BMCR 1998.08.09

Aristotle’s De Interpretatione: Contradiction and Dialectic


Even if we were to judge such things only on the basis of the ratio of text pages to commentary pages, Aristotle’s De Interpretatione would count as one of the most important texts, not just in the Western canon, but in any canon. Weighing in at less than nine pages in Bekker’s Berlin Academy edition (23 pages in the οξτ it has attracted the attention of scholars from the Hellenistic period to our own. Sitting on my shelves are books containing over a thousand pages of commentary on the treatise (including two books on the ninth chapter alone); any well-stocked university library will contain much more. But of course we don’t judge such things on the basis of page-counts: we judge them in terms of the overall contribution of the work to the intellectual and cultural development of the society within which the canon is defined. On this score, too, the treatise fares quite well. It is of interest to logicians, semioticians, psychologists, and linguists as well as to students of rhetoric and of the history of philosophy; it has long been admired by scholars in all of these fields.

In light of this it comes as something of a surprise to find that the treatise has not fared so well among English writers. Ackrill produced an excellent short commentary on the work, along with the Categories, for the Clarendon Aristotle series, and Arens contributes one original chapter in his excellent collection of ancient and medieval commentaries on the treatise, but there have been no other book-length treatments of the whole work in English until the present volume. This revision of Whitaker’s 1993 Cambridge University doctoral dissertation shows some signs of its origins (a two-page chapter on the authenticity of the title of the De Interpretatione, for example, material that ought to have been worked into a footnote or appendix; and a line-by-line analysis of the text that sometimes seems pedantic), but it is, nevertheless, a remarkably clear and persuasively argued explication of the argument of the De Interpretatione.

The book is a chapter-by-chapter commentary on the De Interpretatione, but it presents a unified interpretation. Whitaker’s argument departs in two interesting and important ways from mainstream interpretations of the treatise. Many commentators have viewed the work as part of a suite of treatises including the Categories, Topics, Sophistici Elenchi, and Analytics. On this view, the Categories provides groundwork in the nature of subject and predicate terms, the fundamental units of meaning; the De Interpretatione is supposed to take the account of terms from the Categories and show how they can be combined (literally “interwoven” [ συμπλοκή, a word that seems deliberately to echo Plato’s use of the same imagery at Sophist 262c2-7; cf. Cat. 1a16ff.]) to form assertions, the fundamental units of truth-value; the Topics and Analytics then show how assertions can be combined into demonstrations, the fundamental units of validity, analyzable either informally ( Topics) or formally ( Prior Analytics). Thus the Organon, as traditionally understood, is a package deal providing training in logical theory and practice. Moreover, the traditional view regards the De Interpretatione as something of a patchwork, embodying either several smaller treatises or quite distinct discussions of a variety of topics: chapters 1-4 give Aristotle’s thinking on linguistics, semiotics, and signification; chapter 9, the famous Sea Battle chapter, is taken to be a study of fatalism and its consequences for deliberation; chapters 12 and 13 appear to be about modal logic; and chapter 14 is a foray into epistemological psychology. Consequently a fair amount of work has to be done to show how these apparently disparate topics are relevant to each other and to the overall project of the Organon. Whitaker rejects both aspects of the traditional interpretation: he argues that there is no compelling reason to read the De Interpretatione as a component of the package described above and that the treatise is not a patchwork but a coherent unity. Rather than being seen as sandwiched between the Categories and the Analytics, the De Interpretatione is best viewed as a propaedeutic to the dialectical exercises described in the Topics and the Sophistici Elenchi. The traditional account, according to Whitaker, sees the chapters of the De Interpretatione as disconnected because it has missed the principal connection that holds them together: Aristotle’s account of the various relations possible between contradictory pairs of assertions. Whitaker reads the first five chapters as establishing a framework within which a fundamental relation that he calls the Rule of Contradictory Pairs (RCP) is to be understood. This framework is an account of signification, names and verbs, phrases and assertions, and the compounding of assertions. Once these fundamental components of Aristotle’s theory have been laid out, RCP is defined: “Of every contradictory pair, one member is true and the other false” (79). Chapters 7 through 9 of the De Interpretatione consider three exceptions to RCP: singular vs. universal assertions, complex assertions, and future singular assertions. The remainder of the treatise, chapters 10 through 14, integrates RCP and its related principles into the broader fabric of Aristotle’s dialectical method: since dialectical procedures at the Academy proceeded by question-and-answer debates within which it was required that one force one’s opponent into a contradiction of his own initial position by means of disjunctively exhaustive questions, it was essential that the verbal combatants be able to recognize what sorts of assertions are genuinely contradictory pairs and under what sorts of conditions these pairs can legitimately be used to force an answer of either yes or no.

The thematic nature of Whitaker’s interpretation has the great advantage of showing how the De Interpretatione can be read as a unity, but as often happens with such programmatic approaches to Aristotle’s work it sometimes leaves one wishing for more. A case in point, worth discussing in detail, is Whitaker’s treatment of the famous ninth chapter, where (on Whitaker’s view) Aristotle considers the problems posed for RCP by future contingent assertions (FCA). The traditional interpretation of this chapter has two aspects, one metaphysical and one logical. Both aspects, according to the traditional view, have to do with the question of fatalism. Suppose I assert “There will be a sea battle tomorrow.” Some fatalists had suggested that regardless of whether we regard my assertion as true or as false, the future is fixed, because if it is true today to make such an assertion then the sea battle will occur regardless of what we decide to do, and if it is false today to make such an assertion then no sea battle will occur regardless of what we decide to do. According to one variant of this traditional interpretation, Aristotle’s work in chapter 9 is in response to an argument similar to one reported by Arrian in his summaries of the discourses of Epictetus (2.19.1) and sometimes attributed to Diodorus of Cronus. Although there is some evidence that Diodorus is actually a younger contemporary of Aristotle, it is not inconceivable that the argument as he is reported as presenting it was already current in discussions of fatalism. This argument, known (from Epictetus) as the “Master Argument” ( ὁ κυριεύων λόγος), tries to import necessity into assertions about the future on the basis of an alleged necessity to be found in assertions about the past. The argument as it has been preserved is rather muddled, and better versions of it have been reconstructed (notably by Arthur Prior, Past, Present and Future [Oxford, 1967]), but the gist of it is this. Suppose there was a sea battle yesterday. If I assert, today, “There was a sea battle on 19 August 1997”, then my assertion is not only true, it is necessarily true in the sense that it cannot possibly be false. If it is necessarily true, according to the Master Argument, then it has always been true. This means that it is not only true to say it today, but it would have been true to say it yesterday, or four days ago, or four years ago, or ten thousand years ago. But if it was true to say it ten thousand years ago, then the sea battle was a fixed part of the future for that distant time, and human deliberation and action could have no power to alter the course of the future. Generalize this argument and you have fatalism.

For Aristotle it is precisely this deliberation and human action that makes the future indeterminate rather than fixed, and the traditional interpretation takes him to be responding to the fatalist challenge in our chapter. Metaphysically, according to Aristotle’s well-known view about the criteria of truth (“To say of what is that it is not, or of what is not that it is, is false, while to say of what is that it is, and of what is not that it is not, is true”; cf. Metaphysics IV 7, 1011b25-28), an assertion is made true or false by corresponding or not corresponding with some actually obtaining state of affairs in the world. Let us return to my first assertion, the one about the future sea battle (“There will be a sea battle tomorrow”). Since my assertion refers to a state of affairs that does not yet exist, it appears as though we cannot say that it is true according to the correspondence theory, but if we say that it is false, aren’t we saying that there is some other state of affairs — the one in which there is no sea battle tomorrow — that makes the assertion false? Perhaps it is neither true nor false: perhaps we need to wait until tomorrow to determine its truth value. But if we say that the assertion has no truth value we are rejecting the Principle of Bivalence, a principle that Aristotle is known to have accepted (indeed, he is often regarded as the inventor of the principle), even for future assertions. The Principle of Bivalence (PB) is the semantic thesis that, necessarily, every meaningful, assertoric statement is either true or false; it is to be distinguished from the Law of Excluded Middle (λεμ the syntactic thesis that “either p or not- p” holds for any substitution of an assertoric statement for p. The traditional interpretation has often maintained that Aristotle is more committed to the ontological thesis about the criteria of truth than he is to the semantic Principle of Bivalence, so in the ninth chapter of De Interpretatione he rejects PB for FCAs in order to avoid the fatalist conclusion (the most recent, and in my view most exhaustive, treatment of the traditional view is Richard Gaskin, The Sea Battle and the Master Argument: Aristotle and Diodorus Cronus on the Metaphysics of the Future [Leiden, 1995]).

According to Whitaker, Aristotle does not reject PB for FCAs, nor does he commit himself to fatalism; indeed, on Whitaker’s view the chapter is not about PB or the metaphysics of the future at all. Rather the chapter is about RCP and the fact that, precisely because the future is not yet determined, we cannot know which of the pair of contradictory assertions is true and which false, though necessarily one of them is true and the other false. This is not an epistemic claim, however: “we should note that Aristotle does not mean that it is merely a limit on our knowledge that means that we cannot pick out the true member of a future contingent contradictory pair. It is not just that we have no means of knowing which is which. Rather, it is genuinely still open. Otherwise, fatalism would not have been refuted, and we would merely have the illusion of being able to deliberate and make decisions” (pp. 124-5). Whitaker’s illustrative example is helpful for understanding what he intends here: when there are two candidates in an election, we can claim with certainty that one or the other of them will be the officeholder, but until the election is held we do not know which it will be, because that has not yet been determined.

It is not entirely clear that this emphasis on RCP will always solve every interpretive puzzle in the text, however, or that it treats sufficiently the question of PB and the metaphysics of the future. For example, at 18b17-25 Aristotle considers, as a possible answer to the fatalist, the possibility that both members of a FCA pair might be false. According to Whitaker, “the move of denying that there are true future singular assertions, far from destroying the fixedness of the future and releasing future events to go either way, would actually destroy the reality of the future altogether” (p. 118). But given Aristotle’s emphasis on meeting the fatalist’s challenge, the point of this passage seems to be not that the reality of the future would be destroyed by such a response, but rather that if both members of the pair are false then the fatalist’s argument will still go through. For example, suppose we use the letter p to stand for the assertion “There will be a sea-battle tomorrow.” Then the denial of p may be written ~p (“It is not the case that there will be a sea-battle tomorrow”). Suppose both p and ~p are false. According to PB, which Aristotle tacitly assumes in this passage (as Whitaker rightly notes), every assertion must be either true or false, so if p is false then the denial of p ( ~p) is true, and if the denial of p ( ~p) is false then p is true. The fatalist can claim that if both members of a FCA pair are false then fatalism has not been blocked, since it is possible to infer (by PB) the truth of either member from the falsity of the other, and the future is still determined.

Part of the puzzle here has to do with what makes assertions true or false. Aristotle is known for his realism about the correspondence theory of truth: an assertion is true just in case it corresponds with that aspect of reality that it picks out. It is true to say “the cat is on the mat” just in case the cat actually is on the mat. So what can possibly make an assertion about the future either true or false, given that the future does not exist? There is nothing for such assertions to refer to, no ontological foundation for their truth or falsity. This has led some commentators (from antiquity to the present) to suggest that Aristotle is rejecting PB for FCA pairs. Although Whitaker makes a good case for the relevance of RCP in this chapter, he does not explain sufficiently why PB is not at issue. What does it mean to say that it is “not fixed” which member of a FCA pair is true and which false, while it is nevertheless still the case that one member of the pair is true and the other false? If Whitaker is right, then FCA pairs present an exception to RCP precisely because “one member of the contradictory pair is true, while the other is false, but it is not yet settled which is which” (124). But on the correspondence theory of truth, it is the ontological state of affairs that makes assertions either true or false (indeed, that is precisely why PB normally holds: things cannot be in orthogonal states), so how can it possibly be “not yet settled” which assertion is true and which false unless there is no ontological basis for making that determination? On the other hand, if there is no ontological basis for that determination then how, on the correspondence theory, can it be the case that one of the assertions is true and the other is false (and Whitaker has already noted that this lack of fixity is not an epistemic limitation)? Whitaker rejects the view of some of the ancient commentators that by οὐ μέντοι ἤδη ἀληθῆ ἢ ψευδῆ at 19a39 Aristotle means that in the present, when the claim about the future is made, there is “not yet” a truth value for the assertion (this would amount to a rejection of PB for FCA pairs), yet it seems difficult to avoid the conclusion of these commentators that this passage has something to do with PB.

The ninth chapter of De Interpretatione is, of course, extremely controversial, and I do not mean to suggest that Whitaker has overlooked an obvious solution to the difficulties that it poses; I use the chapter merely as an example of how a programmatic interpretation such as Whitaker’s will inevitably leave certain puzzles unsolved. That said, the careful reader will still gain much of value from Whitaker’s treatment, and I expect that it will be a standard work on the treatise for some time. I recommend it very highly not only for its clear, succinct, and scholarly presentation, but for the challenges it will present to anyone interested in these issues.