Aristotle and Augustine

1 Introduction

In his Preface to the Proceedings of the First Congress of the International Association of Semiotic Studies, Umberto Eco listed among our tasks to “provid[e] the discipline with a unified methodology and a unified objective” (IASS §1). Just five years later, “the multitude of topics and approaches” presented at the Second Congress led to the often-repeated question posed in the Closing Session: “Can semiotics be unified?” (IASS §2). By the Fifth Congress, titled Synthesis in Diversity, the organizers would suggest that “openness and tolerance” were inherent in the semiotic method and that “differences serve to strengthen rather than to divide” (IASS §5). The theme of the 2017 IASS Congress – Cross-Inter-Multi-Trans – also suggests a celebration of difference and, of course, I agree that differences can be liberating and productive. However, I would also assert that semiotics today is a very widely dispersed field with a great deal of activity on the frontiers but very little common ground in the center. What is most obviously and painfully absent from semiotics today is a common theorization of the sign.

I have addressed this problem once before, in a paper called “The difference between semiotics and semiology” (Daylight 2012), where I argued that Peirce and Saussure pursued such different phenomena that they could not even be described as two branches of the same field. In the essay presented here, I will work toward locating the origin of this problem, defined as the failure to produce a common theorization of the sign. My hope is that identifying the origin of the schism between the semiotic and semiological traditions will give us the tools to repair that wound and to move forward with “a unified methodology and a unified objective.”

2 Aristotle’s theory of language

Anyone seriously interested in semiotics is familiar with Aristotle’s On Interpretation and the six lines which have launched so much inquiry in the philosophy of language and signs. Here are the lines in Akrill’s translation:

Ἔστι μὲν οὖν τὰ ἐν τῇ φωνῇ τῶν ἐν τῇ ψυχῇ παθημάτων σύμβολα, καὶ τὰ γραφόμενα τῶν ἐν τῇ φωνῇ. καὶ ὥσπερ οὐδὲ γράμματα πᾶσι τὰ αὐτά, οὐδὲ φωναὶ αἱ αὐταί. ὧν μέντοι ταῦτα σημεῖα πρώτων, ταὐτὰ πᾶσι παθήματα τῆς ψυχῆς, καὶ ὧν ταῦτα ὁμοιώματα πράγματα ἤδη ταὐτά.

Now spoken sounds are symbols of affections in the soul, and written marks symbols of spoken sounds. And just as written marks are not the same for all men, neither are spoken sounds. But what these are in the first place signs of – affections of the soul – are the same for all; and what these affections are likenesses of – actual things – are also the same. (16a3–8)

These lines are dense in meaning, and assertive rather than argumentative in tone. Modrak observes that “this text is, on the most charitable reading, compressed and elliptical” (2001: 13), which must be how we like it, because these few lines “constitute the most influential text in the history of semantics” (Kretzmann 1972: 3). To unravel Aristotle’s claims, it helps to reverse the sequence as it is written. In the first place, at least ontologically, there are real things in the world (πράγματα) and those things exist in the same way for all men. Second, the affections^[1] (παθημάτων) made upon a human soul, psyche, or lifeforce (ψυχῇ) by this reality are also the same for all men, as they are an image or likeness (ὁμοιώματα) of reality. Third, the spoken sounds (φωνῇ), which are symbols (σύμβολα) or signs (σημεῖα) of these affections, are different for different people. And fourth, the same symbolic relationship exists between speech and writing (γραφόμενα).

Each of these statements is made as an assertion; Aristotle makes no attempt to justify or argue for the position. They are held to be self-evident: Ἔστι μὲν οὖν literally means “It is certainly true, then, that…” However, each of the four statements introduces interesting and important problems for semiotics to solve. And if semiotics has been trying to solve these problems for more than 2,000 years, then it demonstrates the power of Aristotle’s formulation.

2.4 Written words are symbols of spoken words

Aristotle’s final statement asserts that just as spoken sounds are symbols (σύμβολα) of affections in the soul, then written words are symbols of spoken sounds (16a3–4). In many ways this assertion is unproblematic: linguists today believe that the sounds we use are conventionally related to thoughts and that writing is conventionally related to sounds. Of course, this is to privilege alphabetic writing, which does not symbolize mental concepts directly, but indirectly via vocal sounds. Other writing systems exist such as hieroglyphics and Chinese which mix together the symbolization of ideas and sounds. Morse code and braille are alphabetic, while sign languages are generally not. Other kinds of graphic marks can symbolize ideas directly, such as ✞, ∞, €, and ♀. It might be possible to vocalize these graphic marks in English as “Christianity,” “infinity,” “Euro,” and “female,” respectively, but they operate primarily as direct symbolizations of ideas, independent of any natural language.

This privilege that Aristotle grants to alphabetic writing, and to spoken language over written language, was labelled “phonocentrism” by Jacques Derrida in a deep critique of Western philosophy (1967). Derrida argues that in works such as Plato’s Phaedrus and Aristotle’s On Interpretation, writing is externalized from the “natural” relationship between the voice and the mind. Writing represents, or reminds us of, the symbolic and conventional nature of language. Any philosophy seeking to substitute words for things in order to reveal the truth in the world is forced to externalize symbolism and internalize the “natural” relationship between things and their impression in the mind. All of which is why it ought to be interesting to a semiotician that Aristotle’s theory of language is generally reported as a triangle.

Exactly why Aristotle’s model of language should be presented as a triangle has always escaped me. For a start, there are four terms, so to produce a triangle the author is forced to omit writing completely, or place it to one side, like the handle of a saucepan. Sometimes the spoken and written are combined to produce something called a “sign-vehicle,” as in Nöth’s summary of Of Interpretation:

Aristotle’s definition of words as signs of the soul, and the latter as likelinesses of actual things, gives the outline of the standard order of the triad: (1) sign-vehicle – (2) sense – (3) referent (Fig. Si 2 below). Sense is the mediator of the referent. (1990: 89)

Nöth’s précis of Aristotle is so bizarrely inaccurate that it can only be explained by an excessive willingness to retrospectively fit Aristotle’s words into Peirce’s model, and to confirm the “standard order of the triad.”

The “semiotic triangle” has become so widely accepted that when Manetti sets out to demonstrate the longevity of the idea over the history of semiotics, he begins by “leaving aside the graphological level” (1993: 72). Manetti’s master concept of triangularity produces this (abbreviated) list of semiotic models (1996: 15):

Aristotle:	Symbolon	pathemata	pragmata
Stoics:	Semainon	lekton	tugchanon
St Augustine:	vox articulata	dicibile	res
Ockham:	Terminus	conceptus	res
Locke:	Name	nominal essence	thing
Peirce:	Representamen	immediate object	dynamical object
Ogden/Richards:	symbol	reference	referent
Saussure:	Significant	signifié	(outside language)

Whatever applicability the semiotic triangle has to Peirce, or to anyone else on that list, it is a very poor representation of Aristotle. As Kretzmann shows (1972: 4), a linear sequence of the four items is a far more accurate representation of the text. Aristotle’s theory of language reads unambiguously as a linear (and unidirectional) sequence, from things to writing:

Figure 1

Aristotle’s theory of language

This correction to the triangular account is more than pedantic, as we shall see in the next section. Rather, it forms the first element in a major reappraisal of our understanding of classical semiotics.

3 Aristotle’s theory of signs

As interesting and important as Of Interpretation is, it does not contain Aristotle’s theory of signs. Aristotle’s definition of σημεῖα appears in the Prior Analytics, also forming part of the Organon. A complete account of the place of signs in Aristotelian logic is not possible here, and indeed has probably not yet been written. But a brief account helps us to understand the meaning of “sign” for Aristotle, and how it might be compared or contrasted with his theory of language laid out in Of Interpretation. It is helpful to begin with Aristotle’s principal examples and then to work backwards toward a definition of each of the elements in his system of logic. In this passage, Aristotle lays out how logical proofs proceed through one of three possible figures [σχήματος]:

For instance, proving that a woman is pregnant because she has milk is from the first figure, for the middle term is having milk (let A stand for being pregnant, B having milk, C for a woman). But ‘The wise are good, for Pittakos was good’ is through the last figure. A stands for good, B stands for the wise, C stands for Pittakos. So it is true to predicate both A and B of C, except that people do not state the latter premise because they know it, though they do take the former. And ‘She is pregnant because she is pale’ is intended to be through the middle figure: for since paleness follows pregnant and also follows this woman, people think it has been proved that she is pregnant. A stands for pale, B for being pregnant, C stands for a woman. (70a13–23)

The proof that a woman is pregnant because she has milk proceeds through the first figure:

In Aristotle’s logical terminology, the demonstration proceeds by stating that some C have the property of B, and all B have the property of A, so if C has the property of B, then C also has the property of A. In ordinary language, if a woman has milk then she is pregnant. The entirety of this demonstration is called a syllogism, and the middle term is “having milk.” Aristotle defines the middle term as “that which both is contained in another and contains another in itself” (25b35).

Aristotle then provides an example of demonstration by the third figure, in which we assert that all wise men are good because Pittakos is good:

In the third figure, the demonstration proceeds by stating that all C have the property of A, and all C also have the property of B, so it is possible that B has the property of A. In ordinary language, if a man is wise, it is possible that he is good. Demonstrations in the third figure fail to establish a universal conclusion.

Aristotle’s demonstration by the second figure provides the proof that if a woman is pale then she may be pregnant:

In the second figure, the proof proceeds by stating that some C have the property of B, and some B have the property of A, so it is possible that if C has the property of A, then C also has the property of B. In ordinary language, if a woman is pale, it is possible that she is pregnant. Demonstrations in the second figure are always non-binding.

So now to the definition of sign:

A sign […] is a necessary or an accepted demonstrative premise. For whatever is such that if it is, a certain thing is, or if it happened earlier or later the thing in question would have happened, that is a sign of this thing’s happening or being. (70a, 6–7)

A sign is a demonstrative premise that establishes a positive or negative correspondence between a subject and a predicate at the current, past or future point in time. For example, The bus will not arrive connects the subject “bus” negatively to the future predication of “arrival.” As Smith phrases it, a sign “is just a statement predicating A of B, offered in support of the claim that C is predicated of B” (1989: 226).

Furthermore, a sign is one element of a syllogism: “If one premise alone is stated, then it is only a sign, but if the other premise is also taken in addition, it is a deduction [syllogism]” (70a24–26). Tredennick, in his translation of Aristotle, rephrases this as: “A sign may be regarded as a syllogism with one premise suppressed” (1938: 525). And this is, indeed, how we think about signs today, as a single term within a syllogism. For example, the notion that “there is smoke” is a demonstrative premise of a syllogism in the first figure which would prove that “there is fire.”

The final technical note is that if Aristotle uses the word “sign” (σημεῖον) for the demonstrative premise in a syllogism, he reserves the word “index” (τεκμήριον) for the demonstrative premise in proofs of the first figure, i.e. in which an irrefutable conclusion can be reached: “Signs may be classified as irrefutable (1^st figure) and refutable (2^nd and 3^rd figure), and the name ‘index’ may be attached to their middle terms, either in all figures or (more probably) only in the first” (Tredennick 1938: 326–327), “for the name ‘index’ is given to that which causes us to know” (70b1–5). A sign causes us to conjecture a conclusion but not prove it, whereas the index causes us to know, to have proof of a conclusion.

The standard translation of τεκμήριον as “index” is provocative but problematic. In Peircean terms, we would think of an index as that class of signs with a natural or causative relation with their objects. As such, both smoke (as a sign of fire) and spots (as a sign of measles) would be classified as indices. However, the class of sign that Aristotle calls τεκμήριον depends on a logical, rather than physical, relation. The proof that smoke is a sign of fire would be a demonstration by the first figure, and hence a necessary sign, or τεκμήριον. However, since other diseases can cause spots on the skin, the demonstration that spots are a sign of measles would proceed by the second figure (as paleness is a sign of pregnancy) and hence would be classified as an ordinary sign.

Aristotle’s complete theory of logic, as it is described in the Prior Analytics, includes: premises which are universal (belonging to all), particular (belonging to some) or indefinite; and premises which are positive or negative; and proofs in which the middle term can fall into the first, second, or third figures. All of this creates an extraordinarily rich logical system which remained productive until at least the nineteenth century. But for our purposes of making a history and critique of semiotics, we already have everything we need: a definition of the sign as a demonstrative premise in a syllogism.

4 Language and signs

Having examined Aristotle’s theory of language and his theory of signs, we are in a position to understand how the two systems interact. As we start, it’s important to remember that language and signs were treated as quite distinct phenomena in Greek philosophy:

It is necessary to recall that the semiotics of the Greeks, from the Corpus Hippocraticum to the Stoics, made a clear cut distinction between a theory of verbal language and a theory of signs. Signs (sēmeîa) are natural givens, which today we would call symptoms or indexes, and they entertain with that which they signify, or designate, a relation based on the mechanism of inference: if such a symptom, then such a sickness; if this one has milk, then birth has been given; if smoke, then fire. Words, however, stand in a different relation with the things they signify, or designate, and this relation is that which is sanctioned by the Aristotelian theory of definition. It is a relation of equivalence and of biconditionality. (Eco et al. 1989: 4).

Manetti goes further in furnishing a symbolic language for this comparison. The linguistic relation between words and things “is that of equivalence (p ≅ q)” (1993: xiii). Whereas the function of the sign “works from implication (p ⊃ q) rather than equivalence” (1993: xiv). Manetti intends “implication” in this case to mean “if p then q.” To this schema, Eco adds an example: “A red flag with a Hammer and Sickle is equivalent to Communism (p ≡ q). But if someone carries a red flag with a Hammer and Sickle, then that person is probably a Communist (p ⊃ q)” (1984: 18). Bringing these two domains together presents us with two systems of relationship which intersect at the point of the thing itself. In the axis of implication are logical relations, and in the axis of equivalence are linguistic relations:

Figure 2

Aristotle’s theories of language and signs

The linguistic relation is one of equivalence and substitution. In each case, the written word can be exchanged for the spoken word, then the spoken word exchanged for the impression upon the soul, then the impression upon the soul exchanged for the thing itself. Having effaced language, we are now in a position to understand the relationship between objects in the real world, which operate in a relationship of implication. The two domains operate in wholly independent realms, here represented as the vertical and horizontal axes, intersecting at the point of the thing itself. In Aristotelian logic, terms such as “pale” and “woman” operate on the same ontological level as a relationship of implication between different objects in the world. Conversely, Aristotelian linguistics operates on different ontological levels as a relationship of equivalence of the same object.

This much is uncontroversial: I might be the first to diagram this schema in this way but the contrast between signs-as-implication and language-as-equivalence is demonstrated by the great historians of classical semiotics such as Eco, Manetti, Marmo, and Tabarroni. But what follows from Aristotle’s schema is perhaps the most important and influential moment in the history of semiotics: the collapse of the linguistic and logical axes into a single theory through Augustinian semiotics. To quote Meier-Oeser: “Augustine’s doctrine also has to be seen as a decisive turning point in the history of semiotics” (2011 §2). And that is what we turn to now.

5 Augustine’s doctrine of signs

The first Latin scholars of signs appear in late antiquity, namely Augustine of Hippo (354–430 AD), and Boethius (480–528 AD). Their works on signs and language are written as commentaries on Aristotle’s Organon, in particular the introductory passages of Of Interpretation and the definition of sign that appears in the last part of Prior Analytics (Meier-Oeser 2011 §1). Although both authors use Aristotle as their primary source text, they are separated from Aristotle not just by 800 years, but by a major shift in context. From Augustine to the eighteenth century, semiotic enquiry was shaped by its theological orientation. Theorizing the interpretation of signs was a method toward understanding the word of God.

Unquestionably, Augustine’s most important contribution is the unification of language and signs:

With Augustine we reach for the first time an explicit fusion of the theory of the sign with the theory of language. Such a rigorous and important theoretical development remains unmatched for at least the following fifteen centuries. (Manetti 1993: 157)

Aristotle, as well as the Stoics and Epicureans who followed him, define the sign as a proposition leading toward a proof, or revealing knowledge. Augustine modifies this definition in a way that will come to revolutionize semiotics, and in ways still not well understood. He defines the sign as quod se ipsum sensui, et praeter se aliquid animo ostendit: ‘something that shows itself to the senses and something other than itself to the mind’ (De Dialectica V.9–10). In other words, the sign is a sensible which gives rise to an intelligible. This model allows smoke to be a sign of fire, but no longer in the classical manner. For Aristotle and the Stoics, all of smoke is a sign of all of fire: they are real entities at the same ontological level. For Augustine, by contrast, it is the sensible part of smoke which brings to mind the intelligible part of fire. As Augustine shows, this definition of sign also allows words to be considered signs: the spoken sound “tree” is a sensible which brings to mind the mental impression of a tree.

Figure 3

Aristotle and Augustine’s models of the sign

For the first time, the relationship of equivalence between words and their ideas is brought under the same theory of the logical or inferential relationship between smoke and fire, milk and pregnant, spots and measles:

For a sign is a thing which, over and above the impression it makes on the senses, causes something else to come into the mind as a consequence of itself: as when we see a footprint, we conclude that an animal whose footprint this is has passed by; and when we see smoke, we know that there is fire beneath; and when we hear the voice of a living man, we think of the feeling in his mind; and when the trumpet sounds, soldiers know that they are to advance or retreat, or do whatever else the state of the battle requires. (De Doctrina Christiana II.2)

One interesting aspect of the unification of language and signs is the immediate need for a taxonomy of signs. It is clear that sensibles must bring to mind intelligibles in different ways. In response, Augustine divides the totality of signs into natural signs (signa naturalia) and given sign (signa data), or signs produced with the intention to communicate. He further divides given signs into the sense which the sign stimulates – whether by sound, sight, taste, touch or smell – so that spoken words are given signs of sound. This division does not easily classify the calls of animals as intentional or natural, and hence Augustine’s classification highlights the problematic category of “animal sounds” – most notably latratus canis, the barking of dogs – which continues to provoke new taxonomies throughout the middle ages (cf. Eco et al 1989).

For Eco and the classical historians, Augustine’s redefinition of the sign is accepted as a progressive development. For the first time we have unified definition of the sign which can account for all sorts of different phenomena. Eco celebrates Augustine in saying that: “Fifteen centuries before Saussure, he will be the one to recognize the genus of signs, of which linguistic signs are a species” (1984: 33). Under Augustine, the sign is defined as a sensible phenomenon which brings to mind something else – an intelligible idea. But Manetti and Eco collectively take this unification further than Augustine intends by making all linguistic relations an implicative relation:

In earlier theories of language, the relationship between linguistic expressions and their content was conceived of as a relation of equivalence. […] In Augustine, the unification of the two perspectives is achieved at the level of the individual word without any need to resort to equivalence relations. (Manetti 1993: 159–160)

In other words, the linguistic relation becomes a relation of implication and causation: we are caused to think about trees when someone says the word “tree.” Eco et al. find a justification for this in Of Interpretation, insisting that Aristotle’s famous six lines “must be interpreted as follows: words are symbols of concepts, but the fact that words are uttered is an index, a proof or a symptom that there exist concepts in the soul of the utterer” (Eco et al. 1989: 6). The linguistic relation would be a species of sign under the genus of implication. However, it is quite clear from Augustine’s taxonomies that both logical and linguistic relations are considered species of the general category of sign, which is a sensible giving rise to an intelligible. Neither the logical (relation of implication) nor linguistic (relationship of equivalence) sign is privileged. Meier-Oeser, for one, demonstrates that it is not until Roger Bacon in the thirteenth century that intra-linguistic relations are considered implicative or causal signs in the sense suggested by Eco and Manetti (Meier-Oeser 2011 §4.2).

But if Manetti considers Augustine’s unification of the sign as a positive development, then the ongoing medieval focus (from Anselm to Ockham) on linguistic signs represents the slow unravelling of this advance:

The strength and importance of language mean that relations with other sign systems are inverted, and that language, rather than being a species, becomes a genus. Slowly but surely the model of the linguistic sign will end up as the semiotic model par excellence. (Manetti 1993: 160)

In other words, the fact that language is such an intuitively central and imposing part of the generation of meaning – particularly in regard to the interpretation of religious texts – allows language to gradually assume the primary place in the study of signs. This trend reaches its summit in Saussure, where all signs are expected to take on the model of language as the sign system par excellence. Eco’s corrective, expressed in works like Semiotics and the Philosophy of Language (1984), is that language should be considered merely as a subset – and a poor one at that – of general semiotics, defined as something which gives rise to (or stands for) something else. This is the position that was adopted by mainstream semiotics in the 1980s and remains the dominant position today, under the guise of the “stand-for” relation^[3].

6 Conclusion

Let us first summarize the historical steps toward our present situation. On Interpretation leaves us with a brief and unsatisfactory account of affections in the soul, and how they are a likeness of real things. To some extent, On the Soul provides an expansion of this assertion of what Whitaker calls “the theory of how thoughts match things” (1996: 14). However, one is left with the impression that Aristotle’s theory of language was far less developed than those of his contemporaries. For Aristotle, a sign is a demonstrative premise leading toward a proof. In contrast, the relationship between language and things is natural, but not semiotic. What Aristotle offers, in its essence, is a rich and productive system of logic underpinned by an anachronistically simplistic theory of language.

Eight centuries later, Augustine redefines the sign as a sensible which gives rise to an intelligible. For the first time, relations of implication (as between smoke and fire) and relations of equivalence and substitution (as with language) can be brought together under the same mechanism. It is a shame that much of Augustine’s subtlety is lost on contemporary semiotics, and in this regard Eco and Manetti’s interpretation – where the entire field of language, signs, and meaning becomes a relation of implication (something that stands for something else) – is decidedly unhelpful.

So what is to be done? My proposal in this paper is that Augustine’s unification of language and logic creates for semiotics an insurmountable problem of internal coherence. These fields were distinct for the Ancient Greeks and have grown further apart since then. The advent of symbolic logic in the nineteenth century meant that logic could finally untangle itself from the syntax of natural languages. And the revolution in linguistics through Saussure, Chomsky, and cognitive science has meant that there is virtually nothing left of classical models of language. Each field is developing according to its own needs and its own discoveries, and neither will be bound to an implicative theory of sign.

But if so, then where does that leave the promise of this paper: to work toward a common theorization of the sign? Peirce once claimed that “the entire universe … is perfused with signs, if it is not composed exclusively of signs” (CP 5.449). But if we say that the universe is composed of signs, then signs must exist prior to the universe, or at the very least, of our comprehension of the universe. In the same way, if language and logic are composed of signs then signs must pre-exist language and logic. But not only language and logic, but evolutionary biology, and communications theory, and every other field that semiotics is entangled with today. Our approach to signs must not depend on the preconceptions and categories of biology, psychology, or the philosophy of consciousness. The answer is that a unified semiotics is not the sum total of all inquiry into meaning making: of human culture, of the communication of bees, the interactions in cells, and the procedures of computer programs. Rather, it is the conditions which allow language and logic, evolutionary biology and communication theory, to appear.

Semiotics should not rely on the latest theory on the chemistry of the brain, nor on the latest trends in linguistics or communications. Biosemiotics and cognitive semiotics must not trail new discoveries in science but provide the foundation for them. If “signs” exist at all, then they must be primary: the basis of all other fields of inquiry which depend upon signs as the phenomena under investigation. The study of signs, therefore, needs to be situated among other fundamental questions, like that of being (ontology) and knowledge (epistemology). Indeed, semiotics will be situated prior to knowledge if we discover – as seems likely – that knowledge is built from signs. Only ontology – the study of existence itself – might rival the study of signs as a primary field of inquiry.

Figure 4

Primordial semiotics

Saussure’s genius, according to Benveniste (1969), was to take the disparate facts of language – the physical, psychological and social – and to distil from them the identity of the linguistic sign as a differential and psychological unit, at the same time arbitrary and socially constrained. Today, we need to do the same for signs. Out of the disparate facts of communication, meaning, stimulus and reaction – among humans and animals, in the interactions of cells and neurons, in the telecommunications of computer sub-programs, in gravity and quantum mechanics – we must distil the identity and unit of the sign. My essay “The Semiotic Abstraction” (Daylight 2017) presents one approach to this foundational task.