Fragment of the Month: August 2022

The Greek background of the grammatical fragments attributed to Ḥunayn b. Isḥāq (ENA 3173.1, T-S Ar.31.30, Rylands B 3652-3653, AIU IX.A.6)

Julia G. Krivoruchko

T-S Ar.31.30 verso

In 2020 N. Vidro [2020b, see also 2020a] published four Genizah fragments, which she identified as belonging to Arabic Inflexion According to the System of the Greeks (Aḥkām al-i‛rāb ‛alâ mad̠hab al-yūnāniyyin̄) احكام الاعراب على مذهب اليونانيين, a work of the ninth-century Nestorian intellectual Ḥunayn b. Isḥāq. Following the editor, I presume that the above fragments come from a single text, which for the sake of compactness I will call ‘Greek-inspired Grammar’ (GIG).¹ GIG reached us as five folios in an eleventh-twelfth-century hand: a well-preserved New York JTS ENA 3173.1 outlines the author’s summary of a previous discourse and his intentions as to the next one;² four other folios (a well-preserved CUL T-S Ar.31.30 and fragmentary Manchester, Rylands B 3652, 3653 and Paris AIU IX.A.6) deal with matters of inflection and classification. 

Below I do not discuss the attribution of GIG’s prototype to Ḥunayn b. Isḥāq, neither do I deal with the details of the edition irrelevant to my observations. Having read GIG for the first time, I felt a need to get justice for its Greek roots and to contextualize it inside modern extended grammar studies. As I continued, the logic of research repeatedly pushed me towards Syriac material, into which I was reluctant to plunge, persisting in Greek-centred methodology. Vidro’s pioneering publication will certainly attract more attention in future, and specialists in Syriac and Arabic grammar will fill the gaps.³ 

I am grateful to N. Vidro for the opportunity to think and read about material I would have not considered otherwise.

 

Parts of speech: GIG and Dionysius Thrax 

In the introduction Vidro explores the theoretical content of GIG comparing it to standard Arabic grammars and remarks on its similarity with the Kitāb al-Alfāẓ al-mustaʿmala fī l-manṭiq by Abū Naṣr Al-Fārābī: both have a classification of parts of speech that is unusual for Arabic grammars and use strange terminology [2020b: 27–29]. She further suggests that “[t]he listed categories and their names derive from a description of Greek in Tekhnē Grammatikē attributed to Dionysius Thrax (ca. 170–90 BCE)” [ibid.: 28].⁴ Generally, both Al-Fārābī and Dionysius Thrax figure prominently in Vidro’s study: “The hypernym ‘numbers ’ (arithmoi) is used in Tekhnē Grammatikē” [2020b: 30]; “the ‘three dots’ in the indirect quotation from Galen are reminiscent of Dionysius Thrax’s discussion of punctuation” [ibid.: 31], etc. 

In as far as any list of parts of speech west of India ultimately derives from Dionysius Thrax, Vidro’s statement is true, but hardly useful. Dionysius Thrax was indeed as fundamental to grammar as Euclid to geometry, and is known to have been widely used in teaching and translated into Syriac, Georgian and Armenian.⁵ But how important was he for the GIG? Using ἀριθμοί ‘numbers’ is hardly an indicator of any connection, as the word is in no way typical to Dionysius Thrax and has been employed by just about every Greek grammarian from the invention of the discipline until nowadays. Neither can the list of parts of speech (PoS) from Ars grammatica be taken lightly.⁶ One should be aware that PoS had symbolic and metaphysical significance for ancient and early medieval culture: the controversies about them were fierce and sowed dissension between schools, since the number of PoS was a scholarly opinion and a pledge of allegiance. The very history of ancient linguistics was conceptualised as the history of the PoS theories. Some grammarians are only known by the number of PoS they championed, which indirectly shows the value attached to this information.⁷ PoS perceptions were integrated into the Weltanschauung, which is reflected in the very terminology applied to them, e.g.:

The list adduced in the name of Dionysius Thrax and containing eight parts of speech was probably accepted in this form already by Aristarchus of Byzantium. Scholars agree that it had been further recognized by Apollonius Dyscolus and became standard. Since Latin lacked the article, Priscian had to elevate the interjection to the status of major PoS in order to preserve the total of eight.⁹ GIG’s declaration of seven parts of speech (Rylands B 3653v, l. 1–2), for this reason alone, should be a part of an altogether different story. Worse, GIG lacks the actual list of PoS, giving only their total, so we have less data on which to base our conclusions about the text’s pedigree.

Vidro wants us to believe that because GIG uses the terms ism, kalima, rābiṭ, ḫālifa, wāsiṭa, wāṣila, ḥāšiya, those are the parts of speech that the author considers major categories. It is possible, but not at all certain, because some of them could have been perceived as subcategories of one another. Neither can we presume that the order of the GIG’s list or the content of the terms is identical to that of Al-Fārābī: the author of GIG never mentions Al-Fārābī, and it is hardly possible to prove that he was familiar with his work.

Does the text of GIG provide extra clues about the system of PoS espoused? The author promises to deal with the material in order of PoS:  

... ואקצד אלי מא ילזם קסמא קסמא מן

وأقصد إلى ما يلزم قسما قسما من

אקסאם אלכלאם ואפ֗רדה עלי [ח?]דה ואצף גִُמל

أقسام الكلام وأفرده على حدة؟ وأصف جُمل

אעראבה ...

...إعرابه ب 

but the number and state of the fragments hardly allows for confident ordering based on the internal criteria.¹⁰

It seems that the mention of seven PoS was preceded by a discussion of some sort: 

ולד֗לך קלת אן אצנאף֗ אקסאם אלכלאם פ֗י

ولذلك قلت إن أصناف أقسام الكلام في

אלערביה סבעה

العربية سبعة

Also, the wording of

וקד יגִוז אן ידכ֗ל הד֗א אלבא[ב] פי ק

وقد يجوز أن يدخل هذا الباب في

קואנין אעראב [א]לאסם

قوانين إعراب الاسم 

may show that the author was unsure, to which section, i.e., to which PoS, the material belongs.¹¹

While the material is insufficient for firm conclusions, moving away from Dionysius Thrax and viewing GIG as a text engaged in intellectual dialogue with alternative taxonomies certainly helps. Consider, e.g., a convenient summary of (some) Stoic classifications by a scholiast:

Comparable terminology is found in Rylands B 3652v:

about which the editor remarks: “The expressions dāl (or ḏāl) of the definite and dāl (or ḏāl) of the indefinite … are not clear. The reference may, perhaps, be to the demonstrative ḏā and related forms” [2020b: note 53 p. 53]. The Stoic stance clarifies the issue. Moreover, a Stoic-centred approach adds certainty to Vidro’s suggestion that “the relative pronouns allatī and allaḏayni, as well as the vocative particles yā and ayyuhā are articles, too” [2020b: 29]. On the other hand, the classification espoused by GIG is hardly identical to the one quoted above, since there is no indication that GIG treats the prepositions and the conjunctions in the way suggested by the scholiast, and earlier Stoic lists of PoS contain less than seven elements. On the totality of evidence considered numerically and chronologically, it seems that the PoS of GIG are closer to the Syriac tradition than to the Greek one.

“Three dots”: GIG, Galen and Aristoteles

Occasional disagreements with Dionysius Thrax cannot be taken as evidence that Ars grammatica was unknown to GIG’s author. On the contrary, the probability that he was familiar with it is significant, since the passage on the ‘three dots’, sc. the punctuation signs used to mark pauses of different length, is very characteristic of Dionysius.¹² Στιγμή ‘dot’, ‘full stop’ occurs only in 9 authors among almost two hundred ‘Grammarians’, 166 times in total, of which 128 (6+122) belong respectively to Dionysius Thrax and his commentators.¹³ Yet before searching for the three dots in the commonplace school trivia, one should look into the context itself:

חתי אן גאלינוס יקול

حتى إن جالينوس يقول

Nay even Galen says

אנה מתי וצׄעת אעלאם תלך אלמ[קאט]ע והי נקט

إنه متى وضعت أعلام تلك المقاطع وهي نقط

that when signs were set for these units, i.e. the three

ת֗ל[ת] תערפהא אליונ[אניון] ............. בהא פ֗י

ثلث تعرفها اليونانيون ............... بها في

dots, the Greeks acquainted themselves with them and ...... them in

אלכתב אלג֗אמצׄה אלמסת[גל]קה ען אלפ֗ הם ואלמוקף֗

الكتب الغامضة المستغلقة عن الفهم والموقف

obscure books that are difficult to understand and comprehend. This is

לאנהא תסדדה ותרשדה ותדלה עלי מעאני אל

لأنها تسدده وترشده وتدله على معاني

because they show one the right way, guide one and indicate to him the meanings

כלאם ותחצרה ותחוטה ותמנעה מן אן יזול עמא

الكلام وتحصره وتحوطه وتمنعه من أن يزول عما

of speech while deterring, protecting and preventing him from moving away from

קצד אליה מנהא צאחב אלכלאם אלי ג֗ירה

قصد إليه منها صاحب الكلام إلى غيره وأبلغ من شرح

what was intended by the author of a statement over to something different.

Galen (b. A.D. 129) is the only author whom GIG mentions by name.¹⁴ An educated man of considerable means and wide interests, he was one of the most prolific writers of his time: about 500 titles were ascribed to him, but a great proportion of them vanished. Among those remaining, GIG’s seems to relate closest to Περὶ ἀλυπίας "On avoiding sorrow", a work considered lost but rediscovered by chance in 2005 in Vlatadon Monastery, Thessaloniki. This treatise is an answer to a fellow medicine student from Pergamon who asked why Galen did not seem to grieve over the great calamity that befell him: in a fire Galen lost all his possessions, including all the books he had written, all his medical tools, medicines and recipes, and his library. It is hypothesised to have been written in A.D. 193, a year after the great fire swept through Rome.

If the GIG’s passage is indeed related to Περὶ ἀλυπίας, it is not an “indirect quotation” [2020b: 31], but rather a bow to the authority: Galen, a great physician, personally cared about having his texts properly punctuated. The mention of Aristotle among the manuscripts destroyed by fire could lead to another association, that with Rhetoric, one of few texts where comprehensibility is directly linked to proper punctuation, and certainly the most famous of them: 

Heraclitus became a standard of incomprehensibility already by the time of Socrates and has kept this position ever since, earning the nickname Σκοτεινός ‘dark, obscure’. אלכתב אלג֗אמצׄה would be an exceptionally appropriate way of referring to his works.

Third person: GIG and Apollonius Dyscolus

Among the odd features of GIG, the editor mentions the definition of the persons of the verb:

אן אלכלמה ת֗לתה אוגִה אלאול ואלת֗אני ואלת֗אלת ואן אלאול הו אלד֗י מנה אלקול

ואלת֗אני הו אלד֗י יואגִד באלקול ואלת֗אלת הו אלד֗י יסתקפי באלקול

إن الكلمة ثلاتة أوجه الأول والثاني والثالثوإن الأول هو الذي منه القول والثاني هو الذي

يواجه بالقول والثالثهو الذي يستقفى بالقول

A verb has three persons (awǧuh): the first, the second and the third. The first is from whom the utterance is. The second is who is addressed with the utterance. The third is who is pursued by the utterance before: [2020b: 52–53; 2020a: 297], where

Considering the context through the prism of its Greek parallels, it is evident that the Arabic verb is a rendition of the classical Gr. ἀναπολέω, translated by the fourth-century grammarian Dositheus Magister as recordor (Ars grammatica, par. 71, l. 65) and reminiscor (ibid., par. 73, l. 20), cf. DGE s.v. ἀναπολέω ‘1. repetir … ; 2. recordar … ; 3. volver a trabajar’; LSJ s.v. ἀναπολέω ‘turn up the ground again…, hence, go over again, repeat’.

The above lexicographical descriptions are only partially helpful, and the exact translation of the word complicated: both ‘repetir’ and ‘recordar’ have their problems. The idea of ‘repetition’ in modern European languages implies reproducing the same (sequence of) actions, while the semantic structure of Greek does not require precise iteration or reoccurrence: that is why the notion ‘repeat’ is further specified through αὐταῖς λέξεσι ‘in the same words’ and similar expressions. Greek implies only loose similarity or association, which makes ‘recall’ with its meanings ‘to remember; recollect; to be reminiscent of; seem similar to; to summon back to awareness …’ (AHDEL, s.v. recall 2.a–c.) a somewhat better equivalent. If we are to reflect the semantics of the GIG on the understanding of the Maimonidean passage quoted by Blau [2006, s.v. قغواققي קפו/קפי]

אד̇א אסתקפיתה פי ג̇מיע אלכתב וג̇דתה

׳אם תחפשנו בכל הספרים, תמצאנו׳,

it should mean ‘go (back) to the books you know/have read/looked through, etc., making your way with the help of your associations and memory’ rather than actively and possibly mechanically ‘pursue, examine, study a (possibly unknown) material’.¹⁷

ἀναπολέω is a derivative of πολέω LSJ I. ‘go about, … , haunt’,¹⁸ where the seme of potentially repetitive action is vividly present, and itself a base for another derivative ἐπαναπολέω LSJ s.v. ‘repeat yet again’.¹⁹ Apollonius makes most of the meaning of ἀναπολέω when explaining the anaphoric functions of (what he calls) articles and pronouns. A good illustration is the context from De pronominibus:

Cf. also:

In TLG, a corpus of Greek literature exceeding 110 million words, ἀναπολέω occurs in 312 contexts in 127 authors, of whom 94 are later than GIG’s time, and ἐπαναπολέω occurs in six contexts of four authors (once in Apollonius Dyscolus and Methodius of Olympus (3–4 cent.), twice in Plato and Homeric scholia). Among the ‘grammarians’, Apollonius Dyscolus is by far the most frequent user of the verb (15 times); his son Aelius Herodianus used it twice; Aristonicus, Epimerismi, and a commentator to Dionysius Thrax’s Ars grammatica once each. Consequently, even if no definition of the third person as such has been preserved under the name of Apollonius, his specific predilection for the verb allows us to point at him as a source for the relevant part of GIG.

Exemplification: GIG and others

From what has been said above it should already be clear that GIG belongs to the category of so called ‘extended Greek grammars’. According to Aussant & Chevillard, “‘[e]xtended [g]rammars’ are grammatical descriptions of various languages which make use of tools initially developed in and/or for another language” (2020: 3). The term ‘extended Latin grammar’ was first proposed by Sylvain Auroux in 1992 and prompted the emergence of a new research field, particularly productive in recent years. The inquiry into extended grammars deals mostly with “transfer of grammatical models” (Aussant & Chevillard 2020:4), i.e. with pre-existing theoretical categories applied to newly described languages. However, it could have benefited from a wider approach that would cover not only theory, but also exemplification used for illustrating grammatical notions.

From the viewpoint of Greek, the grammatical examples in Arabic grammars can be divided into: 

•    those of proven Greek origin (= related to established Greek sources);

•    others (= those with unproven Greek connections; invented or quoted without any connection to Greek, etc.).

Obviously, decisions about the originality of the examples are made on a probabilistic basis, whether formally calculated or intuitively perceived. They are about the probability of sets of interlingual synonyms (as opposed to random words) occurring in two texts for specific purposes of grammar illustration. Theoretically, proving the continuity between Greek and Arabic texts could be facilitated by the fact that the same sets have been previously transferred into Latin and Syriac, yet the lack of manageable corpora on the non-Greek side hinders the inquiry, limiting it to individual researcher’s knowledge and semantic sensitivity.

Once borrowed, the examples acquire a life of their own inside the new traditions whose adepts may no longer be conscious of their prehistory. Such ‘detached’ and especially ‘twice detached’ Greek examples can hardly ever rise to the category of certain borrowings because the quantity of data surviving from the early periods of receiving cultures is normally insufficient.²¹

The exemplification words do not simply form a convenient pool of morphophonologically unproblematic semantically basic vocabulary, but have their own specialization that developed throughout the centuries of teaching practice: each such word is associated with a specific theoretical topic or topics, and the topics have their favourites. Effective examples, once discovered, have a long life expectancy.

Many researchers engaging with the thorny question of the dependence of Semitic grammars on the Greek ones distinguish between grammatical and so called ‘philosophical’ examples. Yet assessing grammatical development through the prism of such a division is not always helpful: a fragment of the recipient grammar can be traced to a single grammatical or philosophical Greek text, but it is necessary to look into the prehistory of the Greek prototype as well. Upon examination, one may discover Greek philosophical texts behind the grammars and vice versa. Strict separation between the disciplines is often anachronistic and may be problematic in as far as the scope of author’s interests is concerned. Of course, the exemplification per se never guarantees that the author subscribes to some specific school of thought or theoretical beliefs, but it is a useful clue to his upbringing.

Before discussing the examples of GIG, it is important to observe how they are used. As appears from the fragmentary material at our disposal, the author tends to group the examples into couples and triplets. The examples in such groups differ only minimally, so that to make the point of comparison more prominent; the principle of minimalism is implemented consistently. 

The most classical of GIG’s examples is the verb ضرب , considered a Paradenbeispiel of the dependence on Greek sources since the nineteenth century. It translates τύπτω ‘to hit’, which is the prototypical verb, in as far as it has been used by Theodosius of Alexandria (4–5 cent.) to illustrate the complete paradigm of Greek verbal morphology, something which cannot be done on a randomly selected verb because it is likely to lack (necessary subsets of) forms. Theodosius was not the first to select τύπτω: it is found in Dionysius Thrax, who would have inherited it from his Alexandrian or even earlier precursors: already in Ars grammatica τύπτω illustrates categories as diverse as voice (διάθεσις) and number.²² It is likely that Theodosius chose τύπτω precisely because Dionysius Thrax had previously featured it as both an active and passive verb. In its turn, the predilection of Dionysius may be due to the appearance of the verb in random action examples (e.g., Aristoteles Analytica priora et posteriora, Ross, p. 49a) and in contexts that draw attention to its subject-to-object semantic structure, such as:

Commentators to Dionysius Thrax well into Byzantine period continued to use τύπτω both as a paradigm verb and as an illustration of the accusative object. Extremely wide was its utilization by Apollonius Dyscolus: 108 instances cover all the range of the syntax of the accusative, including enclitic pronominal and non-enclitic objects, reflexive and passive constructions, infinitives, etc.²³ 

In GIG ضرب appears as the second or third in the series of examples:²⁴ 

לא תצ֗רב זידא

لا تضرب زيدا

או תאכ֗ד מאלה

وتأخذ ماله (T-S Ar.31.30r).

This pair of missing and surviving examples illustrate that the change of vocalisation results in altering the relationship between the clauses, and so do the following couple of examples. To illustrate this point one requires that the semantic structure of both verbs would allow for mutually simultaneous and sequential interpretation, a very loose constraint that most verb pairs would satisfy. For the sake of didactic clarity, actions would be preferable to states and simple syntactic structures to more complex. The default action verb τύπτω, i.e. ضرب, is good enough for the purpose.

Naturally, in most cases the requirements for the examples are more specific, but the Greek tradition provides a reliable resort of tried and tested solutions. Occasionally it is even possible to trace how this exemplification practice emerges in a culture. E.g., the verb τιμάω 'to honour, revere, respect’ occurs in ‘Grammarians’ in 66 forms and 311 contexts in total.²⁵ Mostly it functions as a normal non-exemplifying verb, e.g., in the identifications of toponyms through the cult (‘X is a place where the deity Y is worshiped’), or Homeric quotations. Sporadically, it appears as an example of a contracted conjugation or an item of orthographical or accentological clarification. The usage of Apollonius Dyscolus stands out with its 46 forms, of which 21 are active, and a third of them explicitly mentions the terms related to nominative, accusative or direct object: ὀρθή ‘straight’, εὐθεία ‘direct’, αἰτιατική ‘accusative’. It is therefore likely that by the time of Apollonius τιμάω was emerging as a handy traditional example for illustrating accusatives and direct objects, e.g.: οὔτε τιμῶ σε οὔτε λαλῶ σοι ‘neither do I respect you, nor talk to you’ (De adverbiis, Schneider, part 2, vol. 1,1, p. 182, l. 17). 

Since similar material is found in commentators to Dionysius Thrax, one may conclude that we are dealing either with Apollonius Dyscolus’s usage with possible Dionysian basis, or a projection of the Apollonius’s usage on Dionysius Thrax by later authors.

For our purposes, it is important that τιμάω acquires the status of primary example of the accusative of direct object, e.g.:

It is therefore not surprising that in Rylands B 3652r we encounter the Arabic equivalent of τιμάω instantiating the same category:

ומתי כאנת פ֗

ومتى كانت ف

When it is ....................

[מפ]עול בה נצב מן ד֗לך אנך תקול אבא זי[ד]

مفعول به نصب من ذلك أنك تقول أبا زيد

… a direct object, it is the accusative. For example, you say Abā Zaydin

א]כרמת

أكرمت

akramtu.

In addition to standard examples for transitives, Greek has a number of established examples for the categories ‘action not requiring external objects’ / ‘state minimally related to external objects’. The verbs of walking occupy a prominent place: βαδίζω ‘Α. walk, 2. go about, 3. go, proceed’ (LSJ s.v.) and περιπατέω ‘walk up and down, as in a cloister, opp. βαδίζειν (take a walk)’ (LSJ s.v.).²⁶ Being close semantically, both may appear in the explanations of the same theoretical issues.

Already Plato uses βαδίζω for illustration purposes. E.g., in Sophist, the Stranger intends to show that in order to produce a statement, verbs should be linked to subjects:

Aristoteles uses βαδίζω in 244 contexts.²⁷ Some of them are descriptions of real actions, such as the walking habits of lions and camels in Historia animalium, Louis p. 498b, l. 7, but most are abstract examples. Being a state, ‘walking’ often appears side by side with other states, such as ‘to be white’ (λευκὸν εἶναι) (e.g., Analytica priora et posteriora, Ross, p. 51b, ll. 11–19; p. 73b, ll. 6–7, etc.). Being a human action, walking is suitable for discussing the categories of cause/purpose, the latter being standardly ὑγιαίνειν ‘to be healthy’ (e.g., Ethica Nicomachea, Bywater, p. 1129a, l. 17, etc.). It is useful to illustrate finer details of potentiality and actuality, as well as basic formal categories, e.g.,

Note that in this example, βαδίζω functions simply as a verb with an explicit subject.

It seems, however, that at a certain stage the exemplification through βαδίζω started to become unpopular outside Academia and Lyceum. For this or another reason, examples (but not occurrences) with this verb are missing from Chrysippus and the grammarians influenced by Stoic authorities.²⁸

The competitor verb, περιπατέω, appears in the Aristotelian corpus 21 times, most of which do not represent abstract predicates, but still occasionally occur in discussing causes, e.g.:

Basic syntactic categories of subject-predicate relationship and transitivity are illustrated by περιπατέω and the already familiar τύπτω: