Introduction
Within psychology, the analysis of narrative has been directed by an information processing approach, in particular by different versions of schema theory (Rumelhart & Ortony, 1977; Graesser, 1981) or by related models, such as story grammar or propositional analysis (Mandler, 1984; Van Dijk, 1980). Only more recently has affect been incorporated explicitly and given status as an element in comprehension (Lehnert & Vine, 1987; Dyer, 1983). Each model of narrative analysis proposed so far, including those that incorporate affect, assigns narrative elements to categories, such as propositions, goals, or plot units. Story structure is then generated by mapping connections between elements according to specific rules of inference.
As a measure of their success such models possess a degree of predictive power, indicating that they capture some of the salient features of the reader's comprehension process (e.g. Black & Bower, 1980; Yekovich & Thorndyke, 1981). Their limits are suggested by the relatively simple narratives that have been explicated by their means, either folk tales or elementary stories for children, or specially devised narratives. Although some authors claim that the principles of their model will extend to encompass any narrative, in practice, the narratives that are amenable to their methods of analysis are comparatively simple. None of the available models seems adequate to the analysis of such literary narratives as short stories or novels. Here, text elements often show a marked degree of indeterminacy, in the sense that it is impossible to assign text elements to categories with any degree of certainty.
That the models have so far been confined to explicating simple narratives, it might be argued, is merely a sign of their relative infancy. As the principles of such models are better understood (and as computational power increases, in the case of models instantiated in computer programs), it will be possible to extend their methods of analysis to increasingly complex and more interesting types of narrative. Their successful extension to literary narratives, however, seems inherently unlikely. As Spiro (1982) argued, a schema-based approach to complex literary narratives cannot succeed, as the comprehension of such texts goes beyond the schemata activated during comprehension. I argue that a fundamentally different principle must be involved in comprehension. Because literary texts are indeterminate, the reading process is directed towards the creation of schemata. I propose that the key agent in this process is affect.
Because the elements of a literary text do not exhibit stable meanings, only the more superficial and less interesting aspects of a text are amenable to description within predetermined categories. Thus, no network of relationships and inferences will capture the indeterminacy of a literary text, its complex of shifting and continually developing meanings. Under these circumstances the reader must necessarily have recourse to an alternative principle for guiding the comprehension process: Affect provides the necessary criteria for such a principle. I postulate three main criteria: (1) affect is self-referential: It allows experiential and evaluative aspects of the reader's self concept to be applied to the task of comprehension; (2) affect enables cross-domain categorization of text elements; and (3) it is anticipatory, pre-structuring the reader's understanding of the meaning of a text early in the reading process. These features are explained in greater detail later.
In the next part of the article I point to some problems in applying the standard schema approach to literary texts, and describe an alternative model based on affect. I go on to illustrate the model through an analysis of the opening section of a Virginia Woolf story, then show how data collected from readers helps to support the analysis. I also argue later that the response to literary narrative points more generally to a productive role for affect in cognitive functioning, a phenomenon that is frequently overlooked. I suggest that at the highest level the goals and beliefs of the self are instantiated in the emotions, and that the emotions thus play a determining role in cognitive processing (perception, memory, and reasoning) when this is performed in the service of the self. An analysis of the structure of literary narratives, together with empirical studies of response, provides an important source of information about this process.
The Limitations of Schema Theory
It may be illuminating to indicate some of the difficulties faced in attempting to account for literary narrative using the existing frameworks for analysis. I will illustrate with a particular framework, perhaps one of the more explicit and well developed, that of Graesser (1981). Graesser's model, which is a development of earlier work on schema theory (Bransford & Johnson, 1972; Rumelhart & Ortony, 1977), is described as a "Schema Pointer Plus Tag" model since it addresses the implications of processing typical schema-based elements of a narrative together with elements that are atypical of the schema. In Graesser's work, the application of a schema view to the analysis of narrative is systematically elaborated.
Each statement within a narrative is assigned to one of six categories: An event or a state, which must be either physical or internal to a character; or the statement represents a goal node or a style node (that is, how something was done). A statement can be assigned to only one of these categories. Statements are then related to each other structurally by one of six types of directional arcs, such as "reason," "consequence," or "property." In terms of the "reason" arc, which is the most common connection between statements in the narratives studied by Graesser, it is a rule that only one reason can emanate forward from one goal statement to another. For example, A person buys the plant is connected to its superordinate goal by a reason arc, in order to have the plant. Through these and related methods of analysis, Graesser and his co-workers devised a theory of narrative structure which has strong predictive power in accounting for how readers comprehend and remember simple narratives. It is Graesser's claim that: "In principle, passages in all prose genres should be accommodated by the system" (p. 116).
The limitations of this model of analysis, however, quickly become apparent when it is applied to a literary text. Literary narratives cannot easily be accommodated to these categories and rules. For example, the rule that only one reason arc can emanate forward from a goal is contradicted by the existence of dual or multiple goals, in which characters have several motives or perform actions which are "overdetermined" (to use Freud's useful term), a phenomenon which is rather common in literary texts. Is X going to town to buy a gift for his wife, or is he buying a gift for his wife so that he can go to town? More fundamentally, knowledge about causes and goals, which is basic to a schema model of analysis, is often ambiguous, withheld, or becomes a focus of narrative interest. Thus, a cause that is an assumption or default of a given schema may be modified or displaced; a character's goals may be unknown, illusory, or mistaken. Literary narratives typically pose schemata in a critical framework in order to suggest their inadequacy or to show their inapplicability. This makes literary narratives much less determinate than the relatively simple narratives studied by the schema or story grammar proponents: The principles on which such models are built are thus unlikely to account for comprehension of literary narratives.
This point can be turned around: The indeterminacy itself may be a primary agent in the reading process, driving other systems that control and modify schemata with their apparatus of causes and goals. The argument presented in this article is that such indeterminacy points to affect as the primary process underlying comprehension. This contention is compatible with other research findings on affect, but also goes beyond the functions generally assigned to affect in most theoretical accounts.
A specific example can be given of a phrase from a literary narrative which presents problems for the schema view of Graesser. This narrative, which is the focus of the empirical study of response to be reported below, is a short story by Virginia Woolf (1944), Together and Apart. The section that follows is the beginning of the opening paragraph:
[1] Mrs Dalloway introduced them, [2] saying you will like him. [3] The conversation began some minutes before anything was said, [4] for both Mr Serle and Miss Anning looked at the sky [5] and in both of their minds the sky went on pouring its meaning [6] though very differently. . .
The phrase which presents particular difficulties is No. 3. The first two phrases indicate a social setting, probably a party. Given that an introduction has just taken place, initiating a conversation would seem to be a standard goal. Is this goal being met or not? Because the conversation takes place "before anything was said", the status of the phrase as a goal statement is ambiguous. Although this remains unresolved, the indeterminacy it involves is also likely to transfer to other elements of the narrative. But perhaps the statement should be categorized as an event? If so, another difficulty awaits us over deciding whether the event is an internal or an external one. Does the statement indicate non-verbal communication, such as a meaningful exchange of glances, or a reciprocal sense of empathy? Or is a character conversing in imagination, fantasizing a conversation? Is it an event at all? As the "conversation" began "before anything was said," it might signify a state, such as mutual indecision or embarrassment. At least the cause of the "conversation" seems clear: It follows as a result of the introduction. But, in terms of a cause, this is perhaps the least interesting aspect of the statement, as a cause of a more interesting kind appears to be implied in the next phrase, through the conjunction "for": "for both Mr Serle and Miss Anning looked at the sky . . ." The operation of this cause is, however, left unclear: the sky "went on pouring its meaning," which may indicate some cause for the conversation; but the phrase is itself a metaphor, and thus only compounds the indeterminacy.
Perhaps enough has been said about this phrase to show the size of the problem. The subtlety of Woolf's prose makes such an analysis seem particularly clumsy. The conclusion to which it points is that the process of comprehension is driven by an indeterminacy over causes, goals, the status of events as internal or physical, and other uncertainties. It is a process that frequently serves to unsettle the schemata of the reader at just the moment that the schemata are being identified.
From the point of view of schema theory, therefore, a literary narrative will often be found not to be "well formed," whether in terms of overall plan, coherence, completeness, or conventions (Olson, Mack, & Duffy, 1981). The difficulties of the schema approach can be summarised as follows:
1. Schema identification is necessary to allow the work of understanding to begin, but the application of schemata is likely to be thwarted or disrupted in a variety of ways by a range of textual features.
2. Providing a causal account of states or events is often problematic, due to uncertainties about how text elements relate or to multiple potential relationships within the text. The status of narrative elements as states or events may itself be indeterminate.
3. A goal-directed account of characters is often inadequate. Goals may be multiple, ambiguous or conflicting, so that their status becomes a focus of narrative interest.
The outcome of a literary narrative, therefore, may include the following: Causes are not what the reader might have believed (a cause may be more complex, profound, or obscure); goals may be ineffective or turn out to have been inappropriate, so that the story becomes a critique of the goals as such. Thus, the reader's effort after meaning is directed towards developing a schema for the narrative that transcends the more simple schemata with which he started. The aim of comprehension can be described as the creation of schemata, rather than their application. It is notable, in passing, that few attempts to examine the creation of schemata are available in the research literature (e.g. Rumelhart & Norman, 1978; Vosniadou & Brewer, 1987). Some interesting attempts are now being made to model the process of schema acquisition in artificial intelligence research (e.g. Chein, 1987; DeJong, 1986), using simple narrative materials; but these have depended so far on a plans-goal analysis similar to that of Graesser and are thus subject to similar limitations. One outcome of studying the comprehension of literary narrative may be a better purchase on the elusive process of schema creation.
Beyond the Schema Given
Having suggested the inadequacy of the schema model, an alternative approach to literary narrative will now be described. Readers instantiate schemata to interpret a story, but the story itself defamiliarises the schemata: that is, a familiar or usual schema which would have been applied automatically becomes questionable, and its adequacy may become a focus of narrative interest. Defamiliarisation is frequently cited by critics as a standard effect of literary texts, including narratives, and the present approach is consistent with several theoretical accounts of literary response (Iser, 1980; Perry, 1979; Shklovsky, 1917/1965; Van Peer, 1986). In moving to a more constructive view of the role of the reader, the present approach also sees literary texts as "writerly" rather than "readerly," in Barthes' useful term (Barthes, 1970/1975, p. 4), that is, it places the main emphasis on the constructive role of the reader (noting, however, that the degree to which a reader constructs the text continues to be the subject of contention and ambiguity on the part of literary theorists: cf. Culler, 1983, pp. 69-70).
One effect of defamiliarisation is to require the creation of a new schema adequate to the material presented by the story. The initial schemata are likely to contradict each other in subtle ways, providing the reader with signals of their inadequacy and impelling him to recognise that they have only provisional status. The primary work of the reader is thus to interpret the unfolding sentences of the story for clues to a more adequate schema. The argument of this article is that it is the reader's affective response that guides this process.
The term affect has, of course, different meanings among psychologists, and a variable relationship with such terms as emotion and feeling, which makes it difficult to define (Kleinginna & Kleinginna, 1981). For the purpose of this article I will understand affect to denote the subjective experience of emotions and feelings, including (necessarily for my argument) feelings that have little or no cognitive content but which operate immediately as judgements, preferences, and the like. This is the dimension of affect described by Buck (1985) as the intemal, subjective readout of the current state of the emotion-motivation system. It follows that affect may have "primacy" in relation to cognition (Zajonc, 1984), but the study of literary response indicates a complexity of affect beyond the limitations of Zajonc's earlier account of "preferences" (Zajonc, 1980). As will be seen from readers' responses cited below, the intensity of affect experienced by readers ranges from slight (e.g. a vague unease) to strong (e.g. "an unbearable intensity").
I propose that affect plays the primary role in directing the reading of literary narratives. Three properties of affect which make it appropriate for this task will be discussed: Affect is self-referential, cross-domain, and anticipatory. Their role in the comprehension of narrative will now be outlined; references to the research evidence for attributing these functions to affect will be mentioned briefly and selectively.
While reading a story a reader will judge that not all phrases are equally important. Some carry more weight, acting as the main focus for the reader's developing view of the story; some carry less weight, acting as support or background, or conveying supplementary information. Phrases can be categorised for their salience to evaluating text meaning, as Hunt and Vipond (1986) showed. But as the schemata for the story are called into question the reader must turn elsewhere for clues to make a more adequate interpretation. In addition to processing new phrases, the reader also reassesses his understanding of phrases initially seen as less important. The affective valency of such phrases, which may have been peripheral, now becomes a central resource for construing the meaning of the story. Old schemata gain new feelings and are developed or undermined by the implications that the feelings bring with them. In this respect, affect is cross-domain: It can transfer from schemata in one domain (such as those concerned with a story's setting) to those in another (such as the relationship between two characters). The ability of affect to cross domains has been noted by previous workers (Bower & Cohen, 1982, p. 329; Bruner, 1966, pp. 12-13), although the potentially productive role of this process in cognitive functioning appears generally to have been overlooked. I have suggested elsewhere (Miall, 1987) that it may underlie a number of processes that have been difficult to explain, such as the comprehension of metaphor, dream formation, and bisociation in creative thought.
Affect is also anticipatory. Given the indeterminacy of the reader's experience, the reader must develop some representation of the outcome of the narrative to keep the comprehension process on-line. Because the schemata are defamiliarised, and causes and goals cannot reliably serve to represent the outcome, a representation will be constructed instead from the affective implications of the narrative. For example, given the inadequacy of the introduction schema in the opening of the Woolf story, it is probable that the reader draws on the implications of the phrases about the sky as an alternative source of affect. If this story is to develop a more adequate meaning, it may (perhaps must) be one that is consonant with the weight and resonance of the sky metaphor, for which (at this stage) no clear cognitive meaning is available. Thus, the sky metaphor, together with the connotations of the non-verbal conversation, create an affect which anticipates one or more potential outcomes. In this sense, affect offers a pre-structuring of the meaning of the text as a whole. It has been shown empirically that anticipation is common during the reading of narrative; when reading expository prose it is rather uncommon (Miall, 1988; Olson et al., 1981).
Finally, affect is self-referential. The most obvious self-referential effect is the frequent experience of readers that they "identify" with the experience and motives of the main character or characters in a narrative; that is, the reader comes to share a character's feelings and goals. But affect has a wider scope in narratives than this: Any affective response involves self-concept issues. This provides the interpretative process with a range of potential contexts for attributing meaning to text elements, drawn from the reader's prior experience and concerns.
Evidence that emotion signifies the activation of self-concept issues is provided by recent studies of the self-reference effect. It is now well established that when material is judged in relation to the self, this leads to better recall than if the material is related to a relatively unknown person, or judged during a semantic task (Kuiper & Rogers, 1979; Rogers, Kuiper, & Kirker, 1977). More recently, a series of studies has shown that the self-reference effect is most reliably obtained where the material is imbued with emotion (Bock, 1986; Bock & Klinger, 1986; Miall, 1986); such studies cast doubt on the schema or prototype model of the self (Greenwald & Pratkanis, 1984), suggesting instead that the self is instantiated in terms of the emotions. Thus, emotion during reading not only causes schemata to be reconfigured, but also constitutes the route of access to motives for reading derived from the reader's self concept. Of course, the implications of a text can reflect back on the issues it has activated. Reading is potentially capable of transforming the self, although the extent to which it actually does so will depend upon the concerns that emerge from the reader's prior experience, or, to put it another way, the extent to which the reader's imagination is seized by the text.
Thus, the reader's response to literary narrative may rehearse on a symbolic stage the current concerns (Klinger, 1978) of the self, enabling implications for the self to be anticipated in isolation from the world of action. Readers of narrative are typically more conscious of themselves as readers than readers of expository prose (Miall, 1990), which indicates an awareness of implications for the self in the feelings generated by reading (although self-reference need not always imply self-awareness -- degree of access to private self-consciousness is an established personality variable: Fenigstein, Scheier, & Buss, 1975). Because the self is directly implicated in anticipating the outcome (the meaning which will be attributed to the narrative as a whole), different readers are likely to project different self-related concerns on to the narrative. As teachers of literature are aware, there are often as many "meanings" to a literary text as there are students in the class.
The main features of the comprehension process I have been describing will now be illustrated through an analysis of the opening section of the Virginia Woolf story, shown in Table 1 (the empirical study reported below was focused on this section). In particular, I point to the process of defamiliarisation that may take place in response to a literary narrative, and indicate the role played by affect.
Theoretical Analysis of Woolf's Together and Apart
The opening section, which represents one-fifth of the story, is shown in Table 1. The story describes the introduction of two characters at a party, and follows their halting attempts to start a conversation until the mention of Canterbury (phrase 33) appears to offer the possibility of communication. The remainder of the story alternates between the perspectives of Miss Anning and Mr Serle. We learn that for Mr Serle Canterbury represents "the best years of his life, all his memories." Prompted by Miss Anning's interest he seems on the point of offering genuine information about himself. But the challenge of intimacy is too much, and both withdraw, waiting in paralysed silence until another guest interrupts them and they are able to part. The cause of the failure in communication is not explained in literal terms, so that the reader must depend on figurative, symbolic, and other indirect references to it scattered through the story. Only the opening of the story will now be analysed in detail.
TABLE 1. Opening Section of Virginia Woolf's Together and Apart (Woolf, 1944)
[1] Mrs Dalloway introduced them, [2] saying you will like him. [3] The conversation began some minutes before anything was said, [4] for both Mr Serle and Miss Anning looked at the sky [5] and in both of their minds the sky went on pouring its meaning [6] though very differently, [7] until the presence of Mr Serle by her side [8] became so distinct to Miss Anning [9] that she could not see the sky, simply, itself, any more, [10] but the sky shored up by the tall body, [11] dark eyes, grey hair, clasped hands, [12] the stern melancholy . . . face of Roderick Serle, [13] (but she had been told 'falsely melancholy') [14] and, knowing how foolish it was, [15] she yet felt impelled to say:[16] 'What a beautiful night!'
[17] Foolish! Idiotically foolish! [18] But if one mayn't be foolish at the age of forty [19] in the presence of the sky, [20] which makes the wisest imbecile [21] -- mere wisps of straw -- she and Mr Serle atoms, motes, [22] standing there at Mrs Dalloway's window, [23] and their lives, seen by moonlight, [24] as long as an insect's and no more important.
[25] 'Well!' said Miss Anning, [26] patting the sofa cushion emphatically. [27] And down he sat beside her. [28] Was he 'falsely melancholy', as they said? [29] Prompted by the sky, [30] which seemed to make it all a little futile [31] -- what they said, what they did -- [32] she said something perfectly commonplace again:
[33] 'There was a Miss Serle who lived at Canterbury [34] when I was a girl there.'
[35] With the sky in his mind, [36] all the tombs of his ancestors [37] immediately appeared to Mr Serle in a blue romantic light, [38] and his eyes expanding and darkening, he said: 'Yes.'
[39] 'We are originally a Norman family, [40] who came over with the Conqueror. [41] That is a Richard Serle buried in the Cathedral. [42] He was a knight of the garter.'
[43] Miss Anning felt that she had struck accidentally the true man, [44] upon whom the false was built. [45] Under the influence of the moon [46] (the moon which symbolized man to her, [47] she could see it through a chink of the curtain, [48] and she took sips of the moon) [49] she was capable of saying almost anything [50] and she settled in to disinter the true man [51] who was buried under the false, [52] saying to herself: 'On, Stanley, on' [53] -- which was a watchword of hers, a secret spur, [54] or scourge such as middle-aged people often make [55] to flagellate some inveterate vice, [56] hers being a deplorable timidity . . .
Note. Phrases 12 and 13 in the original read: "the stern melancholy (but she had been told 'falsely melancholy') face of Roderick Serle".
© Copyright 1944 by Quentin Bell and Angelica Garnett. Reprinted by kind permission of Chatto & Windus/The Hogarth Press.
At the outset it seems likely that the reader will identify with the characters' situation, and wish to see some communication which would allow them to move closer. The potential for this is reinforced by Miss Anning's resolve to "disinter the true man" and by Mr Serle's animation at the mention of Canterbury. There are cross-currents in the opening part of the story, however, which are likely to cause the reader to have reservations about both the ability and the understanding of Miss Anning. Against the background of the sky, which is twice associated with Mr Serle, Miss Anning senses that what they say or do may be "a little futile"; and the two characters respond "very differently" to the sky, although at this stage we are not told why. Thus, in the foreground (to use Iser's terminology, 1978, p. 92) is the reader's positive response to the possibility of genuine communication; in the background are a set of inferences which may cut across this possibility. Two opposing sets of affective currents are initiated, in which the more negative (but less prominent) implications of the second are available as a potential re-reading of the hopes represented in the first.
The cue to the more negative reading of the situation which will be made later is already present in the phrases about the sky: For example, the sky makes "the wisest imbecile," and the moon "symbolized man" to Miss Anning. But at a first reading these phrases seem permissive rather than threatening: that is, under the sky the characters may come to share a sense of their littleness, in which what is said matters less than the act of communication itself. The moon which symbolises man also gives Miss Anning the feeling that she can say almost anything. These phrases offer one affective meaning if it is seen that communication can take place, but another when that possibility has been withdrawn. On reading the story again with hindsight, it is the negative affect inherent in the situation which is more apparent in these phrases: They establish the sense of human insignificance, and point to the male principle as a primary cause of the failure in communication. The negative affect also transfers to other key phrases in this section, such as the "blue romantic light" of Mr Serle and Miss Anning's resolve to "disinter the true man," both of which had seemed to promise communication but which now seem illusory.
Although the actual cause underlying the failure of communication is not described directly, the reader is able to grasp a cause at the affective level. The cause is inherent in some deficiency in Miss Anning and Mr Serle, some lack of alignment in their vision of the situation which calls into question the possibility of communication itself. This interpretive schema, which must be created by the reader, is signalled by references to the sky and setting which occur through the story, and by perspectives on the inner self of each character. By the end of the story it is clear that the sky symbolises the negative implications of the failure of communication, making available to the reader his pre-existing affects (if one may use the plural) of response to the sky, such affect acting as a catalyst for transforming his understanding of the relationship postulated in the opening of the story.
The role played by affect is thus critical: Successful interpretation of the story, which involves the creation of a new schema, could not progress without the properties of affect which I have postulated. Whereas a schema is domain-specific, affect can transform the meaning of a schema by transference of inferences across domains, as in this ease the affect induced by the "sky" phrases transfers to the relationship schema. But affect also appears to be self-referential, thus readers are motivated to seek a resolution to the critical implications of such a story for their own concerns (the value placed on relationships, the need for communication). In that the phrases at the opening of such a story are ambivalent and no single schema will account for the range of meanings being felt, several possible outcomes are (at least in a preconscious form) probably entertained by the reader. In this sense, affect anticipates the schema that will be created by the reader, and signals the points in the story at which the new schema begins to be formed. Affect keeps subsequent processing of the story on-line, and provides a reference point by which to judge whether the eventual interpretation is adequate.
Affective Controls on Schema Formation
Understanding of the opening section of Together and Apart thus depends in several ways on the role of affect: Affect enables a wider, self-referential resource of contextual information to be brought to bear on the story, it transfers feelings across domains, and it provides the main vehicle for anticipation. Despite the existence of considerable differences that occur between readers who are reading the same text (another problem for schema-based approaches), some consistent pointers to the anticipatory role of affect on the reading process have been found.
Miall (1988) showed that readers of a fictional text saw phrases that mentioned affective states as significantly more important than readers of an essay containing an equal number of affective phrases. In a second study involving two stories by Woolf and Hemingway, affect ratings for phrases were found to correlate positively with reading times for the phrases at the opening of an episode, while the reader is searching for an appropriate meaning, but to correlate negatively when a schema for the episode has been formulated. Both studies pointed to an anticipatory role for affect in comprehending literary narratives. The study reported later starts at the other end, so to speak, by investigating the schemata that are developed during reading.
The two main schemata relevant to the opening of Woolf's story are shown diagrammatically (and somewhat simplified) in Fig. 1, with schema elements linked by thin arrows. Thick arrows show the influence of the three affective controls that determine the development of the schemata. For example, a set of possible self-referential affects are shown as influences on the two main groups of phrases concerning the relationship schema and the sky and setting. Two potential evaluations of the meaning of the story are also shown, but data from readers (described later) suggests that the "form relationship" anticipation is the stronger one at a first reading. It is also worth noting that the sequence of "relationship" phrases is easier to fit within a conventional schema model, as the use of EVENT and GOAL descriptors indicates. The cross-domain effect of the setting phrases is likely to have its full effect only later, either further on in the story or at a second reading (when, for example, the irony of postulating "the true" Mr Serle becomes clearer).
Figure 1. Affective controls on the comprehension of the opening section of Virginia Woolf's Together and Apart.
An Illustrative Study of Readers' Responses to Together and Apart
If the account of the story summarised in Fig. 1 is correct, then several predictions can be made about readers' responses to the opening section of the story. Readers will attribute affective intensity to the more salient phrases concerning both the relationship schema and the sky and setting. In judging the same phrases for importance, however, readers will initially see the phrases referring to the relationship as the more important. At a second reading it will be the phrases indicating the distance between the characters, especially those which refer to the setting of the sky or moon, which will be seen as the more important. Thus, a reversal in importance can be expected between "relationship" and "sky and setting" phrases: What was foreground now recedes into the background, and the phrases which appeared as background now come into the foreground.
It can also be predicted that because readers exposed only to the first section will construe the story as primarily about a relationship between the two main characters, asked to recall the phrases, such readers will recall more frequently those which offer evidence for a relationship. These hypotheses were examined in the study to be reported. Ratings and recall data was supplemented by eliciting readers' free comments about the meaning they saw in the story, both on reading the first section only, and again after having read the whole story. Such comments provided additional evidence of shifts in interpretation and the creation of new schemata. A number of other hypotheses could also be formulated on the basis of the theory of response I have put forward: The design of the present study is limited to illustrating how readers go beyond their initial schema.
Subjects And Method
The subjects were 28 students enrolled in an English literature component of a combined studies degree at the College of St. Paul & St. Mary, Cheltenham. The study was run with the students in several sessions, during which the 56 phrases shown in Table 1 provided the main focus for collecting response data. The opening section of the story was presented typed normally on a page, with the name of the author and story title as the heading: Subjects read the story in this form first. They then received a version in which each phrase was printed on a separate line, for the purpose of rating.
One group of 12 subjects rated the phrases for affective intensity on a scale of 1 to 6, where 1 signified strong affect; they then attempted to recall as many phrases as possible. Another group of 16 subjects read the first section and rated the phrases for importance (scale 1 to 6, where 1 = very important), then rated the phrases again after reading the story as a whole. Both groups provided written comments on their responses to the story before undertaking the ratings. In the second group, four of the subjects left a significant proportion of the phrases without ratings, due to a misunderstanding of the instructions. For the numerical analyses to be reported these subjects were omitted, and the results are thus based on the ratings of 12 subjects. But the written comments are taken from the protocols of all 16 subjects.
Ratings and Recall
The consistency with which the phrases were rated was checked by carrying out Friedman's analysis of variance, with the 56 sets of phrase ratings as the variables. Each set showed a significant level of between-subject agreement: affect ratings, X2 (12, n = 56) = 113.09, P < 0.001; ratings for importance, 1st reading, X2 (12, n = 56) = 87.93, P < 0.005; 2nd reading, X2 (12, n = 56) = 132.9, P < 0.001. Given this level of agreement it was apparent that reliance could be placed on the mean ratings for each phrase.
The subjects in the first group who rated for affect recalled an average of 14.65 phrases, but "relationship" phrases were recalled significantly more frequently than "sky and setting", as shown by a biserial correlation, r = 0.257, t(49) = 1.86, P < 0.05. This finding indicates that for these readers, who rated for affect, the relationship aspects of the story were more salient, as the phrases referring to this schema were more memorable after reading only the first section. Most of the subjects, for example, recalled the phrases in which Miss Anning and Mr Serle are introduced, in which Miss Anning invites Mr Serle to sit next to her, and where the theme of Canterbury is taken up by both characters (phrases 1-3, 26, 33, and 41). In contrast, the sky and setting phrases were generally recalled by half or fewer of the subjects.
Thus, the main factor influencing recall appears to have been the relationship schema. It is worth noting that affect ratings show no significant correlation with recall. Yet examination of the mean ratings for affect shows that a number of the sky and setting phrases received ratings for affect as high as those for the relationship phrases: The mean affect ratings for the relationship and sky and setting phrases is 3.31 and 2.88, respectively (there is no significant difference between these means). It seems likely that the salience of the affect in such phrases becomes available to readers subsequently: As the relationship schema is found unsatisfactory, so readers will draw upon the sky and setting phrases for their affective potential in re-construing the meaning of the story. The strong affect attached to such phrases, in other words, tends to predict their subsequent importance in understanding the story. An example is phrase No.46, "the moon which symbolized man to her." This obtained a mean affect rating of 2.25, indicating strong affect; its importance increased from 3.09 to 2.64 at the second reading. Several other sky and setting phrases show a similar pattern.
The ratings for importance were examined for evidence of shifts between first and second readings. Overall, the relationship phrases declined in importance whereas the sky and setting phrases increased: A biserial correlation of the mean shift of each phrase against its classification (51 of the phrases were assigned to one of the two groups) showed the shift to be significant, r = 0.419, t(49) = 3.23, P < 0.025. A more detailed study of the 24 phrases receiving the highest ratings for importance at the first reading was carried out, as shown in Table 2. The table shows how the phrases shifted in importance at the second reading. In this set of phrases 13 refer to the relationship schema and 11 to the sky and setting.
TABLE 2. Woolf's Together and Apart: Mean Rank Order of Phrases for Rated Importance at First and Second Readings
No. M Ordl Ord2
-------------------------------------------------------------------------------------------------
r 3. 1.67 1 1 The conversation began some minutes before anything was said
r 43. 1.67 2 3 Miss Anning felt that she had struck accidentally the true man,
r 50. 1.67 3 6 and she settled in to disinter the true man
r 44. 1.91 4 16 upon whom the false was built.
s 5. 2.0 5 2 and in both of their minds the sky went on pouring its meaning
r 30. 2.2 6 24 which seemed to make it all a little futile
r 56. 2.25 7 13 hers being a deplorable timidity . . .
s 45. 2.33 8 9 Under the influence of the moon
s 6. 2.55 9 4 though very differently,
r 13. 2.55 10 23 (but she had been told 'falsely melancholy')
r 37. 2.55 11 32 immediately appeared to Mr Serle in a blue romantic light
s 35. 2.58 13 15 With the sky in his mind,
r 51. 2.58 14 12 who was buried under the false,
s 29. 2.64 15 5 Prompted by the sky,
s 9. 2.83 16 17 that she could not see the sky, simply, itself, any more,
r 49. 2.83 17 33 she was capable of saying almost anything
s 21. 2.91 18 11 -- mere wisps of straw -- she and Mr Serle atoms, motes,
s 23. 2.91 19 8 and their lives, seen by moonlight,
r 2. 3.0 20 10 saying you will like him.
s 24. 3.0 21 18 as long as an insect's and no more important.
s 46. 3.09 22 14 (the moon which symbolized man to her,
r 7. 3.18 23 38 until the presence of Mr Serle by her side
s 4. 3.25 24 7 for both Mr Serle and Miss Anning looked at the sky
-----------------------------------------------------------------------------------------------------
Key. r, "relationship" phrases; s, "sky and setting" phrases; M, mean rating for importance at first reading; Ord1, rank order of phrases at first reading; Ord2, rank order of phrase at second reading.
The shift downwards in importance of the relationship phrases at the second reading is accompanied by a shift upwards in the sky and setting phrases. Comparison of the mean ratings showed that both shifts were significant (P < 0.02). The shift in mean ratings for these groups of phrases (based on the data in Table 2) is shown graphically in Fig. 2. The predicted reversal in importance of the schemata at first and second readings thus appears to have taken place.
FIG 2. Mean ratings for importance of phrases at first and second readings (rating of 1 = most important).
The nature of the shifts among the key phrases in each group can be seen by comparing the order data in Table 2. For example, the phrases suggesting that Miss Anning will set herself to know the true Mr Serle were originally seen as being almost the most important phrases in the opening of the story. At the first reading these phrases, "[43] Miss Anning felt that she had struck accidentally the true man", "[50] and she settled in to disinter the true man", "[144] upon whom the false was built", occur in positions 2, 3, and 4; but at the second reading they have moved to positions 3, 6, and 16. The questioning of Miss Anning about Mr Serle's "melancholy" (phrases 13 and 28), declines from positions 10 and 12 down to 20 and 23. The promise of communication contained in Miss Anning's sense that "[49] she was capable of saying almost anything", slips from 17 to 33. Most dramatically, the romantic potential of Mr Serle, whose ancestral tombs "[37] immediately appeared to Mr Serle in a blue romantic light", is relegated from position 11 to 32.
By contrast, the importance of the sky as an influence on the characters comes to be seen as perhaps the dominant aspect. The most important phrases indicating this, "[5] and in both of their minds the sky went on pouring its meaning", "[6] though very differently", "[29] prompted by the sky", and "[24] for both Mr Serle and Miss Anning looked at the sky", move up in importance from 5, 9, 15, and 24 to 2, 4, 5, and 7 respectively. Several other phrases mentioning the sky and the moon and their effects (numbers 21, 23, and 46), also move from positions 18, 19, and 22 up to 11, 8, and 14.
Thus, the major shift that takes place in the interpretation of the opening part of the story is the reversal in importance of the relationship and the sky and setting schemata. This gives only a first approximation to the nature of the interpretative processes involved; more subtle methods of tracing readers' responses are required to investigate the role of affect in the process of schema formation (such as the use of a talk-aloud method: Kintgen, 1983; Miall, 1990). Readers' written comments, however, show a process of anticipation based on the schemata prevailing at the outset; they also reveal how responses change when the whole story had been read.
Readers' Comments
Most readers anticipated a story about a relationship, whether this was to be satisfactory or not (the numbers preceding the comments refer to the different readers):
1. The story may continue with a more in-depth conversation between the characters, perhaps culminating in the discovery that they have a lot in common. 5. The phrase 'the conversation began some minutes before anything was said' is important as it shows an affinity 'the together bit' between two people. 4. Raises expectations of how story will progress and how the relationship between Miss Anning and Mr Serle will develop.
But some readers were also aware of an undercurrent created by references to the sky and setting, which may either support or cut across the relationship schema. These comments provide one example of the indeterminacy of the sky and setting phrases.
8. They will probably continue to talk and either drift apart or recognise some kind of connection and become closer. Sky and moon will somehow influence how they relate to one another. 7. I sense an exploration of appearance, perception, opposites, apparent contradictions which may or may not be truly so. There is also the introduction of the idea of futility, insignificance. Although only shown so far in relation to the vastness of the sky, I feel the author may well take this further, bring it down to a more internalised level with all its attendant dangers.
The last reader quoted here is anticipating one possible schema which the story as a whole serves to create, concerning the limits of communication. After reading the whole story it is a schema of this kind which the readers note:
2. A waste! Neither character wants to be at Mrs Dalloway's, both seem to force themself into social polite comments, although under this is an unbearable intensity. 16. Story is about communication and the rejection of it, or inability to accept it and its implications. Love akin to dislike -- preconceived notions being destroyed. 7. 'ultimate' communication is transient, comes in flashes and, perhaps, would not be altogether desirable . . . This is seen clearly in the internalization of the narrative, the isolation of the characters even in a social setting, and ultimately, their rejection of, sliding away from true relationship.
Several readers also point in retrospect to the 'sky and setting' aspects of the story. This is seen to symbolise the source of the problem over communication, as in the following comment:
5. Moon and sky -- awesome, tend to show the unimportance of people's lives. They are not particularly romantic symbols -- emotion is stunted and undeveloped. They could be symbols of the unconscious having an important effect but one which is difficult to pin-point.
Through these and other comments it is apparent that readers worked towards a more adequate schema for comprehending the story. The author herself provides only figurative language for transcending the primary schemata of interpretation, thus it is not to be expected that among a group of readers one schema will emerge, or that readers will be able to give a clear description of what schema has been created in response to the story as a whole. Rather, the power of a literary text such as this lies beyond the sentences of the text itself, in activating vectors of concern within the reader that are likely to continue resonating for some time to come. In this sense the overall schema which is being created by the story constitutes a part of the reader's continual creation of the self through the work of the emotions. To put it another way, a schema (in the formal sense) is only a part of what the story brings into being.
Conclusions
Narrative Models and Affect
The evidence analysed above suggests that the cognitive approach to representing narrative is not adequate to representing literary texts. Purely cognitive models, including those (Dyer, 1983; Lehnert & Vine, 1987) that deploy information about affective states, cannot account either for the indeterminacy of literary texts or for the variety of different and often conflicting readings that result from the same text. Such models of narrative imply that the emotions that arise in the reader are an after-effect of reading, playing no role in the comprehension process itself.
But the responses to the Woolf story show significant differences in comprehension from the outset which can be traced in part to distinctive differences in readers' emotions, to prior value systems concerning communication, relationships, parties, and the like. In this respect, a reader is primed to respond in characteristic ways, to read the environment for signals that impinge on issues of current concern. At the outset of a narrative the reader is thus ready to put into operation a set of affective controls to manage the self-relevant issues he has identified. It is these controls that direct subsequent decoding of the narrative, resolving indeterminacies or conflicts between schemata and similar failures at the cognitive level of comprehension. Above all, as literary narratives defamiliarise standard schemata, the affective controls must negotiate the production of new meaning, that is, they manage the process of schema creation.
Clearly, more research is needed to relate this model of affect to existing approaches to narrative, which have convincingly accounted for a range of cognitive features of comprehension in the case of simpler, non-literary narratives. Some major revisions would be required, however. For example, what readers recall following the processing of a literary text is probably reconstructed in part from states of emotion and their implications. Thus emotion provides a higher order control over memory than the narrative summary or gist proposed by Lehnert (1981) or Kintsch and van Dijk (1978). Individual differences in response must also be accounted for by showing the relation of schema knowledge to individual concerns. Put most simply, schema relationships of the type analysed by Graesser (1981) may provide certain building blocks of narrative structure that are independent of any reader, but each reader then colours the structure according to their own emotions, and produces a higher level structure for the whole story that is more or less unique.
Perhaps the most challenging issue is the problem of schema-creation, which none of the cognitive models tackles. As literary narratives tend to defamiliarise the schemata which the reader brings to them, a process which in itself is likely to arouse emotion, the reader must have recourse to an interpretative level which is prior to schemata. The reports of readers show that in response to narrative a reader often becomes self aware, conscious of entertaining hopes or fears for the characters; readers feel curiosity, and respond to the challenge to understand (this includes the readers of the Woolf story reported above, despite the somewhat artificial framework imposed on their reading). In conditions of uncertainty, when narrative elements cannot readily be assigned to existing schemata, the reader's current concerns (Klinger, 1978) are activated. Thus, to read a literary narrative is to rehearse the implications for the self of the situations and events portrayed. Out of the emotions of self-reference, therefore, emerge the schemata which will be adequate to understanding the narrative. But understanding this process better undoubtedly forms one of the most difficult problems confronting empirical research.
Cognition and Emotion
The role I have assigned to affect in the comprehension of narrative also cuts across a range of current views of the emotion-cognition relationship. It argues that affect is primarily a top-down process, directing lower-level cognitive processes involved in decoding language and in memory and reasoning. Although emotion can at times be an outcome of a cognitive process, the consequence of appraising a stimulus (Schachter & Singer, 1962; Lazarus, 1984) -- undoubtedly emotion can arise this way during response to narrative too -- it seems an unduly impoverished way of conceptualizing a reader's encounter with a literary narrative, where the reader's own concerns and anticipations are implicated from the first paragraph. A psychobiological model in which emotions are the "prime" in the system would provide a better framework: such a model is outlined by Buck (1985, 1986).
For Buck, motivation is potential energy; emotion is a readout of motivational potential which has three main functions at the level of bodily maintenance, communication, and subjective awareness. While the system is put into action by emotional "primes," the contexts within which an emotion is relevant are subject to continual updating by a feedback process from experience, learning, and other cognitive activities. Reading a literary narrative can be seen in this context as a learning activity, where the defamiliarising process initiates feedback and correction of the emotions deployed during reading. Thus, the process of reading a literary text is an interaction between an emotion-directed feed-forward mechanism (Buck, 1986, p.285; Nauta, 1971, p.183) that enables constructive responses to be made to forthcoming episodes, and a defamiliarising mechanism that restructures the reader's anticipations and feeds back corrective information to the initiating emotion. This model points to a role for literary texts, an underlying biological rationale, which remains to be worked out in detail.
The present model of response to literary narrative thus suggests several approaches to understanding the relation of emotion to cognitive processes. It may be true that emotion often functions as an appraisal mechanism or stimulus evaluation check (Leventhal & Scherer, 1987), or coordinates switches between cognitive plans in changing circumstances when behaviour is interrupted (Oatley & Johnson-Laird, 1987). But these models of emotion seem less able to account for the affective components of the response to complex narratives of the kind studied in this article. Although the ambivalence or disfunctionality of schemata may be a cause for anxiety in normal expenence, leading to withdrawal or pathological behaviour, it must be assumed that the readers of literary texts seek the experience of defamiliarisation. Narratives allow us to redefine, modify or suspend schemata, but through this process it seems likely that the primary goal of reading is to explore the emotions of the self through engagement with the text. The emotions invoked by narrative episodes and their outcome allow the reader to enact symbolically venous implications for the self. One effect may be to alter the emotional valency of existing schemata, and thus their relationship to other elements within the cognitive system, as well as to bong into being new (and possibly more adaptive) schemata.
Of course, literary texts have a variety of functions. At this point psychological analysis must be supplemented by the resources of the critic, able to reflect on the ways in which readers and texts are situated within a culture. What emotions are drawn upon during response, and what potential schemata are actually realised will be the result of complex interactions between the text and the previous experience of the reader (including experience with other texts of the same kind). The purpose of the present paper has been to examine one text in detail in order to demonstrate psychological processes of response not accounted for by the schema-based models of narrative structure. In this analysis, the primary focus has been on the constructive role of emotion in going beyond the initial schemata of response.
Manuscript received 15 January 1988. Revised manuscript received 29 June 1988.
References
Barthes, R. (1975). S/Z (R. Miller, Trans). London: Jonathan Cape. (Original work published 1970)
Black, J. B. & Bower, G. H. (1980). Episodes as chunks in narrative memory. Journal of Verbal Learning and Verbal Behavior, 18, 309-318.
Bock, M. (1986). The influence of emotional meaning on the recall of words processed for form or self-reference. Psychological Research, 48, 107-112.
Bock, M. & Klinger, E. (1986). Interaction of emotion and cognition in word recall. Psychological Research, 48, 99-106.
Bower, G. H. & Cohen, P. R. (1982). Emotional influences in memory and thinking: Data and theory. In M. S. Clark & S. T. Fiske (Eds), Affect and cognition: The 17th Annual Carnegie Symposium on Cognition. Hillsdale, N.J.: Lawrence Erlbaum Associates Inc. pp. 291-331.
Bransford, J. D. & Johnson, M. K. (1972). Contextual prerequisites for understanding: Some investigations of comprehension and recall. Journal of Verbal Learning and Verbal Behavior, 11, 717-726.
Bruner, J. (1966). On knowing: Essays for the left hand. Cambridge, Mass.: Harvard University Press.
Buck, R. (1985). Prime theory: An integrated view of motivation and emotion. Psychological Review, 92, 389-413.
Buck, R. (1986). The psychology of emotion. In J. E. Le Doux & W. Hirst (Eds), Mind and brain: Dialogues in cognitive neuroscience. Cambridge University Press, pp. 275-300.
Chein, S. A. (1987). Extending explanation-based learning: Failure-driven schema refinement. Proceedings of the 3rd Conference on Artificial Intelligence Applications. Washington, DC: IEEE Computer Society Press, pp. 106-111.
Culler, J. (1983). On Deconstruction: Theory and criticism after structuralism. London: Routledge & Kegan Paul.
DeJong, G. (1986). An approach to learning from observation. In R.S. Michalski, J.G. Carbonell, & T. M. Mitchell (Eds), Machine learning: An artificial intelligence approach, Vol. II. Los Altos, Calif.: Morgan Kaufmann, pp. 571-590.
Dyer, M. G. (1983). The role of affect in narratives. Cognitive Science, 7, 211-242.
Fenigstein, A., Scheier, M. F., & Buss, A. H. (1975). Public and private self-consciousness: Assessment and theory. Journal of Consulting and Clinical Psychology, 43, 522-527.
Graesser, A. C. (1981). Prose comprehension beyond the word. New York: Springer-Verlag.
Greenwald, A. G. & Pratkanis, A. R. (1984). The self. In R. S. Wyer & T. K. Srull (Eds), Handbook of Social Cognition, Vol. 3. Hillsdale, N.J.: Lawrence Erlbaum Associates Inc. pp.120-178.
Iser, W. (1978). The act of reading. London: Routledge & Kegan Paul.
Iser, W. (1980). The reading process: A phenomenological approach. In J. P. Tompkins (Ed.), Reader-response criticism: From formalism to post-structuralism. Baltimore: John Hopkins University Press, pp. 50-69.
Hunt, R. A. & Vipond, D. (1986). Evaluations in literary reading. Text, 6, 53-71.
Kintgen, E. R. (1983). The perception of poetry. Indiana University Press.
Kintsch, W. & van Dijk, T. A. (1978). Toward a model of text comprehension and production. Psychological Review, 85, 363-394.
Kleinginna, P. R. & Kleinginna, A. M. (1981). A categorized list of emotion definitions, with suggestions for a consensual definition. Motivation and Emotion, 5, 345-379.
Klinger, E. (1978). Modes of normal conscious flow. In K. S. Pope & J. L. Singer (Eds), The stream of consciousness: Scientific investigations into the flow of human experience. New York: John Wiley, pp. 225-258.
Kuiper, N. A. & Rogers, T. B. (1979). Encoding of personal information: Self-other differences. Journal of Personality and Social Psychology, 37, 499-514.
Lazarus, R. S. (1984). On the primacy of cognition. American Psychologist, 39, 124-129.
Lehnert, W. G. (1981). Plot units and narrative summarization. Cognitive Science, 4, 293-331.
Lehnert, W. G. & Vine, E. W. (1987). The role of affect in narrative structure. Cognition and Emotion, 1, 299-322.
Leventhal, H. & Scherer, K. (1987). The relationship of emotion to cognition: A functional approach to a semantic controversy. Cognition and Emotion, 1, 3-28.
Mandler, J. M. (1984). Stories, scripts, & scenes: Aspects of schema theory. Hillsdale, N.J.: Lawrence Erlbaum Associates Inc.
Miall, D. S. (1986). Emotion and the self: The context of remembering. British Journal of Psychology, 77, 389-397.
Miall, D. S. (1987). Metaphor and affect: The problem of creative thought Metaphor and Symbolic Activity, 2, 81-96.
Miall, D. S. (1988). Affect and narrative: A model of response to stories. Poetics, 17, 259-272.
Miall, D. S. (1990). Readers' responses to narrative: Evaluating, relating, anticipating. Poetics, 19, 323-339.
Nauta, W. J. H. (1971). The problem of the frontal lobe: A reinterpretation. Journal of Psychiatric Research, 8, 167-187.
Oatley, K. & Johnson-Laird, P. N. (1987). Towards a cognitive theory of emotions. Cognition and Emotion, 1, 29-50.
Olson, G. M., Mack, R. L., & Duffy, S. A. (1981). Cognitive aspects of genre. Poetics, 10, 283-315.
Perry, M. (1979). Literary dynamics: How the order of a text creates its meanings. Poetics Today, 1, 35-61.
Rogers, T. B., Kuiper, N. A. & Kirker, W. S. (1977). Self-reference and the encoding of personal information. Journal of Personality and Social Psychology, 35, 677-688.
Rumelhart, D. E. & Norman, D. A. (1978). Accretion, tuning, and restructuring: Three modes of learning. In J. W. Cotton & R. L. Klatzsky (Eds), Semantic factors in cognition. Hillsdale, N.J.: Lawrence Erlbaum Associates Inc. pp. 37-53.
Rumelhart, D. E. & Ortony, A. (1977). The representation of knowledge in memory. In R. C. Anderson, R. J. Spiro, & W. E. Montague (Eds), Schooling and the acquisition of knowledge. Hillsdale, N.J.: Lawrence Erlbaum Associates Inc. pp.99-135.
Schachter, S. & Singer. J. (1962). Cognitive, social and physiological determinants of emotional state. Psychological Review, 69, 379-399.
Shklovsky, V. (1965). Art as technique. In L.T. Lemon & M.J. Reis (Eds and Trans), Russian formalist criticism: Four essays. Lincoln, N.E.: University of Nebraska Press, pp. 3-24. (Original work published 1917)
Spiro, R. J. (1982). Long-term comprehension: Schema-based versus experiential and evaluative understanding. Poetics, 11, 77-86.
Van Dijk, T. A. (1980). Text and context: Elxplorations in the semantics and pragrnatics of discourse. London: Longman.
Van Peer, W. (1986). Stylistics and psychology: Investigations of foregrounding. London: Croom Helm.
Vosniadou, S. & Brewer, W. F. (1987). Theories of knowledge restructuring in development. Review of Educational Research, 57, 51-67.
Woolf, V. (1944). A haunted house. London: The Hogarth Press.
Yekovich, F. R. & Thorndyke, P. W. (1981). An evaluation of alternative functional models of narrative schema. Journal of Verbal Learning and Verbal Behavior, 20, 454-469.
Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35, 151-175.
Zajonc, R. B. (1984). On primacy of affect. In K. R. Scherer & P. Ekman (Eds), Approaches to emotion. Hillsdale, N.J.: Lawrence Erlbaum Associates Inc, pp. 259-270.