The third chapter in Augustine: A Collection of Critical Essays ed. R. A. Markus (New York: Doubleday, 1972) is titled “St. Augustine on Signs”, written by Markus himself (whom we introduced earlier, and whose book Saeculum we previously summarized). Originally appearing in Phronesis in 1957, it is a curious blend of historical and constructive work — beginning with the historical, then eventually threading in the constructive strands (highlighting roads Augustine suggested, but neglected to travel — I will largely ignore these in this post). The De Magistro (“The Teacher”), De Doctrina Christiana (“On Christian Teaching”), and De Trinitate (“On the Trinity”) are the three works of Augustine’s focused on by Markus here.
The Stoic-Epicurean Debate on Signs
Markus opens with a brief mention of the range of applications enjoyed by Augustine’s notion of signs (Scripture interpretation, models of human language, knowledge acquisition, miracles, and the world’s relation to God). Curiously, though Augustine’s “definition of sacramentum in terms of signum became classical” , his sign-theory is, later in the tradition, “rarely mentioned”  outside of discourse on the sacraments.
The essay then moves on to the development of Greco-Roman philosophy, specifically the Stoic-Epucurean debate, which apparently centered on the nature of signs.  Aristotle had earlier suggested in the Prior Analytics [II.27.70a7] that
[a]nything which involves in its being the being of something else, either at the same time or before or later, is a ‘sign’ of that thing or event. 
The “Aristotelian theory of signs as a means of inference sets the general framework”  for this debate. In both Stoicism and Epicureanism, it is by signs that we move from what is given to what is “non-apparent”. In Stoicism, signs are mental-intellectual: things are connected by rational necessity, so that signs are not mere tools or conventions, but relate to what they signify through the metaphysical mediation of a concept (a rational structure) applying to the signified.  By contrast, Epicureanism sees signs as sensible, and as established through association. “Simple empirical sequence is at the root of the sign-signified relation: a regular and observed sequence establishes the πρόληψις [prolepsis, translated to Latin in Cicero as “ante”, with the sense of the sign “falling” (lepsis) “before” (pro) the signified , as smoke before fire] which enables an inference to be made from one to the other, and such inference is valid only where there is a possibility of verification in sense.” 
The Stoic theory was faulted by figures like Sextus Empiricus for failing to account for being “exclusively propositional and inferential in character” , unable to account for instinctive and non-discursive sign-related engagements, such as animal calls. The Epicurean account could do this.
Augustine “can have had little first -hand knowledge”  of this debate, though it forms the background of his teaching on signs; the Stoic-Epicurean debate framed the “scraps of theory about them to be found in later writers, secular and ecclesiastic alike.”  Signum and σημεȋον have this significance when they’re used by later writers.
[T]hey can normally be translated by ‘evidence’, ‘symptom’, ‘portent’ –a ‘sign’ always allows something else to be inferred. 
Thus, Origen talks about miracles as “signs” and not merely “wonders” because “they direct the mind to their author and his meaning in bringing them about.”  Ecclesiastical writers built upon the sense of sign established from Aristotle onwards. In the scriptural exegesis of Hilary of Poitiers and Ambrose of Milan, signum “acquired a whole range of new resonances” , even if it remained within the classical scope, which was “primarily […] a theory of inference”. 
Curiously, in none of these cases was a theory of signs at the heart of a theory of language.  Aristotle, in his Categoriae VI, treated language under the aspect of discrete quantities (as opposed to continuous ones such as lines, where there are infinite points that are not clearly separated), because words have discrete beginnings and endings, with no overlap. Plotinus, in reply to Aristotle, states that “we may think of speech as action upon a substrate (air) and [passion] within that substrate.” [MacKenna, 513] For Plotinus, the discrete words are secondary to the act of speaking. Speech, for Plotinus, is essentially “meaningful action”.  It is quite possible that Augustine may have read Plotinus. For Augustine, words are “signs par excellence, and his theory of signs is meant to be, from the start, a theory of language as well as of other types of sign.”  The agency of speech is never forgotten in his theory.
The first part of Augustine’s dialogue with his son Adeodatus, the De Magistro (ca. A.D. 389), opens by asking why we use signs. “The purpose of all speaking, we are told, is ‘either to teach or to remind others or ourselves’ ” . This includes prayer, which was given in set forms to teach the apostles not words, but “realities by means of words”.  “[S]peech puts before the mind what was previously either altogether absent from it, or at least not present to it in the sense of being actually thought about.”  Signs are needed to direct the mind to things, and nothing can be learned without them.  What a sign signifies can only be known through other signs, whether synonyms, circumlocution, exemplifying or indicative gesture, or visual representation. Nothing can be shown directly except speech, and “speech itself consists of signs”, as Augustine writes.  Nothing can be known without signs.
In the second part of the dialogue, we read that signs do not always refer to other signs, but can refer to significabilia.  “In general we use words to talk about the things they stand for, in order to gain and to communicate knowledge about them”.  When a word’s meaning is not understood, it is not a sign to the hearer, but mere noise, and passes into significance when the significance of the sign is immediately shown. “Therefore, it is the sign that is learnt from the thing rather than the thing from the sign given” (De Magistro X.33).  “[E]xplanation must ultimately reach a point at which direct acquaintance with the significata of primitive words is presupposed.”  The signs do not give us the realities they signify; we must first learn the reality for a word to become a sign. This does not “contradict the conclusion of the first part of” De Magistro.
[Augustine’s] thesis is precisely that no knowledge can either be acquired or communicated on the basis of the account so far given: in order that I may know the meaning of signs, I have to know, in the last resort, the things they stand for. On the other hand, I have to rely on the words and signs of teachers to receive the direct experience of these things 
Words “bid us look for things.”  “The enquiry after the meaning of symbols is at the same time the enquiry into the reality they speak of”.  Signs and realities require the Logos to bridge them.
Human teachers, on the one hand, can only teach us the meaning of words and signs, and experience, on the other hand, only furnishes us with brute givenness. Only the Interior Teacher, which is Christ dwelling in the mind, can teach by at once displaying to the mind the reality to be known and providing the language for its understanding. He is the source of both the objects encountered and the light which illuminates them for our understanding. 
The dialogue’s conclusion is that “nothing external to the mind can, in the last resort, be regarded as the source of its knowledge.”  (Here is Markus’ first constructive thread. He is unhappy that Augustine does not follow up on his suggestion that some signs “might be signs of indication” , and so point to the signified. Christ as Logos, as Interior Teacher, bridges the “mutually external” things and signs; had this suggestion of signs-as-indicators been followed, however, signs could themselves have bridged the gap, Markus proposes.  However, as Augustine mentioned in his Retractationes, his goal in De Magistro had been to affirm that God alone is our teacher , so this was never likely to happen.)
De Doctrina Christiana
At the end of De Magistro, Augustine promised to return to the topic of “the utility of words” (XIV.46)  This he did in De Doctrina Christiana (Books I-III A.D. 397, Book IV A.D. 426). There, he writes that “all knowledge is of things or signs”, where things are “such things as are not used to signify other things”, though some things, such as Jacob’s stone, are both things and signs of other things.  Signs are also things (or else they’d be nothing), but not all things are signs. Some things, like words, are really only important insofar as they are used as signs.
A sign, then, […] is said to be ‘a thing which, in addition to what it is perceived to be by the senses, […] also brings something else to mind” (II.1.1). [73-74]
That is to say, in Markus’ paraphrase, that a “sign […] is an element in a situation in which three terms are related.”  These are (A) the sign, (B) the signified, and (C) the subject to whom this relationship holds. “Augustine appears to be the first to have stressed this triadic nature of the relation of ‘signifying’ “.  Previously significance was understood dyadically, as pertaining only to the (A) sign and (B) the signified, with “no stress” placed on (C) the subject or interpreter. But for Augustine, “[a] thing is a sign […] precisely in so far as it stands for something to somebody.”  (Here as elsewhere, Augustine stands on our side of a divide that other figures from Late Antiquity so often seem to fall on the far side of.)
Augustine distinguishes between two types of signs, signa naturalia and signa data. Markus calls signa naturalia “symptoms” , though they can be “portents” or “evidence”. It is not conventional. The sign can precede, concur with, or fall after, whatever it signifies. Tracks signify an animal had passed, measles indicate the presence of a complex sickness, certain types of weather can anticipate other meteorological events. Markus calls signa data “symbols”. Augustine defines them as “signs which living organisms make to each other in order to indicate, as far as they are able, what they feel or perceive or understand.” (II.2.3)  These signs are “brought forth” for the sole purpose of a mind communicating to another mind. They are the product of human activity, their meaning the product of sign-making intention, and their significance holds because of human convention. Words are perfect for this, as they are transparent to their meaning in these contexts, unlike gestures, which draw attention away from their meaning to themselves.  Borderline cases such as animal communication (symptom or symbol?), are not what Augustine is dealing with. (II.2.3) Markus proposes that Augustine holds to this symbol/symptom distinction too strictly, and advances that the origins of language are to be found in the “growth of awareness of the nexus of feeling […] and response and of the possibility of the response being reproduced voluntarily” , making the animal speech that Augustine brings up and then shelves quite central to any theory of language. The “by nature” and “by convention” split is, Markus says, conventional, though Augustine “applies this in a novel way”. 
Though he brought the bases of inference and speech-meaning conventions both under the heading of “signs”, his interest is really in the latter. Signa data are “exclusively conventional”.  Such conventions require “social solidarity” among sign users. Signs have no meaning apart from such conventions of use. (II.24.37) He does not ask how conventions became established in the first place. Markus charges Augustine with being “astonishingly blind to the extent that communities are created by the language they speak quite as much as they create it”. 
Augustine never returns to the dichotomy of De Doctrina Christiana, though he never repented of it. He does speak about the “word”, however, especially in the context of anthropological analogues of trinitarian life. In his De Trinitate (ca. A.D. 417), “Augustine uses the notion of the ‘word’ as a key-concept.”  The distinction of outer/inner continues in the utterance/word dichotomy: an utterance is only a word if the sound strikes the mind as having meaning. “The word heard sounding outside is the sign of the word which is luminous within, which is more appropriately called a ‘word’.” (XV.11.20)  The internal word is not though in “the likeness of any sound, and need not, therefore, be of any particular language; it precedes all the signs whereby it is signified and is begotten by the knowledge […] which remains in the mind, when that knowledge is expressed […] as it is (XV.11.20).”  The word within is complete, and “embodied in language”  only for communication. The conventions of spoken speech are “inessential” to the inner word. Markus proposes that, if we “think away […] what we say and hear, and think of it sheerly as meaning” , we can move toward what Augustine means by this. “[T]hinking is, like talking, something we do” , but our thoughts see truth; the “word” specific to each reality is unique to it. Curiously,
Augustine’s theory of the ‘word’ approaches language from the side of the speaker, unlike the sign-theories of the de Magistro and the de Doctrina Christiana.” 
As a result, the gulf between word and utterance is as great as that between sign and signified, if not greater.  There are two theories as there is a speaker-thinker and a hearer in any act of communication.  The bifurcation is relativized in that “[the] ‘word’ is begotten by the knower from the known”  in the species of what is known whenever we can speak about the subject knowing anything. Though verbum is like a concept, the more it is pressed it seems like a judgment.  The “material images which solicit [the mind’s] care and threaten to engulf it” — the mind is sovereign over these, “return[ing] to itself” from the sphere of its engagements: “the verbum mentis is […] a product of the judgement on the material presented by sense, imagination and memory in”  the light of eternity.
Here the light of the eternal truth dwelling in the mind does not shine upon the ‘word’ as upon something opaque and meaningless without its illumination. The ‘word’, in so far as it is anything at all, is meaningful; this illumination is required not to confer upon the ‘word’ its meaning, but rather to generate a verbum of a thing in so far as it is discerned and evaluated in this light. The light is, so to speak, constitutive of the verbum begotten in it.”  Illumination and the Interior Teacher are linked: “Where the Teacher interprets the meaning of signs, illumination […] creates the significance with which it endows its objects” , and in the end, both point “towards the one ineffable source of light”. 
There is an appendix of just over two pages on terms; the endnotes run nearly three.