Paul de Man’s “Semiology and Rhetoric”

A commentary


John Phillips


Form and Content

The main argument of de Man’s seminal essay can be stated as follows: The grounds of literary meaning (and by extension all meaning) must be located in rhetoric rather than in any of the other possible dimensions (form, content, reference, grammar, logic etc.).  But a rhetorical reading cannot guarantee authority over interpretations.  Therefore there is no authority that can guarantee a reading.  This doesn’t license us to read a text just anyway we want to.  Rather it commits us to readings that take full account of the possibilities and limits of reading (and writing) generally.  One name for these possibilities and limits might be deconstruction.


de Man begins by noting a decline in what he calls “formalist and intrinsic criticism.”  And he accounts for this by observing an increasing interest in reference amongst literary critics.  What is at stake?  By “formalist and intrinsic criticism” he designates a wide range of practices that we find dominating literary criticism throughout the middle of the twentieth century from the thirties and forties into the sixties.  Notice that his article is written in 1973.  So what distinguishes these practices?  The word formalism implies a rather conventional but nonetheless very powerful distinction (because it appeals to common sense) between form and content.  Those of us who have read our Ferdinand de Saussure know the distinction in terms of the difference between signifier (form) and signified (content).  How do you make the form your object?  To study the form of a work you study how it gives rise to its meaning.  Imagine we meet each other at breakfast and take turns at giving an account of the party we all attended the night before.  We will have a lot of different accounts of one event, a lot of forms for only one content.  In the same way anyone could have written a poem about school children dancing but only W. B. Yeats could have written “Among School Children.”  The poem is unique not because of its content—what it is about—but because of its form.  The “New Criticism” of the thirties and forties established certain techniques of close reading, especially in the work of its figurehead I. A. Richards, whose Principles of Literary Criticism is now a modern classic. 


Now Richards would perhaps have been surprised to hear his idea of form described in terms of the metaphor of inside and outside.  How does the metaphor work?  Imagine a nut.  A nut has a shell that, once removed, yields a nutritious centre.  This is what de Man means by the following statement: “when form is considered to be the external trappings of literary meaning or content, it seems superficial and expendable.”  The formalists, on the other hand, taught that it is the shell, rather than its content, that is important in literature.  So when de Man observes that the trend in literary criticism has moved from form to reference, what interests him is the underlying metaphor that governs how we have up until now always—without thinking about it too much—imagined meaning to come about.  That is, before we interpret a text we have already accepted an interpretation—based upon a metaphor—of what interpretation is.  It is this unwitting interpretation of interpretation that interests de Man.  He obviously has less concern about whether formalism, structuralism, historicism or author criticism is right or wrong.  Rather he is more interested in the unwitting assumptions that these approaches all share, i.e., the metaphor of inside and outside.  There is more at stake in this than you might have at first realized.  Think about it: most of us (but not all) will have had some experience in what we call close reading.  First year English students at NUS as well as some school students will already have learned to do what we call practical criticism (after I. A. Richards and his school).  This means that we read the texts according to literary forms like figures (metaphors, similes, symbols), narrative structures (first or third person narrators, point of view, character, plot, action, etc.), formal aspects of genre (meter, rhythm and rhyme) and themes (non-referential but thematic constants like death, love, the struggle of good and evil, etc.).  Here form is related to meaning “intrinsically” and no reference to the context of an outside world is necessary.  One might have asked, justifiably: “what is the purpose of it?”  Arguments about how the ability to evaluate a literary text is good for you, even at their most ingenious, ultimately fail to satisfy (and there have been many seemingly persuasive answers of this kind).  Undoubtedly this kind of knowledge counts as a skill and those of us who can do it derive a great deal of pleasure and satisfaction from it, but the question still remains—what good does it do?  How does it apply, if at all, beyond literature?


Perhaps then it would follow that criticism should start looking outside the text to the extra-textual world of real references.  George Orwell’s Animal Farm is a parodic critique of the communist revolution (and by association, all such revolutions).  Shakespeare’s King Lear is a not so subtle warning to King James (it was first played to him and his small court) not to lose his throne.  What we have come to understand as historicism develops as a way of extending the reach of our literary knowledge so that we can talk about its relation to historical events and processes.  This is what we might call extrinsic criticism.  The text now has its meaning located outside itself.  What fundamentally we are left with is a defining distinction—that is not itself fully explicable—between fiction or, more generally, rhetoric and reality.   An example of what often happens in literary criticism would bear this out.  A text by an Asian-American author like Russell Leong features characters who are migrant Chinese in the USA very often reflecting on and getting into situations of the kind Asian-Americans get into.  You might then want to argue that 1) the text in some sense translates the experience of the author; and 2) the text can be read as an engagement with actual situations that Asian-Americans find themselves in and, by extension, as a critique of ideological and historical conditions that help to determine those situations.            


So the rejection of “pure” formalism is not a rejection after all but a repetition that takes the form of a reversal: “The polarities of inside and outside have been reversed, but they are still the same polarities that are at play: internal meaning has become outside reference, and the outer form has become the intrinsic structure.”  The text is regarded either as something that has its meaning inherent in it (formalism) with no need to refer outwards to contexts or other texts, or it has its meaning outside itself, in the reference to author, period, history, social relation, reader or culture (etc.).  What all these approaches to texts share is the unwitting assumption that meaning can be understood on the model of inside and outside, whether the content is outside and the form inside or the form outside and the content inside.


At this stage in the article de Man provides a very important clue as to his approach.  He says he wants to avoid using the terms of the old metaphor (now we know that’s what it is) and instead relocate the problem of literary meaning by examining a couple of terms that, as he says, are “less likely to enter into chiasmic reversals.”  Chiasmus is a rhetorical term (from the Greek: Chiasmus, “a diagonal arrangement”) meaning the repetition of ideas in inverted order.  Shakespeare’s got a good one:

But O, what damned minutes tells he o’er
Who dotes, yet doubts; suspects, yet strong loves. (Othello 3.3)

So instead of this endless repetition of a powerful yet clearly awkward notion of interpretation and meaning, de Man gets his alternative terms “pragmatically from the observation of developments and debates in recent critical methodology.”  What’s he saying?  He will get his new explanation of reading from reading.  Notice that there is no attempt offered to formulate yet another original theory.  The “new” terms are “as old as the hills” and they are to be derived from current critical theory texts.


Semiology, Grammar and Rhetoric

He’s right of course to observe that his alternative terminology is “as old as the hills.”  What should be instructive is that it allows considerable rigor in his textual and theoretical analysis.  Notice, again, that he is not proposing a new theory.  He is analyzing a simultaneously theoretical and practical situation as he finds it.  It is simultaneously theoretical and practical because he refuses to read the theory as if it was a simple meta-language (a vocabulary to be used for discussing language).  He reads it as if it too needs reading.  This is how he was able to tease out the metaphor that lies unheeded at the grounds of most notions of meaning and interpretation.  And he deals with the problems of reading by reading texts that deal with the problems of reading (but which text doesn’t?). 


We don’t, I hope, have to spend too much time on the question of semiology.  Semiology establishes some basic tenets: the arbitrariness of the linguistic sign, the system of differences that gives the sign its value, and the conventional codes that operate as prompts for signification, sometimes making it seem rather culture bound. (What is it that frees language from cultural specificity?  The arbitrariness of the sign and its repeatability: ah, bold and italics, must be important).  Remember this: a sign does not simply refer to its referent (on the model of re-presentation).  A sign is coded according to its system and that’s how it comes to have its particular meanings.  Notice that in passing de Man observes that French writers (poets and novelists) seem always to have been aware of this, while only since structuralism have French critics twigged to it: a first definitive instance of the affirmation of the explanatory power of literature itself.  


Now, grammar.  After de Saussure, whose structural linguistics aims to derive general laws of language, the grammatical laws (which are as structural as anything) tended to become a rather privileged object of structuralist analysis.  A simple grammatical structure (sentence: noun phrase/verb phrase/noun phrase) can generate increasingly complex structures both at the level of the sentence and beyond to the paragraph, the chapter, the book even.  At the level of the sentence alone some complexity is possible.  See the first sentence from the paragraph of Proust (Wolfreys 336), which has four lines of phrases all generated from the model: noun phrase/verb phrase/noun phrase. 


In literary structuralism, especially in France, the analysis of deep grammatical structures went hand in hand with the analysis of rhetorical tropes (figures of discourse).  What this means is that the two axes of language, the syntagmatic (at the level of the generated sentence) and the paradigmatic (the axis of substitutions) can be read as operating together in a discourse.  We can thus explain what de Man means by “assimilations of rhetorical transformations or combinations to syntactical, grammatical patterns” with reference to the coexistence in structuralist theory of patterns of both metonymy (which is syntagmatic) and metaphor (which is paradigmatic).  The syntagmatic axis is composed of the marks (words to you and I) that we find (or put) together in a given text:


O chestnut-tree, great-rooted blossomer,

Are you the leaf, the blossom or the bole?

O body swayed to music, O brightening glance,

How can we know the dancer from the dance?


In this example, which I’ve stolen from de Man, all the elements that we find in the four lines are to be regarded as belonging together only syntagmatically—they are found together because that’s where they’ve been put.  When we think about what they mean, then we inevitably turn to the paradigmatic axis, which we cannot see because it belongs to the system (and not to the parole).  We cannot see it, that is, because it is the axis of possible substitutions (imagine I re-write Yeats’s verse: “O banana tree, little-rooted flourisher,” and you can see what kinds of substitutions are possible).  However to understand metaphor now no longer as just a kind of substitution but more as a kind of combination we find that a possible substitution is given in the third and fourth lines, where the question about the dancing body seems to be a kind of repetition of the question about the tree, thus making the dancer in some metaphorical sense equivalent to the tree.  Here, then, we have a metaphorical substitution on a metonymic axis.  de Man’s point is that we might in this way have chosen to include the metaphor within (and thus subordinate to) the grammatical, linear unfolding without acknowledging that there may be tensions between the two modes of signification in the discourse itself.  That is, the assimilation operates as a kind of smoothing over device to help us finish off the interpretation.


Remember: de Man deals with the problems of reading by reading texts that deal with the problems of reading.  Perhaps its not that obvious to us that “Among School Children” is a text about reading—but does it matter?  de Man can read it as if it was and certainly, then, it would seem to be.


So what is at stake?  The difference between metaphorical substitutions and metonymic combinations (rhetoric and grammar) can be seen as a kind of repetition of a deeper and older opposition: between rhetoric and logic.  But (a big, big but) metonymy is not a grammatical category.  It is no less figurative than metaphor.  The predicative structure of a sentence (noun/verb/phrase) cannot guarantee its meaning—as the example of Archie Bunker’s rhetorical question shows.  In that case, as de Man says, “the same grammatical pattern engenders two meanings that are mutually exclusive: the literal meaning asks for the concept (difference) whose existence is denied by the figurative meaning.”  The question, “what’s the difference?” actually means “there’s no difference.”  Now the point—as de Man points out in the next paragraph—is this: the only way out of the confusion engendered by this paradox is through an intention that cannot be reduced to the grammar of the statement.  What Archie Bunker means by the question is not contained by the question’s grammar.  And nor is it contained by any other verifiable aspect of the statement.  This is the meaning of rhetoric.  When the meaning of a statement cannot be established through an analysis of its grammar we call it rhetorical.  So when de Man says that, “rhetoric radically suspends logic and opens up vertiginous possibilities of referential aberration,” he is drawing attention to the fact that meaning (intentions people have when they make statements or when they read statements) cannot not be based on firm logical grounds.  Rhetoric is abyssal and aberrant.  You can hope to be understood but you cannot guarantee it.  Once we recognize that grammar is subordinate to rhetoric we are in the realm of interpretive decisions.  The structuralist dream of a fully analyzable language is now lost.  But there’s more at stake than that.  The logical grounds of interpretation have gone entirely—especially when we deal with the literary text, which is “above the norm” in rhetorical meaning.  Both logic and grammar are questionable when we read a literary text.  Grammar assumes a simple logical one-to-one relationship between language unit (word, sentence, etc.) and meaning.  Rhetoric contests that assumption.  Logic postulates the possibility of universal truth (a concept that independently of its objects remains unchangeable, eternal and unaffected by rhetoric).  We know from de Saussure that such a concept has no place in a system like the language system, which provides meaning only through the values that the differences between signifiers allow.  In other words, when we make meaningful statements we do so by acting on the combined resources of difference and rhetorical substitution.  This gives us considerable freedom but at a cost—we can no longer hope to control or to limit the structures of linguistic meaning and the multiple possibilities of confusion that always threaten.  But please pay attention to the implications of this last point.  If as readers of literature we can no longer guarantee a fully controllable text, then so long as we can show where these limitations reside—as de Man has done with his examples—we have won considerable interpretive freedom for our rhetorical readings.


Metaphor and Metonymy

It remains for me to say a few words about de Man’s reading of Proust.  He has chosen the example for a simple reason: it thematizes reading (“the most striking aspect of this passage is the juxtaposition of figural and meta-figural language”).  The role of the meta here is very, very, important.  When some faculty (language, consciousness, experience, thought) takes itself as its own topic or object we can identify a self-reflexive or auto-referential role.  Such a role always exhibits—in the form of paradox or contradiction—irrevocable limits to logical, formal or empirical analysis.  Ask me about this—there are many examples of the self-reflexive paradox and each of them can be revealing in different ways.  Now, in the case we have before us, the paradox reveals itself in two different ways.  First we have a meta-figurative discourse and, second, we have a meta-reader-ly discourse, which thematizes reading. 


First we have a passage of fiction (and figurative discourse), which thematizes the role of figurative discourse.  This is the text in its two dimensions overlapping.  The two dimensions of a text are as follows: it is composed first of what we might call its statement.  This is the level of content (whether considered extrinsic or intrinsic).  It is what the text is about.  But all texts are composed of a second dimension, that of their enunciation, the writing or speaking (the “how”) of the text.  In traditional terms this would be its form.  But in de Man’s “new” terminology form would not do, because the word suggests empirical and analyzable elements, and, as we’ve, seen this would miss the rhetorical aspects of meaning and intention.  Here instead of form we can talk about performance.  In this way we can actually make sense of the difference between Archie Bunker’s intention and his wife’s interpretation.  The subject of the statement changes when the subject of its enunciation changes.  The “image repertoire” that Roland Barthes writes about occurs at the level of enunciation.  When you read a text, the subject of (its) enunciation is you.  So reading is just as much a kind of performance as writing, which is why de Man maintains that the difference between literature and criticism is delusive.  (Student: “Are we doing criticism or literature?”  Teacher: “What’s the difference?”). 


The second way that a paradox of self-reflexivity is revealed is in the fact that a reader (Paul de Man) is reading a text in terms of the way it thematizes the problems of reading.  In this way de Man can read the text as rigorously as possible in terms of what the text itself—as a rhetorical entity—makes possible, even necessary.  In other words the text makes a claim (at the level of its statement) on behalf of the value of presence, according to which the most essential figurative tropes are metaphorical as opposed to metonymical.  But in the performance the text reveals a praxis (the Greek word for action or practice in the sense of something that one habitually does)—i.e., it achieves its effects—through metonymic combinations, which ground the metaphorical substitutions.  The metaphorical substitutions of the terms presence, essence, action, truth and beauty are grounded in a metonymic chain (i.e., they are brought together by proximal and thus accidental association).  What this does is to lessen—at the very least—the authority of the rhetorical mode.  But it doesn’t replace that authority with a new one.  Rather it opens up the space of reading as something that cannot be closed, that remains open, undetermined and exposed to chances of its future that no authority could determine or calculate in advance.  It does not do this after the fact but as the very possibility of its own mode of existence (as a rhetorical entity).  This is what de Man means when he points out that Proust’s text cannot simply be reduced to the mystical assertion of the superiority of metaphor over metonymy.  He writes:

The reading is not “our” reading, since it uses only the linguistic elements provided by the text itself; the distinction between the author and the reader is one of the false distinctions that the reading makes evident.  The deconstruction is not something we have added to the text but it constituted it in the first place.  A literary text simultaneously asserts and denies the authority of its own rhetorical mode, and by reading the text as we did we were only trying to come closer to being as rigorous a reader as the author had to be in order to write the sentence in the first place. (339).

Please pay special attention to the meaning of the word deconstruction in this passage.  It doesn’t matter what you want to say about writing because when you write the conditions and possibilities of writing alone determine the limits and possibilities of your statement.  And those conditions and possibilities are revealed when anybody writes about writing or reads a text in terms of the way it thematizes reading.  You could always make counter-factual claims about it but the writing itself would in each case reveal the lie.  So deconstruction is the name that de Man gives for the possibilities and limits of rhetoric (texts, statements and communicative events of all kinds). 


Where does it leave us?  After watching the new Spielberg production, AI, I have a fresh example.  Here is a cinema production that thematizes the relationships between cinema and its audiences.  In this sense it is a very clever film indeed as it is able to include a narrative about narratives (telling stories); the role of mass culture for individuals (the claims in the film are that it is fundamentally benign); the role of the spectator in making the illusion “real”; the persistence and permanence of cinema as a cultural product; (etc., etc.,).  It takes a spectator (like me), who is looking for the figure of the spectator in the film, to begin to see what is going on and, thus, to construct a critique—which I will leave in absentia here but we will come back to it anon.




Paul de Man’s “Semiology and Rhetoric” was first published in Diacritics, 3:3 (Fall 1973) 27-23.  You will also find it in Julian Wolfreys, ed. Literary Theories: A Reader and Guide. Edinburgh: Edinburgh University Press, 1999.