Show Summary Details

Page of

PRINTED FROM OXFORD HANDBOOKS ONLINE ( © Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).

Subscriber: null; date: 17 May 2021

Literary Variation

Abstract and Keywords

This chapter investigates grammatical variation in literary texts. It introduces key concepts in stylistics and discusses topics that stylisticians have been concerned with at the interface between literature and grammar, such as the use in literary texts of non-standard forms of the language and the iconic use of grammar to produce literary meaning and effects. Stylistics is now very much more than the application of linguistic description to the language of literature and has developed theories and models of its own. But at the core of the discipline remains text (both spoken and written) and the notion that text producers have choices as to how they put language into texts. This chapter explains the grammatical aspects of such choices and the literary effects they can have.

Keywords: stylistics, foregrounding, deviation, parallelism, syntactic iconicity, literary style, non-standard varieties

31.1 Introduction

In this chapter I will investigate grammatical variation in literary texts. As a full survey of this field would be a book in itself, I have instead chosen some topics that stylisticians have been concerned with at the interface between literature and grammar. These include the use in literary texts of non-standard forms of the language, either by writing in regional or social varieties or by the stretching of ‘rules’ (section 31.3), and the direct use of grammar to produce literary meaning through the exploitation of its iconic potential (section 31.4). Stylistics is now very much more than the application of linguistic description to the language of literature and has developed theories and models of its own. But at the core of the discipline remains text (both spoken and written) and the notion that text producers have choices as to how they put language into texts. Section 31.2 will introduce some of the background to stylistic discussion of literary texts.

Literature is one creative use of language (others include political speeches and advertising) which provides an opportunity to test the models arising from different grammatical theories. This opportunity arises because literary texts often stretch the norms of everyday usage. Interpreting divergent grammar relies on understanding what is ‘normal’ and can thereby reflect the systematic aspects of grammar. The strategies readers use to understand divergent grammar need explaining just as much as the stages that learners go through in acquiring a language or the loss of grammatical ability in aphasia. As John Sinclair puts it:

no systematic apparatus can claim to describe a language if it does not embrace the literature also; and not as a freakish development, but as a natural specialization of categories which are required in other parts of the descriptive system.

(Sinclair 2004: 51)

(p. 674) 31.2 Stylistics: concepts and approaches

In this section I will introduce some of the fundamental ideas and terminology used in stylistics (see Leech and Short 2007 and Jeffries and McIntyre 2010 for further reading). Section 31.2.1 considers whether there is such a thing as ‘literary style’; section 31.2.2 introduces fundamental concepts in stylistics; section 31.2.3 summarizes the development of cognitive approaches to style; and section 31.2.4 discusses the influence of corpus methods on stylistic analysis of literary grammar.

31.2.1 Is literary language different from everyday language?

It is often popularly assumed that the language of literature has its own set of fundamentally different language features, and in some periods of history and some cultures today, there is indeed an expectation that the language of literature will be regarded as ‘elevated’ or special in some way. However, if there ever was an identifiably separate literary ‘register’ in English, it would be difficult to pin down to more than a handful of lexical items, most of which were regular words which fell out of fashion in everyday language and lingered on mostly in poetry (e.g., yonder and behold).

In fact, although early stylistics, based on Russian formalism (Ehrlich 1965), spent a great deal of time and effort trying to define what was different about literary language, the discipline ultimately came to the view that if there is anything special about literary language, it is nevertheless drawing on exactly the same basic linguistic resources as the language in general. Thus, the idea that one might be able to list or catalogue the formal linguistic features in ‘literary’ as opposed to ‘non-literary’ language was eventually shown to be mistaken and recognition of the overlap between literary and other genres has now reached the point where stylistics tends to see ‘literariness’ as a point on a cline (Carter and Nash 1990: 34) rather than as an absolute. Features that might be seen as more literary (though not limited to literature) would include many of the traditional figures of speech and other literary tropes, such as metaphor, irony, metonymy, and so on, though in stylistics these would be defined more closely by their linguistic nature. Thus personification, for example, would now be shown as the result of a lexico-grammatical choice, such as the decision to combine an inanimate subject with a verb requiring an animate Actor: e.g., the moon yawned.

Scholars of literary language conclude that what makes some literary works unique and/or aesthetically pleasing is the way in which they combine, frame, or contextualize the otherwise ordinary bits of language they are using. This may imply a relatively unusual concentration of one kind of feature (e.g., continuous forms of the verb such as traipsing, trudging, limping) or unusual juxtapositions of otherwise different styles of language (e.g. a mixture of formal and informal language or a mixture of different dialects). (p. 675) Another feature often thought to be literary, though not unusual in other text-types, is parallelism of structure, or even exact repetition (see section 31.2.2). Though such features may indicate literary style, judgements about literary value are, in the end, not uniquely linguistic and are at least partly culturally determined as well as changeable (Cook 1994).

31.2.2 Stylistic concepts: foregrounding, deviation, and parallelism

Stylistics began in the early twentieth century, drawing on ideas of defamiliarization from Russian formalism, which considered that works of art were aiming to make the familiar seem new. Stylistics captured this interest from a linguistic perspective by developing the related concepts of foregrounding and deviation. Foregrounding refers to the process by which individual features of text stand out from their surroundings, while deviation refers to one of the means by which this foregrounding is achieved (Douthwaite 2000, Leech 2008: 30). Two kinds of deviation, external and internal, are distinguished. External deviation refers to features that are different in some way to the norms of the language generally. Internal deviation refers to features that differ from the norms of the text in which they appear.

Studies of the psychological reality of foregrounding have demonstrated that what scholars are noticing matches the reader’s experience. Van Peer (1986), for example, used experimental methods to ascertain which features readers were paying special attention to and found that these matched the features identified in stylistic analysis. Miall and Kuiken (1994) found further support for the psychological reality of foregrounding and also produced evidence that the foregrounded features can provoke an emotional response.

There are many potential candidates for prototypically literary features, but none of them would be found uniquely in literature and most derive their power, including their literary effect, by reference to the norms of everyday language or the norms of the text they appear in. We will see some further examples of this phenomenon in section 31.4 on iconicity in grammar, but a straightforward illustration of foregrounding is the use of parallelism in the structuring of literary (and other) texts, as seen in the opening stanza of the classic children’s poem by Alfred Noyes, ‘The Highwayman’ (1906: 244–7):1

  • The wind was a torrent of darkness among the gusty trees,
  • The moon was a ghostly galleon tossed upon cloudy seas,
  • The road was a ribbon of moonlight over the purple moor,
  • And the highwayman came riding –
  • Riding – riding –
  • The highwayman came riding, up to the old inn-door.

(p. 676) There is nothing grammatically unusual about the structure of each of the first three lines of this stanza, which have a Subject–Verb–Complement–Adverbial pattern.2 It is rather the three repetitions of the same structure that both set the musical framework for the poem, and allow the emergence of the highwayman from the distance to be foregrounded, since he arrives at the point where the grammatical pattern set up in the first three lines is disrupted for the first time. There is also a change of pattern from descriptive sentences of the pattern ‘X was Y’, where the copular verb (be) sets up a metaphorical equivalence between the subject and the complement of the sentence. The new pattern from line four by contrast has a human Agent in subject position (the highwayman), followed by an active verb phrase (came riding). This change in pattern foregrounds the highwayman from the surrounding moorland-plus-sky scene, emphasizing that he is the only sentient—and active—being for miles around.

This kind of patterning is not unique to literary genres. The form and function of parallelism in political speech-making, for example, is similar to that in literature, making the language memorable and creating both musicality and foregrounding. An example is the following passage from the famous wartime speech by UK prime minister Winston Churchill, calling the British people to fight:3

  • we shall fight on the beaches,
  • we shall fight on the landing grounds,
  • we shall fight in the fields and in the streets,
  • we shall fight in the hills;
  • we shall never surrender

Advertising also uses parallel structures and in everyday life we sometimes use parallelism to foreground something in the stories we tell each other.

Parallelism (Leech 2008: 22–3) can occur at any level of structure, from phonological to syntactic, and can be internally deviant (i.e. the remainder of the text does not consist of parallel structures) as well as being externally deviant (i.e. we do not normally speak or write using continually repeated structures). Parallelism can also become backgrounded where it is ubiquitous and therefore normalized. All of these effects are seen in ‘The Highwayman’, where the stanza patterns are the same each time, so that although there is a disruption (internal deviation) at line four, this is itself a regularity causing the reader to expect the same pattern in each stanza after the foregrounding of the disruption in verse one.

As this example of parallelism shows, foregrounding depends upon the background, which is also stylistically patterned. Anyone who has ever described an utterance as (p. 677) ‘Pinteresque’, referring to the idiosyncratically sparse style of playwright Harold Pinter, is instinctively responding to these background features of the text. The rise of corpus stylistics (see section 31.2.4 below) has enabled scholars to examine these less stark patterns of style empirically for the first time.

31.2.3 Text and context: author, narrator, character, and reader

The task of describing the style of literary texts combines an understanding of the norms of the language whilst recognizing unusual and abnormal usage for literary effect. In addition to the increased accuracy of description enabled by linguistic analysis, stylistics has developed a range of theories and models to address questions of textual meaning and where it resides.

These theories need to recognize the special communicative circumstances of literary works. Though readers may believe the author is directly addressing them, there is often a layering of participants in the discourse situation reflected in a distance between authors, readers, narrators, and characters invoked in reading a text. The discourse structure model, proposed by Leech and Short (2007: 206–18), envisages a number of levels of participation, including implied author and implied reader, which are the idealized form of these participants as envisaged by each other. There may also be one or more layers of narration as, for example, in the Joseph Conrad novel Heart of Darkness. The opening of this novel is in the voice of a first person narrator telling a tale about himself and three friends on a yacht with the ‘Director of Companies’ as their captain and host. Very soon, this I-narrative gives way to another first person narrative, when one of the friends, Marlow, starts to tell a tale about his adventures in Africa. The remainder of the novel, with the exception of the last short paragraph, is in Marlow’s voice, though it is being channelled through the voice of the first (unnamed) narrator. Within Marlow’s story, there is also a great deal of quotation of others’ speech, mostly between other characters, but also often with Marlow himself as speaker or addressee, as in the following extract (Conrad 1993: 96):

“‘His end,’ said I, with dull anger stirring in me, ‘was in every way worthy of his life.’

“‘And I was not with him,’ she murmured. My anger subsided before a feeling of infinite pity.

“‘Everything that could be done—’ I mumbled.

Marlow is quoting (to his friends) his conversation with the fiancée of Kurtz, the subject of the novel, and the original narrator quotes him quoting this conversation. The original narrator also implies the existence of an interlocutor (his addressee). In addition, the author (Conrad) is telling the reader this whole narrative and there is therefore an implied (idealized) author and reader as well as the actual author and reader. Thus there are at least five levels at which some kind of communication is taking place.

(p. 678) Such complexity of structure clearly has knock-on effects in the grammar (for example, the patterning of tenses) of the text. Remarkably, perhaps, the average reader seems to find it relatively easy to navigate the different layers of narration, possibly by letting only the most relevant layer take precedence at any one time. Recent developments in stylistics have attempted to explain the reader’s negotiation of such complexity, drawing on cognitive linguistic and psychological theories to explain readers’ construction of textual meaning. Among these, Emmott’s (1997) contextual frame theory, Werth’s (1999) Text World Theory, and Lakoff and Johnson’s (1980) conceptual metaphor theory have influenced our understanding of the reading process. Emmott, for example, proposes a framework where the reader constructs contextual frames as they read, with episodic and non-episodic information being attached to the frames as the text is processed. This information is kept in mind by the reader through two processes, which Emmott labels ‘binding’ and ‘priming’. The features of any scene in the text at any one point will be bound to that scene and may be primed (i.e. brought to the forefront of the reader’s attention) if the scene is the current focus in the text being read. Once the reader moves on to a new passage, the features of that scene become ‘unprimed’ but remain bound into the scene in the reader’s memory. Thus, the reader of Heart of Darkness will at some level be aware of the initial scene on the yacht where Marlow, the original narrator and three other men are sitting, throughout. However, as this scene is not referred to again for a very long period, it is unprimed for most of the main narrative and starts to fade from the reader’s attention. The return of this scene for one final paragraph is both a jolt to the reader’s memory as they recall the features that were bound to the scene early on, and also a renewal of the many narrative layers in the novel, which could perhaps produce some scepticism for the reader as to the accuracy of the story they have just been told.

Another cognitive theory, deictic shift theory (McIntyre 2006), helps to explain how the reader perceives the unfolding action through linguistically determined and changing points-of-view. Each of these cognitive theories envisages how some aspect of the text’s meaning is likely to be produced in the reader’s mind on the basis of features of the text. The grammatical aspects considered by these theories tend to be at levels higher than the sentence, such as textual cohesion (Halliday and Hasan 1976), verb tense sequences, narration types (first, second, third person, etc.) and other textual features such as transitivity or modality patterning that produce particular effects in constructing text worlds (see Werth 1999 and Gavins 2007).

As we shall see below, there are some aspects of grammatical variation in literature where the potential effect on the reader appears to be based on a combination of what we might consider their ‘normal’ experience of the language and the foregrounded and deviant structures they encounter. This extends the cognitive trend in stylistics to include grammatical structure itself and as a result, increasingly stylistics scholars are using cognitive grammar to analyse literary texts (Harrison et al. 2014; see also Taylor, this volume).

(p. 679) 31.2.4 Corpus stylistics

The advent of powerful computer technology and software to process large corpora of linguistic data has impacted practically and theoretically on the field of stylistics. The processing of literary texts using corpus software can not only tell us something about the style of a particular author4 but also show how authors produce characterization through stylistic choices (e.g., Culpeper 2009). Corpus analysis reveals other broader patterns, not necessarily consciously noticed by readers, detailing the background style of texts and genres, against which internally deviant features are foregrounded. Important early work in this field includes Burrows (1987) who took computational analysis of literature to a new level by demonstrating that the behaviour of certain word classes (e.g., the modal verbs) could be shown to differentiate Austen’s characters, her speech and narration and different parts of her oeuvre.

Generic tools available through most corpus linguistic software can be used to identify regular collocational patterns, which involve words that tend to occur together (e.g., boundless energy), and are largely semantic in nature; and also to identify n-grams (see Stubbs 2005), significantly frequent multiword sequences (e.g., it seemed to me), used to discover expressions typical of an author, character or genre. Corpus stylistics of this kind depends on an electronic version of the text in which word forms can be identified. One good example is Ho (2012) who demonstrates the production and testing of hypotheses about the style of John Fowles.

A step up from using word-form recognition software is to use a part-of-speech tagged corpus as in Mahlberg (2013), who, amongst other things, examines the repetition of lexical clusters in the work of Dickens. These consist of repeated sequences of words with lexical gaps as in ‘(with) his/her ___ prep det ____’ (where ‘prep’ and ‘det’ match words tagged as prepositions and determiners, respectively): examples of this pattern include ‘with his eyes on the ground’ or ‘his hand upon his shoulder’. Such clusters demonstrate a link with more lexically-based approaches to communication (Hoey 2005) and developments in grammatical theory such as construction grammar which emphasizes the importance of semi-prefabricated utterances (Croft 2001 and see also Hilpert, this volume, for more on constructional approaches to grammar). Whilst construction grammar attempts to describe the facility with which speakers produce apparently new structures in general language use, however, Mahlberg’s clusters are identified as having what she terms a ‘local textual function’ indicating important features of the plot or character.

Progress towards a reliable automated word class tagging tool has been made (Garside 1987, Fligelstone et al. 1996, Fligelstone et al. 1997, Garside and Smith 1997), but correctly identifying the word class of even high percentages (96–7 per cent) of words is not the same as producing a full grammatical analysis. Very few attempts have been made to study literature with a fully-parsed corpus of data, though Moss (2014) is (p. 680) an exception in creating and analysing a parsed corpus of the work of Henry James. (See also Wallis, this volume for more on corpus approaches to grammatical research.) Automatic parsing (structural analysis) has a much higher error rate than automatic part-of-speech tagging, and the process of manual correction is labour-intensive. Consequently, there are only a small number of fully parsed and corrected corpora, such as ICE-GB ( based at the Survey of English Usage (University College London) (see Wallis, this volume). Though some attempts have been made to semi-automate the tagging of corpora for discourse presentation (see, for example, Mahlberg’s CliC software5), there remain many difficulties with identifying the different functions of some formal structures, such as modal verbs, whose function (e.g., epistemic or deontic) can only be discerned by a human analyst (see Ziegeler, this volume, for discussion of modality).

The other huge advance in corpus methods is the chance to use very large corpora (e.g., the British National Corpus: as a baseline against which smaller corpora or individual texts or features can be compared. Although there remain problems in defining what constitutes any particular norm, the growing number of both general and specific corpora has enabled researchers to establish what might be considered a background style.

31.3 Using non-standard forms in literature

Most recent developments in literary language have tended towards democratization. The use of social and regional dialect forms in literature has challenged the dominance of varieties connected with education, wealth and power (section 31.3.1), and many writers have challenged not only the supremacy of one variety of English, but also the dominance of the written over the spoken mode (section 31.3.2). In addition, there have been attempts by writers to represent thought rather than language itself, and this is seen in the relative ‘ungrammaticality’ of some literary work (section 31.3.3).

31.3.1 Representing regional and social varieties in literature

Perhaps the most obvious question to ask about the grammar of literary works is whether they use a standard form of the language. Authors can be constrained by social (p. 681) and cultural pressures to use a standard form but increasingly in literatures written in English there is a sense that anything goes. Literary fashions at times dictated which variety of a language was acceptable for literary works, but this kind of linguistic snobbery is no longer the norm and geographical, social, and ethnic varieties are increasingly in use in mainstream as well as niche forms of literature. Much contemporary literature attempts to replicate dialect forms in the written language, in the dialogue and sometimes also in the narration, particularly where the narrator is a character in the story. Examples of different dialect use include the novels of James Kelman (e.g., The Busconductor Hines and How Late it Was, How Late), which are written in a Glaswegian dialect; the No. 1 Ladies’ Detective Agency novels by Alexander McCall Smith who represents Botswanan English in his characters’ speech; Roddy Doyle whose novels (e.g., Paddy Clarke Ha Ha Ha; The Commitments) conjure up Irish dialects of English and Cormac McCarthy (e.g., Blood Meridian) whose characters and narration reflect the language of North America’s Southwest. There are many others, of course, and the increasing informality of literary prose means that the boundary between dialect and informal language is increasingly invisible (see Hodson 2014 on the use of dialect in film and literature).

Since the early twentieth century, writers have found themselves increasingly able to exercise choice of variety, yet sometimes these decisions can be politically charged. Writers who operated entirely in their native dialect in the more distant past were perhaps less conscious of making a decision than those using their dialect to contest oppressive regimes or to give voice to people with less access to prestigious forms of the language. The relationship between dialect literature and oppression, however, is not always straightforward. Whilst some writers (e.g., Irvine Welsh, who used code-switching between idiomatic Scots and Standard English in both the dialogue and narration of his novel Trainspotting) might wish to assert their right to write in a non-standard dialect (or a minority language), others have been advised to do so by the powerful elites of dominant culture:

Howells suggests that in order to assure both critical acclaim and commercial success, the poet should dedicate himself to writing verses only in “black” dialect

(Jarrett 2010: 178).

Here, we see Jarrett’s representation of the advice from a benevolent, if patronising, literary reviewer and publisher, William Howells, to the poet and fiction writer Paul Laurence Dunbar (1872–1906), who was ‘torn between, on the one hand, demonstrating his commitment to black political progress and, on the other, representing blacks as slaves and their language as dialect, which are what prominent literary critics and publishers expected of him and black writers in general’ (Jarrett 2010: 177).

The representation of variation in language use has been a significant part of characterization at least since Chaucer’s time, when, as Phillips (2011: 40) says, ‘Chaucer’s fictional language lessons run counter to much conventional wisdom about late medieval multilingual practice, as he celebrates enterprising poseurs instead of diligent scholars or eloquent courtiers’. Whilst Chaucer’s pilgrims were found code-switching repeatedly between English, French, and Latin, Shakespeare’s characters were more prone to (p. 682) being caricatured by isolated and repeated dialect forms such as ‘look you’ for the Welsh characters (e.g., Fluellen in Henry V or Evans in Merry Wives of Windsor) or incorrect Early Modern English grammar, such as ‘this is lunatics’ (Crystal 2008a: 222).

31.3.2 Representation of spoken/informal language in literature

There are many ways in which grammatical variation is used to invoke the spoken language in literary works, often to produce realism, but also to produce foregrounding by internally deviant uses of non-standard spoken forms contrasting with surrounding standard forms. The literary effects of this are wide-ranging, but include the kind of pathos we find in Yorkshire poet Tony Harrison’s poems about his (dead) parents, where the sounds and structures of his father’s voice, for example, are blended with the poet’s own educated standard English style. Thus, we find the opening line of ‘Long Distance II’ includes ‘my mother was already two years dead’ (Harrison 1985: 133), where a standard version would change was dead to had been dead.

Apart from advantages for the scansion and rhyme-scheme, this hints at the humble origins of the poet whose life has taken such a different path to those of his parents. Harrison’s evocation of his father in ‘Long Distance I’ includes non-standard demonstratives preceding nouns such as ‘them sweets’ (instead of those sweets) which evoke the old man whose dialect the poet used to share. Much of Harrison’s early poetry relates his experience of becoming differentiated from his family background by his education and he mostly writes in Standard English where his ‘own’ voice is being represented. To find occasional hints of his earlier dialect in amongst the standard forms, then, directly references his bifurcated identity and is a kind of iconic use of dialect forms to reflect the meaning of the poem directly.

As well as including regional dialect forms, writers have increasingly adopted informal modes of expression, even in novels. Some writers adopt an almost entirely consistent spoken dialect (e.g., Irvine Welsh in Trainspotting) but others represent spoken language only in the reported speech of novels or the script of plays. Modern grammatical description based on spoken corpora of language has developed ways of identifying patterns in this informal usage of speakers (Carter and McCarthy 2006) which show that it is also highly structured. Examples include utterances that are grammatically indeterminate (e.g., Wow); phrases that function communicatively in context without clause structure (e.g., Oh that); aborted or incomplete structures (e.g., A bit) and back-channelling (e.g., Mm or Yeah) where speakers support each other (examples from Carter 2003; see also Dorgeloh and Wanner, this volume).

The direct quoting of spoken language is a staple of narrative fiction and there are also clear grammatical regularities in any indirect quotation, so that present tense verbs take on a past form when quoted indirectly, first person becomes third person and (p. 683) any proximal deixis becomes distal. Thus I am coming home next Tuesday turns into He said he was going home the following Tuesday. These clear categories of direct and indirect speech, however, have been shown by stylistics to be more complex in practice, as there are other categories of speech presentation, including perhaps the most versatile and interesting category, ‘free indirect speech’ (Leech and Short 2007: 260–70). With free indirect speech, some of the lexis and grammatical form of the quoted speaker is included, although some changes in form (i.e. tense, person, and deixis) match those for indirect speech. Thus, we might see He was coming home next Tuesday as a kind of blend of the two examples above, presenting the reader with both a narrated version of the speech and also something of the flavour of the original utterance, with the tense and person changed, but the deixis left intact. More or fewer of the features of indirect speech may be included, resulting in an apparently more or less faithful version of the original utterance. However, context is crucial for understanding whose words are being represented, as the loss of quotation marks and reporting clause can make the utterance appear to belong to the narrator or implied author instead. Whilst not unique to literary texts, free indirect speech is particularly useful to fiction-writers for blending the point-of-view of their characters with that of the narrator and/or implied author.

31.3.3 Representing cognitive patterns through ‘ungrammaticality’

One consequence of increasing variation in literary language is the experimentation with non-dialectal ‘ungrammaticality’ which has featured in many movements including modernist writing (e.g., Ernest Hemingway, William Faulkner, Virginia Woolf), L=A=N=G=U=A=G=E poetry (e.g., Leslie Scalapino, Stephen Rodefer, Bruce Andrews) and feminist poetry (e.g., Fleur Adcock, U. A. Fanthorpe, Carol Rumens). The twentieth century in particular was a time when structuralist views of how human language worked induced in some writers the conviction that the language they spoke and used on a daily basis was inappropriate to their needs as artists, as it was likely to reproduce and reinforce social attitudes and inequalities embedded in the community from which it arose. This view arose from popular versions of the ‘Whorfian hypothesis’ (or ‘Sapir–Whorf hypothesis’; see Aarts et al. 2014) which, in its strong form, claimed that human beings were limited to thinking along lines laid out by the language they spoke. Feminists responded strongly to this theory as helping to explain some aspects of their oppression in a patriarchal society (see, for example, Spender 1980). Creative responses to this hypothesis, now accepted in a much weaker form (that language influences thought rather than determining it), went beyond the more basic reaction against the (standard) language of a recognized oppressor, with the result that writers tried to reflect mental and emotional states directly through a fragmentary and ‘new’ form of the language.

(p. 684) Here is an example from ‘Droplets’ by Carol Rumens, who, like other contemporary poets, has responded to the freedom of challenging grammatical ‘correctness’ in her poems:

  • Tiniest somersaulter
  • through unroofed centuries,
  • puzzling your hooped knees
  • as you helter-skelter,
  • dreamy, careless,

Carol Rumens, ‘Droplets’

(Hex, Bloodaxe Books, 2002, p. 15)6

Not only does Rumens play with the morphology here, creating nouns from verbs (‘somersaulter’) and verbs from nouns (‘helter-skelter’) as well as possibly nonce-derived forms such as ‘unroofed’, she also fails to resolve the syntax by not introducing any main clause verbs until so late that they seem to have no real connection to the long series of images conjured up by multiple noun phrases and subordinate clauses. Her explanation for this (private communication) is that it is an ‘apostrophe’, a poetic form in which the speaker in a poem addresses someone or something not present. This may have influenced the long sequence of address forms, but it does not alter the fact that the syntax, by everyday standards, is stretched to the limit.

These literary responses to language as a kind of socio-cultural straightjacket originally resulted in writing that was deemed ‘too difficult’ by the reading public, though in some cases, such as the poetry of T. S. Eliot, critics celebrated the innovations. Over time, however, writers reverted to exploiting in less extreme ways some of the flexibilities built into the ‘system’ of the language, bending the rules without completely abandoning them and relying on more subtle forms of foregrounding to communicate their meaning. In addition, poetry readers became more accustomed to seeing the grammar stretched in new ways and were better able to process the results.

In a number of cases, such examples of gentle, but often repetitive grammatical abnormality—or spoken forms—have been used to produce a particular kind of characterization to identify the neuro-atypicality of characters and indicate their unusual ways of seeing the world. This technique is labelled ‘mind-style’ in stylistics (Fowler 1977) and, unlike the modernist and other writers expressing their own angst in relation to the world, mind-style is a use of ungrammaticality to express—and to try to understand—the world view of people we may otherwise not comprehend.

There are a few famous and well-studied novels that feature the language of a character whose cognitive processes differ from those of the majority population. These include, for example, characters with learning difficulties such as Benjy in Faulkner’s The Sound and the Fury and characters who appear to lack the normal cognitive abilities of Homo sapiens such as the Neanderthal, Lok, in Golding’s The Inheritors (Leech and Short 2007). Other studies have focussed on characters showing (p. 685) symptoms of high-functioning autism such as Christopher in Haddon’s The Curious Incident of the Dog in the Night-time (Semino 2014).

In the case of Benjy and Lok, from totally different worlds, but sharing a lack of understanding about causation, we find, for example, that transitive verbs are used without their objects, since the connections between the action and its effect are missing (Leech and Short 2007: 25–9 and 162–6 respectively). Here, for example, Benjy is describing a game of golf: ‘Through the fence, between the curling flower spaces, I could see them hitting’. Possibly because of the distance (he is looking through the fence) or because he doesn’t understand the purpose of golf, Benjy just describes the actions of the golfers and leaves the normally transitive verb without its goal. The reader has to work out what is going on from this and similar sentences.

In the case of Christopher, the linguistic correlates of his cognitive style include his lack of use of pro-forms, resulting in repetitive parallel structures (e.g., ‘I do not like proper novels. In proper novels people say…’) and his over-use of coordinated structures rather than subordinated structures (e.g., ‘I said that I wanted to write about something real and I knew people who had died but I did not know any people who had been killed, except Edward’s father from school, Mr Paulson, and that was a gliding accident, not murder, and I didn’t really know him’) (Semino 2014). In all these cases, the grammatical oddness provides clues as to the character’s cognitive state, rather than an attempt to literally reproduce the likely language of such a character.

Similarly, where a short-term rather than a chronic cognitive impairment is depicted by textual means, the apparent breakdown of language may symbolically represent the mental distress of the character, rather than directly representing their language use. An example is the William Golding novel Pincher Martin, whose protagonist is drowning and where the stylistic choices reflect his mental state as his body parts appear to take on a life of their own: ‘His hand let the knife go’. A fuller analysis of this phenomenon in the novel can be found in Simpson (1993: 111–12) who shows that as Pincher Martin gets closer to death, the role of Actor in relation to material action clauses is increasingly taken by either body parts or by inanimate entities such as ‘the lumps of hard water’ or ‘sea water’. Pincher Martin stops having agency and is reduced to an observer of his own demise.

31.4 Iconic grammar, creativity, and the reader

The discussion of grammatical variation in literature in this section will compare literary style with general or statistical norms of grammatical behaviour in English. In considering grammatical aspects of literary style, there are interesting effects that poets can achieve by the subtlest of manipulations of the ‘normal’ grammatical resources of English. The effects examined below do, of course, occur in prose as well as poetry and in non-literary as well as literary texts, but they seem peculiarly concentrated in (p. 686) poems as some of them are likely to have detrimental effects on comprehension in the more functional genres, which, unlike poetry, depend on conformity to norms of structure. Poetic style depends a great deal on the local effects of individual stylistic choices made by the poet, which are therefore more often foregrounded effects, rather than cumulative background effects of a pattern of choices. However, some of these choices also build up over a text to produce a particular topographical effect which may provide the norm against which a later internally deviant feature could be contrasted.

In this section I will use the semiotic concept of ‘iconicity’ to refer to the kind of direct meaning-making that emanates from certain grammatical choices made by poets. Whilst not as directly mimetic as the sound of a word (e.g., miaouw) representing the sound it refers to, this form of iconicity can be said to be mimetic insofar as it exploits the inevitable temporal processing of language (Jeffries 2010b). It is no surprise, then, to find that the speeding up and delaying of time in structural ways forms a large part of the iconicity to be found amongst grammatical choices in poetry.

31.4.1 Nominal versus verbal grammar and literary effect

For convenience, we might argue that language use is largely made up of labelling/naming (Jeffries 2010a) and of the representation of processes and states. Roughly speaking, these divisions of meaning match nominal and verbal structures. The remainder of the language is largely subsumed into these divisions (e.g., adjectives often occur within noun phrases; many adverbs are attached to verbs) and what is not accounted for in this way—mostly adverbials and a few adjectival complements—forms a tiny part of the whole.

Whereas grammatical description is mainly concerned with the structures present in linguistic data, stylisticians have an additional perspective based on their awareness that producers of text repeatedly make choices between different options. One of the most frequent decisions that writers make, albeit subconsciously, is whether to put information into the nominal or the verbal (or sometimes the adverbial) parts of the structure. It may seem surprising that such choices are available on a regular basis, but if we consider the option to nominalize processes, the existence of denominal verbs and the presence of processes within the modification of noun phrases, there is less of a clear division between nominal and verbal information than might seem to be the case at first sight.

So, for example, the choice of a denominal verb in the opening of Sansom’s poem ‘Potatoes’ focuses more on the whole action than would the more standard me, putting them in a bucket:

  • The soil turns neatly as he brings them to light;
  • And me, bucketing them.

Peter Sansom, ‘Potatoes’ (Point of Sale, Carcanet, 2000, p. 45)7

(p. 687) This kind of inventiveness, using recognizable derivational processes, is standard fare for poets, but it demonstrates an important feature of grammatical variation in literature. In order to be able to understand creative uses of language, the reader must be able to refer to some systematic core of the language. In this case, we may be subconsciously aware of all sorts of container-to-action zero derivations of this kind in the standard vocabulary (as in bottle the wine or box the eggs) which give us the pattern for the usage in this line. Note that we are unlikely to link it to an existing verbal meaning as in the rain was bucketing down unless it fits the context.

As with morphology, in syntax there are many examples where the decision to put information either nominally or verbally has a literary effect. One of the reasons that Newlyn’s poem ‘Comfortable box’, about her childhood in Yorkshire, has the impact of a photograph capturing a fleeting moment is because it is almost entirely made up of noun phrases, lacking a main-clause tensed verb, so there is no link to time, present or past:

  • Nothing so cheerfully compact
  • as the full fat cardboard box
  • delivered Fridays, proudly stacked
  • by our man from Groocock’s.

Lucy Newlyn, ‘Comfortable box’

(Ginnel, Carcanet, 2005, p. 43)8

One could argue that there is an ellipted ‘dummy’ clause here (There is nothing so cheerfully compact…) but the effect would surely be different if one were to add the ‘missing’ elements, as the present tense seems at odds with what is clearly a memory while the past tense (There was nothing so cheerfully compact…) lacks the vividness of the original. The choice made by the poet is not to follow the usual clause structure, thus avoiding what is normally expected in main clauses: that it will be located in some kind of time frame, rather than a photo frame.

Noun phrases with no predicate are a common grammatical occurrence in casual conversation, as for example in the answer to a question (Who stole the bike?—The man next door). However, these are normally interpreted in relation to some prior language, in this case the question. By contrast, the use of noun phrases with no further clause structure and no prior language is a staple of poetic language and common too in other literary forms. The reason for this is the variability of effect that can be achieved by naming things and people, with as much description as the author desires, but without fixing them into a time-limited, verbally mediated relationship with the context.

31.4.2 Verbal delay and other clausal variation

If the lack of a main-clause tensed verb can produce in the reader the effect of timelessness that the poem is trying to communicate, this is just one of the ways in (p. 688) which syntax can be iconic. Jeffries (2010b) argues that certain syntactic adjustments which diverge from the norms of English could be said to have a more or less directly iconic effect on the reader:

Through the juxtaposition of subordinate clauses and main clauses, the poet may cause the reader not just to perceive but to actually experience some of the same feelings of frustration and resignation that are being described.

(Jeffries 2010b: 113)

In this case, the Fanthorpe poem ‘The Unprofessionals’ was under scrutiny. In this poem the feelings of someone being visited after something terrible (a bereavement?) had happened are reflected in the syntax as well as the semantics of the poem. Here the main-clause tensed verb together with its subject (‘They come’) arrives only after three lines of adverbial delay beginning ‘When the worst thing happens…’. This has the effect of a structural hiatus representing directly the feeling of disorientation after the ‘worst thing’ and before the relief of friends and neighbours starts to arrive with ‘They come’. The remainder of the poem lacks further main-clause verbs (except a repeat of ‘they come’). The effect is to somewhat undermine the effect of relief since there is a very long list of non-finite verbs (holding, talking, sitting, doing etc.) which imply busyness, possibly without purpose or result. This effect is confirmed by informal discussions with readers of the poem who often vary between seeing the ‘unprofessionals’ as good people (i.e. informal relief from the situation) or as bad people (i.e. busybodies who don’t know when to leave). There are very many similar examples, and not just in poetry, but here is a simple one from a poem by Helen Dunmore:

  • The queue’s essentially
  • docile surges get us
  • very slowly somewhere.

Helen Dunmore, ‘The Queue’s Essentially’

(The Malarkey, Bloodaxe Books, 2012, p. 53)

As argued elsewhere (e.g., Jeffries 2010b: 109), the English reader is keen to arrive at a main-clause verb in a sentence. One of the effects of the long noun phrases discussed in the last section is that the reader will be wondering when the verb is coming. In cases like the Newlyn extract discussed there, it never does and the reader has to learn to live outside time. In this short example from Dunmore, the verb does eventually arrive, but is still quite late after a relatively long subject (up to and including ‘surges’). The topic of the stanza, which observes how queues work, slowly but surely, is reflected in this initial wait (for the verb). Once the verbal element is arrived at, it is then followed immediately not by the obligatory locative complement (‘somewhere’) which the verb requires, but by an optional, and therefore potentially frustrating, adverbial of manner whose own structure includes a structurally unnecessary intensifier (‘very’) which also adds to the impression of waiting too long for syntactic closure.

(p. 689) There are a number of ways of making the reader (of English at least) wait for the main clause verb of the sentence, including the use of a long subject, as we saw above, and also the inclusion of a long optional adverbial or a string of adverbials as we see in the opening of Davidson’s poem ‘Margaret in the Garden’:

  • With your new strappy sandals and summer top
  • and sat in the one garden chair against the fence,
  • with the light breeze lifting the hem of your skirt
  • and the corners of your Saturday newspaper,
  • you remind me of one of those domestic landscapes
  • from between the wars, now so laden with foreboding
  • that we feel worried when we look at them.

Jonathan Davidson, ‘Margaret in the Garden’

(Early Train, smith|doorstop, 2011, p. 16)9

The structure of this sentence is marked, because there is almost an excess of detail in the first four lines which act both informationally and grammatically as the preface to the main proposition (‘you remind me…’). This produces an effect like the opening frames of a film where the shot pans out from the detail of the clothing to show the garden chair, the fence, and then the wind’s minor effects on an otherwise static scene. There are a number of potential effects of waiting for the subject and main verb to arrive. Like the noun phrases with no clause structure, the scene remains somehow timeless, until the verb is arrived at. In this particular case, the verb does not break the spell since it describes a relatively static process (‘remind’), which is then linked to a visual image (‘domestic landscape’), confirming the snapshot or filmic effect of the first four lines.

As we have seen in this section, the effect of playing with what grammarians might call ‘information structure’ (see Kaltenböck, this volume) can be to produce direct experiences for readers responding to abnormal or exaggerated structural effects.

31.5 Conclusion

This chapter has attempted to discuss the wide range of what can be said about grammatical variation in literary texts and also illustrate in a little more detail some of the developments in stylistics to show how grammatical concerns are taken up in relation to literary texts. Literature has been shown to reflect everything from the reality of spoken variation through geographical or social dialects to the individual characteristics of fictional characters or the author’s attempt to convey something of the mental state of the narrator or characters in their work. In addition, studies of the language of (p. 690) literature have tried to demonstrate how it may directly induce certain cognitive states in the reader through the manipulation of the norms of grammatical structure.

Though stylistics has been often portrayed as a kind of applied version of linguistics proper, there are many ways in which what it shows about textual meaning is central to the endeavours of linguistics in understanding and describing human language in all its forms (Jeffries 2015). Having developed a range of theories and models of textual meaning, many of which focus on the reader’s perspective, stylistics has some questions to ask about grammatical structure and meaning which may be testing for theories of grammar.

I have tried to show in this chapter that although there have been some specific literary registers in the past, and these may still exist in some languages, for much of literature, the resources on which it draws are identical to those for any other language use. The literary and aesthetic issues that arise are a kind of secondary effect derived from the normal effects of the same linguistic (including grammatical) features. These secondary effects arise in various ways, such as:

  • using non-standard varieties of language in unexpected contexts, genres, or manners

  • using ungrammaticality for particular literary effect (e.g., to represent breakdown of characters)

  • deviating from grammatical norms (e.g., length of subject) for iconic effect on reader

  • interpreting grammar in the light of the genre of the text.

Grammatically deviant language can occur in a range of text types and genres, including advertising, and unlike the grammatical deviance of, for example, learners or those with speech and language pathologies, this deviation is deliberate (though not always self-conscious) and used for particular meaningful effect. As readers, we rely on our knowledge of some kind of core grammatical system in order to interpret the new structures, and these same techniques are used to interpret unusual or accidental language encountered in non-literary situations. The next stage in understanding the grammar of literary works may well be to expand the range of grammatical and other theories that are used to explore literary meaning, in order to explain the interpretation of more or less deviant language. This has already begun in cognitive stylistics (see, for example, Stockwell 2002) but there is more to be done.

At the beginning of this chapter, we saw Sinclair’s claim that linguistic description needed to ‘embrace’ literary usage as a natural part of human language and I have tried to demonstrate in the discussion above how literary style both depends on, and differs from, ordinary everyday language. If we are to explain how readers decode and interpret literary texts, we either have to posit a huge range of different abilities which are tuned to respond to texts which look a bit like ‘normal’ language, but are different in kind, or we have to assume that the process of decoding and interpreting literary language a) is based on exactly the same skills and abilities as understanding (p. 691) everyday language and b) relies on this ‘normal’ experience to interpret those aspects of literary texts that diverge from the norm.

Of course, I am arguing for the latter, and this brings with it a significant conclusion: that language users possess some kind of mentally stored core grammar on which they rely for understanding both regular and deviant texts. This is not to argue that these internal grammars of speakers are in existence from birth and it does not mean that they are immutable. One possible view of grammatical understanding would be to accept that we develop internal grammars through experience, that these grammars become relatively stable over time, but not completely unchanging—and that nevertheless they are stable enough to use as a baseline against which to judge and interpret new experiences, whether produced accidentally (as in the grammar of learners or impaired speakers) or deliberately (as in literary texts).


Aarts, Bas, Sylvia Chalker, and Edmund Weiner (2014). The Oxford Dictionary of English Grammar, 2nd edn. Oxford: Oxford University Press.Find this resource:

    Burrows, John (1987). Computation into Criticism: A Study of Jane Austen’s Novels and an Experiment in Method. Oxford: Clarendon Press.Find this resource:

      Carter, Ron (2003). ‘The grammar of talk: Spoken English, grammar and the classroom’, in Qualifications and Currriculum Authority (ed), New Perspectives on English in the Classroom. London: Qualifications and Curriculum Authority, 5–13.Find this resource:

        Carter, Ronald, and Michael McCarthy (2006). Cambridge Grammar of English: A Comprehensive Guide. Cambridge: Cambridge University Press.Find this resource:

          Carter, Ronald, and Walter Nash (1990). Seeing Through Language: A Guide to Styles of English Writing. Oxford: Blackwell.Find this resource:

            Conrad, Joseph (1993). Heart of Darkness. New York: Random House.Find this resource:

              Cook, Guy (1994). Discourse and Literature: The Interplay of Form and Mind. Oxford: Oxford University Press.Find this resource:

                Crystal, David (2008a). Think on My Words: Exploring Shakespeare’s Language. Cambridge: Cambridge University Press.Find this resource:

                  Culpeper, Jonathan (2009). ‘Keyness: Words, parts-of-speech and semantic categories in the character-talk of Shakespeare’s Romeo and Juliet.’ International Journal of Corpus Linguistics 14(1): 29–59.Find this resource:

                    Douthwaite, John (2000). Towards a Linguistic Theory of Foregrounding. Alessandria: Edizioni dell’Orso.Find this resource:

                      Ehrlich, Victor (1965) [1955]. Russian Formalism: History and Doctrine. The Hague: Mouton.Find this resource:

                        Emmott, Catherine (1997). Narrative Comprehension: A Discourse Perspective. Oxford: Oxford University Press.Find this resource:

                          Fligelstone, Steven, Mike Pacey, and Paul Rayson (1997). ‘How to generalize the task of annotation’, in Roger Garside, Geoff Leech, and Anthony McEnery (eds), Corpus Annotation: Linguistic Information from Computer Text Corpora. London: Longman, 122–36.Find this resource:

                            Fligelstone, Steven, Paul Rayson, and Nicholas Smith (1996). ‘Template analysis: Bridging the gap between grammar and the lexicon’, in Jenny Thomas and Mick Short (eds), Using Corpora for Language Research. London: Longman, 181–207.Find this resource:

                              Fowler, Roger (1977). Linguistics and the Novel. London: Methuen.Find this resource:

                                Garside, Roger (1987). ‘The CLAWS word-tagging system’, in Roger Garside, Geoff Leech, and Geoff Sampson (eds), The Computational Analysis of English: A Corpus-based Approach. London: Longman.Find this resource:

                                  Garside, Roger, and Nicholas Smith (1997). ‘A hybrid grammatical tagger: CLAWS4,’ in Roger Garside, Geoff Leech, and Anthony McEnery (eds), Corpus Annotation: Linguistic Information from Computer Text Corpora. London: Longman, 102–21.Find this resource:

                                    Gavins, Joanna (2007). Text World Theory: An Introduction. Edinburgh: Edinburgh University Press.Find this resource:

                                      Harrison, Tony (1985). Selected Poems. London: Penguin Books.Find this resource:

                                        Ho, Yufang (2012). Corpus Stylistics in Principles and Practice: A Stylistic Exploration of John Fowles’ The Magus. London: Bloomsbury Publishing.Find this resource:

                                          Hodson, Jane (2014). Dialect in Film and Literature. Basingstoke: Palgrave Macmillan.Find this resource:

                                            Hoey, Michael (2005). Lexical Priming: A New Theory of Words and Language. London: Routledge.Find this resource:

                                              Jarrett, Gene Andrew (2010). Companion to African American Literature. Hoboken, NJ: Wiley-Blackwell.Find this resource:

                                                Jeffries, Lesley (2010a). Critical Stylistics. Basingstoke: Palgrave Macmillan.Find this resource:

                                                  Jeffries, Lesley (2010b). ‘“The Unprofessionals”: Syntactic iconicity and reader interpretation in contemporary poems’, in Dan McIntyre and Beatrix Busse (eds), Language and Style. Basingstoke: Palgrave Macmillan, 95–115.Find this resource:

                                                    Jeffries, Lesley (2015). ‘Textual meaning and its place in a theory of language.’ Topics in Linguistics. 15(1), 1–10.Find this resource:

                                                      Jeffries, Lesley, and Dan McIntyre (2010). Stylistics. Cambridge: Cambridge University Press.Find this resource:

                                                        Kahan, Jeffrey (2015). ‘“I tell you what mine author says”: A Brief History of Stylometrics.’ ELH: English Literary History 82(3): 815–44.Find this resource:

                                                          Leech, Geoffrey N. (2008). Language in Literature: Style and Foregrounding. London: Pearson Education.Find this resource:

                                                            Leech, Geoffrey N., and Mick Short (2007). Style in Fiction. London: Longman.Find this resource:

                                                              Mahlberg, Michaela (2013). Corpus Stylistics and Dickens’s Fiction. New York/London: Routledge.Find this resource:

                                                                Mahlberg, Michaela, Peter Stockwell, Johan de Joode, Catherine Smith, and Matthew Brook O’Donnell (2016). ‘CLiC Dickens: Novel uses of concordances for the integration of corpus.’ stylistics and cognitive poetics.’ Corpora 11(3): 433–63.Find this resource:

                                                                  McIntyre, Dan (2006). Point of View in Plays: A Cognitive Stylistic Approach to Viewpoint in Drama and Other Text-types. Amsterdam/Philadelphia: John Benjamins.Find this resource:

                                                                    Miall, David, and Don Kuiken (1994). ‘Foregrounding, defamiliarization, and affect: Response to literary stories.’ Poetics 22: 389–407.Find this resource:

                                                                      Moss, L. (2014). Corpus Stylistics and Henry James’ Syntax. Unpublished PhD thesis. London: University College London.Find this resource:

                                                                        Noyes, Alfred (1906). ‘The Highwayman.’ Blackwood’s Magazine vol. CLXXX: 244–7. Edinburgh: William Blackwood and Sons.Find this resource:

                                                                          Phillips, Susan (2011). ‘Chaucer’s language lessons’. The Chaucer Review 46(1 and 2): 39–59.Find this resource:

                                                                            Rumens, Carol (2002). Hex. Hexham, Northumberland: Bloodaxe Books.Find this resource:

                                                                              Semino, Elena (2014). ‘Language, mind and autism in Mark Haddon’s The Curious Incident of the Dog in the Night-Time’, in Monika Fludernik and Daniel Jacob (eds), Linguistics and Literary Studies. Berlin/New York: Mouton de Gruyter, 279–303.Find this resource:

                                                                                Simpson, Paul (1993). Language, Ideology and Point of View. London: Routledge.Find this resource:

                                                                                  Sinclair, John (2004). Trust the Text: Language, Corpus and Discourse. London: Routledge.Find this resource:

                                                                                    Spender, Dale (1980). Man Made Language. London: Routledge and Kegan Paul.Find this resource:

                                                                                      Stockwell, Peter (2002). Cognitive Poetics: An Introduction. London: Routledge.Find this resource:

                                                                                        Stubbs, Michael (2005) ‘Conrad in the computer: Examples of quantitative stylistic methods.’ Language and Literature 14(1): 5–24.Find this resource:

                                                                                          Van Peer, Willie (1986). Stylistics and Psychology: Investigations of Foregrounding. London: Croom Helm.Find this resource:

                                                                                            Werth, Paul (1999). Text Worlds: Representing Conceptual Space in Discourse. London: Longman.Find this resource:


                                                                                              (1) With thanks to The Society of Authors as the Literary Representative of the Estate of Alfred Noyes for permission to use this extract.

                                                                                              (2) Arguably, the Adverbials could instead be analysed as postmodifiers within the Complement noun phrases.

                                                                                              (3) See for the full text and audio of this and other speeches by Churchill. Last accessed 14 March 2019.

                                                                                              (4) Note, however, that identifying typical features of an author’s style is not the same as establishing beyond doubt who actually wrote a text. See Kahan (2015) for a sceptical history of stylometrics.

                                                                                              (5) Started at the University of Nottingham, and now a collaboration with the University of Birmingham, this software was first developed to investigate the language of Dickens, but is now being developed for wider application to literary language. One of its features is the ability to locate direct speech in suspended quotations (where the reporting clause interrupts the quoted speech). See Mahlberg et al. (2016) for more information.

                                                                                              (6) With thanks to Carol Rumens for permission to use this extract.

                                                                                              (7) Extract used by permission of Carcanet Press Limited; © Carcanet Press Limited, Manchester, UK.

                                                                                              (8) Extract used by permission of Carcanet Press Limited; © Carcanet Press Limited, Manchester, UK.

                                                                                              (9) Extract used with permission of The Poetry Business, Sheffield, UK.