Arabic-Speaking Christians and Toledo, BCT MS Cajón 99.30 in High Medieval Spain

: This article interprets the Latin and Arabic notes of Christians who made a primer of Latin grammatical texts—primarily Donatus’ Ars grammatica —in order to shore up their Latin vocabulary. The copyists and annotators of this manuscript on the whole offers excellent evidence for Latin learning among Iberia’s Arabic-speaking Christians, who thrived in much of the peninsula between the ninth and thirteenth centuries. 145

In Toledo sometime between the ninth and twelfth centuries, Arabic-speaking Christians made a manuscript of Latin grammar and annotated it in their first language. 1The textbook, now Toledo, Biblioteca Capitular MS 99.30, forms hardly the only evidence we have for Arabicspeaking Christians reading in Latin, but it offers excellent evidence for their learning and/or teaching of the language.Mozarabs, a convenient term for Arabic-speaking Christians in Iberia meaning ‚those who became Arab,‛ employed Arabic as they read Latin codices throughout the Peninsula. 2In general, they thrived in al-Andalus between the ninth and tenth centuries.1 Pieter S.J. Van Koningsveld, The Latin-Arabic Glossary: a contribution to the study of Mozarabic manuscripts and literature.(Leiden: Labor Vincit, 1976), p. 34, who in turn follows Anscari Mundo, ‚La datación de los códices litúrgicos visigóticos toledanos‛, Hispana Sacra Vol.18 (1965): pp.1-25, dates the manuscript to the twelfth century.They went away from the argument of Agustín Millares Carlo, who had dated the manuscript to the eleventh century: ‚Manuscritos Visigóticos‛, Hispana Sacra Vol. 14 (1961): pp.337-444.Louis Holtz, meanwhile, has dated the manuscript to the eighth or ninth century, with a Catalonian origin and Toledan provenance.See Louis Holtz, ed., Donat et La Tradition de L' Enseignement Grammatical: Étude et Édition Critique (Paris: Éditions du CNRS, 1981), pp.384-6.I lean toward an earlier date, and as I discuss below, some of the Arabic notes come from the Latin copyists of the manuscript. 2 I am aware as well of Fernando González Muñoz, Latinidad Mozárabe (A Coruña: Universidad da Coruna, 1996), but the ninth-century Cordoban authors whom he treated, such as Alvarus of Córdoba, left little evidence that they knew Arabic, other than eccentric syntax in Alvarus' case (p.227).Karla Mallette in turn has suggested that this eccentric syntax suggests knowledge of Arabic: Karla Mallette, European Modernity and the Arab Mediterranean: Toward a New Philology and a Counter-Orientalism (Philadelphia: University of Pennsylvania Press, 2010), pp.180-186.Jason Busic has tentatively followed her suggestion, in ‚Between Latin Theology and Arabic Kalām: Samson's Apologeticus contra perfidos (864 CE) and Ḥafṣ ibn Albar al-Qūṭī's Extant Works (fl.Late Ninth/ Early Tenth Centuries)‛, Medieval Encounters Vol. 25 (2019), pp.553-580 (557, n.16).Quite likely, They also moved in some numbers to León, to live under Christian rule, as well as to the monastery of Ripoll in Catalonia.In twelfth and thirteenth-century Toledo, meanwhile, the last large community of Mozarabs helped translate Arabic scientific philosophical texts into Latin, even as Castilians and Franks increasingly held power in the city and its churches. 3henever they lived, the copyists and annotators of this manuscript thus learned Latin for themselves or to teach others in a significant Arab Christian community.They wrote the language in Visigothic script, the standard for Iberian Christians between the ninth and twelfth centuries.The manuscript and its Latin and Arabic notes tell us of more than Latin grammatical texts, but also make clear the deeper roots of Latin education and indeed Latin culture among the Mozarabs.Indeed, Arabic offered them a means to clarify actively their understanding of Latin and its accompanying culture.In what follows, then, I will focus upon these copyists and annotators and their engagement with this textbook.I argue that they employed Arabic to understand Latin words, and indeed, that their primer foremost illuminates an interest in learning vocabulary, as they quite likely already knew some Latin grammar when they began to pore over this textbook.
In general, Mozarabs read and wrote in both Latin and Arabic.These copyists and annotators were no different with their Latin Primer.Indeed, they copied manuscript folios in Latin, from one or several other codices, before annotating them in Arabic and making notes in Latin as well.In a manner reminiscent of Martin Irvine's masterful view of grammar among Latin Christians in northern Europe, their manuscript reflects their own interests in the language: as Irvine made clear, the copyists of Latin grammatical manuscripts ranged widely in the texts which they employed. 4The series of case studies that I offer thus let us watch Arabicspeaking Christians / Mozarabs define the technicalities of Latin grammar and conjugate Latin verbs.Along the way, they also rendered Latin vocabulary in Arabic, showing us how Mozarabs moved between these two very different languages.
With regard to the larger picture of Latin culture among the Mozarabs, this primer has much to offer as well.For while its copyists and annotators likely had some knowledge of Latin before they annotated the primer, as I suggest below, nevertheless the texts offer a Alvarus and others did know something of the language, but their texts do not reflect that supposed knowledge well.means to learn Latin.The excerpts of the stalwart Latin grammarians Donatus (fl.354) and Priscian (fl.6 th century) which they read suggest strongly that Arabic-speaking Christians learned Latin much as did other medieval Europeans, albeit with vocabulary mediated through a second language far different from Latin, in contrast to romance speakers.They thus help us to see how these Arabic-speakers learned the language in which they read the Bible and many other authors, including Pope Gregory I (d.604) and Isidore of Seville (d.636).Put simply, to understand the thought-world of the Mozarabs, one needs to remember that they read deeply in both Arabic and Latin.
I am to my knowledge offering the first article-length reading of the Latin primer, yet my interpretation of the manuscript does build upon earlier scholarly treatments.Pieter Van Koningsveld, in his admirable close reading of a twelfth-century Latin-Arabic glossary, pointed to the importance of this primer in corroborating his broader argument that Mozarabs/Arabicspeaking Christians turned to that glossary not for polemical or missionary purposes, but rather to learn the language for its rich literature. 5Decades later, Cyrille Aillet noted that many of the manuscript's thirty-four folios have the Donatus text over a palimpsest, which certainly makes reading the Latin difficult at times. 6Indeed, these copyists wrote Visigothic script with varying degrees of clarity, and in and around numerous erasures.In general, I have erred on the side of caution with the case studies I put forth for both the Latin text and the Arabic notes, with the understanding that the Latin and Arabic together ought to complement one another. 75 Pieter Van Koningsveld, The Latin-Arabic Glossary, 43-4.Where I have deemed appropriate, I have compared readings from the twelfth-century Latin-Arabic Glossary with this manuscript.For the sake of concision, I refer to Charles Seybold (ed.), Glossarium Latinum-Arabicum (Berlin: Emil Felber, 1900) simply as Seybold or the Latin-Arabic Glossary hereafter.I note as well that my comparison between these readings serves as a service to readers, and that while the two texts share many readings, I am not offering any concrete relationship between them.Van   This means as well that I have had to steer clear of several interesting but problematic passages in the manuscript, a problem to which other scholars of manuscripts doubtless can relate.On editions of the making of editions of Visigothic manuscripts, see Juan Gil, ‚Para la Edición de los Textos Visigodos y Mozárabes,‛ Habis Vol. 2 (1973), pp.189-234.
I have likewise followed Louis Holtz in his description of the manuscript. 8Most importantly, he wrote that the manuscript has many hands, but he did not make clear which copyists did which folios.I suggest that the manuscript is largely the doing of one Copyist A, who also wrote in Arabic, with others who did folios here and there, as well as additional annotators in Latin and Arabic.Certainly, Copyist A did the opening folios, of Donatus' Ars minor, until folio 13v.This copyist wrote Latin in both a lighter and darker brown ink, in the Visigothic miniscule of early medieval Iberia.His erratic hand did not leave uniform word spacing; for example, folios 7r through 8v are wider than many other folios.He also made Arabic notes, in brown ink, which a second annotator, writing in a crisp Arabic hand in brown ink, complemented.Indeed, while I follow Holtz, who did not treat the Arabic hands, I still have doubts as to whether or not this second Arabic hand comes from the copyist, making a second pass through his section of the manuscript, for he passed through his Latin with darker ink too, as on folio 4v.Folios 13v-16v likely come from another copyist.The case for this hand is a little tougher than Louis Holtz's characterization suggests, in part because while the hand here wrote with a condensed script, it at times looks quite like the script of Copyist A. The opening lines of Folio 14r, for example, look quite similar to the preceding ones, although the script condenses in mid-sentence, in darker ink than the rest of the folio.The texts here follow Priscian's Eighteen Books of Grammatical Institutions (hereafter the Institutes), and his treatment of verb conjugation.In the midst of Folio 16v, meanwhile, he put the incipit for the Institutes, although he had already worked with that text.Indeed, Copyist A's hand then picks up at folio 17r, and he copied to the end of the Institutes on folio 22r.
Copyist A likewise did folios 22v-25v, which return to Donatus and his Ars maior and its treatment of nouns: ‚Here begins the Definitions of Nouns and Other Parts of Speech according to Donatus, after which Follow the Rules of Declension according to the Same‛. 9This material on nouns thus fits well with the earlier grammar on verbal conjugations.But of perhaps even greater interest, a word list begins on folio 26r, perhaps in Copyist A's hand, which also wrote that verso folio, while another hand made another list on folio 27r.The copyists added numerous Arabic translations to this Latin lexicon, making it a valuable resource for understanding the linguistic world of Arabic-speaking Christians in Iberia.The Ars maior then picks up until folio 34r, in what appears to be the hand of Copyist A. Folio 34v, the last in the manuscript, opens with another version of the Ars minor, and another hand, lending evidence to Louis Holtz's argument that more than one copyist made the manuscript: someone likely bound together these folios from various copyists, which explains the two copies of the Ars minor.Likewise, throughout the manuscript, still more hands have annotated in both Latin and Arabic.I have thus here aimed to nuance Louis Holtz's valuable description, and point out that one copyist made much of the manuscript, and that he worked in a quite inconsistent manner, with his writing enlarging and shrinking, presumably as he tired or worked quickly.While much of what I treat below deals with Copyist A, it certainly bears mention that he did not work alone, and indeed that the copious Arabic and Latin notes which these hands made blend well together.

Copyist A and the Technical Vocabulary of Latin Education
The foregoing brief details make clear the scope of my interpretation of the manuscript.As we will see, these copyists and annotators offer several excellent case studies in how Arabicspeakers took up Latin grammar.Indeed, their manuscript forms a case study in itself, but I have focused upon interpreting numerous folios in the manuscript in order to do justice to the detail in which they annotated their book.The first of these case studies comes on folios 4r and 4v, where Copyist A worked through the Ars minor.He made numerous Arabic notes in a thick script that has the ductus of the Latin.He often put in diacritical marks, such as a dot under under fāʾ as is characteristic of Maghribi Arabic script, but he pointed almost no short vowels between consonants, much like many other Arabic scribes, copyists, and annotators working with non-Qurʾānic codices. 10ndeed, Copyist A engaged himself especially on folios 4r-4v, in the third part of the Donatus text.The Latin lesson here focused upon ‚what a verb is‛: ‚What is a verb?A part of speech with tense and person, without case, signifying either to do something or to be acted upon or neuter.How many accidents does a verb have?Seven, which are quality, conjugation, type, number, figure, tense, and person.The quality of verbs in which is in modes and in forms: modes which are indicative as 'I read,' imperative as 'read,' optative as 'I should read,' conjunctive as 'when I read,' infinitive as 'to read,' impersonal as 'it is read.'

Donatus' text here puts forth the technical vocabulary for talking about Latin
Yet Arabic offered the means by which to interpret that technical language.For example, the copyist translated qualitas as kayfiyyah, a ‚quality,‛ and modis as nawʿ wa anwāʿ, the singular and plural of ‚kind, sort, type,‛ and the like. 12He wrote on folio 4r of the indicative mood with the root d-l-l, whose many meanings include ‚to demonstrate‛ or ‚to point out,‛ and of the imperative mood as ‚a command‛, an apt description. 13When copying Latin on the subjunctive mood, which he here called the optative, in reference to hopes or wishes, he wrote in his thicker hand: ‚the desired thing,‛ over the Latin obtatibati. 14He furthermore referred to formis as al-ṣifah, by which many Muslims also described God's attributes, such as his ability to speak, see, and judge. 15For this copyist and the separate annotator, language too had attributes.
His notes make clear that he learned how to describe language, whether for his own benefit, or to explain to students.For here as well, Copyist A defined conjugation, when he read of the seven things that a verb has: over ‚conjugation (coniugatio)‛ he wrote ‚arrangements,‛ while another annotator, who wrote numerous other notes that I discuss below, added: ‚and adjustment (takayyuf)‛. 16 over the Latin ‚figura‛ and the thinner hand, of a separate annotator whom I discuss below, added ‚number,‛ over the Latin numerus.17On folio 4v, meanwhile, he copied ‚coniunctiuus ut quum legam‛, ‚conjunctive as when I read‛, with ‚adverbial,‛ while the second annotator added ‚and the junction‛in Arabic above conjunctive. 18So too did he write ‚without limit‛ over the infinitive legere. 19urthermore, this separate annotator worked through this Latin in a thin Arabic script.The Arabic terms which he employed for the above Latin offer striking insight into his movement between the two very different languages.For example, he wrote ‚articulated speech or an utterance (al-nuṭq)‛ above ‚of speech (orationis)‛, and ‚arrangements (al-naẓẓīm)‛, above the Latin ‚person (persona)‛, and ‚without form (illā qālāb )‛ over ‚without case (sine casu)‛. 20His evidence here suggests that he thought of a verb's person as one way of ordering and classifying it.Through illā, meanwhile, he rendered the Latin preposition sine, and he then employed al-qālāb to make clear that Latin verbs have no case, a point someone learning Latin needed to know, in order to separate verbs from nouns or adjectives, and to understand what each does in a sentence.
He likewise thought about verbs in regard to characteristics such as tense and conjugation.or where the Latin on folio 4r reads: ‚how many things does a verb have (accidunt)?Seven, quality, conjugation, kind, number, figure, time, and person,‛ he clarified grammatical vocabulary again. 21Above accidunt, he simply wrote ‚it follows‛ or ‚it attaches to,‛ which reads much like ‚it occurs‛ or ‚it turns out,‛ as we often translate accidere. 22While we cannot know with absolute certainty, the evidence suggests that he wrote yalḥaqu as a way of remembering what an accident is, and either for his teaching or for his own learning.
Both annotators left a wealth of notes on Folio 4v as well.As on the previous folio, for example, Copyist A wrote again that the Latin formis is al-ṣifah, in that thicker Arabic script. 23urthermore, he made a series of notes when he read of first-conjugation verbs, which have an infinitive ending in -are, in the indicative mood: prima que est que indicativo modo: tempore presenti numero singulari•prima secunda personis verbo activo•et neutrali habet productam ante nouissimam litteram: ut amo: amas: ‚and the first [conjugation] is that which is in indicative mood, present tense, singular number, first-and second-person, active and neutral, having been produced before a new letter: as amo, amas‛. 24er neutrali, he interpreted: ‚not the doer and not done‛, a voice goes beyond that which most Latin students learn.Seemingly, he saw the masculine as active and the feminine as passive, so that with the neuter voice, he put neither masculine nor feminine. 25Indeed, when he read of the verb's voice, he also annotated it as fāʿil, meaning active. 26n his thicker Arabic hand, Copyist A also wrote nawʿ when he read of a verb's mood (modo) here. 27His translation thus matches what he wrote on the previous folio, where there too he learned/taught of the indicative, optative/subjunctive, and imperative moods.When he read ‚having been produced‛ or ‚having been led out (productam)‛, he then wrote in his thicker Arabic hand ‚bounded‛ or ‚fixed‛, as in a verb that undergoes a fixed change.Then after this, the hand in thinner Arabic added: ‚and long‛. 28Both ‚bounded‛ and ‚long‛ thus relate to the verb producere, and offer evidence of how these Arabic-speakers thought about the stems of verbs, the letters that do not undergo change, and the endings, through which one sees the conjugation.
Copyist A and the other annotator thus put forth an Arabic vocabulary to describe Latin grammar.They did so while making clear the importance of writing/talking about Latin and its grammatical system in a sophisticated way. 29For example, on the bottom of the folio, when telling of how third-conjugation verbs either have e or i as a vowel before their ending, the copyist wrote of morphology again: ‚These [third-conjugation verbs] both by the imperative and in the infinitive mood are able to be discerned whether the letter I might be shortened, or led out: for having been shortened, the letter I is converted to e, if it is led out, it is not changed.‛ The annotator in a thin Arabic hand again translated ‚led out (ductam)‛ as ‚long (ṭawīl)‛, when explaining how one arrives to a verb with a shortened i, one that undergoes a stem-change.
Above that description of a shortened verb, moreover, he wrote ‚the principal part‛ or ‚the main part‛. 30The qaṣīd here forms the stem of the verb, it seems, rather than the changed syllable or the ending.Donatus here offers examples from legere, whose third-person singular present tense is legit, but whose singular imperative is lege, and audire, a fourth-conjugation verb that he groups with the third, and which does not undergo changes in audit and the imperative audi.
On folios 4r-v, then, Copyist A and his annotator labored on several occasions over the technicalities of Latin grammar.His Arabic notes in a thicker hand have a similar ductus to the Latin, and likewise complement the Arabic notes in a thinner script perfectly. 31Perhaps most importantly, these two employed Arabic, their first language, to explain the technicalities of Latin grammar; for example, what a mood is.They either learned and /or taught how to talk about Latin as a grammarian did, with a command of the necessary vocabulary.And so, we see here see Mozarabs immersed not in practicing the conjugation of verbs -at least as we can tell from the evidence of these folios-but writing of how to talk about the language.
Copyist A and His Annotator in the Second Quire of the Manuscript: Adverbs and Latin Exercises, Folios 9r-13r When he came to Copyist A's work in the Latin primer's second quire on folio 9r, his annotator with the thin hand made many Arabic notes reminiscent of those above.The vocabulary to describe Latin grammar took this annotator's interest again, but, as we will see, he furthermore delved into Latin sentences, complete with Arabic notes on Latin words of all sorts, which offer valuable evidence of how he built his Latin vocabulary either for learning or teaching.To be clear, I am not arguing that one type of Arabic note that he made was more difficult than another, or that he progressed into learning and/or teaching increasingly difficult parts of Latin.We simply do not know what part of Latin -whether the language's case system or vocabulary, for example-posed the greatest difficulty for him.Yet in the second quire, this annotator clearly translated a broader range of vocabulary than in the first.In this second case study of the manuscript, then, we will see the annotator with a thin Arabic script interpreting the Latin text especially between folios 9r and 13v, where he wrote Latin interlinear and marginal notes in his thinner handwriting.Taking stock of where in the manuscript this annotator began to annotate in Latin further helps us understand how he learned or taught Latin grammar, even as he never left Arabic behind as an interpretational tool.
To follow this annotator further into the primer thus lets us watch an Arabic-speaking Christian's Latin learning through his first language.While we still know very little about the larger picture of these Christians' approach to the Latin language, or its accompanying culture for that matter, we need to remember that the focus here is upon individual Mozarabs' annotating, and that annotating in turn offers a view of individual engagement with a book.Then, through a reading of an individual's notes, we can come to broader conclusions.Mapping out a single Mozarab's movement through a manuscript thus helps us to understand the importance of Latin culture among the Mozarabs that scholars such as Francisco Javier Simonet, Pieter S.J. Van Koningsveld, and Cyrille Aillet have made clear. 32Although he argued for Muslim persecution of Christians, Simonet knew well the strength of Latin culture in Córdoba.Van Koningsveld and Aillet, in turn, have argued with more precision for the importance of Latin culture among Spain's Arabic-speaking Christians, and their scholarship offers the means to illuminate further how someone such as these copyists and annotators turned to foundational texts of Latin learning, although we should not dismiss the depth of Simonet's learning.
The content of the Latin text on folio 9r flows seamlessly from a discussion of verbs to conjunctions and adverbs.Whereas before Copyist A treated something that every sentence must have, a finite verb, he now took upon something that no sentence needs but which in reality copyists, scribes, and annotators frequently wrote.The manuscript here has the Ars minor sixth part, on conjunctions:

Pars sexta Coniunctio quid est: pars orationis adnectens ordinansque sententiam: coniunctioni quot accidunt tria: potestas figura ordo:
‚the Sixth Part: What is a conjunction?A part of speech [which is] fastening and ordering a sentence.How many things happen to a conjunction?Three: power (potestas), figure (figura), and order (ordo)‛. 33ch as he did with verbs, Copyist A came to this sixth part of the text when he wrote of what three things cling (accidunt) to a conjunction.Indeed, above the Latin accidunt, the annotator with a thin Arabic hand wrote as earlier yalḥaqu, meaning ‚to attach‛ or to ‚cling to‛. 34If we employ philosophical terms, he wrote of a conjunction's accidents or grammatical properties.More practically, that he translated accidere as yalḥaqu offers strong evidence of his uniform vocabulary for describing the technicalities of Latin grammar.
In reading of the power (potestas) of a conjunction, Copyist A noted five different types or species: coupled, disjunctive, explicative, causal, and rational. 35His later annotator then dutifully put these in Arabic, as those relating to ‚power‛, ‚the hinge‛, ‚the complement‛, ‚conditions,‛ and ‚reasoning‛. 36The -iyyah ending which ends several of these Arabic terms signifies its abstract quality, much as abstract ideas in Latin are often third-declension nouns ending in -tas, such as libertas, liberty, or potestas, power.From the Arabic root s-l-ṭ, whose basic meaning relates to power, he came to the even more abstract al-sulṭaniyyah, while al-burhāniyyah, from the root b-r-h, relates to reasoning.Furthermore, the Latin itself here merits closer interpretation.For here Donatus challenges his readers, whether students or teachers, to give the examples of these different types of conjunctions.For example, the Latin reads: ‚give (da) the rationales: ita, itaque, enim, etenim, enim vero, quum, quam, propter, quippe, ergo, igitur, ideo, scilicet, preterea, idcirco...‛. 37For copulativas conjunctions, moreover, the hand in thinner Arabic wrote what reads from right to left as: ‚order the sulṭaniyyah (conjunctions),‛ complete with a secondform imperative, naẓẓim. 38Yet for whatever reason, he did not translate the conjunctions themselves, other than etenim, a rationalis conjunction, as ‚because‛. 39he treatment of prepositions in the brief seventh part of the Ars minor reads even more interesting than his notes upon conjunctions.The Latin copying rings familiar by this point: 33  ‚Preposition: what is it?A part of speech which is placed before other parts of speech.Their signification either completes or changes (mutat) or lessens‛. 40 one example of the Arabic notes in this seventh part, the thinner Arabic hand wrote ‚sign‛ or ‚sense,‛ over the Latin significatio. 41To my mind, the Arabic here aptly describes the Latin, as it does so often in the primer.The later annotator made an even more interesting note on folio 9v, as he moved through this brief treatment of prepositions.The Latin offers a didactic sentence with numerous prepositions: ‚put a tribunal: through the wall: near the window: because of discipline.‛Over the neuter accusative case noun tribunal, he interpreted: ‚and this is the pulpit of judgment‛. 42Certainly, he thought of the tribunal in legal terms, and al-ḥukm fits well in this context, among other Arabic legal terms that Mozarabs employed, such as alfarḍ, a legal judgement, a term this annotator did not use here.As a point of comparison, a Mozarab translator of the Psalms, Ḥafṣ ibn Albar al-qūṭī (Ḥafṣ, Son of Alvarus the Goth, fl.889), also employed both these words in legal/religious contexts as he wrote a verse prologue to his Arabic verse translation, as did the author of an Arabic prose prologue to an Arabic prose translation of the Psalms, whose Psalter Ḥafṣ aimed to improve. 43 as he covered Latin prepositions. 44Indeed, when he shortly thereafter read the Latin ‚belonging to the eye-witnesses (penes arbitros),‛ he wrote in Arabic, ‚arbiters‛. 45For a passing moment, the evidence suggests, he thought with legal terms, much as earlier in the manuscript he employed terms which we often see in philosophy, such as al-ṣifah, to describe the parts of speech.
In addition to this second annotator commenting upon conjunctions and prepositions, on folio 9v as well Copyist A wrote of interjections [Plate 1]. 46Here too, he thought in Arabic.Rather than translating the Latin's ‚The Eighth Part‛, Copyist A wrote in Arabic the ‚chapter of exclamations‛ in both the left-and right-hand margins.The effect of his Arabic makes it easy for him or another Mozarab to find this section upon further reading.The Arabic root, from which he worked, ʿ-j-b, covers in its most basic meaning wonderment and amazement.His Arabic note thus offers insight into how at least some Mozarabs thought about interjections as utterances of amazement or surprise. 47he most interesting Arabic notes on folio 9v, however, treat a Latin passage in the lower margin.A separate hand wrote the passage in thinner Visigothic script.The Latin has a thin ductus, and I suggest it comes from the Arabic annotator with the thin script, who made many notes in the opening folios of the manuscript as well.Perhaps most importantly, although erasures muddy the meaning of this Latin exercise that I discuss below, nevertheless it illuminates how an Arabic-speaking Christian approached that language by way of Arabic.In contrast to what we have above, with interesting but at times random notes on Donatus's and Priscian's treatment of Latin grammar, this annotator here put into Arabic nearly every Latin word of the exercise.
To try and make sense of its disjointed content, I have divided the passage into four parts/clauses below.But here I offer the note in full, one that does not come from the Donatus text, complete with its many Arabic notes: ‚Thus, lest he might worry, the helper sends away the weak, lest you know of riches.I shall not be moved once by usury, by the wickedness, by a dart.Then, direct the grimaces from an asp as a refuge of compunction.You repelled us, she [the asp] will be present while you turn your back.Lest at any time…‛ These confusing clauses read somewhat coherently together, but clearly formed a Latin sentence exercise.The Arabic notes which he put above the Latin effectively confirm the individual readings.For after writing ‚thus‛ in both Latin and Arabic, the Latin has a partial erasure, but the Arabic note, ‚worry‛, effectively confirms anxiaretur. 48Above ‚the helper dissipates the weak,‛ he wrote in Arabic ‚the weak ones‛, ‚he dissipates,‛ and ‚the helper‛. 49hese notes read more complicated than they might seem.Note, for example, how he rendered inutiles, meaning ‚useless‛ but also ‚unprofitable‛, as al-fāšlīn or ‚weak‛.Al-nāṣir furthermore describes the Latin susceptor perfectly; indeed, the Arabic reading confirms for us that the Latin does not read susceptos. 50The sense here is the helper takes those in need under his care, for a susceptor is also someone who takes someone or something under their wing.Indeed, while I have not found this exact Latin phrase in a text, I strongly suspect it comes from the cobbling together of Psalm 53:6, in which God is susceptor of a soul, and Psalm 52:6, where he dissipates the bones of his enemies. 51Ne dispexeris, a probable reading which an erasure obscures, has a likely Arabic gloss of lam tataʿallam, in the jussive mood.While the Latin means ‚to see clearly‛ or ‚to perceive,‛ the Arabic more often means ‚to know,‛ in the sense of knowledge rather than recognizing someone, as in the Latin conocere. 52n the second part of the exercise, he again began with a conjunction, writing in both Latin and Arabic ‚one time‛. 53 not leave,‛ ‚usury,‛ and ‚evil‛. 54The passive verb non migrabor he rendered with lā ʾubraḥu: although he did not vocalize the word, I have translated it as a passive to fit better with the Latin, and since the Arabic jibes so well with the Latin throughout the passage.Both verbs have the meaning ‚to leave‛.To my mind, it thus seems that he worked through this Latin not only clause-by-clause, but also word-for-word.He did not write out Arabic sentences translating each of these Latin clauses / sentences that do not fit well together, but rather defined one Latin term after another.
The clause relating to the asp furthermore puzzles.The verb here is dirige, an imperative meaning ‚direct‛ or ‚guide,‛ which he translated with the Arabic tenth-form imperative ‚stick in‛. 55Although someone erased part of the Latin graph, nevertheless the final -e, strong evidence of an imperative mood verb, remains sufficiently visible.So too has someone erased partly sannas, meaning grimaces, but it is the most likely direct object, which he rendered in Arabic as ‚crookedness‛. 56I have here transcribed it as ‚al-ʿiwaj‛ or ‚alʿawaj‛ in large part because I think the annotator expressed the crookedness of one's face as the asp bites.Above tunc, he wrote in Arabic ‚at that time‛. 57Asp he likewise rendered with precision. 58With this note, he wrote a letter above the final jīm, perhaps a yāʾ, and then erased that letter, in the process making the whole word more difficult to decipher.His Arabic notes for refuge and compunction read straightforward: a ‚refuge,‛ and literally ‚it is the repenting‛ for compunction. 59he passage ends with the helper protecting these people from the asp.Aderit, ‚she will be among‛ or ‚she will be present‛ he rendered as laḥiqat, ‚she clung‛ or ‚she entered,‛ which has a feminine subject-quite likely the asp--, and was a verb the other annotators employed on earlier occasions. 60Furthermore, the Latin aderit is in the future tense, while this Arabic annotator most certainly rendered the Arabic in the māḍi perfect tense.Dum he translated as iḏā, both of which mean ‚when‛ or ‚while,‛ and from this particular note, we see that he translated mundane conjunctions that one employs often in both Latin and Arabic.Then, for avertis, ‚you turn your back‛, he employed ‚you turn away‛ in Arabic. 61As so often in this primer, we cannot know exactly what the annotator thought as he wrote his final Arabic note in the exercise, but with it he set a more precise timeframe in the Arabic ‚at night‛, than he did in the Latin, where he wrote ‚at any time‛. 62n addition to this Latin exercise in four clauses, this annotator of Latin and Arabic in the margins furthermore did a translation of sorts with a verse from the Vetus Latina Psalms, that is, one from a series of Greek to Latin translations predating Jerome's Hebrew to Latin Vulgate Psalter. 63The verse is Ps.36:32: ‚considerat peccator•iustum•et querit perdere eum,‛ or ‚the sinner considers the just man, and wants to destroy him‛. 64He translated sinner, peccator, as al-ḫāṭiʾ, precisely that in Arabic. 65Furthermore, he put the preposition bi ‫)ﺒ(‬ before al-sadq or alsidq, which translates to ‚concerning‛ or ‚about‛ the just man.Bi thus works much like the Latin preposition de, which does not figure in the Latin here. 66For querit, he employed an eighth-form verb: yultamisu, which indeed means ‚he seeks,‛ complete with an, the conjunction ‚that.‛Over perdere eum, meanwhile, he wrote a verb here meaning ‚he kills him‛. 67Indeed, if one takes into account the preposition bi marking the direct object of the verb, here reads something like a sentence in Arabic as well.
In sum, this annotator's Latin exercises illuminate his word choices and his movement between two very different languages.In the larger framework of the manuscript, he furthermore turned to these translations seemingly after he had made his notes upon Donatus' explanation of Latin adverbs and conjunctions.That is, he went from annotating Donatus' explanation of the Latin language, to rendering Latin clauses/phrases as well.He translated the Arabic faithfully, moving word-for-word, so much so that we can nearly see him switching between Latin and Arabic repeatedly.But he did not do these clause-length exercises elsewhere in the manuscript, as he kept translating through Donatus' and later Priscian's text.What we therefore see in these examples illuminates a shift in how this annotator employed Arabic to understand his Latin better, whether for his own learning or for teaching other Arabicspeakers. 62BCT MS Cajón 99.30, fol.9v.laylan; umquam; Seybold, p. 541, umquam aliquando [lacking Arabic]. 63The translation most likely comes from the Vetus Latina Hispana. 64 Likewise, a separate Latin hand wrote in the folio's margins.The copyist, whether Copyist A or another, and his annotators, worked much as did the others, here reading of the four conjugations of regular Latin verbs.These four conjugations stem from their infinitive ending: -āre, -ēre, -ere, and -īre. 69Knowing the conjugation of a verb in turn helps one remember its four principal parts, the four parts of a verb from which one can arrive at the other forms.The folio marks an important shift in texts.The copyist of the folio here built upon Priscian's treatment of first-conjugation perfect tense verbs in his Institutes. Louis Holtz has pointed to the similarities between the manuscript and Books 9 and 10 of Priscian's grammar.Yet I note at the outset here that the copyist of this folio, and its numerous annotators, did not follow Priscian completely, but rather added in forms of the perfect passive participle to the text as well, where Priscian had only written the first-person singular present and perfect tenses, such as ‚domo domui‛. 70More precisely, much of what he copied deals with Book IX, 32 of his Institutes, where the grammarian notes that first-conjugation verbs beginning with a preposition, a prefix, tend in have perfect endings in -ui rather than -avi. 71 As he read about first-conjugation verbs on folio 13v, the Arabic annotator with a thin hand, I suspect, thought of how to render Latin morphology in that language.It reads a little odd that this important lesson follows the treatment of adverbs, but that oddity stems in part from the shift to Priscian, after copying the Ars minor.As one example of a first-conjugation verb, the copyist worked through the verb ‚to create,‛ creare, although he did not write the infinitive, but rather ‚creo te, creavi, creatum‛ or ‚I create you, I created, having been created‛. 73bove creo te, the Arabic annotator with a thin hand wrote ‚we create‛ in Arabic, naḫluqu, which I suggest is effectively the royal we in Arabic.It likewise bears mention here that in their vernacular Arabic Moroccans now use the first-person plural in place of the first-person singular. 74He certainly employed this form on numerous occasions, although he left no further Arabic notes for the first-person singular perfect or the perfect passive participle, the other two principal parts which he copied.
This copyist and annotator also learned through Priscian of exceptions to the rules of Latin verb conjugation.When the annotator with thin Arabic script read the verb domare, to pacify, he annotated in much the same way as with creare.After reading: ‚exceptis domo te•domui•domitus‛ or ‚with I pacify you, I pacified, having been pacified having been excepted‛, he wrote narwuḍu above the first-personal singular present tense.This Arabic verb does indeed mean to pacify, as well as to tame an animal.He thus understood the Latin quite well, above all in that domare forms an exception to the first-person conjugations, those with an -are infinitive, since domui and domitus are its third and fourth principal parts, and end in-ui and -itus, rather than the -avi and -atus endings which almost all other first-conjugation verbs have. 75The Latin verb increpare also forms an exception, with increpui and increpitus for its third-and fourth-principal parts.The Arabic annotator wrote the Arabic inkasara, a seventh-form verb meaning ‚to be broken‛, which aptly describes the Latin increpo: ‚I crash‛ or ‚snap‛. 76I suggest that he employed a seventh-form verb here because he wanted to capture the Latin's intransitive meaning.Note as well, that here the Latin lacks a direct object such as te, ‚you‛, in contrast to the above example with the transitive verb creare, where the first principal part reads te creo.
This annotator worked his way through verbs whose meanings can trip up readers as well.When he copied ‚I wash, I washed, having been washed‛, he also made clear that lavatum is ‚different with regard to its source (causa) on account of latum‛, which is the fourth principal 73 BCT MS Cajón 99.30, fol.13v. 74 part for ferre, to carry, and thus means ‚having been carried‛. 77That is, the Latin copyist warned readers not to mix-up lavatum with latum, and he did so, he offered an interpretation of sorts on Priscian, who likewise treats this Latin.The annotator then wrote in Arabic ‚this is the sources‛ above ‚different with regard to its source‛. 78Al-ʿuyūn, the plural of al-ʿayn, can also mean substance or essence, which fits well with what the copyist wrote of lavare's sources, that is, its principal parts.It also bears mention that as lavare deals with water, so too ʿayn, ʿuyūn has ‚fountain or spring‛ as one of its most basic meanings.Without speculating too much, we can wonder if the annotator remembered this as he wrote of lavare's principal parts, or if he simply played a word game as he learned and/or taught. 79n addition to the Arabic annotator's treatment of the Latin, someone, quite likely the same annotator, wrote Latin synonyms to Priscian's examples.For mico, with a lesser meaning of ‚to shine,‛ this annotator wrote fulgeo. 80He thus offers evidence that he worked with the meaning of these verbs in some depth, as mico's primary meaning is ‚to dash‛ or ‚to flee‛.The ductus of his Latin furthermore looks similar to that of the Arabic, for example with reference to the thickness of the letterforms, suggesting that he held his pen the same way as he wrote in his two languages.
This Latin annotator kept up this mode of reading over the course of the folio, in which he also treated the other verb conjugations.As he read, ‚he ran out the spirit (deficit spiritum) [is also] for I fight (dimico)‛, he wrote pugno, another verb for ‚to fight‛.This Latin admittedly reads strangely, as the verb deficit most often is intransitive, along the lines of ‚the spirit (spiritus) expired,‛ but the copyist has spiritum in the accusative. 81The annotator may have made a note here to explain a seemingly lesser-known word, dimico, through a more common one such as pugno.In the bigger picture, what thus makes folio 13v so important is that it illuminates Arabic-speaking Christians moving back and forth between Latin and Arabic, not only employing Arabic to understand Latin, but now also noting Latin synonyms as well.
In a similar manner, this annotator read of verbs that have multiple preterit forms and perfect passive participles: ‚I thunder, I thundered (intonavi) or I thundered (intonui), having been thundered (intonatum), or having been thundered (intonitum)‛. 82 he wrote facere sonitum, ‚I make a noise‛. 83Again, he offered up a Latin synonym, in this case employing a transitive verb and its direct object to explain an intransitive verb, intono.Note as well how the Latin lesson that Priscian offered here is that intonare has two forms each for its perfect tense and the perfect passive participle, similar to the exceptions which he treated above.Yet the evidence which this annotator left us suggests he wanted to explain the word's meaning, rather than focusing upon how intonare breaks the rules of first-conjugation verbs. 84s yet another example of this synonym annotating, he wrote iaceo above the verb cubo.Both of these verbs mean ‚to lie down‛. 85It seems that here on folio this annotator keenly worked through his Latin vocabulary.He did so while reading of a verb's principal parts and the exceptions to the rules by which one conjugated these verbs, as when he read of intonavi or intonui as ‚I thundered‛. 86The evidence he has left us shows him turning to the writing of Latin notes, which in turn complement the Arabic ones.In general, these annotators in Latin and Arabic, who likely are the same person, wrote more in Donatus and Priscian's language, indeed the language in which Arabic-speaking Christians also read many theological works, as they moved deeper into the primer.
In much the same manner as above, he wrote a Latin interlinear note, ‚I stand•I confirm,‛ when he read the Latin ‚I agree•I agreed•having been agreed‛. 87Here he explained that ‚I stand‛, sto, effectively forms the base of the verb ‚I agree‛, consto, with the prefix constrengthening the word, while also making clear that confirmo is a synonym of consto.He thereby offers insight into how he saw Latin words complementary to one another, and how a prefix such as con-changed the meaning of a word. 88Perhaps most importantly, the note reads much as others from the other annotators, leaving us with a view of hermeneutical continuity across several Arabic-speaking Christians.
As another example, here of the second declension, he interpreted ardeo⸱arsi in the Priscian text with ‚diligo uel cupio‛.While the primary meaning of the verb ardere is ‚to be on fire,‛ this annotator saw it as being in love, the meaning of diligo and cupio. 89He thus reminds us of the subtlety with which these Christians approached their Latin primer, and more specifically the 83 BCT MS Cajón 99.30, fol.13v.intono; sonitum facio.He abbreviated sonitum here, so that the graph reads soni(tum).The abbreviation which made has a horizontal line running from the foot of the n, crossing through an i.Since there is an unabbreviated i, precision they brought to his Arabic and Latin annotating.While this annotator made many Latin synonym notes that we can group together, we also need to remember that he put much thought into each of them: he was not working mindlessly.
As a final example, he commented in both thin Latin and Arabic upon the verb adstare, ‚to stand at‛.He in Latin ‚I am present (presento)‛ when he read ‚I stand at•I stood at•Having been stood/placed‛. 90This annotator furthermore wrote in Arabic, with naḵṭiru, ‚we are present in one's mind‛. 91Naḵṭiru in turn looks very much reminiscent of the other notes he made, and reads like those in which he conjugated Arabic verbs in the first-person plural muḍāriʿ, which we have here, rather than the first-person singular. 92This annotator, who almost certainly made earlier Arabic notes in thin script, thus turned increasingly to writing Latin, but never left his Arabic as well.
In addition to these interlinear Arabic and Latin notes, yet another annotator also filled the margins of folio 13v with Latin.One such note treats first-person and at times second-person singular verbs: Propello and velo do not relate to each other, but this evidence suggests that he practiced conjugating verbs yet again, although he did not write out the third-person singular or any plural forms.He then wrote merito, either a verb here meaning ‚to earn frequently‛ or the adverb ‚worthily‛, and neco, ‚I kill‛. 93Why he departed here from conjugating verbs in the second-person remains a mystery, but he returned to the practice with Iugulo•iugulas, or ‚I slit the throat• you slit the throat‛. 94is note thus covered a range of verb types here.Velo offers a standard first-conjugation verb.But he has written loquor, ‚I speak,‛ offering an example of a deponent verb, one with passive morphology but active meaning.So too did he work with verbs that have prefixes, much as the Priscilian text teaches, with propello and assero and assentio, where the preposition ad assimilates into as.What he wrote for assentio reads asse(n)tio, with a horizontal line marking the abbreviated -n.This word thus cannot be assertio, the noun which derives from assero.Brief as it is, this note lets us watch Latin learning in practice, but what the annotator wrote here remains difficult to interpret-it reads now much like random thoughts, with examples sharing little in common.
The end of the note puzzles, but offers hints of the depth in which this Latin annotator worked through a lesson on verbs.Metablo may well be an abbreviation of metabolum, which the Dictionary of Medieval Latin in British Sources attests as a change of place or journey.That is, it likely comes from a syncopated Greek word, transliterated into Latin.The final word reads peruello, to pluck, and fits with the above verbal forms with prefixes, as per can mean through but also intensifies the meaning of a verb.The whole note too illuminates the eccentric character of copying in this manuscript.He clearly thought of how to form verbs, but the words by which he did so read quite random.
In sum, I have focused upon folio 13v because it shows numerous Iberian Christians, at least one of whom was a Mozarab, learning and/or teaching Latin verbs with numerous meanings.For scholars, working with the Arabic and Latin of this folio lets us watch Arabicspeaking Christians moving between two very different languages.These annotators worked through these verbs in some depth, with the annotator who wrote thin Latin and Arabic notes likely doing so after noting the technical language of Latin grammar in the opening folios of the manuscript, where he was in dialogue with Copyist A. His thorough annotating thus offers us a window into how the Mozarabs turned to Latin, a liturgical language that they revered, while speaking Arabic and a romance dialect.In a very real way, the annotators' poring over these Latin verbs, among many other notes, furthermore reminds us that Latin grammatica lay at the root of the Mozarabs' reading of Latin Bibles, as well as works of Pope Gregory I (d.604), Isidore of Seville (d.636), and others.This is no small point, as scholars need to consider both Latin and Arabic evidence to understand the Mozarabs' intellectual communities.While Mozarab, a term of convenience as I noted at the outset, means ‚those having become Arab,‛ they nevertheless read Latin as well. 95

Flora and Fauna in the Latin Word List on Folios 26r-27r
Quite strikingly, these annotators wrote in Arabic throughout the primer.So much so, indeed, that I cannot treat the whole manuscript in the depth with which I have covered the foregoing case studies.Yet briefly I should note that they worked through Donatus, alongside the above attributed excerpts from Priscian, before a copyist put an incipit to that sixth-century grammarian's Institutes in the middle of folio 16v, which text runs until folio 22r.Indeed, whoever bound these quires put the Institutes between sections of Donatus, whose ‚Definitions of nouns and other parts of speech,‛ readings from his Ars maior, begins on folio 22v. 96Yet more interestingly, to my mind, at least two hands copied two Latin word lists into folios 26r-27r, each full of arcane vocabulary that has little to do with grammar per se, and which while lacking an incipit, nevertheless follow the explicit for a section of Donatus on interjections, one which adds to the earlier treatment of interjections in the manuscript.The hand on folio 26r-26v looks quite like that of Copyist A, although perhaps a little rushed, and markedly different from folio 27r. 97he lists do not appear in the critical edition of Donatus' Ars maior, but rather very likely come from the Roman Suetonius (d. after 122 CE). 98They complement well the annotating which these copyists did throughout the manuscript, with this Latin vocabulary supplementing Donatus' and Priscian's grammars, and further illuminating the ties between these Arabicspeaking Christians and classical education.This Latin vocabulary thus sheds light upon how copyists made these grammarians' influential works into a text of their own, all the while 95 As Arabic-speaking Christians, they read both Islamic texts and those of Eastern Christians as well, on which see Thomas E. Burman, Religious Polemic and the Intellectual History of the Mozarabs, 1050-1200 (Leiden: E.J. Brill, 1994), esp.Chapters 2-3. 96BCT MS Cajón 99.30, fol.22v Incipiunt definitiones nominum ceterarumque partium orationis.I should note here as well that this same copyist wrote ‚It is done, thanks to God time after time,‛ Finit deo gratias crebro, on folio 22r.He then began to copy folio 22v. 97His Visigothic script widened as he wrote, as if he tired while at his task, so that it looks much like the script on folio 13v. 98I discuss this authorship further below, after treating the terms in the list.
following the traditions of Latin education in the Middle Ages.The lists also suggest, to my mind, that someone later bound folios from numerous copyists into a single Latin primer, which may also explain the earlier unattributed excerpts from Priscian. 99s I demonstrate below, many of the terms that the copyists of these lists wrote in Latin describe fauna.They thus added to texts of Latin grammar and verb conjugation, as the lists form a tool which one memorized or consulted to better their Latin, or to work through a difficult text.As at least two hands made the list, it surprises little that they are not in alphabetical order, although the copying dovetails with the Latin-Arabic glossary which Pieter S. J. Van Koningsveld brought to light in the 1970s, in which readers turned to Arabic to aid their Latin. 100While these copyists did not make a Latin-Arabic glossary, we will see that they did in fact note some Latin terms in Arabic, which language they planned into the body of the text. 101In sum, this back-and-forth between the word lists and the earlier conjugation of Latin verbs illuminates how these students or teachers made their primer a personal book, one that suited their own interests.
When this copyist -again, perhaps Copyist A-wrote folio 26r, he effectively created it in two parts.The upper half of the folio has Donatus' grammar, with a section on interjections.He ended with an explicit for that part, and then, in the middle of the folio, he began to copy vocabulary with little regard for alphabetical order.His first list then moves across the folio, but after eleven lines, roughly halfway down the folio, he made a column in the middle of the folio, in which he wrote a second list marked off from the first list.The second fauna list thus stands out well in the middle of the folio, with the first list surrounding it on three sides: left, above, and right.In both lists, he wrote of birds and mammals complete with the calls/noises that they make. 102ithout doubt, readers of this list, including the copyist, learned and/or taught how to describe the natural world.In the lower-left quadrant of the manuscript, in the first list, he focused only upon fauna and their calls in Latin.He noted, for example, that ‚eagles o scream‛. 103Here and elsewhere he wrote an underlined o after the bird's name, likely standing for omnes, in that all of a particular species make a certain call.In general, what he wrote here reads as a coherent whole: hawks, for example, screech, a noise the copyist rendered with the 99 Louis Holtz treats the wordlists briefly in his manuscript description, Donat et La Tradition, p. 385. 100 Van Koningsveld, The Latin-Arabic Glossary, esp.pp.5-43. 101 verb plipiare. 104Among many other calls, he furthermore noted: ‚thrushes rouse or recite,‛ which captures this sound well. 105For those bees, meanwhile, he employed vombire and vombilare.He switched v for b, as these Christians did, and so, he thought of bombire and bombilare, both meaning ‚to buzz‛. 106These onomatopoeias do not translate well into Arabic, which may explain the absence of Arabic notes here.
As I noted above, this copyist also made a second list, in the middle of folio 26r's lower half.Here too he filled this space with vocabulary of natural world, complete with animals and their sounds.Deer, he remarked, ‚roar‛ or ‚bellow,‛ a term which again shows us the copyist describing their sound/call. 107He put hares in his list too, and their ‚wailing‛ or ‚murmurs‛. 108e wrote of domestic animals as well, treating the sound of horses with the Latin innire, and cats with meolare, neighing (hinnire) and meowing. 109hat he made these lists puts him firmly in a tradition of Latin learning.Alvarus of Córdoba (fl.850s-60s) refers to animal sounds in his fourth carmen.He writes of tigers raging, for example. 110It bears mention as well that in another letter, to Iohannes Hispalensis, mid-ninth century bishop of Córdoba, he humbly states that his simple prose does not follow the art of Donatus.Iohannes replied that holy men had shunned Latin education, and that Christians ought to take notice of Jerome's life, although he does not mention that Donatus had taught Jerome (d.420) his Latin.Alvarus then argues at some length against Donatus and Latin grammar in Christian education, and likewise seemingly makes a reference to Donatus' teaching Latin to Jerome: ‚whence the holy and apostolic men seem to speak through the liberal art of Donatus, they must not be believed to have been instructed by him, but by he who gathered this same thing slightly for the same gentile‛. 111Alvarus famously lamented the decline of Latin learning in mid-ninth century Córdoba, and his response to Iohannes puts forth a careful view of how Christian authors took up Latin education.The copyists in this manuscript, whenever they lived in Toledo, thereby offer an excellent view of how closely he followed a traditional Latin curriculum.
But the list in this manuscript has many animals of which Alvarus does not speak, and it may well come from an earlier source.Suetonius' list, for example, matches much of what the copyist put forth. 112Bees in the Roman's work likewise bombire or bombolire, and thrushes (turdos) trucilare, similar to truciare in the manuscript.The lists read similar enough to suggest that the copyist worked from a tradition passing through Suetonius, although I have not pushed my research further to see if he copied from another author or manuscript which also holds Suetonius' text. 113hat this copyist wrote in Latin thus reads quite differently than the earlier Latin material concerning adverbs and verb conjugation.Indeed, his Latin illuminates exactly what these Arabic-speakers taught and learned when they turned to the liturgical language of northern Europe.They read of the Psalms, as we saw earlier, and furthermore absorbed a large amount of vocabulary that teaches little about Latin grammar or the language's morphology, but rather arcane topics such as a thrush's call.And while these copyists and annotators may well have thought about the manuscript in a different way than I have interpreted it, even as I have read closely the evidence they left, nevertheless we should note how their primer's material changes over the course of its folios.
In the lower-right quadrant of folio 26r, moreover, he made several Arabic notes interpreting those Latin words [Plate 3].They have little to do with the natural world, in contrast to much of the folio.In the larger picture of the primer, his Arabic notes add relatively little to what we have already seen, but they clarify just how frequently these Arabic-Christian copyists turned to their first language in order to understand Latin, a conclusion we can make without belittling his abilities in either language.For example, after he wrote the Latin ‚pushing‛ or ‚undisciplined‛ with its nominative and genitive forms as one finds in a dictionary, he wrote the Arabic ‚ravenous‛. 114Certainly, his Arabic and Latin do blend here as elsewhere, but he seemingly sought to define procax more precisely, perhaps as relating to appetites or temperament, as al-ḫātif suggests.
Other notes on folio 26v, likely from the same copyist, have clearer script and equally interesting content.Yet as we see when he annotated ‚low-bred (vernaculus)‛ as ‚jocular (hazlī)‛, the interpretation of these notes still proves difficult. 115Here, for example, one needs to know that vernacular does not simply refer to a vernacular language, but more specifically a language of lower register, or indeed to low-born people. 116Doubtless, the Mozarab who taught and learned from this primer thought of Arabic and Latin as higher-register languages, in which he read his scripture, while speaking a Romance dialect and colloquial Arabic. 117Yet judging from the Arabic and Latin evidence together, he thought of vernaculus as signifying a joke and thus rendered it so into Arabic. 118ust after this note, meanwhile, he annotated ‚to be sustained (sustineri)‛ as ‚bearing‛ or ‚toleration (iḥtimāl)‛. 119In a wise translating move, he captured the meaning of a Latin passive infinitive with the eighth-form of an Arabic root, which commonly has an intransitive/passive meaning.Rather than an Arabic verb, however, he put an Arabic noun for this Latin passive infinitive.Furthermore, while the primary meaning of sustinere is ‚to sustain‛, his Arabic fits much better with ‚to put up with‛, a lesser meaning of the Latin verb.As in the above translation of vernaculus, he here employed Arabic, his first language, to help him define the Latin more precisely.His attention to detail was in sum admirable, and he certainly left scholars of Arabic-speaking Christians -and not just in Iberia-valuable evidence for language learning and teaching. 120n folio 27r, another copyist furthermore made Arabic notes of similar tone.Here, Pieter Van Koningsveld noted that the Latin words all begin with b, as part of a Latin glossary, although I add here that folio 26v also has entries beginning with c, such as ‚Celibates, who [are] without a living conjugal partner‛, so that in the end product from various copyists, we have here a partially alphabetical list. 121But first, with regard to the Latin on which he commented, he wrote Visigothic script with a bulkier ductus than that of folio 26r-v.Brief as it likewise wrote of the allegorical mixing of blood and wine during Jesus' passion. 126Martin Irvine, moreover, masterfully showed that medieval Christians made Donatus' grammar, a text that effectively taught Latin through Virgil, into an ecclesiastical Latin text. 127In the larger picture, the Latin vocabulary that he copied reads rare enough to merit interpretation, so that his explanations of vocabulary do not mean that he lacked intelligence-far from it.As a scholar of the pre-modern world often has shelves of dictionaries, so too did Arabic-speaking Christians have grammars and word-lists for Latin. 128his list, in sum, illuminates the tools by which Arabic-speaking Christians / Mozarabs sharpened their vocabularies.Indeed, quite likely these copyists rendered their Latin vocabulary in Arabic when they did not find a way to do so in Latin, as if they lacked the Latin vocabulary to describe a heifer, baccula/vaccula, to cite one above example.Yet they more often than not had the ability to describe difficult/new Latin vocabulary through Latin words that they knew, or at the least knew where to look up the terms in a dictionary or other word-list.To my mind, the list offers rather strong evidence that they already knew some Latin when they made the primer, and likely employed it for review or reference.Indeed, when we remember that they also copied their material on adverbs and conjunctions before delving into verbs, it seems quite likely that they already knew how to read Donatus' Latin text on those adverbs and conjunctions.That fact, along with his list of relatively obscure vocabulary, suggests to me students with some prior knowledge of the language, or masters preparing lessons for students who already knew some Latin.

Copyist A on Latin Nouns and Adjectives
The copyists of these lists thus worked through obscure Latin nouns and commented upon them in both Latin and Arabic.More precisely, the evidence from the lists suggests they took less interest in the morphology of Latin nouns, as opposed to the building of their vocabulary.Folios 33v and 34r, however, on both of which Copyist A worked with Donatus' lengthier Ars maior, illuminate engagement with declensions and the case system for nouns and adjectives.Moving slightly into speculation, I want to suggest that these annotators and copyists, in general, may have known the case system when they began earlier to work with Latin verbs, the heart of the primer, more thoroughly.The annotator who did practice sentences / Psalter verses in the margins of earlier folios, among his many other Arabic notes, for example, had at least some inkling of Latin morphology and how to translate it to Arabic.
Copyist A, here near the end of the primer, did not underestimate the importance of nouns to a sentence, even if the annotators on verbs offer the strongest evidence of Arabic engagement with the Latin primer.Donatus opens folio 33v by explaining what a noun is: ‚a part of speech with case, a body, or a thing of its own or communally, properly as Rome Tiber, communally as urbs flumen…‛. 129Then, he noted here that nouns have five accidents: quality, comparison, gender, figure, and case.Over figure, he wrote ‚the apparition‛ in Arabic, while comparison he glossed as ‚the equals and the negation which is a friend‛. 130The evidence suggests that the apparition, or figura, is the word which one saw.His note for conparatio seems to mean that nouns and adjectives in a particular declension, a group of nouns or adjectives, take particular case endings.When he wrote of the ‚negation which is a friend,‛ he likely wanted to say that some words have the same endings, despite being in different declensions: as in urbes (3 rd Declension) and res (5 th Declension).The passage on the whole reads very much like the introductions to verbs and adverbs, where he opened with the vocabulary that one needed to talk about the language intelligently.
In these last few folios, he moved rather quickly from topic to topic.Underneath his discussion of the accidents of a noun, for example, he also made a list of endings for the genitive case, dealing with possession, and the dative case for indirect objects, and ablative case, in which nouns and adjectives act as an agent of means with many verbs, and also the object of some prepositions.He copied in fidelibus, diebus, manibus, for examples of the -ibus ending which helps one see nouns in the third, fourth, and fifth declensions in the ablative and dative plural, in addition to third-declension adjectives in the ablative and dative plural.Yet he also made a peculiar statement about reading biblical languages: ‚all Latin or Hebrew or Greek nouns which come (III) to the rule (II) of Latinity (I)  Perhaps he thought of a first-form maṣdar, or verbal noun, which can also mean ‚to be spread‛ or ‚to circulate‛, but whose form ought to be al-rawāj.The root in its second-form, which often intensifies the meaning of the first-form, also means ‚to circulate,‛ but then the verbal noun ought to read al-tarwīj, and I see no evidence here of a tāʾ nor yāʾ.Nevertheless, the Latin and the Arabic together strongly suggest Copyist A thought of the circulation of coinage as he made his note.
In a very concrete way, folio 34r brings us back to the beginning of the manuscript.For while Copyist A and the others engaged with Latin grammar from adverbs to verbs and nouns, among other topics, they throughout the manuscript employed Arabic to understand their Latin better.They ran through material which one now learns in a beginning Latin course, such as the principal parts of a verb, but also clearly aimed to grow their vocabulary, with words that one does not see frequently.Perhaps, then, they as students had already learned some Latin before they created the primer, which helped them to learn more words while reviewing Latin grammar.Likewise, if they taught, Arabic offered the means to impart Latin vocabulary to their pupils.
This manuscript on the whole offers excellent evidence for Latin learning among the Mozarabs.The copyists and annotators clear that at least some of these Arabic-speaking Christians delved into the learning of the language, with Arabic offering a tool for understanding difficult terms as needed.Doubtless we can think of Copyist A, the Arabic annotator with thin script, and the others working through the language in a rather intense way, as Donatus and Priscian guided them.We see as well, then, that at least some Mozarabs learned Latin with reference to the classical world; indeed, they did so through a late Roman text.Certainly, the vocabulary on folio 34r, with emperors such as Galba and Nerva, along with the dariga coin, lends weight to this classically-tinged learning.On the whole, they thus offer us a case study in which we see something of how these Mozarabs approached the learning of their other liturgical language.

Conclusion
Broadly speaking, the Mozarabs read in both Arabic and Latin, much as Eastern Christians delved into Greek, Syriac, Coptic, and Arabic. 137While the scholarship of Francisco Javier Simonet, Pieter Van Koningsveld, and Cyrille Aillet, among others, has made clear on a large 137 As the readers of Collectanea Christiana Orientalia know quite well, although we have only begun to compare the experiences of Arabic-speaking Christians in the Eastern Mediterranean with those in al-Andalus / Medieval Spain.
scale that the Mozarabs read in these two languages, I in turn have further shown how an individual copyists and annotators taught and/or learned Latin. 138Like so many other Arabicspeaking Christians in Iberia, they employed their first language to understand better the language of Latin Christendom, which the evidence suggests they knew to some degree before they made the primer.In addition to working with Priscian, the copyists hint at making Donatus's grammar, a commentary of sorts upon Virgil, into a Christian text, much as Martin Irvine showed for other medieval Christians. 139Pieter Van Koningsveld, moreover, was absolutely correct to place this primer alongside the Latin-Arabic glossary which he interpreted at length, and to note that it demands a more thorough reading than he gave it.These copyists and annotators furthermore illuminate the Mozarabs' intellectual life between roughly the ninth and twelfth centuries.Latin culture did not die off during these centuries, as Míkel de Epalza argued, but rather thrived, even as they translated their scripture into Arabic. 140The Mozarabs turned to Bibles, chronicles, the Etymologiae of Isidore of Seville, and Pope Gregory I's Moralia in Iob, among other codices. 141This primer, in turn, lets us watch Mozarabs either learning or teaching the language at the root of those codices.The ecclesiastical education of these Arabic-speaking Christians lay in grammar, in a manner reminiscent Augustine of Hippo (d.430) in On Christian Teaching, where he moved from the signification of sounds, for example, to working through difficult scriptural readings.To be clear, we have no evidence of these Mozarabs turning to Augustine -although we have evidence that other Mozarabs did so. 142Yet their copying and annotating fits well with the broader contours of what Augustine and numerous other grammarians believed between the tenth and twelfth centuries: that linguistic learning formed the root of knowledge.
For these Arabic-speakers, moreover, Latin fueled the culture of al-ʿajam.This term literally means ‚the barbarians,‛ and in eastern contexts often referred to Persian Muslims, but in al-Andalus, when Christians deployed it, they meant ‚the Latins.‛As Travis Zadeh has shown, 138 Francisco Javier Simonet, Historia de los Mozárabes; Van  Arabic offered a means for Persian elites to promote their own vernacular writings; I suggest something similar happened here with Arabic-speaking Christians and their Latin culture. 143hat they taught and learned the language of the Latins so thoroughly demonstrates that these Arabic-speakers did not think of their northern neighbors pejoratively; rather, they revered their culture, including scripture. 144In turning to the language of the ʿajam, these anonymous copyists and annotators therefore illuminate how one entered into a rich thought-world built upon luminaries such as Gregory I and Isidore of Seville.To understand these Arabic-speaking Christians, or Mozarabs, we therefore must bear in mind their intellectual debts to the Latin and Arabic languages. 143Cf.Travis Zadeh, The Vernacular Qurʾan: translation and the rise of Persian exegesis (London: The Institute for Ismaili Studies, 2012), p. 166. 144In this regard, I differ from Jason Busic's argument that Ḥafṣ ibn Albar put himself in a ‚linguistically awkward position‛ as he saw Arabs as his opponents with regard to translation, while calling Latin the foreign language.ʿAjam, ʿajamī, and the like had little pejorative force when referring to Latin.
Koningsveld has more recently suggested that Ḥafṣ ibn Albar al-Qūṭī may have authored the original version of the glossary in the ninth or tenth century, with the twelfth-century glossary an expansion of that text.See Van Koningsveld (ed.),The Arabic Psalter of Ḥafṣ ibn Albar al-Qūṭī: Prolegomena for a Critical Edition (Leiden: Aurora, 2016): pp.61-82.Fernando González Muñoz, meanwhile, wrote briefly of a Donatus manuscript in the monastery of Cosme y Damian in Abellar, in the kingdom of Castile-León.A catalog of medieval books in El Escorial, Real Biblioteca de San Lorenzo de El Escorial, MS R-II-18 mentions the Liber conlationum artis grammatice of Donatus.See Fernando González Muñoz, Latinidad Mozárabe, p. 18, n.26.Les Mozarabes, p. 321.I have not attempted to read the erased text.
Here in the Latin primer, in turn, we see a Mozarab who clearly knew some Arabic legal vocabulary, employing it 40 BCT MS Cajón 99.30, fol.9r.Donatus, Ars minor, 8, p. 600 reads ‚quae praeposita aliis partibus orationis…‛ 41 BCT MS Cajón 99.30, fol.9r.al-dalālah; I have not found significatio in Seybold.Marie-Thérèse Urvoy, ‚Quelle est la part d' originalité dans la production écrite mozarab?‛, in Matthias Maser and Klaus Herbers (eds.),Die Mozaraber: Definitionen und Perspektiven der Forschung.Geschichte und Kultur der Iberischen Welt Vol. 7 (Berlin: LIT Verlag, 2011), p. 74; see as well Geoffrey Martin, ‚An Anonymous Mozarab Translator at Work‛, in Miriam Lindgren Hjälm (ed.), Senses of Scripture, Treasures of Tradition (Leiden: E.J. Brill, 2017), 125-152, which interprets author of an earlier Arabic prose prologue and translation in greater depth than did Urvoy, who focused upon Ḥafṣ.Pieter S.J. Van Koningsveld has argued that the prose prologue comes from Ḥafṣ ibn Albar as well, with Biblioteca Apostolica Vaticana MS. ar. 5, which has the prose prologue, as an Arabic prose translation from an earlier translator.I thus now refer to the author of the prose prologue, accepting that it may be Ḥafṣ ibn Albar, although it seems equally probable that Ḥafṣ incorporated the earlier prologue of another Mozarab, the translator of the Psalms in Arabic prose as in BAV MS ar. 5, into his own translation project.Cf.The Arabic Psalter of Ḥafṣ ibn Albar al-Qūṭī: Prolegomena for a Critical Edition, edited and translated by Pieter S.J. Van Koningsveld (Leiden: Aurora, 2016).
BCT MS Cajón 99.30, fol.11r.consideratpeccator•iustum•etqueritperdereeum.See Ps 36:32 (Vulgate): considerat peccator iustum et quaerit mortificare eum.He also wrote exercises on folios 10r-v.65BCTMSCajón99.30,fol.11r.al-ḫāṭiʾ;Seybold,p. 370, peccator al-mudnib.66Thereareclearlytwodots under this initial bāʾ, which I suggest are its normal one dot and a kasrah marking its vowel sound.Furthermore, this cannot be the exclamatory particle yā, as the graph is enclitic, while yā is not, and the noun following yā never takes the definite article.Above, we saw Arabic-speaking Christians learning how to write about Latin verb conjugation.Folio 13v, meanwhile, offers evidence of them putting the earlier readings to use.68The Latin copyist of these folios is difficult to pin down: while Copyist A did folio 13r, the script on folio 13v looks more compact, save for a few lines which do have more word spacing [Plate 2].I thus cannot say definitively if a new copyist made this folio and those immediately following it, or if Copyist A wrote with a more disciplined script here.But the earlier annotator wrote in his thinner Arabic script here too, as well as Latin interlinear notes commenting upon vocabulary.
have transcribed the graph as sonitum.If he had written son(um), I would expect to see a horizontal line above the n.
See his very interesting work ‚Between Latin Theology and Arabic Kalām‛, p. 575; see as well Geoffrey Martin, ‚An Anonymous Mozarab Translator at Work,‛ in Miriam Lindgren Hjälm (ed.), Senses of Scripture, Treasures of Tradition, pp.125-152.