The Glasgow Review Issue 4

Back to list of Issues

Real Grammar in Fictional Contexts

Catherine Emmott

Grammarians have traditionally confined their studies to isolated sample sentences or phrases and have typically drawn on their intuitions about how they believe language works rather than on observations of how language actually does work. Using this method, a grammatical system may be developed without the analyst ever considering how linguistic features occur in “real” 1 naturally-occurring acts of communication such as novels, textbooks, newspapers and everyday conversation. There is, however, a strong assumption amongst certain linguists (e.g. Chafe, 1980, 1994; Thompson, 1987; Sinclair (ed.), 1987, 1990; Werth, forthcoming; Emmott, forthcoming) that the study of “real” acts of communication should provide the raw material for creating grammatical theories, rather than being viewed as an “after the event” application. The “real grammar” which emerges from studying this type of data may be significantly different from findings based on artificially-constructed data in terms of both specific details and the overall theoretical approach.

This study examines one particular genre, narrative fiction, looking at how language actually operates in these texts. The main object of this work is not to provide an analysis of the stylistic properties of specific texts or to comment on the language of the genre (although such observations will arise from the analysis), but to look at some of the canonical assumptions of grammatical theory, seeing how applicable these assumptions are in the face of real data and exploring topics on which the standard grammar textbooks remain silent. In particular, the study suggests that the interpretation of particular grammatical relationships in narrative texts can only be explained by such cognitive factors as the involvement, perspective and orientation of the reader. This raises questions about the nature of grammatical theory and about what should be included in grammar textbooks.


Utilising unreal data has a long tradition in Linguistics and also in Social Science subjects such as Psychology. Most grammar textbooks illustrate their observations about language using examples that have been made up by the grammarians themselves and which they assume will be recognised as representative of how language works in general. Often this has been due either to convenience or to an undue reliance on intuition. Sometimes, however, there are more deliberate reasons, both theoretical and practical, for the choice of data. Generative linguists, for example, have shunned real texts in the belief that they are replete with “performance errors”, such as slips of the tongue in spoken communication, and so do not adequately reflect the underlying knowledge of a language possessed by a speaker. This point of view is made explicit by Chomsky (1965, p. 4) when he comments that: “Observed use of language or hypothesized dispositions to respond, habits, and so on, may provide evidence as to the nature of this mental reality, but surely cannot constitute the actual subject matter of linguistics, if this is to be a serious discipline”. For different reasons, many psychologists will also opt to use self-created language materials in their experiments on language comprehension. This type of data is considered by experimenters to be more easy to “control” empirically than sentences extracted from real acts of communication.

In contrast to this approach, discourse analysts argue that artificially-constructed data lacks important properties of real language. This is partly due to the brevity of artificial data, which is often only one or two sentences in length. Certain grammatical features have been shown only to be understood properly when they are studied alongside related linguistic items spanning much larger stretches of text, as shown by Halliday and Hasan’s seminal (1976) work on textual cohesion. Brief made-up examples are also likely to lack overall discourse structure, such as topic shifts in non-narrative and flashbacks or changes in point of view in narrative. Much research in discourse analysis has demonstrated that major discourse boundaries can have an effect on grammatical form (e.g. Chafe, 1980; Clancy, 1980, 1992; Fox, 1987a,b, 1988; Emmott, 1989).

Even if a single sentence of real text is compared with a single sentence of made-up text, the differences can be considerable. Chafe (1994, p. 108) has found that the patterns of given and new information in constructed examples are often quite unlike those found in a corpus of real texts. Traditional grammar textbooks may also give undue emphasis to simple structures which are comparatively rare in reality (Chafe 1994, p. 17), as Sinclair (1990, p. xi) points out in relation to two-word sentences comprising just a subject and verb. Conversely, structures which have a reasonable frequency in real text might never be mentioned in a grammar textbook based on constructed examples alone.

As an increasing amount has been learnt about the properties of real discourse, approaches to writing grammar textbooks have been changing, with some of these textbooks now citing extracts from real texts (e.g. Freeborn, 1987; Downing & Locke, 1992). The purpose of such extracts does, nevertheless, often seem to be to illustrate a pre-existing model derived from artificial data or to fill in minor gaps in such a model. This is very different from utilising real data to build a model of grammar from first principles.

More revolutionary in this respect are recent attempts to analyse linguistic usage utilising large text corpora which are analysed with the assistance of computer concordances (e.g. Sinclair, 1987, 1990). Such projects are changing many of our assumptions about the language. This type of analysis is not, however, the only way of analysing grammatical items in real data and can be complemented by other approaches. The following section discusses how grammatical items can be studied by tracing their usage through specific passages and observing how linguistic forms are influenced by the development of the texts in which they occur.


The approach taken to grammar in this article can be described as cognitive-functional and text-orientated, in the tradition of American discourse analysts such as Chafe (1980, 1994), Du Bois (1980), Clancy (1980, 1992), Hopper (1979), Hopper and Thompson (1980), Thompson (1987) and Longacre (1983). This approach involves identifying cognitive reasons for the use of a particular grammatical form at a specific point in a text. This section illustrates this type of analysis by examining instances of three grammatical features (the past perfect, the indefinite article and anaphoric/cataphoric pronouns) in selected examples of narrative text. I argue that the textual examples cited can only be accounted for by postulating cognitive factors such as the following:

(i) Knowledge: for example, information which is available to enable the reader to identify the referent of a referring expression or to make inferences about the possible identity of the referent (e.g. Emmott, 1994; Werth, 1995, forthcoming).
(ii) Orientation and re-orientation: for example, the reader’s need to track the continuity of events in a spatio-temporal unit or to recognise switches from one spatio-temporal unit to another (e.g. Chafe, 1980; Clancy, 1980; Emmott, 1989, 1995b, forthcoming; Werth, 1995, forthcoming).
(iii) Involvement: for example, the perception of a scene as being more or less vivid by virtue of its narrative presentation (e.g. Longacre, 1970, 1983).
(iv) Perspective: for example, the identification with the point of view of one character rather than another (e.g. Fowler, 1977, 1986; Fludernik, 1993; Duchan et al (eds.), 1995; Werth, 1995, forthcoming).
(v) Salience: the degree of importance of knowledge (see (i) above), due to factors such as the status of a character or the centrality of events in terms of plot expectations (Clancy, 1980, 1992; Grimes, 1975; Grimes (ed.), 1978).
(vi) Distance: the amount of text/discourse that has elapsed since relevant knowledge (see (i) above) was acquired from the text (e.g. Clancy, 1980; Emmott, 1995a).

These categories will be explained further in the remainder of the section, as the text examples are analysed. The list above is not intended as a comprehensive one since a full analysis of a text would require a much finer-grained system (Emmott, in preparation). It will, nevertheless, serve to illustrate the general point that cognitive factors are important in grammatical analysis and, furthermore, to demonstrate how a number of cognitive factors may weigh against each other in the production/interpretation of a particular grammatical form. In each case, this explanation is compared and contrasted with that of standard grammar textbooks and the differences identified are then discussed in the following section.


One generalization that is often made in grammar textbooks is that the past perfect (e.g. had spoken) is used to signal “remote” or “past in the past” time, which is “anterior” (Quirk et al (1985, pp 195-196) to that denoted by the simple past tense (e.g. spoke). In describing the language of narrative texts, it would seem natural, therefore, to assume that when the main narrative is in the simple past tense, flashbacks will be in the past perfect. In fact, an examination of real narratives shows that substantial portions of flashbacks may utilise the same verb form as the main narrative. Indeed, the past perfect is often found only at the beginning of a flashback and sometimes intermittently throughout 2. During much of a flashback the simple past may be used, as shown below:

Example 1

This account [of her childhood] Vivien had given in a not at all self-pitying way but speculating as to how many siblings she might actually by then have. Zosie had been there and had listened with a kind of staring intensity, her elbows on the table and her little pale face held in the cup of her hands.
“My mother’s abandoned me too,” she said.
That was before she had told Vivien about the baby. She was still the mystery girl, come out of nowhere.
“My mother doesn’t know where I am,” she said. “She doesn’t care, does she? She hasn’t tried to find me, she hasn’t looked for me, she hasn’t told the police. I’m missing but she doesn’t care.”
“How do you know?” Rufus said. “It was you ran away from her not she from you. Or so one gathers. How do you know she’s not going spare?”
“We’ve had the radio on every day and there’s been nothing. I bought a paper while we were in London. I’ve looked at papers every time we’ve been in Sudbury and there’s never been a word. She doesn’t care, she’s glad I’ve gone.”
“So what?” said reasonable Rufus. “Isn’t that what you want? I thought you said that the last thing you wanted was to go home. You don’t want your mother fussing around you, do you?”
Shiva thought he had understood. Vivien certainly had. Vivien said it was one thing a young girl running away from home and being glad to leave her parents but quite another for her to find out the parents were relieved she’d gone. And Zosie said:
“Don’t you see how terrible it is? I’m missing from home and my mother isn’t worried. I might have been murdered. For Christ’s sake, I’m only seventeen.”
She began to cry, tearing sobs. Vivien sat down beside her and put an arm round her, then she turned her round and held her in her arms. It was later that day that Zosie had told Vivien everything – or almost everything. At any rate she had told her about the baby.

(Barbara Vine, 1987, A Fatal Inversion, pp. 194-195)

In this extract, all the events take place several years prior to those of the main narrative, recalling events which happened to the characters in their youths. The extract begins in the past perfect (had given, had been, had listened), but switches to simple past whilst still describing the same time period (said, began, sat down, etc.). Towards the end of the extract, the verb form switches to the past perfect again (Zosie had told Vivien everything), signalling a time which is still anterior to the main narrative, but which is nevertheless subsequent to that of the immediately preceding passage. Clearly, there is no pervasive correlation between anterior events and the past perfect verb form. Quirk et al (1985, p. 196), which is one of the most detailed grammar textbooks, does, in part, prepare us for this:

But of course, the past perfective does not have to refer to a more remote time than that referred to by the simple past. In some cases, particularly in a clause introduced by after, the two constructions can be more or less interchangeable:


[i] I ate [T2] my lunch after Sandra had come [T3] back from her shopping.


[ii] I ate [T2] my lunch after Sandra came [T3] back from her shopping.

After places the eating (T2) after Sandra’s return (which we may call T3), so the past perfective, which places T3 before T2, is redundant. What difference it does make is a matter of the “standpoint” of the speaker. In [sentence (i)] the “past in past” time T3 is identified as being earlier than T2 by the past perfective; but in [sentence (ii)] it is left to the conjunction after to signal this temporal relation.

This explanation alerts us to the possibility of variability but does not adequately account for why in one of Quirk et al’s examples the conjunction is followed by the past perfect (or past perfective as it is termed in Quirk et al) whereas in the other it is omitted as redundant. To ascertain this type of information, it would be necessary to look at the communicative context of real examples of this structure to see what motivates the use of a particular verb form.

Variability of the verb form in the narrative extract in Example 1 might be partially explained by redundancy after the initial set-up of the scene. The initial events would therefore be in the past perfect to establish the anteriority of the ensuing sequence of events, but once established all events in the sequence might be assumed to belong to the same time period without the need for a linguistic indicator. In addition, the simple past might be regarded as a more immediate form, more appropriate to a scene incorporating a fair amount of direct speech and from a practical point of view the simple past may also be easier to sustain over long stretches of text since it has a less cumbersome structure than the past perfect 3. The switch to the past perfect at the end of the extract might be seen as a reaffirmation of the anteriority of the flashback time frame after a minor shift to the parameters (later that day), even though, at this particular point, time is moving forwards rather than backwards. The use of past perfect verb forms at this point also coincides with summarising of the conversation rather than the more “mimetic” direct speech that has come earlier.

If this analysis is correct, the past perfect can be seen as a characteristic of reorientation rather than as a time marker of orientation. The reader’s assumption of the continuity of events within the “past in the past” time band would make unnecessary repeated grammatical signalling and a lack of repeated signalling would allow greater involvement in the action, making the events “come to life” for the reader. However, in cases where a flashback may intermittently signal orientation, there may be a tension between reminding the reader of the time frame of the activity and increasing involvement by making the presentation more vivid.


Examining real texts shows the articles to have a more complex range of functions than suggested by the standard grammar books. Consider, for example, standard statements about the indefinite article, as found in Quirk et al (1985, p. 272):

...a/an X will be used where the reference of X is not uniquely identifiable in the shared knowledge of speaker and hearer. Hence a/an is typically used when the referent has not been mentioned before, and is assumed to be unfamiliar to the speaker or hearer:
An intruder has stolen a vase. The intruder stole the vase from a locked case. The case was smashed open.

Although common, this is not the only pattern to be found in narrative. An entity may also be denoted with an indefinite article when it has been previously mentioned in the text and can therefore be assumed to be familiar to both writer and reader. Consider the following situation in Mervyn Peake’s Titus Groan. Steerpike has been climbing across the roof of Gormenghast for several hours trying to find a way of entering the castle. He eventually manages to force his way through a window and, in exhaustion, lands unconscious on the floor of Fuchsia’s secret attic. Some time later, Fuchsia enters the attic:

Example 2

Her long coloured candles were by the door and she lit one of them immediately from the little white one in her hand. Turning to place it on the table, her heart stopped beating, for she found that she was staring across the room at a body lying huddled beneath her window.

>(Mervyn Peake, 1985, Titus Groan, pp. 147-148)

The indefinite article a is used in the noun phrase a body even though the reader must know both that there is a body in the attic and that the character is Steerpike. This knowledge comes partly from the previous account of Steerpike’s entry into the room and partly from the chapter heading (incorporating a definite article), “The body by the window”. Clearly, the reason for the indefinite article is that Steerpike is at this point unknown to one of the other characters, Fuchsia, through whose eyes the reader is witnessing the action. This use of the indefinite article often accompanies a switch in perspective, either from one character to another or as the physical distance increases between the perceiver and the perceived.

Example 3

...He [Matty] thrust his books into their hands and limped quickly away. They held on to each other, laughing like apes. They broke apart, clamorously collected their fellows. The whole troupe clattered up the stone stairs, up, up, one, two, three storeys to the landing by the great window. They pushed and shoved against the great bar that ran from one end to the other at boy-height, and held the verticals that were less than a boy’s width apart. Fifty yards away and fifty feet down a boy limped quickly towards the forbidden tree.

(William Golding, 1980, Darkness Visible, p. 24)

In this example, Matty is represented by the indefinite expression a boy at the point at which he moves into the distance and this device also serves to highlight his alienation from the other boys. This linguistic signalling of perspective, by means of the indefinite article, therefore overcomes the usual requirement for a definite article when the reader has knowledge of the character’s identity (particularly when the individual has been so recently mentioned, so the linguistic distance is small).

Of course, narratologists and text linguists are increasingly studying the relationship between language and point of view (e.g. Fludernik, 1993; Duchan et al (eds.), 1995; Werth, 1995, forthcoming), so these observations about perspective are not new. However, there does need to be some transfer of these ideas from the domain of literary linguistics to the grammatical canon. This use of the indefinite article is not just found in literary texts. Du Bois (1980, pp. 258-272) finds similar features in oral narratives he examines and comments that “The literature on definiteness....has isolated definiteness from larger considerations of discourse structure as a whole.....many crucial phenomena related to definiteness are either not found or not easily recognized within the domain of one-sentence or two-sentence examples which are typically used”.


The distinction between anaphora and cataphora is rarely examined critically in the grammar textbooks. Quirk et al (1985, p. 347) offer the following explanations:

... we distinguish between Anaphoric and Cataphoric uses of a 3rd person pronoun, according to whether the element with which it co-refers (the Antecedent) precedes or follows it.
and also (p. 351):
Cataphoric reference occurs less frequently [than anaphoric reference], and under limited conditions. Where it does occur, anaphoric reference is also possible, so that we can equate two synonymous sentences such as [i] and [ia] in which the positions of pronoun and antecedent are reversed.
[i - cataphoric] Before he joined the Navy, Gerald made peace with his family.
[ia - anaphoric] Before Gerald joined the Navy, he made peace with his family.

This explanation takes no account of the fact that a sentence may be preceded by prior text. The following example would be cataphoric by Quirk et al’s criterion, but is, arguably, anaphoric in its natural context:

Example 4

In addition to his five famous novels and collection of short stories available as Penguins, E.M. Forster has published about fourteen other works...

(Opening biography in E.M. Forster , 1954, Collected Short Stories)

The pronoun his precedes the name E.M. Forster in this particular sentence, as he precedes Gerald in Quirk et al’s example. The biography has, however, begun with the statement that Edward Morgan Forster was born in London in 1879 and the intervening text contains a succession of anaphoric pronouns. The pronoun his can be interpreted automatically if we have already read the preceding text and can therefore be analysed as anaphoric to the antecedent noun phrase Edward Morgan Forster in the first sentence rather than cataphoric to the succeeding E.M. Forster in the sentence cited in Example 4.

This may seem as if the notion of anaphora simply needs to be expanded to take account of antecedents in previous sentences of the text. This is the approach suggested by cohesion analysts (e.g. Halliday and Hasan, 1976) who see texts as being comprised of inter-sentential ties or chains of linguistic items. The problem with this approach is that there are cases where there is no recent antecedent, but where there are still clues to the identity of the referent, as in the following extract:

Example 5

“I’m going to work on that kid every hour of the day until I get something.” She rose formidably and moved across the restaurant, like a warship going into action, a warship on the right side in a war to end wars, the signal flags proclaiming that every man would do his duty. Her big breasts, which had never suckled a child of her own, felt a merciless compassion. Rose fled at the sight of her, but Ida moved relentlessly towards the service door.

(Graham Greene, 1943, Brighton Rock, pp. 120-121)

This example begins in medias res at the start of a new chapter. The previous episode (pp. 116-120) does not mention the referent, Ida, at all. In terms of the proximity of an antecedent, the pronouns in this extract would have to be considered cataphoric, with the antecedent being the name Ida in the final line. It is, nevertheless, relatively easy to guess, before encountering the name, that these pronouns refer to Ida. The mention of “going to work on that kid every hour of the day” and the warship metaphor link in with the fact that she is pursuing Pinkie, a character who is referred to throughout the story as “the Boy” or described as a “kid”. This is so central to the plot of the story that the reader must be aware of this information. The physical description also corresponds to previous descriptions of Ida. Although the final mention of the name may confirm the identity of the referent, there seems a fair possibility that the reader will have already ascertained it. Making assumptions of this kind is part of reading, but is not taken account of in grammatical theory. If the above example is classified according to Quirk et al’s criterion (or by that of Halliday and Hasan), it cannot be anaphoric. If, however, anaphora is taken to mean that information is available to decode a pronoun (whether or not in the form of an explicit antecedent), then the pronouns could be classified as anaphoric, depending on the interpretive strategy of the reader. As far as cognitive functions are concerned, the reader’s knowledge of the character, reinforced by the salience of the individual due to her plot centrality, means that the distance of the last mention can be overcome.

Sometimes, the ability to identify a character in such circumstances may be due not just to general plot salience but to the reinstatement of a specific “contextual frame” (Emmott, 1989, 1994), as in the following extract:

Example 6

“But he was Fred. He told me he was Fred.”
“He was Charles. You can read it there. Charles Hale.”
“That don’t signify”, Ida said. “A man always has a different name for strangers. You aren’t telling me your real name’s Clarence. And a man don’t have a different name for every girl. He’d get confused. You know you always stick to Clarence. You can’t tell me much about men I don’t know.”
“It don’t mean anything. You can read how it was. They just happened to mention it. Nobody took any notice of that.”
She said sadly, “Nobody’s taken any notice of anything. You can read it here. He hadn’t got any folks to make a fuss. “The coroner asked if any relation of the deceased was present, and the police witness stated that they could trace no relations other than a second cousin in Middlesborough.” It sounds sort of lonely, “Nobody there to ask questions.”
“I know what loneliness is, Ida,” the sombre man said. “I’ve been alone a month now.”
She took no notice of him: she was back at Brighton on Whit Monday, thinking how while she waited there, he must have been dying, walking along the front to Hove, dying, and the cheap drama and pathos of the thought weakened her heart towards him.

(Graham Greene, 1943, Brighton Rock, pp. 31-32)

In the final paragraph, the first male pronoun (him) refers to the sombre man, but the second one (he) refers to the dead man who is called both Charles Hale and Fred. Since there is no antecedent between these two pronouns, there needs to be some explanation of how the second pronoun can be interpreted. It could be argued that there is a direct anaphoric connection between the last mention of the character (hadn’t got any folks) and this one (he must have been dying), made possible by the lexical cohesion between the words deceased and dying. Arguably, however, the shift of contextual frame (to the past Whit Monday in Brighton), recalling a prior salient scene involving Ida waiting for the deceased man, reinstates these characters in our minds before we even reach the pronoun under analysis. If this is the case, the frame switch is taking over the role that the antecedent is supposed to take, providing the necessary information for the reader to interpret the pronoun 4. The relevant cognitive factors in Example 5 are as follows. The reorientation produced by the flashback to Whit Monday allows a break between the pronouns, so that the second pronoun is not likely to co-refer with the first one (denoting the sombre man). The reader’s knowledge of the Whit Monday episode and the salience of that episode to the plot make it possible for the scene to be reinstated and the character set to be inferred.


Some linguists might argue that the type of analysis performed in the previous section is more a matter of usage or style than of grammatical theory. Many grammar books concentrate on decontextualised sentences first and then, if they look at texts at all, move on to connected sentences at a later stage. Hence Quirk et al (1985) finish their book with a chapter entitled “From sentence to text” (pp. 1421-1514), as if connected text is only to be looked at after a grammar of sentences has been proposed. An alternative, however, is to create the initial grammar directly from contextualised sentences, moving from texts to grammar. Although standard grammar textbooks do include some points of this type within their canon of topics, they do not generally seem to be prepared to go beyond this traditional repertoire of observations to take account of new findings from text linguistics. Most grammar textbooks therefore include the observation that texts can begin in medias res with the definite article, but do not note the opposite point that referents can become indefinite part way through a story. These observations do, however, have very similar status and it is difficult to argue that one has a place in a grammar book and the other does not. It is only a historical matter that in medias res openings have been observed and perspective switches have not, probably due to the prominent position of in medias res openings at the beginning of texts. Similar ad hoc textual observations can be found with relation to verb forms. Virtually every textbook includes a discussion of the historic present in oral storytelling, explaining that there may be a switch from the past tense to the present tense at climactic moments in a story. The observations made earlier in this paper about the past perfect are very similar in nature, but do not seem to be noted. Moreover, the point that the past perfect may signal reorientation rather than a straightforward time signal is fundamental to understanding the basic meaning of the verb.

In addition to looking at how grammatical features are actually used in real texts, it is also possible to observe how grammar reflects the “dynamic” unfolding of a text 5. This approach observes the text unravelling, simulating the reading process. As Fillmore (1982, p. 254) points out:

The interpreter’s experience always has a clear dynamic aspect, to which our work has to pay close attention. We need to show, for example, that a text can create expectations in the reader’s mind at one point which it then satisfies or subverts at a later point......In order to present the dynamic aspect of a reader’s experience with a text, we have developed a method of text analysis which takes the text one segment at a time, asking ourselves at each point in this unrolling of the text something like, “Having read this far, what would it have figured out, or be puzzled by, or be expecting?”

This method might appear only to be valid for each specific text, but it is possible to move from an analysis of a number of specific texts to make broad statements about how certain textual functions are achieved. So although the exact nature of the knowledge that a reader has about a character will be different for every text, the way in which knowledge can interact or override other factors (such as the distance between a pronoun and the last mention of a character) can be observed.

This approach produces a grammar which takes account of cognitive factors such as the need for orientation and the ability to switch perspective. It can be used to explain both the writer’s and reader’s relationship with the text. The writer may be assumed to follow the “communicative contract” of producing text which takes account of the cognitive abilities and limitations of an imagined reader. Of course, each reader will respond differently as a result of factors such as their ability to remember details, their attentiveness, their reading strategies, etc. Hence an analyst can only say which factors are available to be taken into account rather than which are actually utilised in the reading process and how they are weighted.


Once cognitive factors are incorporated into the analysis, grammatical theory becomes an extension of textual processing in general, with the reader’s awareness of the development of a fictional world being relevant in both cases. Grammatical theory then begins to reflect cognitive constraints and abilities. Some grammatical items provide explicit linguistic signalling to re-orientate the reader or identify a specific referent. Other items provide less explicit information but rely on the reader 's knowledge of the fictional world, particularly where information which is salient to the plot is concerned. Often, basic storytelling techniques, such as switching perspective or creating a more vivid representation of past events, will have an effect on the form used.

The assumption that grammatical relations are linked to the dynamic unfolding of a text is rather different from traditional grammatical theory and cohesive approaches to linguistic patterns in texts. Such approaches have focused on the linguistic surface of texts. So anaphora and cataphora have been seen in terms of linguistic antecedence, whilst definiteness has been a matter of whether a referent has or has not been mentioned. By contrast, a cognitive approach puts the reader back into the equation and views the interpretation of grammatical items as being a matter of knowledge and inferencing. Ultimately, therefore, a cognitive approach is likely to have more relevance for practical applications of grammar such as artificial intelligence and education.


Chafe, W.L. (ed.) (1980) The Pear Stories: Cognitive, Cultural and Linguistic Aspects of Narrative Production, Norwood, N.J.: Ablex.

Chafe, W.L. (1980) “The deployment of consciousness in the production of a narrative” in W.L. Chafe (ed.) (1980a), 9-50.

Chafe, W.L. (1994) Discourse, Consciousness and Time: The Flow and Displacement of Conscious Experience in Speaking and Writing, Chicago: University of Chicago Press.

Chomsky, N. (1965) Aspects of the Theory of Syntax, Cambridge, Mass.:MIT Press.

Clancy, P.M. (1980) “Referential choice in English and Japanese narrative” in W.L. Chafe (ed.) (1980), 127-202.

Clancy, P.M. (1992) “Referential strategies in the narratives of Japanese children”, Discourse Processes, 15, 441-467.

Downing, A. and Locke, P. (1992) A University Course in English Grammar, New York: Prentice Hall.

DT>Du Bois, J.W. (1980) “The trace of identity in discourse” in W.L. Chafe (ed.) (1980), 203-274.

Duchan, J.F., Bruder, G.A. and Hewitt, L.E. (eds.) (1995) Deixis in Narrative: A Cognitive Science Perspective, Hillsdale, N.J.: Lawrence Erlbaum.

Emmott, C. (1989) Reading between the Lines: Building a Comprehensive Model of Participant Reference in Real Narrative, Unpublished Ph.D. thesis: University of Birmingham.

Emmott, C. (1992) “Splitting the referent: An introduction to narrative enactors” in M. Davies and L. Ravelli (eds.), Advances in Systemic Linguistics: Recent Theory and Practice, London: Pinter, 221-228.

Emmott, C. (1994) “Frames of reference: contextual monitoring and narrative discourse” in M. Coulthard (ed.), Advances in Written Text Analysis, London: Routledge, 157-166.

Emmott, C. (1995a) “Lexical density and narrative intensity: some problems with using computer concordances in stylistic analysis”, paper presented at the European Society for the Study of English, University of Glasgow, September 8-12, 1995.

Emmott, C. (1995b) “Consciousness and context-building: narrative inferences and anaphoric theory” in K. Green (ed.), New Essays on Deixis: Discourse, Narrative, Literature, Amsterdam: Rodopi, 81-97.

Emmott, C. (forthcoming 1996/7) Narrative Comprehension: A Discourse Perspective, Oxford: Oxford University Press.

Emmott, C. (in preparation), “On the complexity of participant reference: designating the major characters in written narrative fiction”.

Fillmore, C.J. (1982) “Ideal readers and real readers” in D. Tannen (ed.) (1982) Analyzing Discourse: Text and Talk, Georgetown University Round Table on Languages and Linguistics 1981, Washington D.C.: Georgetown University Press, 248-270.

Fludernik, M. (1993) The Fictions Of Language and the Languages of Fiction: The Linguistic Representation of Speech and Consciousness, London: Routledge.

Forster, E.M. (1954) Collected Short Stories, Harmondsworth: Penguin.

Fowler, R. (1977) Linguistics and the Novel, London: Methuen.

Fowler, R. (1986) Linguistic Criticism, Oxford: Oxford University Press.

Fox, B.A. (1987a) Discourse Structure and Anaphora. Written and Conversational English, Cambridge: Cambridge University Press.

Fox, B.A. (1987b) “Anaphora in popular written English narratives” in R.S. Tomlin (ed.) (1987) Coherence and Grounding in Discourse, Amsterdam: John Benjamins, 157-174.

Fox, B.A. (1988) “Anaphora in popular stories: implications for narrative theory”, Empirical Studies of the Arts, 6, 2, 149-169.

Freeborn, D. (1987) A Course Book in English Grammar, London: Macmillan.

Golding, W. (1980) Darkness Visible, London: Faber and Faber.

Greene, G. (1943) Brighton Rock, Harmondsworth: Penguin.

Grimes, J.E. (1975) The Thread of Discourse, The Hague: Mouton.

Grimes, J.E. (ed.) (1978) Papers on Discourse, Dallas, Texas: Summer Institute of Linguistics.

Halliday, M.A.K. and Hasan, R. (1976) Cohesion in English, London: Longman.

Hopper, P.J. (1979) “Aspect and foregrounding in discourse” in T. Givon (ed.), Syntax and Semantics, Volume 12: Syntax and Semantics, New York: Academic Press, 213-241.

Hopper, P.J. and Thompson, S.A. (1980) “Transitivity in grammar and discourse”, Language, 56, 251-299.

Longacre, R.E. (1970) “Narrative versus other discourse genres” in R.M. Brend (ed.), (1970) Advances in Tagmemics, Amsterdam: North Holland, 357-376.

Longacre, R.E. (1983) The Grammar of Discourse, New York: Plenum Press.

Milroy, J and Milroy, L.(eds.) (1993) Real English: The Grammar of British Dialects in the British Isles, London: Longman.

Peake, M. (1985) Titus Groan, London: Methuen.

Quirk, R., Greenbaum, S., Leech, G. and Svartvik, J. (1985) A Comprehensive Grammar of the English Language, London: Longman.

Ravelli, L.J. (1991) Language from a Dynamic Perspective: Models in General and Grammar in Particular, Unpublished Ph.D. thesis, University of Birmingham.

Sinclair, J.McH. (1983) “Planes of discourse” in S.N.A. Rizvil (ed.), The Twofold Voice: Essays in Honour of Ramesh Mohan, Universitat Salzburg: Salzburg Studies in English Literature, 70-91.

Sinclair, J.McH. (ed.) (1987) Collins COBUILD English Language Dictionary, London: Collins.

Sinclair, J.McH. (ed.) (1990) Collins COBUILD English Grammar, London: Collins.

Vine, B. (1987) A Fatal Inversion, Harmondsworth: Penguin.

Thompson, S.A. (1987) “ “Subordination” and narrative event structure” in R.S. Tomlin (ed.) (1987) Coherence and Grounding in Discourse, Amsterdam: John Benjamins, 435-454.

Werth, P. (1995) “How to build a world (in a lot less than six days, using only what’s in your head)” in Green, K. (ed.) New Essays on Deixis: Discourse, Narrative, Literature, Amsterdam: Rodopi, 49-80.

Werth, P. (forthcoming) Text Worlds: Representing Conceptual Space in Discourse, London: Longman.


1 - Sinclair (ed.) (1987, 1990) uses the term "real English" to describe the corpus examples drawn on in the COBUILD dictionary and grammar. More recently, Milroy and Milroy (eds.) (1993) have used the same term to describe sociolinguistic variation in grammatical usage. For detailed discussions of the distinction between constructed and natural data see Chafe (1994) and Emmott (forthcoming, Chapter 3).
Back to Text

2 - See Emmott 1989, 1992, 1994.
Back to Text

3 - The use of the simple past for the flashback also allows the contrasting use of the past perfect as a means of signalling events which are either prior to the "past in the past" events or earlier on in the flashback event sequence.
Back to Text

4 - In some texts, examples can be found in which a scene shift can cognitively reinstate a referent when there has been no lexical mention of the character for a considerable stretch of text (Emmott, 1994).
Back to Text

5 - See Sinclair (1983, pp. 71-75) for a discussion of the "dynamic approach"; also Ravelli (1991) for a detailed study.