Article

1. Introduction(edited)
2. Nabataean Aramaic, originating in the kingdom once ruled from Petra, closely resembles Imperial Aramaic. The latter, spread far and wide by the Achaemenid Empire, is now well known to us from Persian-period corpora like the Elephantine Papyri (cf. Folmer, 1995, 2022; Gzella, 2015, pp. 157–211; Muraoka & Porten, 2003). But Nabataean Aramaic does show some differences from Imperial Aramaic, most strikingly a fair number of grammatical features and loanwords taken from Arabic (see the discussion below). This matches both the frequent occurrence of Arabic-derived names in the Nabataean onomasticon and ancient sources which refer to the Nabataeans as Ἄραβες ‘Arabs’. The general consensus since Cantineau (1930–1932, 1934–1935) has therefore been that ‘the Nabataeans’ may have written Aramaic, but that they spoke Arabic (see the works cited in Butts, 2018, p. 40n7). Less attention has been paid to the ways in which Nabataean Aramaic diverges from Imperial Aramaic that cannot be due to Arabic influence. The most broadly accepted scenario, whereby the Nabataeans used Aramaic as a purely written language and modified it by introducing features from their spoken Arabic, cannot account for these. What can these non-Arabic innovations in Nabataean tell us about the linguistic status of Aramaic in the Nabataean kingdom and ancient North Arabia? And how does this affect our understanding of Pre-Islamic Arabic?
3. But first, we must define our terms. The meaning of Aramaic is uncontroversial: (any variety belonging to) a certain family of closely related Semitic languages/dialects that descend from a shared ancestor presumably spoken in late second-millennium bce Syria, with many shared features and a long written history (e.g., Healey, 2013, pp. 23–24; Huehnergard, 1995; Gzella, 2015). Some of the other terms that are crucial to this investigation have been used in different or vague ways and require clarification.
4. Both with reference to the present day and with reference to Antiquity, the term Arabic is used with different meanings. One thing that everyone agrees on is that the term includes Classical Arabic as written during the first centuries of the Islamic era as well as the Modern Standard Arabic used primarily in writing in the Arab world today (in Arabic, these varieties are often collectively referred to as Fuṣḥā ‘most eloquent [language]’). Whether other language varieties are counted as Arabic then reflects a judgement on their closeness to these prototypical forms of Arabic on linguistic, but also social, cultural, and political grounds.2 In Antiquity, the question revolves around the classification of the various Ancient North Arabian corpora, defined as Arabian languages, other than the Ancient South Arabian ones, written in the South Semitic alphabet. The different groups comprising this category are separately known as Safaitic, Hismaic, Taymanitic, Dadanitic, Dumaitic, Hasaitic, dispersed Oasis North Arabian, and Thamudic B, C, and D (Macdonald, 2000). Many of the languages written in these scripts are too poorly attested or understood to establish their exact relationship with other Semitic languages. Of the rest, Taymanitic apparently shares some features with Northwest Semitic languages like Aramaic and Canaanite (Kootstra, 2016), Dadanitic shares some features with what is generally recognised as Arabic, and especially Safaitic and Hismaic are quite close to it (Al-Jallad, 2018b). Macdonald (2000, pp. 48–49, 2009, pp. 312–313) distinguishes these from what he calls Old Arabic, barely attested in the epigraphic record, based on a few grammatical features that set them apart from prototypical Arabic. Given their overall similarity, Macdonald does employ the term North Arabian to refer to both Ancient North Arabian and Arabic collectively, something in which he has been followed by authors such as Butts (2018).
5. Al-Jallad (2018b), on the other hand, points out the many linguistic features shared by prototypical Arabic, Safaitic, and Hismaic and argues that all these language varieties can collectively be called Arabic. Some further support comes from a recently published Safaitic inscription which may attest to the use of the word ʿrb or ʾʿrb as an ethnic self-designation (Al-Jallad, 2020b); certainly, many of the Ancient North Arabian texts are associated with geographic areas and ways of life that have traditionally been associated with Arabs (see below). Ultimately, the difference between these uses of North Arabian and Arabic is one of semantics. The present study follows Al-Jallad’s convention of using Arabic to include Safaitic and Hismaic. We will use North Arabian to include Dadanitic, which is linguistically close to prototypical Arabic, Safaitic, and Hismaic, but lacks certain innovative features that they all share.
6. The question of defining Arabic has already led us to touch on the question of defining Arabs. Up to the late 20th century, the question of whether the Nabataeans spoke Arabic or Aramaic was rarely kept completely separate from the question of whether they were Arabs or Arameans. This essentialisation of ethnicity, where speaking Arabic, having an Arabic-derived name, a historical link to Arabia (the peninsula or the Roman province), nomadic pastoralism, or worshipping a particular god are all just superficial diagnostics of a deeper, independently existing Arabness, should be rejected. As is the case today, it was not at all necessary for all of these factors to occur together in Antiquity either.3 We cannot infer the languages someone used from their name, occupation, or religious practices. Hence, the focus of this paper will lie on the concrete use of Arabic or Aramaic language, not on a deeper and ill-defined Arab or Aramean ethnicity as a proxy. We will, however, discuss some of the cultural and ethnic arguments that have been made in the literature for the sake of completeness.
7. The final concept that must be problematised at the outset is that of the Nabataeans themselves. Macdonald (1998) cautions against a facile extrapolation from script to ethnicity: it is highly unlikely that everyone who wrote in what we call the Nabataean script ‘was’ Nabataean in any meaningful sense. In the Nabataean corpus, the only ways the term is used are in the frequent phrase mlk nbṭw ‘king of the Nabataeans’ or ‘king of Nabataea’ and with reference to ḥlyqt ḥrm (…) nbṭw wšlmw ‘the custom(s) of inviolability (…) of the Nabataeans and the Salamaeans’. In other corpora, nbṭy and nbṭwy ‘Nabataean’ are attested as self-designations a few times,4 but it is unclear what the term would have meant exactly to the writers who used it. As emphasised by Anderson (2005) and Alpass (2011), the Nabataean kings ruled over a realm that was characterised by diversity, and we cannot simply assume that all their subjects shared one language, religion, or any other potential markers of identity. Hence, we will primarily focus on Nabataean Aramaic inscriptions as defined by the use of the typically Nabataean script and the Aramaic language, while taking note of other texts by authors who explicitly identify themselves as Nabataeans.
8. 2 EVIDENCE FOR THE USE OF ARABIC
9. 2.1 Evidence from culture/ethnicity
10. Besides the discussion in his grammar of Nabataean, Cantineau explores the role of Arabs, Arabic and Aramaic in the Nabataean realm further in a separate paper (1934–1935 https://onlinelibrary.wiley.com/doi/10.1111/aae.12234#aae12234-bib-0020). Here (p. 77), he explicitly states at the outset that the Nabataeans were ‘not only of Arab origin, but probably also Arabic-speaking, the Aramaic dialect referred to as “Nabataean” being a learned language used besides an Arabic vernacular’, and that it is not certain that Nabataean was ever in spoken use. He provides a number of arguments for this thesis. Cantineau traces the recorded history of ‘the Arabs’ during the first millennium bce. He notes that they are represented as nomads up to the second century when the weakened Seleucid Empire ceded control over cultivated lands to Arab dynasties in certain areas. Cantineau bases his assessment of these dynasties and states as Arab on onomastics. He also notes the supposed arrival of Safaitic speakers (les Ṣafaïtes) in and near the Hauran around the same time.6 It is inconceivable to Cantineau that a state that depended on caravans and desert trade routes, such as the Nabataean kingdom, should be run by a sedentary population rather than a nomadic one; in this period, he asserts, this amounts to a distinction between Arameans and Arabs. Second, Cantineau cites the many places where Greek authors from Classical Antiquity refer to the Nabataeans as Arabs. These range from Diodorus Siculus (first century bce) to Flavius Josephus (late first century ce).(edited)
11. Third, he states that with the exception of bʿl šmn ‘the Lord of Heaven’ and his temple in the far north of the Nabataean realm, all the major Nabataean deities are Arabian: Dusares, Allat, Manot, al-Uzza, and Hobal. He argues that this is a much more one-sided situation than at Palmyra, where he concedes that the significant worship of the Mesopotamian god Bel does not mean that the Palmyrenes were actually Babylonian. Returning to the importance of onomastics for his argument, Cantineau argues against the likelihood of non-Arabs bearing Arabic names at this time and provides some criteria for identifying a name as Arabic in origin (pp. 83–84). He then conducts a survey of attested Nabataean names starting with g- (to arrive at a manageable sample). The results (p. 91) are presented in Table 1. Cantineau has counted the minimum number of distinct persons bearing each name, based on genealogy and context, and separated these out by attestation in the Sinai Peninsula or elsewhere in the Nabataean realm. No name in his sample is unambiguously Aramaic in origin.

1. Cantineau concludes that the Nabataean onomasticon was overwhelmingly Arabic, supporting his identification of the Nabataeans as Arabs. He qualifies this (p. 92) by saying that non-Arab populations were present in part of the Nabatean kingdom, namely, the Hauran, and perhaps also in northern and central Transjordan. But in ‘Nabataea proper’ (la Nabatène proprement dite), that is, southern Transjordan up to the Hijaz, Cantineau maintains that the sedentary population was of Arab origin. He draws parallels with the cultivation of the desert oases by seminomadic or sedentary populations ‘of Arab stock’ (de race arabe) and sees ‘no reason to believe that it was not the same in the properly Arab part of the Nabataean empire’.7 Besides the circularity of this argument—the properly Arab part of the empire, that is, where Arabs lived, was populated by Arabs—Cantineau does not seem to notice how this contradicts his equation of sedentariness with Arameans and nomads with Arabs.
2. Cantineau is sufficiently aware of the distinction between language and ethnicity that he explicitly and separately considers the case for Arabic as a spoken language among the Nabataeans (pp. 92–93). The strongest argument, in his view, comes from the references to the Nabataeans as Arabs, as for the ancient authors, ‘race and language are intertwined’ (race et langue se confondent). Had the Nabataeans spoken Aramaic, they would have gone down in Greek literature as Σύριοι (‘Syrians’, ‘Arameans’). A second argument comes from the linguistic evidence for contact with Arabic described in Cantineau’s grammar, which we will discuss below. Notably, he does not cite the onomastic evidence in support of an Arabic vernacular (pace Macdonald, 2000, p. 47).8 As for the status of Aramaic, Cantineau goes so far as to deny the existence of Nabataean Aramaic as a separate dialect: it is merely a collection of Aramaic texts written in a certain time and place, comparable to the corpus of French texts written in Mandatory Syria, which do not constitute a separate dialect either.

1. 2.2 Evidence explicitly connecting Nabataea and Arabic languages
2. A first- or second-century ce inscription found in 1979 at En Ovdat (or En Avdat, also Ayn Abadah; Negev & Shaked, 1986) contains valuable evidence of the use of Arabic. The Aramaic text, which refers to the deified Nabataean king Obodas, securely establishes the Nabataean connection, while the three lines of Arabic poetry attest the definite article ʾl- (on ʾlmwt or ʾlmwtw ‘death’) also known from Nabataean Arabic names, as opposed to the other definite articles attested in the Ancient North Arabian corpora. The origin of the Arabic text is debated, leaving questions about how its language relates to the vernacular unanswered. Another important datum on the language use of ‘the Nabataeans’ comes from a Safaitic inscription that Macdonald (1998, p. 186) discusses in the context of the difficulty of establishing ethnicity based on epigraphic texts. This inscription was written in Safaitic by a man who identifies himself as hnbṭy ‘the Nabataean’.9 Two other such Safaitic inscriptions are mentioned by Al-Jallad (2020b, p. 157), both by different individuals than the first.10 Here, we have hard evidence that at least three people who considered themselves Nabataean—for whatever reason—wrote Arabic, although the content of these inscriptions is minimal. Supporting evidence comes from a number of Nabataean inscriptions occurring next to Hismaic ones by the same author (e.g., Hayajneh, 2009; Norris & Al-Manaser, 2020), showing that some writers were proficient in Arabic as well as Nabataean Aramaic.
3. 2.3 Contact features in Nabataean Aramaic
4. Cantineau (1930–1932, vol. 2, pp. 171–173) provides a list of linguistic borrowings from Arabic as part of his appendix on vocabulary. In terms of phonology, he identifies six features. But these are either limited to Arabic personal names and possible loanwords (preservation of -t in the feminine ending; spelling of ḍ [i.e.,ṣ́] with ṣ; preservation of word-initial w-); shared with other contemporary varieties of Aramaic (see Section 3.3; the features are the interchange between s and š and the occasional shift of ā toō); or hard to link to Arabic specifically (interchange between n and l). No clear phonological Arabisms seem to have affected Nabataean Aramaic as a whole. Morphologically, Cantineau lists a number of features, which are limited to incidental occurrences in graffiti, several of which occur in what is arguably an Arabic text (JSNab17), and some features that are more accurately classed as loanwords, but nothing that affects the morphology of Nabataean Aramaic as a whole. The only syntactic feature he notes is again limited to the linguistically Arabic inscription JSNab17, besides some features that are typical of Western Aramaic as opposed to Imperial Aramaic. In contrast with this lack of clear grammatical influence, Cantineau does cite a considerable number of loanwords, which we will discuss below (Section 4.5).
5. O’Connor identifies the remaining loanwords as mostly relating to funerary practice and related legal issues and concludes (p. 220):
   • The class of Arabic loanwords does not, it seems, testify to the strong influence of Arabic on the Nabatean language as a whole, as Cantineau thought. Rather, the class of words reflects rather closely the fact that the most important finds at Madāʾin Ṣāliḥ, as at Petra, are funerary.
   (edited)
6. Despite O’Connor’s otherwise sound analysis, this conclusion amounts to an argument from silence. In the absence of significant numbers of non-funerary texts of considerable length, we do not know what the distribution of Arabic loanwords over other semantic fields looks like and cannot conclude that it was strongly related to burial practices. Indeed, the subsequent publication of Nabataean legal papyri (Yadin et al., 2002; Yardeni, 2001) has revealed a fair number of loanwords from other domains. Macdonald (1998, p. 184; similarly 2000, p. 47; and earlier Levinson, 1974) draws attention to the distribution of the Arabic or Arabian loanwords in Nabataean. He notes that the bulk of these occurs in texts from North Arabia, namely, Mada’in Salih/Hegra and Rawwafa. Three of the four remaining words occur in just two texts from Petra, while the remaining word is limited to texts with strong connections to the Safaitic corpus, where the same word (ʾl ‘lineage, tribe’) is highly frequent. ‘Thus’, he writes,
7. the “Nabataean” language as a whole, and one should be very careful how one defines that, is not permeated with loan-words from Arabic; they are confined to the dialect used in North Arabia, which is what one would expect.
8. Butts (2018) studies the North Arabian elements in Nabataean Aramaic from a contact-linguistic perspective, using the framework of Van Coetsem (1988). Butts’s study is limited to the tomb inscriptions from Mada’in Salih/Hegra, comprehensively published by Healey (1993). Out of O’Connor’s narrowed-down list of loanwords, 12 nouns occur in these texts. These account for nearly 11% of the total noun types attested in the inscriptions, a fairly high concentration (citing Butts, 2016, p. 208), similar to the concentration of Greek loanwords that are nouns in the Syriac Life of John of Tella (10.47%), which has a relatively high concentration of Greek loanwords for a Syriac text, and just less than those found in the Jewish Palestinian Aramaic portion of Genesis Rabbah (13.70%), a text in which the Greek language is highly thematized.
9. Butts concludes (p. 53) that the presence of grammatical borrowings points toward what Van Coetsem terms source language agentivity or imposition, consistent with a scenario where Arabic was the authors’ dominant language. But according to Van Coetsem’s model, lexical borrowing is not normally expected in this scenario. This leads Butts to propose ‘a case of more extreme imperfect learning in which the speaker (or: writer) not only fails to adequately produce the phonology and/or (morpho-)syntax of the recipient language—the expected changes—but also must occasionally resort to their more dominant language for lexical items’ (p. 54), employing a concept (imperfect learning) from another major contact-linguistic framework (Thomason & Kaufman, 1988). As it is uncertain that the construction of mn without dy reflects North Arabian influence, it may be easier to identify what Van Coetsem calls recipient language agentivity or borrowing: Nabataean Aramaic as attested at Mada’in Salih results from the selective incorporation of certain North Arabian words and phrases by language users for whom Aramaic was dominant. This can also account for the optative use of the perfect, which is largely limited to the expression lʿn ‘may … curse’, a phrase that also contains a lexical borrowing. This supports the possibility that the phrase ‘may … curse’ was borrowed as a whole. Perhaps together with a reanalysis of passive participles like dkyr ‘remembered (be)’ and bryk ‘blessed (be)’ suggested above, this may then have enabled the construction to spread to other lexemes, as in šhdw ‘may … (pl.) witness’.
10. 2.4 Conclusions on the use of Arabic
11. There is ample evidence for the use of Arabic in the Nabataean realm. This is clear from the attestation of three lines of Arabic religious poetry in a Nabataean text from the Negev, from the self-designation of certain writers of Safaitic inscriptions as Nabataeans, from parallel Nabataean Aramaic and Hismaic texts by one and the same author, and from references by outsiders to Arabic being used in the former Nabataean kingdom (Section 2.2), as well as a high number of North Arabian lexical borrowings in the Nabataean inscriptions and papyri (Section 2.3). The borrowings in Nabataean Aramaic reflect the use of Arabic for legal purposes and in the context of local customs and modes of subsistence. The En Ovdat inscription attests to the use of Arabic for religious purposes. Finally, the Ancient North Arabian inscriptions may well attest to the use of Arabic as a spoken vernacular. All in all, Arabic seems to have been used in a wide range of contexts.
12. EVIDENCE FOR THE USE OF ARAMAIC
13. In a now classic overview of the languages of the Arabian Peninsula in Antiquity, Macdonald (2000, p. 48) provides an argument for the spoken status of Nabataean Aramaic based on the attested forms of the divine name Dusares, etymologically probably ‘the one of the Shara (mountains north of Petra)’. In the Hismaic inscriptions from Wadi Ramm, near Petra, this name is usually spelled in its expected Ancient North Arabian form, ḏs2ry. In the Safaitic corpus of the Hauran, however, the name is usually spelled ds2r. As inherited ḏ (as inḏū ‘the one of’) is normally spelled ḏ in Safaitic, the spelling with d suggests borrowing via an Aramaic source, where *ḏ had merged with d. Macdonald explains (note his distinction between Old Arabic and Safaitic, discussed in Section 1):
14. If these were speakers of Old Arabic there would have been no point in their communicating with the nomads in Aramaic, since Old Arabic and Safaitic would have been mutually intelligible. In this case, the name Dushara would have come into Safaitic with an initial /ḏ/ and a final, consonantal /y/. Instead, it was borrowed in its Aramaic form with an initial /d/ and (presumably) a final vowel, and this suggests that those Nabataeans from whom the cult was adopted spoke a dialect of Aramaic.
15. This scenario is complicated by Macdonald’s (2018) more recent discussion of a Hismaic inscription where the author consistently spells *ḏ with d instead of ḏ. Macdonald suggests that this reflects an Aramaic accent in the author’s Hismaic, but it could just as well show the merger of the two sounds in some variety of Arabic at the time, which could also have influenced the Safaitic spelling of Dusares. On possible borrowings from Aramaic into Arabic reflected in the sixth-century CE papyri from Petra, see the following section.
16. 3.2 Evidence from usage
17. Of course, the written use of Aramaic in the Nabataean realm is evident from the many Aramaic texts that have reached us, but this does not show that it was also in spoken use. Macdonald (1998, p. 188) notes that while the Nabataean graffiti from Sinai are largely formulaic, the occasional non-formulaic information they contain is also written in Aramaic. He rightly questions whether a purely literary language would be used in such cases. Owens (2018, pp. 448–451) argues for Arabic–Aramaic bilingualism in the Nabataean sphere, predominantly based on the large overlap between Arabic and Aramaic speakers in the Ancient Near East more generally and the apparent use of both languages in JSNab17. Neither argument is convincing: what holds for, for example, Palmyra cannot automatically be applied to Nabataea, and JSNab17 can plausibly be interpreted as an Arabic text that makes heavy use of written Aramaic conventions. Finally, an analysis of personal and place names in Greek papyri from sixth-century CE Petra by Al-Jallad (2018a) shows evidence for the use of both Arabic and Aramaic at that time. Certain place names occur with both Arabic and Aramaic morphology (p. 41): αραμ (ʾārām orʾaʾrām) besides εραμαεια (ʾiram-ayyā) ‘(the) field markers’, αλνασβα (al-naṣbah) besides νασβαθα (naṣbat-ā) ‘the farm’, and αλκεσεβ (al-qiṣb) besides κ(ε)ισβα (*qiṣb-ā), perhaps ‘the canal’. Certain words with Aramaic etymologies also occur with Arabic morphology, indicating lexical borrowing from Aramaic into local Arabic. While this evidence postdates the latest Nabataean inscriptions, the suggestion of a continued presence of Aramaic at Petra up to this time is intriguing.
18. 3.3 Evidence from language change

1. Cantineau (1934–1935) already remarks upon some of the grammatical differences between Nabataean and Imperial Aramaic and notes that they are shared with other forms of Middle Aramaic. In his view, this shows that Nabataean Aramaic was not an independent dialect. Instead, Nabataean scribes were trained in Aramaic-speaking centres elsewhere in the Levant (e.g., Judea, Palmyra, Edessa), learning the language as part of their scribal education. As a result, they brought some of the linguistic innovations that the local forms of Aramaic had undergone back home with them, introducing them to the Nabataean written language. Linguistically, this is plausible enough, but Cantineau does not explain how this can be reconciled with the existence of a distinct Nabataean script and orthography. If Nabataean scribes were trained elsewhere, why do the Nabataean inscriptions feature their own alphabet and not (for instance) Jewish script, Palmyrene, or Old Syriac? The next study to focus on language change is that by Morgenstern (1999). He compares the texts that can be securely linked to the Nabataean kingdom (thus excluding the graffiti from Sinai) to Imperial Aramaic as attested in official documents from Elephantine and Samaria. He notes the following differences that are not due to Arabic influence, all of which should be interpreted as innovations in the written standard of Nabataean vis-à-vis Imperial Aramaic:
2. • 1. Phonetic spelling of Proto-Aramaic *ḏ, which had shifted to /d/, with d instead of z (most significantly in the relativizer dy and singular near demonstratives dnh and dʾ);
   • 2. plene spelling of the masculine plural ending *-īn as -yn besides the defective Imperial Aramaic spelling -n;
   • 3. occasional assimilation of n to a following consonant, as inʾantatā > ʾttʾ ‘the wife’ and mṣbʾ ‘the stele’ (from the root n-ṣ-b);
   • 4. rounding of ā toō in ʾināš > ʾnwš ‘person’ and some nouns in-ān like pqdwn ‘responsibility’ (also in *tamānē > tmwnh ‘eight [f.]’);
   • 5. the use of the plural demonstratives ʾln and ʾnw besides Imperial Aramaic ʾlh;
   • 6. the use of the adverb blḥwd ‘alone’ (possibly borrowed from Phoenician);
   • 7. the C-stem forms ʾqṭl (suffix conjugation) and yqṭl (prefix conjugation) besides the forms that are more common in Imperial Aramaic, hqṭl and yhqṭl;
   • 8. t-stem forms with a sibilant first radical that does not show metathesis, like ytzbn ‘it may be sold’, besides metathesised and assimilated yzdbn;
   • 9. the use of the direct object marker yt instead of Imperial Aramaic l-;
   • 10. the use of the root yhb ‘to give’ in the prefix conjugation (once), besides Imperial Aramaic yntn from another root.
3. We may add a few more features:
4. • 11. Normally, the expression ‘the life of PN’ is expressed as a construct chain, ḥyy PN, while Imperial Aramaic expresses inalienable possession of this type with a periphrastic construction (noted for Biblical Aramaic by Garr, 1990) only attested in the oldest known Nabataean inscription: ḥywhy zy PN literally ‘his life of PN’ or ‘his life, PN’s’ (Suchard, forthcoming);
   • 12. the loss of the second radical *w in the prefix conjugation of hwy ‘to be’, attested once in yhw ‘they may be’ (Alzoubi & Smadi, 2016), besides Imperial Aramaic yhww(n).
   • After the Nabataean period proper, additional innovations occur (Suchard, forthcoming):
   • 13. Loss of an unstressed final vowel in *ʾaqīmū > ʾqym ‘they erected’ in an inscription from Tayma dated to 203 ce (Al-Najem & Macdonald, 2009) if this is not to be explained otherwise (see Section 4.4);
   • 14. spelling of ʿalōhī ‘over him’ as ʿlhwy in the same inscription, possibly indicating a pronunciation asʿalōy with loss of the *h;
   • 15. phonetic spelling of *rēš ‘head, chief’ as ryš (twice) instead of older rʾš in an inscription from Hegra dated to 356/7 ce (Stiehl, 1970).
5. Likely and possible Nabataean loanwords from North Arabian:

1. Conclusion
2. Good arguments can be found to support the spoken use of both Arabic and Aramaic in the areas once ruled by the Nabataean kingdom. The primary interaction between these languages seems to have been the exchange of loanwords. As a result, Nabataean Aramaic can serve as an early witness for certain Arabic nouns, verbs, and the occasional particle, but does not provide a direct window on Arabic as it was spoken at the time. Indirect evidence for Arabic grammar is sometimes available in the way in which these loanwords were borrowed: we have seen evidence for a somewhat retracted realisation of /s/, for contraction of -ay- but retention of-āy-, and, less securely, hints of broken pluralisation, verbal number incongruence, and the use of the subjunctive.