Demonstratives in Musandam Arabic: Distinctive Archaisms and Innovations

(2022) 675-702 Abstract This study introduces and analyzes proximal and distal singular demonstratives in fourteen varieties of Musandam Arabic, a little-documented dialect group located on Musandam Peninsula in northern Oman and neighbouring areas of the United Arab Emirates. Following an overview of the dialect group in its regional context, the study provides a description of singular demonstratives from the point of view of phonology, morphology, and geographical distribution. The study then focuses on two salient features found in several of the varieties under investigation: gender distinction based on consonantal alternation ( d -based masculine forms vs t -based feminine forms); and gemination of the feminine t -element. While the former is attested, albeit rarely, in other Arabic dialects, the latter is unheard of. In the last section of the article, some hypotheses are put forward as to how these forms could have developed from a historical point of view, in light of data from different Arabic and Semitic varieties. While the gemination of the t -element is best regarded as a Musandam-internal innovation, the d : t consonantal alternation reinforces the putative historical link between south-western Arabia and Oman.

* The contribution of each author is as follows: Erik Anonby -data collection and archiving, linguistic analysis, article drafting and revision, article maps; Simone Bettega -linguistic analysis, article drafting and revision; Stephan Procházka -linguistic analysis, article drafting and revision. We wish to express our thanks to those who made this research possible: first, to the speakers of Musandam Arabic consulted in this study, who generously shared with us oral texts and elicited language samples (see Appendix for a listing of consultants); to the local authorities who welcomed us to carry out research on the language and introduced us to potential consultants; to the editors and anonymous referees, who provided meticulous critique and commentary on the initial draft of this study; and to other colleagues who assisted in significant ways, in particular Janet Watson and colleagues at the Centre for Endangered Languages, Cultures and Ecosystems (CELCE) at the University of Leeds, Esther Ravier at École Normale Supérieure/LATTICE, Paris, and Wolfgang Zimmermann, who enabled Erik Anonby to revisit Musandam Peninsula, sharing his historical, geographic and cultural insights about the region in light of work and experiences spanning many decades.
Note on transcriptions: in keeping with common practice in linguistic description, in this article we transcribe the glottal stop and the voiced pharyngeal fricative using the respective International Phonetic Alphabet (IPA) symbols ʔ and ʕ, rather than the conventional Arabica symbols ʾ and ʿ.

Introduction
In this paper, we present proximal and distal singular demonstratives collected from the little-documented Arabic varieties that are spoken on the Musandam Peninsula, and examine unexpected features in their morphological and phonological structures. The mountainous Musandam Peninsula extends northwards from the eastern corner of Arabia into the Strait of Hormuz (Map 1). The Sultanate of Oman now governs most of the region, and this administrative exclave is surrounded by the United Arab Emirates and the sea. Arabic is the main language of most of its inhabitants, but in several towns, speakers of the endangered Kumzari language are also found. The Arabic varieties spoken on the Musandam Peninsula are often referred to as Shihhi (šiḥḥī) Arabic, but alongside the Shihuh, the Dhohuri and Hadheri (urban) people are two other important Arabic-speaking groups.1 For this reason, we group the regional Arabic varieties here under the geographic label of "Musandam Arabic." Although initial publications point to a number of typologically distinctive features, Musandam Arabic has yet to be systematically documented.
The present study is organized as follows: in Section 2, we outline the state of research on Musandam Arabic, and present proximal and distal singular demonstratives in 14 varieties of this dialect group. In Section 3, we discuss structural characteristics of the demonstratives, identifying the most salient features and delineating their geographic distribution. While known social factors are introduced to help explain some of the broader geographic patterns, as well as exceptions to these patterns, an integrated sociolinguistic analysis of variation as it correlates to demographics of the region is beyond the scope of this study. Section 4 sets the data in comparative and historical perspective, with discussion of whether these features represent innovations internal to Musandam Arabic, or have originated from elsewhere on the Arabian Peninsula. The results of our analysis are summarized and discussed in Section 5. An Appendix documents details of field research and linguistic data sources, and Arabic orthography of toponyms mentioned in the article is also provided there.
phonological and morphological sketch of the variety spoken in the village of al-Jadi.3 Syntactic structures, as well as internal dialectal diversity, are essentially unexplored.
This paper forms part of a larger programme of research on the linguistic geography of the Musandam Peninsula. While earlier publications focused on the endangered Kumzari language,4 most of the fieldwork on the peninsula's Arabic dialects was carried out only recently, in 2018. The data presented in this paper are taken from a standardized wordlist5 elicited by Erik Anonby in twelve Musandam Arabic varieties across the peninsula. The Appendix inventories the field research locations and introduces the speakers we consulted, along with personal demographic details. All sound files were recorded in .wav format using an Olympus LS-14 sound recorder. We also include here data from Bernabela's (2010) field notes from Ghubb and al-Jadi and from his (2011) phonological and morphological sketch of the latter variety (see Appendix). The general research process as well as wider findings of the project, in particular those related to linguistic geography, are detailed in a concurrent study.6 This study focuses on one of the more surprising findings of the research, namely unexpected and otherwise rarely or not at all attested features of demonstratives in Musandam Arabic.

Demonstratives in Musandam Arabic: Structural Features and Geographic Patterning
In Table 1 we present the masculine and feminine forms of both the distal and proximal singular demonstratives from fourteen locations on the Musandam Peninsula. As shown in the Appendix, data from twelve locations are from our own field research,7 and two (marked "[Bern.]") are from Bernabela.8 Recordings of all data from our own field research are available at: https://borealisdata.ca/dataverse/arabia.9 Table 1 reveals three remarkable traits which are peculiar to Musandam Arabic demonstratives. First, in contrast to the vast majority of Arabic dialects, including adjacent dialects of the Gulf and Oman (cf. Table 2 below), in which gender opposition in demonstratives is based on the value of the final vowel (typically m. a/ā vs f. i/ī), these vocalic contrasts are in many cases absent in Musandam Arabic demonstratives. Secondly, gender contrast is often marked instead by the alternation of a voiced (m.) versus voiceless (f.) second consonant (d : t). Thirdly, and related to this, in most of the varieties that show this type of consonantal alternation, the characteristic voiceless t of the feminine singular forms is geminated.
The distribution of these typical Musandam Arabic traits within the peninsula itself shows geographically significant patterns. Social factors are also operational in the distribution of structural traits. Although we do not attempt to provide a systematic variationist analysis of the data in the present article, we introduce relevant known social factors whenever they can help shed Given the other forms elicited in this location, we expect the form hōdōk here. However, the initial vowel recorded here sounds like a high vowel ū. This item and the corresponding items from KHW immediately below thus show instability between ō and ū in the demonstratives, even though these two vowels clearly contrast elsewhere in the language.
light on geographic patterning of particular linguistic structures.10 The data from Table 1, "Demonstratives in 14 Musandam Arabic varieties," are localized in Map 2, "Geographic distribution of gender opposition features" and Map 3, "Geographic variation in C 2 in distal feminine demonstratives." We have selected distal feminine demonstratives for the latter map, since they more frequently show retention of regionally distinctive features than their proximal counterparts. Further aspects of the geographic distribution of these gender opposition-related features are discussed following the maps. Map 2 reveals that Musandam Arabic varieties which differentiate gender in the same way as the adjacent Arabic dialects of the Gulf and Oman -i.e., by vowel contrast only -are located in the central neighbourhoods of the administrative capital Khasab (KHC1/2), in Daba (DBA), and in Jadi (JAD). Essential as points of social orientation, Khasab and Daba (also commonly romanized as Dibba) are the two largest cities of the peninsula, and between them account for a majority of its population. Khasab, as the administrative capital of Oman's Musandam Governorate, has just under 20,000 inhabitants.  central Khasab, Hadheri (urban) people inhabit most of the neighbourhoods. Although their presence in the town is longstanding, generally dating back many decades or even centuries, many of these people can trace their ultimate origins to regions beyond the peninsula. As for Daba, at about 60,000 inhabitants, this latter town is divided by an international border, with about 10,000 people living on the north side in Oman, and the remainder in the United Arab Emirates to the south.12 Culturally, Hadheri people in Daba fall into the linguistic and cultural area of the wider Gulf rather than Musandam Peninsula, but their presence here has always, within memory, been part of the town's constitution alongside its Shihuh and Kumzari inhabitants. And in both cities, the trade-centred livelihoods of Hadheri people have enabled intensive contact with speakers of Arabic dialects from beyond Musandam, whether through traders visiting the peninsula, or through their own travel to other regions. Jadi, on the north-western coast of Musandam, is geographically closer to Ra ʾs al-Khaymah and other major cities of the Emirates' north coast. These factors explain well the occurrence of forms in these locations that are common in Arabia outside Musandam (see Table 2).13 This pattern of accommodation to common Arabic structures diffuses outwards from the core urban locations. In the outer districts of Khasab and the nearby town of Qada we observe the typically Musandam Arabic d : tt contrast (see Map 3) in all forms, but in the distal feminine forms the gender distinction is reinforced by the vowel ī. Such hybrid forms can be seen as a first sign of adaptation to the forms used in the nearby city.14 In Haffa, for which Daba is the nearest large settlement and a place of residence in the summer, the proximal demonstratives have largely aligned to the common 12 Sources: https://data.gov.om/OMPOP2016/population; http://citypopulation.de/en/oman /admin; http://citypopulation.de/en/uae/cities, consulted on 9th September 2022. 13 This is not a completely recent phenomenon; more than a hundred years ago, Jayakar, "The Shahee Dialect of Arabic," p. 252-253, documented forms which are typical of the wider Gulf region (repeated from footnote 10 -proximal: Given that similar mixed forms, exhibiting both consonantal and vowel contrasts, are attested in pre-Classical Arabic (see Table 5 Arabic forms outside of Musandam. The distal feminine form, however, is of the same hybrid character as found in Qada. Another type of mixed system is attested in the speech of Shesa, where only the proximal forms exhibit the common Arabic patterns. A further point of background is that the speaker representing Shesa, like most of the people originating in Musandam's villages, now lives in the city -in her case, Daba -and only visits her home village seasonally.15 In all three of these cases of mixed systems, contact with a large city inside Musandam is a determining factor in the structural changes that are taking place. A further key point, also common to all three locations, is that the proximal demonstratives, likely also filling a spatially neutral demonstrative role (as is not uncommon in other languages)16 and ostensibly appearing with the highest frequency in discourse, succumb first to the structural adaptations.
In the remaining villages, generally situated farther away from the cities and in less contact with common Arabic forms, the typical Musandam Arabic forms predominate. Thus, geographical and social distance from centres of diffusion is the main criterion for the retention of structurally distinctive features in the Arabic dialects of Musandam.

Distinctive Gender-Marking Traits in Musandam Arabic Demonstratives: Comparative and Historical Perspective
Here, we carry out a comparative and historical examination of the two areally distinctive structural traits introduced above which are used to distinguish gender in demonstratives in Musandam Arabic: consonant alternation (4.1), and gemination of the alternating consonant (4.2).

Consonant Alternation
If we focus on the first of these two traits -consonant alternation, or more precisely, voicing alternation in consonants -we see that across the entire Arabic-speaking world it is only shared by a handful of dialects today, mostly 15 Further, the speaker is (unique among the consultants) female, educated, and works in the tourist industry. Beyond geography, these are all factors that could facilitate contactinduced changes in her speech. 16 Stephen C. Levinson spoken in a relatively small area that encompasses western Yemen and south-western Saudi Arabia.17 It may well be that this feature also exists in the main bloc of Oman, but has not yet been detected because many Omani Arabic varieties, particularly those spoken in mountainous regions, are still undescribed.
In south-western Arabia, the southernmost documented attestation of a ḏ : t consonantal alternation in the proximal forms is in al-Ḥusayniyah, near Zabid in the southern Yemeni Tihamah, while the northernmost location appears to be Bilād Ġāmid in Saudi Arabia (see Map 065 in Behnstedt's atlas of North Yemen18 for a complete mapping of the phenomenon). Roughly speaking, all these dialects are spoken in the coastal plain of western Arabia and the adjacent mountain range, with the only exception being the Yemeni capital, Sanaa, which lies further inland.
Regarding the distal forms, this phenomenon is roughly found in the same area of south-western Arabia, but somewhat more widespread, extending to some regions farther east: specifically, in the southern Yemeni region of Dathina; in the city of Najran (Saudi Arabia); and to the north, in the southern Najd.19 In the Najd, it is found in Wādī Dawāsir, which explains its existence among the Dōsiri tribe of Kuwait whose members moved there from the Wādī Dawāsir.20 In the areas of south-western Arabia that we have delimited here, a large variety of forms exhibiting different patterns is found. Examples of some demonstratives which most closely resemble those of Musandam Arabic are given in Table 3. It is striking that -in contrast to Musandam Arabic, and with it, many varieties across the Arabic-speaking world -these and all other forms attested in south-western Arabia lack the demonstrative element hā-. 17 In these dialects, the actual opposition features a contrast between a voiced dental fricative ḏ and a voiceless alveolar occlusive t. In Musandam Arabic all dental fricatives have become alveolar occlusives, so the contrast is now a simple opposition based on voicing.  Brill, 1920Brill, -1942Brill, , p. 2842: there, the t-based forms only occur in the distal demonstratives hatāk and hatīk. This has an interesting parallel in Musandam Arabic, where t-based distal demonstratives are also geographically more widespread than their proximal counterparts (see Table 1). 20 See also Alexander Magidow, "Diachronic Dialect Classification with Demonstratives," Al-ʿArabiyya, 49 (2016), p. 91-115.
Attestations of t-based feminine demonstratives already occur in early varieties of Arabic. A clear example is the form ty /tī/ that is found in the famous Namara inscription (dated 328 CE),21 but they occasionally also appear in Safaitic inscriptions.22 However, demonstrative pronouns that use t as a feminine marker are not restricted to peripheral dialects or ancient layers of Arabic, but also feature in Classical and Modern Standard Arabic (see Table 4).
In fact, a much larger variety of demonstratives which differentiate gender by the opposition ḏ : t is attested in pre-Classical Arabic and summarized in   tribes.24 Chaim Rabin, for instance, maintains that tā, which the grammarians connect with the tribe of Ṭayyiʾ, was also the original proximal feminine form in west Arabia and only later replaced by hāḏih.25 However, as is often the case with the information provided by the ancient grammarians, the details are contradictory and the same form is sometimes attributed to different tribes. Although many of the forms may have originally been limited to the speech of a certain tribe, some became more widespread later. Hans Kofler astutely points out that the mere fact of these variants being extensively described in the works of the grammarians and lexicographers indicates their common supra-regional usage.26 As can be seen in Table 5, some of these forms combine two strategies for marking gender opposition, namely, vowel alternation is found alongside consonantal alternation. This is rare in modern dialects, but not unheard of: for proximal demonstratives, it is attested in Ġurāz in north-eastern Yemen  (ḏā : tih); and for distal demonstratives, it is found in Abhā and Bani Mālik in south-western Saudi Arabia (ḏāk : tīk) and as-Suwwādīyeh in south-eastern Yemen (hāḏāk : hātēk).27 As has been shown in Table 1, this mixture of strategies occurs -only with distal demonstratives -in four of the Musandam Arabic varieties as well (QAD, KHW, KHE, HAF).28 A first question that presents itself when returning to consonant alternation in the Musandam Arabic data is: why do we find such a characteristically "south-western" Arabian feature on the eastern edge of the Arabian Peninsula? Scholars familiar with the peculiarities of eastern Arabian dialects are aware that originally south-western traits are not uncommon in the area. Perhaps the best-documented case is the 2nd person feminine singular suffix pronoun -iš, found in almost all Omani dialects and in certain Arabic-speaking communities of Bahrain, the United Arab Emirates, eastern Saudi Arabia and even Iran.29 Clive Holes, who has investigated this phenomenon in depth, explains: At some point […], just before and for many centuries after the coming of Islam, the N. Yemeni Arabic of the period, with its morphologised substrate 2nd fem. sing. -(i)š began to be "exported" to other parts of the peninsula. Yemeni migrations have tended to be in two directions: north-east into central Arabia, and thence east to ancient Bahrain and south into the northern Oman mountains via Tuwam and directly east along the coast into Hadramawt, thence Dhofar and eventually northern Oman. This would explain the at first sight puzzling fact that the -(i)š 2nd person 27 Behnstedt, Dialect Atlas of North Yemen and Adjacent Areas,[152][153][154] The presence of vowel alternation in distal demonstratives in Musandam Arabic alongside consonant alternations (QAD, KHW, KHE m. hōdōk, f. hōttīk; HAF m. hādāk, f. hātīk; see Table 1) could be attributable to renewal of an old pattern of vowel alternation through influence from nearby urban dialects, or wider regional dialects such as Gulf Arabic. However, we feel that such patterns would be more susceptible (in cases of difference between proximal and distal forms) to diffuse first into the proximal forms, which likely have a higher frequency in all varieties. This "proximal-first" tendency has ostensibly transpired in HAF -where full assimilation to the urban system has taken place for proximal demonstratives -but not in the other locations. Another feature commonly found in eastern Arabia that might ultimately find its origin in the south-western Peninsula is the post-participial -in(n)-infix31 (interestingly, our initial fieldwork in the region confirms that both these traits are also found in Musandam).32 One can also add here some lexical peculiarities shared by Yemeni and Omani dialects, for example, fuḫāḫ(i) 'chameleon' and šall/ištall 'to go, set off.'33 In light of the above, it may well be the case that consonant-alternating demonstratives represent another of these originally south-western Arabian traits that has been exported to the coasts of the Gulf. It needs to be remarked, however, that -apart from Musandam Arabic -there is not a single documented dialect in eastern Arabia that makes use of consonantal alternation to distinguish gender in singular demonstratives. This fact, while highlighting Musandam Arabic demonstratives as all the more worthy of attention, begs the question of why we find the -iš suffix and -in(n)-infix spread across a vast area that spans from Oman to al-Hasa and Bahrain, whereas consonant-alternating demonstratives appear to be confined to the tiny peninsula of Musandam.
One possible explanation is that Musandam was the only region historically reached by people of south-western ancestry who spoke a dialect with consonant-alternating demonstrative. This is, however, ad hoc and ultimately unconvincing as it contradicts the historical Arabic sources which agree in stating that large parts of the population of today's Oman originated from Yemen. A more likely scenario is that this feature was originally more widespread across eastern Arabia, but had its area of distribution reduced due to contact with vowel-alternating varieties. The territory of Musandam remains to this day largely inaccessible, consisting mostly of high, rugged limestone mountains surrounded by the sea on three sides and separated from the wider expanse of Arabia by dunes and lower mountains. As a consequence, the dialects of the area have always enjoyed a remarkable degree of linguistic isolation, and their conservative nature should come as no surprise. As already mentioned above, the few Musandam Arabic varieties that do not show the consonant-alternating demonstratives are those that have been subject to greater external influence across history.
A third possible explanation for similarities between the Musandam Arabic and western Arabian forms is, of course, parallel independent innovation. After all, while it is unlikely that morphological innovations as specific as the -iš suffix pronoun and the intrusive -in(n)-infix emerged more than once as the byproducts of random developments,34 feminine-marking t elements are solidly attested with verbs, nouns and adjectives in essentially all varieties of Arabic,35 so it is possible that several separate dialects extended their use to the realm of demonstratives.
As a matter of fact, if we accept Rebecca Hasselbach's reconstruction of the history of demonstrative elements across Semitic languages,36 we see that 34 As noted by an anonymous reviewer, the shift from k to š in palatal environments is a common process cross-linguistically. While accepting this, it should be pointed out that this process is not, in fact, common in Arabic: to our knowledge, this full shift is only attested in a handful of dialects scattered across a relatively limited area (mostly coastal south-eastern Arabia), and it is confined to a single morphological element, namely the 2nd person feminine singular suffix pronoun. Holes, in "Kashkasha and the Fronting and Affrication of the Velar Stops Revisited," suggests that a widespread k > š shift occurred in certain South Arabian languages more than a millennium ago, and that, when Arabic eventually swamped the area, the -iš suffix pronoun was incorporated into its system due to substratal influence. A fitting parallel can be observed in the dialect of Deir ez-Zor in Syria, which is a sedentary qəltu-dialect (in which OA *k=k) that, because of contact with Bedouin speakers, adopted an affricated form for the 2nd person feminine singular suffix pronoun (-či, e.g., ḏ ̣ arab-či 'he hit you [f.]'). This form is actually a hybrid between typical qəltu non-affricated -ki and Šāwi gender distinction based on the presence or absence of a final t is a feature reconstructed for Proto-Semitic itself. However, the -t suffix does not appear in Hasselbach's reconstruction of the Proto-West Semitic (PWS) forms, which had gender distinctions in demonstratives based on vowel quality.37 One must therefore infer that this element was lost in a pre-Proto-West Semitic phase, and that a subgroup of several languages of the West Semitic group spontaneously reintroduced it at a later stage. In particular, all the Semitic languages in which a t-marked feminine demonstrative is attested belong to the Central Semitic group: Hebrew, Old Aramaic, Ancient South Arabian and, of course, Arabic.38 The latter was actually characterized by a unique development: while in the other "innovative" languages the feminine singular demonstrative is marked by a -t suffix (compare Ancient South Arabian ḏt, Hebrew zō(Ɂ)t, Old Aramaic zɁt),39 in certain Arabic forms the t itself constitutes the base of the demonstrative (see Table 5 above).40 This peculiarity has not gone unnoticed among scholars of Semitic linguistics, and it has been pointed out by, among others, John Huehnergard ("The feminine singular demonstrative element, t-, as in tilka, hātā, (ʔ)allatī, is found almost nowhere else in Semitic")41 and Ahmad Al-Jallad (who notes the forms "Classical Arabic tā, hātā, ʔallatī, and Old Arabic ty /tī/").42 As we have seen, contemporary Arabic dialects employ one of two strategies to mark gender on demonstratives: a (common) contrast based on vowel opposition, or a (rarer) contrast based on consonantal alternation in the demonstrative element itself. On the development of these two systems, Magidow writes: On a very speculative level, it is possible that an ancestor of modern Arabic might have generalized a single demonstrative to mark both masculine and feminine singular, a change that has subsequently occurred in many 37 Ibid., p. 22. Note that this opposition based on vowel quality is of a different nature from the one that we find today in most Arabic dialects. 38 Ibid., p. 3. 39 Ibid., p. 8, 12, 13. 40 Ibid., p. 9. Hasselbach notes how the proximal forms with initial t are more rarely attested than the widespread Classical hāḏihi. She also points out that a ḏāt feminine element existed in Classical Arabic, "where it does not function as a relative pronoun but indicates possession" (ibid., p. 18). dialects around the Arabic-speaking world […]. Pressure to re-establish gender marking as in other paradigms (pronouns, verbs) could explain why not one but two strategies arose for marking gender. The vowel alternating dialects were probably produced by paradigmatic pressure from the other paradigms in Arabic that contrast a low vowel /a/ for masculine forms (*ʔanta "you (m.)," faʕalta "you (m.) did") against a high vowel /i/ for feminine forms (*ʔanti "you (f.)," faʕalti "you (f.) did"). The origins of the consonant-alternating dialects are less clear but might be inspired by the Semitic feminine marker *-t/-at. This change only affected the singular demonstratives, although both changes appear to be innovations versus the Proto-West Semitic forms.43 In light of Hasselbach's tentative reconstruction, that we have reported above, it might be the case that Alexander Magidow's hypothesis is too conservative in that the loss of gender distinction in demonstratives happened not in a proto-Arabic phase, but earlier in time (a proto-Central Semitic phase?).
Be that as it may, when the distinction was re-introduced, Arabic developed two different marking strategies, one of which has no equivalent elsewhere in Semitic.44 This, we think, speaks against the hypothesis of parallel independent innovations in south-western Arabian dialects and in Musandam Arabic.
As we have seen, feminine demonstratives of the t-type are already rare from a pan-Semitic perspective, so it is unlikely that these forms developed independently in two separate groups of Arabic dialects. It is more plausible that the innovation occurred only once, at a proto-Arabic stage -so that we would only speak of an innovation when we contrast these forms with the other languages of the Central Semitic group. Though it is hard to pinpoint an exact point of origin, we put forward the idea that t-based feminine demonstratives have always coexisted, within the Arabic family, with the competing forms based on vocalic alternation. Through the course of time, these forms have undoubtedly expanded and receded, moving along the routes of human migrations and across spaces of social interaction between communities. One of these migrations, as we have seen, has probably brought the linguistic precursors of Musandam Arabic from south-western Arabia to the east. The presence of other common traits between the two areas lends further support to this 43 Magidow, "Diachronic Dialect Classification with Demonstratives," p. 98. 44 The t-was most likely another demonstrative element in the Semitic languages which only later has been restricted to feminine forms in Arabic because of the similarity with the nominal and verbal feminine markers. See also Jakob Barth, Die Pronominalbildung in den semitischen Sprachen, Leipzig, Hinrich, 1913, p. 83-85. hypothesis, as does what is more generally known about historical population movements across the Arabian Peninsula.45

4.2
Gemination of t Focusing now on the second unusual trait that characterizes demonstratives in Musandam Arabic -the gemination of t in the feminine singular forms -we find a unicum among dialects from across the Arabic-speaking world: to the best of our knowledge, this feature has not been documented in any other Arabic variety. As with the d : t consonantal alternation, this phenomenon invites substantial probing. At this stage of our research, a couple of different hypotheses can be put forward to account for the emergence of t-gemination in feminine singular demonstratives, though none can be ultimately proven true.
A first possible explanation for t-gemination relies on the idea that Musandam Arabic feminine demonstratives have evolved from an original proximal form, analogous to Classical Arabic hāḏihi, with a second h. Such forms are extremely rare in dialects today, but not entirely unheard of (compare, for instance, the form hāḏihī in the dialect of Banū ʿAbādil, in north-western Yemen).46 Loss of the unstressed short vowel in the middle open syllable would have produced *hāḏhi. From *hāḏhi the emergence of a *hātti form can be postulated via voice assimilation in one direction and place assimilation in the other (either preceded or followed by the across-the-board dental > alveolar shift seen in Musandam Arabic, as mentioned above). Geminated t would then have spread to the distal forms by analogy.47 This hypothesis would need to account for cross-dialectal variability in gemination of t: in three, non-adjacent locations from a total of 10 locations with feminine t, this consonant is not geminated (in proximal or distal forms), which would imply sporadic loss of gemination, perhaps for reasons of articulatory simplification. It would also have to acknowledge assimilation of the second vowel to the value of the first48 45 In line with evidence from the works of Holes cited above, inhabitants of the Musandam Peninsula themselves trace their ancestry to south-western Arabia. Or, of course, there could also have been a prior distal demonstrative form with *dh, but there is no evidence for any distal forms of this type in Arabic. 48 This possible diachronic vowel assimilation process is discussed in footnote 37 above. Synchronic processes documented in Bernabela, A Phonology and Morphology Sketch of the Šiħħi Arabic Dialect of əlǦēdih, Musandam (Oman), reinforce the plausibility of diachronic assimilation of a suffixed h to a preceding voiceless consonant as well as and the presence of a final k, perhaps adopted in imitation of neighbouring varieties. Beyond the historical gymnastics that this explanation depends oneach part possible on its own but requiring the succession of too great a number of steps -its greatest weakness is that it is dependent upon a prior form which is hardly attested in south-western Arabia, and overlooks the existing ḏ : t contrasts which are well-represented there.
A second, and more straightforward, explanation is that Musandam Arabic demonstrative elements with a geminated t have originated in an identical but non-geminated form. As shown in Table 1, such non-geminated forms are presently attested in three of the 10 locations where t is used to mark feminine demonstratives.
According to what was said in the preceding sections, we assume that the Musandam Arabic forms are directly related to those found in south-west Arabia. It is striking that all demonstratives in that region which exhibit consonantal gender contrast also lack the prefix hā-.49 Thus, the Musandam forms are more or less unique in the whole Arabic Sprachraum and therefore it is likely that the hā-prefix is a later innovation resulting from contact with dialects spoken in adjacent regions. Due to the shift of the interdental fricative ḏ to the alveolar plosive d, the original ḏ : t opposition was acoustically weakened as the gender opposition in Musandam Arabic singular demonstratives became solely based on a voicing contrast. Because of the adoption of the hā-prefix, the alveolar consonants additionally found themselves in intervocalic position where voicing of underlyingly voiceless consonants is cross-linguistically frequent. Although we can exclude complete merger of t with the voiced d, we can imagine that speakers felt the need to make the distinction clearer by lengthening the characteristic consonant of the feminine form.
This development may have been reinforced by a tendency, found in other Arabic dialects, to intensify the deictic function of the demonstrative pronoun diachronic vowel assimilation in Musandam Arabic. On page 48, for instance, we read: "The initial h of 3fsg -hi and 3cpl -ham usually assimilates with preceding voiceless stops and fricatives, as in the following examples: […] ʔōdōt + ham → ʔōdōttam 'their customs' […]. When attached to the preposition li, 3cpl -ham may display vowel harmony: līhim 'for them'". Simplification of th clusters to tt is also documented for certain dialects of the UAE (see Hamdi A. Qafisheh, Gulf Arabic English Dictionary, Chicago, NTC, 1997, p. xix). 49 In only two locations, distal demonstratives exhibiting the prefix are attested: hāḏāk: hātēk in as-Suwwādīyeh and hīḏ ̣ āk: hīṭāk in ˁUbāl (see Behnstedt, Dialect Atlas of North Yemen and Adjacent Areas, p. 152-153, numbers 119 and 30 on the maps). The first village is situated within the hā-form region, and the second exactly in the border zone between hā-and zero-marked demonstratives. Therefore, we proceed from the assumption that hā-in these items is not original. by phonological means.50 For example, many Bedouin-type dialects apply velarization to masculine proximal demonstratives, e.g., m. hāḏ ̣ a (but cf. f. hāḏi).51 The surrounding low vowels are certainly conducive to the appearance of emphasis on the masculine form,52 but the intrusion of deictic intensification is likely crucial in the diachronic change. Even more directly applicable to understanding the Musandam Arabic forms, gemination has emerged on demonstratives in other Arabic dialects. Examples for distal pronouns are found in the south-western corner of Yemen (ḏakkuh (m. sg.)/ḏakkeh (f. sg.), ḏukkuh (m. sg.)/ḏikkeh (f. sg.), ḏukkah (m. sg.)/ḏikkah (f. sg.)) as well as in certain Libyan dialects (hādākkāh (m. sg.) and hādīkkāh (f. sg.)).53 The geminated kk is not etymologically explainable in any of the forms. However, as can be seen from the examples, the gemination is not related to a distinction in gender. Alternatively, for a case of gemination used to mark gender contrast on demonstratives, one can look to the proximal plural demonstratives in some central Yemeni dialects: ḏawla (m. pl.)/ḏalla (f. pl.), hawla (m. pl.)/halla (f. pl.).54 It should be added that at least the forms with doubled kk may have emerged from the combination of the pronoun with another demonstrative element *ha(h) and/or independent personal pronouns. 55 The cluster *kh could thus have yielded kk, a phonological development which can be seen as a parallel to the hypothesis presented above, that the geminated tt is the product of a prior *th. The asymmetry in the second consonant of Musandam Arabic demonstratives, both in voicing and gemination (d : tt), thereby ensures an abundantly (if superfluously) clear phonological signalling of gender contrast. As a final piece in this puzzle, a possible explanation for selective gemination -in the feminine but not the masculine form -could reside in the fortis nature of voiceless consonants, which makes them more susceptible to gemination.56 Conclusion In this paper, we have investigated the structural properties and wider historical significance of proximal and distal singular demonstratives in Musandam Arabic. In doing so, we have endeavoured to further nudge open the window into one of the significant remaining blind spots in the dialectology and history of Arabic.
Analysis of data collected from 14 locations across Musandam Peninsula reveals a set of structurally distinctive demonstratives. Although the distribution of forms varies across the peninsula, two remarkable areal features recur: 1) Demonstratives in most Musandam Arabic varieties distinguish gender by means of a voiced/voiceless (d : t) consonantal alternation, sometimes accompanying and sometimes supplanting the vocalic oppositions commonly used elsewhere in Arabic; 2) In half of the dialects surveyed, the voiceless t of the feminine demonstratives is geminated. While the first feature occurs elsewhere rarely, in a swath of Arabic dialects in south-western Arabia and in pre-Classical Arabic, the second feature appears to be unique to the Musandam dialect bundle. Based on what is known about population movements in the Arabian Peninsula over the last two millennia, and on the structural similarities that exist between varieties of Arabic in south-western and eastern Arabia, it is reasonable to conclude that feature (1) was originally a south-western trait that was later exported to the Musandam Peninsula as a result of human migrations across southern Arabia. (It is possible that this first trait was once more widespread across eastern Arabia, and now only survives in the dialects of Musandam.) Feature (2), however, appears to be a unique innovation, and as such, one that probably has its origins within Musandam itself. While the factors that led to the appearance of a geminated consonant in Musandam Arabic feminine demonstratives are still open to investigation, we see reinforcement of the phonological opposition between masculine and feminine elements as the most likely explanation. As Musandam Arabic is the only hitherto described Arabic dialect exhibiting t-based feminine demonstratives along with the prefixed formative hā-, we can imagine that the d : t contrast in intervocalic position was perceived as too weak for the gender contrast and therefore triggered the secondary gemination of the t. Moreover, the reinforcement of the deictic function of demonstratives via phonological means has been witnessed in other Arabic dialects and, although it does not involve gender distinction there, it nonetheless lends further plausibility to the scenario postulated in this study.