Interart Narrative: (Un)Reliability and Ekphrasis

As defined in my previous work, (un)reliability is one of five types of hypothesis or integration mechanism, whereby readers account for textual incongruities. In principle, we may always appeal to alternative logics of resolution: ambiguous data may be attributed to generic convention, as in the detective story, and/or to functional drives, like surprise, no less than to the teller’s blind spots. What distinguishes the perspectival or the unreliability hypothesis is that it brings discordant elements into pattern by attributing them to the peculiarities of the speaker through whom the world is mediated. This links up with Meir Sternberg’s model of how fictional communication works through quotation.To infer unreliability is to make sense of the text at a quotational remove, with the mediator as ironized quotee set into the overall narrative for a purpose. And the ultimate quoter along the chain of transmission is the author, our frame-sharer and our normative reference-point for every intermediate viewpoint. To test and illustrate these relations anew, they are extended here beyond the traditional semiotic boundaries of language, so that mediation extends to the very medium. My test case is a dramatic monologue in which the narrator manifests (un)reliability through verbalization of a visual artwork (through ekphrasis, in short). Here the difficulties of translating between media or art forms intersect with those of managing a quote within a quote. Semiotic and perspectival complexities arise The first version of this article was presented in a session titled ‘‘The Unreliable Narrator Revisited’’ during the Society for the Study of Narrative Literature conference at Dartmouth College, April –May , . My special thanks to Monika Fludernik, who organized and chaired the session, and tomy two cospeakers, Dorrit Cohn and Bruno Zerweck.This version has benefited from Brian McHale’s detailed and suggestive comments. Poetics Today : (Winter ). Copyright ©  by the Porter Institute for Poetics and Semiotics. 712 Poetics Today 21:4 together. Furthermore, to quote or re-present an artwork in words is to frame one global act of communication within another.The perspectives involved in visual discourse—painter versus beholder—are thereby reset into the verbal discourse about the visual discourse. So, in ekphrasis, the quoting chain goes from literary author to narrator to plastic artist, with their respective ends, skills, vehicles, audiences, and at timeswith furthermediators en route. Ekphrastic speakers can therefore betray themselves—as this one does—in more ways than usual, against more diverse (e.g., aesthetic) norms and to richer ironic effects. In my example, Blake Morrison’s ‘‘Teeth’’ (appended below), the issues and disparities of point of view multiply even further. Due to a strategic allusion to Robert Browning’s ‘‘My Last Duchess,’’ the modern speaker unknowingly echoes the RenaissanceDuke, and theDuchess’s picture hovers behind the re-presentation of themodern artwork. In such light, readers infer that the monologist has similarly murdered his wife once we catch and decode the underlying intertextual cum intermedia allusion. This goes to show not only that (un)reliability hinges on reading but also that different orders of (un)reliability (e.g., between flat and fine judgments, as on the speaker here) correlate with different orders of reading. 1. Back to the Question of (Un)Reliability: Personal Trait or Inferential Resource? My title brackets two large issues, even fields, that may well appear unrelated. But I want to argue that they converge on perspective—above all, mediated or ‘‘quoted’’ viewpoint—with results that newly illuminate both, as well as drawing notice to a juncture of intrinsic significance.1Of the two issues, that long associated with narrative point of view (if anything, all too mechanically) comes first. Ever sinceWayne Booth () focused the concept, ‘‘(un)reliability’’ has been accepted by most scholars as a character-trait that attaches to the figure of the narrator (or reflector, hence all speaking/thinking subjects). It is then deemed one feature among others, narratorial and existential. Within this personifying model, to put it simply, narrators are reliable or unreliable as they are tall or short. Their being one or the other, moreover, depends on the rest of their features, that is, on the character type to which they allegedly belong as a whole. So reliability would necessarily go with omniscience and exteriority to the narrated world, unreliability with limited access to knowledge and intraworld existence. The first package is evidently modeled on the divine (‘‘Olympian’’), the second on the human condition, with the teller’s authority polarized to suit. My earlier work on narrative has repeatedly challenged this automatic . On a more personal note, an added bonus of this twinning is its suitability to the occasion of the Special Festschrift Issues in honor of Benjamin Harshav. He has himself recently explored, in another context, the literature/painting nexus (e.g., Harshav ). Yacobi • Interart Narrative 713 personifying model, laying bare its theoretical preconceptions and demonstrating its empirical inadequacy (to such baffling narrators as William Makepeace Thackeray’s showman of Vanity Fair, who is at times Olympian in knowledge and judgment, at times all too human). Instead, I developed there a communication-oriented approach that would seem to meet the varieties andworkings of narrative.Thereby (un)reliability turns froma trait attached to the narrator, among other data bound up with it, into an inference about the narrator, flexibly (even reversibly) responding to whatever textual givens and pressures we encounter as we read.2 But themisconception of (un)reliability as part of a stereotypical identikit has, alas, lately gained fresh currency, sometimes under new guises. It has even come to infect some theoretical work ostensibly aligned with mine, as if the two extremes were reconcilable.3 In the ensuing test case of interart storytelling, therefore, the need to revisit the issue also meets the opportunity to extend and ‘‘estrange’’ it beyond its traditional domain. Perspectival trouble is nomore specific to discourse in language than to literature proper and fiction. It rather doubles in transfer between the (e.g., visual and verbal) media, with their cumulative mediators (e.g., painter → author → speaker) on the way to the finished ekphrasis. Briefly, I define (un)reliability as a hypothesis that readers make and, if necessary, adjust to account for apparent incongruities in the text by reference to transmissional features: to the mediator’s (e.g., speaker’s, narrator’s, observer’s) high or low capabilities and performance. The oddities, we then infer, reflect somehidden design on the part of a reliablemediator or the selfbetrayal of the unreliable. Possibilities of mixture ensue, of course, to the detriment of packages again. In narrative fiction—itself, as will emerge, a variety of quotation—either reliability judgment also entails a well-defined relation, concord or discord, with what we take to be the implied author: the latter supports (‘‘authorizes’’) or betrays his creature via the imaginary telling. Throughout discourse, however, the emphasis falls not only on the coherence-making function of the judgment, but also on its interpretive, conjectural status. Moreover, since (un)reliability no longer figures amongnarratorial givens and attachments but among logics of resolution, it enjoys no priority over any competing logic. On encountering textual infelicities, we may in prin. SeeYacobi , , a, b, and forthcoming, with references to earlier and newer scholarship. Also contrast these untenable ‘‘package deals’’ with Sternberg : esp. ff.; a; : esp. –; : esp. –, to which my approach is related; or with Cohn , which brings it to bear on a Thomas Mann crux. . For examples, see Yacobi forthcoming and below. 714 Poetics Today 21:4 ciple appeal (or shift) to four equally universal ‘‘mechanisms of integration’’:4 . The geneticmechanism explains oddities in terms of the causal factors that produced the text and left theirmark on it—from a typographical error to the mishaps of sloppy writing. . The generic framework (for instance, comedy) dictates or enables a referential stylization, leading to divergences from the generally accepted picture of ‘‘real life.’’ . The existential mechanism reconciles incongruities by ascribing them to the norms of the represented world—whether a preexisting realitymodel, as in science fiction, or one peculiar to the work, as in Franz Kafka’s ‘‘Metamorphosis.’’ . The functional principle imposes order on the textual divergences in terms of the ends they serve.The work’s aesthetic, thematic, and persuasive goals invariably operate as amajor guideline for making sense of its peculiarities. The converse is also true: such peculiarities serve as pointers to the work’s goals and functional design. . Finally, the perspectival mechanism of integration, or the unreliability hypothesis, brings discordant as well as otherwise unrelated elements into pattern by attributing them to the peculiarities and circumstances of the observer through whom the world is refracted. The fictional teller, a creature of the author, provides a natural and notorious target for such attribution.This mediator’s deficiencies, we then go on to explain—for instance, incompetence, untruthfulness, lack of knowledge and understanding, distorted judgment—are manipulated for a variety of ends: rhetorical (notably irony of some kind), psychological, thematic, and so forth. So indeed, though by a reverse hypothesis of congruity and for equally diverse ends, is reliable narration, too. Instead of presuming, far less imposing, a link between the teller’s (un)reliability and his or hermakeup as awhole (typical) character, the reader then opts for this hypothesis wherever it does the integrative job best, or best on the reading in question. From the interpreter’s viewpoint, the five integration procedures are always available.On top of other advantages, principled or practical, this versatility explains and legitimizes interpretive disagreements within a com. To avoid confusion, let me point out that this term bears the all-inclusive sense given to it in Sternberg , embracing the whole repertory of patterning resources given to humans. (For the extent and variety of the repertory, see esp. ibid.: –.) Loose usages apart, compare the term’s more specialized meaning in Harshav : there ‘‘integration’’ is officially associated with ‘‘semantics,’’ precluding the appeal to genesis, for example. Yacobi • Interart Narrative 715 mon theoretical frame, even a common single mechanism. (For example, does a given odd twist betray the mediating twister’s unreliability or reliably connote some purpose?) This also disables anything like a fixed correspondence between trigger and mechanism. For the same textual signs of tension, the same deviances from whatever norm we bring to the reading (ideological, actional, psychological, grammatical), may in principle be resolved via any one or more logic(s) of integration. Thus, ambiguous data are attributable to generic conventions, as in the detective story, and/or to functional drives, such as the use of gaps for narrative interest, no less than to the teller’s blind spots or evasiveness. Even an outright lie, seemingly the starkest giveaway of unreliability, can elicit the opposite judgment: think of white lies, heroic lies, ironic lies, all possibly trust-inspiring beyond the domain of fact itself. The offense against truth (distortion, distraction, withholding, postponement) may a fortiori serve, or co-serve, generic and rhetorical goals. For instance, in Agatha Christie’s The Murder of Roger Ackroyd, the question of Dr. Sheppard’s (un)reliability as a narrator is not reducible either to his being the murderer or to his long suppression of the fact. In retrospect, with his two offenses sprung on us together, we enchain rather than equate and compound them. Although Sheppard is doubtless a criminal as an agent, how will the same ethical judgment fit the narrating-I, who decides to finish the tale, after all, and confess the crime hitherto suppressed from the record? He could, instead, destroy the document, or persist in denying his guilt. The confession—located, moreover, at the expected, conventional point, near the end of the detective sujet—complicates our judgment of his telling.The blame for unfair (unreliable?) play, if any, fallsmore onChristie than on Sheppard. If even a speaker’s lie may turn—in context—laudable or problematic or multipurpose, what hope remains for an inventory of constant (i.e., automatic) textual giveaways? As elsewhere, especially in literary art, the bid for package-dealing goes against reason.Hence the futility of itemizing ‘‘clues’’ to unreliable narration (Amoros : –), or the groundlessness of Ansgar Nünning’s () assuming the general applicability of the ‘‘indications or signals’’ in Kathleen Wall (), or of his embracing Monika Fludernik’s () ‘‘list of categories of expressivity and subjectivity’’ as ‘‘a list of grammatical signals of unreliability.’’ The very idea of an exhaustive or ‘‘systematic account of these signals’’ (Nünning : –) is misguided.5The Janus-faced working of various indicators that can be associated with perspective—from the traditional one of ‘‘knowledge’’ to the neglected one of . See also further references there, including a possibly misleading acknowledgment of my own early work. 716 Poetics Today 21:4 allusion—will run through the analysis below.Only within the poetics of an author, a school, or any analogous family can surface textual phenomena be expected to correlate ad hoc with dominant integration-mechanisms, perspectival and otherwise: even so, these do not always correlate, nor on all readings. Finally, the emphasis laid here on the reader’s inferential and integrative activity does not by any means rule out (as often among ‘‘reader response’’ proponents) the communication with the author. On the contrary, to hypothesize the narrator’s (un)reliability is also to form a complementary hypothesis about the implied authorial normative hierarchy that (dis)agrees with the narrator. Nor, and as importantly, does this emphasis privilege, or preclude, any specific communicative channel. The parties concerned and their affairs find equivalents in media other than language (e.g., visual art) or wider than language (e.g., interart transfers and composites), where terms such as reader, narrator, and author must be adjusted accordingly. Those two broadening thrusts, toward the communicative and toward the cross-medial, link up via Meir Sternberg’s theory of quotation as representation (; a; b; : esp. ff.; ; ). In the theory’s terms, the overall discourse of, say, the literary narrator counts (which would also mean reads and elicits judgment) as an inset, secondorder discourse. For, while the teller frames and quotes at will the speeches (and sometimes the thoughts) of the dramatis personae, he himself is in turn necessarily, if implicitly, quoted within the framing act of communication between author and reader. Similarly, of course, with local quotes across language use. But the same holds true in principle for all mediated discourse, regardless of code: a painting within a painting, or one attached to or invoked by a text, or vice versa, and so on, to the last frame/inset makeup. Accordingly, what Sternberg terms the ‘‘four universals of quotation’’ operate not only in the linguistic medium, nor only in the familiar and typically small-scale varieties there (direct, indirect, and free indirect discourse), but also govern the case of the implied author quoting the global narrator or their nonlinguistic equivalents. Let me briefly explain how the universal quartet both generalizes and focuses the points of contact—hence the sources of indeterminacy—between these two among other quotational parties. . In every instance of quotation, ‘‘representational (‘mimetic’) relations are established not between a discourse and an objective piece of reality . . . but between two discourse events. For the object of representation is here itself subjective, hence discoursive. . . . What the reporter . . . cites (to us) in the subject’s name belongs to one discourse; what that subject originally spoke (to his own addressee) or thought (to himself ) belongs to Yacobi • Interart Narrative 717 another’’ (Sternberg : ). Of all representations, in short, the quotational category has the peculiar feature of mediating, transmitting, representing another’s discourse. As such, it (along with its further universal peculiarities below) cuts across all re-presentational variables: in the magnitude, the existential status, the semiotic coding, or the overtness of either the re-presenting or the re-presented element. (For a fuller list of these ‘‘protean’’ variables, see Sternberg a: – and passim.) Notice that to ask whether the implied quoter endorses or disowns the message of the quotee is another version of the question of the narrator’s (un)reliability. . At the same time, ‘‘the two [discourse events] always assume the form of discourse within discourse. The (piece of ) discourse quoted becomes an inset within the frame of the reporting discourse’’ (Sternberg : ). In direct and indirect report, the quote often announces itself ‘‘with the help of an introductory clause or transformer’’; while the ‘‘dispensability of the transformer in the free indirect pattern’’—so that nothingmodulates, hence nothing visibly distinguishes, between frame and inset—is equally or even more open to other codes and scales of representation (ibid.: –). But the difference here lies between the presence and absence, between the salience and the withdrawal, of overtmodulating devices, rather than between modulation and nonmodulation. From the universal of quoted asmediated, re-presented discourse, it follows that we cannot think of quotation except by appeal to frame, inset, and transformer, with the latter (given or supplied by ourselves) intermediate between the two discourses. So any painting, like any written text, can visibly or allusively frame another. And in a fictional text, the discrepancy between the title page and the speaking itself—between the respective names (‘‘Call me Ishmael’’) or times (the date that precedes the Duke’s monologue in Robert Browning’s ‘‘My Last Duchess’’)— marks the transformation: it establishes the one as frame, the other as inset. To put it differently, the fictionality of the global act of speech means that this discourse, for all its seeming independence, is a quote, framed within the implicit authorial discourse. . ‘‘This new mode of existence [the original getting framed, becoming an inset] entails a communicative subordination of the part to the whole that encloses it.To quote is to re-contextualize, if not to re-textualize, hence to interfere with the original (con)text. All reporters, in short, subject the original to their own rules, needs, ends, indeed ‘framing’ it in the process’’ (Sternberg : ). This is at times more apparent, at times less. More apparent in indirect discourse, for example, where grammatical suggests deeper, communicative subordinacy, as well as the license to interfere with the original wording; or in intermedia quotation, with its open shift of code. Less apparent in global fictional narrative, because of its implicit and di718 Poetics Today 21:4 rect quoting. The narrator’s own (often even voiced) communicative aims and norms are silently but decisively subordinated to the authorial communicative frame. Like any other quotee (verbal, fictional, intramedium, or otherwise), the narrator is therefore exposed to diverse authorial manipulation, which, at its most extreme and discordant, reduces the creature to unreliability. . ‘‘Finally, the represented object being a subject with a perspective of his own, quotation involves a distinctive point of view: a perspectival montage between quoter and quotee, where it is sometimes easier and sometimes more difficult (if at all possible) to tell where the one’s viewpoint or subjectivity ends and the other’s begins. . . . Whether divergent or convergent, the same inset piece of language reflects a multiple point of view, at least a two-in-one, peculiar to reported discourse’’ (ibid: –). As a matter of fact, the perspectival montage between implied author and explicit narrator is the reader’s interpretive problem put in a nutshell: a discordant montage implies narrative unreliability, while concord of viewpoints implies reliability, always vis-à-vis the framing authorial norms as we reconstruct them.6 Also, here precisely (amid our overall attempts at disentangling the montage) enter the questions with which we began, that is, the status of the narrative features and their correlation with (un)reliability. Nowhere does this assignment and interplay of features show such dynamism as in fiction. Here the authorial figure’s power as highest-level quoter definitionally goes with omnipotence, omniscience, and omnireliability; these privileges are at will delegated or denied, partly or wholly, to the quotees, with the overall narrator first, against all package dealing (Sternberg : ff.; a: –; : –; Yacobi , a, forthcoming). However concordantor discordant-looking, for example, the delegation of omniscience to the narrator signifies overall reliability, or its denial unreliability, only in the light of the frame as a whole. Hence the narratological attempts to erase the implied author from the model on various grounds (whether the difficulties of reconstructing the implicit norms or the objections to the anthropomorphic figure of an ‘‘author’’) literallymake no sense.7 . Below I will illustrate how we work toward such resolutions or weigh the alternatives. For some general analysis of the problems, aids, and compensation systems involved, see my b. . For the latest call for dispensing with the implied author on such grounds, and in the context of the unreliable narrator, see Nünning : – (point ) and – (point ).There is also some irony in the discomfort of various scholars with the anthropomorphic metaphor of the creator as controller, responsible for all devices and effects. They have no problem with the analogous metaphor as applied to either the reader or the narrator—specifically including the anthropomorphism behind the prevalent notion of unreliability. Yacobi • Interart Narrative 719 2. Extending the Scope: Perspectival Relations in Ekphrasis To test and illustrate these relations anew, I would like to extend them beyond the traditional semiotic boundaries of language. Let me focus on a case even more intricate than usual, generically so. It entails both an object coded outside language and its recoding into linguistic discourse, so that the mediation in replay overtakes the very medium and newly tests the mediator (the concept of mediacy included). By this case I mean the discourse of ekphrasis, where the speaker (describer, narrator) manifests (un)reliability in and through a verbal representation of visual art, possibly of more than one work, style, even art. Here the difficulties of translating between media or art forms intersect with those of managing a quote within a quote. Semiotic and perspectival complexities arise together. Take a speaker who wants to render a painting or a Grecian urn or (in Book  of the Iliad ) a well-wrought shield. He then does not simply face the challenge of transforming a spatial object (e.g., a figure or a landscape) into a sequence of verbal signs, whose linearity presses for narrativity at that. Of course, this mismatch in extension between object and sign system poses quite a problem by itself, especially to a speaker intent on describing. But the problem doubles (at least) for every maker of ekphrasis, whose spatial object comes with, or in, a spatial signifying form. Distinctively, the ekphrastic maker as remaker must unfold a second-order representation of what the visual artist has already represented in his own medium. Homer recodes into a line of words the piece of world that Hephaestus has already visually (and, given the epic’s truth claim, historically) depicted on the shield; Keats poeticizes the scene allegedly pictured on the urn. Ekphrastic transposition is thus a compounded and accordingly tense re-presentation, an intermedia quotation in Sternberg’s sense. Furthermore, to quote or re-present an artwork in language is to frame one act of communication within another—resetting it into a new medial context on top of everything else, with duly enlarged options for play and interplay. The perspectives built into the visual discourse (painter versus beholder, e.g., the cunning Olympian smith as against the viewing Thetis, the mother of the recipient Achilles) are thereby translated and incorporated into the verbal discourse about that discourse. Larger and higher-level anyway than the inset part, the frame can expand (or lessen or otherwise modify) that part in re-presentation to suit its own ends as a whole, and now a verbal whole. For instance, in framing the various objects depicted on the shield, Homer’s teller need not recreate them across the media or convey how they did or would appear to the original beholder as such.Rather, in transferring the 720 Poetics Today 21:4 peace-time lawsuit held in the marketplace, he exploits his epic license of omniscience to specify (‘‘quote’’) the viewpoints of the parties or the judges. Nor are inside views out of his power. In larger-scale framing, he may also combine ( join, juxtapose) new perspectives on the visual inset with those adopted from the original by the visual inset to serve the more intricate, even revisionary, aims of the verbal whole. For example, the dialogue translated from Hephaestus’s lawsuit on the shield enables Homer to give his listeners an insight beyond the reach of any viewer—through a contrastive analogy between the peaceful resolution of disagreements on the shield and the bitter war for which this shield is being prepared in the epic. A vantage point, this, due to the setting of the imported within a higher viewpoint. But then, like any other reader-elevating effect, the contrast here assumes the reliability of the ekphrasis: the concord between the narrator who frames anew the scenes depicted on the shield and the epic poet himself. Replace (or even qualify) the concordwith difference, let alone discord, and the given literary frame’s viewpoint need no longer be higher than that of anyone else involved, as far as insight, aesthetics, or value scheme are concerned. The narrator at ekphrasis may then sink below the painter/viewer implied in the original, a fortiori below the author/reader judging the narratorial transfer. As all ekphrastic mediators re-present the visual according to their lights and for their own goals, they often reveal their subjective viewpoint in the process. And for such a revelation to arise, we must infer a still higher, authorial frame where the narrator’s discordant inset becomes a part: an interart within an intralinguistic quote, unreliable within reliable mediation of the visual.The re-presenter gets represented in turn, the original re-re-presented for our benefit. The range of voicing open to ekphrasis, from the authorized or objective to the untenably, ironically subjective, is illustrated in a scene from Muriel Spark’s novel Symposium. A BBC team televise an Anglican convent and are shown an unfinished mural that was painted by one of the sisters. What does it represent, they wonder. A group attempt at art interpretation naturally ensues, motivating a variety of figural ekphrases on the part of the dramatized beholders. From the frame above them all, moreover, Spark’s omniscient teller joins the fun in the role of immediate cross-art quoter— privileging us with her asides about the mural—as well as quoter, dramatizer, and orchestrator of the characters’ transfers. She smuggles in her own reliable (if enigmatic) re-presentations between the stereotypical guesses of the viewers from TV and the correctives by the nuns in the know (including the painter), so that the juxtapositions deepen the irony of subjective or projective ekphrastic exegesis. The team director, ‘‘avid for symbolism,’’ leads the way. She first identiYacobi • Interart Narrative 721 fies the ‘‘long, huge, antiquated monster, blowing clouds of smoke’’ (in the narrator’s objective re-presentation) as a dragon. No, the Deputy Superior responds, it’s a real ‘‘steamengine’’ (Spark : ). ‘‘Freudian,’’ then? (The symbol-lover changes ground.) ‘‘Freudian my arse,’’ the artistic Sister roars out.Yet the linkage which the would-be interpreters assume among the surroundings (a convent), the painter (a nun), and the topic (object, reality) of the mural will persist: it now elicits the conjecture that the figures depicted are saints. Even when they learn from a novice (in the artist’s name) that the train depicts ‘‘the scene at the railway station in St. Petersburg on  April , when . . . Lenin [was] met by a great crowd of comrades’’ (ibid.: ), they don’t give up their preconceptions of what a mural in a convent should represent. Instead, the woman at the head of the team comes back with the surmise that the ‘‘half-naked figure with the beard and the loincloth . . . near the refectory’s ceiling’’ ‘‘must be God’’ (ibid.). In this she has turned iconographist to salvage her typecasting of the artist and the artwork: the hypothesis now adds the logic of visual (and predictably religious) allusion to her constant rule of environmental decorum. As ‘‘Lenin was looking up at it with his raised arm, so that his finger touched the pointing finger of the beardedman,’’ the painter-nunmust in her opinion have alluded toMichelangelo’s Creation. A communist practicing and exhibited in a convent being inconceivable, the sacred world thus, hopefully, underlies the profane figuration hostile to it.Wrong guess again, since the allegorical figure represents ‘‘Karl Marx. . . . You must get your points of reference right’’ (ibid.). All along, the clash between the poor interpreters and the painter’s intention (as executed or explicated by those who know better) specifically focuses on the distance between automatic and open-minded ekphrasis. This focus promotes the author’s themes (art interpretation, art and its context, religion and/as political ideology, political and sacred art,TV stereotyping). But the staging and the diversity and the comedy of the visual exegesis suggest amore general lesson.The (un)reliability of these figural voices dramatizes how ekphrastic production is ameeting point between the viewing subject and the object viewed, and how the product verbalized as an inset stands or falls by its harmony with the frame that we reconstruct as the highest. Getting our ‘‘points of reference right,’’ though, often proves harder than in Spark’s convent, where the wrong ones are so transparently wrong, the right ones so vocal in the narrative as well as the dialogic frame, and their counterpointing so ironic. In short, in ekphrasis the quoting chain goes from literary author to narrator to plastic artist, with their respective norms, ends, skills, vehicles, audiences, and at times with further mediators (the parties to the lawsuit in Homer or to the decoding in Spark) en route. The issues of point of 722 Poetics Today 21:4 viewmultiply to match.8 I will now detail these factors, issues, and relations through an example that complicates them all and can accordingly test my approach—especially with respect to the perspectivizing of the speaker as an integrative mechanism. Not only does our reliability judgment of the ekphrastic (re)describer/teller hinge, as ever, on how one reads the whole text, or intertext, but different orders of (un)reliability (e.g., between flat and fine judgments) also correlate with different orders of reading. 3. An Ekphrastic Dramatic Monologue My case study, Blake Morrison’s ‘‘Teeth,’’ appeared in an illustrated anthology of ekphrases, With a Poet’s Eye ( []), all inspired by artworks exhibited in the Tate Gallery. The poem is a dramatic monologue, addressed by an art collector to a business associate visiting him. (See Appendix.) As literalized by the abrupt ‘‘And here’’ opening, the monologue starts where the guided tour of the house culminates—with the ekphrasis, in which the owner parades his treasure, ‘‘my Francis Bacon.’’ Echoing ‘‘the experts,’’ he points out some of the picture’s features in relation to the painter’s hallmarks: . . . Those sharp eyes And the gaoling of the figure in a boxroom Are as typical of Bacon as the scream.