FROM ANTONY OF TAĠRIT TO THE ARABIC VERSION: THE SYRIAC TECHNICAL VOCABULARY OF RHETORIC AND THE MIGRATION OF

Rhetoric was part of the borrowed Greek educational system of the Enkyklios paideia, together with logic and grammar. However, the technical terminology had to be adapted, so the question is which strategies were used to create the vocabulary of Syriac rhetoric. This paper aims at analyzing some meaningful loanwords, adaptations, calques and native words used to build this specialized lexis. A manuscript containing the Syriac version of Aristotle’s Rhetoric was never found, and we need to rely upon other texts that dealt with this topic: Antony of Taġrit’s Books of Rhetoric, a part of Bar Šakko’s Book of dialogues and a part of Bar Hebraeus’s Cream of wisdom. If we wish to have a deeper appreciation of this vocabulary, however, we need to take into account the Arabic version of Aristotle’s Rhetoric, since it displays some interesting peculiarities that can help us in the reconstructive process of what the Syriac version might have looked like. A very useful tool would be the compilation of a database comparing the Syriac lexis


found in the abovementioned texts with the ones used in the
Greek and Arabic versions of Aristotle's book: actually, this method already highlighted some surprising phenomena that will be presented in this paper, in order to show how trilingual comparison can be an important tool to shed some light on the translation movement of both texts and cultural patterns that started in Late Antiquity.
The path that brought Aristotelian rhetoric towards the Arabic milieu is a long and complicated one, though it has occasionally been conceived as straightforward, from Greek to Arabic, diminishing the role played by the Syriac world. To weaken the position of Syriac in this field there is the fact that no manuscript containing the Syriac version of Aristotle's Techne Rhetorike has been found so far. At the present time, then, the first work dealing with rhetoric in Syriac belongs to, apparently, the 9 th century and this absence of relevant works on the topic composed during the so-called 'classical period' unfortunately led some scholars to the wrong impression that the Syriac world was somehow unaware of Aristotelian rhetoric until late, and that a translation of this work was never produced.
This paper aims at showing that, despite the late date of the Syriac works, they can probably be considered the products of a tradition of interpretation of Aristotle's books and also the result of a lively discussion on the topic. This interest in Aristotelian sciences, and specifically Rhetoric, is testified, for instance, by the letter written by Patriarch Timothy I to Mar Pethion, in which he asks his delegate to find out whether the monastery of Mar Mattai possesses commentaries or scholia in Syriac (or other languages) on Poetics, Topics, Sophistical Refutations and Rhetoric. 1

SYRIAC WORKS ON RHETORIC
Concerning the Syriac contributions to rhetorical studies, the most important -and yet still little known -one is the work of Antony of Taġrit: the author composed what looks like a handbook designed for students of rhetoric, divided into five volumes, that explains in details various aspects of the 'art of speaking'. 2 The books are based on Aristotelian material but also display some important differences that show influences of other Greek philosophers like Plato. 3 Antony's efforts also Professor John Watt for his invaluable support and his enlightening remarks, Professor Lucas Van Rompay for his precious insights and comments, and Aaron Butts for his suggestions and advice.  van Bekkum, et al. (Leuven: Peeters, 2007), 136. Moreover, as J.W. Watt, "Rhetorical education and Florilegia in Syriac", in Les auteurs syriaques et leur langue, Études Syriaques 15, ed. M. Farina (Paris: Geuthner 2018), 96, ft. 2, recently remarked, it would be more appropriate to say that Antony composed two treatises on rhetoric, a longer one in four books called 'On the science of Rhetoric', and a shorter one, that coincides with what is normally accounted as The fifth book of Rhetoric, called 'On the ornamentation and decoration of words'. Regardless, since the five books division is widespread and accepted, this paper, as Watt's, will use the conventional reference for the sake of commodity. 3 On the points of contact between Antony and the Greek philosophers see J.W. Watt, "The Syriac reception of Platonic and Aristotelian rhetoric", ARAM 5 (1993), J.W. Watt, "Eastward and westward transmission of classical rhetoric, in Centres of learning and location served as a starting point for portions of at least two other works, namely Bar Šakko's Book of dialogues 4 and Bar Hebraeus's Cream of wisdom. The credit for finding out that a summary of Antony's works was hidden behind a certain part of Bar Šakko's Dialogues belongs to Martin Sprengling 5 and Rubens Duval at the beginning of last century, 6 while for the printed version of this part one can rely upon abbé Martin's edition. 7 John Watt completed an edition, translation, commentary and provisional glossary of the rhetorical part of the aforementioned Cream of wisdom. 8 When we talk about Antony of Taġrit's works, some preliminary remarks are due.  147-150. 4 The first part of this book deals with grammar, rhetoric and poetics. The rhetorical part is based on section one and five of Antony of Taġrit's treatises. J. Bendrat (1968). "Der Dialog über die Rhetorik des Jacob bar Shakko", in Paul de Lagarde und die syrischen Kirchengeschichte, ed. Göttingen Arbeitskreis für syrischen Kirchengeschichte (Göttingen: Lagarde-Haus, 1968). 5 M. Sprengling, "Antonius Rhetor on versification with an introduction and two appendices", The American Journal of Semitic Languages and Literatures 32: 3(1916), and "Severus Bar Shakko's Poetics, Part II", The American Journal of Semitic Languages and Literatures 32: 4 (1916 Dionysius of Tell Maḥre, 9 a remark that was embraced by Ignatius Barsoum. 10 However, there appears to be no concrete evidence to prove or reject this date, except for what Sebastian Brock and Lucas Van Rompay found inside the Deir al-Surian collection: 11 a portion of a manuscript (Ms. Deir al-Surian, Syr. 32) containing Antony's works, that turned out to be the lost part of the British Library manuscript Add. 17208. The two scholars have interpreted the handwriting as belonging to the 9 th century, confirming the idea previously put forward by William Wright about the date of the BL portion. 12 Concerning Antony's treatises on Rhetoric, which are complex and still not entirely studied, the translation and edition of the fifth book has been provided by Watt,13  PhD thesis, which unfortunately was never published, and made a provisional translation and commentary, followed by a reproduction of the Harvard manuscript that she used for her study. 14 Sadly, the other books never experienced a translation or a proper critical edition, having also suffered from the complicated history of their manuscripts' transmission and the poor state of conservation of some of their parts. 15 An attempt has been made by Eliya Sewan d-Bet Qermez, who published the complete set of Antony's books: 16 the author used as his sources a copy, created in 1947, of the Jerusalem manuscript MS 230 17 collated with another manuscript written by Asmar al-Khoury of ʿAinwardo (and, for the fifth book, with Watt's edition). 18 The work is by no means a critical edition, but it still represents the only extant printed sample of the whole set of treatises. 19 A problem posed by Antony's texts and their relationship with Aristotle's works is that the Syriac author never mentions the Greek philosopher by name. We can also assume, 14 P.E. Eskenasy, Antony of Tagrit's Rhetoric Book One: Introduction, partial translation and commentary, Unpublished PhD thesis (Harvard University, 1991). 15 For a detailed account on the manuscripts' transmission see J.W. Watt, "Antony of Tagrit as a student", 264-266, J.W. Watt considering, for instance, the conceptual differences in Antony's understanding of Aristotelian categories compared to the Greek "archetype", that Antony might have owned a different version of Techné Rhetoriké, which has not been transmitted to us. 20 As Watt phrased it: If Antony learned his theory directly from a Greek handbook, it must have been very different from any manual known to us, or he must have treated it with considerable freedom. Since in philosophy Syriac has preserved versions of late antique compilations of which the originals have been lost, it is by no means impossible that behind Antony's treatise lies some lost Greek rhetorical Art. But the curious mixture of the familiar and the strange which this work presents to us makes it possible to suppose that in the Syro-Mesopotamian area there was a tradition of rhetorical teaching which had developed in its own particular way. 21 It is also necessary to establish whether or not Antony was able to read Greek, and to what extent. However, the scanty information on Antony that we have at our disposal makes it hard to answer this question with certainty. Moreover, as Aaron Butts highlighted, 22 the use of loanwords by a writer tells nothing about whether or not he was exposed to the second language, since it only testifies that the language in use, in this case Syriac, had contacts, at a certain moment, with another language, that provided it with the loanwords.
The reading of Antony's treatises may represent an additional resource for determining the period he lived in, specifically with regard to his prose and the language he uses. The impression one gets is that of a quite late Syriac, clearly not in its early classical stage: the texts display, for instance, plenty of finite verbs instead of the abundance of participles characterizing Late Aramaic style. 23 His use of the first person singular (alternated sometimes with the plural) is another interesting feature, probably to be explained in the light of the ultimate aim of his work, which was intended as a hard copy for students. This idea is also supported by the richness of examples that the author provides, which would not have been necessary if we were to exclude a didactic aim: this abundance ultimately makes the work not easy to read. In the light of all this, Antony's prose often appears heavy and repetitive, but testifies to a certain knowledge of the works of his forerunners, that are frequently quoted. 24 Barsoum, on the other hand, had a different opinion, accounting Antony's style as 'grand and eloquent', and reporting that he also studied Greek. However, he did not provide evidence to sustain these statements and he did not state the basis on which they rest. 25

THE ARABIC VERSION
As previously stated, Antony's books are based, to a certain extent, on Aristotelian material, even though it was probably not written in Greek: actually, Antony might also have relied upon a Syriac translation of Aristotle's Rhetoric, 26 as did the author of the only extant Arabic version. Actually, this Arabic translation is preserved in a single manuscript, the Bibliothèque Nationale de France Ms. Arabe 2346 in Paris, 27 and displays some glosses in which the author explains that the work is based on two other Arabic manuscripts and a Syriac version. 28 The 25 I.E. Barsoum, The scattered pearls, 383-384, and I.E. Barsoum,Geschichte ,[298][299]. 26 Watt ( production of this text must have required a certain 'critical' method, given that the author claims to have had at hand a good Arabic copy and a poorer (but still valuable) other Arabic one, while, whenever the two were not sufficient to him, he turned to a Syriac copy, apparently considered the soundest by him. 29 The extant Arabic version was also the one used by Hermannus Alemannus to compose his Latin version of the text, a circumstance that provides us with some help to decrypt certain obscure passages comprised in the Arabic manuscript. 30 If the author, or rather the editor, 31 of the Arabic manuscript was, as it seems, Ibn al-Samḥ, 32 the text must have been written before 1027, which is the year of his death, and this would imply that the work significantly predates those of both Bar Hebraeus and Bar Šakko. The rhetorical portion of the Parisian manuscript has been edited for the first time by 29 U. Vagelpohl, Aristotle's Rhetoric, 51-61, expresses some doubts about the existence of a Syriac version, saying only that 'it is not unlikely that the Rhetoric and Poetics were at some point translated into Syriac' (p. 57). Admittedly though, later he says that 'So far, all we can say with some assurance is that Syriac authors were probably acquainted with the text but that does not require the existence of written or textual knowledge. As it is, the only reliable witness for the existence of a Syriac translation -which we are not in a position to date -is the Arabic Rhetoric itself. There are two types of evidence that have been derived from the Arabic translation: Ibn al-Samḥ's marginal notes referring to a Syriac version; and terminological features of the translation that point to a Syriac source text' (p. 59). Vagelpohl then criticizes M.C. Lyons, Aristotle's Ars Rhetorica, II-VIII and XXV-XXVI, and E. Panoussi, "The unique Arabic manuscript", for taking the existence of the Syriac version for granted, and concludes that the evidences provided by them are far from being conclusive (pp. 60-61).
ʿAbdulraḥmān Badawī 33 and later by Malcom Lyons, who also provided a Greek-based commentary and a bilingual Greek-Arabic and Arabic-Greek glossary. 34 None of these works display a translation, though a partial translation was made by Uwe Vagelpohl, 35 leaving the task of the realization of a comprehensive version in a modern language still a desideratum. Additionally, the lack of an Arabic-based commentary, that uses the actual strings of text displayed by the edition, makes this work hard to handle. Besides, the extant Arabic text is complicated and occasionally impenetrable: moreover, it differs from the alleged Greek archetype in numberless cases, probably due also to the misunderstanding of certain Syriac words that were mistranslated into Arabic. 36  traces of so-called 'demotic forms' and colloquial variants, mixed with inexplicable grammatical mistakes. 37 This, combined with Aristotle's elliptic prose, which must have puzzled both the Syriac and the Arabic translators, and that the matter might have not been too familiar to them both, brought about the extant text and its critical issues. 38 The first Arabic translation must have been produced during the so-called 'Greek-to-Arabic Translation Movement', a label used to denote a massive wave of interest, started around the 8 th century and continuing roughly until the end of the 10 th , among Arab intellectuals towards Greek knowledge and works, which triggered the translation of a great deal of books. The major contribution to this topic has been provided by Dimitri Gutas, who is however skeptical about the importance of the role played by the Syrians as the channel through which the Greek knowledge passed to the Arabs. The same attitude, specifically about the role of a Syriac version of Rhetoric as the source of the Arabic one, is expressed by Vagelpohl, who concludes that: In the end, the textual evidence may be sufficient to make a Syriac intermediary "likely", but it does not amount to conclusive proof. Irrespective of Watt's arguments for a Syriac translation of the Rhetoric antedating Ḥunayn ibn Isḥāq and its potential role in the production of our Arabic version, the contribution of the Syriac translators to the Aristotelian rhetorical tradition in the Islamic world seems negligible beyond faint echoes and vague influences. 40 Watt's arguments in question are those put forward in the lucid analysis hosted in the introduction to his edition and translation of the Cream of wisdom: 41 in short, Watt writes that the Syriac translation was composed before the famous translator Ḥunayn ibn Isḥāq (808-873 CE) 42  much closer to the Old Arabic 43 than to any Greek recension known to us, yet quite frequently it uses the very same Greek (loan) words as are present in the Greek text of Aristotle, where the Old Arabic (and Ibn Sīnā) use a native Arabic term (p. 8).
This paper will use Watt's solid argument 44 in order to show how the presence of loanwords vs. native words points to the use of a Syriac text as the source of the Arabic version. Moreover, in the light of the massive work of translation from Greek into Syriac, and later into Arabic, that started in Late Antiquity, there is no reason to infer that the lack of a manuscript containing a Syriac version of Aristotle's Rhetoric implies that it never existed at all.

A TRILINGUAL APPROACH AND SOME EXAMPLES
In order to understand how the technical vocabulary of rhetoric developed throughout the Syriac and Arabic environments, a comparative trilingual approach, including also Greek terminology, proves to be a good starting point. Actually, rhetoric has not been investigated extensively as other sciences have been, probably also due to the fact that, normally, it does not fit into those categories such as, for instance, philosophy (at least in its modern concept), that have received greater attention by scholars. Admittedly, in a recent paper about the creation of a Syriac philosophical lexicon, King discussed the opportunity of including the subject within this kind of corpus, 45 arguing that Bar Šakko and Bar Hebraeus treated Rhetoric as part of the Organon, as was customary for the users of the Alexandrian extended version of Aristotle's Organon, 46 even though it would not be considered today as philosophy strictu sensu, and he concludes by saying that Arabic philosophy can serve as a tool to understand late Syriac authors. 47 Regardless of the contemporary position of rhetoric inside the philosophical sciences, a linguistic analysis comparing the vocabulary of the three languages -Greek, Syriac and Arabic -allows new perspectives about the history of this science and helps to trace the path of some phenomena involving the migration of words throughout centuries.
This kind of analysis provides, inter alia, new evidences about the knowledge that Antony of Taġrit might have had of Greek, and about what kind of texts he was looking at while composing his treatises. Since he was, most likely, a teacher of rhetoric, he should have been at least acquainted with Greek tradition, but considering his (apparently unaware) use of Greek technical terms, he does not appear too familiar with rhetorical works composed in this language. He might also have used a Syriac version of Aristotle's work rather than its Greek original, as already stated, a conclusion that might be supported by the confusion he makes between certain concepts, which he uses as interchangeable and that would not have been misinterpreted if his source had been a Greek one. An interesting example is provided by the words ‫ܐܦܘܕܝܟܣܝܣ/ܐܦܘܕܝܟܣܐ‬ and ‫ܬܚܘܝܬܐ‬ used, for instance, in The fifth book of Rhetoric. Antony uses them both with the meaning of 'demonstration', even though they correspond to two different Greek technical terms: the first one is a loanword from ἀπόδειξις, while the second one corresponds to Greek παράδειγμα. 48 Antony uses ‫ܐܦܘܕܝܟܣܐ‬ three times in his fifth book, while he uses ‫ܬܚܘܝܬܐ‬ constantly, with the meaning of 'example', 'argument', and 'epidictic' 49 as well. Therefore, a couple of times, he uses it with the clear meaning of 'demonstration', suggesting that he did not have the Greek technical word in mind, that would have suggested ἀπόδειξις instead. ‫ܐܦܘܕܝܟܣܐ‬ can also be found in the Arabic translation of Rhetoric as ‫ا‬ ّ ‫,افودقطيقي‬ 50 rather a loanword from ἀποδεικτική, but more often this text features ‫.تثبيت‬ Attempted conclusions to explain this latter phenomenon will be provided later. The situation is represented also in Bar Hebraeus's Cream of wisdom: the author uses ‫ܐܦܘܕܝܩܛܝܩܐ‬ -again a loanword from ἀποδεικτική -and ‫ܬܚܘܝܬܐ‬ as synonyms, indicating that there was no clear perception of the Greek technical term at this stage as well. 51 Therefore, the correspondence between Greek, Syriac and Arabic words, in this case, fluctuates and appears not to be fixed. 48 It should be stressed, though, that Syriac displays as well the loanword ‫,ܦܪܕܝܓܡܐ‬ meaning 'paradigm'. However, this loanword is absent from Antony's fifth treatise, but it can be spotted in Bar Hebraeus However, going back to Antony's fifth treatise, the author seems at least partly aware of Greek technical vocabulary when he uses loanwords such as ‫ܠܟܣܝܣ‬ < λέξις 'style, way of speaking' or ‫ܛܘܦܣܐ‬ < τύπος 'figure, metaphor', even though it is possible to spot some other examples looking quite straightforward while concealing interesting phenomena: this is the case of some loanwords bearing initial ῥ. According to Butts, 52 the history of Greek loanwords bearing ρ with spiritus asper mirrors their chronology. To be precise, initial spiritus asper was pronounced in Attic Greek as a voiceless glottal fricative /h/ until Late Antiquity, when it apparently ceased to be pronounced. 53 Butts states, for initial ῥ, that it 'was realized as a voiceless alveolar trill /r̥ / in Attic Greek' that turned into a voiced alveolar trill /r/ in the Roman period. 54 The Syriac graphic reflex of this phoneme was ‫ܪܗ‬ for loanwords acquired by the 5 th century, while later it is represented by a simple ‫,ܪ‬ since, according to Butts, the pronunciation of ῥ changed in the meantime. This ῥ phenomenon is represented in the orthography of the word for 'rhetoric' itself, as Antony's text preserves both ‫ܪܗܛܪܘ‬ ‫ܬܐ‬ and ‫.ܪܝܛܘܪܝܩܝ‬ According to the spelling of the reflection of the initial ῥ, the former example is the more ancient form, which entered Syriac before the 5 th century, while the latter is a new loanword reflecting the shift in the realization of initial ῥ. Under the morphological point of view, the two borrowing strategies appear very different: ‫ܪܗܛܪܘ‬ ‫ܬܐ‬ , the older one, bears the Syriac suffix -uṯā used to derive abstract substantives, 55 meaning that Syriac borrowed the word ῥήτωρ in the first place and only afterwards derived from it a word for 'rhetoric'. 56 The more recent loanword ‫ܪܝܛܘܪܝܩܝ‬ appears to be the true borrowed reflex of ῥητορική, which does not need any derivational suffix. Combining the statements given so far, it is possible to trace the following timeline: considering that, according to Butts, ‫ܪܗܛܪܘ‬ ‫ܬܐ‬ form should be placed around the 5 th century, and considering the  -Boston: Brill, 2008), 358-360) collected a series of nouns, belonging to the Christian Neo-Aramaic dialect of Barwar, that use this strategy: the derivational suffix works both with native nouns and with loanwords, like the Arabic adjective wājib ('necessary') that turns into wajəbuṯa ('duty'). 56 S.P. Brock, "Greek words in Ephrem and Narsai: a comparative sample", ARAM 11-12 (1999-2000), 441, and A.M. Butts, "The use of Syriac derivational suffixes", 213. T. Harviainen, "On the loss", 29, failed to recognize this adaptation strategy, saying instead that ‫ܪܗܛܪܘܬܐ‬ is a loanword from ῥητορῆσαι, even though there is no reason to account a loanword of an abstract noun as coming from a verbal form. orthography of the initial ῥ as simple ‫,ܪ‬ the terminus post quem of the ‫ܪܝܛܘܪܝܩܝ‬ loanword is to be placed after the 5 th century. 57 Moreover, the presence of both loanwords in Antony's text and the interchangeable way in which they are used mean that they were not perceived as having different connotations.
Later in time, the situation changes even more, as represented in Bar Hebraeus's Cream of wisdom: the author uses sparsely ‫,ܪܗܛܪܘܬܐ‬ ‫ܪܝܛܘܪܝܩܐ‬ and ‫,ܪܗܝܛܘܪܝܩܐ‬ since, probably, at that point the origin of loanwords was not meaningful anymore and the reasons for writing ‫ܗ‬ were not clear. Considering also that the writers and scribes were acquainted with both the ‫ܪܗ‬ and the ‫ܪ‬ reflexes in "rhetoric", virtually nothing prevented them from creating another outcome, which looks like a recent loanword according to explanations given above, but bears the orthography ‫.ܪܗ‬ The same oscillation between two spellings of the word for rhetoric is to be found in the Arabic text as well: 58 we have thus ‫ة‬ ّ ‫ريطوري‬ and ‫.ريطوريقا‬ Hence, the first one, ‫ة‬ ّ ‫,ريطوري‬ appears to bear the same borrowing strategy of ‫,ܪܗܛܪܘܬܐ‬ employing a derivational suffix belonging to the recipient language, while ‫ريطوريقا‬ looks very much like Syriac ‫,ܪܝܛܘܪܝܩܝ‬ preserving the final <q>. Since the two loanwords show borrowing patterns resembling the ones found in Syriac, and since they are not used anywhere else in the Arabic tradition, they are likely to have reached the Arabic version of the Parisian manuscript by means of a Syriac text rather than directly from Greek. Moreover, none of these loanwords is used in the Arabic commentaries to Rhetoric, such as, for instance, Ibn Sīnā's, Ibn Rušd's, or even Ibn Ṭumlūs's, since they all feature ‫,خطابة‬ the same term used also by Badawī 57 A.M. Butts, "The use of Syriac derivational suffixes", 213. 58 There is no trace of the Greek aspiration in the Arabic spelling.
in entitling his edition of the Parisian manuscript. 59 Actually, ‫خطابة‬ is the substantive commonly used in Arabic to indicate 'rhetoric', instead of the Greek loanword: the presence of ‫ريطوريقا‬ and ‫ة‬ ّ ‫ريطوري‬ in the Parisian manuscript looks like an unicum. Indeed, this treatise is not entitled ‫الخطابة'‬ ‫'كتاب‬ as the other treatises on the subject, but rather ‫ريطوريقا'‬ ‫المسمي‬ ‫ارسطوطالس‬ ‫,'كتاب‬ followed later by the specification 'which means ‫.'الخطابة‬ About the outcome of Greek ῥ, something resembling the aforementioned situation occurs when ῥ is in middle position, specifically double ῥ (ῤῥ). Butts explains how the spelling of the Syriac form of παῤῥησία, meaning 'to speak freely', changed across the centuries, thus representing a variation of the results of Greek gemination in Syriac: 60 before the 4 th century the word was written as ‫,ܦܪܪܣܝܐ‬ in the 4 th , 5 th and 6 th centuries we have ‫,ܦܪܗܣܝܐ‬ 61 while the 7 th century displays a  13 (Paris: Vrin, 2006). As a matter of fact, though, Ibn Rušd uses ‫ريطوريقى‬ in the title of the first of the three books of his commentary, but immediately followed by its Arabic equivalent. 60 A.M. Butts, Language change , 68-69 and 80-82. 61 This is the only form reported by T. Harviainen, "On the loss", 29, who states that the word was used in the early 4 th century, but A.M. Butts, Language change, 78, convincingly argued that, in Syriac, the outcome of the Syriac phoneme was not represented by a voiceless glottal stop until the 6 th century. To explain the reasons why the graphic representation of this phoneme changes from word to word in Syriac, Harviainen says that the pronunciation of Greek ῥ started to be identified with ρ at a certain point (p. 32). He also adds that it is not possible to provide an earlier date than the 4 th century in which the change occurred, and that it is impossible to study the development of ῥ in various positions inside the word due to the variation between ‫,ܦܪܗܣܝܐ‬ ‫ܦܪܪܣܝܐ‬ and ‫.ܦܐܪܪܣܝܐ‬ However, Antony of Taġrit's work preserves another form: in the final part of the fifth book we find ‫.ܦܪܝܣܝܐ‬ The word is preserved in this orthography in all the manuscripts used by Watt for his edition, namely the Harvard one, Barsoum's and Rahmani's. 62 Since it would be unlikely that an error made in the antigraph is repeated in all later copies without a single attempt at emendation, what we have might correspond to the orthography employed around the 8 th -10 th centuries.
There is another notable example of the orthography ‫ܦܪܝܣܝܐ‬ in a painted inscription preserved on the walls of the monastery of Deir al-Surian, in Egypt. 63 The inscription is dedicated to the abbot Maqari, native of the city of Taġrit, who passed away the 10 th of May 889 CE: the whole painted group, including both written parts and images, was commissioned 'scantiness of the material' (later he adds that it is likely that the change took place in the beginning of the 4 th century for the medial position and developed later for the initial one, p. 50). Some decades later, luckily, Butts succeeded in this task. See A.M. Butts, "The integration" and A.M. Butts, Language change. Moreover, it must be highlighted that A. Wasserstein, "A note on the phonetic and graphic representation of Greek vowels and of the spiritus asper in the Aramaic transcription of Greek loanwords", Scripta Classica Israelica 12 (1993), 206, wrote that the Syriac ‫ܗ‬ is probably the reflex of the Greek η following ῥ, rather than the aspiration of the ῥ. This hypothesis was rejected by Syriac scholars such as Brock, in S.P. Brock "Greek words in Syriac: some general features", Scripta Classica Israelica 15 (1996), 256. 62 Moreover, the aforementioned Jerusalem manuscript 230, belonging to the 14 th century, shows the same orthography ‫,ܦܪܝܣܝܐ‬ suggesting that this is not an accident or a single scribal mistake. For a discussion on Watt's sources see J.W. Watt, The fifth book, edition, XI-XXV. 63 The inscription and its related paintings have been studied by K. Innemée, G. Ochała and L. Van Rompay, "A memorial for Abbot Maqari of Deir al-Surian (Egypt)", Hugoye 18: 1 (2015). shortly after his dead by his son Yuḥannon. 64 The Syriac inscription, at line 22, displays our loanword in the same orthography used by Antony of Taġrit. What seems noteworthy is that the spelling is linked to well educated people from Taġrit: Yuḥannon, who acquired the title of abbot after Maqari's death, knew both Syriac and Coptic, since he added a second inscription in this latter language to his father's epitaph, and appears able to translate from one language into another. What we have here is then an educated Taġritan that uses the orthography ‫ܦܪܝܣܝܐ‬ in a formal context, rather than the more classical ‫ܦܐܪܪܝܣܝܐ‬ or ‫ܦܪܗܣܝܐ‬ spellings. From another Taġritan context, belonging again to the (early) 9 th century, comes another occurrence of ‫:ܦܪܝܣܝܐ‬ the Book of Divine Providence by Cyriacus of Taġrit. 65 The manuscript containing this text, Jerusalem, St. Mark's monastery 129, 66 bears the date 806 and, thanks to its colophon, we know that it was copied from Cyriacus' autograph only few years after its composition, while Cyriacus was still alive. The text displays the substantive ‫,ܦܪܝܣܝܐ‬ the adjectival form ‫ܦܪܝܣܝܝܬܐ‬ and the adverb ‫.ܦܪܝܣܝܐܝܬ‬ 67 Therefore, the orthography was certainly in use in the Taġritan area during the 9 th century.
However, it should be mentioned that Bar Šakko's Book of dialogues preserves a different reading of the word ‫:ܦܪܝܣܝܐ‬ even though he summarized Antony's work, Bar Šakko was composing in an autonomous way and might have normalized the word according to the 7 th century Yaʿqub of Edessa's 64 K. Innemée As partly shown so far, Antony's text shows plenty of Greek loanwords, normally accommodated and arranged in various ways, but, as already remarked, sometimes the same concept conveyed by a loanword can be found in its Syriac translation and used side by side with the other one, in apparent free variation. Occasionally, Antony uses the Greek loanword first, followed by its Syriac translation a little later, in what can be interpreted as a 'reader-oriented' composing strategy. A good example of this tendency is provided by the word for 'syllable' within the part of the treatise devoted to metrical patterns. Antony provides a long excursus on metrical units in Syriac poetry, with examples and explanations of the various meters based on syllabic patterns and on the way in which they can be combined and exchanged. Greek words", 254, states that a possible explanation of the presence of -s-at the end of ‫ܐܣܛܘܟܣܐ‬ would be the analogy with the Syriac result of Greek words ending in -os. However, even though the Greek technical term is indeed στοιχεῖον, the Syriac and consequently Arabic loanwords may not descend from it, since there would be little means of explaining, apart from Brock's idea of analogy, the fall of the Greek finaln. Alternatively, one might posit a case of semantic contamination with the words στίχος or στοῖχος, which would offer an explanation for the consonantal pattern. It looks like a merging of terms, which have different meanings in Greek, into a single one in Syriac, which has the meaning of the first one, στοιχεῖον, but the aspect of the second one, στίχος or στοῖχος. Moreover, in this scenario, the realization in Syriac of the final -s would be in harmony with other cases given by Brock. It should be also noted that Brock (ibid.,257) states that the form ‫ܐܣܛܘܟܣܐ‬ is the older one, while in texts from the 6 th century onwards the word is rather spelled as sṭukaye. Hence, the spelling used by Antony of Taġrit would be the more archaic one. On the use of a prosthetic ‫ܐ‬ before loanwords beginning with ‫,ܣ‬ see A.M. Butts, "The integration", 26. 72 G. Endress and D. Gutas, GALex, fascicle 2, 218-220.
‫.اسطقساة‬ 73 This strategy applies also to the aforementioned case of ‫ا‬ ّ ‫افودقطيقي‬ and ‫:تثبيت‬ it might be inferred that certain Greek loanwords were perceived as the proper 'technical term' indeed, but were no longer understood by the readers, making it necessary to clarify them. 74 This tendency of pairing translations and transliterations of foreign words appears to be used by translators of all eras, even though, in the specific case, the incidence of the phenomenon probably increases a bit after the 8 th century, as a consequence of the general decrease in the understanding of Greek. Some concluding remarks should address the final portion of Antony of Taġrit's fifth treatise: apart from involving metric and rhetorical figures, this book contains also a small but interesting final chapter on the law of assonant letters, not entirely preserved. 75 This last part, dealing with what we could call 'rhyming strategies', displays a lexis on the edge between the grammatical and the rhetorical domains: in a complex and intricate way, the author explains which are the best ways to keep a constant rhyme at the end of each verse, and how to make words alliterate the most. According to him, the more letters words share, the better the result would be. He engages then in a series of examples, starting from the keeping of the ‫ܦܪܨܘܦܐ‬ at the end of each verse-ending word. ‫ܦܪܨܘܦܐ‬ has been translated here as 'personal suffixes', but should rather be translated as 'person', as in the first meaning given by 73 Nevertheless, Ibn Rušd does not seem to use this strategy. See M. Aouad,Averroès,vol. 2,26. 74 H. Hugonnard-Roche, "Lexiques bilingues", 8, in discussing the bilingual Greek-Syriac philosophical lexicons, highlights that, in the Baghdadi lexicon that he is describing, occasionally, after the transliteration of a Greek term, the translation is replaced or followed by an explanation of the word itself. This correspond to the strategy represented in the text under examination here. 75 J.W. Watt, The fifth book, trans., XX.
Moberg, 76 or better as 'facie' (figures), as in Balzaretti, 77 for the reason that follows. 78 Antony lists both various pronominal object suffixes and the verbal conjugations, but also the enclitic forms of the pronouns, which are not written altogether with the noun. To suppose that he was not aware of the differences looks somehow forced, since, to our understanding, he was a teacher and he himself must have been educated according to the Syriac version of the enkyklios paideia (including grammar, logic and rhetoric). 79 What he probably means is that it is more elegant to keep the same ending all along the poem, since a sudden change would alter the structure of rhyme and the alliteration patterns. 80 This explanation is corroborated by the fact that he usually chooses for his examples verbs and substantives with the same number of syllables and differing only by one letter, the so-called 'minimal pairs'.

CONCLUSIVE REMARKS
Hopefully it has emerged from this analysis that the comparative study of Greek, Syriac and Arabic rhetoric is productive and stimulating, apt to change some of the previously fixed perspectives on the topic, which saw the Syriac world as relatively uninterested in rhetoric and, consequently, Arabic rhetoric as detached from any Syriac influence. The data presented here are just an introduction to the potential that a trilingual comparative analysis has in helping the scholarly community in the rhetorical field: the connections and interdependences that have been proposed in the present paper hope to show how much the aforementioned older perspectives need to be revised. Moreover, it should be stressed that the Arabic version preserving Aristotle's translation is an incredibly rich source of vocabulary, since it stands quite different from the one known in other treatises on Arabic rhetoric. It should be borne in mind that rhetoric is a science which has an autonomous tradition in the Arab culture: when referring to the Aristotelian science, ‫خطابة‬ is the word to use, which means 'philosophical rhetoric', while when referring to Arab-Islamic rhetoric, ‫بلاغة‬ is in charge, in the sense of 'literary rhetoric' or, as in Ghersetti, 'rhétorique de l'illocutoire'. 81 However, this distinction is only partly reliable, since, actually, none of the types of Arabic rhetoric appears to be entirely detached from foreign influences. 82 The creation of a dictionary of Syriac rhetoric has been a desideratum for a long time, and the studies on the migration of words from Greek to both Syriac and Arabic are still in their early days. This study hopes to have shown how a triple focus serves as a key to reconstruct the history of the transmission 81 A. Ghersetti, "Quelques notes sur la définition canonique de Balāġa", in Philosophy and arts in the Islamic world. Proceedings of the eighteenth congress of the Union Européenne des Arabisants et Islamisants held at the Katholieke Universiteit Leuven, Orientalia Lovaniensia Analecta 87, eds. U. Vermeulen and D. De Smet (Leuven: Uitgeverij Peeters, 1998), 57. For a detailed account on the differences between ḫiṭāba and balāġa, and on the history of Arabic rhetoric see Ph. Halldén, "What is Arab Islamic Rhetoric? Rethinking the history of Muslim oratory art and homiletics", International Journal of Middle East Studies 37: 1 (2005), who explains how much, in fact, the two branches were both influenced by Greek knowledge and how they interacted one with another. See also P. Larcher, "Mais qu'est-ce donc que la balāġa ?", in Literary and philosophical rhetoric in the Greek, Roman, Syriac and Arabic worlds, Europaea Memoria. Studien und Texte zur Geschichte der europäische Ideen 66, ed. F. Woerther (Hildesheim: Olms, 2009), 197-213. 82 Ph. Halldén, "What is Arab Islamic Rhetoric",28. of words and can be used as a powerful philological tool. 83 See S.P. Brock, "Diachronic aspects of Syriac word-formation: an aid for dating anonymous texts", in V Symposium Syriacum 1988, ed. R. Lavenant (Roma: Pontificium Institutum Studiorum Orientalium). Even though this study concerns "native" Syriac neologisms, one can easily make good use of it even with loanwords, since Brock's categories apply to this field as well. Moreover, the creation of calques based on Greek words, which use Syriac derivational suffixes, can be easily spotted in Syriac rhetorical lexis as well. On this topic see, for instance, one of the possible explanations provided for the word ‫ܩܘܡܕܘܬܐ‬ in M. Nicosia, "La Rhétorique d'Aristote dans les milieux syriaques et arabes: histoire d'un épisode de transmission intellectuelle dans l'Antiquité Tardive", in La philosophie en syriaque, Études Syriaques 16, eds. E. Fiori and H. Hugonnard-Roche (Paris: Geuthner, 2019), 276-277. 84 J.W. Watt, "Syriac rhetorical theory", 249.