Narrative levels (also referred to as diegetic levels) is an analytic notion whose purpose is to describe the relations among the plurality of narrating instances within a narrative, and more specifically the vertical relations between narrating instances. Thus, three narrative levels can be identified in a story where a narrator reports the telling of a story by a narrator-character within his own story: the level within the global text at which the telling of the narrator-character’s story occurs; the level at which the primary narrator’s discourse occurs; the level of the narrative act situated outside the spatiotemporal coordinates of the primary narrator’s discourse. In a broader sense, however, narrative levels also include horizontal relations between narrating instances situated at the same diegetic level, as when a story is told by several narrators. The notion of narrative levels serves to describe the spatiotemporal relations between the various narrating acts occurring in a narrative, and can thus be thought of more accurately as “narration levels” or “narrating levels.”
According to Genette, who first proposed the term, narrative level is one of the three categories forming the narrating situation, the other two being time of the narrating and person (1972: chap. 5). Narrative levels, arranged bottom upwards, are extradiegetic (narrative act external to any diegesis), intradiegetic or diegetic (events presented in the primary narrative), and metadiegetic (narrative embedded within the intradiegetic level). What distinguishes narrative level from the traditional notion of embedding is that it marks a “threshold” in the transition from one diegesis (spatiotemporal universe within which the action takes place) to another (Genette  1988: 84). As every narrative is taken charge of by a narrative act, difference of level can be described “by saying that any event a narrative recounts is at a diegetic level immediately higher than the level at which the narrating act producing this narrative is placed […]. The narrating instance of a first narrative [récit premier] is therefore extradiegetic by definition, as the narrating instance of a second (metadiegetic) narrative [récit second] is diegetic by definition, etc.” (Genette ( 1980: 228–29). Bal (1977: 35) and Rimmon-Kenan ( 2002: 92–3) invert this order, placing the diegetic level in a “subordinate” position in relation to the extradiegetic level. Discussions of narrative level frequently overlook the fact that it is not an isolated category but that, forming part of the narrating situation, it correlates with a second type of diegetic relation, a relation of person: hence a narrator (Uri Margolin → Narrator) is either heterodiegetic (absent from the narrated world), homodiegetic (present in the narrated world) or autodiegetic (identical with the protagonist). Together, level and person form the narrator’s status, broken down into a four-part typology of the narrator (Genette  1980: 248; see 3.1.1 below. On the notion of diegesis, cf. Pier 1986).
Formulated in terms of enunciation, narrative level in effect opposes “who speaks?” and “who acts?,” thus opening the way to a more precise description and analysis of change of level through the identification of textual markers. Genette ( 1980: 232–34) distinguishes three types of relations binding metadiegetic narrative to primary narrative: (a) explanatory, when there is a link of direct causality between the events of the diegesis and those of the metadiegesis; (b) thematic, by way of contrast or analogy between levels, as in an exemplum or in mise en abyme, with a possible effect of the metadiegesis on the diegetic situation; (c) narrational, when the act of (secondary) narrating merges with the present situation, diminishing the prominence of the metadiegetic content (Rimmon-Kenan  2002: 93, names the latter relation “actional”). With reference to Barth (1981), these types were later refined into six “functions” ordered by decreasing thematic relation between primary and second-level narrative with increasing emphasis on the narrative act itself: (a) explicative; (b) predictive; (c) purely thematic; (d) persuasive; (e) distractive; (f) obstructive (Genette  1988: 92–4). And finally, by pushing the narrative act as a means of transition between levels yet further, as when the author or the reader enters the domain of the characters, or vice versa, the boundaries between levels are violated, resulting in metalepsis (John Pier → Metalepsis).
Analogously to focalization (Burkhard Niederhoff → Focalization), a systematization of theories of perspective and point of view (Burkhard Niederhoff → Perspective – Point of View), narrative levels represent a narratological response to the traditional notions of frame stories and embedded stories. Narrative level, however, is both conceptually more global than either of these practices and more restricted. On the one hand, every narrative, embedded or not, exists by virtue of a narrative act which is necessarily external to the spatiotemporal universe within which the events of that narrative take place, thus situating it in a web of narrating instances. On the other hand, narrative levels come into play only with a shift of voice, which is not always taken into account by the traditional notions (e.g. the dream sequences introduced into Nerval’s “Aurélie” do not represent changes of level since there is no change of narrator). At the same time, narrative levels provide a set of principles that makes it possible to describe both frame stories and embedded stories. Technically, a process of embedding occurs in both types, but whereas frame stories, usually short, serve to bracket the main story (e.g. the expository pages to Marlow’s narrative in Heart of Darkness), embedded stories, of limited duration, remain subordinate to the primary narrative (e.g. the novella “The Curious Impertinent” in Don Quixote). “If the tale is conceptualized as subsidiary to the primary story frame, a relationship of embedding obtains; if the primary story level serves as a mere introduction to the narrative proper, it will be perceived as a framing device” (Fludernik 1996: 343; see 3.2 below).
In a sense that bears on narrative levels only in part, embedding designates one of the three ways in which sequences can be combined syntactically into more complex forms: linking; embedding; alternation (Bremond 1973; Todorov 1966, 1971). Formally, embedding is defined by syntactic subordination, even though it does not necessarily involve a change of narrating instance (a digression can be related by the primary narrator).
By reformulating narrative embedding in terms of the enunciative threshold in the transitions between levels, Genette opened up a debate with far-reaching implications as to the nature of the relations between levels, a debate centered, at least initially, on the prefix meta-. If understood analogously to metalanguage, metanarrative (métarécit or récit métadiégétique) would correspond to the embedding narrative—a primary narrative on or about the second-level narrative. But in fact metanarrative (or better: metadiegetic narrative) corresponds to the events related within diegetic narrative. Genette insisted that just as the narrating instance of the primary narrative is extradiegetic, so that of a metadiegetic (second-level) narrative is diegetic ( 1980: 229). In order to resolve the potential terminological ambiguity, Bal points to three usages of meta-: (a) a quoted discourse is metalinguistic in the sense of being fictional in relation to the quoting discourse (a sense close to Genette’s); (b) from a functionalist perspective, the quoted discourse is a metanarrative commentary on the quoting discourse (metalinguistic textual devices, etc.); (c) an abusive extension of meta- to cover commentary of any kind (Bal 1981: 53–6; on metanarrative commentary, see Nünning 2004). As for embedding proper, this occurs when there is insertion (attributive discourse provides a link between two discourses), subordination (which excludes juxtaposition), and homogeneity (e.g. one sequence inserted into another)—a set of relations that comes under the prefix hypo-. On this basis, it is proposed that “metanarrative” and “metadiegetic” be replaced, respectively, by “hyponarrative” and “hypodiegetic”—a level below rather than in the diegetic level (Bal 1977: 35; 1981: 43–53; cf. Fludernik 1996: 342; Rimmon-Kenan  2002: 92–6). It must be noted, however, that this revision inverts the order of narrative levels in Genette’s presentation, creating a relation of hierarchical subordination with the extradiegetic level situated at the top, and that it does so at the expense of the intended relation of inclusion between primary and embedded narrative. The terminological refinement thus comes at a price, since it prefigures a hierarchical top-down ordering of narrating instances that may not pertain to all narratives, and also because it severs the significant link between metanarrative and metalepsis (Genette  1988: 91–2); it further conflicts with the specific use of hypo- in the study of hypertextual relations where a hypotext (e.g. The Odyssey) is prior to a hypertext (e.g. Ulysses) (Genette 1982). Interestingly, Bal later abandoned her neologisms and radically altered the notion of narrative level itself. Her comments on “levels of narrative,” based on grammatical subordination of the actor’s text by the narrator’s text, are devoted to various forms of speech representation (Brian McHale → Speech Representation), while embedding, which she explains as text interference between actor’s text and narrator’s text, reverts to the traditional concept in which an embedded fabula serves to explain or to explain and determine the primary fabula or in which there is a relation of resemblance between the two (Bal  1997: 43–60). As a result, the threshold marking the transition between diegeses disappears, and with it the vectors of embedding/embedded and narrating instance constitutive of narrative level.
Narrative levels, then, cover the enunciative situation of narrative in general as well as various forms of embedded narrative. A multifaceted concept, embedding can be found in various disciplines including linguistics, logic, psychology, communication, computer science, etc. With reference to the criteria of punctuation and continuum, boundary, and logical levels that characterize the concept in these fields, Füredy (1989) identified the more extreme forms of embedding found in artistic representation: (a) intact and multiplying boundary (e.g. mise en abyme, which in principle is open to infinite recursion); (b) intact but reified boundary (escape from the undecidable and oscillating boundary built into Escher’s Drawing Hands is possible only through access to an otherwise inviolate metalevel); (c) transgressed boundary (metalepsis). In the field of conversation analysis (Monika Fludernik → Conversational Narration – Oral Narration), by contrast, embedding, which is more closely bound up with context, is referred to as “embeddedness.” Thus a narrative of personal experience will be embedded in accordance not with syntactic subordination or logical level so much as it is with surrounding discourse (explanation, prayer, etc.) and social activity (frequency and length of turn-taking, degree of thematic and rhetorical integration into the general conversation) (Ochs & Capps 2001: 36–40; on the performativity (Ute Berns → Performativity) of oral narration as “situated communication,” see Young 1987: chap. 4). In possible worlds narrative theory, on the other hand, embedded narratives are a variety of alternate possible worlds that exist as beliefs, intents, etc. in the form of retrospective interpretations of the past or projections about the future in relation to the actual world, and thus contribute to the intelligibility of the fabula (Ryan 1986).
The possible worlds approach does in fact open the way to a logically consistent model of narrative embedding. Distinguishing between discourse as an illocutionary category and story as an ontological category, Ryan (1991: chap. 9) adopts a cross-classification of three dichotomies: +/- illocutionary; +/- ontological; +/- actual crossing. On this basis, a system of four types of narrative boundaries, organized into a “concentric structure,” is then elaborated: (1) no boundary, as a given speaker describes a same level of reality; (2a) actually crossed illocutionary boundary, as when the first and second speakers are different but refer to the same reality (e.g. dialogue quoted in direct speech); (2b) virtually crossed illocutionary boundary (e.g. character’s narrative presented by the narrator’s discourse in indirect speech); (3a) actually crossed ontological boundary with no change of speaker (change in levels of reality in Alice in Wonderland reported by the primary narrator); (3b) virtually crossed ontological boundary by the same speaker (dream anchored in reality but described from the outside); (4a) actually crossed ontological boundary with change of speaker (a story within a story, as in the Arabian Nights); (4b) virtually crossed ontological boundary with change of speaker (primary narrator projects an imaginary story by a second-level narrator). One advantage of this model of narrative levels (and by implication, Genette’s, though he is not referred to) is that it provides a solution to the difficulty for traditional accounts of embedding and frame tale in marking off discourse boundaries from the boundaries separating different narrative contents. The system of narrative boundaries or frames, which is classificatory and static, is completed with the notion of “stacks,” a metaphor borrowed from computer science (cf. Hofstadter 1980: 127–31) in order to account for the dynamic and sequential ordering of levels in texts. “In a canonical narrative, the building and unbuilding of the stack follows a rigid protocol which restricts the range of legal operations. This protocol requires that levels be kept distinct, that they be pushed or popped on the top of the stack exclusively; that pushing and popping be properly signaled; and that every boundary be crossed twice, once during the building and once during the unbuilding. At the end of the text, the only level left on the stack should be the ground level. This protocol is respected by all standard narrative texts, but not by all texts of literary fiction. Far from being constrained by the conditions of narrativity, the fictional text may subvert the mechanisms of the stack, thus openly taking an antinarrative stance” (Ryan 1991: 187). The author goes on to discuss various “subversions” of the canonical narrative (the endlessly expanding stack, strange loops, contamination of levels, etc.; see also McHale 1987: chap. 8), suggesting in effect that the stack metaphor operates through execution of a code rather than in accordance with the enunciative principle according to which the narrative act occurs in a spatiotemporal universe external to that of the narrative events, and that non-canonical narratives are deviant in relation to “standard” narratives. However, the logical consistency of Ryan’s model notwithstanding, it might be wondered if is not precisely boundary crossings, irregular as well as “legal” (Peter Hühn → Event and Eventfulness), that contribute to a text’s narrativity (H. Porter Abbott → Narrativity).
In contrast to Ryan’s modeling of boundary crossings, derived from the story/discourse dichotomy, Schmid (2005: 72–99) considers narrative levels, together with presence/non-presence of the narrator in the diegesis, a basic element in the elaboration of a typology of narrators. Rejecting traditional typologies, which generally combine first- and third-person narration with internal vs. external perspective, Schmid adopts Genette’s criteria, although with a revision of his terminology. First, diegesis designates the level of the narrated world, and exegesis the level of the narrating. Second, the diegetic narrator belongs to both levels, and the non-diegetic narrator only to the exegesis. The elimination of personal pronouns and the disappearance of the prefixes homo-/auto- and hetero- serve to underscore a differentiation which is current in German narrative theory and implicit in Genette’s system, namely erzählendes Ich/erzähltes Ich, or “narrating I”/“narrated I” (cf. sujet de l’énonciation/de l’énoncé; “subject of the enunciation”/“the enunciated” in French linguistics). These emendations make possible a terminologically and conceptually clarified typology of narrators: primary non-diegetic (=extra- heterodiegetic); primary diegetic (=extra- homodiegetic); secondary non-diegetic (=intra- heterodiegetic); secondary diegetic (=intra- homodiegetic); tertiary non-diegetic (=meta- heterodiegetic); tertiary diegetic (=meta- homodiegetic) (Schmid 2005: 87; cf. Genette  1980: 248). It must be remembered, however, that Genette’s terminology is additionally intended to account for the narrating instance, i.e. the difference of level resulting from the fact that the narrative act necessarily takes place in a spatiotemporal universe which is external to that of the events related.
From a poststructuralist perspective, the notion of narrative levels is symptomatic of a “boxing of narrative,” “a structure of supervision,” and “purity of composition.” According to Gibson (1996: 215): “It is crucial to the Genettian concept of levels that there be no seepage or osmosis across the threshold. The substance composing each stratum must be unadulterated. There must be no hint of ambivalence or paradox in the definition of a given stratum, no irrational features that might trouble its terms. Equally, there must be no anomalies in any of the strata, nothing mixed or hybrid.” However, Gibson’s critique of “narratological geometrics” (which can also be leveled against Ryan and Schmid) remains silent on such limit cases as mise en abyme, metalepsis, and pseudo-diegetic narrative, overlooking the fact that levels exist by virtue of their thresholds and are perpetually exposed to transgressive crossings, just as it fails to mention Genette’s study of “transtextual” relations (1982, 1987). Nor does the critique take into account the potential descriptive utility, widely acknowledged by theoreticians of differing orientations, of narrative levels, embedding, frames, stacks, etc., despite the inevitably metaphorical nature of whatever terminology is employed. In presenting his notion of “narrative laterality” (inspired from Serres, Deleuze, Derrida), Gibson himself makes ample use of the very terminology and concepts he denounces in order to describe the “collapse of hierarchies” (cf. García Landa 1998: 304).
To be sure, formalist/structuralist models of narrative levels, which set out to reformulate the traditional notions of embedding and framing in terms of a general theory of narrative, may not be so rigid and constraining as supposed. As the transgressive and subversive passages between levels noted above make clear, the relations between levels surpass those of subordination and hierarchy. Genette suggests as much when, in redefining these relations, he adopts a functional perspective ( 1988: 92–4; cf. 2 above), stating however that the province of narratology is not that of “interpretation” (87) and thus stopping short of taking full stock of this position. In fact, he implicitly shifts to a speech act approach to narrative levels, but without putting it in those terms: as shown by Shryock (1993: 6–8), the explanatory function (by metadiegetic analepsis) and the predictive function (by metadiegetic prolepsis) of the second-level narrative operate by virtue of their illocutionary force, while the persuasive, distractive, and obstructive functions can be qualified as such only by their perlocutionary effects, the obstructive function in particular binding the two levels together solely by an act of narration (a point disregarded by Rimmon-Kenan when she renames the narrational relation between levels “actional”). In this light, narrative levels are so many ways of appealing to active participation by the addressee, and not a mere “stratagem of presentation” or “conventionality,” as concluded by Genette ( 1988: 95): the way is opened toward a functional approach to narrative levels in place of the more monological information-based model of narrative communication generally adhered to by classical narratology (cf. Chatman 1978: 151; Jan Alber & Monika Fludernik → Mediacy and Narrative Mediation).
One consequence of formulating narrative levels in functional terms is the reordering of the notion of levels itself. Following a critique of Bal’s revisions of Genette, Nelles (1997: 127–43) introduces two distinctive types of embedding: “horizontal” embedding occurs when a story is told by two or more narrators without a change of diegetic level, and “vertical” embedding when there is a change of level and of speaker and/or of narratee. These forms can be likened, respectively, to Ryan’s type 2a, 2b and 4a, 4b boundary crossings. An additional case is the alternate universes created in a character’s mind, as in a dream (cf. Ryan’s type 3b), which Nelles explains not as a change of level but of the spatiotemporal coordinates of the story, or what Young (1987: 24) calls “Taleworld” (“the realm of the events the story is about”) as opposed to the “Storyrealm” (the “region of narrative discourse within the realm of conversation”). With reference to McHale’s (1987) epistemological vs. ontological fictions, he renames horizontal and vertical embedding “verbal” and “modal,” respectively. Nelles contends that the function of embedded narrative is thematic (by contract or analogy) and that the interpretive strategies implemented by embedding can be analyzed on the basis of the hermeneutic, proairetic, and formal codes, adapted from Barthes’ analysis of “Sarrasine.”
Another functional approach to narrative levels has been elaborated by Coste. Rooted in a communicative theory of narrative, this approach emphasizes the role of the narrator not as homo- vs. heterodiegetic, but as the enunciator: “A narrator is the subject of enunciation of one or more utterances that either contain a narrateme or are involved in the production of a narrateme by the reader” (Coste 1989: 166; on the notion of narrateme and the structure of narrative meaning, see chap. 2). Essential here is the functional separation between subjects of enunciation and subjects of the enunciated, splitting the subject as narrating instance between present storyteller and past (or future) character (cf. Schmid above). Subjects of enunciation, always exterior to the enunciated, are thus determined according to their relations with: (a) enunciated utterances; (b) other subjects of enunciation; and (c) addressees, intentional or not (167). On these premises, Coste sets forth two types of narrative embedding: hypotactic, resulting from grammatical subordination and materialized in the form of delegated narration; paratactic (juxtaposition, coordination), forming a system of “parallel” narrators at the same level and related to dialogism (David Shepherd → Dialogism) in which narratives are combined either by sequential relay, concurrent/conflictive versions, or narrational crossfire (167–73). The same distinction is made by García Landa (1998: 302), who has also drawn attention to the link between paratactically embedded literary narratives and face-to-face communication. In this type of narration, addressee roles are more varied than those typically found in written texts: as in conversational narratives, paratactically organized stories and novels may not be restricted to intended addressees (narratee, implied reader), but also fall on the ears of mere auditors or even those of overhearers or eavesdroppers, including narratologists (García Landa 2004; cf. Goffman 1981). To the extent that both types are enunciative, they can be likened to Nelles’s horizontal or verbal embedding and to Ryan’s illocutionary boundary crossings and, respectively, to her types 2b and 2a. Where Coste’s system differs from these models is in the notion of “overall narrator,” a cooperative construct that acts as an organizer or control function which may be textualized (editor in the 18th-century novel) or not (Wolf Schmid → Implied Author), although it must be mentioned that Ryan (2001), in a different spirit and independently of her work on narrative boundaries, has argued in favor of breaking the narrator down into the creative (self-expressive), transmissive (performative), and testimonial (assertive) narratorial functions constitutive of “narratorhood.” Of central interest in Coste’s model are the interdependent, organic relations between the two types of embedding, captured by the image of the “narrational tree”: while the roots grow deeper and the trunk higher (hypotactic or vertical embedding), the branches spread out laterally (paratactic or lateral embedding).
A significant and oft overlooked fact of the principle of narrative levels is that it focuses on formal features of embedding and as such does not—nor is it intended to—distinguish between the relative importance, quantitative or otherwise, of primary and second-level narrative: the process of embedding employed in the Arabian Nights is identical to that of the interpolated narratives in Don Quixote. The deployment of narrative levels and the modalities of transitions between them are extremely variable, both historically and generically (the Decameron, the picaresque novel, the epistolary novel, postmodern fiction, etc.; for a brief historical survey of frame tales, see Kanzog 1966; for embedding in various genres, see Duyfhuizen 1992). As already discussed, there exist several ways of organizing narrative levels including the weight of thematic criteria relative to the degree of prominence of the narrative act (Genette), the vectorization of illocutionary and ontological boundaries (Ryan), the combination of narrating I / narrated I with level in a typology of the narrator (Schmid), and the separation of levels into horizontal and vertical embedding (Nelles, Coste). It is also possible to examine the textual integration of narrative levels according to the length of primary and second-level narratives relative to one another, the two poles of which are the frame tale and mise en abyme.
The simplest definition of the frame tale—“one story encloses another like a frame” (Kanzog  1977: 321)—is ambiguous because it fails to distinguish between the framing and the framed, and it is also misleading in that (a) picture frames (to which the metaphor alludes) rarely form a part of the framed pictorial representation and (b) “framed” narratives do not come forth unmediated but necessarily interact with surrounding discourse. When examined from the perspective of narrative levels, frame tales must be qualified as a particular type of intradiegetic narrative with regard to the narrative in which they are contained (cf. Ryan’s type 4a border crossing) and are thus, however brief they might be, subject to the criteria of narrativity in their own right (cf. Wolf 2006: 181). In addition to change of voice and level and to the potential for multiple levels of embedding, narratives that employ the framing technique—and this accessorily to the principle of narrative embedding properly speaking—can incorporate a single second-level narrative (Heart of Darkness) or multiple second-level narratives (the Arabian Nights) as well as, within a given second-level narrative, additional embedded narratives (as in “The Three Ladies of Baghdad”). A fourth feature of frame stories is their compositional distribution: a framing can be complete (appearing at the beginning and end of the embedded story), incomplete (introductory only or terminal only, possibly producing metaleptic effects), or interpolated (appearing intermittently) (adapted from Wolf 2006: 185–88).
Overall, the frame tale, together with its second-level narrative, relies heavily on compositional means. Most notably, it offers the possibility of linking together an otherwise disparate group of stories and of establishing thematic relations among them, and it thus contributes to textual coherence (Michael Toolan → Coherence). Semiotically, this corresponds to the syntactic dimension of semiosis. Another feature of the frame tale, particularly in its written form, is that it replicates the communicative situation of oral storytelling, indicating a time and place of the narrative act and the audience and buttressing the “narratorial illusionism” of the framed tale (Kanzog  1977: 322; Nünning 2004: 17; Williams 1998; 110, 113; Wolf 2006: 188–89). The communicative specificities of the framing technique thus come within the scope of pragmatics. And finally, the traditional function of the frame tale (carried over, inter alia, to the elaborate prefatory material of the 18th-century novel) is to validate the framed story (which itself may be improbable) with an air of authenticity, thanks to the impartial report by the primary narrator. This does not necessarily mean, however, that the primary narrator vouches for the veracity of the related facts: a potentially rhetorical move (as in the case of an unreliable narrator), authentification by the primary narrator consists in principle in affirming that the second-level narrator related such-and-such, not in asserting what s/he related (cf. Duyfhuizen 1992: 134; Williams 1998: 114; Wolf 2006: 192). This aspect of the framing technique can be assimilated to the semantic dimension of semiosis, although it also merges with pragmatic considerations.
The defining characteristic of mise en abyme is the relation of repetition and reflection the second-level narrative entertains with the quantitatively greater narrative within which it is contained. Iconic in the semiotic sense (cf. Bal 1978) and producing disruptive but potentially significant effects on the progression of the primary narrative, the device exists in three basic forms (Dällenbach 1977): (a) mise en abyme of the utterance (e.g. portions of the romance The Mad Trist that parallel certain incidents in Poe’s “The Fall of the House of Usher”); (b) mise en abyme of the enunciation, or highlighting of the process of narrative communication (e.g. the exemplum, whose aim is to instill in the reader a moral awareness); (c) mise en abyme of the code or text (e.g. Abish’s Alphabetical Africa, where chapter 1 employs only words beginning with letter “a,” chapter 2 only words beginning with the letters “a” and “b,” etc. up to chapter 26, the second half of the novel reversing this order). These varieties of the device also come respectively within the scope of semantics, pragmatics, and syntactics, although in the case of mise en abyme, unlike in the framing technique, these dimensions are modeled iconically into the primary narrative.
It is not by coincidence that Genette’s study of paratext—the “undecided zone” between the interior and the exterior of the text occupied by prefaces, epigraphs, notes, interviews, etc. which constitutes a space of transaction between author and reader—is titled Seuils (thresholds), the very term employed to describe the transitions between narrative levels. One broad area of inquiry for additional study is the interaction of narrative levels with speaker-hearer relations from a sociolinguistic perspective, beginning with “frame analysis” (Goffman 1974, 1981; Ochs & Capps 2001; Young 1987). Another need, within the scope of cognitive narratology (David Herman → Cognitive Narratology), is to gain further insight into the WHAT, WHERE, and WHEN that can be provided by narrative levels in the construction of storyworlds as focused on by research in text worlds (Werth 1999), deictic shifts (Duchan et al. eds. 1995), and contextual frames (Emmott 1997).