PHONETICAL AND PHONOLOGICAL ADAPTATIONS OF HEBREW LOANWORDS IN MOROCCAN JUDEO-ARABIC: THE CASE OF FEZ JONAS SIBONY GEO – University of Strasbourg Abstract: Like all Jewish languages, Judeo-Arabic dialects borrowed numerous terms from Hebrew and Aramaic. Many studies have already been conducted on modern North-African Judeo-Arabic, some concerning specifically Judeo-Moroccan dialects, in their written or oral forms. This study intends to offer a systematic analysis of how the Hebrew component of Judeo-Arabic in Fez behaves in terms of phonological and phonetical adaptation and to look in particular for regularities in that adaptation. Keywords: Judeo-Arabic, Hebrew component, Moroccan Arabic, Phonology, Morphology, Fez Introduction The Judeo-Arabic dialects are Arabic dialects that have been spoken by Jews. Thus, they are both to be considered as Arabic dialects and Jewish languages. And like all Jewish Languages, they borrowed numerous terms from Hebrew and Aramaic. There is already to this day, many studies focusing on those dialects and on the question of the interactions with the Hebrew language1. This study intends to offer a systematic analysis of how the Hebrew component of Judeo-Arabic in Fez behaves in terms of phonological and phonetical adaptation and to look in particular for regularities in that adaptation. To the present day, many authors have focused on specific aspects of this Hebrew component: Bahat (2001) provides a dictionary listing of Hebrew lexemes in use in Moroccan Jewish written Arabic. Leslau (1945), Maman (1999) and Sibony (2019) address the issue of relevant semantic fields while Tedghi (1995, 1999, 2001 and 2003b) deals with the internal semantic developments of Hebrew loan words. Brunot and Malka (1939, 1940), Lévy (1994, 2009) and Chetrit (2007, 2009, 2014, 2015) also refer widely to the Hebrew component in studies that are not exclusively dedicated to it. 1 See Morag 1963 for the pronunciation of Hebrew words into Yemeni Judeo-Arabic ; see Morag 2007 and Maman 2019 for the different traditons of Hebrew around the world; for Judeo-Arabic dialects see Saada 1956 for Sousse, Cohen D. 1964, 1975 for Tunis, Henshke 1991 for Djerba and 2007, 2008 for Tunisia in general, Cohen M. 1912 for Algiers, Tirosh-Becker 1989a,b for Constantine, Bar-Asher 1992 for Tlemcen and Ain Tmouchent, for Morocco (which will be more developed through the article), see Akoun 2013, 2015, Bar-Asher 1978, Heath 2002, Heath & Bar-Asher 1982, Maman 1989, 1999, Stillman 1988. 105 JONAS SIBONY Finally, in key studies, Bar-Asher (1996) and Tedghi (2003a) retrace the entire history of the research on the Hebrew component in modern Judeo-Arabic dialects. For Fez, the particular issue of phonological and phonetical adaptations in Hebrew loanwords has been researched by Leslau (1945), and for Sefrou by Stillman (2008). Lévy (2009) explores and describes a series of Moroccan Judeo-Arabic dialects and touches on phonetical questions for Fez, Meknes, Rabat-Salé, Sefrou, Debdou, the areas of Tafilalt, Midelt, and Draa, Marrakesh, Essaouira, Safi, El-Jdida and Azemmour. Though Lévy makes crucial remarks about how the Hebrew component behaves, he does not draw any specific conclusions from these. This article will bring together these previous studies focusing on the Judeo-Moroccan dialect of Fez (henceforth FJA) in order to propose a general hypothesis concerning the phonetical adaptation process. In order to open up a comparative perspective, the Fasi (from Fez) data will be complemented with examples from other Moroccan cities, taken from Lévy (2009) but also from Chetrit (2007, 2014, 2015) for Meknes, Stillman (1988) for Sefrou, Pellat (1952) for Debdou, and Caubet (1993) for “standard” Moroccan Arabic (henceforth MA). Moroccan Judeo-Arabic dialects The most important and up to date studies concerning the diversity of North African JudeoArabic dialects have been undertaken by Joseph Chetrit. For the present article, I will focus on four of them (2007, 2009, 2014, 2015). According to Chetrit, around the 1950s, “there were about four hundred Jewish communities in North Africa (urban and semi-urban), most of them in Morocco for which Judeo-Arabic was the principal language”2. Dialectologists distinguish mainly between two kinds of North African Arabic dialects: Pre-Hilalian (ancient urban dialects and among them, all Judeo-Arabic dialects3), dating back to the first wave of Arabization of North-Africa, and Hilalian dialects4, from the wave of Arabization that accompanied the arrival of the confederation of tribes of the Banu Hilal in the eleventh century5. Chetrit distinguishes four groups of Judeo-Arabic dialects (all pre-Hilalian) with different backgrounds and which share many characteristics. He calls them Eqa:l, Wqal, kjal and ˀal dialects 6 and shows how each group share specific phonological and morphological features. 2 3 4 5 6 Chetrit 2014: 202. See for example AGUADÉ, Jorge 2018. “The Maghrebi dialects of Arabic”. Arabic Historical Dialectology, Linguistic and Sociolinguistic Approaches. Oxford University Press: 39-63. In fact, this distinction has been lately reconsidered by many scholars, and among them Alexander Magidow. 2013. Towards a Sociohistorical Reconstruction of Pre-Islamic Arabic Dialect Diversity. BA, MA Dissertation. University of Texas, Austin: Chapter 4, History, North Africa: 233-250. See Holes, Clive (eds.). 2018. Arabic Historical Dialectology. Clive Holes (eds.). Oxford University Press: 347. Chetrit 2015. Those names refer to the specific pronunciation in each of them of the verb “to say”, that carries in it two unstable features of the North-African Arabic dialects: pronunciation of phoneme /q/ and of the long vowels. In other words, the pronunciation of this single word tells a lot about the nature of the dialect. 106 PHONETICAL AND PHONOLOGICAL ADAPTATIONS OF HEBREW LOANWORDS IN MOROCCAN JUDEO-ARABIC: THE CASE OF FEZ Chetrit sums up his investigation7 as follows : the first group Eqa:l is very ancient (formed between the ninth and fifteenth century) and was in use in the East (Libya, Tunisia, Annaba, Algiers); the Wqal group concerns the western Moroccan dialects, it is very old as well (tenth-thirteenth century); the third group kjal may be the oldest (eight-eleventh century) for eastern Morocco; and finally the ˀal group shaped between the sixteenth and the eighteenth century in urban parts in northern Morocco. Judeo-Arabic from Fez, the main subject of this article, belongs to the ˀal group. The ˀal dialects and FJA FJA belongs to the “interior urban belt” dialects 8 . William Marçais, in his Comment l’Afrique du Nord a été arabisée? talks about an “old urban vernacular” 9 . In Chetrit’s classification, it is a ˀal dialect. This group includes the varieties of Salé, Rabat, Fez, Meknes, Sefrou, Taza and Wezzane, and are all dialects of Iberian influence. Those dialects carry very ancient features but were widely reshaped in the sixteenth and seventeenth century after important groups of Megorashim10 settled in the region. In terms of language contact, those dialects arose from the meeting of Judeo-Spanish, Andalusi Arabic and ancient Moroccan Judeo-Arabic dialects11. They show some phonetical and phonological specificities such as: glottalization of /q/ > [ʔ], neutralization of sibilants /š/-/s/ > /s/ and /ž/-/z/ > /z/12, affrication of the dental consonant /t/ > [ts] and a significant attenuation trend of /h/13. If those varieties share many linguistic features, they are nonetheless distinct dialects that each evolved in different contexts, places and in contact with different Muslim dialects. For example, FJA was in constant contact with the “Old Fasi” dialect, the ancient dialect of Ahl Fas, described by Hilili14. The Judeo-Arabic dialect of Sefrou, a ˀal dialect also behaves in a similar way to FJA but has developed internal specificities as well15, due to its contact with Berber speaking populations, with Jebli and ‘Old Montagnard’ dialects16 and interferences from Judeo-Arabic dialects from the Tafilalt and Debdou (both kjal dialects)17 7 8 9 10 11 12 13 14 15 16 17 Chetrit 2015 (overview table p. 34). Heath 2002: 21-24. Marçais William. 1938. “Comment l’Afrique du Nord a été arabisée ? I. L’arabisation des villes”. Annales de l’Institut des Etudes Orientales 4: 1-21. Megorashim means “expelled” in Hebrew and is the term used to designate the Jewish population that were expelled from Spain in the fiftennth century before they established to Northern Morocco. In opposition, the ancient local Jewish population is called Toshavim “natives, residents”. Chetrit 2007: 163-164 ; 2015: 29-31. Like almost every other Moroccan Judeo-Arabic dialects. Chetrit 2015: 32. Hilili, A. 1986. “Esquisse de l’arabe parlé dans la medina de Fès”. Revue de la Faculté des Lettres et des Sciences Humaines de Fès 8: 67-77. Stillman 2011: 194. Stillman 1988: 31. Chetrit 2015: 32 . 107 JONAS SIBONY Which Hebrew and how? The Hebrew component is the most common feature of Jewish languages. Jewish communities are in constant contact with Hebrew (and Aramaic) sacred sources such as the Hebrew Bible, the Talmud and the Midrash 18 as well as with Hebrew literatures from different times and regions19. In other words, Jewish life is bathed in Hebrew culture. Jews use non-Hebrew vernacular languages and insert Hebrew vocabulary into it. In North Africa, Jews speak many kinds of specific sociolects of Maghrebi Arabic, with a Hebrew lexical component. Although these loanwords have rather been historically considered as concerning religion or aspects of Jewish life only, recent studies have shown they refer to other semantic fields as well20. The interaction between Hebrew and Judeo-Arabic is not usual language contact because there has never been any bilingualism, rather it is a situation of diglossia resulting in continued influence of written Hebrew texts into the spoken (judeo-)Arabic dialect. This led to different levels of word use, depending mainly on the speaker’s degree of education. Bar-Asher mentions four groups of speakers that behave differently towards Hebrew loanwords. Group one concerns “wise men (Hakhamim) and their students”; group two “men who know how to write and read Hebrew”; group three “men who don’t know Hebrew” and group four “women”21. The more the speaker has contact with Hebrew, the more he uses a Hebrew lexicon in his Arabic dialect. In this study, I focus on fully integrated Hebrew vocabulary, used by every speaker and in which the etymology is often known to all but also sometimes unknown. I will not discuss the Hebrew of the first group since the most educated population can theoretically use any Hebrew word in a personal capacity and therefore this lexicon and its pronunciation (often “corrected”) are not representative of what happens to the usual spoken language. Integrated Hebrew vocabulary should be considered as the category proposed by M. Weinreich of merged Hebrew, in opposition to whole Hebrew22, i.e. not Hebrew in general, but a fully integrated Hebrew component to the vernacular language. Merged Hebrew is sometimes very ancient and even if it is difficult to identify precisely the period of each loan word, aspects of phonology and morphology can offer precious clues. Many scholars have mentioned the fact that in dialects where the Arabic /q/ (‫ )ق‬is produced /ˀ/, Hebrew follows with ʷ /q/ produced /ˀ/23. How did the pronunciation shift occur? There are two ways to understand that phenomenon. The first is to consider that phonetical evolutions are led by the “wise men”. That seems unlikely. The group designated as “wise men” can indeed identify etymological matches between the two languages and partly lead the pronunciation of traditional reading but they can hardly control the natural evolution of pronunciation in everyday speech. If we nevertheless consider this possibility, we would have to assume that some speakers identified both 18 19 20 21 22 23 Chetrit 2014: 206. Namely Iraqi and Andalusi medieval literatures : poems, grammars and philosophical essays. Tedghi 2003b: 689. Bar-Asher 1978: 166-167 Tedghj 2003b: 690 mentions that category from Weinreich, M. 1954. “Pre- and Early History of Yiddish: Facts and Conceptual Frameworks”. U. Weinreich (eds.). The Fields of Yiddish. New York: 89-96. Akoun 2015: 35; Bar-Asher 1978: 172; Morag 2007: 558; Tirosh Beker 1989: 332. 108 PHONETICAL AND PHONOLOGICAL ADAPTATIONS OF HEBREW LOANWORDS IN MOROCCAN JUDEO-ARABIC: THE CASE OF FEZ historical phonemes and consciously adapted the Arabic evolution of pronunciation to the Hebrew. The second way to address this situation is to assume that these were very ancient loans, dating back to when both the Arabic /q/ was produced /q/ as well as ʷ in the Hebrew reading tradition and that when the Arabic shift /q/ > /ˀ/ occurred, the Hebrew component was already fully integrated and thus followed precisely the phonetic evolution as any lexical element would have. In this manner, the historical nature of the Hebrew component can be seen by its articulation and imbrication with the dialect’s phonological and morpho-syntactical systems. A lexeme of Hebrew origin integrates the FJA linguistic mechanism just as any other word would, whether it comes from Classical Arabic, Berber, Spanish or French. Hebrew nouns go with the Arabic article: l-pṛāṣa “the Parasha” (Hebrew pārāšâ ʤ ʕˇʸʕʕ ˝), lˁasirīm “the rich people” (Hebrew ˁăšīrīm ʭ  ʩʸʩʑ ˇʏʑ ˆ), they can have Arabic plurals as ṣḍāḍə̣ṛ24(Arabic broken plural of Hebrew siddur ʸ˒ːʱ ʑ “prayer book”), external plurals as in mămzirēn and səkkāţ25 (FJA plurals of Hebrew mamzēr ʸʒʦʮʍ ʔʮ “bastard” and sukkâ ʤ  ˗ʕ ʱʗ “Sukkah”). Phonological and phonetical adaptations of the Hebrew consonants in FJA The Hebrew lexicon of interest for this study concerns a borrowed lexicon that is fully integrated into the Judeo-Arabic dialect, and available in daily speech, just like loanwords from other languages such as Spanish or French. Due to the fact that those words have mainly been integrated into speech from written sources (in contrast to the Romance loanwords), I will list the Hebrew letters of the alphabet one by one, and see how they were adapted in borrowed lexicon. The written Hebrew sources are most often codified according to the Masoretic tradition. For the purposes of this study it is crucial to remember that there is not nor has there ever been a situation of bilingualism between Hebrew and FJA. What happened here is the extraction of written forms of an ancient language system and adaptation into a living language system. That is to say that ancient Hebrew phonemes did not influence directly the pronunciation of those loans. And even if some aspects of those ancient phonemes have probably been retained in the reading traditions, the effective pronunciations of the loans in FJA are to be considered as interpretations according to what is available in everyday speech, namely, the FJA phonemes. The FJA phonemes have been described, among others, by Leslau (1945) and Lévy (2009). The transcription used here is the one developed in Levy. Here follows the exhaustive list: Labials and labiodentals: /b/, /f/, /m/, /p/, /v/; dentals / alveolars: /d/, /ḍ/, /t/, /ṭ/, /n/; liquids: /l/, /r/; sibilants26: /s/, /ṣ/, /z/; velars: /k/, /g/, /ḫ/, /ġ/; pharyngeals: /ḥ/, /ˁ/; laryngeals: /h/, /ˀ/; 24 25 26 Tedghi 1995: 45. Leslau 1945: 69. In most North African Judeo-Arabic dialects, opposition has been neutralized between /s/ and /š/ which merged to /s/ and between /z/ and /ž/ which merged to /z/. The two remaining phonemes /s/ and /z/ are produced with various degrees of stridency (Chetrit 2015: 6), depending on the dialect and on the speaker. This phenomenon is not absolutely specific to Jewish speech and can be observed in other “uncorrected 109 JONAS SIBONY semi-vowels: /w/, /y/; vowels 27 : /ā/, /ō/, /ī/, /ə/. In terms of phonetics, the available productions are: [b], [bˁ], [f], [m], [mˁ], [p], [v], [d], [dˁ], [t], [tˁ], [n], [l], [lˁ], [r], [rˁ], [s], [sˁ], [z],[zˁ], [k], [g], [x], [ɣ], [ħ], [ʕ], [h], [ʔ], [ε], [a], [ä], [ʊ], [ö], [u], [o], [i], [Iˁ], [eˁ], [ə], [ă]28. The presented lexicon is not the result of a personal research but is extracted from previous works. The aim of the study is not to present specifically original data nor an exhaustive list of the Hebrew loanwords but to understand how those words have been integrated into FJA. Thus, the words are to a very large extent extracted from Levy (2009) and to a lesser extent from Brunot and Malka (1939, 1940), Leslau (1945), Lévy (1994, 2013), Stillman (2011) and Tedghi (1995, 1999, 2001, 2003b). The comparative data from other Judeo-Moroccan dialects is borrowed from Akoun (2015), Bar-Asher (1978, 1992), Chetrit (2007, 2009, 2014, 2015), Heath and Bar-Asher (2002), Lévy (1994, 2009, 2013), Pellat (1952) and Stillman (1988, 2011). Finally, for obvious reasons of consistency, the various transcription methods from the various authors have been standardized. ʠ  (Alef) /ˀ/29 > FJA /ˀ/-/ø/30: [ʔ]-[ø]31 Hebrew ʠ  identifies with FJA /ˀ/, a glottal stop [ʔ]. However, its production is very fragile and it is lost in many positions as in Arabic words like *ˀaḫ > ḫa “brother”, *ˀarḍ > āṛḍ / lāṛḍ “earth” (with assimilation of the definite article). It is artificially maintained (or reintroduced) through influence of written Arabic sources: ˀōṣōl “real property”. The situation regarding the Hebrew lexicon is precisely the same: ʸ ʒˢʱʍ ʓʠ, ˀestēr “Esther” > FJA isţīr, ʸ ʕʣʏʠ, ˀăḏār “Month of Adar” > FJA adār. Some Hebrew words assimilate the Arabic definite article: ʭʩʧʑ ʸˣʠ, ʍ ˀōrəḥīm “guest” often becomes löṛaḥīm and some have two possible forms ʭʩʸʑ ʕʡʩ ʒʠ, ˀēḇārīm “members” is ˀibārīm or libārīm. This situation is the same in other JA dialects in Morocco and restitution seems to be applied on a case-by-case basis. For the reflexes of Arabic *ˀarḍ “earth”, when as in Fez and Sefrou the word is produced lāṛḍ (FJA) or ləṛḍ (Old Fasi), the identified Hebrew cognate is produced leṛeṣ32 (<*ʵʸʓ ʓʠ‫أَل‬, ˀal- 27 28 29 30 31 32 speeches”, as for example in traditional Meknes Muslim speech, in Old-Fasi (Muslim) or in many women speeches as in Fez, Meknes, Tangiers and Rabat (Lévy 2009: 189). The FJA vocalic system is composed of four phonemes, each of which knows multiple productions conditioned by the environment. Lévy (2009: 208-221) describes it as follows: /ā/ is produced as an openmid front vowel, noted /ǟ/ bəzzǟf “a lot”, as a lightly velarized back vowel, noted /ā/ ḥmār “donkey” or as a “middle” a, noted /a/ imma “my mother” (IPA: [ε], [a], [ä]); /ō/ is produced as a close rounded vowel /ō/ in ṛōḥ “soul”, as a close mid-central rounded vowel /ȫ/ in pȫlis, as a close back rounded vowels /ū/ in hūwa “he” or as a close mid rounded /o/ in ṣof “wool” (IPA: [ʊ], [ö], [u], [o]); /ī/ is produced close /ī/ or /ē ̱̣/ in zdīd “new” and ṛē ̣ţ “I saw”, but as the mid-vowel /ẹ/ in rəbbẹ “rabbi” (IPA: [i], [Iˁ], [eˁ]); /ə/ is a central short vowel with neutral timbre. It can be pharyngealized /ə ̱̣/ in kbə̣ṛ or slightly colored /ă/ as in ḥăps “prison” (IPA: [ə], [ă]). List is based on the descriptions found in Lévy 2009: 163-252. Refers to the Hebrew letter and the theoretical ancien Hebrew phoneme. Refers to the FJA phoneme with whom the Hebrew letter identifies. Refers to the effective productions of the FJA phoneme. Stillman 1988: 55. 110 PHONETICAL AND PHONOLOGICAL ADAPTATIONS OF HEBREW LOANWORDS IN MOROCCAN JUDEO-ARABIC: THE CASE OF FEZ ˀereṣ). Chetrit notes that in Meknes, one of the few effective /ˀ/ for Hebrew ʠ  is actually found in that word əl-ˀeṛeṣ precisely in a dialect where /ˀ/ has been maintained (or reinserted) into the Arabic cognate: əl-ˀaṛḍ33, this match could be explained by analogy since, still in Meknes, when no cognate is identifiable: ʺ ʓʮʎʠ, ˀĕmeṯ “truth” is imiţ34. In Fez, there is at least one case of retention or restitution: in the expression ʵʸʓ ʕʠ ʕʤʭʔˆ, ˁam hā-ˀāreṣ > ˁǟm-a-ˀāṛē ̣ṣ “ignorant person”35. Finally, some /ˀ/ are preserved or reinserted into Hebrew words, probably because unlike with the Arabic words, Jewish speakers can compare them to the written forms found in the Jewish sources, thus leading to “corrections”, as in ʲ ʕˎ ʍʶ ʓʠ, ˀeṣbāˁ “finger” > ˀəṣbāˁ, ˂ ʕʠ ʍʬ ʔʮ, malˀaḵ “angel” > mālˀāḫ and ʤ ʕʠʕ˒ ʔʶ, ṣawwāˀâ “testament” > ṣāvāˀa. ˎ-ʡ (Bet) /b/-/ḇ/ > FJA /b/: [b] Hebrew ˎ and ʡ identify with FJA /b/, the voiced bilabial stop [b]. It is never spirantized36 as it is alternatively in Hebrew according to Hebrew grammar and noted in the Masoretic system37. FJA /b/ appears in Arabic words like bǟs “in order to” or bȫy “my father”. /b/ can be emphasized38 [bˁ] by contact *ṭbəl > ṭḅə̣l “tambourine”. Pharyngealized /b/ appears in some Romance borrowings as well lamḅa “lamp” and in at least one example for phonosymbolic reasons39 in ṛə̣ḅḅẹ “God”. /b/ becomes /p/ ([p]) by assimilation with a voiceless consonant *ḥăbs > ḥăps “prison”, kəppāṛ “capers” but kəbbāṛ in Muslim speech. It is found in Hebrew words ʤʸʩ ʕ ʒʡʏˆ, ˁăḇērâ > ˁabera “transgression” or ˆʔ ˒ʡ ʕʶ, ṣāḇūaˁ > sābūwāˁ 40 “hypocrite” . Hebrew adapted ʡ can be pharyngealized too as seen in at least one example of borrowed consonantal Hebrew root fully assimilated to FJA verbal morphology: √ʬʡʨ > ṭḅə̣ḷ/iṭḅə̣ḷ “to have a ritual bath”. 33 34 35 36 37 38 39 40 Chetrit 2007: 170. Chetrit 2007: 198. This irregular retention might be explained by the fact that ˁǟm-a-ˀāṛē ̣ṣ is a whole Hebrew expression and the word isn’t isolated. The word ˀāṛē ̣ṣ comes as the second part of a Hebrew construct state, it cannot absorb the Arabic article. In addition, maybe /ˀ/ is preserved as an artificial retention because Hebrew /h/ is already lost right before and because both loss leads to a too significant change, which draws attention and is being rectified. The spirantized /ḇ/ is absent from the other Moroccan Judeo-Arabic dialects as well, for example in Debdou: ʬ ʓʡ ʕˎ, bāḇel “Babylone” > bābīl and ʭ ʕʤʸʕ ʍʡ ʔʠ, ˀaḇrāhām “Abraham” > ābrāhām. According to the Hebrew grammar, the phonemes /bgdkpt/ represented by the letters ʺʿʴʫʣʢʡ, are spirantized /ḇḡḏḵp̄ṯ/ in some positions in the word. In the Masoretic system, when the consonant has a dot in it, its production is plosive and when not, it is spirantized. “Emphasis” in the context of Arabic linguistics usually refers to pharyngealization of consonants or velarization of vowels. “Emphasis” is noted by a subscripted dot: /b/ = [b] but /ḅ/ = [bˁ]. Lévy 2009: 178 speaks of emphasization due to “affective semantic reasons”. He assumes there is a correlation between the phonetical emphasization, i.e pharyngealization, and the meaningfulness of the designated subject “God”. He rightly notes that the same phenomenon is found in the other designation of God in Arabic, with the same unusual pharyngealization: əḷḷāh. Chetrit (2007: 169) speaks of “semantic emphasis due to the psychological weight of the lexeme” and quotes the additional example ḅaḅa “dad”. Lévy 2009: 192 adds that sābūwāˁ is then the pronunciation of ˆʔ ˒ʡ ʕʶ and ˆʔ ˒ʡ ʕˇ. 111 JONAS SIBONY ˏ-ʢ (Gimel) /g/-/ḡ/ > FJA /g/-/ġ/: [g]-[ɣ] Hebrew ˏ is identified with FJA /g/, a voiced velar stop. In FJA, it is usually found in Romance loanwords such as gāṛṛo “cigarette” but some occurrences come from historical Arabic /q/ *qrōn > gṛōn “horn” (probably borrowed to a Bedouin dialect), *qǟbəl > gǟbəl ّ ǧazzār > “to look after”. Other cases come from ancient dissimilations of Arabic /ž/ ‫ﺟﺰار‬ gəzzāṛ “butcher”. This background has made it an internal phoneme and might have facilitated a possible match with Hebrew Gimel as in ʩˣˏ, gōy “non-Jew” > gȫy. Unlike Hebrew ʡ, non-dotted Gimel ʢ, the spirantized allophone /ḡ/ has managed to find a match with another FJA phoneme: /ġ/, a voiced velar fricative [ɣ] (etymologically corresponding to Hebrew ʲ ). /ġ/ appears in Arabic words ġəss “he cheated”, ġȫla “ogress” and is regularly unvoiced *ġē ̣ṛ > ḫē ̣ṛ “good”. It is found in Hebrew ʭʩˇʑ ʸˣʢ ʕ ʮ, ʍ məḡōrāšīm > meġōrāsīm “expelled (from Spain)”. ː-ʣ (Dalet) /d/-/ḏ/ > FJA /d/: [d] Hebrew ː identifies with FJA /d/, a voiced dental or alveolar stop [d]. Since FJA dialects don’t have interdental consonants (merge of historical Arabic ‫ د‬/d/ and ‫ ذ‬/ð/ > /d/), Hebrewʣ (Hebrew spirantized /ḏ/) is produced [d] as well. /d/ appears in Arabic words dīb “jackal”, dǟz “to pass by” and can be pharyngealized by contact > [dˁ] as in *dār > *dāṛ > ḍāṛ “house”41. There are a few cases of contextual devoicing *dḫəl! > tḫəl! “come in!”. In Hebrew words, Hebrew ː and ʣ are produced /d/: ʸ˒ːʱ, ʑ siddur “prayer book” > səddōṛ and ˇˣʣ ʕʷ, qāḏōš “holy” > ˀādōs42. ʤ (He) /h/ > FJA /h/: [h]-[ø] Hebrew ʤ identifies with FJA /h/ which is a voiced or voiceless glottal fricative [h], found in həm “worry” or həllȫla “pilgrimage”. FJA /h/ is sometimes reduced to /ø/ as in *kəllhȫm “all of them” > kəllȫm, the same happens to Hebrew loans at least once in ʵʸʓ ʕʠ ʕʤʭʔˆ, ˁam hā-ˀāreṣ > ˁǟm-a-ˀāṛē ̣ṣ “ignorant person”. ʥ (Vav) /w/ > FJA /v/: [v] Hebrew ʥ identifies with FJA /v/, a voiced labiodental fricative [v]. In FJA non-Hebrew words, [v] appears in Romance loanwords, Spanish and French as vīlla “villa” or can correspond to an Arabic voiced /f/ *ˁənfəz > ˁənvəz “to tread” or even to a voiced 41 42 In Hebrew script, ḍāṛ is alternatively noted ʸʠʣ, dār - which corresponds to the notation in Arabic script ‫دار‬ - or ʸʠʶ,࿯ with the grapheme ʶ, ࿯ used for /ḍ/. Lévy (2009: 210) mentions the fact that in divorce certificates, “Casablanca” is often ʠʶʩʡʬʠʸʠ ࿯ ʶ,࿯ ḍār l-bīḍa (while Arabic script notes ‫اﻟﺪار اﻟﺒﯿﻀﺎء‬, al-dār al-bayḍāˀ). Same in the Tafilalt Jewish dialects ˀādȫs. 112 PHONETICAL AND PHONOLOGICAL ADAPTATIONS OF HEBREW LOANWORDS IN MOROCCAN JUDEO-ARABIC: THE CASE OF FEZ preposition f- “in” *f-əḍḍāṛ > v-əḍḍāṛ “at home”. Lévy43 suggests that the [v] production of ancient ʥ is due to a Megorashim influence since it is produced [v] in Fez, Meknes, Rabat and Sefrou, all dialects of Spanish influence, but produced [w] everywhere else. In Fez, indeed, ʯˣʥʕˆ, ˁāwwōn is ˁavȫn “sin” and ʸ ʩ ʑʥʏʠ, ˀăwīr is həvvīr “air”. Chetrit44 adds that the [v] pronunciation also concerns the Andalusi production of Hebrew ʥ, which strengthens the Iberian hypothesis. Historically, and as it is still the case in the southern Judeo-Arabic dialects, Hebrew ʥ was identified to Arabic /w/, a voiced labio-velar approximant [w], still effective in FJA Arabic words such as ˁǟwd “horse”. In the south, ʥ has retained the [w] production45. Back in Fez, a few cases of Hebrew ʥ > FJA /b/46 are observed where plural of ˁāwwōn becomes ˁābōnōţ in the formula bāˁābōnōţ, the name ʩ ʑʥʒʬ, lēwī “Levy” is lē bi, ̣ lībẹ or lē bẹ ̣ 47 and ʣʩ ʑʥ ʕː, dāwīḏ “David” is dābid48. It could result once more from a Spanish influence coming from a confusion /v/-/w/-/b/, attested for instance in the Andalusi JudeoArabic version of “David”: dabi49. ʦ (Zayin) /z/ > FJA /z/: [z] (with variant degrees of stridency) Hebrew ʦ identifies with FJA /z/, a voiced alveolar fricative [z] which results from the neutralization of the opposition of Arabic ‫ ز‬/z/ and ‫ ج‬/ž/. The production of FJA /z/ is to be found somewhere between [z] and [ʒ] (voiced post-alveolar fricative), but usually, closer to [z]: zōz “two”, zdǟda “chicken”. /z/ can be unvoiced *zḥǟm > sḥǟm “crowd” or emphasized [zˁ] by contact (*rāžəl >) *rāzəl > ṛāẓəl “man”. /z/ in Hebrew loanwords corresponds to the same articulation and shows the same stridency. It is found in words like ʸʒʦʮʍ ʔʮ, mamzēr “bastard”> mămzīr or ʤʕʦ˒ʦʮ, ʍ məzūzâ “Mezuzah” > mzȫza. ʧ (Chet) /ḥ/ > FJA /ḥ/: [ħ] Hebrew ʧ identifies with FJA /ḥ/, an unvoiced pharyngeal fricative [ħ], seen in ḥȫţ “fish” or dəllǟḥa “watermelon”. It is found in Hebrew loans like ʤʸʕ ʡʍ ʓʧ, ḥeḇrâ > ḥəbṛa / ḥē ̣bṛa “brotherhood”. ʨ (Tet) /ṭ/ > FJA /ṭ/: [tˁ] 43 44 45 46 47 48 49 Lévy 2009: 182. Chetrit 2015: 7 In Tafilalt, ʯˣʥʕˆ, ˁāwwōn is ˁāwōn and ʸʩ ʑʥʏʠ, ˀăwīr is lāwīr, in Marrakesh and El Jadida lāwwīr (but in Meknes laṿṿeṛ). The same occasional development happens in Arabic words as well: *wōṣ(t)-əḍḍār becomes bōṣəḍḍār “patio” and wōh! > bōh! “misfortune”. In Midelt and in the Draa area, ʸʩ ʑʥʏʠ is həbbi and in the Tafilalt ʭʩ ʑʥʕʰˆʕ , ˁānāwīm “modests” is ˁanabim. In Marrakesh lē wi, ̣ in Azemmour līwī. In Marrakesh dāwid. Mentioned in Lévy 2009: 180, from Corriente, Federico. 1977. A Grammatical Sketch of the Spanish Arabic Dialect Bundle. Madrid. 113 JONAS SIBONY Hebrew ʨ identifies with FJA /ṭ/, a pharyngealized voiceless alveolar stop [tˁ] as seen in ṭḷēba “request” or ṭāṣāţ “cups”. Hebrew ʨ appears inʤʸʩ ʕ ʨʑ ʍ˝, pəṭīrâ > pē ̣ṭē ̣ṛa “funeral” and ʡˣʨ, ṭōḇ > ṭōb “good”. ʩ (Yod) /y/ > FJA /y/: [j]-[i] Hebrew ʩ identifies with FJA “semi-vowel” /y/, a voiced palatal approximant [j]. If FJA /y/ is stable in middle and final position as in zǟy “coming”, it usually turns to /ī/ when at the beginning of the word: *yədd “hand” (as in Old-Fasi) > īdd. As intended, same happens to Hebrew loans: ʯ˒ʸ ʗˇ ʍʩ yəšurūn “Jew” > īṣoṛon and ʤ ʕʡʩˇʑ ʍʩ, yəšīḇâ “Yeshiva” > līsība50 with agglutination of the definite article (definite form is l-līsība) as in other nouns beginning with a vowel (after loss of /ˀ/ or /y/): ‫أرض‬, ˀarḍ > lāṛḍ and l-lāṛḍ “the earth”. ˗-ʫ (Kaf) /k/-/ḵ/ > FJA /k/-/ḫ/: [k]-[x] Since both [k] and [x] sounds are available in FJA, Hebrew ˗ and ʫ (/k/ and its spirantized allophonic variant /ḵ/) find easy phonetic matches. However, in FJA, [k] and [x] are reflexes of the distinct phonemes /k/ and /ḫ/. /k/ is thus a voiceless velar stop [k]51 seen in kǟn “he was” or kəllȫm “all of them” and can be voiced by contact *kdəb > gdəb “he lied”. It appears in the Hebrew loan words ʤʕʰ˗ʕ ʔʱ, sakkānâ “danger” > səkkāna, and ʤʸʕʕ ˝˗ʔ , kappārâ > kəppāṛa / kăppāṛa “atonement”, both reshaped into the CəCCāCa FJA stem52. FJA /ḫ/ is a voiceless velar fricative [x] as seen in ḫəbz “bread” or ḫāf “he was scared”. /ḫ/ can be voiced by contact as in *ḫzər > ġzər “he looked”. Since Hebrew ʫ can’t be at the beginning of a word (according to the Hebrew spirantisation rule), FJA /ḫ/ is always in the middle of the word or in final position in Hebrew borrowings53: ʭʕʫ ʕʧ, ḥāḵām > ḥāḫām “wise” and ˂ ʕʠ ʍʬ ʔʮ, malˀāḵ > mālˀāḫ “angel”. 50 51 52 53 lē ̣sība in Azemmour. In the Tafilalt, Arabic /k/ is often produced /t/-/ţ/ as in *kbīr “big” > ţbīr and kəlb “dog” > təlb. It should logically be the same for Hebrew loans but I couldn’t find any example (Heath and Bar-Asher 1982 don’t give any). Although Lévy (2009: 329) mentions ʸ ʕˎ ʍʫʔˆ, ˁaḵbār > ˁəţbār. However, this case is suspicious since it would imply an original pronunciation ˁakbār without spirantization, which is found nowhere in Morocco and since ˗ and ʫ identify to distinct phonemes, the other possibility would imply an adaptation /x/ > /ţ/ which is not attested anywhere either. Moreover, Lévy 2009 indicates p. 329 that ˁəţbār is a Hebrew loan meaning “boring matter”, but says p. 290 that ˁăţbāṛ means “mouse” in Sefrou and is borrowed from Tafilalt Judeo-Arabic… See FJA Arabic words dəllǟḥa “watermelon”, ṭəˀˀāya “pitcher” or nəzzāṛa “woodworkers”. There is at least one exception, that could origin from some kind of hypercorrection l-xalab “the dog” identified as coming from Hebrew ʡʓʬ˗ʓ , keleḇ. In Chetrit’s book of proverbs (2015. Paroles Affables. Matanel Fondation), is mentioned the proverb from Taroudant äs iˁaṛf əl-xäläb l-əţ-ţsis?, ”what does the dog understand to the flour?”, where xäläb is used as a common word for “dog” and yet Chetrit gives as well a proverb where “dog” is said with the usual Arabic word kəlb: mn iddnä kəl l-kəlb “the dog ate from our hands”. Brunot (1940) and Leslau (1945) mention the expression fḥäl l-xalab “like a dog” in use in Fez alongside the usual kəlb once more. It is quite difficult to explain its origin. For Leslau (1945: 76) and Brunot and Malka (1940: 43), production xalab is to be explained by the fact that the Jewish speakers 114 PHONETICAL AND PHONOLOGICAL ADAPTATIONS OF HEBREW LOANWORDS IN MOROCCAN JUDEO-ARABIC: THE CASE OF FEZ ʬ (Lamed) /l/ > FJA /l/: [l] Hebrew ʬ identifies with FJA /l/, an alveolar lateral approximant [l] seen in līla “night” and lǟˀ! “no!”. /l/ can move to /n/ by dissimilation: səlsla > sənsla “chain”. It can be pharyngealized [lˁ] by assimilation *ṣla > ṣḷa “synagogue”. Pharyngealized /l/ is common in Romance borrowings sāḷḷē ̣ṛo “salt shaker” or appears as intensification for phonosymbolic reasons əḷḷāh “God”. /l/ is thus the match for Hebrew ʬ as seen in ʡʕʬ˒ʬ, lūlāḇ > lōlāb “Lulav” and ʬ ʒʤˣʮ, mōhēl > mȫhəl “Mohel”. ʮ (Mem) /m/ > FJA /m/: [m] Hebrew ʮ identifies with FJA /m/, a bilabial nasal [m], as seen in mə̣ṛṛa “once”, məsˁȫd “Messaoud”. /m/ is usually pharyngealized [mˁ] in the neighboring of an emphatic consonant as in *ṣəmṭa > ṣəṃṭa “belt”. It is regularly produced /m/ in Hebrew loans too: ʸʒʦʮʍ ʔʮ, mamzēr > mămzīr “bastard”, ʭˣʬ ʕˇ, šālōm > sǟlȫm. However, Lévy notes at least one case of dissimilation ʮ > n, in the Hebrew masculine name sənṭōb < ʡˣʨʭ ʒˇ, šēm ṭōḇ54. Such a shift is not very common in FJA but appears here and there as in *miftāḥ > nəftāḥ “key”55. ʰ (Nun) /n/ > FJA /n/: [n] Hebrew ʰ identifies with FJA /n/, an alveolar nasal [n], found in nhāṛ “day” and nţē ̣n “you” and can sometimes move to /l/ by dissimilation: ḥsən “good, better” when isolated but ḥsəl mən “better than”. /n/ appears in Hebrew loans as ʯ ʕʱʩ ʑʰ, nīsān > nīsān “Nisan” or ʤʕʰ ʕ˟ ʔˢ, taqqānâ > ţəˀˀāna “ordinance”. ʱ (Samech) /s/ > FJA /s/: [s] with variant degrees of stridency Hebrew Samekh identifies with FJA /s/ which is produced close to a voiceless alveolar fricative [s]. Actually, as in many Judeo-Arabic dialects, /s/ is the result of a merger of /s/ 54 55 identify the correspondence k > x as charecteristic of the Hebrew word. Actually, it could also correspond to a real reading of Hebrew ʡʬʫʬ although the vowels of l-xalab don’t seem to match. It could also imply a truncated Biblical quotation, maybe from Joshua 14:13 lə-ḵālēḇ “to Caleb”? or Ecclesiastes 9-4 lə-ḵeleḇ “to a dog”? In addition, in the Tafilalt, there is a case of /k/ > /x/ without any apparent Hebrew interference in kəzzāb > xəzzāb “liar”. Lévy 2009: 181. Bar-Asher 1978: 168 mentions the case of ʬ˒ˎ ʔʮ, mabbūl “flood”, produced mbbul or məbbul by educated men but nəbbul by women and uneducated men in at least in the Tafilalt, Marrakesh and Salé. He does not specify in the article which Judeo-Moroccan dialect is referred to but I had the chance to ask him personally for more details. Nəbbul is mentioned as well in his study on Aïn-Temouchent and Tlemcen Algerian Jewish dialects of Arabic (1992: 176). Finally, in Azemmour the name ʭʩ ʑ˕ ʔʧ, ḥayyīm is produced ḥiyyəm or ḥiyyȫn. 115 JONAS SIBONY and /š/ and shows a degree of stridency that varies between [s] and to the palato-alveolar [ʃ], although closer to [s]. It is found in words like (*šēḫ >) sē ̣ḫ “chief” and skəţ! “shutup!”. It is voiced at least once in *ˁǟtrūs > ˁətrȫz “goat”. /s/ represents Hebrew Samekh in words like ʱʒʰ, nēs “miracle” > nīs or ʸʒʴˣʱ, sōp̄ ēr “scribe” > sȫfər. ʲ (Ayin) /ˁ/ > FJA /ˁ/: [ʕ] Hebrew ʲ identifies with FJA /ˁ/, a voiced pharyngeal fricative [ʕ] seen in words like ˁǟwd “again”, ˁrōṣa “bride”. /ˁ/ represents Hebrew ʲ in ʭʕʬˣʲ ˁōlām > ˁȫlǟm “world” or else ʸʩˈʕ ʑ ˆ, ˁāśīr > ˁasīr “rich”. ˝-ʴ (Pe) /p/-/p̄/ > FJA /p/-/f/: [p]-[f] Hebrew ˝ is identified to FJA /p/, a voiceless bilabial stop [p], mainly observed in Romance loanwords pȫlis “police”. In fact, historical Arabic does not have a [p] sound and the main reflex of the same historical Semitic phoneme corresponding to the Hebrew Pe is ‫ف‬, /f/ Fa56. The persistence of the production /p/ is quite enigmatic since it is produced [p], at least by educated men, in every region of Morocco, with a few exceptions [f] or [b], regardless of the degree of Romance influence57. And the same can be said for many JudeoArabic dialects, inside and outside of Morocco: except from Tunisia, Libya and Egypt where ʴ > /b/ is common58, /p/ is very stable across the Arab speaking world59 while /p/ is absent from standard Arabic phonology. However, in many of those places, women pronounce /b/. In Fez, Hebrew ˝ is produced /p/ as in ʧ ʔʱʓ˝, pesaḥ > pisǟḥ “Passover” and ʤ ʕˇʸʕʕ ˝, pārāšâ > pṛāṣa “Parasha” and ʴ (/p̄/) identifies with /f/, a voiceless labiodental fricative [f] found in FJA Arabic words like fḥǟl “as” or fōṭa “towel”. For Hebrew words, see ʯʓʴˏʓ , gep̄ en “grapevine” > gīfən “wine blessing”, ʸʓʴ ʒʱ, sēp̄ er > sīfer60. ʶ (Tsade) /ṣ/ > FJA /ṣ/ > [sˁ]-[s] 56 57 58 59 60 See for example Hebrew ʬʩ ʑ˝, pīl but Arabic ‫ﻓﯿﻞ‬, fīl “elephant”. With the opposite idée, Akoun 2015: 47 mentions that according to Bar-Asher, the consonant remained produced as [p] in communities that have been influenced by the Sefardic pronunciation. Djerba and Gabès produce bišáḥ “Pessa’h” (Maman 2019: 604), Tunis says pésaḥ and bišáḥ (Maman 2019: 604) and Tripoli says bísḥak “your Pessa’h” (Maman 2019: 604) and būrīm “Purim” (Yoda 2018: 94). For Egypt, see Khan, Geoffrey. 2018. “Judaeo-Arabic”. Arabic Historical Dialectology. Clive Holes (eds.). Oxford University Press: 166. See Heath and Bar-Asher (1982), Lévy (2009), Maman (2019) and Yoda (2018). For comparison, ʧ ʔʱʓ˝, pesaḥ is pisǟḥ in Meknes, pisäḥ in Taroudant and Inezgan (Essaoui 1990), and pisāḥ in the Tafilalt. For ʭʩʸ˒˝, ʑ pūrīm, Marrakesh has pȫrim. The data for Tafilalt (Lévy 2009: 311-331 and Heath and Bar-Asher 1982) shows it is the only place in Morocco where ʴ has various reflexes /p/-/f/-/b/: ʤ ʕˇʸʕʕ ˝, pārāšâ > pṛāṣa but ʤʸʩʕ ʨʑ ʍ˝, pəṭīrâ > fṭē ṛa ̣ “funeral” (Fez has pē ṭe ̣ ̄ ṛa), ̣ ʤ ʕʱʕʰʸʔʍ ˝ parnāsâ > fərnāsa “salary”. Finally, Heath and Bar-Asher (1982: 44) indicate that women pronounce /b/, without giving any example but Akoun (2015: 46) mentions that women in Tafilalt do pronounce burim and bisaḥ. 116 PHONETICAL AND PHONOLOGICAL ADAPTATIONS OF HEBREW LOANWORDS IN MOROCCAN JUDEO-ARABIC: THE CASE OF FEZ Hebrew ʶ identifies with FJA /ṣ/, a voiceless alveolar sibilant [sˁ] found in ṣḷa “synagogue”, ṣōf “wool” and thus in Hebrew loanwords like ʺ  ʩ ʑʶʩ ʑʶ, ṣīṣīṯ “Tzizit” > ṣē ̣ṣẹţ. Some cases of voicing can be observed *ṣġē ̣ṛ > zġē ̣ṛ “small”. However, the distinction between /s/ and /ṣ/ is not always very clear in FJA. Some occurrences of /ṣ/ are described as semi-emphatics by Bruno and Malka 61 , that is, a minor intensity in terms of pharyngealization. This situation leads to confusion and explains ambiguous doublets like mȫsẹ / moṣṣẹ “Moses” or /s/-/ṣ/ inversions as in *sə̣ṛba (< šə̣ṛba) > ṣə̣ṛba “sirop” or else in Hebrew loans like ˆʔ ˒ʡ ʕʶ, ṣāḇūaˁ > sābūwāˁ “hypocrite”. ʷ (Qof) /q/ > (*/q/ >) FJA /ˀ/: [ʔ] Hebrew ʷ identifies with the historical Arabic ‫ ق‬/q/, which reflex is /ˀ/ in FJA; a glottal stop [ʔ]. This is actually the main reflex of /q/ in many urban dialects across the Arabic speaking world62. For Arabic words */q/ > /ˀ/ is regular: *qom! > ˀom! “stand up!”, *qṛaˀ > ˀṛa “he read”, same for Hebrew loans: ʬ ʕʤ ʕʷ, qāhāl > *qǟhǟl > ˀǟhǟl “community”, ʸ ʕ˟ ʑˆ, ˁiqqār > ˁˀāṛ or else ʤʕʰ ʕ˟ ʔˢ, taqqānâ > ţəˀˀāna “ordinance”63. ʸ (Resh) /r/ > FJA /r/: [r]-[rˁ] Hebrew ʸ identifies with FJA /r/, an alveolar trill [r] found in ryǟl “five cents” and often pharyngealized [rˁ] as in ṛōḥ “soul”. /ṛ/ is a non-completely phonemic counterpart of /r/ and their alternation is complex. The presence of /ṛ/ is sometimes obviously allophonic and conditioned by the proximity of another pharyngealized consonant. Some other cases of conditioning could result from the influence of an adjacent vowel height as in kbə̣ṛ “taller, bigger” compared to kbīr “tall, big”. Finally, even if the distinction is not fully phonemic, the alternation /r/-/ṛ/ might be used in some cases as a distinctive mark to avoid semantic ambiguity: rəbbẹ “rabbi” / ṛə̣ḅḅẹ “God”; kəbbər “to raise”, kə̣bbə̣ṛ “to offer”; bərd “it cooled” / bə̣ṛḍ “he sanded down”64. The Hebrew loans with an original ʸ share the same characteristics: ʸʩˇʕ ʑ ˆ, ˁāšīr “rich” is produced ˁasīr, but the presence of a pharyngealized 61 62 63 64 Brunot and Malka 1939: VIII. In Morocco, */q/ > /ˀ/ is found in Old Fasi (Muslim), Jewish Fez, Meknes, Sefrou, Rabat, Ouezzane (Lévy 2009: 196). For more about reflexes of /q/ in Arabic dialects, Cf. for ex. Behnstedt, Peter and Woidich, Manfred. 2013. “Dialectology”. The Oxford Handbook of Arabic Linguistics. Owens, Jonathan (eds.). Oxford University Press. In Sefrou, where the evolution of Arabic /q/ is /ˀ/ as well, the name ʷ ʕʧ ʍʶ ʑʩ, yiṣḥāq “Isaac” is yiṣḥaˀ in proverbs but rather ṣḥaˀ in everyday use (Stillman 1988: 57). The verbal root √BDQ is borrowed in many JudeoArabic dialects from the use of the Hebrew expression ʵ ʒʮ ʕʧʺ ʔʷʩʣʑ ʍˎ, bəḏīqaṯ ḥāmēṣ “to search for leaven bread”, which gave birth to a FJA verb, among other dialects in Fez and Sefrou bdə̣ˀ/ibdə̣ˀ (<*bdəq): “to do a major clean up”. In Midelt and the Draa area, Arabic /q/ is produced /k/ thus in Midelt, Arabic bəlḥāq “actually” is bəlḥāk, Hebrew ʬ ʕʤ ʕʷ, qāhāl is kǟhǟl and ʷ ʕʧ ʍʶ ʑʩ, yiṣḥāq is isḥāk. In the Draa area, Arabic *qədra “cauldron” is kədra and Hebrew ʤ ʓʷʏˆ ʔʮ, maˁăqê “ballister” is māˁākẹ. In Marrakesh, where Arabic */q/ is /q/, Hebrew ʷ is /q/: ʤʸ˒ʡ ʕ ʍʷ qəḇūrâ > qē bōṛa ̣ “grave”. Lévy 2009: 189, 220. 117 JONAS SIBONY consonant or some specific vowels (mainly /ā/) make Hebrew /r/ behave just like FJA /r/ in the same conditions: ʤ ʸʕ ʕʶ, ṣārâ “trouble” becomes ṣāṛa and ʤ ʸʕ ˢʕ ʔʤ, hattārâ “annulling” is hătţāṛa. ˇ (Shin) /š/ > (*/š/>) FJA /s/: [s] with variant degrees of stridency Hebrew ˇ identifies with FJA /s/, produced close to a voiceless alveolar fricative [s]. It might have identified, as a first step to a proto-FJA phoneme */š/, which merged later with /s/. As a result, both etymological Arabic /š/ and Hebrew /š/ are produced /s/ with a degree of stridency that varies between [s] and [ʃ]. It is found in words like (*šēḫ >) sē ̣ḫ “chief” and skəţ! “shut-up!”. /s/ represents Hebrew ˇ  in words like ʭˣʬ ʕˇ, šālōm > sǟlōm “peace, shalom” or ˆʔ ˒ʡ ʕˇ, šāḇūaˁ “week” > sābūwāˁ 65. ˈ (Sin) /ś/ > FJA /s/: [s] with variant degrees of stridency Hebrew ˈ identifies with FJA /s/ which is produced close to a voiceless alveolar fricative [s], and as seen before, represents Arabic */s/ ‫ س‬and /š/ ‫ ش‬and Hebrew /s/ ʱ, /š/ ˇ  and /ś/ ˈ. Cases for Hebrew ˈ are seen in ʤ ʕʧʮʍ ˈ, ʑ śimḥâ, “joy” > simḥa or ʤʸʕ ʸʕ ˈ, ʍ śərārâ “rulership” > sē ̣ṛāṛa66. ˢ-ʺ (Tav) /t/-/ṯ/ > FJA /t/: [t]-[t͡s] As virtually every North-African Judeo-Arabic dialect67, FJA has lost interdentals. Arabic /t/ (‫ )ت‬and /ṯ/ (‫ )ث‬have merged into /t/, a voiceless alveolar stop [t] which is however produced as a sibilant affricate /ţ/ ([t͡s]) in almost every position in the word except before /l/, /r/ and /n/ and in some pharyngealized environments68: ţəḥţ “under” but tlǟţa “three”. Hebrew ˢ and ʺ identify with FJA /t/. As a second step, it follows the contextual allophonic 65 66 67 68 The phenomenon is common to most Moroccan Judeo-Arabic dialects, except from the dialect of Debdou where /z/-/ž/ and /s/-/š/ didn’t merge and are perfectly distributed as seen in šrāb “wine” but sāˁa “hour” (Pellat 1952: 138) , and of course, the same can be said about the Hebrew loanwords ʺˣʸ ʕʨˇ, ʍ šəṭārōṯ > šṭārōţ “legal acts”. In my research, I had the chance to have access to a study on Demnate Judeo-Arabic (El Karich, Mohamed Abd El Monaim. 1990. Habla Arabe de una Judia de Demnate. C.E.C. dirigida por el profesor Simon Levy. Departamento de Lengua y Literatura Españolas: Universidad Mohamed V). In the dialect described, /s/-/š/ and /z/-/ž/ didn’t merge and are well distributed. However, the study focuses on one speaker only, an isolated old woman, leaving alone in the city. The distribution of the phonemes could be the result of some leveling to the Muslim local dialect. And still, a few cases of /š/ > /s/ are observed, which suggests remnants of an ancient Judeo-Demnati dialect with merged phonemes. In Demnate, /s/ is usually voiced in ʬ ʒʠʸʕ ˈʍ ʑʩ, yiśrāˀēl > izṛạel. Chetrit 2015: 6, Lévy 2009: 182, note 63. Chetrit 2015: 7. 118 PHONETICAL AND PHONOLOGICAL ADAPTATIONS OF HEBREW LOANWORDS IN MOROCCAN JUDEO-ARABIC: THE CASE OF FEZ inclinations of FJA /t/ > /ţ/: ʤʕʰ ʕ˟ ʔˢ, taqqānâ “ordinance” > ţəˀˀānâ. Actual production [t] in Hebrew loans is very rare and I could not find any occurrence for Fez69. After this exhaustive development of how Hebrew consonants are adapted in FJA, it seems quite clear that the pronunciation of the Hebrew consonants have been interpreted on the basis of reading traditions but adapted by identification with what was already available in the FJA phonological system. Since there has never been a situation of bilingualism Hebrew-FJA, the various reflexes of pronunciation might be thought as interpretations using what is available in the FJA phonological system, itself built on historical contact between Arabic, Berber and to a lesser extent Spanish. The choice of identification might have sometimes been influenced by some etymological consciousness. A symptomatic illustration, the ʺ  ʿʴʫʣʢʡ The reflexes of ʺʿʴʫʣʢʡ are, in this regard, noteworthy to observe: for Fez, the actual spirantizations are precisely the ones that find a phonemic match in FJA 70 . Indeed, spirantized ʢ, ʫ and ʴ find a phonemic match with FJA: /ġ/, /ḫ/ and /f/ whereas ʡ, ʣ and ʺ  are always produced /b/, /d/ and /t/. This can easily be explained by the fact that ʡ /ḇ/-[v] cannot be rendered in standard Arabic71, neither can ʣ /ḏ/-[ð] and ʺ /ṯ/-[θ] since the NorthAfrican Judeo-Arabic dialects all lost their interdentals. By comparison, David Cohen describes the exact same situation for Judeo-Arabic from Tunis72. Morpho-phonetical adaptations of the Masoretic Hebrew vowels into FJA73 Unlike the consonants, the adaptation of the Hebrew vowels into FJA are very connected to the syllable structure of the word and more precisely to the available stems and patterns in FJA. As it is the case for Semitic languages in general, the consonants build the root and bear the global meaning of the lexeme but the vowels are defined by the stem and are related to the mechanism of the language. The simplest illustration comes with what happens to the verbs. In FJA, the verbal stem of the first form is CCəC/iCCəC and the second form is CəCCəC/iCəCCəC. When a Hebrew verb is adapted into FJA, the consonants of the original root are reinterpreted as seen above but the vowels adapt to those of the FJA verbal stem. In those cases, it is safer to talk about borrowed lexical or consonantal roots. For example, the Hebrew verb bāḏaq 69 70 71 72 73 Although, I found some for Sefrou (in Stillman 2008 : 54), where the conditions are exactly the same: ʣ˒ʮ ʍʬ ʔˢ, talmūḏ > talmüd whereas ʤʸˣˢ, ʕ tōrâ is ţora and ʺˣʮˣʬʏʧ, ḥălōmōṯ is ḥalömöţ. However, some cases of hyper restitution are observed in the Tafilalt. Hyper restitutions could explain the enigmatic xalab “dog”, usually said to originate from Hebrew keleḇ (among other Leslau 1945: 76). Still in the Tafilalt, it occurs as well in a series of words: ʤʸʩʕ ʨʑ ʍ˝, pəṭīrâ > fṭē ṛa ̣ “funeral, ʤʕʥʠʏ ˏʔ , gaˀăwâ > ġāˁāwa “pride” (alongside gāˁāwa) and ʡ˒ʰʕˏ, gānūḇ (?) > ġnūb and gnūb “theft”. Actually, [v] is present in FJA, probably due to contact with Spanish, but was affiliated to ʥ. Cohen 1964: 30. All example are taken from Lévy (2009), Leslau (1945) and Brunot and Malka (1940). 119 JONAS SIBONY becomes bdə̣ˀ/ibdə̣ˀ (or the Hebrew root √BDQ is verbalized in a FJA stem) and dāraš becomes dṛəṣ/idṛəṣ. The vowels are fully imposed by those of the FJA verbal stem. The mechanism for the nouns is not so different although the borrowings come with consonants and vowels (and not with consonants only) and then are attracted to a close FJA nominal stem according to several criteria: the same number of consonants and close vowels. One of the difficulties is that the number of possible nominal patterns is greater than the number of verbal stems. For Lévy, “the original nature of the Hebrew vowel does not have any importance on its actual production when borrowed. The vocalism adapts to that of the Arabic stem”74. I would not say that the original vowel plays absolutely no role in the choice of the hosting pattern but indeed, it is not the key element, while the nature of the syllables of the pattern is conclusive. Another difficult issue is the fact that FJA only has 4 vocalic phonemes /ā/, /ō/, /ī/, /ə/ and Hebrew has much more. The Hebrew vocalic phonemes therefore have to fit into those four vocalic phonemes. In fact, it seems that the vowels are identified with the FJA vocalic phonemes by two criterions: quality and/or quantity. For example, Hebrew kubutz is identified to a short vowel of /u-o/ coloration. From that starting point, it can match the FJA /ō/ phoneme (quality), and be produced as /ō/, /ȫ/, /ū/ or /o/ according to the consonantal and syllabic environment or it can be identified with the short FJA phoneme /ə/ (quantity). It is possible to address this issue the other way round. FJA shaped many stems with closed unstressed syllables with short vowels (phoneme /ə/). In terms of morphology, Hebrew short vowels are usually found in similar syllable structures. When Hebrew words are reshaped into FJA patterns, those Hebrew short vowels are very likely to be adapted into /ə/. When that does not happen, mainly when the analogy with a FJA stem is difficult, the Hebrew vowel usually keeps its coloration and is then assimilated to the phoneme of same coloration. Although it seems to be the main process for the formation of those loans, many additional elements may result in irregularities. Mainly, since FJA Hebrew vocabulary is continuously in touch with its written, fixed and codified source, the contact may imply an opposite attraction towards the Hebrew shape of the word and, concerning the vowels, lead to a recoloration as in ʤ ʸʕ˝ʔ˗, ʕ kappāra “atonement” > FJA CəCCāCa > kəppāṛa and kăppāṛa. ʔ (Patach) /a/ > FJA /ā/-/ə/: [ε], [a], [ä], [ə], [ă] In the Masoretic system, ʔ is a short /a/ vowel. Both characteristics are important since it leads to two distinct possible projections into FJA phonemes: /ā/ or /ə/ which reflexes are respectively /ǟ/, /ā/, /a/ ([ε], [a], [ä]) and /ə/, /ă/ ([ə], [ă]): ʔ > FJA /ə/ is observed in the first syllable of ʤʕʰ˗ʕ ʔʱ, sakkānâ “danger” which fits FJA CəCCāCa pattern and then becomes səkkāna. CəCCāCa is a very common structure in every Moroccan Arabic dialect 75 , and corresponds to many stems76. It can be found in FJA Arabic words like zəllāba “djellaba”, 74 75 76 Lévy 2009: 213. All examples of “standard” MA are taken from Caubet 1993 (1). We are dealing here with a generic structure CəCCāCa and since the Hebrew words attracted by it were not built at first in the FJA morphological system but in the Hebrew one, it is not exactly a proper stem. In MA 120 PHONETICAL AND PHONOLOGICAL ADAPTATIONS OF HEBREW LOANWORDS IN MOROCCAN JUDEO-ARABIC: THE CASE OF FEZ or ḥəffāṛa “gravedigger” or in MA words such as žəlbāna “pea”, dənžāla “eggplant”. In this structure, /ə/ is regularly produced /ă/ when close to a pharyngeal or a laryngeal consonant as in MA măˁgāza “lazy”, šăˁˁāla “flame”, ḥămmāla “cloth part”77. In Hebrew ʤʸʕ ʕˢ ʔʤ, hattārâ “annulling” > FJA hătţāṛa, production /ă/ of phoneme /ə/ is either the result of the guttural environment or a re-attraction towards the Hebrew original vowel. The hypothesis of re-attraction towards the source language is strengthened by examples coming from the Romance languages, such as the case of sănḍāla “sandal”, a Spanish loan in FJA with production /ă/ of phoneme /ə/ without a guttural environment. Other Hebrew structures with ʔ can identify to different FJA structures, see for example ʷʩːʑ ʔʶ, ṣaddīq > CəCCīC78 > ṣəddē’ “saint” or ʤ ʕ˞ ʔʮ, maṣṣâ > CəCCa79 > məṣṣa “Matza”80. Hebrew ʔ can identify with FJA /ā/ as in ˂ ʕʠ ʍʬ ʔʮ, malˀāḵ “angel” > mālˀāḫ in the Moroccan CāCCāC structure, which is, according to Caubet81, specifically used to integrate loanwords. ʏ (Hataf Patach82) /ă/ > FJA /ā/-/ø/ : [ε], [a], [ä], [ø] ʏ is in Hebrew a very short /a/ vowel. In the Hebrew context, it is a colored Shwa resulting from the contact with a guttural sound. According to the Hebrew morpho-phonetical rules, it appears at the end of a closed unstressed syllable (Shwa Nah) or after the first consonant of the word (Shwa Na’). In the first case, ʏ will identify with /ā/ as in ʤ ʓˈʏˆ ʔʮ, maˁăśê “event” > FJA CāCāCi > mǟˁǟsẹ. Such a structure is not regular in MA, but similar structures are commonly used for loanwords: CāCāCa māgāna “watch”, bāṭāṭa “potato”, CāCīCa žākēṭa “jacket”, CīCāCa sīmāna “week”. Caubet specifies that “what characterizes those triliteral stems with two long vowels, is that they are not productive and that they are not stems strictly speaking but rather ‘molds’, into which loanwords are poured from French, Spanish, Berber and Classical Arabic”83. Hebrew can be added to that list. When ʏ is a Hebrew colored Shwa Na’, it is usually removed in FJA: ʤ ʕʮʩʺʏʑ ʧ, ḥăţīmâ “signature” > FJA CCīCa > ḥţīma. The result reminds us of many forms in MA like feminine substantives and adjectives, collectives and Masdars: gzīra “island”, sṛēqa “theft” and zbība “raisin”. 77 78 79 80 81 82 83 CəCCāCa is a generic structure that may reflect stems of different kind. It can be the structure of a feminine substantive C1əC2C2āC3-a as in zəllāba “djellaba” a stem C1əC2C2āC3-a of nominalization from a verb (Masdar) ḥəffāṛa “gravedigger”, a feminine adjective MəC1C2āC3-a măˁgāza “lazy”, a stem of names of units or collectives C1əC2C2āC3a or C1əC2C3āC4a dəllāḥa “watermelon”, žəlbāna “pea”, dənžāla “eggplant” and other. Caubet 1993 (1): 89-95. See MA bəttīx “melon”. See MA ḥənna “henna”, məṛṛa “time”, zəbda “butter”. Akoun 2015: 46 mentions that example to show the penetration of Arabic phonetic elements into the Hebrew loan words and states that alongside məṣṣa, maṣṣa is actually in use too, probably depending on the speaker’s level of education. Caubet 1993 (1): 96. The other Hatafim, i.e. colored Shwas, and mainly ʐ and ʎ won’t be listed in this study since they behave like their non-Hataf counterparts. Caubet 1993 (1): 92. 121 JONAS SIBONY ʔ (Furtive Patach) /a/ > FJA /ā/: [a] ʔ (Furtive Patach) is always adapted into /ā/. Its own raison d’être in Hebrew is to link a long vowel to a final guttural consonant: *šābūˁ > ˆʔ ˒ʡ ʕˇ, šāḇūaˁ. Thus, it always comes after a long vowel. The adaptation reflex in FJA is to insert a liaison semi-vowel to facilitate this connection: /w/ when the vowel is re-interpreted phoneme /ō/ and /y/ when the phoneme is /ī/. Since, the MA mechanism creates long vowel structures with many consonants precisely in order to insert loanwords, as CāCūCa nāmūsa “mosquito”, CāCīCa fārīna “flour”, CūCūCīC ṭōmōbīl, the same happens to the elongated Hebrew words with ʔ (Furtive Patach): ˆʔ ˒ʡ ʕˇ, šāḇūaˁ > sābūwāˁ (CāCūCāC), ʔʧʩ ʑʬ ʕˇ, šālīaḥ “emissary” > sālīyāḥ (CāCīCāC). ଉ ʕ (Kamatz Gadol) /ā/ > FJA /ā/: [ε], [a], [ä] ʕ (Kamatz Gadol) is in Hebrew a long /a/ vowel. In FJA, it is globally identified with the phoneme /ā/ and is adapted according to the stems and the phonetical environment: ˇ ʕ˙ ʔˇ, šammāš “beadle” > FJA CəCCāC > səmmās, like MA gəzzāṛ “butcher”, ḥăddād “blacksmith”. When ʕ represents the Hebrew morpheme /-â/ (Kamatz Gadol + Mater Lectionis ʤ, mainly feminine morpheme), it turns to FJA/-a/: ʤʕʰ˗ʕ ʔʱ, sakkānâ > səkkāna and ʤʕʰ ʕ˟ ʔˢ, taqqānâ > ţəˀˀāna. I would like to focus on two specific examples of words with ʕ (Kamatz Gadol) > FJA /ā/. Hebrew ʤʸʕ ʕʨ ʍʴ ʔʤ, hap̄ ṭārâ “Haftara” and ʤʕʬ ʕʣʡʍ ʔʤ, haḇdālâ “Havdala” mentioned by Leslau84 and Brunot and Malka85. They are produced respectively bdǟla and fṭāṛa. This result is due to morphological interpretation and attraction. First, the initial ha- of the words is interpreted as the Hebrew definite article (while it is the stem prefix of Hebrew hap̄ ˁālâ, nominalization of verbal stem hip̄ ˁīl). Thus, once the word had been borrowed, the speakers removed it and replaced it by the Arabic definite article: l-bdǟla and l-fṭāṛa. The other reason for such confusion might be that stem CCāCa86 is a very fruitful structure in MA. The frequency of the structure might have attracted both words into it. ʕ (Kamatz Qatan) /ŏ/ > FJA /ō/: [ʊ], [ö], [u], [o] (and /ə/? > [ə]) ʕ (Kamatz Qatan) is identified with FJA /ō/ in ʤ ʕʮ ʍʫ ʕʧ, ḥŏḵmâ > FJA CūCCa 87 > ḥōḫma “wisdom”. Unfortunately, I could not find any example to illustrate the hypothesis that short Hebrew vowels can match both the same vowel color in FJA and the short phoneme /ə/, maybe because of the small amount of original ʕ (Kamatz Qatan)in the Hebrew lexicon. 84 85 86 87 Leslau 1945: 68. Brunot and Malka 1940: 8. See MA šnāfa “lip”, žrāna “fog”, ktāba “writing”, zyāṛa “visit”. In standard MA, see for example noqṭa “point”, šoṛba “soupe”, mudda “moment”. 122 PHONETICAL AND PHONOLOGICAL ADAPTATIONS OF HEBREW LOANWORDS IN MOROCCAN JUDEO-ARABIC: THE CASE OF FEZ ʩʑ (Hiriq Yod) /ī/ > FJA /ī/: [i], [Iˁ], [eˁ] ʩʑ appears in Hebrew in words that will match FJA structures with /ī/. It is produced /ī/ in ʯʩːʺʩ ʑ ʒˎ, bēṯ dīn > CīC88 > bīţ dīn “law court” and in ʭʩʸ˒˝, ʑ pūrīm > CūCīC89 > pȫrīm. ʩʑ is produced /ē ̱̣/, /ẹ/ (which are regular inclinations of phoneme /ī/) in some specific contexts like next to emphatic /ṣ/: ʺʩ ʑʶʩ ʑʶ, ṣīṣīṯ > CīCīC90 > ṣē ̣ṣẹţ “Tzizit”. ʑ (Hiriq) /i/ > FJA /ī/-/ə/ : [i], [Iˁ], [eˁ], [ə] ʑ can turn to /ī/ or /ə/: ʤʕʧʮʍ ˈ, ʑ śimḥâ, “joy” > CīCCa91 > sīmḥa but ʤʕʰˇʍ ʮ, ʑ mišnâ “Mishna” > 92 CəCCa > məsna. From another Hebrew structure: ʸ˒˝ ʑ˗, kippūr > CəCCūC93 > kəppōr. ʒ (Tzere) /ē/ > FJA /ī/-/ə/ : [i], [Iˁ], [eˁ], [ə] ʒ is identified with FJA /ī/ and sometimes with /ə/. ʱʒʰ, nēs “miracle” > CīC > nīs. Hebrew Segolate nouns with ʒ usually turn to FJA CīCəC: ʸʓʴ ʒʱ, sēp̄ er “book” and ʸ ʓʣ ʒʱ, sēḏer “order” turn to sīfər and sīdər, thus FJA CīCəC94. This stem, with a long vowel in the first syllable does exist in MA for Arabic words but is not very productive. The fact that it integrates Segolate nouns with ʒ might be explained by the original structure of those Hebrew words, where the stress is on the first syllable. Another specific case concerns Hebrew active participles pōˁēl that seem to experience two possible adaptations: ʬ ʒʤˣʮ, mōhēl “Mohel” > mȫhīl or mȫhəl and ʸʒʴˣʱ, sōp̄ ēr “scribe” is sȫfē ̣ṛ or sȫfər. The first production is reshaped on a CūCīC common MA structure, but the second one; CūCəC, seems unattested beyond these Hebrew loans. ʓ (Segol) /e/ > FJA /ī/-/ə/ : [i], [Iˁ], [eˁ], [ə] Just like ʒ (Tzere), ʓ (and ʎ) is identified with FJA phoneme /ī/ in many situations, but it is sometimes /ə/ in original unstressed positions: ʸ ʒˢʱʍ ʓʠ, ˀestēr “Esther” > CīCCīC95 > isţīr, ʯʓʴˏʓ , gep̄ en “grapevine” > CīCəC > gīfən “wine blessing”. 88 89 90 91 92 93 94 95 See MA ˁīn “eye”, zīt “oil”. See MA ṭōbīs “bus”, būlīs “police” (pȫlīs in FJA). MA stem used for loanwords like ḍēṣēṛ “dessert”. See MA fīsta “jacket”, ˁīnba “grape”. See MA žəfna “banquet”, bəlġa “slippers”. See MA fəllūs “chick”, məssūs “tasteless” See MA ˁīnəb “grapes”. Caubet 1993 (1) mentions the stems with 4 consonants and two long vowels which function is to integrate loanwords and she quotes CāCCāC, with kāwkāw “peanut” and CīCCūC šīflōṛ “cauliflower”. CīCCīC could be added to that list. 123 JONAS SIBONY ˒ (Shuruk) /ū/ > FJA /ō/: [ʊ], [ö], [u], [o] ˒ is identified with phoneme /ō/ and thus will be produced as /ō/, /ȫ/, /ū/ or /o/, depending on the environment: ʭʩʸ˒˝, ʑ pūrīm > CūCīC > pȫrīm, ʸ˒ːʱ, ʑ siddūr > CəCCūC > səddōṛ96. ˒ a is always /ū/ before /w/ as in ˆʔ ˒ʡ ʕˇ, šāḇū ˁ “week” > sābūwāˁ, like for Arabic words where phoneme /ō/ is produced /ū/ only before /w/: hūwa “he”. ʗ (Kubutz) /u/ > FJA /ō/-/ə/: [ʊ], [ö], [u], [o], [ə] ʗ is identified with /ō/ or /ə/ as in ʤʕ˝ ʗʷ, quppâ > CūCCa > ˀȫppa “money box” but ʤʕ˗ʱʗ , sukkâ “Sukkah” > CəCCa > səkka and ʤ ʕˎ ʗʺ ʍ˗, kəṯubbâ > CCəCCa > kţəbba. ˣ (Holam or waw holam) /ō/ > FJA /ō/: [ʊ], [ö], [u], [o] ˣʖis identified with /ō/, with contextual inclinations leading to ō/, /ȫ/, /ū/ or /o/: ʩˣˏ, gōy “non-Jew” > CūC97 > gȫy but the feminine plural Hebrew morpheme ʺˣ- /-ōṯ/ is -ȫţ in ndǟbȫţ “gifts” and -ōţ in braḫōţ “blessings”, probably due to the presence of /ḫ/. ʍ (Shwa) /ə/ > FJA /ī/-/ø/: [i], [Iˁ], [eˁ], [ø] ʍ is identified with FJA /ī/ (because of its coloration /e/ in the reading traditions) or /ø/, but never to the FJA short phoneme /ə/. In fact, ʍ is often found after the first consonant of the word in Hebrew and since the presence of phoneme /ī/ in such a syllable structure in FJA is very unlikely to happen, it is mainly seen in borrowed vocabulary as it is here with the Hebrew loans: ʸˣʫˎ, ʍ bəḵōr “first born” > CīCūC98 > bē ̣ḫōṛ; ʸ ʕʡ ʍ˗, kəḇār “already” > CīCāC99 > kē ̣bāṛ. The second possibility for Hebrew ʍ is to disappear: ʤ ʕˎ ʗʺ ʍ˗, kəṯubbâ > CCəCCa100 > kţəbba101. Since the loanwords are a matter of lexicon, and because the Semitic lexicon is organized on lexical consonantal roots, the correspondences between Hebrew consonants and FJA are quite regular. But when it comes to vowels, the Semitic morphological rules apply and the vowels are imposed by the internal stem and pattern system. However, the selected FJA 96 97 98 99 100 101 See the MA loan from Spanish reshaped in CəCCūC > kəppōṭ “coat”. See MA ṣōf “wool”, ḍōw “light”. See MA loanwords šīfōṛ “driver”, nīlōn “nylon”. See MA loanword from Classical Arabic žīhād “Djihad” and (ˁla) qē ̣bāl “according to”. See MA mrəmma “embroidery frame”. For more about the behaviour of Hebrew Shwa (in Tafilalt, Marrakesh and Meknes Judeo-Arabic), see Akoun 2013. 124 PHONETICAL AND PHONOLOGICAL ADAPTATIONS OF HEBREW LOANWORDS IN MOROCCAN JUDEO-ARABIC: THE CASE OF FEZ structures are in most cases similar to the borrowed Hebrew word and in conformity with the Hebrew vowel coloration or quantity. Conclusion Throughout this study, I have identified many regularities made possible by the morphological proximity of the two languages, which facilitated identifications and adaptations. The phonological and the morpho-phonetical system of FJA exerts a very strong magnetic force on the lexical borrowings that are being reshaped according to the internal rules of the host language. The words for which the match is easy, adapt to generic FJA common stems. Those stems come with already established vowels. For the others words, where the Hebrew stem is too different from what is usually found in North African Arabic, FJA, proposes specific existing structures, with long vowels, dedicated to the integration of foreign words. Caubet (1993) listed them and showed how those structures, absorb French and Spanish words. The present study shows additional cases of such structures, but this time, in order to absorb Hebrew words. As a general conclusion, it appears that the integration of Hebrew vocabulary and the phonetical adaptations came about in two stages. The first stage concerns the reading tradition of Masoretic Hebrew and depends very concretely on what is written, that is, mainly how the consonantal script and the niqqud were interpreted. The second stage relies on concrete integration and on how the Hebrew element evolved alongside the other components of vernacular Fasi Judeo-Arabic. Once the words are in concrete everyday use, by speakers who are aware, or not, of the etymology, those words adapt to the internal phonological and morpho-phonetical rules of the host language and follow their evolutions. The fact that the phonetical internal evolutions of the dialect did impact an historical Arabic lexicon and the Hebrew loanwords in exactly the same way (like with Hebrew ʷ, /q/ and Arabic ‫ق‬, /q/ > /ˀ/) may imply that the Hebrew loanwords are very ancient. Bibliography Akoun, Nathalie. 2013. “The Shwa following a long Tiberian Vowel in the Hebrew of Moroccan Jews”. Mehaqarim balashon XIV-XV. Jerusalem: Magnes: 163-174 (in Hebrew). Akoun, Nathalie. 2015. “haaravit hamishtaqefet beivrit shel maroko”. Mehqarim betarbutam shel yehude tsfonafrika, Eds. Moshe Bar-Asher and Steven Fraad. New Haven: Yale University: 29-52 (in Hebrew). Bar-Asher, Moshe. 1978. “Al yesodot haivriim bearavit hameduberet shel yehude maroko”, Leshonenu 42: 163-189 (in Hebrew). Bar-Asher, Moshe. 1992. La composante hébraïque du judéo-arabe algérien. Jerusalem: Magnes. Bar-Asher, Moshe. 1996. “La recherche sur les parlers judéo-arabes modernes du Maghreb: etat de la question”. Histoire Epistémologie Langage 18/I: 167-177. Brunot, Louis, and Malka, Elie. 1939. Textes judéo-arabes de Fès. Rabat: Ecole du Livre. Brunot, Louis, and Malka, Elie. 1940. Glossaire judéo-arabe de Fès. Rabat: Ecole du Livre. Caubet, Dominique. 1993. L’arabe Marocain. 2 vols. Louvain: Editions Peeters. Chetrit, Joseph. 2007. Diglossie, hybridation et diversité intra-linguistique. Paris and Louvain: Peeters. 125 JONAS SIBONY Chetrit, Joseph. 2009. Linguistic Treasury and Textures: Socio-Pragmatic Studies in North African JudaeoArabic and its Hebrew Component. Articles, Poems, Tales and Proverbs. Jerusalem: Bialik Institute (in Hebrew). Chetrit, Joseph. 2014. “Judeo-Arabic Dialects in North Africa as Communal Languages: Lects, Polylects, and Sociolects”. Journal of Jewish Languages 2 (2014): 202-232. Chetrit, Joseph. 2015. “Diversity of Judeo-Arabic Dialects in North Africa: Eqa:l, Wqal, kjal and ˀal Dialects”. Journal of Jewish Languages 4 (2015): 1-43. Cohen, David. 1964. Le parler arabe des Juifs de Tunis, Textes et Documents. Paris, La Haye: Mouton and Co. Cohen, David. 1975. Le parler arabe des Juifs de Tunis, Tome II : étude linguistique. Paris, La Haye: Mouton and Co. Cohen, Marcel. 1912. Le parler arabe des Juifs d’Alger. Paris: H. Champion. Essaoui, Abdessamad. 1990. Dialecto de los Judios en Inezgan. Monografia de Licenciatura. dirigida por el profesor Simon Levy. Departamento de Lengua y Literatura Españolas: Universidad Mohamed V. Heath, Jeffrey. 2002. Jewish and Muslim Dialects of Moroccan Arabic. London: Routledge. Heath, Jeffrey and Bar-Asher, Moshe. 1982. “A Judeo-Arabic Dialect of Tafilalt”. Zeitschrift für Arabische Linguistik, No. 9. Harrassowitz Verlag: 32-78. Henshke, Yehudit. 1991. “Hebrew Elements in the Spoken Arabic of Djerba”. Massorot V-VI Henshke, Yehudit. 2007. Hebrew Elements in Daily Speech: A Grammatical Study and Lexicon of the Hebrew Component of Tunisian Judeo-Arabic. Jerusalem: Mosad Bialik (in Hebrew). Henshke, Yehudit. 2008. “tofaot udrakhim bintiyat hapoal haivri: hapoal bamarkiv haivri shebefi yehude tunisya”. Leshonenu 70: 679-695 (in Hebrew). Leslau, Wolf. 1945. “Hebrew elements in the Judeo-Arabic dialect of Fez”. The Jewish Quatterly Review 36: 61-78. Lévy, Simon. 1994. “De quelques hébraïsmes en arabe marocain et de leurs voies de passage”. Approches linguistiques: universaux et paramètres. Rabat: Mohamed V University: 35-43 (in French). Lévy, Simon. 2009. Les parlers arabes des Juifs du Maroc, histoire, sociolinguistique et géographie dialectale. Zaragoza : Instituto de Estudios Islámicos y del Oriente Próximo (in French). Lévy, Simon. 2013. “Les parlers arabes des Juifs du Maroc”. Langage et société 2013/1, n°143: 41-51 (in French). Maman, Aharon. 1989. “On identifying the Hebrew Element in Judeo-Maghrebian”. Massorot III-IV: 117-201. Maman, Aharon. 1999. “Developments in the Magrebian Judaeo-languages from the Expulsion from Spain to the Present”. Sefunot: Studies and Sources on the History of the Jewish Communities in the East, New Series vol. 7. Jerusalem: Yad Izhak Ben Zvi: 141-190 (in Hebrew). Maman, Aharon. 2019. Synoptic Dictionary of the Hebrew Component in the Jewish Languages. Including the notes of Shelomo Morag. Second revised edition. Jerusalem: Hebrew University (in Hebrew). Morag, Shelomo. 1963. Haivrit shebefi Yehude Teman. Jerusalem: Academy of the Hebrew Language. Morag, Shelomo. 2007. “Pronunciations of Hebrew”. Encyclopaedia Judaica, Second Edition, vol. 16. Detroit: Macmillan Reference USA in association with Keter Pub. House: 547-562. Pellat, Charles. 1952. “Nemrod et Abraham dans le parler arabe ds Juifs de Debdou”. Hesperis XXXIX: 121-145. Saada, Lucienne. 1956. “Introduction à l’étude du parler arabe des Juifs de Sousse”. Cahiers de Tunisie 16: 518-532. Sibony, Jonas. 2019. “Éléments hébreux ou pseudo-hébreux dans le parler arabe des Juifs de Fès”. AIDA 12th Conference. Online Edition: https://books.openedition.org/iremam/4069 (in French). Stillman Norman. 1988. The Language and Culture of the Jews in Sefrou. University of Manchester. Stillman Norman. 2011. “The Functions of Hebrew in Moroccan Judeo-Arabic with Particular Reference to the Vernaculars of Fez, Sefrou and Meknes”. Ḥikrei Maˁarav u-Mizraḥ, Studies in Language and History, Presented to Joseph Chetrit. Yosef Tobi and Dennis Kurzon (eds.), Karmel, Jerusalem: University of Haifa, Matanel: 193-203. Tedghi, Joseph. 1995. “Les interférences de l’hébreu dans les langues judéo-arabes d’orient et d’occident musulman”. Les interférences de l’hébreu dans les langues juives. Paris : Publications CEHJMC: 41-66 (in French). Tedghi, Joseph. 1999. “Comparaisons hébraïques dans le judéo-arabe marocain”, Vena Hebraica in Judaeorum Linguis. Proceedings of the 2nd International Conference of the Hebrew and Aramaic Elements in Jewish Languages. Eds S. Morag et al. Milan: Centro Studi Camito-Semiitici di Milano: 321-351. Tedghi Joseph. 2001. “shimushim metaforiim bamarkiv haivri bearavit hayehudit shel hamaarav”. BIVO 13. University of Madrid: 44-51 (in Hebrew). Tedghi, Joseph. 2003a. “Évolution des recherches sur la composante hébraïque dans les parlers judéo-arabes maghrébins modernes”. Linguistique des langues juives et linguistique Générale. Alvarez-Péreyre, F. and Baumgarten, J. (eds.). Paris: CNRS Editions: 157-190. 126 PHONETICAL AND PHONOLOGICAL ADAPTATIONS OF HEBREW LOANWORDS IN MOROCCAN JUDEO-ARABIC: THE CASE OF FEZ Tedghi, J. 2003b. “De l’évolution sémantique des lexèmes hébreux dans le judéo-arabe marocain”. Mélanges David Cohen. Maisonneuve and Larose: 689-698. Tirosh-Becker, Ofra. 1989a. “Hamarkiv haivri baravit hayehudit”. Leshonenu laam 40-41, Kovetz lishnat halashon: 331-337 (in Hebrew). Tirosh-Becker, Ofra. 1989b. “Sugiot leshoniot min haaravit hayehudit shel Konstantin”. Masorot III-IV: 285-312. Yoda, Sumikazu. 2018. “Libyan Judeo-Arabic”. Jewish Libya. Jacques Roumani, Judith Roumani, David Meghnagi (eds.). Syracuse University Press. 127