Philosophy of Language in the Medieval Arabic Tradition

4 Philosophy of Language in the Medieval Arabic Tradition Peter Adamson and Alexander Key In the Arabic-speaking intellectual world from the seventh century onwards, Hellenic analyses of linguistic content found fertile ground. But the tripartite theory of meaning consisting of sounds, thoughts, and things (phônai, noêmata, and pragmata) based on Aristotle’s De interpretatione was not universally welcomed. An autochthonous and pre-existing Arabic bipartite theory of meaning, consisting solely of ( vocal form (lafz·) and mental content (ma nā), provided an alternative. This Arabic pairing was the predominant model used to relate mental content to linguistic content, and it was in play across all available genres, from poetry to exegetical hermeneutics and legal theory. When Hellenic knowledge arrived on the scene in the eighth century, the bipartite Arabic model swiftly became part of the vocabulary for philosophy as well. Despite this agreement on elementary terminology, there was no sense of shared endeavour among the scholars who worked in Arabic on linguistic content across Hellenic philosophy, on the one hand, and grammar, poetics, and legal theory, on the other. Instead, there was an often bitter division between those who followed the Organon and those who stayed within the autochthonous disciplines of Arabic scholarship. In this chapter, we will analyse a notorious debate between two scholars of this period, a logician and a grammarian, who represented the philhellenic and phil-Arabic approaches to the philosophy of language. Our analysis will show that the distinction between Hellenic philosophy and the autochthonous Arabic disciplines was not the result of any prima facie incompatibility between the two epistemologies. In fact, once the sensitivities of politics and culture had died down, the tripartite and bipartite theories of meaning that each tradition brought to the table proved perfectly compatible and mutually productive. But, before that point was reached in the eleventh century, the two sides in our debate needed to work through their political and cultural differences. The polemical context drove them to make exaggerated claims for their respective disciplines: either logic was the only way to think, or it was grammar. THE MEDIEVAL ARABIC TRADITION 75 The self-evidently false nature of that dichotomy contributed to its abandonment, but not before casting a long shadow on the philosophy of language in the medieval Arabic tradition. Our chapter will, through its analysis of the debate, provide a clarification of the autochthonous Arabic position, which has received less attention than the philhellenic discourse. There was a pre-existing theory of meaning in Arabic before Hellenic philosophy appeared, and we will show how that bipartite Arabic theory of meaning, consisting solely of vocal form and mental content, provided the terminology and epistemological architecture for a whole series of ideas about how the mental content could, and should, be turned into linguistic content. We begin with a brief exercise in historical scene-setting. Philosophers working in the medieval Arabic tradition had good reason to be interested in language. The opportunity to read Hellenic philosophy depended on works that had to be translated from Greek into Arabic.1 At the same time, the Arabic cultural context was literary. It was dominated by a poetic and oratorical tradition that stretched back before Islam, penetrated almost all scholarly disciplines, and gave rise to a pronounced sensitivity towards semantic concerns.2 Practitioners of Hellenic philosophy engaged closely with Arabic versions of Aristotle and other Hellenic thinkers throughout the ninth, tenth, and eleventh centuries ad.3 This naturally led them to reflect on, and sometimes boldly assert, the possibility of rendering Greek philosophical ideas in the Arabic language. These early thinkers were especially engaged with the Aristotelian logical corpus, or Organon. And the issues of polysemy and naming raised by the Categories and On Interpretation had already appeared in the Arabic grammatical and exegetical traditions that sprang up around both the pre-Islamic poetic corpus and the new divine text of the Quran in the seventh and eighth centuries. As Arabic intellectual civilization developed, so these debates about language, using the bipartite theory of vocal forms that interact with mental content, became more complex. Three primary drivers for discussion about the philosophy of language in the Middle East from the seventh century onwards were therefore the philhellenic analyses of Aristotle’s logical works in Arabic, the dominance of poetry and oratory in the Arabic cultural context, and hermeneutical responses to that newest of monotheistic revelations, the Arabic Quran. These three streams flowed into the debate over the relative merits of logic and grammar. One side of this debate looked to ancient Greece for its inspiration, while the other gave precedence to the desert Bedouin environment and the Arabic language chosen by God for the Quranic 1 Often with Syriac as an intermediary, and increasingly as the centuries progressed directly from the Greek. On this translation movement, see Gutas (1998, 2010), in addition to Endress (1987). 2 For a review of the literary tradition, see Heinrichs (1987, 2012). An invaluable resource for the ideas and people of the literary tradition is Meisami and Starkey (1998). For further reading, see Key (2013). 3 For an introduction to Arabic philosophy, see Adamson and Taylor (2005) and McGinnis and Reisman (2007). And for the Greek texts that became available, see Gutas (2010) and D’Ancona (2013). 76 PETER ADAMSON AND ALEXANDER KEY revelation. The contrast is well exemplified by the choice of al-Khuwārizmı̄, the scholar in whose works algebra first appears, to divide his survey of the ninth-century intellectual landscape into two halves: “Arabic and Islamic”, and “Greek and foreign”.4 In the year 937 or 938 the vizier Ibn al-Furāt,5 sitting in his Baghdad court, called for a grammarian to contest logic’s claim to possess a tool without which the truth could not be known. The logician and philosopher Abū Bishr Mattā was asked to defend ( his subject against the criticism of the grammarian and polymath Abū Sa ı̄d al-Sı̄rāfı̄. Mattā was a founding member of the “Baghdad school” of Aristotelian philosophers, ( which included the somewhat more famous Yahyā Ibn Adı̄ and the much more ( ˙ famous al-Fārābı̄.6 Mattā, like Ibn Adı̄, was a Christian of Syriac extraction, a fact that provides al-Sı̄rāfı̄ with opportunities for sarcasm during the debate (he points out that expertise in logic apparently does not prevent one from thinking that the same thing can be both one and three). We have only one account of this debate, and it comes from the literary, political, and sociological anthology of the littérateur and philosopher Abū H.ayyān al-Tawhı̄dı̄, who was writing somewhat less than a century later.7 Abū H.ayyān was ˙ comfortable with Hellenic philosophy, as witnessed by his intellectual love affair with ( Abū Sulaymān al-Sijistānı̄ “the Logician”, a pupil of Yahyā Ibn Adı̄.8 However, as an ˙ Arabic littérateur given a choice between a famous Arab polymath and a Syriacspeaking logician who was happily professing ignorance of the culturally totemic subject of Arabic grammar, Abū H.ayyān supported al-Sı̄rāfı̄.9 Abū H.ayyān had also heard brief anecdotes about the debate from al-Sı̄rāfı̄ himself, while more detail had ( been provided by the Mu tazilı̄ grammarian al-Rummānı̄, Abū H.ayyān’s beloved teacher and a pupil alongside al-Sı̄rafı̄ of the commentator on Sı̄bawayh’s grammar, Ibn al-Sarrāj.10 Abū H.ayyān’s unsurprisingly partisan reportage of the debate also ( Abū Ja far Muhammad al-Khuwārazmı̄ (fl. c.830) wrote that he was dividing his survey of the disciplinary landscape into ˙“the disciplines of the [Islamic] revelation and those Arabic disciplines associated with it” and “the foreign disciplines from the Greeks and other nations” (al-Ḫ uwārizmı̄ 1895). On the culture clash, see inter alia, Zimmerman (1981) and ( Key (2012). For the matter of algebra, see Brentjes (2012). 5 Abū al-Fath al-Fadl b. Ja far Ibn al-Furāt (Ibn H . inzāba), d. 938. Appointed to the vizierate by the ˙ caliph al-Muqtadir (reg.˙ 908–32) in 932 and for a second time by the caliph al-Rādı̄ (reg. 934–40) in 937. ˙ For the history of this period, see Kennedy (2004). ( And for its historiography, El-Hibri ((1999). 6 Abū Bishr Mattā b. Yūnus, d. 940. Abū Sa ı̄d al-H asan al-Sı̄rāfı̄, d. 979. Yah yā Ibn Adı̄, d. 974. Abū . ˙ Nasr Muhammad al-Fārābı̄, d. 950. ˙7 Abū˙ Hayyān, d. 1032. On the debate, see Mahdi (1970), Endress (1977, 1986), Elamrani-Jamal (1983), . and Kühn (1986). For English translation, see Margoliouth (1905). Margoliouth, however, did not have access to a critical edition of the Arabic text, which subsequently became available through the publication of Abū H . ayyān’s original anthology (al-Tawh˙ı̄dı̄ 1965: i. 104–5, 107). 8 On Abū Sulaymān (d. c.991) and these philosophical/literary circles more generally, see Kraemer (1986a, b). 9 On the other hand, Joseph E. Lowry has profitably suggested, in personal communication, that Abū H. ayyān’s report of the debate can be read as an ironic presentation of al-Sı̄rāfı̄ as a sophist, using as he does all the classic rhetorical tools( of sophistry, from deliberate misunderstanding to reductio ad absurdum via homonymy. 10 Abū al-H . asan Alı̄ al-Rummānı̄, d. 994. Abū Bakr Muh˙ammad Ibn al-Sarrāj, d. 929. For an accessible review of Sı̄bawayh’s work and importance, see Carter (2004). 4 THE MEDIEVAL ARABIC TRADITION 77 mirrors the original context in which it was held: Ibn al-Furāt’s exact request to the court had been for someone to “break down Mattā’s position” that truth could be found only through logic and Aristotle.11 Despite the partiality of the reportage, we gain insight into Mattā’s philosophy of language, and a careful reading shows that he and al-Sı̄rāfı̄ share a surprising amount of common ground. Both subscribe to a fundamental bipartite distinction between language and the thoughts or meanings that language expresses. They even employ the same terminology for articulating this distinction: the linguistic entity is referred to as a lafz· (vocal utterance or form), while the mental content the word expresses is ( called a ma nā (meaning, idea, content).12 A workable analogue for this Arabic pairing is the distinction that Ferdinand de Saussure made between two bands of wavy lines representing sound and thought.13 But, whereas Saussure’s theory of meaning used a third element, the linguistic sign, to explain how people said and meant things, the Arabic theory of meaning uses only vocal form, mental content, and the connections made between them by each act of language use. Both al-Sı̄rāfı̄ and Mattā describe mental contents as objects of the mind or ( intellect; all are objects of reason (ma qūlāt). They are universal; Mattā explicitly says that intelligible mental content is common to people of different nations—he gives the example of grasping that 4 + 4 = 8—yet different nations express that same mental content with completely different vocal forms.14 Al-Sı̄rāfı̄ is even more explicit about the stark differences between mental content and its corresponding vocal form. The former endures through time, whereas the latter is composite and fleeting, disappearing even as it is spoken. In a strikingly Platonist remark, he ( explains this by saying that the vocal form, being “natural” or “physical” (t.abı̄ ı̄), is subject to constant disappearance, whereas mental content endures because it involves no matter and belongs to the intellect, which is divine.15 Mattā and al-Sı̄rāfı̄ disagree, however, about the lesson to be drawn from this basic distinction. Mattā is given little to say in the debate as reported, yet his position appears in clearer focus than that of al-Sı̄rāfı̄. At its heart is a claim about the subject matter of logic and the indispensability of Hellenic logic for thought. 11 Margoliouth (1905); al-Tawhı̄dı̄ (1965: i. 108). ( ( ˙ The plural form of lafz· is alfāz· , and the plural form of ma nā is ma ānı̄. They will ( be rendered in this chapter by “vocal form” and “mental content” respectively. On the terms lafz· and ma nā, see Frank (1981), Heinrichs (1998), and Key (2010). 13 Saussure (1949: 156) has a diagram showing these two bands, the plan indéfini des idées confuses and the [plan] non moins indéterminé des sons (“[p]lane of vague, amorphous thought” and “equally featureless plane of sound”). See also Saussure (2005: 110–11). 14 Margoliouth (1905: 113); al-Tawhı̄dı̄ (1965: i. 111). ˙ 15 Margoliouth (1905: 117); al-Tawhı̄dı̄ (1965: i. 115). Al-Sı̄rāfı̄ also explicitly agrees that mental content is not different for speakers ˙of different languages: Margoliouth (1905: 118); al-Tawhı̄dı̄ ˙ (1965: i. 116). 12 78 PETER ADAMSON AND ALEXANDER KEY By “logic” I mean one of the tools of speech by which one knows its correct usage from its incorrect usage, and by which one knows unsound mental content from sound mental content. It is like a balance, with which I know which side goes up and which side goes down.16 Subsequently, when asked to explain the different aspects of mental content that the Arabic particle wāw (a conjunction that usually but not exclusively means “and”) engenders in the discipline of logic, Mattā says: This is grammar, which I have not investigated. For the logician does not need grammar, whereas the grammarian does need logic, because logic examines mental content whereas grammar examines vocal form. If a logician passes by a vocal form, he does so accidentally, and if the grammarian stumbles upon mental content, this is also accidental. Mental content is more noble than vocal form.17 Notice that the second passage introduces a substantial caveat to the first. He has claimed that logic is a way of determining true from false speech, but now it becomes clear that the relation of logic to speech is an indirect, “accidental” one. This is because logic properly deals only with the relationships between ideas in the mind.18 Furthermore, both grammarians and logicians need to understand mental content, so logic is essential for both undertakings despite its neglect by grammarians. Language is not similarly essential; it is the central focus of grammar but only an accidental aspect of logic. Mental content on Mattā’s view is not an inward rehearsal of a natural language. But it is (at least sometimes) propositional. This is clear not only from the example he gives (4 +4 =8) but also from the association of logic with mental content: logic allows us to combine (propositional) mental content into valid syllogisms. Unsurprisingly, the position he is defending comes from Aristotle’s Organon, which was of intense interest to the Baghdad school.19 It can be traced ultimately to a passage at the beginning of On Interpretation: Now spoken sounds are symbols of affections in the soul, and written marks symbols of spoken sounds. And just as written marks are not the same for all men, neither are spoken sounds. But what these are in the first place signs of—affections of the soul—are the same for all.20 Here in On Interpretation thoughts are called by Aristotle affections of the soul, but in the commentary tradition they are sometimes called “objects of the mind” (noêmata, 16 Margoliouth (1905: 112); al-Tawhı̄dı̄ (1965: i. 109). In a not insubstantial irony of which Abū H . ayyān ˙ was most likely aware, Mattā is speaking in the faultless parallel cadences of an Arab littérateur—because his speech is being reported second hand by one such expert. 17 Margoliouth (1905: 116); al-Tawhı̄dı̄ (1965: i. 114). ˙ 18 See Margoliouth (1905: 113); al-Tawhı̄dı̄ (1965: i. 111). ˙ 19 Mattā himself translated the Poetics, which was considered a part of the Organon (on this see Black 1990; Aristotle 2012). 20 16a3–7 (Ackrill trans.):
‚Ø b s a K B fi çøB fi H K B fi łıåB fi ÆŁÅ ø   ºÆ, ŒÆd a ªæÆç  Æ H K B fi çøB fi . ŒÆd u æ Pb ªæ ÆÆ AØ a ÆP , Pb çøÆd ƃ ÆPÆ· z Ø ÆFÆ Å EÆ æø, ÆPa AØ ÆŁÆÆ B łıåB. THE MEDIEVAL ARABIC TRADITION 79 ( which corresponds to the Arabic “reasoned or intellected things”, ma qūlāt).21 This gives us the tripartite theory, according to which vocal form refers to things, but only through the intermediary of mental content.22 Mattā accepts this and, in his debate with al-Sı̄rāfı̄, insists that logic operates at the level of the mental content. Grammar studies only the external vocal forms of this mental content, and, since it is at the level of thought that knowledge and truth primarily reside, grammar is not really very important. Al-Sı̄rāfı̄’s first response to Mattā’s claim that logic is the balance with which mental content can be weighed is to keep “logic” tightly circumscribed as the practice pursued by Mattā and his fellow enthusiasts of a foreign falsafa. For al-Sı̄rāfı̄, logic is ( a tool of the logicians, whereas everyone else simply uses human reason (al- aql) to decide what is what. This is more a cultural and terminological than a substantive critique; Mattā might say (but does not) that logic is precisely the way that reason works. But al-Sı̄rāfı̄ quotes the second hemistich of a line from a wine poem by the famous poet Abū Nuwās to the effect that anyone who claims to be a philosopher has only a partial apprehension of the problem at hand.23 Al-Sı̄rāfı̄ then objects that logic lacks the universality claimed for it by Mattā. It is, rather, a Greek phenomenon (indeed the invention of only one Greek, namely Aristotle) and therefore subject to the conventions of the Greek language.24 Neither, therefore, can it be a universal tool for distinguishing truth from falsity, nor can it hope to resolve all disagreements or advance the cause of human understanding: “the world after Aristotle’s logic remained the same as it had been before his logic.”25 Certainly Mattā’s position would be undermined if it were true that Aristotelian logic were still wedded to the linguistic and conceptual particularities of Greek and the Greeks. But al-Sı̄rāfı̄ does not pursue this potentially lethal point. Instead, he switches to a line of attack that admits the universal potential of logic, saying that logic does manage to abstract away from the particularities of natural language: “grammar is a logic drawn out from Arabic, and logic is a grammar understood through language.”26 With this somewhat Delphic statement al-Sı̄rāfı̄ claims grammar as a synonym for logic, and implies that, while (presumably Arabic) grammar has to be extracted from the Arabic language, logic can be universal. Yet, while logic can be a universal tool for structuring mental content, it can only ever be understood through a language. The intentions about which we reason and the mental content 21 See, e.g., Simplicius in Cat. 10. On the possible origins of this theory with Boethus of Sidon, see Griffin (2012). 23 Abū Nuwās (al-H . asan al-H . akamı̄, d. c.814): [say to anyone who claims to know falsafa/] ‘you have learnt a little and missed a lot’. al-Tawhı̄dı̄ (1965: i. 110 (n. 6)); Abū Nuwās (1958–2006: iii. 4). ˙ 24 Margoliouth (1905: 113); al-Tawhı̄dı̄ (1965: i. 110). ˙ 25 Margoliouth (1905: 115); al-Tawhı̄dı̄ (1965: i. 113). This comes in response to Mattā’s statement that ˙ the Greeks, alone among all other nations, devoted themselves to the intellectual pursuit of wisdom (hikma) and this makes them worthy teachers. Margoliouth (1905: 114); al-Tawhı̄dı̄ (1965: i. 112). ˙ ˙ 26 Margoliouth (1905: 117); al-Tawhı̄dı̄ (1965: i. 115). ˙ 22 80 PETER ADAMSON AND ALEXANDER KEY that we know “can only be accessed by a language made up of nouns, verbs, and particles”.27 If grammar has no pretentions to universality, then its importance must come from the fact that the mental tool of logic is useless without mastery of vocal form. Such mastery enables us to avoid logical errors that could happen to a speaker of any language as well as those errors that are specific to the language one is speaking. Al-Sı̄rāfı̄ catches out Mattā for thinking two Arabic constructions are equivalent when they are not: “Zayd is one of the brothers” and “Zayd is one of his brothers” (a mistake, since no one can be his own brother). Al-Sı̄rāfı̄ describes this as “a question that has more to do with reasoned mental content than with the appearance of words”.28 Because Mattā lacks grammatical expertise, he falls into mistakes that are relevant to the level of thought. In this way, al-Sı̄rāfı̄ gives grammar at least some of the responsibility for shielding us from what Mattā would consider to be “logical” errors. An even more powerful point made by al-Sı̄rāfı̄ is that the logician must put his thoughts into language if he is to communicate: Why do you claim that grammarians only investigate vocal form to the exclusion of mental content, and logicians the reverse? That would only be true if the logician was silent, circulating his ideas only on the level of mental content, and organizing whatever he wanted with imaginative intimations, passing thoughts, and sudden guesses. However, when this logician wants to justify to a student or interlocutor that which upon consideration and careful reflection seems correct to him, he must use the vocal form that encompasses his intention, fits his aim, and corresponds with his goal.29 For al-Sı̄rāfı̄, thought and language are inextricably bound up in each other, and their mutual relationship is not as simple as Mattā assumes: If that which is sought by reason and that which is mentioned in vocal form could be taken back through their different branches and variant paths to that clear level on which four plus four equals eight, then our argument would be over and we would agree. But that is not the case.30 Because language is no mere mirror of mental content, putting thoughts into language can hardly be the trivial afterthought that Mattā makes it out to be. One example of the differences between language and thought is that language is limited to a smaller scope than the mental level: It is now clear that composite vocal forms do not encompass the extent of human reason, that mental content is reasoned, and that they are both intensely connected and expansively spread. 27 28 29 30 Margoliouth (1905: 113); al-Tawhı̄dı̄ (1965: i. 111). ˙ Margoliouth (1905: 121–2); al-Tawhı̄dı̄ (1965: i. 118–19). Margoliouth (1905: 121); al-Tawhı̄dı̄˙ (1965: i. 119). Margoliouth (1905: 113); al-Tawh˙ ı̄dı̄ (1965: i. 111). ˙ THE MEDIEVAL ARABIC TRADITION 81 The vocal form, in whatever language, has the power neither to command nor to contain that expanse.31 This idea did not originate with al-Sı̄rāfı̄. A century before our debate, one of the most ( renowned Arabic theorists, al-Jāhiz· , wrote that mental content (ma ānı̄) is different ( ˙ from the linguistic names given to it (asmāʾ al-ma ānı̄) “because mental content is spread out without end, and stretched out infinitely, whereas its names are curtailed and numbered, collectable and limited”.32 The inevitable consequence, albeit one al-Sı̄rāfı̄ does not spell out as clearly as some later authors would do, was polysemy. A limited corpus of vocal forms has to do the work of expressing an infinity of mental content. Some vocal forms therefore had to mean more than one thing (homonymy). Conversely, some mental content could be expressed with a multiplicity of vocal forms (synonymy). In the debate this is first revealed in the examples that al-Sı̄rāfı̄ uses to get from Mattā’s position that grammar only accidentally addresses mental content to his own statement that grammar is in fact a logic drawn out from the Arabic language.33 Al-Sı̄rāfı̄ gives a long stream of synonyms that can be used to refer to language, from “speech” and “conversation” to “presentation” and “demand”, before demonstrating that, while these vocal forms are different, their meanings are “in the same space when it comes to resemblance and similarity”. One cannot, therefore, say that “Zayd spoke the truth but did not talk truthfully”: it would be an error that “used the vocal form contrary to the reason of both the speaker and the listener”.34 A command of polysemy, or more simply of the various ways that vocal form can interact with mental content, is therefore a prerequisite for any intellectual exercise. It “was the way the truth was determined before logic was devised and it remains the way the truth is determined”.35 This orientation produces a particular sort of intellectual endeavour: one in which language takes centre stage and the potential for semantic breadth is exploited. As al-Sı̄rāfı̄ says, “make your mental content clear through eloquence!”36 When a good scholar makes the audience think, they appreciate their apprehension of that scholar’s intention all the more, and yet the scholar should also, at the same time, provide explanations and commentary sufficient to ensure complete understanding.37 The manipulation of homonyms and synonyms, 31 Margoliouth (1905: 127); al-Tawhı̄dı̄ (1965: i. 126). Cf. Hume’s denial of the infinity of the mind as ˙ quoted by Griffin (2012: 72 n. 14): “The capacity of the mind is not infinite . . . no idea of extension or duration consists of an infinite number of parts or inferior ideas, but of a finite number, and these simple and indivisible” (Treatise 1.2.4). 32 al-Jāhiz· (1960: i. 76). This passage in al-Jāhiz· (d. 869) supports our reading of b-s-t. in the debate ˙ ˙ passage as “spread out” rather than “simple”, pace Margoliouth’s translation, Mahdi (1970: 83), and Kü hn (1986: 358). 33 Both statements already translated; see nn. 17 and 26. 34 Because “speak” and “talk” are synonymous verbs. Margoliouth (1905: 116–77); al-Tawhı̄dı̄ (1965: i. ˙ 114–15). 35 Margoliouth (1905: 127); al-Tawhı̄dı̄ (1965: i. 126). ˙ ı̄dı̄ (1965: i. 125). 36 Margoliouth (1905: 125–6); al-Tawh 37 Margoliouth (1905: 126); al-Tawhı̄dı̄˙ (1965: i. 125). ˙ 82 PETER ADAMSON AND ALEXANDER KEY for instance through metaphor, is indispensable for effective language use, and analysis of these manipulations is the only way to determine the truth. As this might suggest, for al-Sı̄rāfı̄ poetics is a central case of language use.38 This is why the literary nature of the Arabic culture into which Greek philosophy was translated was so important. Opinions like al-Sı̄rāfı̄’s were culturally, politically, and professionally dominant and they explain why Mattā’s claim that words are too superficial to matter would tend to fall on deaf ears. But al-Sı̄rāfı̄’s claim was more than just persuasive rhetoric or an attempt to save the art of poetry. He sees understanding of language as the route to truth, and thus claims for grammar what Mattā claimed for logic. He offers this illustrative challenge to Mattā: Here is a problem that has caused disagreement, so solve it with your logic. Someone says ‘soand-so owns from the wall to the wall’. What is the correct legal ruling? How much land should be approportioned to so-and-so?39 This is a matter of law (fiqh), which like grammar was a discipline of paramount importance in the centuries immediately before and after our debate. So it counts as another point in al-Sı̄rāfı̄’s favour if grammar, and not the supposedly universal tool that is logic, can be used to reach correct legal decisions. The solution to the problem of the walls is, as with al-Sı̄rāfı̄’s other examples, to be found in a lexical analysis of word meanings and their usages in context. For al-Sı̄rāfı̄, Mattā is handicapped in his ability to deal with reality by his failure adequately to consider the clouding effect that language has on logic. Mattā lost the debate before Ibn al-Furāt, and his embarrassment stung other members of the Baghdad school into a response. The question of how logic relates to grammar received particular attention from two members of the school: the ( Muslim al-Fārābı̄ and his Christian student Yahyā Ibn Adı̄. The latter wrote a ˙ combative treatise On the Difference between Logic and Grammar, to which we shall return.40 As for al-Fārābı̄, he addresses the question throughout his writings, often defining logic in part by way of contrast to grammar. A good example appears in his Introductory Epistle on Logic: [Logic’s] role in relation to the intellect is the same as the art of grammar in relation to language. Just as grammar rectifies language among the people for whose language the grammar has been made, so logic rectifies the intellect in order that when there is the possibility of error it intellects only what is right. Thus the relation of the science of grammar to language and vocal form is like that of the science of logic to the intellect and the intelligibles ( (ma qūlāt).41 38 39 Cf. Kü hn (1986: 344–5). Margoliouth (1905: 126–7); al-Tawhı̄dı̄ (1965: i. 125–6). ( ˙ Yahyā b. Adı̄ (1988: 414–24). French translation: Elamrani-Jamal (1982). German translation: ˙ Endress (1986). 41 Dunlop’s translation (1956: }1, 225.5–9), modified. 40 THE MEDIEVAL ARABIC TRADITION 83 Here al-Fārābı̄ seems to take over Mattā’s position more or less unaltered. Logic operates at the level of the intelligibles, grammar at the level of vocal form.42 Logic and grammar are complementary, and each discipline is equally indispensable in its own sphere.43 Later in the epistle, however, we find a passage that complicates matters. Here al-Fārābı̄ ponders the etymological connection between the words for “reason” or “speech” and “logic”, a connection present in both Greek (logos, logikê) and Arabic (nut.q, mant.iq). Logic is the only route to certainty of truth with regard to anything that we desire to know. And the name ‘logic’ is derived from ‘discourse’. According to the ancients, the expression ‘logic’ indicates three things: [first,] the faculty with which man intellects the intelligibles, acquires disciplines of knowledge and arts, and distinguishes between admirable and repugnant actions; second, the intelligibles obtained in the soul of man through comprehension, which are called interior discourse (al-nut.q al-dākhil); third, the expression by language of what is in the mind, which is called exterior discourse (al-nut.q al-khārij). This art of logic is called ‘logic’ because it provides the rational faculty with rules for the interior discourse that is the intelligibles, because it provides all languages with shared rules for the exterior discourse that is the vocal forms, because it guides the rational faculty towards what is correct in both discourses, and because it protects the rational faculty from error in both discourses. Grammar is similar in some respects and yet also different because it only provides rules for the vocal forms that are specific to whichever nation and the people who speak its language. Logic, on the other hand, provides rules for the vocal forms that are common to all languages.44 This passage provides a detail that was lacking in the presentation of Mattā’s position in the debate—namely, that the intelligibles or mental content take the form of “interior discourse”. Of course this goes back to Plato (Theaetetus 189e–190a). We need to be careful, though: when al-Fārābı̄ uses the word nut.q for the phenomenon internal to the mind, he does not mean language (lisān).45 Language comes under the third heading of “exterior discourse”. It is hard to find a good translation of nut.q here (just as it would be difficult to translate the Greek logos in analogous contexts). We have chosen “discourse” for its relative neutrality, but even this goes too far in a linguistic direction. In fact al-Fārābı̄ must mean something similar to what we ascribed to Mattā: mental content that has a propositional character, but is not in any particular language.46 42 Al-Fārābı̄ makes the same statement in his Catalogue of the Sciences, and subsequently notes that, just as a grammarian would insist that the rules of grammar must be known before poetry and oratory can be memorized or used, so a logician would make a parallel claim that logic must be known in order for thoughts and beliefs to be known as true or false. al-Fārābı̄ (1949: 54, 58–9; 1953: 28–31, cf. 20 (Arabic text)). 43 al-Fārābı̄ (1949: 59). 44 Dunlop (1956: }4, 227.23–228.10); Gutas (1998: 272). Cf. Avicenna (2007a: i. 20) on the same matter. 45 Pace Dunlop (1956), who translates nut.q here as “speech”. 46 Gutas (1998: 172) chooses “articulation”. 84 PETER ADAMSON AND ALEXANDER KEY However, al-Fārābı̄ adds a new qualification to the Mattā view: logic also operates at the level of vocal form, in so far as it deals with aspects of exterior discourse common to all languages. The restriction of grammar to the level of vocal form is retained, but logic is given a wider remit, spanning the divide between mental content and vocal form. One can readily see why al-Fārābı̄ wants logic to apply to vocal forms. After all, Aristotle’s organon is the canonical group of texts on logic, and these texts do concern themselves with language. The Baghdad school accordingly characterizes logic as dealing with vocal form. For instance, the last member of the school, Abū l-Farāj Ibn al-Tayyib (d. 1044), says in his commentary on Porphyry’s Isagoge ˙ that “the subject-matter of the art of logic is the simple vocal forms that refer to ( ( common ideas [al-suwar al- āmma]”.47 Ibn Adı̄ likewise says that both logic and ˙ grammar have “vocal forms” as their subject matter.48 He stresses, however, that logic deals with vocal forms in so far as these are referring, unlike grammar, which deals only with their more superficial linguistic form (for example, vocalization to show case).49 Al-Fārābı̄ and Ibn al-Tayyib make the same point.50 But there still ˙ seems a tension, if not an outright contradiction, with Mattā’s claim that logic deals with mental content and not vocal form. Al-Fārābı̄ has not just one, but two, ways of resolving this tension. First, as we have seen, he allows logic to deal with universal features of language. We can find examples in his commentary on On Interpretation: the tenses of the verb, and even a certain range of uses for the definite article, are said to occur in all languages.51 Second, al-Fārābı̄ explains that, although logic does deal properly with mental content and not vocal form, Aristotle nonetheless writes as if language were at stake because “mental content is difficult to grasp”.52 Actually, though, language is being used only as a convenient substitute for mental content: it is easier to work with the sentence “Socrates is mortal” than the thought that Socrates is mortal. This habit of Aristotle’s might suggest that thoughts and bits of language combine in exactly the same ways.53 But in fact language sometimes fails to capture mental content adequately. One can be led into error by linguistic features that do not reflect what is happening at the level of thought. Al-Fārābı̄ is happy to exploit this point for polemical purposes, as we can see from another passage in the On Interpretation commentary. Here al-Fārābı̄ is considering the phenomenon of tense, which had long played an important role in Aristotelian logic. In On Interpretation, chapter 3, Aristotle distinguishes nouns from verbs in ( 48 Ibn al-Tayyib (1975: }173; 1979: }85). Yahyā b. Adı̄ (1988: 416.7, 421.9). ˙ ( ˙ Yahyā b. Adı̄ (1988: 421.10). ˙ 50 Zimmermann (1981: 10); al-Fārābı̄ (1986: 24). Ferrari notes that the issue already arises in the Greek commentators (Ferrari 2006: 24.8–9106–7). 51 Dunlop (1959: }9); Zimmermann (1981: 38, 63); al-Fārābı̄ (1986: 46, 69). 52 Zimmermann (1981: 14); al-Fārābı̄ (1986: 25). 53 Zimmermann (1981: 13–14); al-Fārābı̄ (1986: 25–6). 47 49 THE MEDIEVAL ARABIC TRADITION 85 part by their lack of tense.54 He then adds that past and future tense are only an inflexion (ptôsis) of a basic verb; the basic verb “additionally indicates the present”, whereas past and future verbs indicate what is “around” the present.55 In commenting on this passage, al-Fārābı̄ considers a view denying that there are in fact present-tense verbs. The view is bound up with a standard sceptical worry about time itself—namely, that nothing can happen at the present time, but only over an extended period of time. But it is not philosophers of time who are particularly prone to the misconception: it is scholars of grammar. Some Arabic grammarians have, says al-Fārābı̄, proposed that, in the absence of a proper present-tense verb, derived nouns should instead be taken as expressing the present tense. Al-Fārābı̄ traces this unfortunate proposal to a feature of Arabic—namely, its (supposed) lack of present-tense verbs.56 Whatever we make of al-Fārābı̄’s analysis of this particular point,57 the broader implication is clear. Anyone who confines her attention to the level of vocal form is liable to make mistakes, being misled by features of the particular vocal forms she is considering. Grammarians are particularly, perhaps uniquely, likely to commit such errors. Notice that the same point need not apply to logic in so far as it deals with language. For, on al-Fārābı̄’s account, it considers only universal aspects of language—that is, features that appear in every language.58 Nowhere does he suggest that such features might be misleading in the way that Arabic’s handling of the present tense is misleading. Rather he seems to presume that universal features of exterior discourse correspond to features of interior discourse and, ultimately, the world. Indeed, al-Fārābı̄’s confidence that every language does share these features is obviously not empirically grounded in a representative anthropological survey.59 So he must assert it on the strength of a tacit assumption, along the following lines: any language that lacked these features would be unfit for purpose, and would therefore never arise. Al-Fārābı̄ thus has the resources both to attack grammar, which studies the frequently parochial features of specific languages, and to defend logic, which has 54 On the late ancient reception of this in Boethius, see Suto (2012). It is interesting to note that Boethius also endorses the notion of an “inner discourse” that is not in any particular language, and explores the contrast between logic and grammar: Suto (2012: 82, 119). 55 De int 16b16–18: e ªÆ  j e ªØÆ E P ÞBÆ, Iººa HØ ÞÆ· ØÆçæ Ø b F ÞÆ, ‹Ø e b e Ææ Æ æÅÆ Ø åæ , a b e æØ. 56 Zimmermann (1981: 30–3); al-Fārābı̄ (1986: 40–2). The passage is paralleled in an introduction to logic by al-Fārābı̄, which, however, complains about mistakes made by certain ancients, rather than contemporary grammarians. Dunlop (1955: }5). 57 For critical remarks, see: Zimmermann (1981: pp. cxxxi–v). 58 In this respect Zimmermann (1981: p. cxxvii) is right to call logic a “universal grammar”, though this leaves out its (in fact more important) role in studying thought. A “grammar of thought” would be a contradiction in terms, because grammar is by definition the study of language and thought is for al-Fārābı̄ not linguistic. 59 Al-Fārābı̄ does, however, appear to have attempted some limited form of survey, as can be seen from his Arabic/Persian/Greek comparison at al-Fārābı̄ (1969: 61–2). 86 PETER ADAMSON AND ALEXANDER KEY emerged as the indispensible study of both thought and the universal aspects of language. This substantiates the Greek commentators’ claim that “philosophy uses logic to show, in the theoretical domain, what is true and what false, and in the ( practical domain what is good and what is bad”.60 Ibn Adı̄ fills out the picture further. For him logic is the instrument that allows us to extend our knowledge from immediate truths—first principles of the intellect and the deliverances of sensation— to demonstrated conclusions.61 This is why, in his treatise On the Difference between ( Logic and Grammar, Ibn Adı̄ can say that “the subject-matter for the art of logic is vocal forms that refer to universal things”.62 Logic’s purpose is after all the discovery of syllogistic demonstrations whose conclusions are necessary, certain, and universal.63 Logic’s exalted aims are again in stark contrast to the purpose of grammar. Grammar seeks nothing more than correct vocalization, correctness being measured against culturally specific custom.64 ( Another work by Ibn Adı̄, On the Existence of Common Things, sets the Baghdad school’s theory of logic within a broader metaphysical context, by considering the ( universal thought in the soul as one of three types of common ( āmm) thing.65 Human, for instance, can exist as a particular, material thing (like Zayd or ʾAmr), or ( as a form in the soul. Ibn Adı̄ calls these two ways that human can exist “physical” and “logical” respectively, and says that the vocal form “human” signifies the latter.66 This indicates his allegiance to the tripartite theory of words (“vocal forms”), mental content (“logical” commonality), and things (“physical” object instantiating a commonality). But, following the Neoplatonic commentary tradition, he also recognizes a more fundamental way that the commonality might exist: not in individual bodily things nor as an idea in the soul, but as “divine” and “essential”. It is this sort of commonality that is the real target of a definition like “rational mortal animal”.67 ( Bringing this together with his other remarks about logic, we can say that for Ibn Adı̄ logic is the instrument by which the soul arrives at demonstrative knowledge that conforms the soul, in so far as is possible, to that which is “divine”. We take this to be a particularly vivid instance of a more general phenomenon. The philosophers of the Baghdad school, in part because of their enthusiasm for logic and in part because of their rivalry with the grammarians, present logic (and by extension philosophy) as an art that transcends language. The diversity of human languages is an inevitable consequence of the fact that vocal forms are only conventional 60 Westerink (1961: 134, ( ll. 23–4). Notice the similarity to Abū Bishr’s opening characterization of logic in his debate with Abū Sa ı̄d. 61 He sets out this view in a short work called The Four Scientific Questions Regarding the Art of Logic. For translation(and discussion, see Shehadi and Rescher (1964); Adamson (2011). 62 63 Yahyā b. Adı̄ (1988: 422.9). See further Adamson (2007). ˙ 64 “[Grammar] is an art that concerns itself with vocal forms, in order to give rise to motions and rests ( in the way that Arabs do” ( (Yah˙yā b. Adı̄ 1988: 421.3). 65 Edited at Yahyā b. Adı̄ (1988: 148–59). On the treatise, see Adamson (2007), and more importantly ˙ (including a French translation) Rashed (2004). ( ( 66 67 Yahyā b. Adı̄ (1988: 148.12–16). Yahyā b. Adı̄ (1988: 154.17–20). ˙ ˙ THE MEDIEVAL ARABIC TRADITION 87 signs for mental content. The features of a given language are no more revelatory of reality than a given convention regarding table manners is revelatory of the moral law. By contrast, mental content relates to its objects naturally and is therefore the same for anyone who has an understanding of reality.68 This means that the human intellect is adequately related to reality, including divine reality (for instance Ibn ( Adı̄’s “divine” commonalities, and one could extend the point to the ability of metaphysics to discern the nature of God). None of this can be said for language. We saw that al-Sı̄rāfı̄’s defence of grammar appealed to practice, and in particular to the role of language in legal disputes, as with the example of the boundary dispute and the wall. Some further investigation of legal practice will be useful for our chapter, an illuminating complement for al-Sı̄rāfı̄’s ideas, just as the broader Baghdad school complements Mattā’s position. The claim of the Baghdad school that language is inadequate to deal with reality while logic shows both what is true and false in the theoretical domain and what is good and bad in the practical69 was met with a demonstration of how theory could rest on poetics and how practice could rely on semantics. We have already seen that, a century before the debate, Arabic scholars had developed a model of finite vocal forms interacting with the infinite potential of mental content. This remained one of the central ways in which the art of poetics was understood. However, the results of the analysis of poetry would never constitute a sufficiently practical science to combat the claims of certainty put forward by the logicians. Jurisprudence offered itself as a practical discipline that sat alongside grammar on the “Arabic” side of the cultural divide. From the very beginning, Arabic legal scholars had faced the problem of an Arabic-language revelation, a new religion, a literary culture, and the socio-political imperative to create order and certainty. The first major scholar that we know of to put forward a comprehensive theoretical framework to deal with these interlocking ( problems was the legal scholar al-Shāfi ı̄ (d. 820).70 In the following passage he sets out the relationship between God, Arabic, and law: God addressed the Arabs with his book in their language. He did so because the Arabs knew the mental content behind their language. They knew that the language had semantic breadth, and that it was in God’s nature to address them in the Quran with explicit language of general application while intending that language to be indeed explicit, unrestricted, and independent of the words around it, or alternatively intending the explicit and unrestricted language to contain specificity that rested on the surrounding words, or intending the explicit and 68 69 Zimmermann (1981: 12); al-Fārābı̄ (1986: 27). Shehadi and Rescher (1964: 574). ( ( Al-Shāfi ı̄ has been the subject of substantial recent scholarship in English: Lowry (2007); al-Šāfi ı̄ (2013). Also the relevant sections of Vishanoff (2011). For a readable translation (with historical introduction)( of a twentieth-century manual of Islamic law that represents the continuity of the tradition after al-Shāfi ı̄, see as-Ṣadr (2003; repr. 2005). ˙ 70 88 PETER ADAMSON AND ALEXANDER KEY unrestricted language to be in fact specific, or intending explicit language to be non-explicit, this being known from its context. The Arabs begin their speech with vocal forms that clarify what follows, and they begin speech with vocal forms that need to be clarified by what follows. The Arabs say something and they make the mental content behind it understood without this being manifest in the vocal form. This is comparable to the way that they indicate something with a physical gesture. They consider this way of speaking to be the most elevated, because only those who know it can do it and those who are ignorant of it cannot. The Arabs call a single thing many names, and they call many ideas with a single name.71 Furthermore, Arabic has a privileged position among languages as “the most vast in scope”.72 Within it, both God and Arabs use words like “man” to mean either all men, or a single specific man.73 Both also use words that make no sense on their own, but rather require context. Homonymy and synonymy likewise feature in both divine and human language. The confessional (and political) message is that Islam is a religion for which God chose the Arabic language. Philosophically, the message is that allusion and intimation are part of clear and effective communication, and consequently that semantic and linguistic analysis is central to determining both what is theoretically true and false, and what is practically good and bad—precisely the function claimed by the philosophers for logic.74 The reliance of law upon language led to (and was fed by) a sustained commitment to both grammar and lexicography. Grammar made it possible to understand rigorously how sentences fit together, allowing scholars to determine exactly what God had told people in the Quran, or exactly what commitments arose from legally binding declarations. But this presupposed the possibility of discerning the meaning of individual words in the relevant speech acts.75 For that purpose, lexicographers drew on ethnic and literary sources, predominantly histories of language use among the Bedouin Arabs and the poetic corpus that lay at the heart of Arabic culture. ( Scholars such as al-Shāfi ı̄ and al-Sı̄rāfı̄ thus had their own model of how language and reality interacted. It assumed polyvalency and was underpinned by an enumeration of lexical and grammatical rules and precedents. They could use linguistic analysis to make sense of the world, making claims of epistemic certainty regarding practical matters of human action and theoretical matters of divine creation and command. The Arabic set of hermeneutical and analytical tools that was being developed was capable of dealing with hermeneutics, poetics, semantics, and law. It ( al-Shāfi ı̄ (1938: 51–2 (}}173–6)).This passage is translated and analysed by Montgomery (2006), Lowry (2007), El Shamsy (2009), and Vishanoff (2011), and we have (profited from their efforts. 72 Lowry’s translation. Lowry (2007: 253, 251–4); see also aš-Shāfi ı̄ (1938: 42 (}138)). 73 See the discussions of this problem in Schöck (2006, 2007). ( 74 Al-Shāfi ı̄’s belief in the “idiosyncratic . . . irreducible semantic and structural vastness and complexity” of the Arabic language has been read as a response to incipient legal formalism in Iraq at his time. The central thrust of that legal formalism was that “language and its rules and structures yield meanings in predictable ways” (Lowry 2007: 251–2; see also Jackson 2002). 75 For example, in the lexicographical sub-genre of addād works, Arabic words with two opposite ˙ Key (2012: 221–3). meanings were enumerated and analysed. See Bettini (2010); 71 THE MEDIEVAL ARABIC TRADITION 89 was also steadily accumulating cultural capital. In this context it is scarcely suprising that a scholar like al-Sı̄rāfı̄ felt no need for another organon, especially one that appeared to claim that it alone could carve reality at the joints.76 We can now see how much was at stake in the tenth-century debate over the relative merits of logic and grammar. A fundamental disagreement about the path to truth, in both practical and theoretical contexts, ran parallel to a cultural and political divide. The success or failure of our two debaters at the court of Ibn Furāt ultimately depended more on the prejudices of their audience than on the substance of their ( arguments; al-Sı̄rāfı̄, Abū H.ayyān (and earlier al-Shāfi ı̄) felt strongly about the privileged position of Arabic literary culture. As for the beleaguered party at Ibn Furāt’s court, scholars such as Mattā and al-Fārābı̄ could console themselves with the thought that, however much the currency of grammar might outstrip that of logic in the culture of their day, the superiority of thinkers like Aristotle was timeless. This divide may seem insurmountable, yet a coming-together was inevitable, especially once intellectual life began to be systematically institutionalized in the madrasa from the eleventh century onwards. As in so many other fields, the pivotal figure who synthesized what had come before and set the stage for later developments was Avicenna (d. 1037).77 His logical writings proved immensely influential in the subsequent centuries. In particular, the Pointers and Reminders (al-Ishārāt wa-lTanbı̄hāt), a popular object of commentary in the centuries after Avicenna’s death, presented logic in a way that did not raise the hackles of scholars in the way Mattā had alienated al-Sı̄rāfı̄. Take, for example, the section on the relationship of vocal form and mental content in Pointers and Reminders: There is a certain relationship between the vocal form and the mental content. And the patterns of the vocal form may affect the patterns of the mental content.78 One of the most influential commentators on Pointers and Reminders, Nası̄r al-Dı̄n ˙ al-Tūsı̄ (d. 1274), used this statement to distinguish between the existence of something ˙ in the mind, its existence in speech (and writing), and its existence in external physical reality (the Aristotelian tripartite division). He then made a further distinction 76 For a discussion of this phrase, originating in Plato’s Phaedrus (265e), see Griffin (2012: 69 n. 3, 87, 92). ( 77 Abū Alı̄ al-H . usayn Ibn Sı̄nā (Avicenna), on whom see Gutas (1998); Wisnovsky (2003); Street (2008). Honourable mention must also be made at this point of another bold synthesizer, albeit one whose theories did not have as much . azm of ( impact as Avicenna’s. In Islamic Spain the legal theorist Ibn H Cordoba (Abū Muhammad Alı̄, d. 1064) developed an account of law, theology, and language that has ˙ unfairly been characterized as simply literalist. In fact, it included a thoroughgoing account of Aristotelian logic, and rested on a commitment never to exceed the meaning contained in one’s premisses. “A proposition”, he wrote in a statement that applies to both logic and law, “cannot give you anything more than itself ” (Vishanoff 2011: 95, 88–9; see also Ibn H . azm 2003). 78 Avicenna and al-Tūsı̄ (1983–94: i. 131). ˙ 90 PETER ADAMSON AND ALEXANDER KEY between the natural and unchanging connection between things in the mind and their counterparts outside it, and the non-natural connection between things in the mind and their vocal forms. The indicative connection that links mental content and vocal form (the Arabic bipartite division) changes according to circumstance. Because the mind may well think with its own internal language (“mental vocal forms”), the problems that affect language may come to affect thought. Homonymy is one of those linguistic problems, and it therefore becomes a problem for logic.79 Al-Tūsı̄ has taken the linguistic stick that al-Sı̄rāfı̄ used to beat Mattā (that his logic ˙ failed to account for polysemy) and fitted it back into what was now fast becoming (an) Arabic logic,80 rather than just an iteration of Aristotle’s Organon. We say “fitted back” because polysemy had, of course, been present in the Categories and On Interpretation at the outset. But the post-Porphyrean position that logic is only about vocal forms in so far as they signify (universal) mental content had, in the heated climate of debate with an established discipline of grammar, led Arabic logicians away from the kind of analysis of linguistic ambiguity that was championed by al-Sı̄rāfı̄. The Baghdad school sought to insulate both logic and mental content from the polysemy of natural languages. As we saw, al-Fārābı̄ saw logic as dealing with only the universal aspects of these languages, and with the unambiguous sphere of mental content. In The Cure, Avicenna also takes up this question of the extent to which the logician must concern himself with the features of language: An investigation of vocal form is a matter of necessity, and yet vocal forms are not something that the logician qua logician should concern himself with, unless [he is simply using them] in conversation or discussion with others. If it were possible to study logic through pure thought with only the mental contents themselves observed, then that would be enough. If the logician were to be able to apprise his interlocutor of that which is in his soul through some contrivance, then he would be able to dispense with vocal forms.81 This passage is taken from the Eisagoge, with which The Cure begins. Subsequently, at the start of his discussion of the Categories,82 Avicenna writes that logic is not concerned with the totality of language, but rather only with certain combinations of words: the syllogism, the definition, and the description. As for the question of whether a given mental content deserves the vocal form used for it, this is no part of logic, but rather the concern of the lexicographers.83 79 Avicenna and al-Tūsı̄ (1983–94: i. 131). ˙ On the subsequent development of which, see Wisnovsky (2004); El-Rouayheb (2010). 81 Avicenna (2007a: 22–3). See also Sabra (1980). 82 Avicenna was highly critical of the late antique position, discussed above, that Aristotle’s Categories had a linguistic aspect. See Avicenna (2007b: 1–8, esp. 8.4–9). 83 Avicenna (2007b: 5, ll. 10–11). 80 THE MEDIEVAL ARABIC TRADITION 91 To return to the same passage in Pointers and Reminders: [avicenna] The patterns of the vocal form may affect the patterns of the mental content. For this reason, the logician must take heed of absolute vocal forms, inasmuch as they are unrestricted by any specific language. [al-t.ūsı̄] This means that Avicenna’s investigation of mental content is his primary goal, while the investigation of vocal form is his secondary goal. His investigation of vocal forms, inasmuch as they are unrestricted by any specific language, is knowledge of their single, combined, homonymous, or ambiguous state, and indeed any of the other patterns that affect the way in which they refer, such as the effect of a negative on a necessarily negative copula, its necessary conversion to metathesis,84 the effect of these two on a mode, and the effect of a mode on them both.85 Avicenna has enabled al-Tūsı̄ to combine classical logical analysis—extolled by Mattā ˙ and dismissed by al-Sı̄rāfı̄—with the polysemic analyses of linguistic ambiguity, regarding which Mattā was ignorant and al-Sı̄rāfı̄ an expert. Mattā had been led to downplay the role of language in logic, but al-Tūsı̄ felt no such pressure. ˙ Avicenna rehabilitates the traditional ground-clearing exercise of considering polysemy, presenting the study of “the patterns of the vocal form” as a necessity (even if a regrettable one) for the logician, rather than a mere source of confusion and error. In Pointers and Reminders, Avicenna furthermore reclassifies the ways that vocal form refers to mental content, and vice versa. He writes that vocal form indicates mental content in one of three ways. First, congruence (mut.ābaqah), where each vocal form directly connects to its mental content. An example would be the relationship between the vocal form “triangle” and the idea of a three-sided shape. Second, implication (tadammun) between a vocal form and a mental content, such as ˙ the word “shape” and the idea of a triangle. Third, concomitance (al-istitbāʾ wa-liltizām), in which one idea requires that another follow it. An example is the relationship between the word “ceiling” and the idea of a wall supporting it.86 This third type allows the sort of connections between mental content and vocal form that tended to concern literary non-philosophers. What is a metaphor, if not an expression that calls to mind an idea that is not directly connected to it? (We are not concerned here with the Arabic reception of Aristotle’s own work on poetics and metaphor in the Poetics and Rhetoric. Commentaries on these texts were substantially 84 85 86 Metathesis: reversion of the order. See Zimmermann (1981: pp. lxiii, 98 n. 1). Avicenna and al-Tūsı̄ (1983–94: i. 131). ˙ and al-Tūsı̄ (1983–94: i. 139). Avicenna (Ibn Sı̄nā) ˙ 92 PETER ADAMSON AND ALEXANDER KEY disconnected from Arabic literary theory,87 despite their use of the poetic syllogism as a way to analyse metaphor.88) Another example of Avicenna’s attention to the relation between verbal form and mental content can be found in his discussion of the indefinite nouns in Aristotle’s On Interpretation (16a20f). Here he compares the way that nouns like “human” signify to the way that definitions signify: The composition can be . . . composed of two mental contents . . . for which a single vocal form can be substituted. For example “Zayd is a rational, mortal, animal”. The part of this phrase “rational, mortal, animal” is a composition that acts in this way, and a single vocal form like “human” can be substituted for it.89 Here we have a case where a composite at the level of thought can be represented by either a composite vocal form or a single word. Elsewhere, a composition at the linguistic level tracks a composition at the level of mental content. When explaining Aristotle’s inclusion of composite negative nouns such as “not human” or “not seeing” alongside simple nouns like “human”, Avicenna remarks: “the vocal form ‘not’ and the vocal form ‘seeing’ both indicate a mental content, and the combination of their two mental contents is the mental content of the ‘not seeing’ whole.”90 Of course Avicenna’s model is constructed to meet a very different need from that of the literary tradition. He seeks to establish the types of words that can be used to construct syllogisms, and to establish the relationships between words’ single or composite natures. Linguistic scholars instead want to give both a hermeneutical account of linguistic ambiguity and a poetic account of metaphor and imagery. Nevertheless, the results of the two processes were analogous. Both models used the pairing of vocal form and mental content to represent the interaction of language 87 The rhetoric and poetics of Hellenic philosophy were dealt with at length by practitioners of falsafa such as al-Fārābı̄ and Avicenna. Their relative lack of impact outside falsafa was due to the prior existence of a functional theory for reading and thinking about poetry, an indigenous theory untroubled by puzzling references to Greek tragedy, or other persistent difficulties of cross-cultural literary translation. The definitive study of that transmission is Vagelpohl (2008). There was, for example, no dialogue between Aristotle’s statement at Rhetoric 1404b–1405a that homonymy was primarily used by sophists to mislead and the very different role that al-Sı̄rāfı̄, as we have seen, accorded to homonymy. This must have been in part due to the difficulties (detailed by Vagelpohl) experienced by the translator of the Rhetoric. For the homonymy in the Rhetoric’s discussion of metaphor he used the same Arabic (ittifāq al-ism) as had been used for the very different homonymy of the Categories. Wansborough (1984); Aristotle (1948: i. 1; 1982: i. 176.19–20, ii. 106). Abu Deeb (1979: 303–22, esp. 310–11) shows the scale of the disconnect between Aristotle’s Poetics and Rhetoric and Arabic literary theory, with a discussion of Avicenna’s commentary on the Poetics and Rhetoric and the work of al-Jurjānı̄. An illustrative excerpt of Avicenna’s analysis of poetry is translated in van Gelder and Hammond (2008: 26–8). See also Black (1990). 88 Both al-Fārābı̄ and Avicenna sought to address poetry’s dominance of their cultural context in part by analysing it in terms of syllogisms that rested on imagination. For referenced analysis of this logic of poetics, see Heinrichs (2008: 3–10). 89 Avicenna (2007c: 32–3). See also Black (1991: 67). 90 On Interpretation, 16a30. Avicenna (2007c: 12). Cf. Black (1991: 67). THE MEDIEVAL ARABIC TRADITION 93 and mind, and both models emphasized the possibility that a single vocal form can denote multiple mental contents. As we have already discussed, al-Sı̄rāfı̄’s advocacy for his linguistic analysis of polysemy presented a challenge to the discipline of logic. Mental content was infinite and its vocal form finite, so only a technique that came to terms with the linguistic ambiguity of homonymy could suffice. Avicenna’s remarks imply an answer to this challenge: logic can handle the idea that finite forms capture the more expansive mental content that lies behind them, because single words and compound vocal forms can both indicate combination of several thoughts. Avicenna takes one of the dominant modes of analysis in Arabic intellectual culture and integrates it into the Organon, a typically syncretic step that was bound to be more culturally successful than the aggressive claims of supremacy on behalf of logic made by the Baghdad school. Nevertheless, as the commentators on Pointers and Reminders make clear,91 the relationships with which Avicenna was concerned were between mental contents more than they were between mental content and vocal form. Language comes into the logical picture only as an accident, by accident. To understand the relation between “human” and “not-human” is to understand the implications of one piece of mental content for another piece of mental content, not one bit of language for another bit of language. Nevertheless, Avicenna’s focus on the relationships between the mental content behind a vocal form may have enabled Arabic literary theory to progress beyond the binary of vocal forms and mental content to a recognition of the mental processes that could be prompted by encountering a literary expression. For al-Sı̄rāfı̄, vocal forms were simply mapped onto a range of individual static mental contents. Now Avicenna had applied this traditional terminology of vocal form and mental content to understand how mental contents (internal language) interact. The resulting analysis of dynamic thought processes had potential for poetics. ( That potential was heightened by the work of such figures as Abd al-Qāhir al-Jurjānı̄, the dominant figure in the analyses of secular and profane language from the eleventh century onwards.92 Al-Jurjānı̄ was the first literary theorist to investigate the mental processes that take place when a line of poetry, or a Quranic verse, is heard or read.93 He wrote that: 91 See Avicenna and al-Tūsı̄ (1983–94: i. 139). ˙ For more on al-Jurjānı̄, see Abu Deeb (1979); Larkin (1995); van Gelder and Hammond (2008: 29–69). Van Gelder and Hammond’s translation includes a section in which al-Jurjānı̄ addresses the relationships of poetry and logic to truth (van Gelder and Hammond 2008: 30–8). 93 Some previous steps had been taken in this direction by al-Rāghib al-Isfahānı̄ (fl. in or before 1018), ˙ target and the source not who described complex poetic metaphors in terms of the connection between being mentioned in any vocal expression, but instead being left to be reasoned by the audience. Key (2010: 58; 2012: 232–3); Abū al-Qāsim al-H . usayn b. Muhammad Ragib al-Is˙fahānı̄, ‘[Ragib on the New Style] Afānı̄n al-Balāġah’, MS 165 in Landberg Collection,˙ Yale University, Beinecke Rare Book and Manuscript Library, New Haven, fo. 12b. 92 94 PETER ADAMSON AND ALEXANDER KEY Language is divided into two types. In one, you grasp the intended aim with the indication of the vocal form alone, for example if you intended to inform [someone] about Zayd’s literal departure, you would say ‘Zayd left’. In the other type of language you do not grasp the intended aim with the indication of the vocal form alone. Instead, the vocal form indicates its mental content to you through a process that is constrained by the vocal form’s lexical position. You then find that mental content to contain a second indication that leads you to the intended aim. This type of language relies on allusion . . . and analogy.94 Language is either simple indication, or complex reference. For al-Jurjānı̄, complex reference is defined by the second level of mental effort on the part of the audience that goes into deciphering it. He was aware of the theoretical originality of this claim: If you have understood all of the above, then here is an abbreviated way to express it. You can just say ‘the mental content and ‘the mental content’s mental content’. By ‘the mental content’ you will mean that content directly understood from the vocal form itself without an intermediary. By ‘the mental content’s mental content’ you will mean that you reason mental content from the vocal form and then that mental content leads you to further mental content . . . 95 We do not know that al-Jurjānı̄ had indeed read the work of Avicenna or other logicians, but we can see that their two models had the potential for productive interaction. No longer was logic claimed to have universal domination over all thought and language, as in Mattā and to a lesser extent in al-Fārābı̄. Avicenna limited the scope of logic to the syllogism, the definition, and the description. In doing so he enabled logic to take what would be an increasingly central place in the Islamic madrasa in the coming centuries, even as he left room for the autochthonous linguistic sciences of Arabic grammar, lexicography, and literary theory. Al-Jurjānı̄ did not solve, or indeed address, the problems of logic. Nor did Avicenna answer the questions of Arabic literary theory. But the poisonous atmosphere of the debate between Mattā and al-Sı̄rāfı̄ had dissipated; the war between grammar and logic was over.96 References Abu Deeb, Kamal (1979). Al-Jurjānı̄’s Theory of Poetic Imagery. Warminster: Aris & Phillips. Abū Nuwās, al-H . asan b. Hānı̄ (1958–2006). Der Diwān des Abū Nuwās, ed. Ewald Wagner. Bibliotheca Islamica. 7 vols. Wiesbaden: Franz Steiner; Berlin: Klaus Schwarz. Adamson, Peter (2007). ‘Knowledge of Universals and Particulars in the Baghdad School’, Documenti e studi sulla tradizione filosofica medievale, 18: 141–64. 94 al-Jurǧānı̄ (1978: 202; 1992: 262); Key (2012: 233). Also translated and edited by Larkin (1995: 74, 184–5). 95 al-Jurǧānı̄ (1978: 203; 1992: 263); Key (2012: 233). Also translated by Abu Deeb (1979: 75–6). 96 Other cultural, political, and intellectual debates of course continued to rage, perhaps most notably the philosophical argument about the status of Hellenic falsafa, in which al-Ghazālı̄ wrote philosophy that attacked falsafa. For al-Ghazālı̄, see Griffel (2009). THE MEDIEVAL ARABIC TRADITION 95 Adamson, Peter (2011). ‘The Last Philosophers of Late Antiquity in the Arabic Tradition’, in U. Rudolph and R. Goulet (eds), Entre Orient et Occident: La Philosophie et la science grécoromaines. Vandoeuvres: Fondation Hardt, 1–43. Adamson, Peter, and Taylor, Richard C. (2005). The Cambridge Companion to Arabic Philosophy. Cambridge: Cambridge University Press. ( Aristotle (1948). Organon Aristotelis in versione Arabica antiqua / Mant.iq Aristū. ed. Abd ar-Rahmān Badawı̄. 3 vols. Cairo: Dār al-Kutub al-Misrı̄yah. ˙ ˙ Aristotle (1982). Aristotle’s Ars Rhetorica. The Arabic Version, ed. M. C. Lyons. 2 vols. Cambridge: Pembroke Arabic Texts. Aristotle (2012). Aristotle Poetics: Editio maior of the Greek Text With Historical Introductions and Philological Commentaries, ed. Leonardo Tarán and Dimitri Gutas. Leiden: Brill. ( Avicenna (Ibn Sı̄nā, Abū Alı̄ al-H al-Mant.iq I (al-Madḫal)’, in . usayn) (2007a). ‘al-Shifāʾ: ( ( al-Šifāʾ, ed. Ibrāhı̄m Madkūr. Cairo: Wizārat al-Ma ārif al- Umūmı̄yah, 1952; repr. Qum: Ẕ ovı̄ el-Qurbā. ( Avicenna (Ibn Sı̄nā, Abū Alı̄ al-H . usayn) (2007b). ‘al-Shifāʾ: al-Mant.iq II (al-Maqūlāt)’, in al-Šifāʾ, ed. Ibrāhı̄m Madkūr. Cairo: Wizārat aṯ -Ṯaqāfah wa-l-Iršād al-Qawmı̄, 1959; repr. Qum: Ẕ ovı̄ el-Qurbā. ( ( al-Mant.iq III (al- Ibārah)’, in Avicenna (Ibn Sı̄nā, Abū Alı̄ al-H . usayn) (2007c). ‘al-Shifāʾ: ( al-Shifāʾ, ed. Ibrāhı̄m Madkūr. Cairo: Dār al-Kātib al- Arabı̄, 1970; repr. Qum: Ẕ ovı̄ elQurbā. ( Avicenna (Ibn Sı̄nā, Abū Alı̄ al-H al-Išārāt wa-l. usayn), and Nas˙ı̄r al-Dı̄n at.-T˙ūsı̄ (1983–1994). ( Tanbı̄hāt., ed. Sulaymān Dunyā. 3rd edn. 4 vols. Cairo: Dār al-Ma ārif. Bettini, Lidia (2010). ‘Ḍidd’, in Kees Versteegh, Lutz Edzard, and Rudolf de Jong (eds), Encyclopedia of Arabic Language and Linguistics Online Edition. Brill Online: Brill. Black, Deborah L. (1990). Logic and Aristotle’s Rhetoric and Poetics in Medieval Arabic Philosophy. Leiden: Brill. Black, Deborah L. (1991). ‘Aristotle’s “Peri hermeneias” in Medieval Latin and Arabic Philosophy: Logic and the Linguistic Arts’, Canadian Journal of Philosophy: Supplementary Volume 17: 25–83. Brentjes, Sonja (2012). ‘Algebra’, in Gudrun Krämer, Denis Matringe, John Nawas, and Everett Rowson (eds), Encyclopaedia of Islam, THREE. Brill Online: Brill. Carter, M. G. (2004). Sibawayhi. London: I. B. Tauris. D’Ancona, Cristina (2013). ‘Greek Sources in Arabic and Islamic Philosophy’, in Edward N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Winter 2013 edn), <http://plato. stanford.edu/archives/win2013/entries/arabic-islamic-greek/> (accessed 11 August 2014). Dunlop, D. M. (1955). ‘Al-Fārābı̄’s Introductory Sections on Logic [I]’, Islamic Quarterly, 2: 264–82. Dunlop, D. M. (1956). ‘Al-Fārābı̄’s Introductory Sections on Logic [II]’, Islamic Quarterly, 3: 224–35. Dunlop, D. M. (1959). ‘Al-Fārābı̄’s Paraphrase of the Categories of Aristotle’, Islamic Quarterly, 5: 21–54. Elamrani-Jamal, A. (1982). ‘Grammaire et logique d’après le philosophe arabe chrétien Yahyā ( ˙ b. Adı̄ (280–364 H/893–974)’, Arabica, 29: 1–15. Elamrani-Jamal, A. (1983). Logique Aristotélienne et grammaire arabe: Étude et documents. Paris: J. Vrin. 96 PETER ADAMSON AND ALEXANDER KEY El-Hibri, Tayeb (1999). Reinterpreting Islamic Historiography: Hārūn al-Rashı̄d and the Narrative of the ʻAbbasid Caliphate. Cambridge: Cambridge University Press. El-Rouayheb, Khaled (2010). Relational Syllogisms and the History of Arabic Logic, 900–1900 (Leiden: Brill). El Shamsy, Ahmed (2009). ‘From Tradition to Law: The Origins and Early Development of the ( Shāfi ı̄ School of Law in Ninth-Century Egypt’, Ph.D. thesis, Harvard University. Endress, Gerhard (1977). ‘La Controverse entre la logique philosophique et la grammaire arabe au temps des khalifs’, Journal for the History of Arabic Science, 1: 339–51. Endress, Gerhard (1986). ‘Grammatik und Logik: Arabische Philologie und griechische Philosophie im Widerstreit’, in B. Mojsisch (ed.), Sprachphilosophie in Antike und Mittelalter. Amsterdam: Grüner, 163–299. Endress, Gerhard (1987). ‘Die wissenschaftliche Literatur’, in Helmut Gätje (ed.), Grundriss der Arabischen Philologie II. Wiesbaden: Dr Ludwig Reichert, 24–61. ( ( al-Fārābı̄, Abū Nasr Muhammad (1949). Ihsāʾ al- Ulūm. Uṯ mān Amı̄n. Cairo: Dār al-Fikr ( ˙ ˙ ˙˙ al- Arabı̄. ( al-Fārābı̄, Abū Nasr Muhammad (1953). Catálogo de las Ciencias [Ihsāʾ al- Ulūm], trans. ˙ ˙ ˙˙ ( A. Gonzalez Palencia, ed. A. Gonzalez Palencia. Madrid: Dār al-Fikr al- Arabı̄. al-Fārābı̄, Abū Nasr Muhammad (1969). Kitāb al-H . urūf, ed. Muhsin Mahdi. Beirut: Dār ˙ ˙ al-Mašriq. ( al-Fārābı̄, Abū Nasr Muhammad (1986). Sharh al-Fārābı̄ li-Kitāb Arist.ūt.ālı̄s fı̄-l- Ibārah, ed. ˙ ˙ ˙ Wilhelm Kutsch and Stanley Marrow. Beirut: Dār al-Mašriq. ( Ferrari, C. (2006). Die Kategorienkommentar von Abū l-Farāj Abdallāh ibn at.-Tayyib. Leiden: ˙ Brill. Frank, Richard M. (1981). ‘Meanings are Spoken of in Many Ways: The Earlier Arab Grammarians’, Le Museon, 94/3–4: 259–319. ( ( ( al-Ǧāhiz· , Abū Uṯmān Amr b. Bahr (1960). al-Bayān wa-l-Tabyı̄n, ed. Abd al-Salām Muham˙ ˙ ˙ mad Hārūn. 4 vols. Cairo: Maktabat al-Khānǧı̄. Griffel, Frank (2009). Al-Ghazālı̄’s Philosophical Theology. Oxford: Oxford University Press. Griffin, Michael J. (2012). ‘What Does Aristotle Categorize? Semantics and the Early Peripatetic Reading of the Categories’, Bulletin of the Institute of Classical Studies (The Peripatetic School through Alexander of Aphrodisias (Issue in Honour of R. W. Sharples)), ed. Peter Adamson, 69–108. Gutas, Dimitri (1998). Avicenna and the Aristotelian Tradition: Introduction to Reading Avicenna’s Philosophical Works. Leiden: Brill. Gutas, Dimitri (2010). ‘Greek Philosophical Works Translated into Arabic’, in Robert Pasnau and Christina van Dyke (eds), The Cambridge History of Medieval Philosophy. Cambridge: Cambridge University Press, ii. 801–14. Heinrichs, Wolfhart (1987). ‘Poetik, Rhetorik, Literaturkritik, Metrik und Reimlehre’, in Helmut Gätje (ed.), Grundriss der Arabischen Philologie II. Wiesbaden: Dr. Ludwig Reichert, 177–207. ( Heinrichs, Wolfhart (1998). ‘Lafz· and ma nā’, in Julie Scott Meisami and Paul Starkey (eds), Encyclopedia of Arabic Literature. London and New York: Routledge, 461–2. Heinrichs, Wolfhart (2008). ‘Takhyı̄l: Make-Believe and Image Creation in Arabic Literary Theory’, in G. J. H. van Gelder and Marlé Hammond (eds), Takhyı̄l: The Imaginary in Classical Arabic Poetics. Cambridge: Gibb Memorial Trust, 1–14. THE MEDIEVAL ARABIC TRADITION 97 Heinrichs, Wolfhart (2012). ‘Takhyı̄l’, in P. Bearman, Th. Bianquis, C. E. Bosworth, E. van Donzel, and W. P. Heinrichs (eds), Encyclopaedia of Islam, 2nd edn. Brill Online. ( ( al-Ḫ uwārizmı̄, Abū Abdallāh Muhammad b. Ahmad b. Yusūf (1895). Mafātı̄h al- Ulūm, ed. ( ˙ ˙ ˙( G. Van Vloten. Leiden: Brill. Facsimile edition, ed. Muhammad H . usayn Abd al- Azı̄z. ( ˙ Cairo: al-Hay’ah al- Āmmah li-Qusūr al-Ṯaqāfah, 2004. ( ˙ Ibn al-Tayyib, Abū al-Faraǧ Abdallāh (1975). Ibn al-Tayyib’s Commentary on Porphyry’s ˙ ˙ Eisagoge [Arabic text], ed. Kwame Gyekye. Beirut: Dār al-Mašriq. ( Ibn al-Tayyib, Abū al-Faraǧ Abdallāh (1979). Arabic Logic: Ibn al-Tayyib’s Commentary on ˙ ˙ Porphyry’s Eisagoge [English translation], ed. Kwame Gyekye. Albany: State University of New York Press. ( Ibn H Alı̄ b. Ahmad (2003). al-Taqrı̄b li-H . azm, Abū Muh˙ammad . add al-Mant.iq wa-l-Madḫal ( ˙ ilayhi bi-l-Alfāz· al- Āmı̄yah wa-l-Amṯ ilah al-Fiqhı̄yah wa-yalı̄hi Mihakk an-Naz· r fı̄-l-Mant. ˙ iq taʾlı̄f Abı̄ H . āmid Muh˙ammad al-Ġazālı̄, ed. Ah˙mad Farı̄d al-Mazı̄dı̄. Beirut: Dār al-Kutub ( al- Ilmı̄yah. Jackson, Sherman (2002). ‘Fiction and Formalism: Toward a Functional Analysis of usūl ˙ al-fiqh’, in Bernard G. Weiss (ed.), Studies in Islamic Legal Theory. Leiden: Brill, 177–201, 406–11. ( ( ( al-Jurǧānı̄, Abd al-Qāhir b. Abd al-Rahmān (1978). Dalā’il al-I ǧāz, ed. Muhammad Mah( ˙ ˙ ˙ mūd al-Šinqı̄t.ı̄. Beirut: Dār al-Ma rifah; repr. 1987. ( ( ( al-Jurǧānı̄, Abd al-Qāhir b. Abd al-Rahmān (1992). Dalā’il al-I ǧāz, ed. Mahmūd Muham˙ ˙ ˙ mad Šākir. Cairo: Dār al-Madanı̄; repr. 1987. Kennedy, Hugh (2004). The Prophet and the Age of the Caliphates: The Islamic Near East from the Sixth to the Eleventh Century. 2nd edn. Harlow: Pearson. Key, Alexander (2010). ‘Language and Literature in al-Rāġib al-Isfahānı̄’, in Reflections on ˙ Language and Knowledge in Middle Eastern Societies. Cambridge: Cambridge Scholars Publishing, 32–62. Key, Alexander (2012). ‘A Linguistic Frame of Mind: al-Rāġib al-Isfahānı̄ and what it Meant to ˙ be Ambiguous’. Ph.D. thesis, Harvard University. Key, Alexander (2013). ‘English-Language Resources for Arabic Poetics’, Stanford University Division of Literatures, Cultures, and Languages <http://arcade.stanford.edu/english-lan guage-resources-arabic-poetics> (accessed 11 August 2014). Kraemer, Joel L. (1986a). Philosophy in the Renaissance of Islam: Abū Sulaymān Al-Sijistānı̄ and his Circle. Brill: Leiden. Kraemer, Joel L. (1986b). Humanism in the Renaissance of Islam. Studies in Islamic Culture and History Series. Brill: Leiden. Kü hn, Wilfried (1986). ‘Die Rehabilitierung der Sprache durch den Arabischen Philologen alSirafi’, in B. Mojsisch (ed.), Sprachphilosophie in Antike und Mittelalter. Amsterdam: Grüner, 301–402. ( Larkin, Margaret (1995). The Theology of Meaning: Abd al-Qāhir al-Jurjānı̄’s Theory of Discourse. New Haven: American Oriental Society. Lowry, Joseph E. (2007). Early Islamic Legal Theory: The Risāla of Muhammad ibn Idrı̄s ( ˙ al-Shāfi ı̄. Leiden: Brill. McGinnis, Jon, and Reisman, David C. (2007). Classical Arabic Philosophy: An anthology of sources translated from the Arabic. Indianapolis: Hackett. 98 PETER ADAMSON AND ALEXANDER KEY Mahdi, Muhsin (1970). ‘Language and Logic in Classical Islam’, in Gustave E. von Grunebaum (ed.), Logic in Classical Islamic Culture. Wiesbaden: O. Harrassowitz, 51–83. ( Margoliouth, D. S. (1905). ‘The Discussion between Abū Bishr Mattā and Abū Sa ı̄d al-Sı̄rāfı̄ on the Merits of Logic and Grammar’, Journal of the Royal Asiatic Society of Great Britain and Ireland, 79–129. Meisami, Julie Scott, and Starkey, Paul (1998). Encyclopedia of Arabic Literature. London and New York: Routledge. Montgomery, James E. (2006). ‘Al-Jāhiz· ’s Kitāb al-Bayān wa al-Tabyı̄n’, in Julia Bray (ed.), ˙ Writing and Representation in Medieval Islam: Muslim Horizons. New York: Routledge, 91–152. Ragib, Abū al-Qāsim al-H . usayn b. Muh˙ammad al-Is˙fahānı̄ (n.d.). ‘[Ragib on the New Style] Afānı̄n al-Balāġah’. MS 165 in Landberg Collection, Yale University, Beinecke Rare Book and Manuscript Library, New Haven. ( Rashed, M. (2004). ‘Ibn Adi et Avicenne: Sur les types d’existants’, in V. Celluprica and C. D’Ancona (eds), Aristotele e i suoi esegeti Neoplatonici. Naples: Bibliopolis, 109–71. Sabra, A. I. (1980). ‘Avicenna on the Subject Matter of Logic’, Journal of Philosophy, 77/11: 746–64. as-Ṣadr, Muhammad Bāqir (2003). Lessons in Islamic Jurisprudence, trans. Roy Mottahedeh. ˙ ˙ Oxford: Oneworld; repr. 2005. ( al-Šāfi ı̄, Muhammad Idrı̄s (1938). al-Risālah, ed. Ahmad Muhammad Šākir. Cairo: Must.afā al˙ ˙ ˙ ˙ Bābı̄ al-H . alabı̄. ( al-Šāfi ı̄, Muhammad Idrı̄s (2013). The Epistle on Legal Theory, trans. Joseph E. Lowry. New ˙ York: New York University Press. Saussure, Ferdinand de (1949). Cours de linguistique générale, ed. Charles Bally, Albert Sechehaye, and Albert Riedlinger. Paris: Payot. Saussure, Ferdinand de (1983/2005). Course in General Linguistics, trans. Roy Harris. London: Duckworth. Schöck, Cornelia (2006). Koranexegese, Grammatik und Logik. Leiden: Brill. Schöck, Cornelia (2007). ‘Discussions on Conditional Sentences from the Year ah 17/ad 638 to Avicenna (d. ah 428/ ad 1037)’, in Peter Adamson (ed.), Classical Arabic Philosophy: Sources and Reception. Warburg Institute Colloquia. London: Warburg Institute, 55–73. ( Shehadi, N., and Rescher, F. (1964). ‘Yahyā Ibn Adı̄’s Treatise on the Four Scientific Questions ˙ Regarding the Art of Logic’, Journal of the History of Ideas, 25: 572–8. Street, Tony (2008). ‘Arabic and Islamic Philosophy of Language and Logic’, in Edward N. Zalta (ed.), The Stanford Encyclopedia of Philosophy <http://plato.stanford.edu/>. Stanford: The Metaphysics Research Lab, Stanford University. Suto, T. (2012). Boethius on Mind, Grammar and Logic: A Study of Boethius’ Commentary on Peri Hermeneias. Leiden: Brill. ( al-Tawhı̄dı̄, Abū H . ayyān (1965). Kitāb al-Imtā wa-l-Muʾānasah, ed. Ah˙mad Amı̄n and ˙ Ahmad al-Zayn. 3 in 1 vols. Beirut: Dār Maktabat al-H . ayāh. ˙ Vagelpohl, Uwe (2008). Aristotle’s Rhetoric in the East: The Syriac and Arabic Translation and Commentary Tradition. Leiden: Brill. van Gelder, Geert Jan, and Hammond, Marlé (2008). Takhyı̄l: The Imaginary in Classical Arabic Poetics. Cambridge: Gibb Memorial Trust. THE MEDIEVAL ARABIC TRADITION 99 Vishanoff, David R. (2011). The Formation of Islamic Hermeneutics: How Sunni Legal Theorists Imagined a Revealed Law. American Oriental Series, vol. 93. New Haven: American Oriental Society. Wansborough, John (1984). ‘Review of Aristotle’s Ars Rhetorica: The Arabic Version. A New Edition with Commentary and Glossary by M. C. Lyons’, Bulletin of the School of Oriental and African Studies, 47/3: 550–1. Westerink, L. G. (1961). ‘Elias on the Prior Analytics’, Mnemosyne, 14: 126–39. Wisnovsky, Robert (2003). Avicenna’s Metaphysics in Context. London: Duckworth. Wisnovsky, Robert (2004). ‘One Aspect of the Avicennian Turn in Sunni Theology’, Arabic Science and Philosophy, 14: 65–100. ( Yahyā b. Adı̄ (1988). Maqālāt Yahyā b. ʻAdı̄ al-Falsafı̄yah, ed. Sahbān Ḫ ulayfāt. Amman: ˙ ˙ ˙ Manšūrāt al-Ǧ āmiʻah al-Urdunı̄yah. Zimmermann, F. W. (1981). Al-Farabi’s Commentary and Short Treatise on Aristotle’s De Interpretatione. Oxford: Oxford University Press for the British Academy.