The following is an excerpt from this webpage:

Manfred Jahn
Narratology: A Guide to the Theory of Narrative

Full reference: Jahn, Manfred. 2005. Narratology: A Guide to the Theory of Narrative. English Department, University of Cologne.


N2.3.5. Transgression of levels


 Normally, the levels of action, fictional mediation, and nonfictional communication (as shown in the graphic above, N2.3.1) are hermetically sealed domains indicating crucial thresholds of control and awareness. Any agent situated on a higher-level dominates and frames all lower-level agents, while lower-level agents are unaware of the existence of the higher-level agents. For instance, the characters at the level of action do not know that they are characters in some narrator's story, and they cannot complain if their acts or motives are misrepresented by this narrator. Similarly, a narrator such as Holden Caulfield is not aware of the fact that he is a fictional figure in a novel written by J.D. Salinger (the point is spelled out in more detail in N1.6).

Occasionally, however, one finds some playful and not-so-playful transgressions of levels, which Genette calls 'metalepses' (Genette 1980 [1972]: 234-237). Typical cases cited in the literature are (1) characters attempting to establish communicative contact with either audience or author (see the device of the 'aside ad spectatores' in drama and film -- D3.4, also actors 'acting out of character'), and (2) narrators and narratees seemingly joining the characters in the action. Slightly modifying the terms used in Malina (2000), the first could be called a 'diegetic-to-extradiegetic metalepsis', while the second would be 'extradiegetic-to-diegetic' (these terms differ slightly from the ones actually used by Malina because I want them to dovetail with the Genettean terms listed in N2.4, below). Here is a famous example of the second type:

Clearly, a metalepsis can either be playful and harmlessly metaphorical (as in the example above) or else a serious transgression violating the "sacred frontier between two worlds, the world in which one tells, the world of which one tells" (Genette 1980 [1972]: 236) -- in other words, the domain of the discourse and the domain of the story. See D. Herman (1997) for a formal description of metalepsis and Malina (2000) for an in-depth exploration of functions, effects, and types of 'reconstructive', 'deconstructive', 'subversive', and 'transformative' metalepses.

Whoever is interested in another batch of recent studies of the phenomenon should watch out for the proceedings of the International Colloquium "Metalepsis Today" held at the Goethe Institut, Paris, on 29-30 November 2002 (ed. John Pier). Related phenomena include alterations in prose narratives (N3.3.15), the alienation effect in drama (D6.1), the device goof in film (F5.3.3), and parabasis in classical rhetoric (the latter term refers to a character directly addressing the audience).

N2.4. Narrative Levels

N2.4.1. Story-telling can occur on many different levels. As Barth (1984 [1981]) puts it, there are "tales within tales within tales". The model presented in N2.3.1, above, provides a general framework which can easily be adapted to more complex circumstances. One such circumstance arises when a character in a story begins to tell a story of his or her own, creating a narrative within a narrative, or a tale within a tale. The original narrative now becomes a 'frame' or 'matrix' narrative, and the story told by the narrating character becomes an 'embedded' or 'hyponarrative' (Bal 1981a: 43):

N2.4.2. For a more elaborate analysis of embedded narratives, Rimmon-Kenan (1983: 91) suggests the following terms:

See Genette (1980 [1972]: 228-234; 1988 [1983]: ch. 14) [extradiegetic, diegetic, intradiegetic, metadiegetic]; Bal (1981: 48-50) [on 'hypo-' vs. 'meta-']; Lanser (1981); Rimmon-Kenan (1983: 91-94) ['graded' narrators and narratives]; Duyfhuizen 1992; O'Neill (1994: ch. 3); Nelles (1997: ch. 5).

N2.4.3. Genette has illustrated the basic structure of embedded narratives with the help of a naive drawing using stick-figure narrators and speech-bubble narratives (Genette 1988 [1983]: 85). In graphic (a), below, first-degree narrative A contains a second-degree story B. The other examples in the graphic are 'Chinese-boxes models' which can be drawn to great accuracy, indicating both the relative lengths of the various narratives as well as their potentially 'open' status (Lintvelt 1978; Ryan 1991: 178; Branigan 1992: 114).


In example (b), A is a first-degree narrative, B1 and B2 are second-degree narratives, and C is a third-degree narrative (Question: which ones of these are matrix narratives?). Finally, example (c) illustrates the embedding structure of Henry James's The Turn of the Screw. James's novel ends on the conclusion of a third-degree narrative (the Governess's tale) without explicitly closing its two superordinate matrix narratives.

There are a number of texts which are famous for their multiply embedded narratives: The Thousand and One Nights, Chaucer's The Canterbury Tales, Jan Potocki's The Saragossa Manuscript, Emily Brontė's Wuthering Heights, John Barth's "Menelaiad". See also Chatman (1978: 255-257), Barth (1984 [1981]), Ryan (1991: ch. 9), Baker (1992).

N2.4.4. As an exercise, work out the following problems. Some of them are quite tricky; use simple Chinese-boxes models to argue your answers.

1. Can a hyponarrative be a matrix narrative?

2. Can a matrix narrative be a hyponarrative?

3. Must a first-degree narrative be a matrix narrative?

4. Can a text have more than one first-degree narrative?

5. Can a single character be both a second-degree narrator and a third-degree narrator?

N2.4.5. Comment. The foregoing account makes short shrift of a host of rather unhappy terms that haunt the narratological literature, including the term 'frame narrative' itself (does it refer to a narrative that has a frame or one that is or acts as a frame?). With reference to graphic (a) in N2.4.3, above, Genette calls the narrator of A an 'extradiegetic narrator' whose narrative constitutes a 'diegetic' level, while B is a 'metadiegetic narrative' told by an 'intradiegetic' (or, confusingly, 'diegetic') narrator. On the next level of embedding, one would get a meta-metadiegetic narrative told by an intra-intradiegetic narrator. Against this, Bal (1981a: 43) and Rimmon-Kenan (1983: 91-93) have argued that hypo- (from Greek 'under') is a more adequate prefix than meta- (from Greek 'on, between, with') to refer to what are, at least technically (though not necessarily functionally), subordinate narratives. Oddly, however, in their system, B (in graphic [a]) is a 'hyponarrative' told by a 'diegetic narrator', and if there were an additional level, Bal and Rimmon-Kenan would be happy to have a 'hypo-hyponarrative' told by a 'hypodiegetic narrator', and so on. Although the hypo- concept is a useful one, correlating hypodiegetic narrators with hypo-hyponarratives is both awkward and counterintuitive. More drawbacks of the nomenclature become apparent when one tries to tackle the problems set in N2.4.4.

N2.4.6. Embedded narratives can serve one or several of the following functions:

N2.4.7. Hyponarratives are also often used to create an effect of 'mise en abyme', a favorite feature of postmodernist narratives (Dällenbach 1981; Ron 1987; McHale 1987: ch. 8; Wolf 1993). The graphic on the right shows a visual example.


Spence (1987: 188) cites the following example: