4
Temporality, Particularity, and Narrative
An Excursion into the Theory of Text Types
(ii) Event sequencing. Narrative representations cue interpreters to draw inferences about a structured time-course of particularized events.
From Contexts of Narration to Narrative as a Type of Text
To approach the second basic element of narrative identified in my thumbnail sketch in chapter 1, the present chapter turns from discussion of my main case studies to more general considerations of the concept of “text types” on which the overall approach is based. To explore where stories fit within a larger constellation of textual kinds – to identify the patterns of event sequencing that are distinctively associated with narrative – I examine commonalities and contrasts among describing, narrating, and explaining, viewed both as cognitive activities and as forms of communication, that is, text types embedded within interactional, social, institutional, and other contexts for communicative practice. Further, the chapter relates issues of text types to research on categorization processes in cognitive science, exploring what might constitute the “basic” level of a taxonomy of text types (i.e., the most cognitively fundamental level), and also what might be considered prototypical features of descriptions, explanations, and stories (i.e., the features found fully realized in exemplars or standard cases of these types). My account suggests that, like examples of birds, cups, and chairs, instances of the category narrative adhere to a logic of graded centrality, with specific story artifacts (or story-like mental representations) being more or less prototypically narrative in nature.
Having discussed in my previous chapter the dialectical relationship between narrative texts and their contexts of production and interpretation – the way narrative transactions, regardless of medium, are always and irreducibly grounded in particular discourse contexts or communicative occasions – in this chapter and the ones that follow I shift my emphasis to the text side of this text–context dialectic, without, however, assuming that texts can be accessed or interpreted apart from particular contexts of interpretation.1 Rather, my focus shifts from consideration of how communicative contexts shape and are shaped by the narratives circulating within them to analysis of the critical properties of texts that do circulate in the manner characteristic of stories. To put this point still another way, although contexts or occasions help determine the meaning or function of storytelling acts, analysts can work to identify structural properties of those (situated) acts themselves – properties thanks to which they function as instances of narrative rather than as syllogisms, recipes, warehouse inventories, sayings in greeting cards, or mathematical theorems, as the case might be.
The elements of event sequencing, worldmaking/world disruption, and what it’s like, the focus of this and my next two chapters, are meant to capture some of these critical properties of narrative viewed as a type of text as well as a cognitive structure.2 In the present chapter, I begin the process of characterizing key traits of narrative by sketching a map of the place of stories within a broader ecology of kinds of texts. More specifically, to explore where stories fit within a larger constellation of text types – to determine what methods for sequencing events might be distinctively narrative in nature – I examine commonalities and contrasts among describing, narrating, and explaining, viewed both as cognitive activities and as forms of communication, that is, text types embedded within interactional, social, institutional, and other contexts for communicative practice. Although my analysis considers only three text types and is meant to be illustrative rather than exhaustive, it will allow me to zoom in on distinctive features of stories. More specifically, a text-type approach can bring into relief key issues associated with the element of event sequencing; the approach highlights the specific kind of temporal structure that functions as a critical property of narrative representations, but not descriptions, as well as the concern with particularized events (rather than general patterns and trends) that sets stories apart from certain types of explanations. Building on this discussion of narrative’s distinctive temporal structure and its focus on particularity, my next two chapters turn to other key properties of the narrative text type, including the way it functions as a blueprint for worldmaking and its foregrounding of an experiencing consciousness within the storyworld thereby evoked.
As I discuss in more detail in my next section, text types such as description, narrative, and explanation stand in complex relations with one another and with other representational strategies and artifacts. In line with research in the cognitive sciences that I go on to review – research suggesting that people make sense of things in the world by grouping them into multi-level systems of categories – this network of relationships among kinds of texts can be plotted along horizontal and vertical axes. Vertically, the text types in question are members of a category subordinate to the category that contains all the phenomena classifiable as texts, but superordinate to the category containing, say, particular kinds of narratives or explanations.3 In other words, (kinds of) text types have a place within a hierarchical taxonomic system, with the categories becoming narrower or more specific as you move downward through the hierarchy:
texts
text types (e.g., narrative)
genres (e.g., science fiction)
subgenres (e.g., cyberpunk fiction)
Horizontally, realizations of text types can be more or less clearly distinct from one another, depending on how many features a given artifact or representation shares with central, prototypical instances of the type of which it is a token. Furthermore a single text can simultaneously embody several text types through operations of conjunction, alternation, embedding, etc. Novels, for example, can contain many descriptive passages yet still be considered narratives; in this case, a text that is globally narrative embeds local instances of description, with narrative segments of the text being conjoined to or alternating with descriptive segments.4 Conversely, describing a person’s everyday behavior might entail quoting the humorous anecdotes he or she likes to tell; in this case, a text or discourse that globally realizes the category of description embeds local instances of narration. Explanations of physical or social processes may also embed detailed descriptions of their initial conditions and their effects. Yet a text, discourse, or mental representation that addresses the central concern of explanations, namely, the question of why?, differs in crucial respects from one that addresses the central concern of descriptions, namely, the question of what? (see Hempel [1948] 1998). Less intuitively clear is how telling a story about a situation or an event relates to explaining that situation or event.5 I consider the nature of the relationship between narrative and explanation in a “coda” on narrative and science provided in the final part of this chapter.
In the sections that follow, after discussing recent work on text types, I relate that work to research on what cognitive theorists have termed basic-level and prototype effects in categorization processes; here I explore what might constitute the “basic” level of a taxonomy of text types (i.e., the most cognitively fundamental level), and also what might be considered prototypical features of descriptions, explanations, and stories (i.e., the features found fully realized in standard cases of these types). I argue that, like other members of a category, instances of the category narrative adhere to a logic of graded centrality, with specific story artifacts (as well as story-like mental representations) being better or worse examples of narrative. Accordingly, whereas prototypical instances of the category narrative (what Hogan 2003 terms exemplars) share relatively few features with those of description, more peripheral cases are less clearly separable from that text type, allowing for hybrid forms such as “descriptivized narrations” and “narrativized descriptions” (Mosher 1991). A further question is whether narrative shares a similarly porous boundary with the text-type category explanation, making possible “narrative explanations” (Adams 1996) that blend attributes of these two kinds of texts. In any case, the existence of hybridized or “fuzzy” forms does not mean that narrative is, at a deep level, equivalent to description or explanation. Rather, such forms again suggest the graded versus binarized, either-or nature of text-type categories, and the way non-prototypical instances of those categories can verge on neighboring textual kinds.
Building on the taxonomy outlined earlier in the chapter, especially the account of core features of explanation, the coda at the end of the chapter uses the taxonomy to investigate how narrative relates to description and explanation in contexts of scientific inquiry in particular. Here my discussion uncovers something of a paradox or at least a tension: whereas some aspects of scientific inquiry may be indissolubly bound up with narrative modes of representation, other kinds of scientific explanation appear to be not just non-narratively structured but also antithetical or at least resistant to narrativization – to being conceptualized in narrative terms.
Text Types and Categorization Processes
The first part of this section reviews work on the concept of text types. I then connect that work with research on categorization processes and the vertical as well as horizontal relations between categories within taxonomies, or systems for organizing the world into kinds and instances of those kinds. I focus on two key concepts: basic-level effects, whereby one category within a hierarchy of levels (such as texts > text types > genres > subgenres) functions in a more cognitively basic way than the categories at other levels, and prototype effects, by which instances of the same category may be more or less prototypical examples (or “exemplars”) of that category. My argument is that both sorts of effects are relevant for understanding how narrative provides a specific method for sequencing events – this method being one of the distinctive properties of stories that in turn accounts for how narrative relates to other types of texts.
Categorizing texts: an overview
In Chatman’s account, a text can be defined as “any communication that temporally controls its reception by the audience” (1990: 7), i.e., “a time-regulating structure” (1990: 8). But different kinds of texts regulate time in different ways, and one of the motivations of the theory of text types is to capture the differences at issue.
Theorists of texts types have proposed a number of taxonomic principles and nomenclatures (see, e.g., Adam 1985; Chatman 1990; Fludernik 2000; Görlach 2004; Virtanen 1992; Virtanen and Wårvik 1987; Werlich 1975, 1983). In this book I assume that text types are broader in scope than literary genres (Bildungsroman, the psychological novel). Instead, they can be compared with the “primary speech genres” characterized by Bakhtin ([1953] 1986: 60) as relatively stable types of utterance which develop within particular spheres of language use, and from which “[s]econdary (complex) speech genres – novels, dramas, all kinds of scientific research, major genres of commentary, and so forth – arise in more complex and comparatively highly developed and organized cultural communication (primarily written) that is artistic, scientific, sociopolitical and so on” ([1953] 1986: 62). For his part, Görlach (2004) defines text type as
a specific linguistic pattern in which formal/structural characteristics [e.g. the lexical patterns found in patient histories written by physicians, or the question-answer turn-taking sequence of police interrogations] have been conventionalized in a specific culture for certain well-defined and standardized uses of language so that a speaker/hearer or writer/reader can judge:
a) the correct use of linguistic features obligatory or expected in a specific text type...
b) the adequate use of the formula with regard to topic, situation, addressee, medium, register, etc.;
c) the identification of intentionally or inadvertently mixed types, or their misuse;
d) the designation of the text type [e.g., speakers not only know what features characterize a political speech but also know the name]. (2004: 105)
As Görlach notes, judgments of this sort form part of language users’ overall communicative competence; hence “the lack of knowledge of textual conventions can carry the same degree of stigmatization as the incorrect use of syntax or pronunciation” (2004: 105). In an academic debate, for example, foul-mouthed exchanges of insults are neither obligatory nor expected (a), nor am I likely to address my interlocutor in such a debate using the formula Hey, bud or Yo, dude (b), unless I am inadvertently or intentionally (humorously) mixing the text types of public academic debate and streetcorner dispute (c). Further, my ability to frame this example, and to refer to academic debates versus street-corner disputes by name, stems from my knowledge of the repertoire of text types that constitutes my overall communicative competence (d), i.e., my knowledge of what to say, how, and under what circumstances (cf. Hymes 1974; Saville-Troike 2002).
Insofar as text types are heuristic constructs used to make sense of more or less heterogeneous semiotic practices, text-type categories can be related to categorization processes in general – that is, the processes whereby people use categories to make sense of the world. Originating in the work of cognitive anthropologists, anthropological linguists, and psychologists examining categorization processes both intraculturally and interculturally, the study of category systems has highlighted the extent to which humans rely on categories in everyday perception, reasoning, and communication. Inspired by the pathbreaking work of Eleanor Rosch and her colleagues (Rosch 1973, [1978] 2004; Rosch et al. 1976; cf. Lakoff 1987, [1987] 2004), theorists working in this tradition have focused on two key aspects of categorization processes; these aspects have been termed basic-level effects and prototype effects.6 On the one hand, basic-level effects arise from the way categories are arranged hierarchically in terms of degrees of inclusiveness; in such hierarchies, one level of the system can be described as being more cognitively basic than the others, and certain “effects” flow from this hierarchical structure: e.g., in empirical tests, it is the level at which members of the category are most rapidly identified. Such basic-level effects, discussed more fully below, thus manifest themselves along the “vertical” axis of a category system and apply to the basic category tree versus subordinate categories maple or red maple – and also versus the superordinate category living thing. On the other hand, prototype effects concern not the hierarchical relations among more and less inclusive categories but rather the relations among more or less prototypical instances of the same category – and also among neighboring categories themselves. Research on this second kind of effect suggests that certain members of categories tend to be perceived as more central or prototypical than other members (cf. robins versus emus as instances of the category bird). Further, the research suggests that boundaries between categories are permeable, such that less standard cases of neighboring categories can be difficult to situate in one category versus the other – as is the case with certain non-prototypical instances of the category tree as compared with exemplars of the category shrub.
In my next subsection, I explore how accounts of these two kinds of effects within category systems can help illuminate relations among text types such as narrative, description, and explanation.
Are text types cognitively basic?
Cognitive-anthropological research (Atran 1990; Berlin 1992; Berlin, Breedlove, and Raven 1973; Ellen 1993) has revealed cross-cultural regularities in the structure of folk taxonomies of living organisms – regularities pointing to basic-level effects in category systems. In Atran’s characterization, a culture’s folk taxonomy consists of “a stable hierarchy of inclusive groups of organisms, or taxa, which are mutually exclusive at each level of the hierarchy” (1999: 317). These levels or ranks encompass FOLK KINNGDOMS (e.g., animal, plant), LIFE FORMS (insect, mammal, bird, tree), FOLK GENERICS (mosquito, dog, starling, oak), FOLK SPECIFICS (schnauzer, white oak), and FOLK VARIETALS (miniature schnauzer, swamp white oak). Across cultures hierarchically differentiated ranks, not the taxa (groups of organisms) that they contain, are universal. Further, names for taxa of the same rank tend to have (across different languages) similar morphological/lexical properties. Whereas most folk generics are assigned names consisting of unanalyzable lexical stems (such as oak or dog, in English) subordinate taxa usually receive binomial or polynomial labels, which consist of an attributive word or phrase attached to a lexical stem: willow oak, southern red oak (for a more detailed account of linguistic reflexes of folk taxonomies, see Herman and Moss 2007).
The similar structure of names for folk generics across different languages (as attested in Malt 1995) provides evidence that categories at this (intermediate) level of taxonomic systems are cognitively fundamental, that is, “functionally and epistemologically primary with respect to the following factors: gestalt perception, image formation, motor movement, knowledge organization, ease of cognitive processing (learning, recognition, memory, etc.), and ease of linguistic expression” (Lakoff [1987] 2004: 144, my emphasis). In other words, the research on basic-level effects suggests that “categories are not merely organized in a hierarchy from the most general to the most specific, but are also organized so that categories that are cognitively basic are ‘in the middle’ of a general-to-specific hierarchy. Generalization proceeds ‘upward’ from the basic level and specialization proceeds ‘downward’ ” (Lakoff [1987] 2004: 144). In this model, the middle level of a taxonomic hierarchy such as
superordinate level = animal
basic level = dog
subordinate level = retriever
has the following attributes – it is:
the highest level at which category members have a similarly perceived overall shape;
the highest level at which a single mental image can reflect the entire category;
the highest level at which a person uses similar motor actions for interacting with category members;
the level at which subjects are fastest at identifying category members;
the level with the most commonly used labels for category members;
the first level named and understood by children; and
the first level to enter the lexicon of a language. (Lakoff [1987] 2004: 168)
But how might this work supporting the notion of basic-level categorization bear on text-type classifications – and on narrative texts in particular? Prior to being made the object of academic study (for example, through accounts of literary genres or analyses such as the one outlined in the present chapter), a culture’s system for classifying text types and their superordinate and subordinate categories is likewise a folk taxonomy. In parallel with other taxonomies, such classificatory systems are in principle subject to basic-level effects. But consider whether a taxonomy like the following captures categorization processes underlying the use of texts, at least in some cultures and subcultures:
superordinate level = “time-regulating” phenomena (Chatman 1990: 7) interpretable as texts (print texts, cinematic texts, spoken discourses, structured mental representations that could in principle be textualized, etc.)
basic level = text types (narrative, description, explanation, instruction, etc.)
subordinate level = genres (detective novel, Bildungsroman, legal explanation, forensic explanation, etc.)
Some reflexes of basic-level effects appear to be absent from this taxonomy, including reflexes associated with category labels. Hence among the terms that Görlach (2004: 79) lists as definitions of story, several are in actuality competing labels for this text type, used in more or less free variation with story: narrative, account, tale. In other words, it would be difficult to establish that any of the lexical items in the set story, narrative, account, tale is more “basic” than any of the other items. Yet in other respects – e.g., categorical representativeness and motor responses – basic-level effects might in fact be attributed to members of the middle or text-type-level category in this taxonomy. It could be argued that the text-type level is the highest level at which a single mental image can reflect the entire category, with the concept “narrative” reflecting key properties of all the narrative genres and subgenres that can be ranged under that category. Likewise, it could be argued that what Meir Sternberg (1990, 1992, 2001) characterizes as the narrative universals of curiosity, suspense, and surprise orient my (simulated) motor responses to agents and events when I transpose my spatial and temporal coordinates to those orienting characters within mentally projected storyworlds (no matter what the specific genre or subgenre of narrative involved) – whereas descriptions and explanations do not trigger responses of this kind.7 Then again, when I consult my own intuitions about the highest level at which inferences about overall shape first become possible, I find that I am more likely to draw those inferences at a lower level of the taxonomy – in connection with the plot structure of detective novels, say, or the processes of identity formation in the Bildungsroman.8 So which is more basic, in the sense at issue: the text-type level or the level of genres?
The mixed picture that emerges from this taxonomic thought-experiment may account for the proliferation of text-type classifications among text linguists, for example.9 The research reveals a lack of convergence in this context – i.e., a lack of consensus about how to create a hierarchy of levels in this domain of categorization, let alone a common list of items within each such level. We may therefore speculate that ways of understanding the domain of textual phenomena are changing more rapidly than ways of understanding and categorizing the domain of biological organisms, for example. Similar intuitions seem to motivate what might be characterized as the anti-taxonomy sketched by Barthes ([1968] 1977), who uses the capitalized word Text as a mass noun, like water or space, and the lower-case word text as a count noun, like cat or pencil. In this way, Barthes suggests particular texts are merely strategic demarcations of a generalized textual field or, rather, process – a citational process by which ever more Text is produced. The work, any particular textual instantiation, now becomes epiphenomenal or “the imaginary tail of the text” (Barthes [1968] 1977: 156–7). As Barthes puts it, “the Text does not stop at (good) Literature; it cannot be contained in a hierarchy, even in a simple division of genres. What constitutes the Text is, on the contrary (or precisely), its subversive force in respect of the old classifications” ([1968] 1977: 157).
In characterizing the Text as superordinate to specific literary genres, Barthes builds on his own earlier account of narrative as a transor super-generic phenomenon in “Introduction to the Structuralist Analysis of Narratives” ([1966] 1977), where narrative stands out as an object of inquiry precisely because of the way it occupies a level above particular story genres and media, whether popular or elite, traditional or avant-garde. Situating narrative at a super-generic level allowed Barthes to explore what he took to be critical properties of narrative in general as opposed to novels, autobiographies, news broadcasts, etc. But this strategy invites, in turn, questions about prototype effects arising from “horizontal” relations among narratives and other types of texts, including descriptions and explanations.
Text types, category gradience, and prototype effects
Just as basic-level effects are evident along the vertical axis of text-type taxonomies, prototype effects are evident along the horizontal axis. They concern not the hierarchical relations among more and less inclusive categories but rather the relations among more or less prototypical instances of the same category – and also among neighboring categories themselves.
As Lakoff (1987: 12–22) points out, and as sketched preliminarily in chapter 1, along the horizontal dimension of taxonomic systems, relations among categories and category members are subject to two forms of gradience, membership gradience and centrality gradience. These two forms of gradience are defined, respectively, by “the idea that at least some categories have degrees of membership and no clear boundaries” and by “the idea that members (or subcategories) which are clearly within the category boundaries may still be more or less central” (Lakoff [1987] 2004: 144). For example, members of the category tall people will belong to that category on a more-or-less basis (is someone who stands 5 feet 11 inches [1.8 meters] tall or not?), and there will be no clear dividing line between the categories tall people and people of medium height. Rather, a person who is 6 feet 6 inches will, by the standards of current-day North America, be “more” (or more strongly) a member of the category of tall people than will someone who is 5 feet 11 inches. By contrast, unlike the category tall people the category bird is a bounded one; despite having the ability to get airborne, neither a plane, nor a human cannonball at the local carnival, nor a flying squirrel is a bird. However, particular birds will still be more or less central or prototypical realizations of the category, with robins and sparrows having more of the prototypical features of bird than emus or penguins, neither of which can fly. Bird, then, is subject to centrality gradience but not membership gradience, whereas tall people is subject to membership gradience; this in turn entails that certain instances of the category tall people will be more representative or “central” than others, with centrality now being defined as full possession of the properties that members of the gradient category can have on a more-or-less basis.
Hogan (2003: 45–7, 71–6) offers an excellent synopsis of some of the issues raised by research in this area. As Hogan notes, prototypes are in essence standard cases. But the cognitive process by which those cases come to be viewed as standard is richly structured. As Hogan puts it:
prototypes add “average” properties to defaults.... [This] averaging is “weighted” by saliency. The prototypical man for any given person will involve average properties, not of all men, but of men who are highly salient in that person’s experience.10 . . . Perhaps more interestingly, the weighting of averages is also bound up with contrast effects. In other words, the saliency of particular instances of one category is in part a function of their contrast with instances of some opposed category.... Thus, if a square jaw is associated with men and not women, then the prototypical man will have a squarer jaw than will the actual statistically average man. (2003: 46)
As Hogan’s characterization suggests, although processes of categorization in general and assessments of relative prototypicality in particular may be grounded in basic cognitive abilities and dispositions, contexts of interpretation affect judgments about how specific instances of categories pertain to prototypes and where to draw the boundary between a non-prototypical instance of one category and an equally peripheral member of a neighboring category. Prototype effects, accordingly, may emanate from underlying cognitive capacities and tendencies, but they are also anchored to features of the situations in which those abilities and dispositions are deployed. This way of thinking about prototypes raises key questions: how are judgments of what constitutes a prototypical narrative affected by the contexts in which those judgments are formulated, and what sorts of contrast effects come into play with judgments about (or analyses of) standard cases of stories vis-à-vis descriptions, explanations, or other text-type categories?
The contextual grounding of prototype effects can be illustrated by dropping down one level in our hierarchy of textual phenomena, from text types to genres. Theorists of literary genre have long recognized the presence of prototype effects in systems for classifying genres. According to Jacques Derrida, in an argument that perhaps underestimates the role of exemplars or standard cases in generic systems, a given text submits itself only more or less to the law of any particular genre (Derrida [1980] 1991), orienting itself to multiple generic norms simultaneously. Conversely, genre distinctions make their presence felt even (or perhaps especially) in texts that actively violate dominant generic conventions (cf. Dubrow 1982: 3–4). Suppose a western were to include a scene with alien invaders or a hardboiled detective novel were to recruit extensively from the techniques used in eighteenth-century sentimental fiction. The subversive force of such generic transpositions becomes palpable only because they cut against the grain of established categorization practices, which organize people’s understanding of the textual domain and establish cuts or boundaries between textual kinds. Yet as reception theorists as well as students of genre have discussed, such boundaries between genres may change over time, since particular texts can claim generic status while also pushing against the boundaries of established generic norms.11 To paraphrase Jauss (1982): the novel was never the same after Joyce’s Ulysses.
The appearance of generically creative or subversive works like Joyce’s can reset the system of generic classifications, demonstrating how context impinges on assessments of prototypicality. In other words, what counts as a novel (for example) can be altered by texts eventually if not immediately recognized as instances of the novel form. Ulysses altered the conditions for novel-writing in general, as attested by such follow-up works as Woolf’s Mrs Dalloway (1925), Döblin’s Berlin Alexanderplatz (1929), and Faulkner’s The Sound and the Fury (1929). This process suggests that different contrast effects came into play at different points in the history of Ulysses’ reception. Its degree of prototypicality as a novel increased as the generic categories that formed the main comparison set were themselves reconfigured by the publication of Joyce’s novel and other, affiliated, novels; those novels’ focus on the flux of consciousness reduced the contrast effect between e.g. novelistic discourse and the expressive dimensions of lyric, and thus the saliency of features (action, plot) which had been factored into the determination of what made for a prototypical novel.
In short, generic categories function more like tall people than bird, with cut-off points between one genre and another being subject to variation over time. But, moving back up one level in the hierarchical systems, are instances of text-type as well as generic categories also gradient in nature? Is it the case that texts, discourses, or representations can be more or less narrative in kind, such that the frontiers of narrative are less like borders between autonomous nation-states than contested property lines between neighboring communities, which share some practices but not others? When compared with central, prototypical instances of the narrative text type, are certain non-prototypical narrative representations less readily distinguished from descriptions – in the same way that some trees are quite shrub-like? Further, to return to Hogan’s (2003) synopsis of recent research on prototypes, how do the issues of contextual grounding and contrast effects play out at the level of text types? Are different kinds of features going to be identified as prototypically narrative or descriptive or explanatory depending on context? In parallel with the way the standard case of dog changes when one moves from the farm in Maine to an apartment in Manhattan, does the prime exemplar of the category story change when one moves from the family dinner table to a scriptwriting class? And will different sorts of contrast sets cause different features of narrative to be weighted with special salience? Does my own analysis weight particular aspects of narrative more heavily because I have chosen description and explanation as my contrast set, whereas different critical properties might be foregrounded if the contrast set consisted of drama and lyric?
In the following sections I signal the relevance of these questions while moving forward with my more general account of how research on categorization might be synthesized with text-type theory to explore basic elements of narrative.
Narrative as a Text-Type Category: Descriptions versus Stories versus Explanations
In this section, I explore conceptual underpinnings of the approach developed in this book, a thumbnail sketch of which was presented in chapter 1. Specifically, I outline what can be construed as features of prototypical instances or standard cases of the three text-type categories under consideration, description, narrative, and explanation – though, again, I acknowledge the relevance of context and of contrast effects when it comes to the identification of prototypes. Attempting to isolate core features of each textual kind, I also examine how forms of category gradience might apply in this domain. To reiterate, since my next two chapters investigate the elements of worldmaking/world disruption and what it’s like, I focus here on the role of event sequencing in stories – and discuss the extent to which a specific method of sequencing events distinguishes the narrative text type from descriptions and explanations.
Description
Description can be conceived as a cognitive activity that may or may not be realized as a discourse or text falling within the text type description. If textualized, descriptions can in turn be embodied in a variety of media and shaped by diverse representational conventions, from mathematical symbolism and flow charts to inventories of personality traits and ethnographic practices.
As Chatman notes, description has long held a special fascination for narratologists, who tend to treat it as “the most interesting of the other text-types because its relation to Narrative is the most subtle and complex” (1990: 16). As I go on to discuss, the area of overlap between narrative and description can indeed be substantial, since representations and discourses falling under both rubrics have the net effect of coupling properties with situations, events, or objects (compare ascribing to someone the trait of being poor against telling the story of how he or she came to be impoverished). But Lukács ([1936] 1970), in his critique of modernist fiction, emphasized the difference. For Lukács, to describe constitutes in some contexts a refusal to narrate, when “mere” description is measured against the realist narrative practice of emplotting events within a larger chain of occurrences in a richly social context. In making this argument, Lukács builds on a (rhetorical) tradition that, as Hamon (1982) and Chatman (1990: 23) point out, long viewed description as secondary or derivative, i.e., as inferior to (or an interruption of) what is properly narrative. Again, however, contrast Sternberg’s (1981) functionalist account, according to which descriptive and narrative functions can be fulfilled by one and the same textual structure – and reciprocally, many different structures can serve, say, a descriptive function – depending on the particularities of communicative contexts. From this perspective, attempts to draw an invidious distinction between narrative and description are misguided, especially since, as discussed below, context alone determines whether one and the same statement functions as a description of some aspect of a storyworld or rather as narration via a character’s thought-processes.
For his part, Pflugmacher suggests that “[d]escription is a text-type which identifies the properties of places, objects, or persons” (2005: 101). Pflugmacher’s definition might be expanded as follows: representations and discourses that are central instances of this text-type category entail the ascription of properties to entities within a mental model of the world (whatever the modality status of that mentally projected world – e.g., real, fictional, dreamed, etc.). Furthermore, such ascriptions can be either static or dynamic. On the one hand, stative propositions involve property ascriptions at a moment in time or, as in (1), ascriptions of enduring attributes over an undifferentiated span of time.
1 Water is H2O.
On the other hand, active propositions, or process statements, involve property ascriptions over a differentiated time-course within the mentally projected world; cf. (2) and (3):
2 The trees lost their leaves during the transition from fall to winter.
3 Mondays, Wednesdays, and Fridays I have toast for breakfast, but on Tuesdays and Thursdays I have cereal.
(2) and (3) can be captured by Pflugmacher’s definition if they are translated into their underlying logical structure, along the lines of “At time t the trees possess property L, and at time t + 1 property ~L (where L = being fully leaved and ~L = being leafless).” Here, however, the limits of both the classical accounts of description and the classical model of categorization become evident. Precisely because descriptions can focus on temporally emergent phenomena, the boundary between description and narration should be thought of as porous and variable rather than as impermeable and fixed (cf. Genette [1966] 1982; Kittay 1981; Ronen 1997). Likewise, specific instances of description and narrative will pertain to these text-type categories in a gradient, more-or-less rather than binarized, all-or-nothing manner.
Pointing to forms of category gradience, Mosher (1991) identified mixed modes that he termed descriptivized narration and narrativized description, already discussed in chapter 1. Mosher’s analysis suggests the relevance of centrality gradience for members of text-type categories: narrativized descriptions and descriptivized narrations are neither prototypically descriptive nor prototypically narrative. But what is more, it is not always clear which text type is dominant and which is subservient in a given discourse context, raising the question of whether a description is being narrativized or a narrative descriptivized. In these cases membership gradience also makes its relevance felt. If I read a clause such as the house was dark in a novel, only context can determine whether this is a statement by the narrator ascribing a property to the house or, rather, a report of some perceiving agent’s reaction to the house at a specific point in the unfolding action (cf. Jahn 1997). In other words, is this clause a piece of descriptive text embedded within and subserving a more global narrative, or is the clause itself a narrative report of a character’s perceptions – with the character’s consciousness suffusing the third-person voice?
Chatman is surely right to suggest that (thanks to a kind of higher-order grammar that applies to texts versus words, phrases, and clauses) readers and speakers are able “to distinguish between Narrative and Description as text-types, on the one hand, and sentences in the surface of a text which are loosely called ‘narrative’ or ‘descriptive’, on the other” (1990: 16). Arguably, however, membership gradience affects classifications at both of these levels. As discussed in chapter 3, interpretations of surface-level structures are shaped by their situation in a larger discourse context; so are assessments of global text-type categories – for example, when one has to determine the communicative function of the first utterance spoken by one’s interlocutor at the beginning of a conversational interchange. Thus, to get a better sense of the (fuzzy) boundary between description and narrative, it is necessary to work toward an account of the elements that are prototypical for narrative.
Narrative
Like description, narrative is a cognitive activity, which may or may not be realized as an artifact falling within the text type narrative. That text-type category in turn encompasses a variety of media and representational conventions, ranging from those used in sign language and cinematic narrative to face-to-face storytelling and avant-garde literary narratives.
Again, my analysis in this chapter focuses on just one of what I have presented as a set of basic elements or critical properties of narrative:
(ii) Narrative representations cue interpreters to draw inferences about a structured time-course of particularized events.
As discussed more fully in my next subsection, the degree to which represented events are particularized provides a parameter along which narratives can be distinguished from explanations. Whereas stories are prototypically concerned with particular situations and events, it can be argued that explanations by their nature concern themselves with ways in which, in general, the world tends to be. Particularity is, however, a scalar, more-or-less notion, with context determining whether a text or a discourse counts as more or less particularistic. I am expected to go into detail when telling ghost stories around the camp-fire; the threshold for sufficient particularity is high in such contexts. But the same level of particularity would be judged excessive in the give-and-take of normal conversation (see the final section of my next chapter, and also Norrick 2007: 135–6). Again the contextual grounding of judgments about prototypes or standard cases manifests itself.
If particularity sets narrative apart from explanation, the kind of time-course represented in stories serves to distinguish the prototypical narrative from instances of description. But how can the temporal profile of narrative, its distinctive method of sequencing events, best be characterized? Example (3) suggests that a fairly widely used definition of narrative as a sequential representation of a sequence of events (cf. Chatman 1990; Genette [1972] 1980; Rimmon-Kenan [1983] 2002) is too broad:
3 Mondays, Wednesdays, and Fridays I have toast for breakfast, but on Tuesdays and Thursdays I have cereal.
Example (3) is a sequentially ordered representation of things that happen in a sequence, but arguably it is still not a narrative – taken on its own, at least. For one thing, the representation lacks particularity; like a recipe or a list of measurements read off a scientific instrument, it concerns a general pattern or protocol for activity or behavior as opposed to a series of particularized events. Further – to anticipate my discussion of the element of worldmaking/world disruption below and, more fully, in my next chapter – no disruptive or noncanonical events figure in (3), of the sort that are prototypical for narrative representations. The absence of the noncanonical explains why converting (3) to a series of statements about particular breakfasts, as in (3′), would not in itself produce a narrative representation:
3′ On Monday, Wednesday, and Friday I had toast for breakfast, but on Tuesday and Thursday I had cereal.
However, with a shift in context – and in accordance with the Proteus Principle (Sternberg 1982; see note 1 of this chapter) – both (3) and (3′) might start to acquire narrativity. Thus, if either representation occurred in the context of a discourse in which a cruel foster-parent had just said to his or her new foster-child, “In this house, we do not eat breakfast,” and the child had replied by uttering (3) or (3′), the result would be challenge to the dominant order, a breaking out of the disruptive or the noncanonical, that would serve to shift this representation across the (fuzzy) boundary separating descriptions from stories. Likewise, in (3′′),
3′′ From Monday to Thursday I had toast for breakfast. But on Friday I had cereal.
a modicum of world-disruptiveness accounts for the intuition that the representation has shifted its position along the continuum or cline linking description with narrative.12
In any case, as discussed in chapter 2, the tradition of thinking about narrative as a collation or coordination of two levels of temporality has its roots in the fabula/sjuzhet or story/discourse distinction proposed by Russian Formalist theorists such Viktor Shklovskii ([1929] 1990) and further developed by structuralist narratologists. At issue is the distinction between the what and the how, or what is being told versus the manner in which it is told. From this perspective, narratives feature two different layers or levels of temporal sequence that can be more or less (dis)aligned, namely, the sequence of events in the storyworld evoked by the narrative, and the sequence in which those events are ordered in the narrative representation itself. In strictly chronological narration (ABC), these sequences match up; in what Genette ([1972] 1980) termed analepses (= flashbacks of the form BCA or BAC) and prolepses (= flashforwards of the form ACB), the two sequences diverge.
Drawing on this work, Sternberg (1990, 1992, 2001) points to the peculiarly double temporality of stories as the constitutive condition for narrativity, or what makes narrative narrative, as well as the basis for the three narrative universals he names: suspense, curiosity, and surprise. As Sternberg puts it,
narrativity lives between the processes uniquely run together by the genre [or, in the terms used here, the text type]: actional and communicative, told and telling/reading sequence. This interplay between temporalities generates the three universal narrative effects/interests/dynamics of prospection, retrospection, and recognition – suspense, curiosity, and surprise, for short. Suspense arises from rival scenarios for the future....Its fellow universals rather involve manipulations of the past, which the tale communicates in a sequence discontinuous with the happening. Perceptibly so, for curiosity: knowing that we do not know, we go forward with our mind on the gapped antecedents, trying to infer (bridge, compose) them in retrospect. For surprise, however, the narrative first unobtrusively gaps or twists its chronology, then unexpectedly discloses to us our misreading and enforces a corrective rereading in late re-cognition. (2001: 117)
The question, however, is whether the two layers of temporality operative in stories are sufficient to account for suspense, curiosity, and surprise as distinctively narrative phenomena or effects. As already noted, example (3) likewise involves both a sequence of represented events and a sequential structure in the representation itself; however, in representations like (3), opportunities for suspense, curiosity, and surprise would seem to be minimal or nonexistent. For this very reason, perhaps, (3) is a better candidate for inclusion in the text-type category of description than that of narrative; but my larger point is that double-layered temporality cannot be used as a criterion for narrativity – nor as an explanation for the presence of Sternberg’s three key narrative effects – if layered temporality can obtain in descriptions as well as stories.
In this connection, it should be noted that Sternberg (1981) has elsewhere sketched a continuous, more-or-less rather than discrete, either-or relationship between narrative and description – in a way that harmonizes with the account developed in the present book. In Sternberg’s terms, because “actional mimesis presupposes a descriptive element, however implicit or even camouflaged” (1981: 72), “action and description form not givens but inferences, constructs, opposed but not divorced frames of coherence. Whether in tense or harmonious opposition, they may cohabit in the very same piece of text; and it is only according to the dominant function – or primary frame of intelligibility – that we can reasonably speak of actional or descriptive writing” (1981: 73). Hence, “[a]ctional and descriptive discourse . . . form a polar rather than ungradable contrast; and the position of a given textual piece on that continuum can be determined not in formal but in functional terms alone, involving all the contextual operations of reading” (1981: 76). But what combination of textual and contextual factors accounts for judgments about where a given text or discourse falls along this continuum stretching from descriptive to actional (= narrative) representations? Granted that, in accordance with the Proteus Principle (Sternberg 1982), neither temporal sequence nor any other formal feature serves as a guarantee for narrativity, do certain kinds of temporal structures inhibit the interpretation of texts as narratives?
For his part, Prince (1973, 1982) argues that event sequences are a necessary but not a sufficient condition for stories; i.e., narratively organized sequences have a higher-order structure not found in all strings of events classifiable as descriptive sequences. More precisely, narratives represent time- and place-specific transitions from some source state S to a target state S′. In Prince’s (1973) account of “minimal narratives,” the target state S′ is the inverse of the source state S, with an event mediating between them. Thus contrast the higher-order structure evident in (5) but not (4):
4 The politician had a reputation for hypocritical self-righteousness. The politician had a reputation for integrity. The politician’s illegal acts came to light.
5 The politician had a reputation for integrity. Then the politician’s illegal acts came to light. As a result, the politician now had a reputation for hypocritical self-righteousness.
The structure manifest in (5) but not (4) at least affords the possibility for the sort of choice, risk, consequence, and irreversibility that Kittay (1981) and Chatman (1990) characterize as distinguishing features of narrative versus description – specifically, the risk assumed by the politician in committing illegal acts in the first place, and the irreversibility of the consequences of his choosing to commit those acts. Bremond (1980) addresses similar issues in his discussion of the logic of narrative sequences, which, he argues, pass through three phases: the opening of a possibility (the possibility of committing illegal acts); the actualization or non-actualization of that possibility (the politician’s commission of the acts); and, if the possibility is actualized, the end result (the politician’s losing his reputation for integrity and gaining one for hypocritical self-righteousness instead). Accordingly, narrative does not merely involve a (dual) temporal sequence – the sequentially organized representation (= sjuzhet) of a sequence of events (= fabula) – but also traces paths taken by particularized individuals faced with decision points at one or more temporal junctures in a storyworld; those paths lead to consequences that take shape against a larger backdrop of consequences in which other possible paths might have eventuated, but did not. Hence, what Sternberg characterizes as the narrative universal of suspense arises, not just because of narrative’s double temporality, but also because of the structure of risk or irreversible consequence that certain kinds of temporal structures afford. Likewise, surprise is rooted not just in temporality but also in expectation, or rather in the violation of what is expected based on a standard or canonical pattern of events (cf. Bruner 1990, 1991).
To put these same points another way, prototypical instances of narrative involve more than particularized events unfolding within more or less richly detailed storyworlds. Thus, as noted in my preliminary discussion in chapter 1, Todorov (1968) argued that narratives characteristically follow a trajectory leading from an initial state of equilibrium, through a phase of disequilibrium, to an endpoint at which equilibrium is restored (on a different footing) because of intermediary events – though not every narrative will trace the entirety of this path (cf. Bremond 1980; Kafalenos 2006; Propp [1928] 1968). In this account, stories prototypically involve a more or less marked disruption of what is expected or canonical, and being able to recognize such disruptions depends on forming inferences about the kinds of agency characters have in storyworlds, as role-bearing or position-occupying individuals sometimes acting at cross-purposes with their own interests and goals or those of other such individuals. Thus, in example (5), the disruptive event is the politician’s own self-destructive conduct.
By the same token, the narrative universal of surprise (Sternberg 1990, 2001) is rooted not just in temporality but also in expectation, or rather in the violation of what is expected based on a standard or canonical pattern of events – as when stories portray politicians as engaged in precisely those forms of illegality on whose detection and prosecution they had staked their earlier careers. Yet it is not only that narrative can be recognized as such because of the way it represents non-normal situations and events; more than this, narrative is a cognitive and communicative strategy for navigating the gap, in everyday experience, between what was expected and what actually takes place. Thus Bruner (1990) characterizes narrative as the primary resource for “folk psychology” – that is, people’s everyday understanding of how thinking works, the rough-and-ready heuristics to which they resort in thinking about thinking itself. From this perspective, as discussed more fully in chapter 6 (cf. Herman 2008b), narrative affords a kind of discourse scaffolding for formulating reasons about why people engage in the actions they do, or else fail to engage in actions that we expect them to pursue. Accordingly a further narrative about the politician’s upbringing and familial relationships, or perhaps about a problem connected with addiction or some other disability, would be the most appropriate instrument for accounting for the turn of events represented in example (5).
My next chapter revisits issues of event sequencing and narrativity by providing something of a primer on how to build – and disrupt – a storyworld. The chapter also argues that, by starting with worldmaking/ world disruption as a basic cognitive and communicative function served by storytelling, and then working backward to the formal structures that support this root function of narrative, it is easier to motivate – to provide warrant for – fine-grained analyses of both the temporal and the spatial dimensions of storyworlds. Meanwhile, I return now to the factor of particularity vis-à-vis distinctively narrative methods of sequencing events. This factor is of key importance when it comes to studying the relationship between stories and explanations.
Explanation
Explanation, like narrative and description, is a cognitive activity that may or may not be embodied in a material artifact belonging to this text-type category, which encompasses multiple representational media and proof procedures. In this subsection I review a prominent theory of explanation as well as attempts to reconcile narrative and explanation, that is, to assimilate (at least some) narratives to the text-type category explanation, and vice versa.
In the philosophy of science, a focal point for recent research on explanation has been the Covering Law Model (CLM) that was originally developed by Carl Hempel and Paul Oppenheim (cf. Hempel [1948] 1998), refined in later work by Hempel during the 1950s and 1960s, and then disputed or at least recontextualized by subsequent theorists (see Klemke et al. 1998 for important contributions to the debate). In essence the CLM suggests that phenomena can be explained if they are characterized as instances of more general patterns or trends capturable as laws, whether deductive (or “nomological”) or inductive (or statistical) (Lambert and Britten [1970] 1998). From this perspective, the explanandum, or thing to be explained, is accounted for by virtue of its being subsumed under or “covered” by a law-like regularity that applies in analogous circumstances, all other things being equal. If the circumstance in question obtains, then that initial condition, coupled with its falling under the scope of the covering law, provides the explanans, or principle of explanation for the circumstance at issue. Hence the predictive power of covering laws, as when one predicts that if there is a pool of water and the temperature reaches 0 degrees Celsius, then all other things being equal that water will freeze.13 The nomological and statistical types of explanation are exemplified by (6) and (7), respectively:
6 All human beings are mortal. I am a human being. Therefore [by modus ponens], I am mortal.
7 Human beings tend to live longer than cats. Baby X, born on the same day as kitten Y, is a human being. Therefore [by modus ponens], baby X will probably go on to live longer than the cat that kitten Y will become.
Classical accounts of the CLM do not address how explanation might relate to narrative, however. Can there even be such a thing as a narrative explanation – given that narrative concerns itself with the particular and the contingent, with how specific things were, are, or will be (Margolin 1999) versus how in general they have to be? Arguing that “the reason for telling a narrative is to explain what happened” (1996: 3), Adams answers this question in the affirmative; he attempts to reconcile the notion “narrative explanation” with the CLM. Adams suggests that
Narrative is a type of explanation that has a past event (or state of affairs) as its explanandum, and a sequence of events as its explanans: narrative explains an explanandum, a single event, and tells an explanans, a sequence of events. The logic of narrative explanation lies in the assumption that a sequence of events explains a single event by leading up to it. (1996: 110)
However, in contrast with Adams’s attempt to link narrative to CLM explanations, researchers such as Ankersmit (2005a, 2005b), Bruner (1986, 1991), Danto (1985), and Mink (1978) have drawn a broad contrast between these two explanatory modes, suggesting that if narratives in fact explain the world they do so in a way that differs from instances of explanations that are affiliated with the CLM.
For example, as noted in chapter 1, Bruner (1991) distinguishes abstract, logico-deductive, or paradigmatic reasoning from narrative reasoning; he argues that, just as basic and general principles of reasoning ground the domain of logical-scientific reality construction,
the construction of the social order is [also] well buttressed by principles and procedures. It has an available cultural tool kit or tradition on which its procedures are modelled ...we organize our experience and memory of human happenings mainly in the form of narrative – stories, excuses, myths, reasons for doing and not doing, and so on. (Bruner 1991: 4)
For Bruner, narrative explanations are a kind of original, “folk” explanatory mode from which academic and scientific explanations have evolved to create more technical, specialized types of explanatory accounts (see Herman 1998 for fuller discussion; cf. Lyotard [1979] 1984 on the complex historical interactions between narrative and scientific forms of knowledge). A key question in this context is whether, having bifurcated from a common root, the two explanatory modes have now effectively become distinct text types separated by firm boundaries, or whether their common origin has resulted in a more porous border between them, comparable to that separating description and narrative. Or, to use Bruner’s terms, is it only the comparatively recent ascendancy of paradigmatic reasoning that has made CLM-type explanations the litmus test for explanation in general, and banished narrative explanations to peripheral status at best – on the grounds that they are “soft,” lacking scope, rigor, and generality? In short, it may be time to conduct a genealogical investigation of the story and the syllogism as competing varieties of and models for explanation – an inquiry into their intertwined histories as means of accounting-for.
One way to explore the border between narrative and explanation is to examine the commonalities and contrasts between quantitative and qualitative explanations. Once we stop trying to answer questions about how much (the degree to which) and how often (the frequency with which) data display a given property or set of properties, and begin to address instead questions about how and why those data have the character that they do (Johnstone 2000), it can be argued that we are shifting from quantitative explanation to something different – namely, a qualitative, case-study mode of explanation. But stories, too, are told in order to address questions about how and why. Hence, mapping the boundary between narrative and explanation requires coming to terms with what may be an internal split within explanation itself, assuming that qualitative and quantitative explanations are both bonafide instances of this text type. More precisely, studying how narrative relates to explanation will entail a two-pronged investigation: an inquiry into the relation between quantitative and qualitative modes of explanation, coupled with an inquiry into the relation between stories and qualitative explanations in particular.
The concluding section of this chapter explores some further puzzles about relationship between narrative and explanation.
Coda: Text Types, Communicative Competence, and the Role of Stories in Science
In this concluding section, I relate the fuzzy, more-or-less rather than binary, either-or logic of text types to people’s everyday communicative competence – that is, their ability to recognize, understand, and create not just prototypical exemplars of text-type categories but also borderline instances and hybrids or blends. Furthermore, I draw on work in the philosophy of science and the sociology of scientific knowledge to tease out further issues pertaining to the study of narrative vis-à-vis the neighboring text types of description and explanation.
My analysis assumes that at least a baseline ability to correlate individual texts with text-type categories, and to recognize degrees of membership within such categories, is a necessary ingredient of everyday discourse competence. For example, my discourse competence, my familiarity with a higher-order grammar of texts, enables me to avoid conflating descriptive and narrative sequences. I do not find myself reading the phone book as if it were a novel by Zola, or, conversely, fleeing from Martian invaders after watching War of the Worlds. As I have also suggested, however, the grammar of text types itself instantiates the category of “fuzzy grammars” (Aarts et al. 2004). As the complicated relationships among description, narrative, and explanation suggest, the very notion “kind of text” is a gradient, more-or-less affair, rather than being binarized, i.e., all or nothing. Thus, beyond licensing a variety of text-type combinations and embeddings, discourse competence requires that speakers and writers of a language be fluent in their use and interpretation of fuzzy forms such as descriptivized narrations and narrativized descriptions. Indeed, a fundamental mechanism of human creativity appears to be the ability to engage in the strategic deformation of prototypes – for example, through the blending of properties associated with categories usually viewed as separate or discrete (Hogan 2003: 70–86; Turner 2006). Phenomena such as metaphor and allegory, as well as roleplaying and performative re-enactment, can be explained in these terms.
The discreteness of the categories narrative and description can also be questioned from the vantage-point of research on the nature and origins of scientific knowledge. Based in part on Quine’s (1951) holistic model of scientific knowledge as a web of beliefs, according to which all scientific theories are underdetermined by experience and observation statements are thus necessarily theory-laden (since experience is not given as theory-neutral data but rather constructed by virtue of its relation to some conceptual framework or another), recent inquiry into science has disputed what Barnes (1990) characterizes as the rationalist account of scientific knowledge. This account holds that “[a]ny reasonable agent could infer which beliefs deserved to count as established scientific knowledge by comparing their logical implications with the indications of observation and experiment” (Barnes 1990: 62). By contrast, holistic models of scientific knowledge lead away from this “individualistic rationalist” account toward a “collective convention-alist” account (Barnes 1990: 63; cf. Longino 1990, 2002; Nelson 1990). Insofar as narrative constitutes a primary resource for constructing and disseminating more or less collectively held conventions for interpretation, descriptions or observational reports can be viewed as necessarily embedded in stories, rather than as raw experiential inputs out of which stories and other kinds of representations are subsequently built up.14 For example, as noted earlier in this chapter, storytelling practices interconnect with the process of ascribing motives, or reasons for acting, to self and others. There are thus grounds for arguing that, rather than assembling observations of individual behaviors into narratives of human conduct, people make sense of observed behaviors in terms of prior narrative templates – what Hutto (2008), building on Bruner’s (1990) work, has characterized as folk-psychological narratives. From this perspective, the chief analytic goal is not to disentangle descriptions from narratives (an impossible task), but instead to study narratives as both causes and symptoms of the webs of belief in which any (scientific or other) description of the world must be situated.
By the same token, the boundary between the text types of narrative and explanation may differ from that separating descriptions and narratives. From a microanalytic perspective at least, narrative sequences can be conceived as building additional structure (for example, a certain kind of temporal structure) on foundations provided by the elements contained in descriptive sequences. By contrast, some modes of scientific explanation, especially those of the inductive-statistical variety, may be radically resistant to narrativization, that is, radically at odds with narrative explanation. Abbott (2003), for example, discusses narrative-resistant aspects of Darwin’s theory of evolution by natural selection. Abbott suggests that it is not possible to tell a “story” about evolution at the species level, insofar as, in Darwin’s account, neither natural selection nor species function as entities with agency, but are rather part of a non-goal-directed process involving random mutation and selective adaptation. In other words, even though Darwin’s theory of natural selection is a way of understanding change over time and thus invites attempts at narrativization, it simultaneously defeats those attempts by assuming that evolution stems from a cascade of processes that operate in truly random fashion. In a different way, accounts of the so-called “explanatory gap” in cognitive science (Levine 1983) may also prove resistant to narrativization, defeating any attempt to map story or story-like templates onto causal processes of literally mind-boggling complexity. At issue is the gap between physical brain states and the condition of conscious awareness that may or may not be supervenient on those states. Even if cognitive neuroscience develops an inductive-statistical explanation of the link between complex neuronal activity in the brain and the equally complex phenomenality of conscious awareness, it is doubtful that an explanation taking into account the multivariate statistical relationships involved could ever lend itself to narrativization.
The previous examples raise general questions about the status that narrative, as a radically particularized, non-quantitative mode of accounting-for, might have in fields of inquiry that traditionally rely on quantitative methods, such as evolutionary biology, physics, and neuroscience. But further research in this area should also explore whether stories shape (overtly or covertly) what might on first blush appear to be non-narrative modes of explanation. In the domain of theoretical physics, for example, it may be that explanations based on the fourth basic element of narrative, what it’s like, need to be construed as intrinsic to the work of science, given the key role accorded the observer in quantum theory – for example, the observer of Schrödinger’s Cat, or the observer to whom Heisenberg’s uncertainty principle applies. Such explanations involve the impact of events on the real or imagined consciousnesses that register them; they can thus be argued to share with stories the consciousness factor characterized as a basic element of narrative in chapter 6. Does the ineliminability of the observer’s role in quantum-theoretical experiments – a structural requirement that invites comparison with the impossibility of narration without a viewpoint – give the lie to accounts of scientific knowledge claims as detached from particular perspectives and their attendant biases?
In any case, assessing how narrative bears on the production and organization of scientific descriptions and explanations will require more extensive interchange among philosophers and sociologists of science, science practitioners, text linguists/discourse theorists, and analysts of narrative. A key question is whether scientists use narrative as a vehicle for what remains at heart a descriptive and explanatory enterprise, or whether stories in fact play a fundamental, even constitutive, role in the work of science.
The present chapter has made a foray into the theory of text types and research on the cognitive dimensions of categories to unpack what I have characterized as the second basic element of narrative, namely, event sequencing. At issue are narrative’s distinctive temporal structure and its special concern with particularized situations and events, rather than general trends. I have also taken this opportunity to discuss aspects of a key concept on which my analysis relies, namely, the idea of the prototype or standard case. My next two chapters shift the focus from the conceptual underpinnings of the model back to a more direct exposition of the approach itself, focusing on what I have labeled as the third and fourth basic elements of narrative: respectively, the representation of disruption or disequilibrium in a storyworld involving human or human-like agents, and an emphasis on how real or imagined consciousnesses are affected by what goes on in such storyworlds-in-flux.15