SOURCE: "Avicenna on the Subject Matter of Logic," in The Journal of Philosophy, Vol. LXXVII, No. 10, October, 1980, pp. 746-64.

[Here, Sabra outlines Avicenna's influential conception of logic as a part of philosophy that can lead one to "knowledge of the unknown. "]

I think it is true to say that modern logicians have no great interest in the ancient debate about whether logic was a part or an instrument of philosophy. They are of the opinion that the debate, at least in the form it took in the ancient schools of Greek philosophy, raised a question the solution of which was largely a matter of convention. Avicenna would readily agree, and for the same reason; in one place at least he characterized the question as nothing more than a quibble about the meaning of words. But in both ancient and medieval discussions the question was often linked with another concerning the subject matter of logic. If, as the Platonists and the Stoics maintained, logic is a part of philosophy and the various parts of philosophy are studies of various portions or aspects of being, then what portion or aspect of being should be assigned to logic? This was not a verbal question. And since Avicenna decided to come down on the side of the Academy and not on the side of his "friends" the Peripatetics who maintained that logic was only an instrument, it is not surprising that he should take the trouble in his Kitab al-Shifa' to expound his views on these two interrelated questions.

The dispute as to whether logic was a theory or an instrument has a further significance for the historian of Islamic thought: it became part of a continuing struggle of far-reaching consequences between the champions of Arabic and Islamic learning and the followers of an imported Hellenistic tradition. It should be remembered that in Islam the trivium was not an accepted category; logic and grammar stood on opposite sides of the fence, supported by rival groups. The word chosen by the translators of Aristotle for the art of logic, mantiq (speech), could naturally be used, and was sometimes used, as a title of works on grammer; and this alone was bound to impose the question as to which group, the grammarians or the logicians, was to be regarded as the true custodian of sound discourse. The ensuing controversies may have been motivated in part by religious or nationalistic impulses, but they were not devoid of philosophical interest—being ultimately concerned with the relation between language and thought. The documents that have survived from the early period of this debate appear to indicate that it was the grammarians who had the better of the argument. Logicians were content at first to claim that they were concerned with meanings whereas grammar was competent to deal only with words. This was too simple a view of the task that the grammarians had set for themselves. In the famous debate that took place in the tenth century between the logician Matta ibn Yunus and the grammarian Abu Sa'id al-Sirafi, both recognized leaders in their respective disciplines, the case for logic appears to be vulnerable and lacking in sophistication. With regard to the logicians' view of grammar, Sirafi cites example after example of how subtle shades of meaning are reflected in linguistic features which it was the business of the grammarian to identify. It is true that Matta—whose translations of Greek philosophical works into Arabic were done not directly from the Greek but from Syriac—is being repeatedly put on the defensive by being reminded of his ignorance of Greek, the original language of the logic he now claims to explicate in another language (Arabic) which he has not mastered. The audience is clearly unsympathetic to him, and he is rarely given the chance to answer the questions put to him. But even with the admission that this was perhaps not a fair trial, one comes away from reading it with the clear impression that it was Sirafi who had a deeper appreciation than his adversary of the incongruences generated by the "creation of a language within a language," as he aptly described the logicians' enterprise.

Abu Nasr al-Farabi, at one time a student of Matta, was the first Arabic logician to take seriously the questions of the relation of logic to grammar and of language to thought. His Enumeration of the Sciences briefly formulates the idea of logic as universal grammar (grammar furnishes the rules proper to the utterances of a given language; logic furnishes the rules common to the utterances of all languages), and states some of the implications of the connection between "inner" and "outer" speech. His book on The Utterances Employed in Logic can be viewed as an attempt to face the problems of introducing an artificial mode of speaking into natural Arabic. In his so-called Short Commentary on Aristotle's Prior Analytics, he tried to convince other Muslims of the usefulness of Greek logic by illustrating Aristotelian forms of inference in terms taken from Islamic theology and jurisprudence. He thus preceded al-Ghazali in formulating the kind of argument that finally succeeded in securing for Aristotelian logic a permanent place in Muslim education.

It is one of the paradoxes of Islamic intellectual life that the man most responsible for admitting Aristotelian logic into the scheme of traditional learning was an opponent of Greek philosophy who wrote a powerful book in its refutation. The great religious thinker Abu Hamid al-Ghazali (d. 1111) not only rejected the metaphysical doctrines of Peripateticism and Neoplatonism but also warned against the dangers of studying astronomy and mathematics. Nevertheless he wrote several books on Aristotelian logic and, what was more crucial for the history of logic in Islam, prefaced his influential work on Islamic jurisprudence (al-Mustas-fd) with a lengthy introduction in which he went so far as to say that without logic one could not be sure of any part of knowledge (whether secular or religious, theoretical or practical). Al-Ghazali was able to do this because he understood logic as a mere instrument, a kind of "balance" for weighing arguments which did not commit its user to any doctrine or belief.

Ghazali's attitude was in great contrast to that of another religious thinker who lived some two hundred years after him. Ibn Taymiyya was a strict jurisconsult who had no use for mystical or speculative approaches to religion, let alone an unbridled habit of thought such as philosophy. He could not tolerate the fact that Greek logic had appeared to gain a firm foothold in the field of religious learning, and he singled it out for a concentrated and persistent attack. Unlike Ghazali, Ibn Taymiyya believed that Aristotelian logic was committed to certain metaphysical doctrines from which it could not be detached. One could not adopt Aristotelian logic without being contaminated by its false presuppositions. One of these was a pervasive belief in universals or essences. According to Ibn Taymiyya, only individuals exist, and what is called "essence" is nothing but a conventional device for grouping individuals together for practical or theoretical purposes of our own. A universal term like 'man' does not refer to something shared by individuals, for no such thing exists. The inevitable conclusion was a total rejection of Aristotle's theories of definition and syllogism.

Ibn Taymiyya's criticisms, often acute and original, constituted the most radical critique of Aristotelian logic in the Arabic language. But they were not the work of a concerned logician; their aim was not to reform logic but to destroy it. They came too late in history to achieve that aim on a wide scale; the trend initiated by Ghazālī had taken root and was already well established. But the logic that Ghazālī had made acceptable was not a part of philosophical inquiry, a search for new truths, but an instrumental discipline consisting of a fairly fixed set of rules that one learned in order to apply them in other disciplines.

Avicenna, too, appreciated the instrumental character of logic, but his perspective differed from that of any of the thinkers I have mentioned. In broad terms he belonged to the same philosophical tradition to which al-Fārābī and al-Kindī before him also belonged. But he was more independent of mind than either of his two predecessors, and he spoke more often than they with an individual voice of his own. He was frequently critical or skeptical of the Aristotelianism he embraced and modified, and openly dissociated himself from the Peripatetic school of Baghdad, feeling himself the equal of the ancient commentators whom he read of course in translation. In matters of logic, as in other parts of his philosophy, he helped himself more freely than members of that school to Platonic (and Stoic) doctrines which had already been fused together in the late Greek writings that became the common heritage of Islamic philosophers. He did not always do this in the spirit of eclecticism, but often as the work of an independent thinker who felt able and obliged to make up his own mind.

Logic occupies a major part of Kitāb al-Shifā', the huge philosophical summa which Avicenna completed just before he reached the age of forty. It is known from Avicenna's own account and from a supplementary account provided by his pupil al-Jūzjānī, that the book was composed in varying and sometimes difficult circumstances which had their effect on the character of its various parts. Al-Jūzjānī says, for example, that Avicenna dictated most of the Physics and the Metaphysics during a period of only twenty days without referring to other writings. When he came to write the Logic, however, he was able to consult the books of others "whose order [of treatment] he followed and whose objectionable views he discussed"—a fact which the historian of Avicenna's logic must bear in mind. The logical part of the Shifā' is divided into sections corresponding generally to the parts of Aristotle's Organon. Only the first section, the Introduction paralleling Porphyry's Isagoge, was translated into Latin in the middle ages. It happens to be the section in which Avicenna directly addresses the question of the subject matter of logic. Avicenna's discussions thus continue a tradition that goes back to the Greek commentators, and his own treatment of these questions entered into the stream of philosophical thought in the West. In this [essay] I shall not in general be concerned to reconstruct the complex process of transmission of Greek ideas into Arabic, or follow Avicenna's discussions into the writings of medieval thinkers; my main object will be to identify and clarify Avicenna's views. I hope that the following remarks, despite their preliminary character, will not fail to show that Avicenna's style of thinking and writing does lend itself profitably to the kind of analytical approach attempted here.

To avoid confusion I shall refer to sections of Avicenna's Logic in the Shifā' by English titles, such as "Introduction" or "Interpretation," reserving for Greek works their commonly used titles in Latin or Greek.

Chapter 2 of the Introduction, on the chief divisions of the sciences, ends with a longish passage giving Avicenna's first statement in the Shifā' on the nature of logic and its relation to the other sciences. The essences of things (māhiyyāt al-ashyā'—māhiyya …) may exist in the actual things (a'yān al-ashyā') or in thought (fīal-tasawwur.) Certain accidents (a'rād) attach themselves to the essences when these possess one or the other of the two modes of existence. We may therefore examine the essences in themselves, without reference to their existence in individuals or in thought, or our examination of them may involve those adventitious properties which accrue to them in consequence of their external or mental existence. Avicenna here gives some examples of the kind of accidents that may attach to essences as mental entities: being subject or predicate, universality or particularity of predication, essential or accidental predication. He explains his examples by briefly remarking that "in external things there is no essential or accidental predication, nor is a thing a subject or a predicate, a premiss or a syllogism or the like."

These opening sentences are misleading in that they give the impression that the accidents exemplified here come into being simply as a result of bringing the essences into one's mind. But Avicenna seems to be struggling to dispel this misunderstanding in the following words:

If we wish to investigate things and gain knowledge of them we must conceive them; thus they necessarily acquire certain states (ahwāl) that come to be in conception: we must therefore consider those states which belong to them in conception, especially as we seek by thought to arrive at things unknown from those that are known. Now things can be unknown or known only in relation to a mind; and it is as concepts that they acquire what they do acquire in order that we move from what is known to what is unknown regarding them, without however losing what belongs to them in themselves; we ought, therefore, to have knowledge of these states and of their quantity and quality and of how they may be examined in this new circumstance (emphasis added).

I take this to mean that, although the properties of being a subject or a predicate or the like can attach only to concepts and not to external things, they do so only when the concepts are manipulated for the purpose of arriving at (or conveying) a piece of knowledge. Thus, in addition to the two varieties of investigation whose aim is to gain knowledge of external and mental things as such, there exists an inquiry whose aim is to be of use in carrying out the other two investigations. Such an inquiry is called "logic."

Avicenna is thus arguing that logic has its own subject matter which it does not share with any other science. But because of the very nature of this subject matter (properties acquired by concepts when organized for the purpose of attaining or transmitting knowledge), he maintains at the same time that the goal of logical investigation is to help in other investigations. He concludes this passage by saying that if philosophy is understood as the investigation of external and conceptual things as such, then logic is not a part of philosophy, but, as an aid in other investigations, it is an instrument of philosophy. If, however, the term 'philosophy' is applied to "all manner of theoretical investigation," then logic is a part of philosophy and a tool for the other parts. To Avicenna's mind, the question whether logic is a part or an instrument of philosophy is both false and futile—false because it presupposes a nonexistent contradiction between the two roles and futile because "to busy oneself with such matters serves no purpose." But this brief discussion at least allows him to offer something like a definition of logic: it is an inquiry into concepts, and into their properties, insofar as they can be made to lead to knowledge of the unknown.

In the Introduction Avicenna has no name for those concepts which, on account of certain properties that attach to them in the context of proof, he sets apart as the proper object of logic. He does, however, provide such a name in his Metaphysics, in a passage which [William and Martha Kneale, in their The Development of Logic] believed to be "the origin of that discussion of first and second intentions which continued until the end of medieval logic":

As you have known, the object of logic was the secondary intelligible concepts (al-ma'ānī al-ma'gūla al-thāniya)—those that depend upon (tastanid ilā) the primary intelligible concepts— insofar as they may be of use in arriving at the unknown from the known, and not insofar as they are thoughts (ma'gūla) having an intellectual existence that is not attached to matter at all or attached to non-corporeal matter.

As has been noted more than once, Avicenna's doctrine had a precursor in the Porphyrian distinction between terms in first position and terms in second position …, a distinction which we do find in Arabic writers before and contemporary with Avicenna. A look at some of these writers will show the wider scope of Avicenna's remarks, brief though they are.

Al-Fārābī's Commentary on Aristotle's De interpretatione makes the standard observation: 'name' and 'verb' are terms in second position, whereas, he implies, the categories are terms in first position. The notes (ta'liqāt) that Ibn Bājja (Avempace, d. 1138) wrote on Fārābī's account (?the same as the just-mentioned Commentary or rather a separate paraphrase of Aristotle's De int.) furnish a longer list of terms in second position including 'particle', 'definite' …, 'indefinite' …, 'straight', …, 'oblique' …, 'derivative' as well as 'name' and 'verb'.

Let us look next at the relevant sentences in the notes that a late tenth-century translator and commentator of Aristotle, the Syrian Christian al-Hasan ibn Suwār, has written on the Categories. He states first that Aristotle's aim in the Categories is to discuss those "single utterances in first position (fīal-wad' al-awwal) which signify the highest genera of things (al-umūr) by means of the affections (āthār) [produced] by them in the soul, and [to discuss] things insofar as they are signified by the utterance." In regard to the expression 'first position' al-Hasan explains:

We say utterances in the first position in order to distinguish them from utterances in second position; for utterances in first position are those names and labels (?) that are first applied to things as signs (simāt, 'alāmāt) that signify them in a general way (dalāla mujmala,) such as calling this "silver" and this "copper" and this "gold," while utterances in second position are those that signify what we have set apart as utterances in first position, such as calling ["name"] every utterance signifying something definite, without time, as "Zayd" and "'Amr," and calling "verb" everything that additionally signifies time, as "stood up" and "will stand up." These are utterances in second position because we have posited them subsequent to the existence of the others.

Finally, here are two examples of what a leading logician and teacher of logic in eleventh-century Baghdad, Abū al-Faraj ibn al-Tayyib (d. 1043), had to say about the two expressions in question. In his Commentary on Porphyry's Isagoge, Abū al-Faraj wrote:

… utterances are investigated in two ways, as utterances in first position and as utterances in second position. Utterances in first position are those that signify things (al-umūr), such as "Zayd," "'Amr," and "has struck." Utterances in second position are those that signify utterances in first position.… And [Porphyry's] concern here is with utterances in first position.

The only examples given by Abū al-Faraj of words in second position are 'name' and 'verb', which, as he also observes, are discussed at the beginning of Aristotle's De interpretatione.

What is lacking in all these examples is any statement to the effect that terms (or concepts) in second position constitute the specific subject matter of logic.

Al-Fārābī in his Commentary on Aristotle's De interpretatione does employ the phrase "secondary concepts" (al-ma'qūlāt al-thawānī), the very same expression which we encountered in Avicenna's Metaphysics. I shall here paraphrase Fārābī's text without attempting to do full justice to his rather difficult arguments, my aim being simply to indicate the context in which he introduces that phrase. The occasion is Aristotle's statement at 16^b 19 ff which prompts Fārābī to speculate about the combinative function of "existential verbs" (is, exists). An existential verb, he says, indicates three things: a time, a combination or connection, and an unspecified subject. A question may arise as to how an existential verb, whether used existentially or copulatively, can perform the combination. The problem (as presented by Fārābī) is that a non-existential verb like 'walks' is analyzable into 'is walking', so that "man walks" is equivalent to "man is walking," where 'is' performs the combinative function. Should we then say that "man is (exists)" is also analyzable into "man is existing," where existence would occur twice—once as a connector and again as a predicate?

Fārābī answers that in the case of the existential 'is' ("when 'is' is predicated by itself") no absurdity would result from such a repetition. But no repetition would need to be involved in the case of the copulative 'is' (when the latter is "predicated for the sake of something else"). In this last case 'is' signifies only time, an (unspecified) subject, and "the notion of a copulative existence," the predicate (say, white) being something apart from that.

Fārābī then goes on to pose the unexpected question of whether the copulative 'is' did not itself require a connector which in turn required a connector and so on to infinity—to which question he gives the following enigmatic answer in terms of "secondary concepts":

There would be nothing impossible or absurd in this consequence [of an infinite series of connectors], for the notion of connector is here one of the secondary concepts, and it is neither impossible nor absurd for secondary concepts to go on to infinity, as you have heard me say many times and as I have set down in writing.

He finally adds that repetition of one and the same secondary concept is not necessary, but does no harm if it occurs.

All this is rather baffling. But the character of Fārābī's arguments, the sudden but surprisingly brief appearance of the idea of an infinite chain of connectors regarded as secondary concepts, and the reference to his previous teachings and writings, all this is clear indication that we do not yet possess all we need to have to penetrate the thoughts of early Arabic logicians on the subject that has concerned us. Avicenna's remarks in the Introduction and in the Metaphysics thus remain the clearest and fullest statement on the topic of the subject matter of logic which has come down to us from the period between the translation of Aristotle's logical works into Arabic and the middle of the eleventh century. It is remarkable that they also seem to have become henceforward the standard doctrine to which later Arabic logicians turned for a ready answer to the question of what logic was about. To illustrate this last point I shall quote a passage [from 'Alā' ibn 'Alī al-Tahānawī's Kashshāf istilāhāt al-funūn] that describes the situation as it appeared to an erudit living in the eighteenth century:

The authorities are of the opinion that the subject matter [of logic] comprises the secondary concepts (al-ma'qūlāt al-thāniya), not in respect of what they are in themselves, nor insofar as they exist in the mind (for this [inquiry] is a function of philosophy), but insofar as they lead or can be of use in leading to the unknown. Thus a universal concept in the mind, when compared to the particulars under it, will be considered essential or accidental to them according as it enters into or lies outside their essences, and it will be considered a species if it coincides with those essences.… Now for a universal concept to be essential, accidental or a species or the like, is not something external but something that arises in universal natures when they exist in the mind. It is so with a proposition's being predicative or conditional and with an argument's being a syllogism, an induction or an example.… The logician [also] investigates tertiary and higher-level concepts, for these are essential attributes of secondary concepts. "Proposition," for example, is a secondary concept which may be investigated in regard to its division, conversion, or conclusiveness when combined with other propositions. Thus "conversion," "conclusiveness," "division," "contradiction" are concepts on the third level of thought; and if, in a logical inquiry, something is judged to be one of the divided parts or contradictories, then that thing will belong to the fourth level of thought, and so on.

The gist of all this had already been said by Avicenna; only the idea of a multi-level hierarchy of concepts is lacking in Avicenna's writings. When and in what context did that idea become articulated; and with what consequences, if any? These are questions that must await further search of the enormous bulk of logical writings that relentlessly piled up in the centuries separating Avicenna from the author of the above passage.

Avicenna further develops his views in chapters 3 and 4 of the Introduction, on the utility and the subject matter of logic, respectively. His discussion in chapter 3 begins with the famous distinction between taṣawwur and taṣdīq which we find in almost every Arabic writer on logic after Avicenna. The same terms have been found in the logical writings of al-Fārābī, but their ultimate provenance remains somewhat uncertain. Paul Kraus suggested [in Recherches Philosophiques V, 1935-36] in 1936 that they translated the Stoic terms ραντασία and σヅγκατάθϵσιϛ, and, on the basis of this suggestion, M.-D. Chenu has argued [in "Un Vestige du stoïcisme," "Revue des Sciences philosophiques et théologiques XXVII (1938)] for a Stoic infiltration of medieval Latin thought by way of translating into Latin certain Arabic texts containing the words taṣawwur and taṣdīq.

Literally, taṣawwur is the act of grasping or receiving a form (ṣūra …) in the mind, and taṣdīq is the act of taking or believing something to be true (ṣādiq). Both are described by Arabic logicians as acts of knowledge ('ilm), but truth and falsity are said to have to do only with the second. Often the two words are also applied, respectively, to the form received or the proposition believed to be true. 'Concept' and 'conception' usually do well as translations of 'taṣawwur', ('thought' is also appropriate in some contexts); for rendering "taṣdīq" one has to oscillate between a number of words such as 'assertion', 'belief', 'judgment', 'proposition'. The medieval Latin translation of Avicenna's Introduction has intellectus for taṣawwur and credulitas for taṣdīq.

Whatever the origin of these two terms, Stoic or otherwise, a similar distinction to what they are meant to convey can be easily found in Aristotle, and it appears that Avicenna at least conflated the two distinctions. We read in Aristotle's De interpretatione:

Just as some thoughts in the soul are neither true nor false while some are necessarily one or the other, so also with spoken sounds. For falsity and truth have to do with combination and separation. Thus names and verbs by themselves—for instance 'man' or 'white' when nothing further is added—are like the thoughts that are without combination and separation; for so far they are neither true nor false.

The ninth-century Arabic translation of this passage (by Ishāq ibn Hunayn) does not use the words taṣawwur and taṣdīq. But in the corresponding chapter in Avicenna's Interpretation the plural taṣawwurāt is used interchangeably with āthār, the equivalent in Ishāq's translation of Aristotle's παθήματα and in a passage of the same chapter that parallels the lines just quoted from Aristotle, ma'qūl … and i'tiqād (belief) are made to stand for taṣawwur and taṣdīq: a single thought (ma'qūl), says Avicenna, is neither true nor false; only the belief (i'tiqād) associated with relating one thought to another affirmatively or negatively is true or false. Now there is no term in Aristotle's text that corresponds to Avicenna's i'tiqād (a word which usually rendered the Greek πίστιϛ); only combination and separation of thoughts are said by Aristotle to have truth or falsity. But Avicenna clearly understands Aristotle's remarks in terms of a distinction between acts of conceiving single thoughts and acts of belief applied to the conceived relations between thoughts. How Avicenna himself understood the distinction is made abundantly clear in chapter 3 of the Introduction:

… a thing is knowable in two ways: one of them is for the thing to be merely conceived (yutaṣawwar: intelligatur) so that when the name of the thing is uttered, its meaning (ma'na: intentio) becomes present in the mind without there being truth or falsity, as when someone says 'man', or 'do this!' For when you understand the meaning of what has been said to you, you will have conceived it. The second is for the conception to be [accompanied] with belief (taṣdīq: credulitas), so that if someone says to you, for example, "every whiteness is an accident," you do not only have a conception (taṣawwur) of the meaning of this statement, but [also] believe it (ṣaddagta) to be so. If, however, you doubt whether it is so or not, then you have conceived what is said, for you cannot doubt what you do not conceive or understand, but what you have gained through conception in this [latter] case is that the form of this composition and what it is composed of, such as "whiteness" and "accident," have been produced in the mind. Assertion (taṣdīq), however, occurs when there takes place in the mind a relating (nisba: comparatio) of this form to the things themselves as being in accordance with them; denial (takdhīb: mentiri) is the opposite of that.

It is clear from this text that taṣdīq is not the relation between subject and predicate in a predicative proposition. Such a relation is here called "form of composition" which (as in the case of doubting) can be entertained in the mind without truth or falsity being applied to it; that is, it can be the subject of mere conception (although, of course, unlike the conception of a single thought, it is capable of being described as true or false). Taṣdīq is the attribution of this relation or form to the things themselves.

The role of belief or assertion is again emphasized by Avicenna in his Interpretation, in an account of what a predicative statement is made of. "A predicative proposition (qaḍiyya ḥamliyya), he says, consists of three things, a subject-concept, a predicate-concept, and a relation (nisba) between the two. Concepts (ma'ānī) do not, however, become subjects and predicates by being gathered together in the mind; in addition to this mind must believe (ya'taqid) affirmatively or negatively the relation between the two concepts." He goes on to insist that mere concatenation (tatālī) of terms does not make up a statement. To be a complete expression of a predicative preposition a sentence must therefore contain, in addition to the terms indicating the subject and predicate, a sign that indicates the relation or connection between these. Such a sign is of course the copula (rābita: connector) which, he says, may take the form of a verb (as in Greek or Persian or, sometimes, Arabic), or a noun (the Arabic pronoun huwa,) or a vowel change (modifying the predicate term or both subject and predicate terms). In any case, three linguistic elements are needed to correspond, one to one, with the three essential components of a predicative proposition.

But if assertion is something apart from the relation to which it is applied in predicative propositions, should not such propositions be analyzed into four, rather than three, components, and should not their complete verbal expressions contain four, rather than three, elements? As far as I can see, the question is nowhere broached by Avicenna himself, but it was raised by later Arabic logicians, no doubt led to do so by Avicenna's own remarks. Some argued that since a "relation of judgment" (nisba ḥukmiyya) is found equally in the affirmation and negation of that relation, it must be clearly distinguished from both; four components must therefore be recognized in the make-up of a predicative proposition. Others maintained that the copula would not be able to perform its function as a connector unless it signified both the judgment relation and its affirmation or negation. The copula would thus give expression to both the assertion (taṣdīq) made and the concept (taṣawwur) to which the assertion is applied, and there would thus be no need for a separate assertion sign. This is not the occasion to pursue this discussion in the various writers, but it seems that it was the latter view that finally prevailed. It was also the latter view that very likely expressed Avicenna's own implicit opinion.

I have dwelt at some length on the distinction between taṣawwur and taṣdīq because it became the accepted doctrine of all Arabic logicians. As pointed out earlier, taṣawwur and taṣdīq divided between them the whole sphere of knowledge, the first being attainable by definition, the second by argument. Logic, being concerned with the appropriate means of acquiring knowledge, therefore divided into two parts: a theory of definition (mabḥath al-taṣawwurāt) and a theory of proof (mabḥath al-taṣdīqāt). The following passage succinctly expresses this pervasive doctrine. It comes from Avicenna's Kitāb al-Najāh, a summary account of Kitāb al-Shifā':

Every knowledge is either conception (taṣawwur) or belief (taṣdiq). Conception is the prior knowledge (al-'ilm al-awwal) and it is acquired by means of definition or the like.… Belief is acquired only by means of syllogism or the like.… Thus definition and syllogism are two instruments by means of which knowledge of unknown things is acquired through discursive thought (al-rawiyya).… Now every syllogism and every definition is made up by bringing intelligible notions (ma'ānī ma'qūla) into a definite composition so that each would have a matter from which it is composed and by means of which the composition is effected. And just as a house or seat cannot properly be made from any chance matter or in any chance form, but rather every thing has its own matter and form which are proper to it, so also there belong to every object of knowledge (ma'lūm) which is knowable by means of discursive thought a proper matter and form by means of which that object may be grasped (taḥaqquq). And just as a house may be improperly built because of deficient matter or form or both, so also discursive thought may be vitiated (fasād) on account of its matter even if the form is valid (ṣāliḥa), or on account of the form even if the matter is appropriate (ṣāliḥa), or on account of both.

Fārābī had said in his Enumeration of the Sciences that the objects with which the rules of logic are concerned are "the thoughts (ma'qūlāt) in so far as they are indicated by the utterances, and the utterances in so far as they indicate the thoughts." We establish an opinion in ourselves by setting up in our minds those thoughts which are apt to verify it. This process is called by the ancients "inner speech" (al-nuṭq al-dākhil). To impart the truth of an opinion to someone else we employ the forms of speech (aqāwīl) suitable for achieving that purpose. This is called "outer speech" (nuṭq khārij bi-al-ṣawṭ). It is the function of logic to provide the rules (qawānīn) that guide us toward the proper conduct (called "qiyās," reasoning) of both kinds of speech.

Avicenna seems to have had some such statements in mind when he wrote in chapter 3 of the Introduction (= chapter 4 in the Latin edition) that "there is no value in the doctrine of those who say that the subject of logic is to investigate utterances in so far as they indicate notions (al-ma'ānī.)" It is noticeable that in the two previous chapters the question of language and its relation to logic is nowhere brought into the discussion. Now that Avicenna has put forward in those chapters his own view of the subject and use of logic, he feels he can settle that question without much belaboring of words. Unfortunately his remarks are much too brief. The logician, he says, would have been able to dispense with utterances only if it were possible to learn logic by means of "pure thought." But we are forced to use utterances,

… especially as it is not possible (muta'adhdhir: non potest) for the reasoning faculty (al-rawiyya: ratio) to arrange notions (al-ma'ānī: intellecta) without imagining the utterances corresponding to them, reasoning being rather a dialogue with oneself by means of imagined utterances. It follows that utterances have various modes (aḥwāl) on account of which the modes of the notions corresponding to them in the soul vary so as to acquire qualifications (aḥkām) which would not have existed without the utterances [seguitur ut verba habeant diversas dispositiones per guas differant dispositiones intentionum gue concomitantur esse in anima, ita quod fiant eis indicia gue non haberentur nisi per verba]. It is for this reason that the art of logic must be concerned in part with investigating the modes of utterances.… But there is no value in the doctrine of those who say that the subject of logic is to investigate utterances in so far as they indicate notions … but rather the matter should be understood in the way we described.

The modes mentioned here are of course those secondary properties which concepts acquire when they constitute definitions and arguments. They are thoughts (ma'qūlāt) of a second order, twice removed from the things of the material world to which outward speech belongs. Avicenna now tells us that reasoning is impossible without utterances, whether spoken or imagined. By itself this is not a new thing to say. But the consequence he draws from this statement is not the simple language-thought parallelism noted by the writers whose views he found inadequate. He clearly says that the conceptual modifications are brought about by modifications in the utterances. This means that the secondary concepts, the proper object of logic, not only are reflected in language but are generated by it. Is this so because logical concepts arise only in the context of a process, reasoning, which is dependent on language in a peculiar way? In any case, however one interprets his words, and I am not sure I quite understand them, he seems to be making a stronger claim for the role of utterance in logic than I have encountered in any writer before and up to his time. Avicenna is not just saying that utterances are important in the study of logic. Having already pointed out that logic was not a (psychological) inquiry concerned with mental entities as such, he is now telling us that it is an inquiry primarily concerned with language. The reason is, or seems to be, that the properties constituting the subject matter of logic would be inconceivable without the exercise of a particular function of language.

Avicenna returns to the question of the relation of logic to philosophy and the related question of the subject matter of logic in other parts of the Logic in Kitāb al-Shifā'. In the section on Categories, for example, he again asserts his independence from the Peripatetics (including Fārābī and Ibn al-Tayyib) by emphatically excluding the doctrine of categories from the proper domain of logic. This agrees with his understanding of logic as concerned with second-order concepts. And in the section of Syllogism he devotes a chapter to showing how the function of logic as an instrument is to be understood. His interesting views in these and in other places are, however, too detailed and too complex to be dealt with adequately here.