Perspective in narrative may be defined as the way the representation of the story is influenced by the position, personality and values of the narrator, the characters and, possibly, other, more hypothetical entities in the storyworld. The more common term in Anglo-American criticism, which will be treated as equivalent here, is "point of view."
In the visual arts, perspective refers to a method that presents a scene as perceived from a "single fixed viewpoint" (Carter 1970: 840), creating the impression of three-dimensional depth on a two-dimensional surface. In a painting of this sort, parallel lines converge as they recede from the viewer; objects gain or lose in size depending on whether they are near or far; and in the background, colors lose their intensity and acquire a bluish tinge. That the concept of perspective can also be applied to language is made evident by the following sentence, assumed to be spoken by a boy: "My father towered above me." The man in question need not be a giant; the impression of his great height might simply result from the boy's viewpoint. The example also shows that the concept of perspective may be extended from vision in the literal sense to vision in the figurative sense, i.e. to interpretation and evaluation. Thus the verb "towered" suggests that the father is threatening the boy. Again, this impression need not be shared by other observers, as it might be an interpretation of the father's body language by a son who has a difficult relationship with his parent. Most narratologists use perspective in the broader sense that includes visual data without being limited to them.
The concept of perspective is especially pertinent to narrative. Narratives have at least one narrator and usually more than one character and thus offer the possibility for a range of, and a change of, perspectives. A narrator may tell the story from his own point of view, as in the following example: "A long time ago, little Stephen Dedalus, an inhabitant of Dublin, was eagerly listening to a story told to him by his father." But a narrator may also tell the story from the point of view of a character, as is shown by Joyce's A Portrait of the Artist as a Young Man; the Joycean narrator adopts the perspective of little Stephen: "His father told him that story: his father looked at him through a glass: he had a hairy face" ( 1926: 7). The point of view of a small child is indicated by the simple, repetitive syntax and by the periphrases "glass" for monocle and "hairy face" for beard.
Perspective is a complex and controversial concept, as is attested by the proliferation of rival metaphors such as "reflector" (James  1972: 247), "focalization" (Genette  1980; Niederhoff → Focalization), "slant," "filter," and "interest- focus" (Chatman 1990), or "window" (Jahn 1996; Fludernik 1996). One source of confusion and controversy, which is related to the spatio-visual origin of the term, is the ambiguity of the attributes "external" and "internal," pointed out by a number of scholars (e.g. Edmiston 1991: 155) but ignored by many more. In narratology, these terms are not used with reference to well-defined spaces (inside or outside a box) but with reference to minds (inside or outside a character's consciousness). However, the boundaries of a mind are less easily determined than those of a box. A character's consciousness can be directed inwards, as in meditation, but it can also be directed outwards, as in perception. In the latter case, the "internal" perspective pulls us straight back into the "external" world. A further difficulty is that the terms may refer both to points from which the action is viewed and to regions that are viewed from these points. Describing a point of view as "external," for example, suggests that we are viewing a character from the outside, from a spatial and possibly from an emotional and ideological distance. But this does not tell us how far our vision extends. In the case of the so-called camera perspective, it is extremely limited: we only learn what a newcomer to the scene might observe and thus have no way of knowing what the characters feel or think. In the case of omniscient narration, our vision is not limited at all. We have access to the characters' thoughts and feelings, including subconscious ones, as well as to every other conceivable region of the storyworld. Thus it is important, in analyzing perspective, to indicate not only a point or position from which the events are viewed, but also the kind of mind located at this position and the kind of "privilege" (Booth  1983: 160–63) this mind enjoys, i.e. its access, or lack of such, to the different regions of the storyworld.
A second reason why perspective is a difficult concept has been pointed out by Lanser. "Unlike such textual elements as character, plot, or imagery, point of view is essentially a relationship rather than a concrete entity. As it tends to evade stabilization into the language of 'things', it has been difficult to grasp and codify" (1981: 13). Analyzing an image in terms of perspective means analyzing it as a view, i.e. as the result of a relation between a viewing subject and a viewed object. Narratologists have occasionally succumbed to the temptation of simplifying things by reducing the relation to one of the elements connected by it.
Point of view is used in its technical sense, with reference to a narrative method, as early as 1866 (Stang  1961: 107–11). The first sustained discussion of the subject in English is to be found in the writings of James. However, "point of view" in James usually refers to a writer's temperament and outlook on life (cf. Morrison 1961/62: 247–48). When James discusses narrative method, he uses such related spatio-visual metaphors as "centre (of consciousness)," "window," "reflector," or "mirror," all of which refer to a character whose experience governs the presentation of the story. James prefers this kind of presentation to a first-person narrator ( 1972: 249), and he also advocates consistency in point of view, deploring his own deviation from such consistency in one of his tales as a "lapse from artistic dignity" ( 1972: 244).
James's disciple Lubbock ( 1972) systematized the master's critical observations into a coherent theory organized around an opposition between telling and showing, i.e. the traditional method of relating a story, in which the narrator is prominent (Plato's diegesis), and a new, quasi-dramatic method, in which the narrator retreats to the background (Plato's mimesis). Lubbock distinguishes four points of view, arranged here in a sequence from telling to showing and paraphrased in more up-to-date terms: (a) third-person narration with a prominent or authorial narrator; (b) first-person narration; (c) third-person narration from the point of view of a character, a Jamesian "reflector;" (d) third-person narration without comments or inside views (called "the dramatic method"). Lubbock does not recommend the fourth type, as one might expect an advocate of showing to do. He points out the sacrifices that this type entails, such as the difficulty of depicting the mental life of characters (256–57), and he comes down in favor of the third type, the reflector mode, which is also preferred by James. This type combines access to the mental life of the reflector character with a withdrawal of the narrator.
Lubbock is a spokesman for the Zeitgeist, especially as regards his predilection for showing over telling and the withdrawal of the narrator. The only conspicuous dissenter is Forster, who argues that novelists need not be consistent in their point of view and that narratorial comments and intrusions are legitimate ( 1990: 81–84). But this is a minority opinion. Even three decades later the premises and preferences established by James and Lubbock are still going strong. Friedman continues to advocate consistency in point of view and expresses a somewhat qualified predilection for showing as against telling. Like Lubbock, he uses this opposition as the principle underlying a range of no less than eight points of view ( 1967: 119–31):
(1) "editorial omniscience" (third-person narration with an intrusive narrator);
(2) "neutral omniscience" (similar to the first, with a less intrusive narrator);
(3) "'I' as witness" (minor character as first-person narrator);
(4) "'I' as protagonist" (protagonist as first-person narrator);
(5) "multiple selective omniscience" (third-person narration from the point of view of several characters in succession, as in Woolf's Mrs Dalloway);
(6) "selective omniscience" (third-person narration from the point of view of one character);
(7) "the dramatic mode" (third-person narration in scenic mode without inside views);
(8) "the camera" (like the previous, without a clear distinction).
Friedman's typology includes several different criteria: the knowledge of the narrator (which distinguishes 1 and 2 from 5 and 6); the frequency with which a narrator comments on or interrupts the story (1 vs. 2); the question of whether a narrator is also a character (3 and 4 vs. the rest); the narrator's importance as a character (3 vs. 4); constancy or change of point of view (5 vs. 6); etc. It is a moot point whether all of these criteria should be subsumed under the one umbrella term of point of view. Furthermore, it may be doubted whether each of the eight types can be situated at a particular point on a scale ranging from telling to showing. Why the "'I' as witness" should tell more and show less than the "'I' as protagonist" is far from evident. Nor is it obvious why these two are more remote from the telling mode than types (1) and (2). A first-person narrator who tells the story with hindsight and frequently comments on the action is a better example of the telling mode than "neutral omniscience," Friedman's second type.
A major objective of the Lubbock-Friedman school is the elimination of the narrator, in particular the avoidance of narratorial comment, which they regard as intrusive moralizing. A novel, according to these critics, should make the readers see or experience the story instead of telling them what to think about it. Booth delivers a trenchant critique of such claims in The Rhetoric of Fiction ( 1983), arguing that the elimination of ideology envisaged by the advocates of showing is a delusion. Narrative has, as the title of his book implies, a rhetorical dimension: it communicates views and values. Doing so in an overt way, with a visible narrator making explicit comments, is just as legitimate as doing so in a covert way, by opting for a first-person narrator or adopting the point of view of a character. In a similar vein, Weimann (1962) traces the historical development from narrators who speak their minds to narrators who adopt the point of view of a character; to Weimann, this development is a story of decadence and decline. Twenty years after these critics, Lanser (1981) restates their arguments with some new inflections. While Weimann argues from a Marxist standpoint, Lanser is inspired by feminism, and where Booth draws on rhetoric to situate the techniques of fiction within a broader framework, Lanser relies on speech act theory. Furthermore, she is no longer concerned with repudiating Lubbock and Friedman, but rather responds to structuralists such as Chatman and Genette. These differences notwithstanding, Lanser continues the case made by Booth and Weimann in that she endorses a study of point of view that includes its socio-political implications and the writer's ideological agenda.
A model that has been highly influential in the German-speaking world is Stanzel's typological circle, which was first proposed in  1971 and presented in its most elaborate form in  1984. In this version, the circle is organized around three diametrical lines (see illustr.). They represent three criteria, each of which results in a binary opposition yielding two terms: mode (narrator vs. reflector); perspective (internal vs. external); person (identity vs. non-identity of the narrator's and the characters' realms, i.e. first person vs. third person). The six terms resulting from the three criteria are placed at equidistant points on the typological circle. Three of them define the "narrative situations" that are privileged in that, empirically speaking, most extant narratives cluster around them. The external perspective corresponds to the authorial situation, the reflector mode to the figural situation, and the identity of the realms of existence of narrator and characters to the first-person situation. Thus each narrative situation is defined by one of the poles in the binary opposition resulting from the three criteria and also, to a lesser extent, by the two adjacent poles. The figural situation, for example, consists in the dominance of the reflector mode and is additionally characterized by an internal perspective and by the non-identity of the worlds of narrator and character ( 1984: 55).
Stanzel has always been given credit as an eloquent critic; his typological system, however, has not won much approval. Cohn, for example, points out that the criteria of mode and perspective are so close that they can be regarded as equivalent: a reflectorial mode implies an internal perspective, a narratorial mode an external one (1981: 176–80; cf. Genette  1988: 78–79). Cohn and other critics, such as Leibfried (1970: 246), have also suggested that Stanzel should allow for a free combination of his oppositions instead of enclosing them in a circle. This is especially obvious in the case of first-person narration, which comes in two different forms: an authorial one, in which narrators tell the story as they see it at the time of the narration, i.e. with hindsight; and a figural one, in which they render it the way they experienced it as characters in the story. In the typological circle, these two forms can be accommodated only as intermediate cases between the narrative situations, which is awkward. While it makes sense to posit a range of transitional cases between the authorial and the figural situation, no such range exists between the I-situation and the two other situations. A narrative may be a perfect example of both first-person and figural narration. Cohn, for one, has shown that free indirect thought, a form of thought presentation associated with the figural narrative situation, occurs in first-person narrative (1978: 166–72).
The free combination of distinctions is the hallmark of Genette's Narrative Discourse, the most influential contribution to narrative theory from the quarters of French structuralism. Genette attacks theorists like Friedman and Stanzel for locating such terms as first-person narration, the dramatic mode or figural narration within the same category as "points of view" or "narrative situations." Genette insists on separating questions and distinctions relating to the narrator ("voice" in his terminology) from those relating to perspective, arguing in favor of a free combination of narrator types and point-of-view types. Furthermore, Genette introduces a neologism, replacing perspective with focalization ( 1980: 185–211). He distinguishes three types of focalization, which differ primarily in the amount of information they allow the narrator to communicate. Zero or non-focalization, a reformulation of the traditional idea of omniscience, grants the narrator access to every conceivable region of the storyworld; internal focalization, roughly equivalent to Stanzel's figural narrative situation, means a restriction to the experience of one character; external focalization, similar to Friedman's camera, imposes an even greater restriction, for it precludes inside views and limits narration to what an ignorant and uninvolved observer might perceive. Genette adds a further distinction to the second or internal type, which may be either fixed (adhering to one character throughout the text), variable (shifting between different characters) or multiple (shifting between different characters while retelling the same event).
Genette's rigorous separation of narrators and focalizations has more than once been hailed as a Copernican breakthrough in narrative theory, but surprisingly few narratologists have been willing to accept the consequences of this separation. After all, it makes sense only if narrators and perspectives are distinct categories, in other words if the choice of a particular kind of narrator does not entail a particular perspective. Genette suggests that, in principle at least, his three focalizations may occur in first-person narration just as much as in third-person narration ( 1988: 114–29). However, scholars such as Fludernik (2001b: 621) or Cordesse (1988) disagree with this homological model. They argue that omniscience or zero focalization is not an option for first-person narrators, since they do not have access to other minds and are restricted to what they have learnt in the course of the story. Furthermore, Fludernik claims, following a suggestion by Cohn, that first-person narrators cannot exclude their own thoughts and feelings (Cohn & Genette  1992: 263). Even when a first-person narrator does not reveal them, rendering the story in the camera mode, the reader will attribute thoughts and feelings to him or her in the process of reading (Fludernik 2001a: 103).
A comprehensive treatment of focalization or perspective in first-person narrative is given by Edmiston, who comes to the following conclusions (1991: 168): zero focalization is possible (but has to be regarded as an infraction of a literary norm); internal focalization is also possible, with the experiencing I as the point-of-view character; external focalization in the Genettean sense is not an option, but there is the additional option of telling the story from the point of view of the narrating I (for which Edmiston enlists the now-unemployed term of external focalization). While these conclusions do not precisely confirm the homological model suggested by Genette, they would appear to corroborate his general stance of allowing for a relatively free combination of narrator and point-of-view options. It should also be kept in mind that the case for a restriction of point of view or focalization in first-person narrative is always based on the knowledge of the narrator. This, however, is only one facet or parameter of point of view. Furthermore, this case rests on rather commonsensical or realistic assumptions. Since most of us are willing to abandon such assumptions when it comes to narrative content, it is hard to see why we should be less broad-minded about narrative discourse. If we are willing to be entertained by invisibility cloaks, we should not demur at first-person narrators who are omniscient.
In addition to the debate about the applicability of Genette's classification of focalizations to first-person narration, there has also been a more general debate about the triple nature of this typology. Most narratologists seem to prefer a dual model to a triple one: see, e.g., Bal ( 1997: 148), Vitoux (1982), Rabatel (1997) or Schmid ( 2008: 137–38), all of whom distinguish, in different terms, between a narratorial and a figural perspective. What is eliminated from these dual typologies is the camera mode (Genette's external focalization), which, however, has been defended by Broich (1983). Interestingly, even some of those who are skeptical about the camera mode make subordinate concessions or distinctions which would appear to indicate that this mode is not a figment of the narratological imagination. While Bal compensates for the elimination of Genette's external focalization by introducing the concept of the focalized object, Vitoux grants the narrator a "play of focalization" (359), which includes external focalization as one of its options. Finally, Rabatel allows for an external vision both within narratorial and figural focalization (101–02).
A major tendency in recent work on perspective is an increasing awareness of the diversity of the phenomenon. Scholars elaborate on the basic types of the various classifications by discussing changes from one type to another, intermediate cases, embeddings, transgressions or unusual combinations. One method of accounting for the complexity of narrative perspective is to distinguish its different facets or parameters. Schmid, who builds on earlier studies along these lines by Uspenskij, Lintvelt and Rimmon-Kenan, discerns five such parameters: space; ideology; time; language; perception ( 2008: 123–37). The point of distinguishing these parameters is that they are not necessarily in line with each other. A narrative may report events as they are perceived by a character, while at the same time using language that is very remote from that of the character. This is the case of James, as was pointed out long ago by Scholes et al. ( 2006: 270) and recently reiterated by Aczel (1998). James's novel What Maisie Knew tells us what Maisie knew, but it does not show us how Maisie spoke. Of course, the different parameters may also be in line, as in the beginning of A Portrait of the Artist, where the narrator renders a child's perceptions in a child's language. The alignment of parameters is referred to as "compact perspective" by Schmid, their dissociation as "distributive perspective" ( 2008: 151–52). It should be added that scholars who favor the parameter approach to perspective are not in full agreement about the distinction and the number of parameters. Thus Fowler, who reviews Uspenskij in similarly favorable terms as Schmid, argues that the parameter of "phraseology" (corresponding to "language" in Schmid's quintuple division) is not a separate parameter, but is inextricably bound up with the others. "By separating off 'phraseology,' the theorist simply expresses nostalgia for the text as decorative form" (Fowler 1982: 226).
An interesting recent development initiated by Nünning (2001) and followed up by Surkamp (2003) is an attempt to enlist Pfister's theory of perspective in drama for the study of narrative. Ironically, this theory was initially motivated by the inverse attempt to enlist a narratological concept for the study of drama (Pfister  2000: 57). According to Pfister, the perspective of a character in a play is constituted by psychological disposition, ideology and the awareness of what the other characters are up to. As the combination of these three factors will differ from one character to another, they will view and judge the same event in different ways. Pfister's interest is not so much in individual perspectives as it is in the differences or similarities between them. The point is to establish the structure of perspectives, to hear the more or less harmonious concert that is performed by the voices of a play. Structure may be non-perspectival, approximately realized in some medieval moralities (all of the characters function as authorial mouthpieces, lacking individual perspectives); it may be closed (different perspectives are hierarchically structured around a privileged perspective, which is either explicit, i.e. articulated by one of the characters, or implicit); finally, it may be open (a hierarchy between the different perspectives is lacking so that no privileged perspective emerges). Nünning transfers Pfister's theory to narrative, with some minor adjustments concerning the relations between the perspectives and one major adjustment. Narrative features not only characters, but also a narrator whose perspective is defined, in similarly broad terms as that of a character, as a set of "psychological idiosyncrasies, attitudes, norms and values, a set of mental properties, and a world-model" (Nünning 2001: 213).
Nünning certainly has a point, for one thing because the texts that most narratologists deal with include dialogue and are thus partly dramatic. Even as far as the strictly narrative portions of the text are concerned, Nünning's approach is valuable in that it alerts us to the potential plurality and diversity of perspectives. However, the dramatic analogy can also be misleading. In drama there are roughly as many perspectives as there are characters who speak. In narrative, however, the mere existence of a character does not imply that his or her perspective is of any importance. If we learn that a character is a teenage girl, we can make certain assumptions about her knowledge, her interests, her values, etc. But this only turns into a perspective when we learn about her views of the world around her. How prominent her perspective becomes also depends, of course, on the way her views are represented—with lofty disdain, with amusement, or with sympathy. Another problem in Nünning's approach to perspective is a potential loss of the relational quality of the concept. When he writes that character perspective "embraces everything that exists in the mind of the character" (2001: 211), there is a strong shift in the direction of the viewing subject and a danger of abandoning the relational character of the concept pointed out by Lanser (1981: 13). To sum up, perspective structure provides us with a chart of the potential perspectival reference points of a text, whereas the more traditional narratological accounts of perspective analyze where the narrator situates the representation of the story in relation to these points or how he or she makes it move between them.
An intriguing question about perspective is asked by Bonheim: Why is it that in some narratives point of view seems to be highly important and significant, while in others it appears to be negligible? (1990: 300). In other words, can there be narrative without perspective or with a less conspicuous one? If we think of the concept in purely spatio-visual terms, the answer is not difficult. One can tell a story without a fixed viewpoint in the literal sense, just as one can paint a landscape without perspective. This has been demonstrated by Stanzel ( 1984: 117–22) and by Jahn (1999: 95–100), who elaborates on Stanzel's binary distinction between perspectival and non-perspectival depiction of space with a scalar model. Similarly, if one thinks of focalization in terms of restriction of information, then Genette's zero focalization, the equivalent of non-perspectival narrative, would appear to be a possibility; at the very least, focalizations may be more or less restricted. Based on these premises, the answer to Bonheim's question is that narrative can be perspectival and non-perspectival, focalized or non-focalized, and also something in-between. However, the answer becomes more difficult if we follow Uspenskij or Lanser in defining perspective in a more inclusive manner. If perspective also has an ideological dimension, a narrative without perspective is hardly possible. However, even on the premise of a more inclusive definition, perspective may be more or less conspicuous. A writer can foreground it by assigning it to a character not usually selected for such purposes, e.g. a toddler, as in A Portrait of the Artist, or an animal, an interesting case discussed by Nelles (2001) and Burns (2002). Arguably, every shift of perspective from a narrator to a character has a foregrounding effect, even if the character is of a thoroughly unremarkable sort. Schmid argues that in comparison with narrator perspective, character perspective is "marked" in the linguistic sense ( 2008: 138). Perspective is foregrounded precisely when it is perceived as a perspective, i.e. as a limited or partial view among other views of the matter that are equally possible. When a narrator adopts a character's perspective, the latter's view will be contextualized and qualified by the mere fact of the narrator's presence: it will appear not as the view, but as one view. Of all literary genres or modes, narrative seems to be the one most suited to create this effect, which is not the least of its attractions.
(a) It has been observed that camera narration alias external focalization is employed only for part of a text, very often the beginning. It would be interesting to study the transitions where this mode is abandoned. Is the transition to a narratorial or to a figural point of view? The former possibility would confirm Vitoux's (1982) claim that camera narration is merely an option in the play of narratorial focalization. (b) The study of perspective has focused almost exclusively on fictional narrative. An exploration of the subject in non-fictional narrative genres might yield interesting results in its own right and also throw new light on the phenomenon in fiction. (c) When narratologists review the work of their predecessors, they usually focus on the gaps and the mistakes. Previous theories are demolished or quarried for the purpose of building a new one. This does not make for a fair appraisal of the critical tradition. Perhaps it is time for a non-partisan history of theories of point of view and related metaphors from James (or earlier) to the present day, preferably by someone who makes a vow not to conclude the study with a new theory or typology of their own.