The Stoic Theory of Case

1 The Metaphysical Status of Case

Before attempting to understand its place in the semantic theory, we must establish what case is for the Stoics. The Stoics distinguish between semainonta, things that signify, which are at the level of what they call phōnē, voice, and semainomena, things signified.^[4] One might expect to find case described as a purely linguistic feature, at the level of phōnē, rather than as something signified. In this respect, the Stoic notion of case differs from the standard modern notion of grammatical case, as well as from many ancient conceptions of it.^[5] Evidence suggests that the Stoics thought of case as part of the signified, and not merely as a feature of words at the level of phōnē.^[6] Most notably, in Diogenes Laertius’ account of Stoic doctrines, case is mentioned not in the discussion about the signifying (peri phōnēs) but about the signified.^[7] Similarly, in his description of so-called categorical propositions, Diogenes describes them as composed of a nominative case and a predicate (κατηγορικὸν δέ ἐστι τὸ συνεστὸς ἐκ πτώσεως ὀρθῆς καὶ κατηγορήματος, Diogenes Laertius (DL) VII.70). Propositions are signified, as are predicates. Cases are thus put on a par with signified components, rather than components of the sentences expressing them.

To be sure, this needn’t mean that case is thought by the Stoics to be only at the level of signified. It is a fact about the Greek language that it is inflected and most linguistically inclined thinkers both before and after the Stoics spoke of case as a property of words, at the level of the signifying.^[8] In reporting Stoic views, Diogenes Laertius (VII.58) attributes to some a definition of certain parts of speech (verbs, conjunctions) as caseless (aptōton) while others (the article) are said to admit of case (ptōtikos), all apparently at the level of the signifying. We therefore needn’t infer that case was, for the Stoics, solely a signified, while it was for almost everyone else a morphological feature of words. Rather the Stoics likely thought of case as twofold: both signified and a property of phōnē.^[9] In fact, it is likely that they took linguistic case to be a marker of the semantic content expressed by words that bore it: grammatical case, as a feature of phōnē, reveals case at the level of signified. At any rate, it is the latter which is of interest here and what I will mean by “case” in the following discussion will be, unless otherwise noted, the signified case.

The first salient question, then, is what kind of signified is case? Some of the most prominent signified items (sēmainomena), in the Stoic taxonomy are lekta, “sayables”, which are expressed by written and spoken words, and which form the content of thoughts and beliefs.^[10] Notably, lekta are incorporeal.^[11] In establishing the status of case, we might therefore begin with whether cases are incorporeal, which would put them among the lekta. Several sources do in fact suggest that cases are incorporeal.^[12] Among them, a treacherous passage in Clement states the incorporeality of case in no uncertain terms:^[13]

ἡ πτῶσις δὲ ἀσώματος εἶναι ὁμολογεῖται· διὸ καὶ τὸ σόφισμα ἐκεῖνο οὕτως λύεται· “ὃ λέγεις, διέρχεταί σου διὰ τοῦ στόματος”, ὅπερ ἀληθές, “οἰκίαν δὲ λέγεις, οἰκία ἄρα διὰ τοῦ στόματός σου διέρχεται”, ὅπερ ψεῦδος· οὐδὲ γὰρ τὴν οἰκίαν λέγομεν σῶμα οὖσαν, ἀλλὰ τὴν πτῶσιν ἀσώματον οὖσαν, ἧς οἰκία τυγχάνει.

Case is agreed to be incorporeal; and hence the famous sophism is solved as follows: “What you say passes through your mouth.” This is true. “You say: A house. Therefore a house passes through your mouth.” This is false. For what we say is not the house, which is a body, but the case, which is incorporeal and which the house bears.

Clement Strom. VIII.9.26.5 (translation Long and Sedley 1987, modified)

According to this – difficult and potentially dubious – passage, case is signified and incorporeal. Taken at face value, the passage therefore suggests that case must be a lekton of some kind, since it must fall under one of the four classes of incorporeals (place, time, void, and lekta) and lekta is the only plausible candidate.

Although our evidence provides many examples of lekta, both complete and incomplete, case is rarely if ever listed as one of them.^[14] Nonetheless, it is mentioned immediately following the enumeration of predicates (katēgorēmata) in Diogenes Laertius, where cases appear to be to nouns what predicates are to verbs. Predicates are said to be incomplete lekta (ἔστι δὲ τὸ κατηγόρημα… λεκτὸν ἐλλιπὲς DL VII.64). They are the incorporeal counterparts, at the level of lekta, to the corporeal verbs at the level of phōnē. Predicates are also said to be constructed with cases (συντακτὸν ὀρθῇ πτώσει DL VII.64). This would suggest that verbs express predicates while nouns express cases.^[15] Cases, then, would be lekta much in the same way as predicates.

This view is rejected by scholars including Frede, Schubert, and, to some extent, Long.^[16] While all agree that cases are not properties of expressions and accept the overwhelming evidence in favour of the view that they are part of what is signified by expressions, they nonetheless deny that cases are lekta. Long argues that the meaning of a noun is not a lekton but rather the very thing to which it refers.^[17] Cases on his view are “the means of referring [deficient lekta, namely predicates] to a subject or actual thing”. This leaves a few unanswered questions: what it means to refer lekta to things, how cases do this, or why they are necessary in order to do it. The usual distinction is between the corporeal signifying expressions and the incorporeal signified (the lekton), which bears some relation to reality. Propositions, which are expressed by signifying expressions, correspond to facts or states of the world: the proposition ‘Socrates walks’ corresponds to Socrates walking. At a more granular level, predicates, as constituents of propositions, seem to correspond to constituents of states of the world: ‘walks’ in ‘Socrates walks’ corresponds to the disposition walking realised in Socrates.^[18] On Long’s view it would seem that the intermediary, the incorporeal signified, is left out of the semantics of nouns, and replaced with the case, ptōsis. The exact metaphysical status of ptōseis is not well accounted for, nor is the mechanism by which they act as a means of referring (which I take to mean relating in some sense) lekta to things. Long’s view appears to leave us with an under-specified entity in the already complex Stoic theory.^[19]

Frede’s view is not entirely dissimilar to Long’s, though it is a little more elaborate. He similarly thinks that case is a non-linguistic item, neither a property of expressions nor a lekton. Instead, he argues that case should be understood as a particular instance of a universal.^[20] “Man”, on this view, signifies a case of an individual man, which Frede identifies as the instantiation of the common quality (man-ness) in the individual man.^[21] Unlike on Long’s view, the case does not seem to have a mediating role between the word and an object, here an instantiation of a quality. Instead, case just is the instantiation of the quality in the individual to which the noun refers. In other words, cases and qualities coincide so that saying that a noun signifies a case is just the same as saying that a noun signifies the instantiation of a quality.

One of Frede’s arguments in favour of this is that qualities are distinct from the objects they qualify, fitting nicely with evidence suggesting that words, cases, and objects are distinct from each other. Frede indeed bases his view on the report that nouns such as “Socrates” signify qualities (DLVII.78) together with the claim that propositions such as “Socrates walks” are composed of a case and a predicate (DL VII.63-4). This presents two difficulties. Firstly, it is not clear that qualities and qualified objects are sufficiently distinct for the view to gain traction. It is certainly true that a quality is conceptually distinct from the object it qualifies, but the distinction is notably imperceptible to the human eye. Referring to the quality and referring to the object which has the quality will likely be, for any human, one and the same process. To justify introducing both case and quality into the semantic theory in the way the Stoics do, both ought to have more of a role to play, if only an explanatory one. Secondly, and by his own admission, Frede’s view cannot account for words other than nouns. For example, pronouns, which do not signify qualities, nonetheless seem to express case, since propositions of the form “this one walks” are said to be comprise a “deictic nominative case” (DL.70).^[22] This speaks against a view on which case and quality are one and the same thing.

In addition, Frede gets to his view by dismissing much of the evidence for the incorporeality of cases as either confused or untrustworthy.^[23] While he is right that many of our sources are either untrustworthy or confused (or both), the details of his diagnoses seem worth prodding. In particular, Frede dismisses Clement’s testimony (above) as possibly reflecting a non-stoic view and adapting the Stoic notion of case to a Platonic theory.^[24] Yet the context here is clearly a Stoic one. We know from Diogenes Laertius that Chrysippus did deal with this particular sophism.^[25] Clement has been discussing the Stoic theory of causation.^[26] The idea that Clement would report the Stoic solution to the sophism using a Platonic adaptation of a Stoic concept is not sufficient to dismiss the report that case is incorporeal. Frede also argues that Clement may be confused, since a more plausible answer to the sophism would be to say that the sound, not the incorporeal meaning, goes through one’s mouth. This is certainly true, and it is unlikely that Chrysippus would have said that the incorporeal passes through the mouth. Nonetheless, an alternative diagnosis of the confusion is available. It is quite possible that Clement could be mistaken about the answer to the sophism, but not about the incorporeality of case. In fact, he may well be conflating – a common mistake – the signifying expression (the phōnē) with what it signifies (an incorporeal lekton) and giving a garbled report of the answer to the sophism on the basis of what he knows to be true: that case is incorporeal.^[27] I am therefore disinclined to dismiss this passage outright, and rather think that we ought to attempt to make sense of it.

Another reason to think that Clement’s report may well be right is given to us by Diogenes Laertius, in his definition of predicates. Predicates, he tells us, are incomplete lekta which are to be constructed, or combined, with cases in various ways:

ἔστι δὲ τὸ κατηγόρημα τὸ κατά τινος ἀγορευόμενον ἢ πρᾶγμα συντακτὸν περί τινος ἢ τινῶν, ὡς οἱ περὶ Ἀπολλόδωρόν φασιν, ἢ λεκτὸν ἐλλιπὲς συντακτὸν ὀρθῇ πτώσει πρὸς ἀξιώματος γένεσιν. (…) οἷον τὸ ‘διὰ πέτρας πλεῖν.’ καὶ τὰ μέν ἐστι τῶν κατηγορημάτων ὀρθά, ἃ δὲ ὕπτια, ἃ δ’ οὐδέτερα. ὀρθὰ μὲν οὖν ἐστι τὰ συντασσόμενα μιᾷ τῶν πλαγίων πτώσεων πρὸς κατηγορήματος γένεσιν, οἷον ‘ἀκούει,’ ‘ὁρᾷ,’ ‘διαλέγεται·’

A predicate is what is said of something, or a thing to be constructed with something or things, according to Apollodorus, or an incomplete sayable to be constructed with a nominative case to form a proposition. (…) And of predicates, some are active, some passive, and some neither. Active are those which are construed with an oblique case so as to produce a predicate, for example ‘hears’, ‘sees’, ‘converses’.^[28]

Diogenes Laertius VII.64

This passage, too, is difficult and contentious. For our purposes, we only need to note that predicates are clearly thought of as being in some way constructed with (suntakton) cases, either nominative or oblique. On Frede’s view, this will require a corporeal (a case) and an incorporeal (a predicate) being constructed or joined with one another to form another incorporeal (a proposition). How this might work is unclear. How can a body be constructed with an incorporeal and how can an incorporeal have a corporeal constituent?^[29] Frede does not address this worry. It seems that he does not think it implausible that the Stoics would have held such a view. Yet, the Stoics are staunch physicalists. On their view, incorporeals do not exist and only subsist. Even if we were to think that their subsistence does not mean that they are necessarily dependent on existing things, we cannot deny that they have a relevantly different ontological and metaphysical status from bodies.^[30] We may think that lekta are “items in the world”, as Bronowski argues,^[31] but they are not in the world in the same sense that bodies are in the world. The idea that they could combine in any meaningful way with corporeal things seems difficult to square with Stoic doctrine. Incorporeals, after all, unlike bodies, cannot act or be acted upon, and Nemesius attributes to Cleanthes the view that “no incorporeal interacts with a body and no body with an incorporeal” (81.6-10).^[32] It is even more implausible to think that this combination of a corporeal and an incorporeal could together form an incorporeal, that is, that incorporeal lekta could have corporeal constituents. Diogenes tells us that predicates can join with cases to form other predicates. A two-place predicate, ‘hears’, will construct with a case, for example ‘Plato’ (in the accusative) and this will yield the one-place predicate ‘hears Plato’.^[33] This new predicate can, in turn, be constructed with a nominative case, for example ‘Socrates’, to form a proposition, ‘Socrates hears Plato’. Both predicates and propositions are incorporeal. Yet, on this view, they would be partly composed of corporeals. This seems unlikely, given Stoic ontological commitments.

I am much more inclined to agree with those who argue that cases are in fact lekta.^[34] The argument from silence – based on the fact that ptōsis is not explicitly listed as a lekton^[35] – is perhaps the strongest one against this view, and it is weak, given the state of the evidence. On the other hand, the evidence that cases are incorporeals, and such as to be connected to other lekta is clear and speaks in favour of the idea that cases, too, are lekta, since it is likely that an incorporeal lekton can only be constructed with another incorporeal lekton. This in turn supports taking Clement’s albeit otherwise confused report as reliable in this particular respect.

This leaves open the questions of whether cases are expressed by nouns as a whole and whether they exhaust the lekta expressed by nouns. We could perhaps think that just as a noun can be broken down into a nominal root and an ending, e.g. “Socrat-es”, the lektical bits can similarly be broken up so that the nominal root, “Socrat-” expresses one thing (or nothing) and the ending “-es”, which marks case at the level of phōnē, expresses another, namely case.^[36] The latter might be thought to be a bit of the lekton expressed by the noun as a whole, which would be divisible into ‘Socrat-’ and ‘-es’, mirroring the division of the noun. We would then have to account for what the expressed ‘Socrat-’ is, if anything. Recall that the examples of propositions given by Diogenes Laertius and Sextus Empiricus suggest that ‘Socrates walks’ provides an adequate rendering of the content of a middle proposition, and that, according to the argument so far, this content must be lektical, so that ‘Socrates’, even if it is to be broken up, must be broken up into lekta. This would have the undesirable result of leaving ‘Socrat-’ as a new and unaccounted for type of lekta. In addition, the evidence according to which the binding of a case and a predicate yields a proposition again suggests that the constituents of the proposition ‘Socrates walks’ just are the predicate (‘walks’) and the case. This all speaks in favour of taking the case (at the level of the lekton) to be ‘Socrates’, and not merely ‘-es’.^[37] At any rate, it seems possible to conclude broadly that cases are lekta, paradigmatically expressed by nouns, and counterparts to predicates, the lekta expressed by verbs; the two, joined together, can form complete lekta, such as propositions.

2 The Semantic Function of Case

As transpires from the discussion so far, cases are just one of several items that are said to be in some way signified by nouns, which appear to be the paradigmatic and best attested signifiers of case. We must therefore now account for the role cases, as lekta, might play in the larger semantic picture.

Let us survey some of the evidence relating to the semantics of nouns. The Stoics distinguished between three main types of propositions: definite, indefinite, and middle, following Sextus Empiricus’ terminology.^[38] Sextus in fact distinguishes the different propositions by the semantics of their respective subjects and what expresses them and describes middle propositions in the following way:

μέσα δὲ τὰ οὕτως ἔχοντα ‘ἄνθρωπος κάθηται’ ἢ ‘Σωκράτης περιπατεῖ’. (…) τὸ δὲ ‘Σωκράτης κάθηται’ μέσον ὑπῆρχεν, ἐπείπερ οὔτε ἀόριστόν ἐστιν (ἀφώρικε γὰρ τὸ εἶδος), οὔτε ὡρισμένον (οὐ γὰρ μετὰ δείξεως ἐκφέρεται), ἀλλ’ ἔοικε μέσον ἀμφοτέρων ὑπάρχειν, τοῦ τε ἀορίστου καὶ τοῦ ὡρισμένου.

Middle propositions are ones such as ‘man is sitting’ or ‘Socrates is walking’. (…) And ‘Socrates is sitting’ is middle, according to them, since it is neither indefinite (for it has delineated a species), nor definite (for it is not expressed with deixis), but it seems to be between the indefinite and the definite.

Sextus Empiricus M VIII.97.2-3,9-11

What the subjects of the sentences that express them have delineated (aphōrike), that is, what they pick out, and how (with or without deixis), is what sets different types of propositions apart. Proper nouns (onomata) and common noun (prosēgoriai) form sentences that express the same kind of propositions because they share a semantic feature: they delineate a “species” (aphōrike to eidos).

Diogenes Laertius adds that proper nouns and common nouns signify qualities, a proper noun a proper quality, and a common noun a common quality: ^[39]

ἔστι δὲ προσηγορία μὲν κατὰ τὸν Διογένη μέρος λόγου σημαῖνον κοινὴν ποιότητα, οἷον ‘ἄνθρωπος,’ ‘ἵππος’· ὄνομα δέ ἐστι μέρος λόγου δηλοῦν ἰδίαν ποιότητα, οἷον ‘Διογένης’, ‘Σωκράτης’.

According to Diogenes [of Babylon], a common noun is a part of speech which signifies a common quality, for example “man”, “horse”. A proper noun is a part of speech which designates a proper quality, for example “Diogenes”, “Socrates”.

DL VII.58

As we have seen, Diogenes Laertius also later reports that that the propositions expressed by sentences formed by nouns are ones “composed of a nominative case and a predicate, for example ‘Dion is walking’” (τὸ συνεστὸς ἐκ πτώσεως ὀρθῆς καὶ κατηγορήματος, οἷον ‘Δίων περιπατεῖ’, DL VII.70) and I have argued that this is because case is the lekton expressed by the noun. That is, in this report about the syntax of propositions, the case appears to be the lektical correlate to the noun, just as the predicate is the lektical correlate to the verb, so that the case at the lektical level is ‘Dion’, the lekton expressed by the proper noun “Dion”.

These passages are thus reporting different facts about proper and common nouns, all couched in slightly different terms, providing (i) an explicit report about what they “signify” or “designate”; (ii) an indirect report of what they – or more precisely the propositions they contribute to expressing - “delineate”; (iii) an implicit indication of what they express, that is what they contribute to the proposition at the lektical level. All three thereby appear to attest to different aspects of what proper and common nouns signify in a broad sense. That is, all are in some broad sense about the semainomena (signifieds) related to proper and common nouns. This therefore suggests that case is part of a larger semantic picture which somehow also includes species and quality. What is less clear is how case, species, and quality all relate to each other qua signifieds, and in what way or in what sense, they are signified (or designated, or expressed, or delineated) by proper or common nouns. Let us therefore try to understand how they might relate and where they might fit in this overall semantic picture for proper and common nouns.

Species and quality are closely related. ^[40] Quality – or perhaps more accurately the qualified (poion) – is one of the four Stoic ontological categories. It is corporeal and with matter jointly constitutes qualified objects. Species is a similarly technical notion for the Stoics. It includes species with just one individual, so that both man and Socrates are species. Importantly, eidē are special kinds of ennōemata, concepts, namely universals.^[41] For any qualified individual, there are thus both a corresponding quality and a corresponding eidos. Just as man-ness and Plato-ness are qualities, the concepts <man> and <Plato> are eidē.^[42] These eidē are items perceived or tended to by our minds, which correspond to qualities in the world. In this sense, they are mental representations of qualities.^[43] The concept <Plato> is a mental counterpart to the corporeal quality Plato-ness.

Returning to semantics, we can think of the link between quality and eidos as playing a role in reference-fixing. The reference of a proper noun such as “Plato” is neither a quality nor a concept, but rather an object in the world, namely Plato himself, who has the proper quality of being Plato. Nonetheless, the concept <Plato> and the quality Plato-ness both play a part in fixing the reference of the noun “Plato” to the qualified individual Plato. The quality delineated by the noun enables reference to the object in the world because the object has the quality of which this concept is a mental representation. By using the proper noun “Plato”, then, I refer to Plato by means of the concept <Plato> and insofar as he has the quality of being Plato. Concept and quality thus work together to fix reference to objects. The relationship of species and quality to the referent of a noun can thus be illustrated as in the following diagram:

Nouns refer to the qualified object, signify (or designate) the quality, and delineate the species. It is precisely by doing the latter that they do the former. And it is in this sense that both species and quality feature in the semantics of nouns.

We must now account for the role of cases in this picture. Here, Frede’s work may help. His view is founded on evidence that suggests that ptōseis are so called because they have fallen (peptōkasi) from a concept. In particular, Ammonius tells us that the Stoics disagreed on this matter with Peripatetics. Peripatetics called cases ptōseis because they had fallen from the nominative, and therefore did not call the nominative a case. The Stoics, on the other hand, did call the nominative a case and justified it in the following way:

ἀποκρίνονται <οἱ ἀπὸ τῆς Στοᾶς> ὡς ἀπὸ τοῦ νοήματος τοῦ ἐν τῇ ψυχῇ καὶ αὕτη πέπτωκεν· ὃ γὰρ ἐν ἑαυτοῖς ἔχομεν τοῦ Σωκράτους νόημα δηλῶσαι βουλόμενοι τὸ “Σωκράτης” ὄνομα προφερόμεθα· καθάπερ οὖν τὸ ἄνωθεν ἀφεθὲν γραφεῖον καὶ ὀρθὸν παγὲν πεπτωκέναι τε λέγεται καὶ τὴν πτῶσιν ὀρθὴν ἐσχηκέναι, τὸν αὐτὸν τρόπον καὶ τὴν εὐθεῖαν πεπτωκέναι μὲν ἀξιοῦμεν ἀπὸ τῆς ἐννοίας, ὀρθὴν δὲ εἶναι διὰ τὸ ἀρχέτυπον τῆς κατὰ τὴν ἐκφώνησιν προφορᾶς.

The Stoics reply that it [sc. the nominative case] too has fallen, namely from the thought (noēma) in the soul: ‘For when we want to make clear the thought of Socrates which we have in ourselves, we utter the name “Socrates”. So, just as a stylus released from a height and lodged upright is said both to “have fallen” and to have “the upright fall (=case)” (orthē ptōsis), in the same way do we <Stoics> think that the nominative “has fallen” from the conception (ennoia), and that it is “upright” because it is the archetype of the utterance used in the expression <of its other forms>.’

Ammonius in Ar. de Int. 42.30 (translation Blank 2014, modified)^[44]

The nominative, according to Ammonius, has fallen from a conception (ennoia) in our soul.^[45] This is promising, if we want to link cases to eidē, which are concepts. However, ennoiai are not identical to ennoēmata: ennoiai are psychological states which have ennoēmata as their object.^[46] Nonetheless, it is possible that Ammonius is conflating the two here.^[47] In fact, other passages on the subject of case suggest that incorporeal cases fall not from our psychological states, but from the non-existent objects towards which our soul is directed.^[48] The scholiast Stephanus, in his commentary on the Tēchnē, attributes to the Stoics the view that cases fall from the general into the particular:

Εἰ ὀρθή, πῶς πτῶσις; Ὅτι πέπτωκεν ἐκ τοῦ ἀσωμάτου καὶ γενικοῦ εἰς τὸ εἰδικόν· ὀρθὴ δέ, ὅτι οὔπω ἐκινήθη εἰς πλάγιον, ἢ ὅτι ἐξ αὐτῆς τὰ καλούμενα παρὰ τοῖς <Στωϊκοῖς> ὀρθὰ ῥήματα, ἅ εἰσιν ἐνεργητικά, οἷον ‘Σωκράτης τύπτει’.

If it [sc. the nominative] is upright (orthē), why is it a case? Because it has fallen from what is non-corporeal (asōmaton) and generic (genikon) into what is specific (eidikon). But it is upright, because it has not yet been altered into an oblique [case], or because it is the foundation of what the Stoics call upright, that is active, verbs, e.g. “Socrates strikes”.

Scholia in DThrax 230.25-33 (translation Long and Sedley 1987)

καθὸ μὲν γὰρ πέπτωκεν ἀπὸ τοῦ κοινοῦ εἰς ἰδιότητα, ὀνομάζεται πτῶσις.

Since it has fallen from the general (koinon) into a particular property (idiotēs), it is called a case.

Scholia in DThrax 231.22^[49]

These passages tell us that what cases fall from is non-corporeal (asōmatos) and generic (genikon) or common (koinon). They are difficult because Stephanus is not using technical Stoic terminology accurately. In particular, he is not using asōmatos (incorporeal) in the technical Stoic sense. That is, case does not fall from one of the four Stoic incorporeals. It could, however, have fallen from something that is not corporeal, namely an eidos. In other words, we must assume that the scholiast is here describing the concept as “incorporeal”, even though the Stoics did not include concepts among the incorporeals.^[50] The metaphysical distinctions underlying these Stoic views and their use of terminology are tenuous and it should not surprise us if they confused scholiasts, who did not share their view, much less the basis for them, and might easily have conflated what is not-corporeal with incorporeals, as I suggest Stephanus did here.

The formulation in Stephanus is also echoed by another passage from the Scholia, which adds that case falls upon bodies:

Ἡ δὲ εὐθεῖα καλεῖται πτῶσις, καθὸ μεταπίπτει εἰς τὰς λοιπὰς πτώσεις, ἢ καθὸ ἀπὸ τοῦ γενικοῦ μεταπίπτει εἰς τὸ εἰδικόν, ἤγουν ὅτι ἐπιπίπτει τοῖς σώμασι.

It is called the straight (eutheia) case, because it falls into the remaining cases or because it falls from the generic (genikon) into the specific (eidikon), that is to say, because it falls upon (epiptei) bodies.

Scholia in DThrax 550.25-27^[51]

What it means for a case to “fall upon” a body, or to “fall into” the specific is unclear. It seems likely that compounds of the verb piptein, “to fall”, are used here precisely because the matter at hand is the word “case”, ptōsis (literally “fall”), which comes from the verb. We needn’t take the falling image too literally: it is used to explain the fact that cases are so called, more than as a serious analogy or illustration.^[52] Nor should these texts be interpreted as suggesting that case goes from the specific or the particular into a body, or renders it into a body, as Frede (1994a, 19–20) argues.^[53] Rather case appears to be something that falls or happens upon a body. This is consistent with the passage in Clement (Strom. VIII.9.26.5, above) where the house, a particular, is said to bear or meet (tunchanein) a case.^[54] All three passages taken together, then, suggest that cases link or mediate between something non-corporeal and something corporeal: a case falls from the non-corporeal (πέπτωκεν ἐκ τοῦ ἀσωμάτου, 250.25) and it falls upon bodies (ἐπιπίπτει τοῖς σώμασι, 550.26).^[55] Alongside Ammonius’ testimony, this suggests that eidos is a likely candidate for the point of origin of cases’ so-called fall, since eidē are concepts, which are non-existent (and non-corporeal) universals.

What they fall into, as it were, is less clear. They are said to fall into the specific (eidikon) or a particular property (idiotēs). The latter might suggest that they fall into qualities. However, Stephanus is not using the Stoic terminology entirely accurately. Idiotēs is not the technical term for Stoic proper quality, idia poiētēs. Rather, it parallels to eidikon in the two other passages, and is set in contrast with to koinon, what is common, so that it is likely that it is used for to idion, what is particular, rather than for the Stoic proper quality. Similarly, to eidikon is set in contrast with to genikon, which parallels to koinon at 231.22, so that it is plausible that it is used not for what belongs to a species, but rather for what is specific, as opposed to general, that is, what is particular. If this is right, the texts are then all suggesting that cases bear relations to concepts, which are common or universal, on the one hand, and to particulars, on the other.^[56]

Still, this dual relationship needs to be defined further. I suggest, speculatively, that this relationship which operates as a link from concepts, towards which we direct our minds, to particulars in the world is akin to reference-fixing. It operates in the following way. When I utter a noun, I make use of the associated concept: as I utter the noun, I activate the ennoia which has as its object the ennoēma associated with the noun. When I say “Socrates”, I invoke, as it were, my concept of Socrates-ness. This concept is what enables reference, because it is linked to the ptōsis, the lekton ‘Socrates’, expressed by the noun “Socrates”. When expressed, the ptōsis falls from the concept.^[57] By saying “Socrates”, I put the concept <Socrates> to work and from it falls the ptōsis ‘Socrates’. Now, the concept <Socrates> has a special relationship with the peculiar quality Socrates-ness, which qualifies Socrates’ matter. It is, as we have seen, a mental representation of it. The noun I utter, “Socrates”, which expresses the ptōsis ‘Socrates’, is linked to the quality Socrates-ness in the world, by the concept <Socrates>.

It is precisely through the concept that the noun is able to refer to Socrates.^[58] The ptōsis thus expressed calls the same concept <Socrates> to my interlocutor’s mind: it gives rise to an ennoia, a psychological state, directed at the concept and thus delineates the eidos <Socrates>, as Sextus Empiricus suggests when he claims that middle propositions ‘delineate a species’ (aphōrike to eidos, SE M VIII.97, above). The relationship of cases to concept, the so-called “fall”, should therefore be thought as operating in both directions: when I utter the noun, activating my ennoiai, and appealing to the concept, the case falls from the concept, and is expressed by the noun. Once it has been expressed, it can be grasped by my interlocutor. Grasping the case will take her to the concept from which it has fallen, which will in turn take her to the referent, by allowing her to pick out the corresponding quality in the world, and to understand that I am saying something about Socrates.

To sum up, when I say “Socrates” and express ‘Socrates’,^[59] invoking, or directing my soul towards <Socrates>, I thereby invoke <Socrates> in my interlocutor’s mind, allowing her to pick out Socrates, the individual qualified by the quality Socrates-ness. By this process, cases make particular objects qualified by given qualities distinguishable or identifiable by means of the concept from which they fall and enable us to communicate about them. The full semantic picture, then, could be represented thus:^[60]

The complexity here explains the difficulties encountered in our sources. There are a number of elements at play: the expression, the signified case, the concept, the quality, and the qualified object.^[61] Several of these (the case, the concept, and the quality) are conceived of in very specific ways by the Stoics, ways which may not be clear or intuitive to some of our sources. It is then not hard to see why some of them may have been confused, forgetting a step in the process or collapsing two elements. This reconstruction of the theory has the benefit of incorporating all the moving parts mentioned in the various reports, without the need to conflate entities with different metaphysical status or adding additional entities,^[62] while also explaining how they fit together.

3 Oblique Cases

One aspect is nonetheless yet to be addressed: my account has not so far made any mention of oblique cases. If case is, as I have suggested, at the level of both phōnē and lekta, then just as nouns can be used in several cases, there are, for any noun and associated quality and concept, several lektical cases expressible. While at the level of phōnē, case is marked by the ending of the noun, things cannot be quite so simple at the level of lekta. As we have established, cases are lekta and constituents of other lekta. The constituents of the proposition ‘Socrates walks’ are the predicate ‘walks’ and the (nominative) case ‘Socrates’, in line with Diogenes’ report that a middle or categoric proposition is composed of a nominative case and a predicate (κατηγορικὸν δέ ἐστι τὸ συνεστὸς ἐκ πτώσεως ὀρθῆς καὶ κατηγορήματος, DL VII.70). Similarly, we might think that in the less-than-predicate ‘loves Plato’, the constituents are the predicate ‘loves’ and the (accusative) case ‘Plato’. In both instances, the lektical case is the lekton expressed by the noun.^[63] If this is right, although all of the cases of a given noun will delineate one and the same species, signify one and the same quality, and refer to one and the same qualified object, a noun in different (phonetic) cases will nonetheless express different (lektical) cases and therefore contribute different content to different propositions.

From a purely mechanical perspective, the account and diagram above can easily accommodate this. When I utter a noun in the nominative, a nominative case will fall from the concept; when I utter a noun in the accusative, an accusative case will fall from the concept, etc.^[64] There is nothing in the account above to prevent a many-to-one relation of cases to qualities and species. Nonetheless, we ought to explain the need for different cases at the level of lekta and just how just how these lektical contents differ.

Here, Frede’s view is on the right track. He suggests that cases specify how the object “enters into the proposition”.^[65] Cases, according to Frede, represent the way in which the objects to which they relate are constitutive of truths. Frede gives the example of “a man hits a ball” and explains that the man and the ball are part of the truth in different ways: the ball on the receiving side and the man on the active side. Their differing cases in Greek (nominative for “man” and accusative for “ball”) reflect this. This is supported by the evidence. For example, Apollonius Dyscolus, a grammarian heavily influenced by Stoic philosophy of language, clearly describes cases as indicating certain relationships between subjects and objects:^[66]

Ὁμοίως ἐπὶ γενικὴν φέρεται καὶ ὅσα ἐπικράτειάν τινων σημαίνει εἰς τὴν τῶν ὑπερεχόντων ἢ κυριευόντων διάθεσιν κατὰ λόγον οὐκ ἀπίθανον. προφανὲς γάρ ἐστιν ὡς χωρὶς γενικῆς κτῆμα οὐκ ἔστιν ἐπινοῆσαι. διὰ τοῦτο τὰ κτητικὰ εἰς γενικὰς ἀναλύεται καὶ ἀπὸ γενικῶν παράγεται ἔν τε ὀνόμασιν ἔν τε ἀντωνυμίαις, ἁπάντων τῶν δυναμένων κτῆσιν ἀναδέξασθαι.

All [sc. verbs] which signify domination similarly govern the genitive to indicate the function of superiors or masters, according to a plausible account. For it is clear that possession cannot be conceived without the genitive. Because of this, possessives are analysed into genitives and derived from genitives in both pronouns and nouns, since they all designate things able to possess.

Ap. Dysc. S III.174 = GG 2.2.419.13-420.5^[67]

Ἔφαμεν ἐν τοῖς προκειμένοις τὰς μὲν πλαγίας πτώσεις ἀναφέρεσθαι ἐπ’ εὐθείας τῶν μεταξὺ ῥημάτων τὴν μὲν ἐνέργειαν ταῖς εὐθείαις ἀναπεμπόντων, τὸ δὲ πάθος ταῖς πλαγίαις, ὡς ‘Διονύσιος ἔτυψεν Θέωνα, ἐγώ σε ἐτίμησα’. (τὸ δὲ πάθος ἐγγενόμενον κατὰ τὰς πλαγίας μεθίστησιν αὐτὰς εἰς εὐθεῖαν, τῆς προκατειλεγμένης εὐθείας τρεπομένης εἰς γενικὴν μετὰ τῆς ὑπό προθέσεως, ‘ἐγώ σε ἔδειρα, σὺ ἐδάρης ὑπ’ ἐμοῦ’.)

We have said previously that oblique cases refer back to straight ones as middle verbs assign activity to direct cases and passivity to oblique cases, e.g. “Dionysius (nom.) hit Theon (acc.); I (nom.) honoured you (acc.)”. (The passivity which happens in the obliques changes them into straight ones, as the primitive nominative turns into the genitive with the preposition hupo (by): “I (nom.) thrashed you (acc.); you (nom.) were thrashed by me (gen.))”

Ap. Dysc. S II.141= GG 2.2.236.8-14

These texts confirm that, according to Apollonius, different cases are used to capture different relationships between the objects to which the words refer. A genitive is used for instances of possession and domination, which would not be rendered by an accusative or a dative. The nominative carries a notion of activity: it is used when the object is performing the action designated by the verb; while the obliques carry notions of passivity: they are on the receiving end of this action. If this attests to a Stoic view, and put in strictly Stoic terms, cases, by specifying syntactic relations at the level of lekta, render the ways in which their referents relate either to each other (e.g. as one possessing the other, in the case of a genitive) or to the action picked out by the verb (e.g. as the one acting or being acted upon).

While Frede is therefore on the right track here, this attractive aspect of his view in fact fits better with the theory as I have reconstructed it thus far than with the rest of his own reconstruction. On his account, case seems to encapsulate two very different things. On the one hand, it is equated with the quality itself. As such, it is a body, something in the world. On the other, it signifies the way in which the object it qualifies enters into the proposition. In this respect, it seems to serve a very important function at the level of the lekton, the truth-bearer. If the case is the quality, and nothing more than the quality, it is not clear how it can perform this second function. Does case vary so as to reflect the way in which the object qualified by it enters a proposition or functions as a constituent of a truth? It is surely not the case that an object has multiple qualities: nominative-Platoness, accusative-Platoness, vocative-Platoness, so that each can be signified depending on how Plato enters the proposition.^[68] Nor can the quality itself change depending on how Plato enters the proposition: Plato’s Plato-ness is not affected by whether Plato is teaching or being taught. The idea then must be that case both is (i) the quality – a body – and furthermore (ii) the specification of the way in which the object qualified by this quality enters the proposition.

This twofold conception is problematic for two reasons. Firstly, it seems odd that case would be the quality and the way in which the object enters the proposition.^[69] Secondly, this seems to require two oddly separate semantic processes. On the one hand, signifying a quality must be thought to be akin to a denoting sort of relationship. On the other, signifying the way in which the object enters the proposition can hardly be such a relationship. It seems to be a semantic relationship between the noun and an incorporeal meaning of some kind. That a noun would bear two such relationships may not seem entirely implausible if the two things to which it is related are somehow also related to each other. Yet, if we follow Frede, the quality and the way in which the object enters the proposition do not seem to be related in any such way; they in fact seem quite unconnected. This mismatch is problematic. Case appears to serve double duty and one would want the two roles to either be attributed to two distinct semantic entities, or to match more closely and work together in a tighter way.

On the view of case presented in the preceding sections, cases are lekta and constituents of propositions. As such, they contribute information to the propositions they compose. This propositional content can – and I suggest does – take the form of both a lektical counterpart, as it were, to its referent and syntactic information. Just as the noun “Socrates” in the nominative refers to Socrates and encodes syntactic information (its being a subject), the lektical case, both captures or represents in some way its referent, enabling reference to it in the way outlined in Section 2, and encodes syntactic information at the lektical level by specifying the way in which the elements of the lekton fit together.

This syntactic or structural specification at the level of lekta in turn goes hand in hand with the idea that lekta are in some way intended to reflect, capture or otherwise be representative of the world. As Bronowski notes, “ce qu’il y a à dire, c’est comment est structuré le réel”.^[70] On Bronowski’s view, language is not so much an assemblage of words supposed to mirror the assemblage of things in the world but an expression of structural elements. Lekta are in other words such as to capture the structure of the world. To do this, there must not only be components of lekta which relate to things in the world, but also components which capture the relationship between the things in the world which words pick out. A case does just that in both enabling reference thanks to its relationship to the relevant concept and encoding in the lekton syntactic information. This lektical syntactic information correlates to the structural relation of the objects captured by the lekton it constitutes. Just as the noun “Socrates” in the nominative marks its being the subject of the sentence, the lektical case ‘Socrates’ not only enables the noun that expresses it to refer to Socrates, it also marks it at the subject of the predicate expressed by the verb. This in turn reflects the structure of the world, namely the relation of Socrates to the property, activity, disposition, relation (etc.) captured by the predicate. On this account, cases are therefore key to lekta’s ability to capture reality in its structural complexity.

4 Cases and Other Parts of Speech

The account presented so far has focused on cases as expressed by nouns, and on the role of case in the semantics of nouns. Yet there is some evidence that some pronouns also express cases.^[71] Recall that we are told, for example, that a proposition of the form ‘this one is walking’ (οὗτος περιπατεῖ) is “one composed of a deictic nominative case and a predicate” (τὸ συνεστὼς ἐκ πτώσεως ὀρθῆς δεικτικῆς καὶ κατηγορήματος, DL VII.70). The demonstrative pronoun (houtos, “this one”) therefore here seems to express a (deictic) nominative case (ptōsis orthē deictikē). The matter is less clear when it comes to indefinite pronouns. We are told that propositions of the form ‘someone is walking’ are composed not of an indefinite case but of an “indefinite part or parts” (τὸ συνεστὸς ἐξ ἀορίστου μορίου ἢ ἀορίστων μορίων, DL VII.70).^[72] What an “indefinite part” is is unclear. “Part” here could be short for “part of speech”. Yet parts of speech are at the level of the signifying. An indefinite part of speech, that is, an indefinite pronoun, can help compose a sentence, and, together with a verb, express a proposition. It cannot compose a proposition, as the parts are said to do here. More likely, then, Diogenes means an indefinite part of a lekton. These indefinite lektical parts, as it were, are indefinite pronouns’ counterparts at the level of lekta. Now, the fact that they are called “parts” not “cases” might well indicate that indefinite pronouns do not in fact express cases, but other kinds of lekta.^[73] Nonetheless, we might think that the fact that both nouns and demonstrative pronouns express cases suggests that all inflected words, that is, all words with case at the level of phōnē, express cases at the level of lekta.^[74] This would especially make sense given the role I have suggested cases play in Section 3. If cases encode syntactic or structural information and thereby enable lekta to capture structural features of reality, they will be required to play this role for the lektical counterparts of all inflected words. In case that is right, I offer here some suggestions as to how the theory I have outlined might be extended to cases expressed by pronouns, if indeed there are some.

Let us start with deictic cases, since those are attested. What might distinguish a deictic case from other cases is not clear. One obvious difference is that a deictic case is one expressed with deixis. Yet if we do take seriously the idea that deictic cases are expressed by demonstrative pronouns, there must be more to it. Unlike nouns, pronouns are not said to signify qualities or delineate eidē. Demonstrative pronouns refer not by appealing to the qualities of their referent, but to its matter.^[75] The diagram presented above, in which cases are linked to objects in the world through species-concepts and qualities can therefore not apply to demonstrative pronouns. Deictic cases differ from cases expressed by nouns not merely in that they are expressed with deixis, but also in that they do not bear the same relationship to the world, namely one mediated by species-concepts. Certainly, to be a case (ptōsis, literally “fall”) at all, it must be thought to fall from something. But perhaps it does not fall from a species-concept, but from something else. In the absence of any evidence on the matter, what that might be must remain entirely speculative. I am inclined to think that perhaps all deictic cases fall from a single and unique concept, one associated with deixis or ostension and accessible to all competent speakers.^[76] If this is right, they will be easily accommodated in the account above.

Indefinite pronouns, in turn, might be thought to behave much as demonstrative pronouns. If the “indefinite parts” are in fact indefinite cases, then perhaps the idea is that indefinite pronouns express special cases, namely indefinite ones. Since, just as demonstrative pronouns, indefinite pronouns do not signify qualities, indefinite cases would not fall from a quality-representing concept. What they might fall from is once again a matter of speculation, though it seems plausible that such cases would fall from a very broad concept, such as the concept of being someone, or existing.

It is clear is that the cases expressed by both definite and indefinite pronouns – if there are such cases – cannot bear the same relationship to species-concepts as the cases expressed by nouns, but this needn’t undermine the account above. These cases do not link them to objects in the world in quite the same way as the cases expressed by nouns, by way of a species-concept and a quality. Nonetheless, it would seem that cases expressed by demonstrative pronouns play a similar role insofar as they are deictic and the deixis which accompanies the case is what enables those pronouns to refer. Discrepancies of this kind – in the exact role of cases in a word’s semantics – are therefore compatible with the theory presented above, which can accommodate variations in the details of the information encoded in the lekta constituted by different cases. In fact, much of the detail of the picture I have painted, for example that cases capture structural elements of the world by encoding syntactic information, will remain unchanged across all inflected parts of speech.

5 Conclusion

The Stoic semantic case, I have argued, is a lekton expressed by nouns and other inflected words. It plays a crucial part in the wonderfully complex semantic content and reference-fixing mechanism of the words that express it. Proper and common nouns refer to objects in the world by means of case, concept, quality (common or peculiar), and finally the qualified object which is the noun’s referent. Cases expressed by nouns fall from eidē. An eidos is a concept, a mental representation corresponding to a quality (common or peculiar) in the world. It is by appeal to this eidos that a qualified individual is referred to. Cases expressed by other parts of speech, such as pronouns, are not linked to qualities in the same way and, I have suggested, fall from other kinds of concepts, and play a similar part in reference-fixing. In addition, case is an essential component of complete lekta and most especially propositions. It encodes syntactic information correlating to structural elements of the world, such as relationships between objects in the world and between objects and actions. The Stoic theory of case was thus both extremely sophisticated and crucial to articulating their overall semantics.