Abstract -- Fludernik's voluminous monograph - a 500+ page Vienna Habilitation - presents a uniquely integrative analysis of literary and non-literary speech and thought representation around the focus of free indirect discourse (style indirect libre, erlebte Rede). Building on the seemingly incompatible models proposed in Pascal's The Dual Voice and Banfield's Unspeakable Sentences, Fludernik draws on pragmatic, frame-theoretical, and narratological postulates to arrive at a remarkable synthesis. Acting according to Wittgenstein's demand in the conclusion to the Tractatus (6.54), Fludernik throws away these ladders after she has ascended them. The paper assesses the methodological and practical consequences arising from Fludernik's argument.
1. Free indirect discourse. This is a subject that continues to be an inexhaustible generator of dissertations; the most recent crop includes von Roncador (1988), Ehrlich (1990), and Wiebe (1990). The current surge of interest can be traced back to the notable insights deriving from the somewhat belated Anglo-Saxon discovery of the subject (Pascal 1977, Cohn 1978, McHale 1978), but mainly also to the very stimulating literature/linguistics controversy arising over Banfield's (1982) Unspeakable Sentences. At the height of this controversy, McHale denounced Banfield's sentence-based generative-syntax approach as "wrong in principle, wrong in . . . orientation" (1983, 17). Although Fludernik starts out by setting herself exactly the same aim as Banfield - to analyze the linguistic forms of subjectivity - she has already charted up a whole-hearted endorsement from McHale, who calls her book no less than a "landmark contribution to the poetics of narrative" (Fludernik 1993b, i). (All subsequent references to this text will be given by page number only).
To call this book "another study of free indirect discourse" (3), as the author does at one point (albeit ironically), is a notable understatement, because it covers much more. As has become increasingly clear, a study on free indirect discourse alone would be patently pointless. Taken on its own, it is an elusive phenomenon which has neither a necessary nor a sufficient formal characteristic. All things being equal (which, of course, they rarely are), Oh, she simply hated her daughter! can as easily be a sentence of narrative report as She was tired can be a piece of free indirect discourse.[1] This is the reason why Culler argues that free indirect discourse is not even "a definable linguistic category" (1978, 612). Its true identity only appears in its pragmatic context, when it, as Fludernik nicely puts it, "materializes in the reading process" (441). As a "reportative" or "quotational" device, it must also be seen against the background of and in its interaction with other types of speech and thought representation.
Fludernik actually surpasses the latter requirement because she analyzes free indirect discourse in relation to all known quotational contexts and techniques, from narrative reports of speech acts and psychonarration to coloured narration, ironic mentionings, hidden quotations, and embedded formulaic phrases. Ultimately, her aim is to cover "as much variety as possible from a generic, historical, narratological, ethnic and gendered point of view, filling, if possible, lacunae left by previous research" (10). At the same time, rather than begin ab ovo and retread familiar ground, Fludernik sensibly decides to leave certain historical periods, authors, and text types un(der)represented because their features and effects have been treated in recent studies. Thus there are few citations from Woolf, Joyce, and Dos Passos; and what many theorists consider to be the standard case (if not the entire extent) of free indirect discourse - third-person, past tense, free indirect thought - is also presumed to have been accounted for. This focus on variety and avoidance of well-trodden paths is an indication of Fludernik's pronounced anti-essentialist stance, which directly opposes any system of hermetically sealed "standard" or "ideal" categories (73).
In presenting her evidence, Fludernik does more than what could reasonably be considered her homework. She adduces literary cases of free indirect discourse in English from some 280 titles, ranging from The Canterbury Tales to postmodernist fiction. To these are added conversational examples culled from UCL's Survey of English Usage corpus, excerpts of journalistic prose from the New York Review of Books, a considerable number of literary and nonliterary German and French citations, and there are even occasional excursions into Russian and Japanese. Whenever a topic requires it, the author reviews the relevant philosophical, linguistic, narratological, and stylistic literature. Perhaps the most graphic indication of her erudition and thoroughness is the number of references in the comprehensive "criticism" bibliography, which contains no fewer than 1250 titles, remarkable by anybody's standards. Yet not once does one have the feeling that the sheer scale of information management overtaxes the author.
Fludernik documents free indirect discourse for first and third-person texts (carefully differentiating authorial and "reflector mode" texts), but, for good measure, she also throws in some cases of second person free indirect discourse. Free indirect discourse is analyzed in its past and present tense realizations in English, its imparfait guise in French, and its preterite and "free subjunctive" forms in German. The technique varies in its representation of observable and nonobservable objects (spoken utterances/written texts vs. thoughts/perceptions) and allows grading according to degrees of "expressivity" (266) and "reflectivity" (378). To some extent, it expresses the reporter's concordant (empathetic), discordant (ironical), or neutral attitude.
Various attractions and restrictions show up under this fine-grained featural grid. For instance, early (i.e., 18th century) German cases split up in a sort of complementary distribution, in which the plain free indirect forms represent thought, and the free subjunctive forms represent speech (96-7, 148). In contrast, English free indirect discourse between the Renaissance and Aphra Behn is exclusively used for the representation of speech (96). Unlike the English form, German free indirect discourse apparently cannot integrate so-called discourse parentheticals; thus a close (i.e., device-preserving) translation of Where were her paints, she wondered? is apparently impossible (196). (Wo waren ihre Farben, fragte sie sich reduces the scope of the question, transforming the discourse parenthetical into a narrative parenthetical). Free indirect discourse goes well with discordant narratorial irony, but representations of observables do not seem to co-occur with a concordant narratorial attitude (312).
This very cursory list of attractions, exclusions, parallels and nonparallels naturally suggests a systematic combinatorial table of feature collocations - one immediately envisions a feature matrix filled with pluses, minuses, and zeroes. Admittedly, the resultant combinatorial explosion would probably be unmanageable; and on the whole one is grateful that Fludernik never asks such burning questions as whether nineteenth-century present-tense German discordant authorial third-person free indirect discourse does or does not exist (and if not, why not). Of course, no-one can say whether this or some other hitherto wholly unremarkable combination may not one day turn out to be an important catalyst, change a paradigm, make an ideal type, or falsify a hypothesis. However, there is some consolation in the thought that, if all else fails, one can always fall back on the combinatorial game to generate more grist for the mills of future dissertations.
2. Imaginary and perceived discourse. An area which does deserve more immediate and detailed attention, but, unfortunately, is only touched on in passing by Fludernik, is that of representational embedding, especially in that intriguing guise of perceived discourse. Although the German term erlebte Rede has often been acknowledged to be an awkward misnomer (Pascal 1977, 30; Füger 1993, 49), it is worth remembering that one of its original objective correlatives is heard (experienced, perceived) speech (von Roncador 1988, 238). Fludernik is generally aware of "perception interpretations" (203, 255, 268-9, 312); she even has a whole chapter (5.5) on "narrated perception." Unfortunately, this does not deal with heard speech at all. Elsewhere, heard speech is usually only mentioned when Fludernik criticizes Banfield for using it as an excuse to evade the issue of non-literary, conversational free indirect discourse (381). By way of refutation, Fludernik presents many examples of reportative (quotational) free indirect speech, especially from her conversational corpus. However, letting the matter rest there effectively sidesteps the issue. The main problem is that, unlike "ordinary" free indirect discourse, perceived discourse does not conform to a quotational or reportative framework. As Banfield notes, and Fludernik duly acknowledges (140-41), free indirect heard speech often uses referential expressions where plain free indirect discourse requires pronouns. Heard speech challenges the partitioning of a text into narration-of-speech and narration-of-events. In true test-case fashion, it also undermines Fludernik's basic assumption that the perspectival orientation of represented speech tokens is "necessarily outside" (136) - clearly, things can be heard in the mind in the absence of any observable signal. By contrast, treating perceived discourse as systematically independent would not only allow a non-reductive approach to the phenomenon itself, it would also create a significant stepping stone towards an analysis of imaginary speech and thought, which, so-far, has been woefully neglected.
3. The deictic centre. This is Fludernik's basic term to refer to a linguistic reference point (going back to Bühler's origo) for elements such as pronouns, proximal and distal adverbs (here, there, now, then), modals (presumably), and tenses. In a more general sense, a deictic centre also serves as an anchoring point for subjective expressions and constructions, in which function it is usually identified as a subject of consciousness, a self (Banfield), a focalizer, a reflector, a narrator, or simply a being capable of having and expressing a point of view (450).
The bulk of the linguistic evidence on free indirect discourse and related techniques of speech and thought representation appears in Fludernik's chapters 2, 3, and 4. Chapter 2 provides an introductory "guidebook on free indirect discourse" (82). Chapter 3 analyzes the "shifted" character of indirect and free indirect discourse, and documents person, tense, and mood aspects as controlled by a narrating or reporting deictic centre. Chapter 4 focuses on the linguistic and rhetorical means of creating the reported deictic centre and draws up a comprehensive inventory of subjective expressions and constructions. It is generally shown that, with very few exceptions (such as the subjectless imperative), most emphatic constructions (e.g., inversions) and subjective expressions found in direct discourse also occur in free indirect discourse and indirect discourse. Fludernik also shows how a reported deictic centre's subjective expressions and constructions can even "colour" or "infect" the diegetic parts of a narrative. This ability is responsible for a plethora of intermediate cases and the blurring of the usual, seemingly clear-cut distinctions. Fludernik's plausible solution is to replace the traditional categories by a scale of forms (see para 5, below).
4. Banfield's ladder. There is no room here to do full justice to Fludernik's very detailed linguistic findings. However, it is interesting to note that her investigation begins by taking the same "revolutionary" (83) point of departure as Banfield, namely the desire to "locate expressivity in grammar" (440). A very basic summary of her central linguistic chapters would be to say that they are an extended contrafact to the main contentions put forward in Banfield. Of course, this continues a well-established tradition; indeed, rarely has a theorist been bombarded with such vast numbers of counterexamples as Banfield, who, among many other oddities, claimed narratorlessness for heterodiegetic texts, banned the bulk of subjective expressions from indirect discourse and the diegetic parts of the narrative, denied the existence of conversational and present tense free indirect discourse, and was especially prone to making highly suspect ungrammaticality judgments (Jahn 1992, 350). What becomes particularly clear through Fludernik's counterevidence is that, directly or indirectly, all of Banfield's suspect postulates are attributable to biased data selection (with a strong predominance of reflector mode texts) and to her rigorous generativist essentialism (whose "Extended Standard Theory" basis has by now become hopelessly dated) (365). As is shown very convincingly by Fludernik, the moment the data base is extended to include authorial and postmodernist narratives as well as conversational data, a host of counterexamples is available to anyone who "starts looking for them" (84). Fludernik's broad data base also rehabilitates many sentences that Banfield predicted as non-occurring or condemned as ungrammatical.
5. Fludernik's scale. Fludernik eventually rejects the canonical "tripartite model" of indirect, free indirect, and direct discourse, and in its stead proposes a "scale of forms" (280). The scalar arrangement counteracts the inherent normativism and reductivism of ideal types, which often lead to "exclusionary taxonomics" (282). In a framework of ideal types, so-called marginal cases are usually treated as inessential exceptions, misfits, and ungrammaticalities - indeed, this is very much the treatment they get in Banfield. Closer reflection shows that an exception does not exist per se, but is actually the counterpart of the ideal case. Conversely, a continuous scale can accomodate a multiplicity of major and minor categories and is especially suitable for placing intermediary cases. It is for this reason that scales have become very popular in literary theory, and there are a number of well-known scales grading speech and thought representations such as the ones introduced in Chatman (1978), McHale (1978), Lanser (1981), and Leech/Short (1981).
Fludernik's own scale arrays six major categories in the following order: pure narrative - narrated perception - speech report and psychonarration - free indirect discourse - indirect discourse - direct discourse. To these, Fludernik adds 24 linking categories which are defined as being "intermediary between" any two (and sometimes more) of the six major categories. Unusually, free indirect discourse does not appear in its traditional position, viz., halfway between indirect and direct discourse. The reason for this lies in Fludernik's frank admission that "there is no stringent order in the arrangement of the scale" (309), a fact that is largely due to the circumstance that neither mimesis nor diegesis are "unitary" elements (314), and that speech representation offers a different grading from thought and perception representation (311). It does not matter, therefore, where free indirect discourse is placed, because its position is, to put it bluntly, as arbitrary as that of the other categories. Technically speaking, this reduces Fludernik's scale to a nominal scale, i.e., a basically unordered collection of categories.[2]
These scalar deficiencies signal the theorist's dilemma: s/he can either essentialize the range of phenomena to a system of ideal cases (and lose a non-biased overview), or accept it in its full variety (and lose its ordinal scalability). Methodologically, it is nevertheless a salutary and creditable move to acknowledge that the phenomena under discussion are so complex and closely interrelated that they neither conform to a simple ordinal ranking nor fall into watertight categories. Consciously opting for the second horn of the dilemma as the lesser evil, Fludernik's scale (even if it should perhaps more appropriately be called a network of interrelated concepts) currently offers the most plausible charting of the data, and is a mandatory starting point for future discussion and research.
6. Dual voice. Unlike Banfield's "unspeakable sentences," Pascal's (1977) "dual voice hypothesis" seems better armed against falsification and counterexamples because it builds on the seemingly natural metaphor of textual voice. (Lest it be misread, this statement raises two challenges, viz., that of being difficult to falsify and that of its metaphorical origin.) According to Pascal, the narrator's and a character's voice "mingle" in free indirect discourse, producing a "mixed" language, a "fusion of two voices" (1977, 12). Interestingly, it is a mingling, a mixture, a fusion, in which for some unexplained reason the voices are not equally audible: whereas the character's voice can run the whole gamut of expressivity, the narrator's voice always remains significantly muted, often to the point of being dubbed "implicit," as in the case of irony (Pascal 1977, 26). Pascal unfailingly regards it as a fault whenever a narrator's voice becomes too noticeable and begins to contest the character's voice - never the other way round (cp. Pascal 1977, 88, 95, 97). Moreover, leaving all pretense to a tertium comparationis, Pascal often takes narratorial "voice" to be a metonymical abbreviation for attitude, style, diction, irony, or something even more abstract such as naked "narratorial presence" (1977, 138).
The slippery slope into the dual voice fly-trap becomes even more apparent if one assumes a deliberately negative mind-set, a casuistic attitude of refusing to play along. Very pedantically speaking, it makes little sense to assign textual material to a voice, let alone several voices. Readers do not normally voice or mouth the texts they read (unless they read them aloud, which may be, but probably is not, comparable to silent reading). Finally (still arguing in a spirit of deliberate contrariness), properly appreciated, an "implicit" voice in a written text amounts to a blatant contradiction, something comparable to, say, the idea of an italicized subtext. On the whole, this is evidently one of those situations in which the opposite position carries at least as much validity as the original hypothesis, and this is never a good thing. Whatever one may think of Banfield's views, her radical claim that free indirect discourse cannot support subjective expressions attributable to different "selves" actually seems to rest on a better foundation than the dual voice hypothesis.
The voice metaphor is additionally complicated by the contention that textual polyphony only works in "a silent register" (quoted McHale 1978, 282). The point is echoed by Rimmon-Kenan (1983, 15), who, precariously sticking her neck out, suggests that "[i]t is perhaps because of the difficulty a speaker would experience in trying to perform orally the co-presence of voices characteristic of FID that the phenomenon seems more congenial to the silent register of writing." Here the metaphor comes full circle, welcoming in its own tail end. Of course, McHale (1978, 282) and Fludernik point out that free indirect discourse is not rare in oral discourse, which is not a silent register. Moreover, it is clearly a standard trick in oral discourse to imitate another person's voice without wholly suppressing one's own (in fact, only a professional imitator can presumably suppress all features of her/his own voice). The resultant real dual voice effect is possible precisely because the register of oral discourse is not silent, because there are idiosyncratic features such as gender, pitch, volume, stress, and intonation that, suitably combined, can be suggestive of more than one voice. This indeed tallies with Fludernik's view, put forward in her discussion of Bakhtinian alterity, that a dual voice effect is most likely to occur in texts which imitate an oral discourse situation, such as skaz. Although this view does not deny dual voice in principle, it does subvert dual voice in ordinary textual free indirect discourse: "If one compares these strategies of double voicing in Russian skaz with the nineteenth century 'realist' English novel, one notices immediately how monologous [my emphasis, M.J.] in Bakhtin's sense Victorian novels still are: they propagate a superior, frequently ironic, narratorial viewpoint which looks down on a fictional world, on characters riddled with moral, intellectual and linguistic foibles" (331). As for the test case of ironical free indirect discourse (351-54), Fludernik eventually concludes that speaking "of a narrator's voice intermingling with the figural idiom - or even juxtaposed to it within the free indirect discourse - is clearly incorrect" (354).[3]
Despite all these reservations, Fludernik does not trash the concepts of voice, dual voice, or textual polyphony. Indeed, she argues, there is one important reason for "recuperating" these concepts, and that is because they are needed in an analysis of "the reader's intuitive perception of discourse" (350). In an important elaboration of this claim, Fludernik adds: "Readers do in fact construct a narrator's (or author's) voice as a default value and, given sufficient linguistic evidence, experience an evocation of figural voices on that background" (350).
Deictic centres and voices are thus assigned to different levels of analysis - one linguistic, one cognitive/pragmatic. This is the move by which Fludernik achieves her almost miraculous synthesis of Banfield's (one self, no voice) and Pascal's (two voices) accounts. Briefly, on the level of linguistic forms, there are only deictic centres, but no voices. On a first level of cognitive processing, the reader constructs one voice, the narrator's (or the narration's, in the absence of a personalized narrator). On yet another level, that of pragmatic-cognitive modelling, the reader "perceives" the voice of a character. However, these voices do not mingle, as is claimed in the dual voice hypothesis. The narrator's voice, albeit in overall control, is "backgrounded" (116), while at the same time the character's voice is "assimilated" to "the reporting text" (115). In free indirect discourse, the narrative voice constitutes the controlling background against which a reader experiences a character's expressive voice.
7. A rhetoric of subjective expressions. Fludernik's final resolution of the linguistic evidence is presented in her chapter 8, in which she attempts to determine the extent to which the representation of other persons' language and thought relies on schematic or "typicalized" subjective expressions. Fludernik presents an exhaustive, or near-exhaustive, tabulation of subjective forms, demonstrating that these forms are not freely generative. Moreover, analyses of real-life conversational and journalistic texts suggest that subjective expressions are often used rhetorically and evocatively rather than strictly mimetically. Indeed their primary function often is to act as "enquotation devices" (419); they flag something as a representation of a speech or thought act, they signal a transition to the character's code, without actually engaging in (or guaranteeing) a facsimilar, verbatim quote. If one adds up Sternberg's (1982), Short's (1991) and now Fludernik's evidence, the very concept of facsimilar, mimetic quotation practically amounts to a fallacy.
To gauge the practical import of this conclusion, briefly reconsider the hackneyed She was tired. Although devoid of any obvious subjectivity markers, given a suitable context this would be interpretable as a free indirect discourse representation of a character's utterance or thought. Naturally, the sentence can also be loaded with expressive forms, say an interjection and an inversion - Oh, was she tired! In this case, and supposing that the narrator is not currently him/herself given to using expressive language of this kind - a possibility not in principle to be ruled out - the narrator would utilize "a schematic indication of alterity" (437) with the subjectivity markers mainly acting as conventional rhetorical cues. The inversion and the exclamation are rhetorical triggers affirming (suggesting, evoking) the deictic anchoring of a linguistic or semi-linguistic activity. The rest - attribution to a character's voice - will be supplied by the properly conditioned reader.
8. Functions and Frames. In order to inquire into the cognitive and pragmatic contribution of co-operative readers, Fludernik draws on the "frame" and "script" theories developed in artificial intelligence and cognitive research (Minsky 1979, Schank/Abelson 1977). Acting on cues given in the textual surface, the reader selects a cognitive model, script or frame, which is usually already loaded with "normal case" or "default" assumptions. In a reader's processing of a narrative text, scripts help to anticipate and understand the story line, bridge gaps, flesh out characterization, and attribute voices, attitudes and judgments to the narrator and the characters. As noted above, it is crucially important to keep apart the level of the linguistic data and the level (or levels) created by the reader's constructive contributions. On this premise, voices as well as narrators and narratees are fictions, cognitive constructs rather than empirical objects. It is the reader's cognitive activity which, from the proverbial marks on the page, creates the narrator and his/her voice, as well as the characters, the action and the characters' voices - this is how the "Fictions of Language" in Fludernik's title are related to the "Languages of Fiction" (2, 463).
Fludernik next turns to a frame-theoretical discussion of "the theoretical apparatus of narratology" (449). Following Pratt (1977), she addresses the question to what extent the frame of oral storytelling - a frame in which the teller is visible and audible, speaking in the presence of his/her audience (441) - also underlies narratological and cognitive models of literary narrative. Surveying the main models of narrative communication, Fludernik argues that these have always been heavily influenced by "natural" conditions; indeed she claims that "one can . . . explain the entire communication analysis of fiction as an (illicit) transferral of the frame of real-life conversational narrative onto literary personae and constructed entities" (448). "Speakers" and "voices," for example, are plain instances of concepts that originate in the oral storytelling frame. The oral frame also clearly underlies presuppositional reasoning like "if there is a story, somebody must needs tell it" (448). On a more subliminal level, its influence can be detected in various narratological anthropomorphizations like "covert narrators," "arrangers," "show-ers," "implied authors," and so forth.
Fludernik's economical (if slightly confusing) way of covering all of these instances as an "(illicit) transferral" naturally raises the question of a frame's adequacy. Some "fictions" created by applying the oral script are perfectly acceptable; this is why narrators and voices are valid concepts. Anybody who did not use the "narrator script" would not only fail to understand a great many literary narratives, but would indeed step outside a large interpretive community. The point here is that this particular transferral is not at all "illicit," but virtually necessary. On the other hand, there are, of course, cases where the script of oral narrative does not apply equally well. For example, Fludernik rightly notes that reflector mode narratives "are structured around the script of experiencing or viewing rather than telling" (449). Moreover, it is one thing to be able to explain why theorists find it tempting to construct implied authors and "show-ers" (Chatman 1990, 113), but it is quite a different question whether these are at all helpful (443; see also Nünning 1993). Obviously, the history of narratology is full of dubious naturalizations based on truly illicit transferrals.
9. Work in progress. Here Fludernik breaks off this extremely pertinent narratological discussion and tantalizingly refers the reader to her next book-length study, already "in preparation," entitled Towards a 'Natural' Narratology. This book, it is promised, will provide a more extensive "presentation of the frame-theoretical model in its application to the theoretical apparatus of narratology" (449). Judging from Fludernik's past and present interests, this will presumably be an effort to ascend Pratt's (1977) ladder, utilizing the script of natural narrative as a paradigm script of literary narratology.
However, a word of caution appears to be in order. First of all, it will be prudent to recall the comments made by Culler (1988) in his critique of Pratt (1977), especially his objections to the supposed primacy of oral narratives, as well as to the "tendentiousness" of the term natural (1988, 211-12). If oral stories are called natural, Culler notes, the suggestion is that novels are somehow unnatural (1988, 214). It is true that Fludernik carefully scarequotes the word, but this will, of course, only raise the justified complaint that using terms under protest is not a good foundation for anything. Clearly a theory that aims to be radical - a key notion in all of Fludernik's endeavours - must proceed on (and if necessary, invent) its own termes justes.
Two other objections suggest themselves, and Fludernik would do well to be aware of them from the beginning. One is that of reductionism: the natural frames might prove too simple for literary narratives. The second is the one identified in her own phrase of potential "illicit transferral": the natural frames might be the wrong frames. On the grounds of Fludernik's past and present performance - her analysis of the historical present (Fludernik 1992), second person narratives (Fludernik 1993a), and now of speech and thought representation (Fludernik 1993b) - it is unlikely that the first objection will have a case. So far, the natural frames found by Fludernik have never been simplistic or obvious; in fact, her simple categories are usually more complex than other theorists' complex categories. Thus there is every reason to suppose that narratology stands to gain in sophistication and differentiation by the extraliterary analogies. The best indication of this is the present study itself, whose very convincing achievement it is to leave the well-trodden paths and widen the horizons of traditional narratological inquiry.
The second objection addresses a more real danger. Here one must hope that (as in her assessment of reflector mode narratives) Fludernik will not forget that frames sometimes fail to fit, that some frames may be suitable points of departure, but less suitable points of arrival, and that, very generally speaking, new situations require new frames. One also hopes that the natural frames will not lead to the recuperation of too many traditional metaphors. One can see why it might be an advantage to recuperate "voices," but one certainly does not want to recuperate authors who do their own narrating, implied authors doubling up as self-fulfilling interpretations, or reflectors acting as narrators. Throwing away these ladders may not be enough - there are too many traditionalists around who want to put them up again.
Notes
1. In fact, "Oh, she simply hated her daughter!" is a sentence of narrative report in Nabokov's Lolita (New York: Van Rees, 1955), p. 82. "She was tired," the third sentence in Joyce's "Eveline," is variously read as free indirect discourse, narrated perception, or ambiguous (see Chatman 1978, 204-5; Füger 1993, 50-52).
2. Despite the popularity of scales in literary theory, literary theorists are usually unaware of the very functional scale concepts that, since the 1940's, have been part of measurement theory in the behavioral sciences. See Siegel (1956, 21-30) for a standard introduction. Whether the concepts of partly ordered scales (Coombs et al. 1954), fuzzy sets (McNeill and Freiberger 1993), or preference rule systems (Jackendoff 1983) could equally be appropriated to literary theory cannot be readily answered here; but the question is certainly worth persuing.
3. There is one isolated lapse of precision in Fludernik's discussion of irony. This is her suggestion that one should make a distinction between "narratorial irony," in which a "textual speaker . . . utilizes contradictions on whatever level," and "authorial irony," in which "the contradictions are recognized only by the reader and the ironic intent is hence attributed to the (implied) author rather than the narrative voice" (352). As far as I can see, these are unproductive and unnecessarily confusing definitions. For one thing, as Fludernik well knows (because she draws attention it herself), "contradictions" do not capture the pragmatic character of irony (352). Second, one must object to the way the implied author - who, usually, has no function whatsoever in Fludernik's approach - manages to insinuate itself here. More importantly, the term authorial irony conflicts with the term authorial narrator - according to Fludernik, an authorial narrator's irony is narratorial irony, not authorial irony. The one example discussed as authorial irony (item 44, p. 352) is an ironical quotation controlled by an authorial narrator who fully conforms to Fludernik's characterization of ironic Victorian narrators (see above), and is therefore a case of narratorial irony. Surely a term like Chatman's "unreliability irony" (1990, 149-51), an irony whose victim is a narrator like Huck Finn or the governess in The Turn of the Screw, would be preferable. In any case, there is no obvious connection between this authorial/unreliability type of irony and textual voice. It is probably the degree of personalization rather than the degree of vocality that correlates with the likelihood of narratorial unreliability.
Works Cited
[26 Aug 2000]