Splitting Definitives: The Separation of the Definite Article in Medieval and Pre-Modern Written Judeo-Arabic


This article explores the reasons behind the orthographic practice of representing the definite article in written Judeo-Arabic as an independent entity, a phenomenon which became widespread in Jewish Arabic-speaking communities in the pre-modern era. Commencing with its representation in fifteenth to nineteenth-century Egyptian Judeo-Arabic manuscripts, the orthographic feature is traced back to Judeo-Arabic texts produced in medieval al-Andalus, Sicily, and the Maġrib, and from there, to post-1492 CE Sephardī Jewish refugees, who settled in North Africa and Egypt. The phenomenon is revealed to be the result of a two-stage process: (i) direct language contact between Romance and Judeo-Arabic; and (ii) the influence of Judeo-Spanish writing on Judeo-Arabic spelling practices in diaspora communities after their expulsion from the Spanish Kingdoms.

4 The transliteration used here and elsewhere represents the written CA spelling, rather than the phonetic realization. Where phonetic transcriptions are used, they are bounded by square brackets []. 5 Written JA is generally categorized into three periods: early JA (ninth-tenth centuries CE); classical JA (tenth-fifteenth centuries CE); and late JA (fifteenth-nineteenth centuries CE) (Khan 2007). This categorization is based on orthographic developments, of which the representation of the definite article, discussed here, is an important feature. For an alternative categorization of JA, which combines written and spoken manifestations of JA, see Hary 1992Hary , 2009. In using the term 'Judeo-Arabic,' I do so in reference to written manifestations of the language, following the example set by Khan (2007) to mean simply 'Arabic written in Hebrew script.' For a recent discussion of spoken JA dialects, see Khan (2018). 6 Blau & Hopkins (1984) differentiate between scribes educated in CA and those ignorant of its principles; the former used classical JA spelling before the tenth century, while the latter universally favored early JA spelling (which Blau & Hopkins state was based on Hebrew and Aramaic spelling practice). Ackerman-Lieberman has convincingly challenged this dichotomy, instead suggesting that the type of spelling used may be determined by the urban/rural divide present in many Arabic-speaking countries at the time ( the definite article is regularly written independently of the noun or adjective it modifies (Khan 1992(Khan :231, 2006(Khan :51, 2010Hary 2009:110, §1.15;Wagner 2010:6, §4.7.1).7 Although the definite article's transformation into a separate entity8 is an oft-noted feature of eighteenth and nineteenth-century JA texts (see, for example, Khan 1992Khan :231, 2006Wagner 2010:66-67), to the best of my knowledge, the cause of its separation has yet to be explored. A potential explanation -twofold in nature -for this orthographic development is proposed in this article. Commencing with a brief examination of the definite article's representation in late Egyptian JA folk tales and letters ( §2.1), I then examine orthographic precedents for the separation of the definite article in Arabic and JA texts ( §2.2). While this exploration provides some significant insights into an orthographic practice that may have precipitated this phenomenon's emergence, it does not offer a comprehensive solution. Thus, I turn my attention to pre-seventeenth-century texts which display this phenomenon, 7 Wagner also records two instances in eleventh-century letters (which she attributes to Egypt and the Maġrib, respectively) in which the definite article is written as an independent entity (2010:66, §4.7.1). The first of these (CUL T-S 13J17.3) was written by Efrayim b. Ismaʿīl al-Ǧawharī, a merchant based in Alexandria, but most probably of Maġribī origin (see Appendix, Table 3, number 85). The second letter (CUL T-S 12.218) contains consistent separation of the definite article from the noun it modifies. This trader's letter -dated to 'the sixth of šawwāl, the year four hundred and eleven' (i.e., 23 January 1021 CE) is written in a distinctly Sephardī script-style (complete with ligatures for final heh) and contains references to Cordoba and the Berbers (see recto, lines [18][19]. I, therefore, suggest that this text may be more reflective of Andalusī writing practices than Maġribī standards of the day. I am grateful to Dr. José Martínez Delgado for discussing my doubts about the Maġribī origins of this text with me. 8 The definite article is categorized as an independent entity here when the space between it and the noun or adjective it governs is equidistant to other morphologically discrete units. So, for example, in the hand of Ismaʿīl b. Isḥāq Al-Andalusī (see Appendix, Table 1, number 2), the definite article is consistently separated from its noun or adjective by a space of approx. 0.2cm; the same distance which separates the majority of words from one another in this writer's letters (see, for example, CUL T-S 10J12.5, recto, line 6). Meanwhile, Efrayim b. Ismaʿīl al-Ǧawharī's (see Appendix, Table 3, number 85) writing has been described as displaying a tendency towards separation because, in some instances, the distance between the definite article and its noun is equidistant (0.2cm) with that between the preceding preposition and the definite article (e.g., CUL T-S 13J17.3 recto, line 6), but more often than not, the definite article is written attached to the noun it modifies, or is at least only 0.1cm apart from the following noun, rendering it closer than other independent morphological units. In those letters which are listed as not containing an independent definite article, the space between the definite article and its noun or adjective is consistently less than the space between morphologically distinct units (see, for example, CUL T-S 13J25.6, written in the hand of Ḥalfon b. Menašše b. al-Qaṭāʾif (for other letters and documents by the same scribe, see Appendix, Table 4, number 115)).
leading me to explore the potential influence of Romance orthography on the writing practices of Andalusī Jews in the medieval era ( §2.3). In a quest to understand the phenomenon's emergence as a stable feature of postseventeenth century written Egyptian JA, I also examine the potential impact of Judeo-Spanish (henceforth JS) on Egyptian JA writing practices post-1492 CE ( §2.4). I contend ( §3) that the definite article's emergence as an independent entity in written JA was a twofold process. The first stage was borne of direct contact with Romance dialects and extended to areas of the Maġrib with close ties to Al-Andalus. The second stage of the process was a consequence of the mass migration of Sephardī Jews from the Iberian Peninsula after 1492 CE, the influence of whom (at least in this regard) is felt as far east as Baghdad.

The Definite Article's Representation in Late JA Folk Narratives and Letters9
The dawn of the fifteenth century is generally considered a turning point in JA orthography (see Khan 2007;10 Hary 199210 Hary , 2009); it signifies the close of the classical JA period, in which JA spelling practices were primarily informed by CA, heralding the advent of an alternative writing style, often designated 'Hebraized' in contemporary scholarship (see Hary 1997Hary :37, 2009 It should be noted that the shift from classical to late JA orthography occurs in different geographical regions at different times (see Avishur 1986:3 regarding the situation in Iraq) and -in the case of Yemen -not at all (Khan 2018:150, 154). Here, the main focus is on Egyptian written JA. 11 The Hebraisms -derived from Rabbinic Hebrew orthographic norms -include: (i) the representation of consonantal waw and yāʾ with double vav and double yod, respectively; (ii) increased plene spelling of the short vowels /i/ and /u/; (iii) and the representation of final long /ā/ with heh rather than ʾalef or yod. 12 While the influence of Rabbinic Hebrew writing practices on late JA orthography is clearly discernible, it is possible that direct Arabic influences, most notably in the use of the diacritical dot, have been overlooked in JA scholarship (Connolly 2018).
A brief survey of late JA texts -dating from the fifteenth to nineteenth centuries -is necessary to: (i) determine the approximate date at which the writing of the definite article as a separate entity becomes a common phenomenon in Egyptian JA texts; and (ii) qualitatively assess the regularity of the feature's use after this date. The following sections summarize findings regarding this feature in (i) fourteenth/fifteenth-and sixteenth-century and (ii) eighteenth-and nineteenth-century JA folk tales and letters.
All the Egyptian JA folk narratives and letters from the fourteenth/fifteenth and sixteenth centuries adhere to the classical JA (and Arabic) practice of attaching the definite article to the following noun or adjective. This practice has all but been abandoned in the eighteenth-and nineteenth-century Egyptian folk tales and letters in favor of the definite article's representation as an independent entity.16 Thus, the seventeenth century (approximately speaking) would appear to be the period in which this orthographic shift became widely adopted among Egyptian Jewish writers. As can be seen, however, from its occurrence in the Sephardī letter, Bodl. MS Heb.c.72/18 (see above), this phenomenon predates the seventeenth century. Does this constitute an idiosyncratic quirk of a Sephardī Jewish Arabic-speaking merchant, or is it 14 This is only the case, however, when the definite article is not preceded by the bound morphemes bi-'in, with' (CA: bi-) or li-'to, for' (CA: li-). In such cases, the definite article forms a separate entity with the bound morpheme, e.g., ‫אסם‬ ‫באל‬ 'in the name' (CA: bi-ʾl-ʾism) (NLI Cairo JC 104, 2 verso, line 5); ‫טוראב‬ ‫באל‬ 'with the earth' (CA: bi-ʾl-turāb) (NLI Cairo JC 104, 7 recto, line 12). This practice is common to all late JA texts that I have examined. 15 It seems, however, that this adherence to Arabic orthographic norms is the exception rather than the rule in eighteenth to twentieth-century JA texts. The tendency to write the definite article as an independent entity is evident even in printed JA material from Baghdad in the early twentieth century CE. 16 Unfortunately, we do not have at our disposal JA texts that can be confidently dated to the seventeenth century CE.
perhaps indicative of a more widespread writing practice, prevalent among Sephardī Jews?

Orthographic Precedents in Arabic and JA
In early Arabic documentary and literary texts, division of words at the end of a text line is a common phenomenon. Borne of a desire to ensure a rectangular written area, with four neat sides (Gacek 2012:146), it was superseded by the elongation (mašq) or contraction (ǧamʿ) of letters, or the superscription (taʿlīq) of the final word(s) of a line (ibid.). The latter practice was particularly prevalent in the Eastern tradition, while the former found favor in the Maġrib. The scribes of earlier texts in which word division is frequently encountered at the end of the text line do not seem to have been concerned with morphological boundaries, so much as with aesthetic uniformity. Yet, the practice of separating words for the purposes of textline justification is not as arbitrary as it may first appear; only those graphemes which do not connect to the following grapheme by means of a ligature -such as ʾalif, wāw, rāʾ, etc. -are used for this purpose. Thus, in an early ḥiǧāzī-script copy of the Qurʾān, BnF Arabe 328(a), recto, lines 13-14, we find the lone ʾalif of the definite article positioned at the end of the text line, while its lām sits at the start of the following text line, con-  These lines read as follows: The same preoccupation with text line justification is apparent in some JA manuscripts. Hebrew graphemes, however, are inherently independent; they are not connected to one another with a ligature. Therefore, the points at which a word may be divided are as many as there are letters. It is noteworthy, therefore, that the division of the word in JA texts most often occurs at the point at which the definite article ends and the noun or adjective it modifies begins; the ʾalef and lamed of the definite article -written as a single entity, often ligatured -are positioned at the end of the text line, while the noun or adjective is written at the start of the following text line (see Fig. 3). This tendency to separate the definite article and noun from one another for the purposes of textline justification may be indicative of its perception as a morphologically distinct entity. This same tendency may also motivate the orthographic practice of writing the lām (or lamed) of the definite article, even when it is assimilated, in CA (and JA). However, while this representation of the 18 The manuscript reads as follows: For another example of this phenomenon, see Bodl. MS Heb.d.11/9, recto, lines 30-32, a letter written by (Abū Zikrī) Yahūda b. Saʿadya ha-Rofe. definite article in classical JA texts may grant us some insight into how the definite article was perceived among Arabic-speaking Jews in the medieval era, it does not answer the question of how the independent definite article emerged as a standard feature of post-seventeenth-century Egyptian JA texts. To answer this question, it is necessary to delve further into the history of orthographic representations of the JA definite article, beginning with the earliest evidence of this phenomenon that we have encountered so far, namely, the fifteenthcentury JA letter Bodl. MS Heb.c.72/18.

Al-Andalus, Romance Orthography, and Sephardī Jews
Bodl. MS Heb.c.72/18 is found among a cache of letters addressed to Mošeh b. Yahūda20 in Alexandria. Amongst these letters, which predominantly appear to be sent from Cairo and the Maġrib, there is at least one other letter composed in the most westerly regions of the Arabic-speaking world (  To assess the veracity of these preliminary impressions, I expanded this search to encompass individuals whose documents are preserved in the Cairo Genizah collections and about whom we have some biographical data (see the Appendix for Tables 1-4).24 The Tables list some 125 Jewish individuals of (i) Andalusī and Sicilian (Table 1), (ii) Maġribī, (Table 2), (iii) Alexandrian (Table 3), and (iv) Egyptian, Palestinian, and Iraqī (Table 4) origins, noting the 23 Here, the first ligatured definite article (prefixed with the preposition bi-) is written independently of the following noun, while the following definite article is attached to the noun. 24 The data available are often very limited. I have relied heavily on the historical works of Stillman (1973), Ben-Sasson (1976), Cohen (1980), Gil (1992, Simonsohn (1997), and Goldberg (2012) for much of this information. Where available, I have also used the writer's nisba -such as ʾal-Andalusī 'the Andalusian' -mentioned in the letters. This is by no means proof of the writer's geographical origin, as it may refer to a historical-geographical association belonging to a previous generation of his/her family, but it is used here in the absence of other details.
classmarks which have been attributed to them, and whether or not the definite article is written as a separate entity in their letter(s) and documents. The overall trends revealed by this search confirm that the separation of the definite article was not simply an orthographic peculiarity of ha-Levi and Maimonides or the two fifteenth-century merchants, but was a widespread orthographic practice among Andalusī and Sicilian Jews (see Table 1, individuals 1-31).
The propensity towards writing the definite article as an independent entity is also evident among Maġribī Jews, albeit to a lesser extent (see Table 2, individuals 32-84). A little over half of the Maġribī writers consistently separate the definite article from the noun/adjective in their letters. In some of the texts (approximately 30%), the writers seem uncertain of the definite articles' status, sometimes writing it as a separate entity, sometimes as a bound morpheme. The remaining minority (approximately 15%) write the article attached to the noun/adjective.
Of those letter-writers who lived and worked in Alexandria25 (see Table 3, individuals 85-92), three merchants wrote the definite article as an independent entity, while the remaining five veered between attached and detached representations of the definite article.
The final category (Table 4, individuals 93-125) contains letters from merchants and Jewish notables who lived in Egypt, many of whom emigrated from Iraq, Palestine, or present-day Syria as adults. Of these, only two separated the definite article from the noun/adjective (see Table 4, numbers 108, and 122), while another individual showed an inconsistent representation of the definite article (see Table 4, number 121).
This review of the Sephardī, Maġribī, Egyptian, Palestinian, and Iraqi Cairo Genizah documents indicates that the separation of the definite article from the noun/adjective it modifies was an enduring practice among Sephardī and -to a lesser extent -Maġribī Jews. Yet, we are still no closer to understanding the motivation behind the phenomenon's development. In order to do so, it is necessary to examine the emergence of the Romance languages in the Iberian Peninsula.
The early Romance dialects -which gradually fragmented and evolved into the independent languages Catalan, Spanish, French, Romanian, Portuguese, 25 It is probable that some of the individuals included in this category -such as Efrayim b. Ismaʿīl ʾal-Ǧawharī (see Appendix, Table 3, number 85) -were in fact of Maġribī origin. However, as little is known for certain about their birthplace and educational backgrounds, I have included them in this category.
and Italian -are documented in written form26 as early as the eighth and ninth centuries CE (see Frank-Job & Selig 2016:24, 27). Therefore, the spoken forms are thought to have emerged as early as the seventh -mid-eighth centuries CE (see Herman 1996 for a discussion of the geographical differences in the development of Romance languages).27 Whether these early manifestations of the Romance dialects were considered independent of Latin by their native speakers and writers is much discussed (see Wright 1976Wright , 2016López-Morillas 2000;Ledgeway 2012; and Clackson 2016), but for our purposes, it is sufficient to state that Romance dialects were in circulation in the Iberian Peninsula prior to the Islamicate conquest of the region in 711 CE. More specifically, one of the most fundamental linguistic innovations of Romance languages -namely, their independent definite and indefinite articles,28 conspicuously absent in their predecessor language, Latin -was also either established or at the very least inchoate in the early eighth century CE (Ledgeway 2012:89-105;Ramat & Ricca 2016:52-53;Giusti 2016:551). When the feature came into wide circulation is a source of much contention; some scholars suggest that its relative morphological uniformity and functionality within the varieties of Romance languages indicates its emergence ca. sixth century CE (Giusti 2016:552); while others -on the basis of the extant written evidence -claim that it did not develop until the eighth century CE (Ledgeway 2016:766; see also Adams 2013). Yet, it is evident that it was at least incipient on the eve of the Islamicate conquest in 711 CE.
Thus, while little is known for certain about the language situation of Al-Andalus during the period of Muslim rule, and scholarly consensus regarding the interpretation of the scant evidence at our disposal is far from being achieved (Zwartjes 2006),29 we can establish with (relative) certainty that pre-711 CE, the inhabitants of the Iberian Peninsula -Christians, Jews, and Berbers, 26 These early eighth to late tenth century CE instances of written Romance vernaculars constitute snippets (such as riddles, anecdotes, glossaries, lists, and marginalia) within Latin texts. It is not until the mid-twelfth century CE that Romance texts are available on a larger scale (Frank-Job & Selig 2016:24). On the important distinction between written and spoken manifestations of the language, see Clackson (2016); and Wright (2016). 27 Even the most conservative estimates regarding the development of the Romance dialects place them in the early ninth century (Ledgeway 2012:1, n.2). 28 The various iterations of the definite article -el/la (Catalan/Spanish); le/la (French); il/ la (Italian), and so forth -are derived from the Latin demonstrative pronoun ILLE 'that' It is also worth bearing in mind that of the extant evidence the majority pertains to the elite, be they Christian, Jewish, or Muslim.
alike -spoke early Romance dialect(s). The literate among them likely wrote Romance and Latin (Wasserstein 1991:4). In 711 CE, the invading troops comprised Muslim Arabs, Jews from North Africa and the Arabian Peninsula, and -by far the largest component -Berbers.30 As such, a multilingual situation arose in Al-Andalus, in which the conquering Muslim elite spoke Arabic (as a minority), Berber, and (perhaps) a Romance dialect, while the conquered Christians, Jews, and native Berbers spoke Romance dialect(s).31 This heterogeneous linguistic situation is partially reflected in the issuing of bilingual Arabic-Latin coins in the earliest years of the conquest (Gallego 2003:110-111) and that circulated for some time after the initial invasion (Thompson 1971:68 cited in Zwartjes 2006); the inclusion of Latin, a concession to the realities of the linguistic situation in which the vanquishers found themselves; the use of Arabic, a declaration of their intention that it would one day dominate.
How rapidly this ambition was realized remains uncertain. It is welldocumented that no notable post-tenth century CE Latin texts survive (Wasserstein 1991:5;Gallego 2003:127) (and, by inference, may not have been produced, at all). That the Arabic language came to be held in high regard by the indigenous inhabitants of the Iberian Peninsula by the ninth and tenth centuries CE is evident from the translation into Arabic of the Psalms (pre c. 889 CE) and the Gospels (946 CE) (López-Morillas 2000:49); its mastery by eminent members of the Jewish community, such as the celebrated Ismaʿīl b. Naġrīla (d. ca. 1056 CE) (Bennison and Gallego 2010:146), and, later, Ibrahīm b. Sahl al-Išbīlī (d. 1251 CE); and the extensive production of grammatical, philosophical, poetic, and scientific texts written in Arabic by Christians and Jews alike during the tenth to thirteenth centuries CE. Scattered references in primary sources, in which contemporaries bemoan the loss of Latin knowledge among the Christian inhabitants of Al-Andalus (see, for example, Alvarus of 30 As to whether the conquering Berbers spoke Arabic already, there is currently no definitive answer. Some scholars have suggested that, in addition to Berber, some of the conquering Berber tribes may have spoken a Romance dialect (Wasserstein 1991:4;Gallego 2003:119). However, few seem to suggest that the Berbers would have already been Arabic-speaking at the time of the invasion, and recent scholarship has challenged the assumption that they would have undergone the process of Arabization immediately after the conquest (see The lack of post-tenth-century Latin sources originating from Al-Andalus does not necessarily preclude the continued use of spoken and written Romance dialects, at least among some strata of society. There is evidence suggesting that Romance was spoken -or at least written -in areas of Al-Andalus throughout the tenth and eleventh centuries. The most oft-cited manifestation of the continued awareness and use of Romance in written form is the muwaššaḥ, a manifestation of strophic poetry (the other being zaǧal, written in colloquial Arabic), for which Al-Andalus' Muslim and Jewish elite were famed (Gallego 2003:129-130;Rosen 2000:165-166). Within these final verses (referred to as the ḵarǧa 'departure') of these CA or -in the case of some of the Jewish community -Hebrew poems, are found a couple of lines written in Andalusī Arabic, Romance, or a combination of the two languages (Barletta 2012:769;Rosen 2000:165). Another example of the inclusion of Romance vocabulary in an Arabic-language text is Ibn Baklarish's Kitāb al-ʾadwiya ʾl-mufrada li-ʾl-Isrāʾīlī (commonly referred to as Kitab ʾal-Mustaʿīnī) (see Khan 2008:95-96).32 Suffice it to say,33 that the Romance languages' definite articles' superficial similarity with the Arabic definite article (pronounced -ʾil in some dialects) is conspicuous. The pre-existing tendency in early JA texts to separate the definite article from its noun for text line justification (see above) may have emphasized its morphological distinctiveness, leading the erudite Jewish writers of JA texts to separate it consistently, irrespective of its place on the text line. As this fashion appears to have been prevalent among the Jewish elite, it is not surprising that it gained currency and was adopted among those who may not have had intimate, first-hand knowledge of Romance dialects.
However, it is clear from Tables 1-4 (see Appendix) that while some Maġribī merchants embraced this orthographic innovation, it was not prevalent among 32 The work is thought to have been composed ca. 1100 CE, and the earliest extant copy is a privately owned manuscript (referred to as the Arcadian Library manuscript), dated 1130 CE (Burnett 2008). For a discussion of the Romance and Latin elements in this text, see Villaverde Amieva 2008. 33 It is not my intention to resolve the issue of Al-Andalus' uncertain linguistic situation. Nor do I claim that the potential influence of Romance orthography on the representation of the definite article in JA as a separate entity should be regarded as evidence for the continued use of Romance languages in Al-Andalus. It is possible that the feature emerged during the tenth or even ninth century, during a period of bilingualism, and was retained by Jews long after they ceased to write or speak Romance dialects.
Egyptian Jews in the Middle Ages.34 Therefore, we must keep searching for the reason(s) as to why it became a popular feature of post-seventeenth-century JA texts in the eastern regions of the Arabic-speaking world.

The Expulsion of Sephardī Jews from Spain
The expulsion of Muslims and Jews from the Castilian and Aragonese Spanish kingdoms in 1492 CE led to a large influx of Spanish-speaking refugees in North Africa, Egypt, and Palestine -estimates vary between 40,000 to 100,000 Sephardī Jewish refugees who arrived in Morocco and then dispersed across the Maġrib towards Egypt and Palestine.35 It is thought that the majority of Sephardī Jewish refugees hailed from the Catholic province of Toledo (conquered in 1085 CE) rather than the Muslim kingdom of Granada (which fell only in 1492 CE). Thus, it is likely that many of them spoke a Spanish or Castilian dialect rather than the Andalusī Arabic dialect prevalent in Granada (Heath 2002:10-11). Spanish, or JS, continued to be spoken among these refugees for at least two centuries after they arrived in the Maġrib, Egypt, and Palestine (Kraemer 1991:246). It is even recorded that some Jewish families in northern Morocco carried on using JS amongst themselves until the early twentieth century (Heath 2002:11). As for evidence of written JS (i.e., Spanish written in the Hebrew script), this exists both pre-and post-1492 CE (see Minervini 1992;Penny 2000;Schwarzwald 2004;andBunis 2011, 2016).36 The Cairo Genizah collections alone comprise JS letters, legal documents, responsa, medical texts, piyyūṭ, and biblical commentaries.37 Although slight in number, these texts, 34 Or at the very least, that this feature was not prevalent among those eastern educated Jews whose texts were deposited in the Ben ʿEzra synagogue's genizah. 35 Prior to the 1492 CE Edict of Expulsion, Jews had already begun to travel from Spain to Arabic-speaking lands due to persecution at the Catholic authorities' hands (Beinart 2002:2). 36 While pre-1492 CE JS texts contain Hebrew and Aramaic vocabulary and are written in Hebrew script, it has been demonstrated that the language (morphology, syntax, lexicon, and limited phonology) did not differ from the dialectal varieties of Christians (see Minervini 1992). It is only after the expulsion of the Jews from the Iberian Peninsula that JS developed its own characteristics: retaining archaic features that ceased to be used elsewhere in Spanish-speaking societies; innovating; and dialect-mixing among Sephardī Jewish refugees from different areas of Spain settling in new regions with one another (see Penny 2000:176-192). 37 This observation is based on a brief survey of JS material in the Cairo Genizah collections on the Friedberg site. A more comprehensive search of the Genizah's contents would, I'm sure, reveal more variety, and perhaps even a greater number of JS texts.

Conclusion
In this article, I have traced the orthographic representation of the definite article in written JA folk tales and letters from pre-modern Egypt to medievalera Al-Andalus in an effort to resolve the question of why the definite article is written as a separate entity in post-seventeenth-century Egyptian JA texts. In the course of this exploration, I have demonstrated that medieval-era manifestations of this orthographic phenomenon may be viewed as the result of JA's contact with Romance, or at least, as a result of pre-tenth-century Romance-JA language contact, the consequences of which were retained as a stylistic feature in post-tenth-century JA writings. By tracing the representation of the definite article in medieval JA texts written by (i) Andalusī and Sicilian, (ii) Maġribī, (iii) Alexandrian, and (iv) Egyptian, Palestinian, and Iraqi Jews, I have shown that this orthographic feature was limited predominantly to Jews of Andalusī, Sicilian, and Maġribī origins during this period. It was not until after 38 The Sephardī Jews who fled to the Maġrib, Egypt, and Palestine were generally considered to be more erudite and socially and economically well-off than the Jews living in Arab lands prior to 1492 CE (Heath 2002:10). This social prestige may have contributed to the assimilation of some of their customs and practices, including in the realm of writing. 39 These manuscripts may be found on the Friedberg Genizah Project website under the classmarks listed here.
Journal of Jewish Languages 9 (2021) 32-76 the expulsion of Jews from the Iberian Peninsula that the feature also became widespread in Egyptian JA writings. In this later period, the feature arose from analogy with the writing of the m.sg. definite article in JS. These findings attest to the significant role orthography may play in disentangling the historical circumstances of language contact, and they challenge and develop the widespread understanding of late written JA as predominantly 'Hebraized.' Table 1 Representation     (Cohen 1980:116).    Cohen (1980:241).