Elamur Research⌕
☰
home / papers / 2026 / On the Shaping of the Andalusi Arabic Lexicon: Semitic and Non-Semitic Borrowing, Phonetic Variation and Lexical Composition (NAḤT)

On the Shaping of the Andalusi Arabic Lexicon: Semitic and Non-Semitic Borrowing, Phonetic Variation and Lexical Composition (NAḤT)

Summary

Arabic has no comprehensive etymological dictionary, even though it is the best documented Semitic language. Corriente argues that a full one would need a funded international team. As a stopgap, he builds etymological coverage for a single well-studied dialect, Andalusi Arabic, as part of his planned Encyclopédie linguistique andalouse. This article previews what that work has turned up. He sorts hard-to-explain roots into three groups:

  • Loanwords from Akkadian, Aramaic, Persian, Old Egyptian and Berber, many not previously noticed.
  • Phonetic variation of roots, of the kind Majzel' called allothesis and metathesis.
  • Lexical composition, where an extra consonant, a prefix or a preposition fuses onto a root, or where whole phrases are compressed into a new root (naḥt).

He argues that many Arabic roots with no apparent Semitic cognates become explainable once these processes are taken into account. The method is comparative and example-driven, and the author openly says it is not exhaustive. The study of semantic change is left for later work. The value is a concrete model for doing dialect-by-dialect etymology and a list of mechanisms that have often hidden true etyma.

Key points

  • The context is the author's planned Encyclopédie linguistique andalouse, written in French, which builds on his earlier Dictionary of Andalusi Arabic (Corriente 1997) and his Arabic-loanword dictionary for Ibero-Romance (Brill 2008).
  • Akkadian borrowings proposed: faḫḫār "potter", argued to be direct from Akkadian because /ḫ/ is kept faithfully, and furātu "Euphrates", where vowel preservation points to direct borrowing.
  • Aramaic and Syriac borrowings: examples include ṭīn "mud", zāwiya "corner" and radana "spin". The Andalusi senses are linked to Syrian settlers in al-Andalus.
  • Persian borrowings: many are compounds with ḫar "donkey", such as ḫardal "mustard" (from "donkey tongue") and ḫirbaq "hellebore" (from "donkey face").
  • Composition types:
    • an added consonant;
    • a fused mimation or nunation;
    • an old causative prefix;
    • a preposition fused at the end of a root (e.g. ǧarraba "to experience" from ǧarà + bi);
    • a preposition fused at the start of a root;
    • full naḥt, constrained by the four-consonant root limit (e.g. armala "widow" from a phrase meaning "one who has no man", with parallels in Hebrew, Syriac, Akkadian and Ge'ez).

Interesting ideas

  • I1. A trait that survives intact, such as an unmerged consonant, can show which of several candidate routes a borrowing took. — For Mova (learn): when silver labels disagree, look for each witness parser's distinctive error signature to trace which source a disputed label really came from.
  • I2. Opaque single words can be read as fossilized verb-plus-preposition phrases whose meaning combines the two parts. — For Mova (frames): treat English particle and preposition verbs as one predicate with its own VerbNet-style roles, rather than as a verb plus a loose adjunct.
  • I3. Unexplained residue is sorted into a fixed set of named mechanisms (loan, phonetic variant, affix fusion, compound) instead of being left as noise. — For Mova (tokens): give every unknown word in the morphology an explicit, inspectable hypothesis type (e.g. foreign loan, spelling variant, blend, compound) so lemmatization failures stay explainable.
  • I4. The same underlying formula recurring independently across several languages supports a reconstruction. — For Mova (back): when translating through the meta-world, check that a reconstructed meaning structure surfaces with matching pieces in each target language as a consistency test.
  • I5. A hard size limit on roots forces compounds to drop material, which is what makes their sources hard to recognize. — For Mova (back): the text → graph → text compressor can record which components were dropped under its length budget, so a compressed form can still be traced back to its source.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Topics: Akkadian and cuneiform NLP

Cite

The paper itself, not this page. BibTeX from the library's record of the work.

@inproceedings{corriente2026shaping,
  title = {{On the Shaping of the Andalusi Arabic Lexicon: Semitic and Non-Semitic Borrowing, Phonetic Variation and Lexical Composition (NAḤT)}},
  author = {Corriente, Federico},
  booktitle = {Folia Orientalia 52},
  year = {2026},
  url = {https://wydawnictwo.pan.pl/index.php/fo/article/view/7027}
}

References

2 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. The frequency of loanwords in Old and Medieval Arabic and Neo-Arabic is known to every student of this language, thanks to works like Jeffery’s and Dozy’s and, in the case of Andalusi Arabic, of Corriente 1997, 3 so we shall herewith confine ourselves to cases insufficiently dealt with in previous literature, 2 Of the kind labeled as “allothesis” and “metathesis” by S.S. Majzel’ in his book entitled Puti rasvitija kornevogo fonda semitskix yazykov (“Developing ways of the root stock in the Semitic languages”, Moscow, 1983), which we reviewed in Estudios de dialectología norteafricana y andalusí 9 (2005) 273-8, and some of whose ideas, often disputable, have nevertheless influenced our thinking in some parts of the present article. 3 Namely, A.F. Jeffery’s The Foreign Vocabulary in the Qur’ān, Baroda, 1938, R. Dozy’s Supplément aux dictionnaires arabes, Leiden 1981, and F. Corriente, A Dictionary of Andalusi Arabic, Leiden. such as direct borrowings from Akkadian, and a heretofore not-witnessed though expectable wealth of Aramaic, Old Egyptian and Iranian loanwords, even Berber ones, though these merely affecting Western Neo-Arabic. 4 1) Akkadian: Only a very old direct borrowing from this language can explain A. izā’a “in front of”, obviously from a pan-Semitic root {ḥḏw}, whence Ar. ḥazā «to see», independent from cognate Ak. izēzum «to stay». Other unsuspected cases of likewise direct borrowing might or appear to be A. ḫabarun “news”, from Ak. ḫabāru(m) “to make noise”, Ar. dīsun “rush, reed” from Ak. dīšu(m) «lush grass», reflecting the pan-Semitic root {dṯ’}, through a South-Arabian phase, 5 Ar. zibbun “penis”, apparently an old euphemism obtained by a metaphorical use of Ak. zibbatu(m) ”tail”, Ar. faḫḫārun “potter”, necessarily borrowed from Ak. paḫāru(m), and not from its Ar. loanword paḥḥārā, on account of the faithful transmission of /ḫ/, while a better preservation of vowels would in a similar way prove the direct borrowing of Ar. furātu from Ak. purattu “Euphrates” and Ar. quwwatun “strand of a rope” from Ak. qû(m) “rope”, although both items existed also in the North-Semitic branch. All in all, perhaps a meager harvest, but not meaningless nor definitive, because our survey of this realm can lay no claim to exhaustiveness. 2) Aramaic: The number of loanwords of this origin in Old Arabic has always been known to be high, but a detailed survey of A. lexicon yields a considerably longer list of new never heretofore suspected cases, like A. ḏaḫīratun “treasure < ammunition”, in a root apparently isolated from any Semitic cognates, probably reflecting Ar. dəkīrā «remembered (when the need for it arises)»; otherwise A. radana “spin” is no doubt borrowed from Sr. rədan, A. rahṭun “family” but Aa. ‘way, manner” closely reflects Sr. rehṭā, 6 while the whole A. root {rwq} is likely to be borrowed from such items as Rb. arēq «to pour» and rǝwwāqā 4 We are using the abbreviations A. = Arabic, Aa. = Andalusi Arabic, Ak. = Akkadian, Ar. = Aramaic, Br. = Berber, Cp. = Coptic, Eg. = Old Egyptian, Esa. = Epigraphic South Arabian, Gz. = Ge‘ez, Hb. = Hebrew, Kb. = Kabyle Berber, Ml. = Maltese, Np. = Neo-Persian, Ph. = Pahlavi, Sr. = Syriac, Rb. = Rabinic Aramaic and Ug. = Ugaritic. 5 Suggested by the evolution of /ṯ/ into /s/, detected by G.M. Grande, Jazyk juzhnoaraviskoj pis’mennosti, Moscow 1966, p. 40, also witnessed by the A. ordinal number sādis “sixth”. Another formerly known case of direct contact between Ak. biltu(m) “tribute” and Esa. is >bltn< “gift” in the latter, origin of Gz. bǝnnät “tribute”, as reported by W. Leslau, Comparative Dictionary of Ge‘ez, Wiesbaden 1987, p. 99. 6 Semantically closer to the Aa. item, which can be explained as a consequence of the linguistic impact of the Syrian settlements in Al-Andalus, to which we consecrated an article, “Los sirismos del árabe andalusí”, in Estudios de dialectología norteafricana y andalusí 4 (1999) [2000] 55–63, as a counterpoint to our previous paper “South Arabian features in Andalusi Arabic”, in Studia lingüística et orientalia memoriae Haim Blanc dedicata (Wiesbaden, Harrassowitz, 1989), 94–103. Curiously enough, Aa. raddána “spindle”, an overhauled variant of A. mirdan, has survived in the Castilian dialect of La Palma, one of the Canary Islands, as redina, registered by Corriente 2008: 418, paralleled by Ml. raddiena “spinning wheel”, after J. Aquilina, Maltese-English Dictionary, Valletta 1990, p. 1173. find
  2. «percolator»; cf. also >zaraba< “to pen (cattle)” < Sr. zərab «to press» and Rb. «to surround», >zināqun< “horse collar” < Sr. zənaq “to tie up (a beast)”, >zāwiyatun< «corner, angle» < Rb. and Sr. zāwītā, >zayyun< “costume, habit” < Rb. >zīw(ā)< «beautiful outlook», >sabara< “to probe or sound” < rb. səbēr «to pierce», Aa. >sannaf< «to slice» < Sr. sannef «to tuck up», A. >ṭamara< “to bury or hide” < Rb. and Sr. ṭəmar, >ṭīnun< “mud” < Rb. ṭīn(ā), >‘ağnana< “to knead” < Rb. ‘ăgan and Sr. ‘əga/en «presser, >‘aṣā< “to disobey” < Rb. ‘ăṣā/ē «to oppress; to withhold payment», and Sr. ‘əṣā «to resist», >‘anita< «perish» < Rb. ‘antā «oppressor», and >qama‘< “to subdue” < Rb. qāma‘ «to press» and Sr. qəma‘ «to tie». 7 3) Persian: the rather impressive number of Iranian loanwords in A., above all Old, Middle and Modern Persian (i.e., Pahlavi and Neo-Persian or Farsi), is generally acknowledged, 8 but its real extent is far from having been established in a comprehensive statement. Amidst a host of heretofore ignored items, hard to etymologize in any other way and detected by our recent research on Aa., we could excerpt, e.g., A. barada “to file” prob. < Ph. burdan or np. bordan «to take away, to remove», 9 tubbān “breeches” < Np. tobbαn < tan bαn “body protector”, ǧawḏar “calf” < Ph. *gaw dar “kind of cow”, ḫirbaqun «hellebore» < Np. ḫar bok, lit. «donkey face» ḫardal “mustard” < Ph. ḫar dil, lit. ”donkey tongue”, ḫuršūf “artichoke” < Ph. ḫar čōb, lit. «spiny stick», ḫarr/nūbun “carobs” < Np. ḫar lup ”donkey jaw”, ḫammana «to think or reckon» < Np. ḫamαnα’i “likeness”, ḫanğarun “dagger” < Np. ḫa/enğir “sharp”, ḫandarīs “exquisite wine” < Ph. *ḫand+i rēš «laughter from the beard», ḫunzuwānah “pride” < Ph. *ḫōn uzwān «proud language», duḫān “smoke” < Ph. *dūd ḫān(ag), represented by Np. dud ḫαne “fireplace”, daydān “habit” < Np. dide dαne “seen and learnt”, dardārun “ash-tree” < Ph. *dard dār «tree of pain» or dard ār «inflicting pain», because its branches were used as scourges, durnuk «kind of rug» < Ph. dō rang, “of two colours”, dahr «time; fate» < Ph. dagr zamān “long time”, ramakatun “mare” < Ph. ramkē, rawnaqun «splendor» < Ph. rōy nēk “beautiful shape”,>zi’birun “down, fuzz” < Ph. az abar “from above”, sābiġ “long garment” < Ph. šabig “ritual shirt in Zoroastrian ceremonies”, whence Np. šabi «night gown», sarābun «mirage» < Ph. *sar āb, «head water, i.e., only imagined», Aa. For which the North Semitic cognates Hb. qāmaṣ «to hold» and Ug. >qmṣ< «to scare» pose a pan-Semitic root {qmḍ}, which would have been preserved by A., if this item were not borrowed from Ar. 8 E.g., in such classical references as A. Shir, Mu‘ǧamu l’alfāẓi lfārisiyyati lmu‘arrabah, Beirut, 1990, to be used with caution, and W. Eilers’ article “Iranisches Sprachgut im Arabischen”, in Actas do IV Congresso de Estudos árabes e islâmicos, Coimbra-Lisboa 1968, 581-550, not to speak of Jeffery’s above mentioned work. 9 Cf. the parallel case of Br. ttǝkkǝs afuhri «whetstone, lit. she takes away the excess», in Dozy’s Supplément I:139, s.v. >tksāfhr<, explained in our “Marginalia on Dozy’s Supplément” in Zeitschrift für Arabische Linguistik 29 (1995) 23–50, esp. 28, nº 52. >barham< “ointment” < Np. bar ham “together”, >ḫafağ< «horse radish» < Np. ḫafanğ, “delicious”, >ḫunnār< “beloved” < Ph. ḫwān “table”, >ḫankarah< “carousal” < Ph. *ḫwān kār “(eating) table work”, with an agentive suffix, Aa. >zinṭār< “strong and brave man” < Np. zende dαr «watching», >ġunbāz< “kind of doublet or jacket” < Np. *gom bαz< “without arms”, and a long etc. 4) Old Egyptian: The presence in A. of this kind of borrowings follows a similar pattern to that of Iranian loanwords, having been traditionally acknowledged, but never thoroughly investigated. 10 Among the items attributable to this origin in A., our research points to previously unsuspected cases like A. aṯāṯ “furniture” < 3ṯt< «bed», asās “foundation, basement” (cf. Cp. esēt), anām “men” < ỉnm «who?» (cf. Cp. nim «somebody»), the very name of Al-Andalus, from Cp. *ament e-rēs “the West by the South”, 11 bān “ben tree” < Cp. p+an «the perfume», bardī “reed-mace” < Cp. pi+roti “undergrowth”, burr «wheat» < >brt< «cereal», >baṣalatun< “onion” < Cp. (e)mčōl, >būmun< «owl» < Cp. bom, >tuḥfatun< “gift” < >ḥtp< «grace, favour», >iṯmidun< “antimony” < >smty<, >ṯaman< “price” < >śmn<, >ǧubbatun< “jubbah” < ḏb’yt “kind of garment”, >ḥida’atun< “kite” < >ḥ’t< “a certain bird”, >ḥaṭṭa< “to put down (a load)” < >ḥtp< “to be put down” (with loss of the final consonant by metanalysis of the Semitic preposition bv+), >ḥuqqatun< “little box” < >ḥ’ḫt< «wooden box», >ḥanūṭun< “embalming ointment” < >ḫnt.t< «ointment», >ḫarağa< “to go out” < >ḥrj< “to be or become far away”, >ḫaṣṣa< “to be bad or mean” < >hsj.t< «to do evil», >rīfun< “cultivated land” < r’pr (or shapes closer to Cp. erphei “temple and surrounding farms”), >si‘r < “price” < Eg., cf. Cp. ša(a) r, >sāqa< “to drive” < Eg., cf. Cp. sōk, ‘išā’un “evening” < >wš3w<, ġadan “tomorrow” < ḥḏ t3 «morning», >qaṭara< “to distill” < >ḳdrt< “frankincense”, Aa. >’wḏim< 12 “cornelian” < ỉdbw, >b/faysārah< “dish of cooked beans” < Cp. pise arō “coction of beans”, >daqqūšah< “oil cruet” < Cp. t+kounčou, etc. 5) Berber: We shall not repeat here the results of our survey of loanwords from this origin in Aa., to which we consecrated some articles, 13 nor is this Not even in the five volumes of Erman & Grapow’s magnificent Wörterbuch der ägyptischen Sprache, Berlin, 1982, our main reference for this survey, which includes a list (V 242) of merely three dozens of Arabic cognates of Eg. words, not necessarily borrowed from these. 11 About which, see Corriente 2008: 179 and our paper “The Coptic loanwords of Egyptian Arabic in comparison with the parallel case of Romance loanwords in Andalusi Arabic. The true Egyptian etymon of Al-Andalus”, en Collectanea Christiana Orientalia 5 (2008) 59–123. 12 Mere transcription in the Leiden Glossary of Hb. ōdem, about which see our paper “The names of the gems in the ephod as reflected in the Leiden Glossary and their reflexes in the Islamic mystical jargon”, in Graeco-Latina et Orientalia. Studia in honorem Angeli Urbani (ed. S.Kh. Samir & J.P. Monferrer-Sala), Córdoba, CNERU-CEDRAC, 2013, 87–99. 13 E.g., “Notas de lexicología hispano-árabe: II. Nuevos berberismos del hispanoárabe”, in Awrāq 4 (1981) 27–30, et “Le berbère à Al-Andalus”, in Études et Documents Berbères 15–16 (1998) [2000] 269–275. Another important contribution to this realm is J. Bustamante & M. Tilmatine, “El léxico amazige contenido en la ‘Umdat aṭ-ṭabīb”, in Al-Andalus-Magreb (Cadix) 7 (1999) 43–64. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/corriente-2026-shaping-andalusi-arabic-lexicon-semitic/