Published online by Cambridge University Press: 01 January 2020
In many fields within the history of medieval philosophy, the comparison of the Latin and Arabic Aristotelian commentary traditions must be concerned in large measure with the influence of Arabic authors, especially Avicenna and Averroes, upon their Latin successors. In the case of the commentary tradition on the Peri hermeneias, however, the question of influence plays little or no part in such comparative considerations. Yet the absence of a direct influence of Arabic philosophers upon their Latin counterparts does have its own peculiar advantages, since it provides an opportunity to explore the effects upon Aristotelian exegesis of the different linguistic backgrounds of Arabic and Latin authors.
1 With the exception of Averroes’s brief Middle Commentary, which was translated by William of Luna sometime during the thirteenth century, none of the Arabic commentaries on the Peri hermeneias was translated into Latin, and even Averroes’s work did not have a wide circulation. On the transmission, see Gauthier’s, R.-A.Introduction to the new Leonine edition of Thomas’s commentary, Expositio libri Peryermenias, vol. 1.1 of Opera Omnia, Aquinas, Thomas , 75-81.Google Scholar Perhaps this lacuna in the transmission of Arabic philosophy to the West is attributable to the long history of Latin philosophical commentary on the logica vetus, which made the aid of the Arabic commentators superfluous. The situation was far different from that pertaining to the assimilation of Aristotle’s unfamiliar metaphysical and physical works, in which the help of Arabic authors was indispensable.
2 For an overview of the Peri hermeneias commentaries prior to the time of Aquinas, see Gauthier’s, Introduction to Thomas Aquinas , 64-75.Google Scholar On the general subject of the Peri hermeneias in the West see Isaac, J., Le ‘Peri hermeneias’ en Occident de Boèce à saint Thomas (Paris: Vrin. 1953).Google Scholar For the most part, I have confined my remarks on the Latin authors to direct commentaries on the Peri hermeneias, and refrained from comparisons with more general logical texts, both by these commentators and by other thirteenth-century authors. I suspect, however, that there are some tensions between the approach to language and logic that is found in the commentary tradition, and that found in independent logical treatises and in works on speculative grammar. But a full investigation of this topic is beyond the scope of the present discussion.
3 I have omitted Averroes’s Middle Commentary even though it is the only Arabic text on the subject that was available in the West, since it is less concerned with the issues under consideration here than are the commentaries of Avicenna and Fārābī.
4 Martin of Dacia , q. 1, 236.1-3; see also 235.26-236.12.
5 For Boethius’s discussion of the term interpretatio, see Boethius , prima editio (hereafter la), 32.8-34.28; secunda editio (hereafter 2a), 4.15–13.24. The passage cited by Martin is la:32.11–12: ‘Interpretatio est vox significativa per se ipsam aliquid significans.’
6 Martin of Dacia , q. 1, 236.3-9
7 Unlike Martin, Aquinas does not add the difference “in which there is truth or falsity” to the Boethian definition of interpretatio. But he does argue that only “he who explains something to be true or false seems to interpret,” and thereby ultimately accepts that “only enunciative speech, in which truth or falsity is found, is called interpretation” (Thomas Aquinas  1.1, 6.48–52). Regarding the interpretation of Boethius, Isaac and Gauthier have both argued that Aquinas only shows awareness of the secunda editio of his commentary. See Isaac, , Le ‘Peri hermeneias’ en occident 100Google Scholar n. 1, and Gauthier’s Introduction, 49. Thus, Gauthier cites the secunda editio’s definition of interpretatio as a vox articulata per se ipsam significans (2a:6.4–5) as Thomas’s source for the definition of interpretatio as vox significatiua que per se aliquid significat, at 1.1, 6.35-37. However, Thomas’s citation seems closer to the prima editio’s definition of interpretatio as a vox significativa per se ipsam aliquid significans (32.11–12). This calls into question Isaac's claim regarding the prima editio that Thomas “n’y a fait aucune allusion.”
8 Aquinas, Thomas. “, 1.1, 5.15–16: “Cum autem logica dicatur rationalis sciencia, necesse est quod eius consideratio uersetur circa ea que pertinent ad tres predictas operationes rationis.”Google Scholar Aquinas is referring to the differences among the understanding of indivisibles, composition and division, and discursive reasoning; the Peri hermeneias is assigned to the second operation, composition and division.
9 Ibid., 1.1, 6, 59-61; see also 6.46–61
10 In the proemia to both hisPeri hermeneias and Posterior Analytics commentaries, Aquinas limits his characterization of logic to that of a rational science, although he does not explicitly deny logic the status of a linguistic science. See Expositio in libros Posteriorum analyticorum, Gauthier, R.-A., ed., vol. 1.2Google Scholar of Thomas Aquinas , Bk. 1, ch. 1 (Proemium), 3.1-5.50, for a more detailed version of the themes treated in the proemium to the Peri hermeneias commentary.
11 Kilwardby, Robert, Notule super Periarmenias Aristotilis, Osmund Lewry, P., ed., in Kilwardby, Robert , 379-1113.Google Scholar All citations of the introduction and first lectio of Kilwardby’s commentary refer to the edition appended to Fr. Lewry’s dissertation. Before his untimely death in 1987. Fr. Lewry was preparing an edition of the entire commentary, and had provided a draft version of the text to members of his graduate seminar at the Pontifical Institute of Mediaeval Studies in Toronto in the spring of 1983. This draft translation forms the basis for my remarks on the remainder of Kilwardby’s commentary, but references are also provided to one of the three manuscripts of the commentary, Venezia, Biblioteca Marciana lat. L.VI.66 , fols. lr-18v. For the reference to Boethius, see Boethius  2a:4.27–8, where the phrase sonum. cum quaedam imaginatione significandi is used as a definition of vox. At 5.22–6.3, it is used to distinguish interpretatio from vox locutio.
12 Robert Kilwardby , 379.11–22
13 None of the medieval commentators seems bothered by Boethius’s explication of signification by reference to imagination rather than intellect. In her translation of Aquinas’s commentary, Jean Oesterle suggests that the roots of this phrase are in Aristotle’s distinction between mere physical sound (psophos) and voice (phōnē) at De anima 2.8.420b27–421al, the latter requiring the presence of a soul capable of having phantasia of some sort. See Aristotle: ‘On Interpretation,’ Commentary by St. Thomas and Cajetan (Peri Hermeneias) (Milwaukee: Marquette University Press 1962), 17 n. 2.
14 Robert Kilwardby , fol. 7v, lines 4–11: Sed dicendum quod logicus considerans ueritatem et falsitatem circa orationem, cum ueritas et falsitas causantur in oratione a rebus significatis, diffinit orationem per res significatas, dicens, Oratio est uox significatiua etc.; set grammaticus, circa orationem considerans congruitatem et incongruitatem, cum congruitas et incongruitas causentur ab ipsis consignificatis, hec autem consequuntur res in quantum eis debetur constructio et ordinatio, quia sunt media construendi siue ordinandi dictionem cum dictione, diffinit orationem per ordinationem, dicens, Oratio est congrua dictionum ordinatio. Et sic diuersa intentio auctorum fecit diuersitatem diffinitionum. For the citation from Priscian, see Priscian  Bk. 2, 15, 1:53.28.
15 In the De ortu scientiarum, Robert Kilwardby , ch. 49, §468, 160.19-161.2, Kilwardby treats signification as a property common to all the sermocinales scientiae included in the trivium, i.e., logic, grammar, and rhetoric. Arguing that every sermo is a signum, he claims that all the linguistic sciences must be concerned with sermo significativus. The difference between the three arts derives from the fact that grammar confines its consideration of language to the representation of what is already nota, whereas logic and rhetoric are inquisitivus of what is ignota. Cf. §474, 162.22–3, where grammar is said to be [d]e sermone. significativo per se, rhetoric and logic de [sermone] ratiocinativo per se; and ch. 53, §493, where Kilwardby insists that logic is both rational and linguistic. This picture accords broadly with the suggestion in the Peri hermeneias commentary that both grammar and logic are rational as well as linguistic (for grammar’s objects are nota), although it obviously suggests that the significative-consignificative distinction is rather arbitrary. But cf. §484, 165.12–15, where grammar is associated with both modi significandi and mental concepts: “Subiectum enim sermo significativus est secundum quod huiusmodi; finis, congruus, et aptus modus significandi omnem mentis conceptionem; definitio, scientia de sermone docens omnem animi conceptionem significare.” The overall impression one gleans is that Kilwardby has a generally consistent approach to the question of the relations between logic and grammar, but is rather fluid in the terminology he uses to define their formal differences as distinct sciences.
16 Magnus, Albertus, Libri 2 PerihermeneiasGoogle Scholar in Albertus Magnus  vol. 1, Bk. 1, tract. 1, ch. 1, 373M6–18.
17 Ibid., 1.1.1, 374a3–27
18 Ibid., 1.1.1,375b10–18; the text cited is 11. 15-18. For the Avicennian background, to which Aquinas also alludes, see below, n. 68.
19 Ibid., 1.1.1, 375bl8–22.
20 Ibid., 1.1.1,375b23–5. The Borgnet edition attributes this position to Andronicus, although Boethius mentions Alexander and Aspasius, as do other thirteenth-century commentators. It is not clear whether the mistake is Albert’s or the editor’s; it is obviously due to Boethius's discussion of Andronicus’s misgivings about the authenticity of the text, which follows immediately after the discussion of the views of Alexander and Aspasius on oratio. For the controversy over oratio, see Boethius , 2a:10.4-11.11.
21 Albertus Magnus , 1.1.1, 375b25–38. Martin of Dacia also uses the couplets of truth-falsehood and congruity-incongruity to contrast the logical and grammatical approaches to speech. See Martin of Dacia  q.14, 248.14-249.30, “utrum pertineat ad logicum considerare veritatem et falsitatem.”
22 It is interesting to compare the Latin commentators with the Arabic philosophers on this point, for the latter always include particles in their discussions of the parts of speech in Peri hermeneias commentaries.
23 Martin of Dacia , q. 17, 253.14–17
24 Ibid., q. 20, 254.20-7. (Question 20 contains the replies to questions 17–20.)
25 Ibid., q. 20, 254.28-255.3
26 One reason for this reluctance to import the unmistakably logical terms of subject and predicate may come from the need to distinguish the approach of the Peri hermeneias from that of the Prior Analytics. Aquinas, for example, argues that the designations of ‘noun’ and ‘verb’ apply to the signs of simple intelligibles insofar as they are considered as parts of enunciations; when treated as parts of a syllogism, these same signs are considered not as nouns and verbs, but as ‘terms,’ and they are treated under this formality in the Prior Analytics (Thomas Aquinas , 1.1, 6.80-7.97). Cf. Robert Kilwardby , 386.26–30. Albertus Magnus argues more generally that since the notions of subject and predicate depend upon the notion of enunciation, they cannot be determined until after Aristotle has determined the nature of the enunciation. See Albertus Magnus  1.3.1, 400b34–401al4.
27 Robert Kilwardby , 386.9–26
28 Ibid., 386.36–40
29 Ibid., fol. 4v, line 3: “Primum dubitabile est propter quid diffinitur hic aliter nomen quam a Prisciano, cum essentia uniuscuiusque sit semel.” Priscian offers more than one definition of the noun, but the medieval commentators take as canonical that of signifying substance with quality. See Institutiones grammaticae, 2.18 (Priscian , 1:55.6): “Proprium est nominis substantiam et qualitatem significare”; and 2.22 (Priscian , 1:56.29-57.4): “Nomen est pars orationis, quae unicuique subiectorum corporum seu rerum communem vel propriam qualitatem distribuit. dicitur autem nomen. quasi notamen, quod hoc notamus uniuscuiusque substantiae qualitatem.”
30 Robert Kilwardby , fol. 4v, II. 4–7: “Set dicendum quod diuersa auctorum intentio fecit diuersitatem diffinitionum. Diffinitur ergo in grammaticis per partes sui essentie in quantum est constructibile, quia per substantiam et qualitatem; hie autem per partes sue essentie in quantum est subicibile. Et est ista diffinitio magis formalis; illa autem magis materialis, cum fiat per partes essentiales.” The claim that the logical definition is more formal, the grammatical more material, reflects the view that the vocal sound, which is more central to grammar, functions as the matter of an utterance.
31 Albertus Magnus , 1.1.3, 379a19-b8
32 Ibid., 1.2.1, 381a27–39
33 Cf. above, nn. 14-15.
34 These remarks in fact apply to the infinite verb, but the Latin commentators generally consider the logical status of infinite nouns and verbs to be equivalent. Ackrill translates the passage in question as follows: ‘Let us call them indefinite verbs, because they hold indifferently of anything whether existent or non-existent’; see The Complete Works of Aristotle: The Revised Oxford Translation 2 vols., Barnes, Jonathan, ed. (Princeton: Princeton University Press 1984) 1:25-38.Google Scholar For the Boethius text in question, see la:52.11–53.4; 2a:62.3–63.14. See also Ammonius , 41.16–42.8; medieval Latin translation, Ammonius , 79.62- 80.88. It should be noted that I am using ‘negation’ very loosely here. The medieval commentators all deny that infinite nouns and verbs constitute negations in the sense of complex, negative, complete enunciations.
35 See for example Thomas Aquinas , 1.4, 23.223–27: ‘Si autem imponeretur a priuatione, requireret subiectum ad minus existens; set quia imponitur a negatione, potest dici de ente et de non ente, ut Boetius et Ammonius dicunt.’
36 Martin of Dacia , q. 22, 260.19–26
37 Ibid., q. 22, 260.28-261.3
38 Simon of Faversham , q. 4, 152.27-153.3
39 Ibid., 165.11–12
40 Ibid., 165.31
41 Ibid., 165.34-166.1
42 Cf. Albertus Magnus’s justification, discussed below at nn. 44-48.
43 See above at n. 38.
44 Albertus Magnus , 126.96.36.1991a21–35
45 On Kilwardby see n, 12 above.
46 Albertus Magnus , 188.8.131.521a41–5. Albert goes on to argue that the infinite noun is not a negation simply, since a negation leaves nothing: Non enim simpliciter negatio est, quia negatio nihil relinquit (391b1–2). Perhaps this is a practical acknowledgement of the fact that infinite nouns are employed in Aristotelian logic.
47 Albertus Magnus  2.1.1, 426b17–35: In praecedentibus autem antecedentis libri dictum est quid est nomen, et quid innominabile est: non homo enim secundum logici intentionem non dico nomen: nec tamen in aliam cadit partem orationis: sed dicitur nomen infinitum, quod est nomen secundum aliquid, quia significat substantiam cum qualitate: sed non habet nominis perfectionem, scilicet quod significet et interpretetur quae sit ilia qualitas qua substantia nominis habet determinari: et quia finitam tollit qualitatem quae in homine est humanitas, et nullam ponit, ideo non interpretatur et significat substantiam nisi infinitam, cujus qualitas ipsa est infinitas: sed talis qualitas sufficit ad hoc quod sit nomen secundum intentionem grammatici, quae modos plusquam res attendit. Note that the last clause of this passage identifies the grammarian’s concern to be with modes, presumably of signification, rather than with things. Yet earlier, we saw Albert identifying grammar as concerned principally with the direct signification of things, in contrast with logic’s concern with things as signified in the mind (see n. 32 above).
48 As to Albert’s position on the problem of the inclusion of infinite terms in the theory of opposition, the passage cited in the preceding note actually occurs in Albert’s discussion of the theory of opposition. Albert has just distinguished (at Albertus Magnus  2.1.1, 426a39-bl6) affirmative enunciations with finite subjects from those with infinite nouns as their subjects. Although Albert reiterates that the infinite noun is innominabile because it is said of both being and non-being, he goes on to accept its inclusion as the subject of an enunciation, on the grounds that the infinite noun does signify something that is in some way one; he ends by declaring, et talis unitas sufficit ad unitatem subjecti in propositione sive in enuntiatione una (2.1.1, 426b37–9). In contrast to Simon, Albert argues the infinite verb cannot enter into an enunciation and retain its infinite status, since only the radix of the verb, the copula, remains in the infinite verb, while its predicative force is removed (2.1.1, 426b39–427b20). Despite this difference, Albert seems to share with Simon a common desire to defend the use of infinite terms in enunciations, and thereby to defend Aristotle’s practice in the later parts of the text. Yet despite this, both commentators are reluctant to reject the accepted interpretation of the alogical status of infinite terms. The result is that the ex professo discussions of questions regarding what makes a linguistic object logically interesting become almost irrelevant to determining specific points of logical doctrine and logical practice.
49 Thomas Aquinas , 1.4, 23.227–29, and in general, 23.207–39
50 Apart from his actual commentaries on the Organon Fārābī’s works dedicated to logic and language include the Kitāb al-ḥurūf (Other of Letters [or Particles]), Kitāb al-alfāẓ al-musta ‘malah fi al-manṭiq (Utterances Employed in Logic), and portions of the Iḥṣā’ al-‘ulūm (Catalogue of the Sciences) and Kitāb al-tanbīh ‘alā sabil al-sa’ādah (Reminder of the Way to Happiness).
51 Arabic texts will be cited initially by their Arabic titles, with an English translation of the title in parentheses. In subsequent citations, a shortened version of the English translation will be used. Where medieval Latin versions of the Arabic texts are available, I have in most cases provided references to these as well as to the Arabic original.
52 The term usually rendered as ‘grammar,’ naḥw, is, strictly speaking, narrower in range than its English counterpart, being closer to ‘syntax.’ See for example Fārābī, , Iḥṣā’ al- ‘ulūm (Catalogue of the Sciences), 3rd ed., Amin, Uthman, ed. (Cairo: Librairie Anglo-Égyptienne 1968), 62.1,Google Scholar where grammar is said to concern itself with a particular part of the science of composite expressions, namely, the study of the proper endings to be used when forming complex expressions. For the medieval Latin translation of this work by Gerard of Cremona, see Catálogo de las ciencias, 2nd ed., Palencia, A. G., ed. (Madrid: Consego Superior de Investigaciones Cientificas 1953), 12419-20.Google Scholar The subject-matter of grammar, then, is not language in general, but ‘irāb, inflection (62.5; Palencia, 124.24–125.1). On this point, cf. Elamrani-Jamal, A., Logique aristotélicienne et grammaire arabe (Paris: Vrin 1983), 98-101;Google ScholarBohas, G., Guillaume, J.-P., and Kouloughli, D.E., The Arabic Linguistic Tradition (New York and London: Routledge 1990), 49-72Google Scholar
53 Catalogue of the Sciences 68.4-7; Palencia, 128.25–129.4. Fārābī also extends the analogy to include prosody at lines 8–10, Palencia lines 4–9. The Latin translator apparently did not know the Arabic term for prosody, al-‘arūḍ, for he simply transliterates the Arabic. According to Palencia’s apparatus, there is, however, a marginal gloss indicating that the Arabic term means de ponderibus uersuum.
54 Catalogue of the Sciences 74.10-15; Palencia, 133.23–134.1. Cf. Al-Tawṭi’ah (Introduction [to Logic]), al-Ajam, Rafiq, vol. 1 of Al-Manṭiq ‘inda al-Fārābī (The Logic of al-Fārābī) (Beirut: Dar el-Mashreq, 1985), 55.9-56.2;Google Scholar there is an English translation of this text, along with an earlier edition, in ‘Al-Fārābī’s Introductory “Risālah” on Logic,’ Dunlop, D. M., ed. and trans., Islamic Quarterly 3 (1956–57) 224-35.Google Scholar The passage in question is translated in §1, 230.
55 Catalogue of the Sciences 76.2-77.15; Palencia, 134.23–136.4. See esp. 77.5–7; Palencia, 135.24–28: “So the science of grammar in every language considers only what is specific to the language of that nation; and [it considers] what is common to [their language] and to other [languages], not insofar as it is common, but insofar as it is found in their language in particular.”
56 On Yahyā see Rescher, Nicholas, The Development of Arabic Logic (Pittsburgh: Pittsburgh University Press 1964), 130–4.Google Scholar The text I will be discussing is ‘Yaḥyā ibn ‘Adī’s Treatise on the Difference between the Arts of Philosophical Logic and of Arabic Grammar (Maqalah fi tabyīn al-faṣl bayna sinā ‘atay al-mantiq al-falṣafi wa-al-naḥw al-‘arab),’ Endress, Gerhard ed., Journal of the History of Arabic Science 2 (1978)Google Scholar Arabic pagination 38–50/English pagination 192–81 (note that the English pagination is in inverse numerical order). There is a French translation in Elamrani-Jamal, , Logique aristotēlicienne et grammaire arabe, 187-97;Google Scholar a German translation is found in Endress, Gerhard, ‘Arabische Philologie und griechische Philosophie im Widerstreit,’ in Mojsisch, Burkhard, ed., Sprachphilosophie in Antike und Mittelalter, (Amsterdam: Verlag B. R. Grüner 1986), 163-299.CrossRefGoogle Scholar
57 Philosophical Logic and Arabic Grammar, §3, E191/A40.1–4
58 Ibid., §§6–9, E190–89/A41–42. Elamrani-Jamal has an excellent discussion of the implications of Yaḥyā’s emphasis upon correct ‘irāb or vocalization. See the reference in n. 52 above.
59 Philosophical Logic and Arabic Grammar, §10, E188/A43.1. Yaḥyā notes in this passage that the grammarians claim that it is their intention to consider expressions as significant of meanings, a claim which Yaḥyā contends is misleading. See §10, E188/A43.1–5.
60 Philosophical Logic and Arabic Grammar, §§11–12, E188-E187/A43–44.
61 On this point in Latin speculative grammar, see Covington, Michael A., Syntactic Theory in the High Middle Ages: Modistic Models of Sentence Structure (Cambridge: Cambridge University Press 1984), 33–5;CrossRefGoogle Scholar and Rosier, Irène, La grammaire spéculative de Modistes (Lille: Presses Universitaires de Lille 1983), 45, 56–7,Google Scholar 212 n. 97. On this sense of meaning, currens currere mean the same thing in Latin, just as do Zaydun Zaydan in Arabic. An example of this view can be found in Boethius of Dacia , 55.60-56.71.
62 As I note below at nn 100–2, there is an analogue to the notion of modi significandi in Avicenna’s Peri hermeneias commentary, but it is not used to provide the basis for a universal science of grammar. Avicenna’s remarks in this context do suggest, however, that he is willing to construe meaning more broadly than is Yaḥyā.
63 This view as it applies to Fārābī is upheld by Zimmermann, F. W., in his Introduction to Al-Fārābī’s Commentary and Short Treatise on Aristotle’s ‘De interpretation’ (Oxford: Oxford University Press and the British Academy 1981),Google Scholar esp. xli-lvii; cxviii-xxii; cxxxi, cxxxviii-ix. It is challenged by Elamrani-Jamal, , Logique aristotelicienne et grammaire arabe, 77, 88.Google Scholar
64 Avicenna, Al-Shifā’ (Healing), vol. 1, Al-Mantiq (Logic), part 1, Al-Madkhal (Isa-goge), Anawati, G., El-Khodeiri, M., Al-Ahwani, F. and Madkour, I., eds. (Cairo: Al-Matba’ah al-Amiriyah 1952),Google Scholar Bk. 1, chap. 4, 23.5–6, 24.3-4. For the medieval Latin translation of this portion of Avicenna’s logic, see Opera philosophica, 2 vols. (Venice 1508), 1:3rb19–21, 41–2. Avicenna’s rather derisive description of the Fārābīan view is considerably toned down in the Latin, which simply reads ideo deliquerunt.
65 Avicenna does qualify this with respect to those logical arts whose function is essentially communicative, i.e., where discussion and debate are involved, as in dialectic, rhetoric, and poetics. See my Logic and Aristotle’s ‘Rhetoric’ and ‘Poetics’ in Medieval Arabic Philosophy (Leiden: Brill 1990), 60-1.
66 Isagoge 1.2,15.1-8; Latin l:2rb29–42
67 Ibid., 1.4, 22.14-17; Latin I:3ra64–3rb4
68 Though one would presume that under these ideal conditions, the art of logic itself would be dispensable, since the circumstances rendering error possible would be entirely eliminated. For, in this case, no process would be needed to progress from the known to the unknown, whereas facilitating the acquisition of knowledge of the unknown is the purpose of logic according to Avicenna. On this last point see, for example, Isagoge, 1.3, 16.15-17.5. For Avicenna’s departure from the Aristotelian theory of the dependence of thought upon imagination, the canonical text is Bk. 5, chap. 5 of the De anima part of the Shifā’. There Avicenna rejects the theory of abstraction from images for a theory of the direct emanation of intelligibles from the Agent Intellect; the images are thereby assigned a purely preparatory, rather than a specifying, function in intellection. See Avicenna’s ‘De anima,’ Being the Psychological Part of Kitāb al-Shifa’ Rahman, F., ed. (London: Oxford University Press 1959), 234.12-236.2;Google Scholar medieval Latin translation in Avicenna , 2:126.27-128.63.
69 Isagoge 1.4, 23.1-4; Latin 1:3rb12–19. Despite the obvious critique of Fārābī in this passage, Avicenna may well have adapted this notion of the primacy of thought over expression from Fārābī himself, who draws upon it in his Sharḥ (Long Commentary) on the Peri hermeneias and in the Utterances Employed in Logic. See below at n. 75; cf. also Zimmermann, Introduction to Fārābī’s Commentary and Short Treatise, xlii.
70 Sharh al-Fārābī li-kitāb Arisṭūṭālīs fi al-‘ibārah (Long Commentary on ‘Interpretation’), Kutsch, W. and Marrow, S., eds. (Beirut: Imprimerie Catholique 1960) 1916-18,Google Scholar 23.11–14; Zimmermann trans., 3, 8–9. I have used Zimmermann’s translation throughout, with some modifications where indicated. On the construal of the title, see Zimmermann, 3–4 n. 5. It should be noted that Zimmermann discerns two different versions of Fārābī’s proemium to the commentary (Introduction, cxliv, and 4 n. 3); my remarks are based on both versions.
71 Introduction to Logic 59.14-15; Dunlop trans., 233. Cf. Catalogue of the Sciences, 78.3–4; Palencia, 134.15–16, where the same phrase is used in the same context.
72 Since grammar is not viewed as a truly universal, and hence truly philosophical, science by Fārābī, the philosophical study of language cannot be carried out by the grammarian.
73 Long Commentary 24.2-7; Zimmermann trans., 10 (modified). The Arabic commentators all refer explicitly to sensibles when discussing what is outside the soul. This seems to stem from Aristotle’s reference at 16a3–4 to ‘passions’ (pathēmata) in the soul, the phrase that reputedly caused Andronicus to dispute the authenticity of the text. The Arabic translation renders this as āthār ‘traces,’ the term commonly used for the forms found in the common sense and imagination insofar as they are remnants of sense perception. Thus, the Arabic translation has led the commentators to assume Aristotle is implicitly invoking the causal account of perception in the De anima, and through it alluding to the ultimate origin of intellectual thought in the senses. For the Arabic version of the text, see Die Hermeneutik des Aristoteles in der arabischen Übersetzung des Isḥāk Ibn Ḥonain, Pollak, Isidor, ed. (Leipzig 1913; reprinted Nendeln, , Liechtenstein: Kraus 1966), 14.Google Scholar
74 Long Commentary25.22; Zimmermann trans., 13 (modified)
75 Ibid., 25.23-26.1; Zimmermann trans., 13–14 (modified). In his Introduction, xlii, Zimmermann draws attention to a parallel passage in the Kitāb al-alfāẓ almusta’malah fi al-manṭiq (Utterances Employed in Logic), Mahdi, M., ed. (Beirut: Dar el-Mashreq 1968) 102.7–15,Google Scholar on the method of instruction known as the ‘substitution of words’ (scil. for thoughts), where the same theme of facilitating conceptual understanding through language is evoked, with somewhat pejorative overtones.
76 Long Commentary, 26.24-25; Zimmermann trans., 14–15
77 Avicenna, Al-Shifā’, vol.1, Al-Manṭiq (Logic), pt. 3, Al- ‘Ibārah (Interpretation), El-Khodeiri, M. and Madkour, I., eds. (Cairo: Dar el-Katib al-‘Arabi 1970),Google Scholar Bk. 1, ch. 1, 5.14–17
78 Ibid., 1.1, 6.1–6
79 The term bāṭil, for example, is used by Avicenna to describe fictional concepts like the phoenix. See, for example, ‘«Epitre sur la disparition des formes vaines» d’Avicenne,’ (French) Michot, J. R., ed. and trans., Bulletin de philosophie méediévale 29 (1987) 152-70,CrossRefGoogle Scholar esp. 155.8; English translation in Avicenna’s, “Letter on the Disappearance of the Vain Intelligible Forms After Death”,’ Michot, J.R., trans., in Bulletin de philosophie méediévale 26–7 (1984–85) 94–103,Google Scholar esp. 98.
80 On the use of this term— which does not mean ‘indefinite’ except in logical contexts— see Zimmermann, Introduction to Fārābī’s Commentary and Short Treatise cxx and n. 2.
81 The Latin commentators’ discussions of the inflections of the noun were not included in the first part of this study in the interest of brevity, although this topic was also a common occasion for the Latin authors’ presentation of their views on the relations between logic and grammar.
82 A proper name in Arabic, literally meaning ‘the king’s servant.’
83 Long Commentary 32.5, my translation (cf. Zimmermann trans., 20).
84 Ibid., 38.4; Zimmermann trans., 28. The topic here is not the indefinite noun, but the indefinite verb, and Aristotle’s claim that it is said of both existent and non-existent things. Fārābī extends his comments here to cover all words of the form ‘non-X,’ be they nouns or verbs.
85 Long Commentary 38.23-39.2; Zimmermann trans., 29. Cf. Fārābī’s ‘Ibārah ([Short Treatise] on Interpretation) in vol. 1 of Al-Manṭiq ‘inda al-Fārābī, 147.12–17; Zimmermann trans., 234. (Note that the page numbers in the margins of Zimmermann’s translation refer to an earlier edition of Fārābī’s Short Treatise.)
86 Long Commentary 39.6-18; Zimmermann trans., 30. Fārābī’s discussion of propositions with indefinite terms is complex and extended, and a full consideration of his views on the topic is beyond the scope of this paper. Zimmermann offers a brief discussion in his Introduction to Fārābī’s Commentary and Short Treatise, lxiii-lxvii. An extended consideration of indefinite terms in a variety of commentators is found in Soreth, M., ‘Zum infiniten Prädikat im zehnten Kapitel der aristotelischen Hermeneutik,’ in Stern, S. M.et al., Islamic Philosophy and the Classical Tradition (Columbia, SC: University of South Carolina Press 1972), 389–424.Google Scholar Averroes’s views on indefinite terms have been discussed in two articles by Wolfson, H.A., ‘Infinite and Privative Judgments in Aristotle, Averroes, and Kant,’ Philosophy and Phenomenological Research 8 (1947) 173-87;CrossRefGoogle Scholar and ‘The Twice-Revealed Averroes,’ in Ross, J. F., ed., Inquiries into Medieval Philosophy: A Collection in Honor of Francis P. Clarke (West-port, CT: Greenwood 1971), 211-41;Google Scholar both articles have been reprinted in Wolf-son’s Studies in the History of Philosophy and Religion, 2 vols., Twersky, I. and Williams, G. H., eds. (Cambridge, MA: Harvard University Press 1973), 2:542-55Google Scholar and 1:371–401, esp. 387–97. All subsequent citations refer to the reprinted versions of these two articles. It ought to be noted that in the Short Treatise, Fārābī does allow that indefinite nouns can be extended from their proper signification of privation in two ways. First, they can be extended to allow their predication of all subjects which share a common genus or species with the subjects to which the privation properly applies, e.g., ‘non-rational’ as applied to horse, or ‘non-bearded’ to women. Second, they can be used in a still wider sense of all existent things, even those outside the genus to which their corresponding possession properly applies. Fārābī’s example is the use of negative predications of God. But Fārābī insists that even in these extended uses, the subject of predication must be something existent. See Short Treatise, 153.10–155.6; Zimmermann trans., 238–40. On this extended and looser reading of indefinite terms, Fārābī’s theory appears closest to that of Aquinas, who argued that the mind must presuppose some suppositum in order to use infinite terms. See above at n. 49.
87 See, for example, Short Treatise 135.17-136.1; Zimmermann trans., 222.
88 Ibid., 136.3-5; Zimmermann trans., 222. Cf. 138.6–9; Zimmermann trans., 225, on the indefinite verb. Fārābī's reference, in the parallel passages of the Long Commentary (39.11–13; Zimmermann trans., 30), to the discussion of privation in Metaphysics 5.22.1022b32–33, may explain his claim that indefinite terms in other languages than Arabic actually inflect like single terms. For Fārābī seems to think that Aristotle is talking about terms with a privative alpha in both the Peri hermeneias and the Metaphysics, and is unaware that the former text in fact discusses terms with the negative particle ou preceding them. This conflation of privative terms with indefinite ones may have been reinforced by the Arabic versions of the Metaphysics: the lemmata in Averroes’s Long Commentary indicate that the privative alpha was rendered into Arabic by lā, the same term used to translate ouin Aristotle’s discussion of indefinite nouns and verbs in the Peri hermeneias. See Averroes, , Tafsīr mā ba ‘d al-tabī ‘ah (Great Commentary on the Metaphysics), 4 vols., 2nd ed., Bouyges, M., ed. (Beirut: Dar el-Machreq [Imprimerie Catholique] 1967, 1973), 2:647.5–6;Google Scholar and see the Arabic version of the Peri hermeneias, Pollak, ed. (cited in n. 73 above), 3.30 and 4.12. Averroes himself, it should be noted, also glosses Aristotle’s remarks in the Metaphysics as referring to ‘metathetic nouns’ (al-asmā’ al-ma’dūlah) (Great Commentary, 2:647.6–7).
89 Al-Ishārāt wa-al-tanbīhāt, Forget, J., ed. as Le livre des théorèmes et des avertissements (Leiden: Brill 1892), 279-11;Google Scholar 28.10–29.2; English translation, Remarks and Admonitions, Part One: Logic, Inati, S.C., trans. (Toronto: Pontifical Institute of Mediaeval Studies 1984), 83, 85–86.Google Scholar Like Fārābī (see n. 86 above), Avicenna allows for a ‘more general’ use of indefinite terms, in which their meaning is taken to be broader than that of a privation. He is also insistent that, whether indefinite terms are taken in a broader or a narrower sense, their predication implies an existent subject. Avicenna admits in his Interpretation, 1.4, 27.9–28.6, that Aristotle’s remarks on indefinite verbs could be taken to imply that they can be said of non-existent subjects, but he adds that Aristotle is wrong if this is what he meant. In the Najāh (Deliverance), Fakhry, M., ed. (Beirut: Dar al-Afaq al-Jadidah 1985), 54.19–26,Google Scholar Avicenna similarly allows the use of indefinite terms in broader and narrower senses, but prohibits their application to fictional entities like the phoenix. Thus, while it is proper to say, ‘The phoenix is not sighted’ (simple negation, ‘Not [S is P]’), it is not proper to say, ‘The phoenix is non-sighted’ (‘S is not-P’). In the Remarks and Admonitions, this same rule is applied to all affirmative statements. Only negative statements can be made about non-existent beings (28.16–29.1; Inati trans., 86). On this cf. Fārābī, Short Treatise, 155.2–4; Zimmermann trans., 240.
90 In the Interpretation, Avicenna presents indefinite nouns as compounds with the particle lā, reflecting the Arabic versions of the Peri hermeneias. In other works, he often reverts to the more natural Arabic construction with ghayr (literally, ‘other than’). On the artificiality of the lā compounds with reference to Fārābī, cf. Zimmermann, , Introduction to Fārābī’s Comentary and Short Treatise, cxxxiii-iv.Google Scholar
91 Interpretation 1.2, 12.13
92 Ibid., 1.5, 32.18-33.3
93 For a discussion of this distinction in Arabic logic, see my Logic and Aristotle’s ‘Rhetoric’ and ‘Poetics,’ 71-8
94 The term ‘judgement,’ (ḥukm) is used in a very broad sense in the Arabic commentary tradition on the Peri hermeneias, where it is roughly equivalent to ‘meaning’ or ‘idea.’ Its use does not imply composition, assent, or truth-value.
95 Interpretation, 1.2, 12.16-13.7
96 Remarks and Admonitions 28.13-29.2; Inati trans., 85–6
97 The long discussion in Fārābī’s Short Treatise of opposition in the metathetic proposition (al-ma‘dul, i.e., one containing an indefinite predicate) displays a similar concern with the applicability of the theory to syllogistic. Thus, Fārābī closes his discussion with a consideration of the logical equivalence of negations said of existent subjects and the corresponding metathetic affirmations, in which he focuses upon the problem of including negative predications as the minor premise of a first figure syllogism (154.16–155.4; Zimmermann trans., 239–40). Fārābī concludes by declaring that the extension of indefinite nouns to this wider meaning (i.e., as simply requiring an existent subject) is “of enormous benefit to the sciences” (155.4–5; Zimmermann trans., 240). For discussions of the use of al-ma‘dūl as a technical term for statements containing indefinite terms, see Zimmermann's Introduction to Farabi‘s Commentary and Short Treatise, lxiii n. 1; and Wolfson, ‘Twice-Revealed,’ 394; ‘Infinite and Privative Judgments,’ 545. Zimmermann and Wolfson argue convincingly that the use of ma ‘dūd ‘udūl represents a translation of Theophrastus’s Greek term metathesis, and thus reflects the secondary meaning of ‘adala ‘to deviate,’ not its meaning ‘to be equal,’ as Inati suggests (Remarks and Admonitions, 85 n. 28).
98 Peri hermeneias, chap. 2,16a32-b5
99 Interpretation, 1.2, 13.8–12
100 On grammar and vocalization, cf. n. 52 above.
101 Interpretation, 1.2, 13.16-14.14
102 Ibid., 1.2, 14.14-16. As with the indefinite noun, it is again unclear whether Avicenna’s remarks are fully compatible with the attitude expressed by Yaḥyā ibn ‘Adī (nn. 56–60 above), in which all grammatical operations are said to be irrelevant to meaning. Since Avicenna accepts the confinement of grammar to what is idiomatic to a particular language, he would appear to accept the general outlines of Yaḥyā’s argument. But he does seem to be more flexible than Yaḥyā in allowing that grammatical operations do affect meaning in some way, by contracting or restricting its extension. Thus, it is difficult to see how, given the passage under consideration, Avicenna could accept Yaliya’s claim that the cases of the noun in no way affect its meaning. However, Avicenna’s analogy with the addition of an accidental property to a definition may offer a means of reconciling his view with that of his predecessor. For Yaḥyā’s basic point is that grammatical operations do not affect the word’s signification of one essence rather than another, a point which seems akin to Avicenna’s claim that the word is altered by its cases only in the way that ‘human being’ is altered by the addition to it of the accident ‘white.’ The modification in both cases is an accidental, not an essential, one.
103 Short Treatise, 136.10–11, my translation (cf. Zimmermann trans., 222).
104 Ibid., 136.11-15. The translations of the technical terms coined by Fārābī are my own here; compare Zimmermann’s translation, 222–3. For further remarks on Fārābī’s deliberate conflation of all the oblique cases with the genitive, cf. Zimmermann, 222–3 n. 13, 224 nn. 1, 7.
105 Short Treatise, 137.3-7; Zimmermann trans., 223–4
106 Catalogue of the Sciences, 77.12-15; Palencia, 135.28–136.4
107 Kitāb al-tanbīh ‘ala sabīl al-sa‘ādah (Reminder of the Way to Happiness), Yasin, J. A., ed. (Beirut: Dar al-Manahel 1987), §19, 83.7-84.4;Google Scholar medieval Latin translation, ‘Le «Liber exercitationis ad viam felicitatis» d‘Alfarabi,’ Salman, H., ed., Récherches de theologie ancienne et medievale 12 (1940) 33–48Google Scholar (the translated passage is found at §40, 47.45–48.7).
108 Catalogue of the Sciences, 76.8-77.4; Palencia, 135.1–17
109 Utterances Employed in Logic, §2, 42.8–12
110 See Isagoge 1.3, 20.14–15: “And the relation of this art to the internal reflection which is called ‘inner speech’ (al-nuṭq al-dākhilī) is like the relation of grammar to the external interpretation which is called ‘external speech’ (al-nuṭq al-khārijī).”
111 De Anima, 3.7.431a16–17; 431b2; 3.8.432a3–10. Cf. De Memoria 1.449b3–450al.
112 For Avicenna’s repudiation of the dependence of thought upon images even as an efficient cause of thought, cf. n. 68 above.