THE STOIC NOTION OF A GRAMMATICAL CASE *

Transcription

1 THE STOIC NOTION OF A GRAMMATICAL CASE * MICHAEL FREDE There is a trivial, but also very vague sense in which all Greeks must always have known that the Greek language distinguishes various cases of the declinable parts of speech, according to traditional grammars at least five such cases, the nominative, the genitive, the dative, the accusative, and the vocative. It is also clear that at least from a certain point onwards, when they began to reflect on their language, the Greeks were aware of the declension of declinable words according to the cases. Thus we find in a fragment of Anacreon (fr. 3) the lines: KkopoVAov p2v Zpy t?p.h, KkopoVAy) 6 Lmpaivopai, KkOpovAov 26iooidu. Kleohoulou men etq6g ere6, Kleohoul6i d epimainomai, Kleohoulon de dioske6 But to be aware of the declension of names and nouns is not yet to have a theoretical notion of the grammatical cases, let alone a doctrine of them. Historians of grammar or linguistics seem to agree that it was the Stoics who introduced the term in the sense of a grammatical case. The first explicit reference to a doctrine of the grammatical cases certainly is constituted by a title of a lost work by Chrysippus On the five cases (D.L. VII 192). And a look at Diocles s account of Stoic logic in Diogenes Laertius shows that this notion of a case played a crucial role in Stoic philosophy of language or linguistics and in Stoic logic (D.L. VII 64-65; 70). Unfortunately, it is quite unclear how the Stoics do conceive of the grammatical cases. This is particularly unfortunate, since already the ancients saw that the traditional grammatical doctrine of the cases has its ultimate origin in Stoicism (Ammonius, In De int. 43,4-5; Leo Magentinus, Scholia in Aiistotelem, Brandis 10411). Given that we are so used to the notion of a case from school grammar, we rarely reflect on, and tend to take for granted, what a case is supposed to be. We certainly assume that we know what the ancients took a case to be: it is a word-form, the particular inflected, more specifically declined, form which a declinable word takes. Like the ancients we say that nominis is a genitive, or that it is the word nomen in the genitive. All this seems so obvious that we are inclined to take it for granted that whatever concept of the grammatical cases the Stoics might have had, they, too, certainly must have thought of the cases as the declined forms of declinable words. Thus, e.g., Pohlenz (Die Stoa I 44) just assumes, as if it were obvious, that the Stoic doctrine of the cases is a doctrine of the declension of nouns. And most recently Long and Sedley (The Hellenistic. *An earlier version of this paper was delivered in February 1993 as a lecture at the Institute for Classical Studies. I would like to thank those who participated in the discussion, in particular David Sedley whom I had hoped to be able to persuade, but also Richard Sorabji, Katerina Ierodiakonou, and Fritz Zimmermann who provided me with written comments. BlCS

2 14 BICS Philosophers 1201) affirm again: In its standard Stoic usage the word case designates not the inflection itself, but the inflected word. A nominative case is a word inflected in the nominative. And in volume I1 203, they claim that cases on any analysis are functions of language. But the matter is not as clear as it tends to be assumed. Certainly not all ancient grammarians agreed that cases are inflections of a certain kind or inflected word-forms. Thus a scholiast on Dionysius Thrax (p. 230, 34-36) explains that the cases belong to the realm of what is signified by expressions, rather than to the realm of expressions themselves. Otherwise, he says, the name Atreides (for which he lists at least four different forms just for the genitive) would have a lot more than five cases. There is no reason to assume that the scholiast s argument is Stoic, but there is abundant evidence that the view he argues for is Stoic. For when we look at the passages in Diogenes Laertius in which Diocles reports Stoic views which involve the cases (VII 64-65; 70), it becomes clear that the Stoics, who carefully distinguish between expressions and what is signified by expressions, invariably treat cases as constituents of what is signified, rather than as constituents of the verbal expression used to signify something. Thus they distinguish between verbs, that is expressions of a certain kind, and predicates, that is what is signified by these expressions (D.L. VII 58). Correspondingly, they seem to distinguish between complex verb-phrases and the corresponding predicates signified by them, e.g. between is sailing through rocks and sailing through rocks, between is hitting the ball and hitting the ball (D.L. VII 64). The predicate corresponding to the verbal form is hitting they call a praedicatum rectum, since it requires completion by an oblique case to generate the (complete) predicate hitting the ball. Here the oblique case clearly is not the expression the ball in the verbal phrase is hitting the ball, but rather a constituent of the predicate, which is distinguished from the verbal expression as its signification. To put the matter in Stoic terminology, cases are not constituents of expressions, e.g. sentences, which on the Stoic view are physical, corporeal items, but rather constituents of the lekta, e.g. propositions, which on the Stoic view are immaterial items. Thus it seems clear that for the Stoics cases cannot be the forms of declinable words or their inflections. In fact, it is not even clear that Stoic cases are, as Long and Sedley assume they ought to be on anybody s view, functions of language. There is an admittedly dark and corrupt passage in Simplicus s commentary on the Categories (p. 209, 10) which compares Stoic and Platonist terminology. It distinguishes between dispositions, concepts, cases, and predicates, as the Stoics are supposed to call them respectively. Now here we have cases as part of an entirely metaphysical distinction of different kinds of items in our ontology. It certainly is not a distinction of linguistic items (or functions of language). And the fact that cases are listed next to predicates should make us hesitant to rule out ah limine that these are the very cases the Stoics talk about in grammar. After all, it is almost invariably in conjunction with predicates that cases are referred to in Diogenes Laertius, VII 64, our main source for the Stoic doctrine of cases. So what are we supposed to think? On the one hand it seems that Stoic cases should be word-forms, on the other hand much of the evidence suggests that they cannot be wordforms. Unfortunately, all attempts so far to clear up this difficulty have met with so little success that one is inclined to give up in despair and hope that one day perhaps a new papyrus will provide us with the evidence to settle this thorny question. But perhaps we should not give up and look at the evidence again, forgetting, as well as we can, what we think we know about cases.

3 MICHAEL FREDE: THE STOIC NOTION OF A GRAMMATICAL CASE 15 If we are to make headway, obviously the first question we have to settle is whether cases, indeed, are linguistic items, or rather items somehow signified by declinable words. What reasons do we have to think that cases are word-forms? There is first of all, and most importantly, the fact that this is how the ancients generally use the word case in grammatical or linguistic contexts. But it should be kept in mind that this is not how they use the term invariably. As already noted, the scholiast on Dionysius Thrax (230, 34-36) rejects the view that cases are word-forms. And we should also keep in mind the following. Stoic grammar is shaped by the idea that differences and distinctions on the level of expressions reflect, and are indicative of, differences and distinctions on the level of what is signified. There are verbs, because there are predicates. And there are verbs in the plural and in the singular, because there are singular and plural predicates. There are past-, present-, and future- tensed verbs, because there are past-, present-, and future- tensed predicates or propositions. We distinguish transitive and intransitive verbs, because there are transitive and intransitive predicates; and similarly with other distinctions (D.L. VII 64). A good deal of our grammatical concepts and terms were first introduced by the Stoics to analyse and to describe what is signified by expressions, and only then were the terms transferred to describe the analogous features of expressions. But grammarians as a rule, and non-stoic philosophers invariably, rejected the Stoic assumption of, and theory about, Iekfu, the items supposedly signified by expressions like sentences. Nevertheless they continued to use the Stoic terms to describe the corresponding features at the level of expressions. So it would not at all be surprising if the grammarians equally continued to talk about cases and the different kinds of cases, like the genitive or the accusative, though these notions originally had been introduced by the Stoics to analyse what is signitied by expressions. Hence the fact that non-stoics use the term case to refer to expressions, though weighty. should not impress us too much. What we should rather look for is direct evidence as to how the Stoics themselves actually used the term. There are a number of passages which seem to imply, and have been taken to imply, that the Stoics use the term case for expressions. To begin with, there is, e.g., Plutarch s explicit testimony (Q.P. 1009C). Plutarch tells us that a statement, or what the Stoics call an ux-i6rna, consists of a name or noun and a verb, and that the Stoic dialecticians call the former a case and the latter a predicate. So here we do have the explicit claim that the Stoics use the term case to refer to words. But closer reflection on this testimony suggests, not only that it does not constitute evidence for what it claims, but that it rather looks like evidence for the contrary. To begin with, as opposed to what Plutarch says. the Stoics meticulously distinguish between a sentence and the corresponding proposition or uxirsrnu ; it is the sentence but not the asiijrna, which - in the most simple case - consists of a noun and a verb. What is more, we have already seen that the Stoics, far from calling verbs predicates, carefully distinguish between verbs and the predicates signified by them. And so we should not put any trust in the claim, either, that the Stoics call nouns or names cases. Rather the contrary, we should suspect that, just as the Stoics distinguished statements or sentences and propositions, and verbs and predicates, they also distinguished cases and nouns. What we definitely do learn from this Plutarch passage, though, is that non-stoics did not bother to be careful about the Stoic distinction between expressions and lekfu, and that, as a result, there is a certain amount of confusion in our testimonies. Further, there is a testimony in Stobaeus which seems to, and has been alleged to, show that Stoic cases are words. Stobaeus (Ed. I 136, 21ff.), in a passage with close parallels to the Simplicius text mentioned above, tells us which position the Stoics take on

4 16 BICS Platonic ideas. They distinguish between ideas which they take to be mere concepts or mental constructs and cases. And we are told that we participate in ideas or concepts, but obtain or bear (tunkhanomen) the cases, which they call appellations (prospgorias; 137, 5-6). Now appellation or appellative is the Stoics and the grammarians term for nouns. So, if we accept Stobaeus s report at face-value, the Stoics distinguish between ideas, or concepts, and nouns, namely cases. But it is difficult not to suspect, in particular in light of the Simplicius passage, that the Stoics in fact distinguished between ideas or concepts and some other non-linguistic kind of items, mainly cases, which somehow fall under these ideas or concepts and hence are called cases. It surely is no accident that Stobaeus a few lines earlier (136, 24f.) had told us that according to the Stoics the ideas are ideas of the things which fall under the concepts. But whatever these things may be, it certainly is not the appellations in the sense of the designations or names of things which fall under their concepts, and the ideas certainly are not ideas of names of nouns. Hence, to say the very least, Stobaeus s testimony can hardly be used to serve as straightforward evidence to the effect that Stoic cases are nouns. Perhaps Stobaeus is suffering from the same confusion, or at least the same lack of care, as Plutarch. But there is yet another possibility. A passage in Clement (Strom. VIII 9, 26, 4) shows that we cannot rely on the assumption that the term appellation refers to expressions, rather than to what is signified by them. Clement reports that the Stoics and the Peripatetics disagreed as to what causes are causes of. The Stoics claimed that causes are causes of predicates; thus the surgical knife is the cause for the flesh of being cut. We may note here in passing that this Stoic doctrine presupposes a careful distinction between verbs and predicates. For obviously the Stoics do not mean to claim that causes are causes of verbal phrases like... is being cut, but rather of something s actually being cut, something s happening to something. The Peripatetics, on the other hand, are supposed to claim that a cause causes an appellation, e.g. a cut (tome). Now, clearly, the Peripatetics do not want to claim that a cause causes a noun rather than a verb. So appellation here can hardly refer to nouns. It rather seems that it refers to what is signified by nouns. The cause does not cause the noun cut, but a cut. So the Peripatetics point seems to be that a cause is not, as the Stoics claim, the cause of a predicate, i.e. of something signified by a verbal phrase, but rather of what is signified by a noun. Hence appellation here is not used for nouns, but rather for something signified by them. That this use does not reflect a misunderstanding or idiosyncrasy on Clement s part, is shown by the fact that we find the very same use in the same context of a discussion of causality in Sextus Empiricus (P.H ). Thus there also is the possibility that this is how Stobaeus in the passage referred to above uses the term appellation. Hence, it is not clear at all that Stobaeus actually does use appellation in the sense of noun. We also should note that Stobaeus does not explicitly say that it is the Stoics who call cases appellations. He might as well be taken to explain the Stoic use of cases by saying that they are what people also call appellations. In short, this text in Stobaeus raises so many difficulties that it hardly can serve as safe evidence concerning the Stoic use of the term case. Another very important text which might be taken to show that Stoic cases are words is Simplicius In Cat. 105, 10. At issue here is Chrysippus s rejection of ideas and universals as things which exist. And Simplicius explains that, to understand the Stoic position, we not only have to take into account their doctrine of the general or the generic like the kind or genus man (the genikon poion), but also how, according to them, the cases are uttered. Now the word used for uttered here is propheresthai. But we are told in D.L. VII 57

5 MICHAEL FREDE: THE STOIC NOTION OF A GRAMMATICAL CASE 17 that the Stoics make the following distinction: it is things which are said, but sounds or expressions which are So, one may think, cases, being uttered, must be expressions. But, of course, we do not know whether Simplicius here uses the term utter in the strict Stoic sense, nor do we know whether all Stoics all the time kept to this narrow use of the term utter, nor is there any reason to take it for granted that Simplicius here, in reporting on the Stoics, is using their language. Hence this passage does not show, either, that Stoic cases are words. Further, there is a passage in Galen s De sophismarihirs (p. 106, Edlow). anthriipos in anrhriipos esti is said to be ambiguous. It might either refer to the ousia or to the case. Now one might argue that the ambiguity Galen has in mind here is the ambiguity between a man and the word man. If this were so, then case here would refer to a word. But, of course, again we cannot be certain whether Galen here is using Stoic language. And, anyway, the ambiguity might also be the ambiguity between a man exists and man (i.e. the kind) exists. So this passage, too, throws little light on our question. It would be tedious to take the reader through further, even more problematic passages, just to conclude in each case that they hardly constitute evidence for the view that Stoic cases are word-forms. Indeed, it seems to me to be fair to say that there is not a single passage which constitutes hard uncontroversial evidence in favour of this view. There is, on the other hand, a good deal of evidence which suggests that cases are non-linguistic items. To be more precise, there are lots of passages, some of which we have already considered, and some of which we will consider later, in which some notion of case is used according to which cases clearly are not words. Consider, e.g., S.E. A.M. xi, Sextus points out that both the Stoics and the followers of Plato and Xenocrates claim that good is used, or meant, in different ways, but that there is a crucial difference between their respective positions. The Platonists take good to apply to both the idea of the good and the things which participate in the idea, but this in such a way that the different things which are signified by good have nothing in common. It is, Sextus explains, like the word dog. The word dog (in Greek) on one occasion signifies the kind of thing under which the animal which barks falls, on another occasion the kind of thing under which a certain type of philosopher falls, etc. And these different significations of the word dog Sextus calls cases. Here clearly a case is not a word, but rather something signified by a word (cf. also Clement, Strom. VIII 4, 12, 5-13, 1). Similarly, there is the notorious passage in Clement (Sirom. VIII 9, 26, 5) in which Clement tries to solve a sophism by introducing the notion of a case, which he clearly distinguishes both from a word and from what the word stands for, i.e. the concrete object. The puzzle is this what you speak of comes through your mouth; you speak of a house; hence a house comes through your mouth. We are supposed to understand that it is not the actual house (nor the sound house which also is something corporeal), but an incorporeal case which is uttered. Here again the case seems to be something like the meaning of a word, and certainly is not itself a word. But all these passages do not help much to settle the issue, either. For though they without a doubt show that case can be used, not for words, but for something signified by them, we do not at this point know whether the notion of a case invoked in them is a Stoic notion, let alone whether it is the Stoic notion of a grammatical case. So we are, after all, left with the passages in Diogenes Laertius s account of Stoic logic as the only texts which indubitably and uncontroversially refer to grammatical cases as the Stoics conceive of them.

6 18 BICS Now in these uncontroversial passages, cases, as we have been claiming from the outset, are elements or constituents of what is signified by expressions, of lekta, rather than of the expressions themselves, that is to say they are elements of non-linguistic items. We should keep in mind, as noted above, that according to the Stoics a cause is a cause of a predicate or a lekton, e.g. of Socrates s falling ill or of Socrates s being ill. And clearly words are not constituents of this sort of item. Nor is the word cold part of catching a cold. But the Stoics do claim that catching a cold is a certain kind of predicate constituted in part by an accusative case (cf. D.L. VII 64). This accusative case cannot possibly be a word. I infer that, to go by these passages, Stoic cases cannot possibly be word-forms. And given that outside these passages there is no uncontroversial evidence to the contrary, I conclude that we should proceed on the assumption that Stoic cases are not linguistic items, however counter-intuitive this at first might seem. But if cases are not declined words, what are they? Instead of speculating about this, we should look at our ancient sources which tell us at least something as to what, according to the Stoics, grammatical cases are supposed to be. What they have to say does not go very far, but it does provide us with a safe and secure basis and starting point, which turns out to remarkably fit, and throw light on, much of our remaining evidence. Several ancient grammarians tell us that the Stoics called the cases, including the nominative, cases, because they fall from the general (genikon), or common (koinon). into the specific (eidikon) or even the individual (idiotts) (cf. Schol. in D.Th. 230, 25f; 231, 22-23; 383, 5; 550, 26-27). In the same vein several Aristotelian commentators explain that for the Stoics cases, including the nominative, are cases because they fall from a thought or a concept (notma) (of. Ammonius In De int. 43, 9-10; 11; 14; Stephanus In De int. 10, 28; Anonymus In De int. 2, 14-15; Leo Magentinus In De int., Scholia in Aristorelern, Brandis 104n). This, as it stands, is very vague, but it immediately reminds us of the contrast the Stoics, according to Stobaeus (Ed. I 137, 5-6), draw between ideas or concepts and cases, and of the distinction Simplicius (In Cat. 105, 10) attributes to the Stoics between the idea, the general or generic, the kind, on the one hand, and the case, on the other. It also reminds us of the distinction between the idea or general concept and the cases Simplicius attributes to the Stoics and to some Platonists (In Car. 209, 1Off). At first it might seem as if cases were just particular objects, or - more generally - particular instances, which fall under some general concept and hence are called cases. Thus Socrates, we might think, is a case of man, falls under the kind, the general concept, man. And this might seem to fit the doctrine of cases very well. Take the Greek sentence anthrcipos tuprei sphairan. It is not man in general, the kind or, as the Stoics say, the genus man, which hits the ball, but a particular man, something which falls under the genus or concept, and it is not a ball in general which gets hit, but a particular ball. Yet particular objects do not exist in isolation by themselves, they enter into facts or truths about them. But they are constituents of truths in quite different ways. The ball is part of the truth that a man hits it in a quite different way from the man who is hitting it. The ball enters the fact, as it were, on the receiving side: something is done to it, something affects it in such a way as to cause an effect upon it. This is why it is called an aitiatik? prosis, an accusative. or better causative, case. The man, on the other hand, enters the fact as the agent, as the grammatical subject of the active transitive predicate hitting, and thus is called a cusus rectus. And there are yet other ways in which a particular object might enter a lekton, e.g. by being invoked or addressed, in which case it is called a vocative.

7 MICHAEL FREDE: THE STOIC NOTION OF A GRAMMATICAL CASE 19 Unfortunately things are more complicated. For it is clear, both from the Stobaeus passage on ideas and from the passage in Simplicius which compares Stoic and Platonist terminology, that the Stoics distinguish cases not only from the general kind, the idea, or the concept, under which they fall, but also from the particular objects, which are members of the kind. The case in both of these passages is something the particular object is said to obtain or to bear (tiuzkhanein), rather than the particular object itself. This fits remarkably well the fact that Clement in his discussion of the sophism about the house which comes through your mouth not only distinguishes between the actual house, the word house, and the case, but also characterizes the case as something which the house bears or obtains (Strom. VIII 9, 26, 5). And it fits the fact that the Stoics call the particular objects which make a proposition true, e.g. the Socrates who makes the proposition that Socrates is the teacher of Plato true, the tunkhanonta (S.E., A.M. VIII 11; 12; 75; Plut. Adv. Colot F; Philop. In An. pi. 243, 2; [Ammon.] In An. pi. 68,4 ff; [Them.] In An. pr. 92, 1-2; Simpl. In Cat. 209, 13; Ps. Alex. In S.E. 20, 27ff). Thus Sextus in a famous passage (A.M. VIII 12) tells us that the Stoics in the case of, say, Dion is walking distinguish between (i) the word Dion ; (ii) what is signified by the word; and (iii) the external object, which, Sextus says, they call runkhanon. There also may be a trace of this view in a scholium on Dionysius Thrax which distinguishes between the body and the case (550, 27). So it does seem that cases are not particular objects, but something particular objects somehow have. I say somehow have, because the passage in Simplicius (In Cat. 209, IOff) concerning Stoic and Platonist terminology shows, as does also, to a limited degree, the Stobaeus passage on ideas, that the Stoics tried, also terminologically, to distinguish between (i) an object s falling under, or participating in, a concept, (ii) its having a property or quality in the Aristotelian sense, like being just, or being ill, and (iii) its having a case. So what are these cases which are neither words, nor concepts, nor particular objects, but rather, it seems, something somehow signified by words, falling under concepts, and being had by objects? A yet further passage in Simplicius s commentary on the Caregories (53, 9ff) might be of some help, though it does not even pretend to report Stoic doctrine. It rather concerns Iamblichus s view as to how statements like Socrates is a man are to be understood. Iamblichus claims that it is not man in general, the universal or kind man, which is predicated here of Socrates, though Socrates does participate in the kind man, but rather a case. So Iamblichus s view seems to be that, insofar as Socrates participates in the kind man, there is a case or instance of humanity to be found in Socrates (something like an inherent form), in virtue of which Socrates is a man, and that it is this case, rather than the kind itself, which is predicated of Socrates. Now this is Iamblichus s metaphysics. It in several important respects differs from Stoic metaphysics. To begin with, man for him, as for all those in the Platonic and the Aristotelian tradition, is a species. For the Stoics, on the other hand, it is a genus, a generic item, whereas it is something like Socrates which most narrowly and strictly speaking, is a species (D.L. VII 61). This at first sight seems extremely puzzling, since we should have thought that Socrates is an individual, a particular object, which belongs to, rather than is, a species. Presumably the solution of the puzzle is this. Kinds or ideas, whether species or genera in the Platonic or the Aristotelian sense, for the Stoics are mere concepts, constructs of the mind. But, in addition to general concepts, they also assume individual concepts. And so, just as talk about the kind or genus man is understood as talk about the general concept man, so talk about the species Socrates is to be considered as talk about the individual concept (of) Socrates. So we have to distinguish between the

8 20 BICS individual Socrates, the species or individual concept Socrates, and the genus or general concept man. Now, what does fall under, is a case of, the individual concept Socrates? We might think that this must be Socrates himself. But we have to keep in mind that a central notion in Stoic metaphysics is the notion of an individual quality, an individual essence, something about, e.g., Socrates which determines him uniquely. So in the case of Socrates, there are three, rather than merely two, things to be distinguished: (i) the species or individual concept Socrates; (ii) the individual essence of Socrates; and (iii) Socrates himself who somehow has this essence. It also turns out that the Stoics characterize proper names by the very fact that they signify such individual essences or qualities (D.L. VII 58). But this means that, at least in the case of proper names like Dion or Socrates, individual qualities fit exactly the characterization of cases we have worked out so far. They are distinguished from the word as something signified by it, they fall under a concept associated with the word, but they also are distinguished from the particular objects which somehow have them. How exactly we are supposed to conceive of these individual essences is a matter we may leave aside here, except to note that later Platonists were quite ready to avail themselves of this notion, perhaps in part under the influence of Porphyry (In Cat. 129, 9-10). Thus, e.g., Boethius (In De int. I1 137, 7) can speak of the Platonitas of Plato. Thus, if we were concerned just with proper names, we would know what to say about cases, how they are distinct from objects, words, and concepts and constitute a quartum quid. But what are we to say about common names? What is it that falls under the general concept man? If we assume a view like Iamblichus s, it will be items like Socrates s humanity, Plato s humanity, Aristotle s humanity etc.. But what is the Stoic view? The Stoics say (D.L. VII 58) that common nouns signify a common quality. Now this can hardly mean that they assume that there actually exists in the ontology one item, the common quality man, which all men share. In this case it would be difficult to see why they should be so eager to claim that ideas, kinds, generic items are mere concepts or mental constructs. So they may assume, as Iamblichus does, that each human being has its own instance or case of humanity, but that, insofar as having this in each case amounts to the same, humanity can be said to be a common quality. But there is yet the further possibility that for the Stoics it is the individual essence itself which, in the case of a human being, is the instance or case of humanity in question, that the individual essence is the concrete and particular form humanity takes in each individual human being. We do not need to decide between these two options. On either interpretation, if we consider common nouns, the case here, too, is an essential characteristic or quality which an object has, a particular instance of some general kind. But we may note in passing that this also gives a more precise content to the remark of a scholiast on Dionysius Thrax (231, 22-23) that a case is called a case, because it falls from the common into individuality (idiptota), and to another scholiast s remark (In D. 7%. 550, 26-27) that a case falls from the general into the specific, that is that it falls upon the (particular) bodies. Now this doctrine of cases may be a more or less plausible piece of metaphysics, but we have to wonder what its point in the context of the theory of language is supposed to be. To understand a sentence like Dion is walking, why is it not sufficient to say that Dion here refers to, or names, Dion, and that, in addition to Dion, we have, if you wish, the individual concept of Dion and, finally, the word Dion? Why do we, fourthly, need a case?

9 MICHAEL FREDE: THE STOIC NOTION OF A GRAMMATICAL CASE 21 Let us take the sentence unrhropos periputei ( man is walking ). What does man here mean or refer to? Aristotle had taken the position that it refers to the species or kind man (Cur. 2b 1-3; cf. also Aristotle s indefinite propositions in De inr. and An. pr.). This, of course, he did not take to mean that the species man is walking. It rather is a truth about the species to the effect that some particular object falling under this species is walking. But it still is a statement about the species. After all, species or genera for Aristotle are beings which can be named and referred to. Now consider the following sophism: if something is in Athens, it is not in Megara; man is in Athens; hence it is not the case that man is in Megara. This is a sophism, because for man to be in Athens it suffices for there to be a particular man in Athens: and for man to be in Megara it suffices for there to be a particular man in Megara; and the fact that there is a particular man in Athens rules out that there be exactly this particular man also in Megara, but it does not rule out that there be any particular man in Megara. But if the kind man is a being, there to be named or to be referred to, why, given the truth of the general conditional if something is in Athens, it is not in Megara, should it not follow that if man is in Athens, man cannot be in Megara? Chrysippus s answer, it seems (cf. Simpl. loc.cir.), was that the kind man is not a something, let alone a being, and that hence the conditional does not apply to it. But a case of man is a something. And hence it does follow that if man is in Athens, he is not in Megara, if, that is, we understand man in man is in Athens, as we can in Greek, to refer to a particular man, though not to any man in particular, or rather to a particular case of man, though not to any case in particular. Perhaps it is this ambiguity of man in man is which Galen had in mind when he, in the passage mentioned above (De soph. p.106, Edlow), said that man might either refer to a case or to the ousia (i.e. the species or kind). But again, as with proper names, the question arises why we need a case, and why we do not just say that (a) man in (a) man is in Athens refers to a particular man, though not to any man in particular, rather than to a case of man? The answer presumably is that the Stoics do not treat Dion or man in Dion (or: man) is in Athens as naming, or referring to, something in the sense in which this is usually understood. In Stoic discussions of statements, unlike in discussions inspired by Plato or, more importantly, Aristotle, talk of subject, subject-expression, what the statement is about is conspicuously absent. There is no suggestion that Dion names, or refers to, an object such that the statement is true, if this object is in Athens. Instead, it seems to be the Stoic view that the statement is true if there is an object such that this object is Dion (or: a man) and it is in Athens (cf. Alexander In An. pr ff). So Dion or (a) man, on their view, is not referring to an object the statement is about, but rather makes reference to a qualification an object has t3 meet, if the statement is to be true in virtue of the predicate s being true of it. And this does explain why the Stoics think that Dion or (a) man make reference to a particular case, rather than to a particular object. Why the Stoics want to understand Dion is in Athens not as referring to Dion, but rather in the sense of there is something which is Dion and which is in Athens is a complicated matter. But here is one, albeit highly speculative, sugstion as to why this may be so. As we can see from Alexander (In An. pr. 402, 3ff.) the Stoics, against Aristotle, take the view that there are two construals of Dion is not walking. On the first reading this is I Cf Simpl. In Cat. 105, 13-14; D.L. VII 70; 186; Pnilop. In Cat. app. crit. p. 72

10 22 BICS the negation of Dion is walking, on the second the negation sign is construed with the predicate only. But this means that on the second construal both Dion is walking and Dion is not walking will be false, if Dion no longer exists since on the second construal to claim that Dion is not walking is to claim that there is something which is Dion and it is not walking. It is only on the first construal that Dion is not walking will be true, since it amounts to it is not the case that there is something that is Dion and it is not walking. This suggests that the distinction between objects and cases is supposed to allow us to deal with terms which have no reference. There may be no Dion, but it is still possible, though false, to claim that there is something which has the essence of Dion. The Stoic view also becomes apparent when we consider their related position on tensed statements (cf. Alexander In An. pr. 403, 14ff). It seems that they assumed that the sentence Dion was in Athens has two readings: (i) there is something which is Dion and it was in Athens; and (ii) there was a time when it was the case that there is something which is Dion and it is in Athens. An important difference between the two readings is that on the second construal there is no need for Dion to exist presently for the statement to be true. Now a possible suggestion is that the Stoics thought, inter alia, that for something to be named, or referred to, it had to be there to be named. Hence Dion in Dion was in Athens, on the second interpretation of this sentence, cannot be taken to name, or refer to, Dion, since possibly Dion does not exist any more. But as the way we construe Dion in this sentence on this reading does not differ from the way we construe it in other sentences, there is no reason to say that it ever names or refers to Dion in the way this is usually understood. Hence sentences of the form Dion is in Athens are understood as saying that there is something which is Dion and it is in Athens. And here, that is in something which is Dion, Dion signifies or indicates the distinctive quality or essence of Dion, something which falls under the concept of Dion, and which something in Athens has to have if the sentence is to be true. This, then, I suggest, is the motivation for the introduction of the metaphysical notion of a case into the analysis of propositions. It is based on a certain understanding of proper names and names for kinds of things, in particular of how they function in subject-place, but also on a metaphysical theory about kinds, objects and cases. A good deal of work would need to be done to explain how this initial notion of a case was extended and refined in such a way as to, e.g., deal with nouns in general or with demonstratives. But this is not my concern here. Nor is it my concern to again discuss the question whether cases, on the Stoic view, are incomplete lekra. I think that, being, at least in the paradigm cases, qualities, they must be bodies, since the Stoics take qualities, properly speaking, to be bodies. Hence Clement s claim that cases are agreed to be incorporeal (Strom. VIII 9, 26, 5), the main piece of evidence to the contrary, possibly should be taken to reflect a non-stoic view, based on the adaptation of the Stoic notion of a case either into a basically Peripatetic or a basically Platonist theory. We have already seen from two passages in Simplicius s commentary on the Categories (53, 17; 209, 13, though in the latter passage they call it differently) that Platonists had adopted the notion of a case for something like their imminent forms which, of course, they would regard as incorporeal. But it also is very possible that Clement or his source are confused. The version of the sophism Clement tries to explain by introducing the notion of a case already in itself seems rather confused. It is a garbled version of a sophism Chrysippus had used already (D.L. VII 186), but in order to show, we may suppose from the wording, that it is a mere sound or word which comes through your mouth, rather than a wagon or house, when you say wagon or house. It is, after all, not some incorporeal meaning which comes

11 MICHAEL FREDE: THE STOIC NOTION OF A GRAMMATICAL CASE 23 through the mouth, as Clement suggests. But there might be a further confusion. We have seen that Sextus Empiricus calls the different meanings of dog cases (A.M. XI 29). Now, when Clement speaks about the very same matter of the different meanings of dog, he does this in the context of a distinction between the objects called in a certain way, say animal, the word used to call them this way, and what is signified by the word. And he characterizes (Strom. VIII 4, 13, 1) this signification or meaning as incorporeal and a lekton and a thing (prdgrna) and a thought and anything but an animal. But it now should be apparent that the notion of a case we find in this passage in Sextus is not the Stoic notion of a grammatical case. And hence it also almost seems as if Clement or his source confused the two notions of a case, a case in the sense of something like the meaning of a word like dog and a case in the sense of what is signified by dog in (a) dog is running down the alley. There are further signs of confusion, but what has been said already should suffice to disqualify Clement s remark as reliable testimony for the Stoic view and hence as evidence against the suggestion that cases are not incomplete lekta, and thus incorporeal. But this is not what I am concerned with here. What I am rather concerned with is how this term case, which in its origins has nothing to do with inflection or declension, came to have its traditional meaning of a characteristically declined word-form. As I mentioned earlier, objects enter facts in different ways. It is not the same thing for the father to follow the child and for the child to follow the father. Correspondingly the cases are ordered and related differently. The case of the child in the case either of something which follows something or of something which is being followed by something. And accordingly it is a nominative or an accusative case. Notice here that, whereas on the traditional notion of a case the accusative is a kind of case in the way in which a lion is a kind of animal, such that an accusative and a dative cannot be the same case, on the Stoic view it is one thing for something to be a case, and a further thing for it to be an accusative, and this in such a way that the same case in one proposition may be an accusative and in another, or even in the same proposition, a dative. On the Stoic view the case is a case independently of its being part of some proposition, and it only becomes an accusative, or whatever, by entering into a proposition in a certain way. Now, as I also mentioned at the outset, the Stoics do assume that sentences mirror or represent what gets said (the lekton) by using these sentences. The Stoics contribution to grammar largely rests on their systematical efforts to observe and to understand how sentences reflect the underlying lekta. And so they also noticed that the fact that cases enter into lekta in different ways is reflected by the phenomenon of declension. Correspondingly, they distinguished between declinable and undeclinable parts of speech (D.L. VII 58). The word for declinable they used was ptfifikon, a word which by its formation indicates the connection between declension and cases. They noticed that the inflection of a declined word marks whether the case signified is, say, a nominative or an accusative. We also already have noticed that, given the assumed correspondence between lektu and sentences, the terms used for the description of the relevant features of lekta tended to be applied to the corresponding features of sentences, as, e.g., the terms transitive or conjunction. So we would expect that even the Stoics themselves would be prepared to also call a noun, depending on the way the underlying case enters into the lekton, nominative or accusative. And it was their language which was taken over by the grammarians. It could be taken over without committing oneself to the Stoic theory of cases or the Stoic theory of Iekta which had given rise to it. In fact, given the metaphysics and the philosophy of language

12 24 BICS and logic underlying the theory, and given the difficulties both of the Stoic theory and of the problems it tries to deal with, we should imagine that the grammarians were more than glad not to have to commit themselves to the Stoic theory of cases, to leave these problems to the philosophers, and to altogether forget about them, as far as possible. In this way accusative or dative cases did become declined word-forms. But this raised the question why they were called cases, what they were cases of. Here one answer, inspired perhaps by Stoicism, was that we have to distinguish between the word in the abstract (to genikon onoma) and the different concrete forms it takes. In this sense the nominative is as much a case as the genitive or the dative (a view attributed to Apollonius Dyscolus and Herodian, Schol. in D. Th. 546, 7ff.; cf. also app. crit. ad Ammon. In De inr. 43, 19 and Priscian Inst. V 68, p. 184, Iff). But Peripatetics, relying on the authority of Aristotle, would insist on taking a position which would explain why Aristotle had called the oblique cases cases and had contrasted them with the nominative form as being not a case, but a noun or name (De int. 16a 32f.). So they claimed that the cases were cases insofar as they were derived and inflected from the standard form of the word in the nominative. After all, Peter s name is not Peter s, but Peter. Peter s is not a name at all. Hence they objected to the Stoic view that the nominative is as much a case as the other cases. Presumably to accommodate the already established usage among grammarians, they were, though, willing to say that, derivatively or catachrestically, the nominative can be called a case, too (cf. Ammon. In De inr. 42, 30ff.; Stephanus In De int. 10, 21ff.; Anon. In De int. 2, 6ff.). Given the increasing influence of Aristotle, and in particular of his Categories and De interpretatione, in late antiquity, this position also found acceptance among grammarians (cf. Schol. in D. Th. 546, 5ff.), presumably all the more easily as the notion of a word in the abstract which does not have any form in particular seemed too abstract to them. And thus we arrive at the notion of a case which we find to the present day, at least in grammars of Greek and Latin, e.g. in Schwyzer-Debrunner (Griechische Grammarik I1 53). What, I think, confuses us in our understanding of the Stoic notion of a case is that we not only tend to project our notion back on the ancients, but also believe to be justified in this, because we think that this is, more or less, the notion which we find already in Aristotle s De interpretatione, and that the notion the Stoics use must be a relative, if not a derivate, of Aristotle s notion, and hence, like ours, the notion of a word-form. But we must not forget that the notion Aristotle uses in De interpretutione is not the notion of declined word-form thought of as being derived from its standard form in the nominative. It rather is the more general notion of a word-form thought to be derived from a standardform which just happens to be applied here to the forms of the noun in the oblique cases. There is, moreover, no reason to suppose that the Stoics when they talked about cases even thought of the use of the word we find in Aristotle. They were interested in a metaphysical entity when they telked about cases. It was, as it were, a historical fluke, that they put this notion to a use in linguistics which made it refer to almost the same thing as Aristotle, using an altogether different notion of case, had been referring to; almost the same, since the Aristotelian term referred to word-forms and the Stoic term to items signified by these word-forms. These the Stoics carefully distinguished from the corresponding word-forms. But when later authors disregarded or denied this distinction, confusion was almost inevitable. And it is this confusion, rather than the lack of sources, which has made it so difficult to understand what the Stoics have to say about cases. Kehle College, Oxford