MICHAEL RIFFATERRE Undecidability as Hermeneutic Constraint One of the most striking developments in recent literary criticism, especially deconstructive criticism, has been the growing popularity of the concept of undecidability." thas always been a truism that literature is an oblique form of communi- cation that both demands a conscious effort of interpretation and hampers it, delaying it, or allowing for more than one interpretation, or preventing readers from preferring one over_another,-or-even-excluding any stable reading, any definitive solution to the problems it raises. Critics have traditionally isolated and catalogued devices aimed at jeopardizing interpret- ation: e.g. obscurity, defined as a block to the identification of the object the sign refers to (a block especially conspicuous in the case of symbolism); ‘anbiguity, the inability to decide among a finite number of alternative meanings having the same phonetic form; sylepsesthat force us to attribute more than one meaning at the same time to a given phonetic form, ete. But there is now a tendency to generalize from these phenomena, and to question the very validity of any interpretation, which is seen as a preconception, arbitrarily imposed on the text, not reflecting textual facts but inventing them (this is Stanley Fish’s position: ‘it is from the perspective of (the reader's} assumptions that the facts [of the text] are specified”, From this viewpoint, the interpretation of each segment of the text is deemed to deconstruct the interpretation of the previous onc. Ie seems to me that this view is untenable for two reasons. First because its advocates all share, perhaps unconsciously, a yearning for a quasi- mystical view of literary phenomena, an assumption that the essence of art™ ‘must elude analysis. Secondly, the same critics who assume that the text itself deconstructs the interpretations it elicits, thus generating a theoretically continuous and practically multiple series’ of readings shifting between textual aporia and metalinguistic contradictions, also seem unable to prove the mobility of self-erasing interpretations except by presupposing a relative immobility of the textual aporias that cause them. I will address only this second problem, it being the one that can be discussed objectively on the evidence of the texts themselves. Aporias are consistently identified withno Michael Riffaterre firmly localized passages, where th ages, where the problems arise from the lexicon i to the changing connections or funtions erties ty to make them A. The oe poe literary reading are paradosically derived from a ic concept of the sign, based unquestioningly on the same segmentation of the veal Sequence tha I suspet i alton fe leographers seems to me, onthe contrary, thatthe terarines: of literary texts doe not reside in the fact that they contain specifically literary signs which would emain sin any context (og tropes markers of irony, parody, ees ers indicating the genre to which the text belongs etc.), Ifsuch perma tent sins existed, we could nt explain why they eat be ebserved toe literary texts (in a sophisticated con r per Eich eee rae sera reinng to ergenrning eases al rg diferent vending Separate an succes te oe ae thm lng the writen Hes they eran whatcha al ate erer Eoonoution(@) out lapels Gioteine Ciba thar ote onvey within the Festritons dhe context may impore on our dscodng® ‘rom another, reti e Cit Sone Mere aoe ol 3s Took back and make comparisons and aa ecineane a liscover among these signs similaritics or relationshi aleaeot cnet - abacistnaepccicd sia orf te sal selon win the lesion Wie bee iF trdighe e rethscnpell to rane epson scent or our sal reaing the eames ac usage and that accounts for those feaures common to nonciterey ard literary texts alike, namely linguistic features. A different. segmentation Imps sel on sta permit alerting whchoubay for the discrete, successive meanings, first perceived and now found ineuE, HBStE souscceeieueeiel bf woe ee ee reading eft Together, these sgnincanes, Bam of Neigicwien generate the text's overall significance, the substance of its lierariness, il mit Tam proposing hs pie two tending or any eee One mn teva el pons te aga sequence: this is the literary reading. i oe ‘Reading rrnpecvely eno frat to Hen the new ier slant Thre srr ate er wy hat al this revision «radical eragure of ue flea Intspreaton od ten oak on th new sgh sem the aby meager ofs wo ae Ths ee it ncn lo ndg ot hey tae psa Tis ewe she Seen anes man ntnurs daa we experi: the st teatig.What aed way an extreme case of eifely as the very Enslty of the and sedccreee suggests. The corollary of this view i hat Ierarines votes ie cae Undecidability as Hermeneutic Constraint mm that cannot be cleared, an assumption I find hard to reconcile with any definition of significance. Tipropose instead to sce undecidability only as what sets the stage for'® second reading; the literary reading proper, as the first step towards recovers ing significance where meaning has failed It seems to me that undecidability derives this capability from the fact that the elements of the text it affects are undecidable only in terms of linguistic 'segmentatiof,, only in terms of @ mistaken decision on the reader's part as to what constitutes a relevant sign in the text. AS the reader’s focus changes through rereading, so does the segmentation, and the formerly irrelevant signs become operative, that is, ‘capable of significance, once they are reassembled to form new’ ad hae signs of which they become components at the expense of their own discrete meanings, and to whose structures they are now subordinate. I shall call these products of the literary scgmentation, these superordinate semiotic compounds, Aypersigns.? Tn accordance with this model, undecidability is the outcome of the linguistic segmentation of the text, and the literary segmentation the out- come of reader response to undecidability. This response therefore accounts for the common empirical perception of literary communication as richer, ‘more complex, more elusive and yet more demanding of our participation. ‘The paradoxical combination of clusive significance and of compelling effect results from the shift from one sign system to another. That shift is not, however, a solution or an end to undecidability, but rather a new focus pointing towards what the text is really about. This new focus, the literary Significance proper, controls and guides reader response, constituting indeed a hermencutic constraint that clearly informs our decision as to the signifi- tance and its aesthetic (and eventually ethical) values, It is perhaps no exaggeration to venture that this constraint is more powerful than that exerted by decidable signs, since it paradoxically inverts factors, forcing the tundecidable to represent decidability elsewhere, cither in an intertext or at a level different from that of the text (at the level of gente, for instance, or tropology). Three types may be distinguished: in the first, the undecidable is left entire but instead of pointing to a solution destined to remain ‘unattainable, it becomes a representation of undecidability, and itis as such, not as a content, that it produces literariness. In the second, undecidability becomes a periphrasis or an implication of decidability, the latter being located in an actual or potential intertext. In the third, undecidability teansfers significance from the text to intertextuality itself, rather than to an intertext. My example for the frst type is a familiar instance of ambiguity which Umberto Eco dusted off and proposed again a few years ago: ‘Charles makes love with his wife twice a week. So does John’.° There is no doubt that this is an undecidable utterance if this utterance is perceived through the successive decoding of its discrete components (several words, two sentences, the first one explicit, the second referring implicitly to the first), for the listener or reader cannot decide whether three or four people are112 Michael Riffaterre involved (two couples, or one couple and an interloper). But such a success: ive, piecemeal perception can never occur, because the whole utterance is perceived only as belonging to a genre expressing an intention on the speaker/writer’s part, as well as describing a situation. The genre is that of the joke (or the subgenre of the risqué joke), the intention a ribald or malicious one. It is made not only possible but necessary by the sociolect intertext: as Christine Brooke-Rose” has pointed out, no ambiguity would arise (in Eco’s terms) and therefore (in my terms), no pseudo or represented ambiguity would be resorted to, if the utterance were ‘Charles walks his dog twice a day. So does John.” There is no intertext at work in the case cof the dog. In the case of a wife, on the other hand, the word mife has a seme monagamy or availability to one person only in z love-making context, whether the seme is actualized positively or negatively, (as ele of brmiscvity). Representation leaves no doubt that the sentence is intended 10 suggest that Mrs Charles is having an affair. It is an unambiguous represen- tation of 2 jocular innuendo, Even if there was hanky panky going on, we would still have an unambigu ous representation of a make-believe ambiguity—one that does not leave any doubt as to the purpose of the sentence. This purpose is not just to hint but to activate the comical potential of a hint what makes a hint a verbal game, in other words, its literariness. It docs not matter whether the triangle hinted at is fact or fiction, The point or intention is to make light ofits implications, To that end, a comical variant of :he marriage descriptive system has been selected over an equally possible tragic version. Equally Possible but not equally treated in literature, let alone in the sociolect. Ever sinee medieval fabliux, the husband's cuckoldry or woman's supposed treachery have been favourite themes. This preference is so pervasive and constant, that the more stereotyped the mimesis of matrimony, the more expected and predictable the comicality. But.it is like a humorous bias in the statement of facts, not a factual statement per se. For comicality in representation remains uttafected whether or not tke statement is true or false. This comicality is therefore independent of reality, non-verbal context, or circumstances. 11 is a way of saying built on an indifferently hypothetical or verified disambiguation of an ambiguity. The ambiguity has been presented only to be eliminated along the line of irony or of the pessimistic, malevolent conventional wisdom of the sociolect — part and parcel of a cynical stercotype of woman in its various but always comical versions (the shrew, the liar, the hot spot, ete As a sentence the verbal sequence is referential and undecidable. As a text, it is selfsufficient and univocal.® And what gives the text its status as 4 unit of significance is the presence of the sociolect, of the intertext the relevance of which is activated by an essential seme of the sememe wife In the second type of undecidability as hermencutic constraint, the ambiguous text is generated by a repetition of a verbal expansion on an undecidable given or matrix. Every iterating sign points to an implicit statement which it refers to while repressing or displacing it into the intertext. In fact, these insistent ambiguities presuppose the intertext. In Undecidability as Hermeneutic Constraint 113 argc ms would seem to establish undecidability. Moreover it looks as if formal undecidabili ey and ethical mystery: what reasons may there be for “ presence of evil in se fhe part it plays in the scheme of things? He wee ad th pac py he he tO gh meets fs ofc thx pets 26 wel, Whether oF 2 to him: The Tyger ‘Tyger! Tyger! burning bright In the forests of the night What immortal hand or eye Could frame thy fearful symmetry In what distant deeps or skies Burnt the fire of thine eyes? On what wings dare he aspire? What the hand dare seize the fire? ‘And what shoulder, & what art, Could twist the sinews of thy heart? ‘And when thy heart began to beat, What dread hand? & what dread feet?