at least one domain, however, in which this belief obtains – phonology. While CAr is unarguably conservative in that it keeps separate 28 of the 29 PS phonemes, merging only *s (s1) and *ts (s3) to [s], the phonetic realization of these phonemes is to be considered highly innovative. A reassessment of the evidence from the 1 The abbreviations used in this paper are – PS: Proto-Semitic; NWS: Northwest Semitic; CAr: Classical Arabic. 2 On these features, see HUEHNERGARD (forthcoming) and AL-JALLAD (forthcoming). 3 See AL-JALLAD (2014; 2015 a, b). * Faculteit der Geesteswetenschappen, Institute for Area Studies (LIAS), Leiden University, a.m.al-jallad@hum.leidenuniv.nl Pre-Hilalian Emphatics 89 pre‑Islamic period, however, suggests that Old Arabic4 was more conservative in this respect than CAr and most of the modern dialects (see AL‑JALLAD 2015b). An exception is some of the pre‑Hilalian dialects of North Africa. This paper will re‑examine the realization of the emphatics of these varieties in light of Old Arabic and PS, and argue that they in fact better re%ect the original Arabic situation. 2 The emphatics, an overview There is a virtual consensus today that the emphatic correlate in PS was glottalization.5 Glottalized consonants are preserved in the Modern South Arabian languages, Ethio-Semitic, and can be reasonably posited for Akkadian and early NWS. Since glottalization requires the full closure of the vocal cords, they cannot be con*gured for voice. Additionally, there are articulatory reasons to argue that the emphatic lateral and interdental had a stop‑onset.6 The following table lists the reconstructed values of the PS emphatic series with their CAr re%exes: Table 1: Proto-Semitic and Classical Arabic re-exes of the emphatics Proto-Semitic CAr (conventional pronunciation) *[tθ’] [ðʕ] *[t’] [tʕ] *[ts’] [sʕ] *[tɬ’] [ɮʕ] *[k’] [q] The exact developmental trajectory of these phonemes in Arabic is still debated by specialists and their Proto-Arabic values remain uncertain. HUEHNERGARD (forthcoming) suggests that pharyngealization was a feature of Proto‑Arabic while AL-JALLAD (2015b) leaves open the possibility 4 Old Arabic is used here to refer to the pre-Islamic dialects of Arabic attested in documentary sources, such as epigraphy and papyri, and not materials attributed to the pre-Islamic period by Arabo-Islamic traditions. 5 KOGAN (2011: 60+). 6 Glottalized sibilants and interdentals are extremely rare cross-linguistically, and so it may have been the case that these sounds were a+ricates, like *ṣ, see KOGAN (2011: 71). 90 A. Al-Jallad for at least some members of the series to have been realized as glottalics.7 In addition to the ancient evidence, there are some contemporary southwestern Arabian dialects of Arabic that realize these consonants as glottalized in certain environments, but this could be due to substrate in%uence from Ancient or Modern South Arabian.8 Our main source for the phonology of Old Arabic comes from Greek transcriptions from the pre‑Islamic period. While the Greek glyphs cannot supply unambiguous evidence for glottalization, they do prove that the Old Arabic series was entirely voiceless, a fact compatible with both glottalization and pharyngealization.9 Thus, the evidence from Syria proves that Proto‑Arabic at least maintained the original voiceless quality of the emphatic series, regardless of whether or not they had become phrayngealized. The following table gives a reconstruction of the Proto‑Arabic emphatics as voiceless, without explicitly taking a position on the quality of the emphatic correlate. Table 2: Proto-Arabic Re-exes of Proto-Semitic Emphatics Proto-Semitic Proto-Arabic *[tθ’] *ṯ ̣ *[t’] *ṭ *[ts’] *ṣ *[tɬ’] *ṣ́ *[k’] *q 2.1 From Proto-Arabic to Sibawayh10 Sibawayh endorses a voiced realization of both *ṭ = [dʕ] and *ṯ ̣ = [ðʕ], as he lists the voiceless counterparts of these phonemes in his compilation of pronunciations which are to be avoided in poetry and Qurʾānic recitation. In Safaitic, the spelling of word boundary *q and *ʾ is sometimes simply q, perhaps suggesting a glottalized realization, /k’/ (AL-JALLAD, 2015a: §3.1.6). 8 See WATSON (2011: 899) for references and a discussion. 9 The dialect of Petra in the 6th c. CE, however, shows that pharyngealization had set in, as it exhibits both voiced re%exes of the emphatics and the lowering of high vowels contiguous with emphatics (AL-JALLAD 2015, §3.7). 7 10 The material from SIBAWAYH is drawn from the online Sibawiki project: http:// sydney.edu.au/arts/research_projects/sibawiki/homepage/ Pre-Hilalian Emphatics 91 For the re%exes of *q, *ṣ́, and *ṣ, the evidence is ambiguous – *ṣ́ had a “weak” counterpart11, but it is hard understand this as having to do with voicing features, and *ṣ has two sub-standard pronunciations, one like the z and the other like the s. I have suggested elsewhere that this points towards an a+ricated realization of the sound Sibawayh endorses, a pronunciation supported by Greek transcriptions from the *rst Islamic century (AL‑JALLAD 2014). The conventional pronunciation of Classical Arabic, on the other hand, does not seem to re%ect the system described by Sibawayh or Old Arabic. Instead, the closest counterpart to this pronunciation is found in the dialects of Anatolia, where only one change was experienced – the merger of *ṣ́ with *ṯ ̣ to [ðʕ], a sound change typical of nearly all modern dialects of Arabic. The Naǧdī dialects also exhibit a similar con*guration, but with a voiced realization of *q. Table 3: Re-exes of the emphatics in Anatolian and Naǧdī Proto-Semitic CAr conventional pronunciation) Anatolian Naǧdī *[tθ’] [ðʕ] [ðʕ] [ðʕ] *[t’] [tʕ] [tʕ] [tʕ] *[ts’] [sʕ] [sʕ] [sʕ] *[tɬ’] [ɮʕ] [ðʕ] [ðʕ] *[k’] [q] [q] [g] Urban dialects of the Levant and most of North Africa can be derived from an antecedent form similar to the Anatolian con*guration through two sound changes: the shift of interdentals to stops and, in some cases, the shift q > ʔ. Table 4: Re-exes of the Emphatics in Urban Dialects Proto-form Cairo Northern Syria Damascus [ðʕ] [dʕ] [dʕ] [dʕ] [tʕ] [tʕ] [tʕ] [tʕ] [sʕ] [sʕ] [sʕ] [sʕ] [ðʕ] [dʕ] [dʕ] [dʕ] [q] [ʔ] [q] [ʔ] 11 SIBAWAYH calls this ‫ﺍﻟﻀﺎﺩ ﺍﻟﻀﻌﻴﻔﺔ‬. 92 A. Al-Jallad 3 Pre-Hilalian Maghrebian The foregone discussion gives the impression that the voiceless con*guration of the emphatics in Proto‑Arabic and the Old Arabic from the Levant does not continue into the modern dialects. Instead, the modern colloquials descend from a mixed series as found in Anatolian Arabic, upon which the conventional pronunciation of CAr may have been based. There is, however, one group of dialects that may constitute an exception – the pre‑Hilalian Maghrebian dialects. As is well known, ḍ, the re%ex of *ṣ́ and *ṯ,̣ and ṭ have merged in a number of Maghrebian Arabic varieties, including Jijel (Algeria), Tangier, Tetuan, Branes, Mtioua (Morocco) (KOSSMANN 2013: 186–7; HEATH 2002: 159). The origin of the ṭ pronunciation is unclear. HEATH assumes a ḍ > ṭ shift and suggests one could posit that the ṭ-pronunciations were once regular across pre‑Hilalian Maghrebian Arabic, but *nds such a scenario ultimately unlikely (2002: 161). KOSSMANN notes that the Berber languages adjacent to these dialects also have a ṭ, but emphasizes that the directionality of in%uence is unclear (2013: 187). The emphatic series of what I will conventionally term modern pre‑Hilalian appears as follows: Table 5: Emphatics in Modern pre-Hilalian Arabic Proto-Arabic Modern pre-Hilalian *ṯ ̣ ṭ [tʕ] *ṭ ṭ [tʕ] *ṣ ṣ [sʕ] *ṣ́ ṭ [tʕ] *q q [q] The traditional view holds that the [tʕ] realization of *ṣ́ and *ṯ ̣ is the result of the devoicing of ḍ [dʕ], but this scenario is based solely on the idea that the voiced pronunciation was original. Moreover, the sound change does not a+ect all instances of ḍ (<*ṣ́ and *ṯ)̣ in any variety, suggesting instead that it is an older feature, which began to give way to the spread of the voiced ḍ of the urban koinés. Of course, one could assume an old ḍ shifted to ṭ and then back to ḍ through lexical di+usion, but such a scenario is needlessly complex. I would instead like to propose a di+erent interpretation, namely, that the modern pre‑Hilalian set continues the voiceless Old Arabic con*guration. An extra stage, however, is required to bridge the gap between the two sets – the loss of interdentals. The scenario needed to ex- Pre-Hilalian Emphatics 93 plain their relationship goes as follows: the forebear of modern pre‑Hilalian, like other dialects, merged *ṯ ̣ and *ṣ́ to the value of *ṯ,̣ which was a pharyngealized voiceless interdental, [θʕ]. Let us call this stage “Old Maghrebian”. Following this merger, the dialect experienced the loss of interdentals, like all pre‑Hilalian dialects. Thus, [θʕ] became [tʕ], along with the shift of [ð] to [d] and [θ] to [t]. This step gives us our contemporary con*guration. Table 6: From Proto-Semitic to Modern pre-Hilalian Proto-Semitic Old Arabic Old pre-Hilalian ʕ Modern pre-Hilalian *[tθ’] *ṯ ̣ *ṯ ̣ [θ ] ṭ [tʕ] *[t’] *ṭ *ṭ [tʕ] ṭ [tʕ] *[ts’] *ṣ *ṣ [sʕ] ṣ [sʕ] *[tɬ’] *ṣ́ *ṯ ̣ [θʕ] ṭ [tʕ] *[k’] *q *q [q] q [q] Evidence for a voiceless re%ex of *ṯ ̣ and *ṣ́ is found in Andalusian Arabic, in, for example, hate < *ḥaẓẓ and almateque < *al-maḍīq (CORRIENTE 1977: 47). The voiced realization of these phonemes is attested as well in both transcriptions and loanwords, e.g. cayált ‘I spent the summer’ < *qāyaẓtu and alcalde ‘judge’ <*al-qāḍī. While these examples have also been used to argue for a lateral realization of the re%ex of *ṣ́ and *ṯ ̣ (CORRIENTE 1989: 98), they do not provide unambiguous evidence for it. The word alcalde contains an etymological lateral, but cayált does not, and so one would have to assume that a merger of ẓ and ḍ took place to a lateral, the opposite direction of this merger in other dialects. While this is not impossible – and there is evidence in Old Arabic12 for such a shift – another explanation is forthcoming. The l in transcription may be an attempt to represent velarization or pharyngealization of the preceding vowel, thus alcalde = [alqa:ʕðʕi:].13 In this case, the spelling cayált provides evidence for a pharyngealized or verlarized realization of the re%ex of *ṯ,̣ but cannot inform us regarding its voicing con*guration, as Iberian 12 See AL-JALLAD (2015b, §3.7.4). 13 While the Castillian l is not velarized, there is little reason to assume that these words were brought in through Castillian. Catalan, for example, has a velarized l in coda position in most dialects, and some dialects have an unconditioned velarized realization of the lateral (RACESENS – ESPINOSA 2005: 20). 94 A. Al-Jallad Romance had only a voiced lateral, [l]. Nevertheless, from the evidence at hand, we can conclude that both voiced and voiceless re%exes of the emphatic lateral and interdental co-existed in Andalusia. This hypothesis, if correct, would mean that a phonologically conservative dialect similar to the Old Arabic of the Levant was implanted in the Maghreb at an early stage. I say similar because it is impossible to know if this dialect was indeed a re%ex of the Old Arabic of southern Syria or of an unattested Arabian dialect with an identical emphatic repertoire. In support of the former theory, one may point towards another peculiar feature shared with the pre‑Islamic Levantine dialects – the relative pronoun di, ddi, d (HEATH 2002: 494–495). Northern Old Arabic preserved the original shape of the relative pronoun ḏ-, which could have continued to in%ect for case or have become frozen as either ḏū or ḏī. In one case it is preceded by the article/demonstrative pre*x h-, hḏ */haḏḏV/.14 Old Ḥigāzī Arabic, on the other hand, seems to have innovated a new relative pronoun based on the demonstrative series ʾallaḏī. Could it be that the Anatolian emphatic con*guration re%ects an Old Ḥigāzī development? It is possible, but we have so far no evidence for the pronunciation of these phonemes in their original context. Aside from a short inscription in the Dadanitic script, JSLih 384, our only evidence for Old Ḥigāzī is the Qurʾānic Consonantal Text (QCT), which, on its own, tells us nothing about the phonetic realization of its consonants independently of the reading traditions imposed upon it. On the other side of the Arabophone world, in what is today Yemen, a voiceless re%ex of *ṯ ̣ is also encountered, but alongside a voiced realization of *q, [g]. Perhaps equally as important, we also *nd re%exes of the relative pronoun *ḏV in this peripheral area as well. Such dialects may also re%ect a development from Old Arabic directly rather than from the intermediate stage of Old Ḥigāzī, and may have been implanted in Yemen during the period of the conquests, if not earlier. Since both the Maghrebian and Yemeni dialects are characterized by archaisms, there is no a priori reason to assume a direct connection between them in order to explain these features. 4 Conclusion The conclusion of this paper is that the voiceless re%exes of the emphatic series, especially *ṣ́ and *ṯ,̣ continue the Proto‑Arabic situation, as re%ected in the Old Arabic material from Syria. The mixed voiced-voiceless series characteristic of most contemporary dialects of Arabic and the conventional pronunciation of Classical Arabic then re%ects a departure from this situa14 See AL-JALLAD (2015a: 88). Pre-Hilalian Emphatics 95 tion, and may be characteristic of Old Ḥigāzī, but this can only be a suggestion. Its ubiquity in the Arabic dialects of today may re%ect a convergence towards this dialect, which would have acquired considerable prestige in the context of Arabo-Islamic civilization. Siglum JSLih Dadanitic inscriptions in Jaussen and Savignac 1909–1922 References AL-JALLAD, A. (2014): Aṣ-ṣādu llatī kas-sīn – Evidence for an a+ricated ṣād in Sibawayh? In: Folia Orientalia 51: 51–57. AL-JALLAD, A. (2015a): An Outline of the Grammar of the Safaitic Inscriptions. Leiden & Boston: Brill. AL-JALLAD, A. (2015b): Graeco-Arabic I: the southern Levant. In: Nehmé, L. et al (eds.): Ecrit et écriture dans la formation des identités en monde syriaque et arabe IIIe-VIIe siècles. Leuven: Peeters. CORRIENTE, F. (1977): A Grammatical Sketch of the Spanish Arabic Dialect Bundle. Madrid: Instituto Hispano-Arabe de Cultura. CORRIENTE, F. (1989): South Arabian Features in Andalusi Arabic. In: WEXLER, P. – BORG, A. – SOMEKH, S. (eds.): Studia linguistica et orientalia memoriae Haim Blanc dedicata. Wiesbaden, 94–103. HEATH, J. (2002): Jewish and Muslim Dialects of Moroccan Arabic. LondonNew York: Routledge. HUEHNERGARD, J. (forthcoming): Arabic in its Semitic Context. In: AL‑JALLAD, A. (ed.): Arabic in Context. Leiden: Brill. JAUSSEN, A. – SAVIGNAC, M. (1909–1922): Mission archéologique en Arabie (6 vols.). Paris: Leroux/Ernest Paul Geuthner. Reprint 1997, Cairo: Institut Français d’Archéologie Orientale. KOGAN, L. (2011): Proto-Semitic Phonology and Phonetics. In: WENINGER, S. et al. (eds.): The Semitic Languages: An International Handbook. Berlin/ Boston: de Gruyter, 54–150. KOSSMANN, M. (2013): The Arabic InEuence on Northern Berber. Leiden: Brill. RECASENS, D. – ESPINOSA, A. (2005): Articulatory, positional and coarticulatory characteristics for clear /l/ and dark /l/: evidence from two Catalan dialects. In: Journal of the International Phonetic Association 35 (1): 1, 20. WATSON, J. (2011): Arabic Dialects (general article). In: WENINGER, S. et al. (eds.): The Semitic Languages: An International Handbook. BerlinBoston: de Gruyter, 851–896.