1 Language and style

1.1 What is stylistics?

Stylistics has been defined as a sub-discipline of linguistics that is concerned with the systematic analysis of style in language and how this can vary according to such factors as, for example, genre, context, historical period and author (Crystal and Davy 1969: 9 and Leech 2008: 54). For instance, there is the individual style that distinguishes one writer from another, the styles associated with particular genres (e.g. ‘newspaper language’ or the gothic novel), or the characteristics of what might constitute ‘literary’ style. In this sense, analysing style means looking systematically at the formal features of a text and determining their functional significance for the interpretation of the text in question (Wales 1989: 438). In fact, the growth of stylistics over the last twenty or so years has meant that this definition no longer captures every aspect of stylistics, and part of our aim in this book will be to outline the remit of stylistics as it stands today. For example, during the 1980s interest began to grow in the role of the reader in interpreting texts (see, for example, Alderson and Short 1989 and Short and van Peer 1989), and recently there has been a surge of interest in the cognitive aspects of text comprehension (see Stockwell 2002a and Gavins and Steen 2003). The connection between stylistics and linguistics is that stylistics uses models of language, analytical techniques and methodologies from linguistics to facilitate the study of style in its widest sense. With some notable exceptions (e.g. Crystal and Davy 1969, Enkvist 1973, McIntyre et al. 2004, Jeffries 2007) stylistics has tended to concentrate on the analysis of literary texts, though there is in fact no reason why this should necessarily be the case. In this book we will concentrate our attention on both literary and non-literary texts.

Stylistics has its roots in the formalist school of literary criticism that emerged in Russia in the early years of the twentieth century, though the term ‘style’ goes back to classical rhetoric and poetics. The prime exponents of Russian formalism were Roman Jakobson, Victor Shklovskii and Boris Tomashevskii, and the aims of the movement were to isolate the properties and characteristics of literary language (notice the predilection for literary texts), and to explore how the concept of defamiliarisation¹ in both art and literature was at the root of the intrinsic aesthetic value of the work in question. At the heart of Russian formalism was the belief that the purpose of all art was to defamiliarise the familiar in order to generate for the viewer or reader a new perspective on the topic of the piece of work under consideration.

As a movement, Russian formalism was hugely influential for a large part of the early twentieth century, though its prime tenets were to prove unsustainable (see Simpson 1996: 7–19). The notion that it should be possible to delineate the formal (i.e. linguistic) features which figure in ‘literary’ as opposed to ‘non-literary’ language was eventually shown to be misconstrued. Indeed, as we shall show throughout this book, features of what we might intuitively think of as ‘literary language’ are equally common in non-literary genres, and stylistics nowadays tends to see ‘literariness’ as a point on a cline (see Carter and Nash 1990: 34) rather than as an absolute. Literariness in this sense is not a quality of a text, rather it is a concept belonging to a specific genre. Similarly, the contextual (social and cultural) aspects of what makes a text literary have been increasingly recognised in stylistics as elsewhere. Nevertheless, the impact of Russian formalism on the development of stylistics was of immense importance, and its influence is seen particularly in the psychological concept of foregrounding (see Mukařovsky 1964, Leech 1969: 56–72, van Peer 1980, 1986, Fowler 1986: 71 and Douthwaite 2000), which we will be concerned with in Chapter 2.

The predominance of literary texts as the focus of study within stylistics is reflected in some of the alternative names that stylistics sometimes goes by. These include literary linguistics, critical linguistics, literary semantics, literary pragmatics and poetics, and are mostly an attempt to find a term for the full range of activities practised by modern stylisticians, as well as an attempt to acknowledge that stylistics is not simply concerned with identifying formal features of style in language.² The predilection among some stylisticians for the analysis of literature also leads to a number of potentially confusing names for particular approaches to stylistics. These include literary stylistics and linguistic stylistics, and it is worth clarifying here the slight differences in focus of these two approaches. Sometimes, a distinction is made between literary and non-literary stylistics, and such a distinction usually refers to the kind of texts commonly studied. Hence, literary stylistics in this sense is concerned with the analysis of literature whereas non-literary stylistics is concerned with the analysis of non-literary texts. However, where the term literary stylistics is used in contrast to linguistic stylistics, the distinction is not between the kinds of texts studied, but between the objectives behind such analysis. Literary stylistics in this case is concerned with using linguistic techniques to assist in the interpretation of texts, whereas linguistic stylistics is about doing stylistic analysis in order to test or refine a linguistic model (Wales 1989: 438) – in effect, to contribute to linguistic theory. Most stylisticians would argue that what they do is a combination of both of these things, and this is the approach that we take in this book.

1.2 The need for stylistics

Stylistics has a firm place within linguistics, providing theories of language and interpretation which complement context-free theories³ generated within other areas of language study. Nevertheless, the suggestion that stylistics is concerned with literature more than linguistics is a common criticism from theoretical linguists, though it is countered here by the corpus linguist, John Sinclair, who emphasises the importance of (literary) stylistics for the study of language when he says:

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)

The alternative criticism – that stylistics is concerned too much with language and not enough with literary concerns – might characterise the censure of stylistics from a literary direction. Our response to these criticisms is to say that whilst literary texts are, typically, the data upon which stylistic theories are developed, tested and applied, in the same way that, for example, spoken conversation tends to be the data used by sociolinguists, nevertheless, the stylistic features we will discuss in the book are not exclusive to any one genre, and stylistic techniques can be applied equally to non-literary texts. In discussing particular stylistic features we will draw on texts that best exhibit the features under discussion, and we will take examples from both literary and non-literary texts to illustrate analytical techniques. In addition, we will demonstrate that the value of a stylistic approach, whether from a literary or a non-literary viewpoint, is the precision and detail with which we can describe the textual effects of literature, whether our focus is the text itself, the reader's contribution or even some notion of authorial meaning. Stylistics has no settled view of the relationship between author, text and reader, but constantly evolves new theories and models of this dynamic relationship, in order to elucidate ever more clearly the processes by which meaning comes about.

Stylistics draws upon theories and models from other fields more frequently than it develops its own unique theories. This is because it is at a point of confluence of many sub-disciplines of linguistics, and other disciplines, such as literary studies and psychology, drawing upon these (sub-)disciplines but not seeking to duplicate or replace them. This versatility of approach and open-mindedness are, of course, characteristic of the humanities in general. Instead, it takes a particular view of the process of communication which places the text at the centre of its concerns, whilst being interested in the relationship between writer and text, and reader and text, as well as the wider contexts of production and reception of texts.

So, as well as drawing upon the descriptive apparatus of context-free linguistics, such as the formal descriptions supplied by structuralists, generativists and other twentieth-century linguistic movements, stylistics also takes into account the concerns of pragmatics on the one hand and sociolinguistics on the other, though these latter two sub-disciplines of linguistics have tended to be more interested in spoken than written language whereas stylistics has traditionally been concerned more with written than spoken texts. In other ways, stylistics also shares boundaries with cognitivist approaches to language, and, increasingly, with corpus linguistics. If what is described in the remainder of this book at times sounds rather similar to pragmatic, sociolinguistic, cognitive or corpus-linguistic accounts of language phenomena, this will be no more than we expect, as stylistics shares many concerns and theories of language and meaning with these fields, sometimes with only slight differences of emphasis. (For readers new to the terminology in this paragraph, we would reassure you that such terms as are needed to understand this book will be introduced and explained as they arise.)

In general terms, then, we have claimed that we need stylistics to provide an angle on language study which places the text at the centre of its concern. But what, in particular, might we expect stylistics to provide in relation to individual texts?

1.2.1 Literary explanations

The first and driving impulse of stylistics was to use the growing field of linguistics to explicate the textual effects that literary scholars may have agreed upon, but had neither the terminology to explain nor in many cases the wish to do so, being interested instead in other approaches to the study of literature which were less concerned with the language and more with the ideas encompassed by literary texts. Though some aspects of classical rhetoric could still be used to analyse the language of literature, this tended to be less appropriate to use with contemporary writing, whose authors were not trained in classical rhetoric and whose goal in writing works of fiction and poetry was not self-consciously rhetorical. Rhetorical analysis also worked at a more general level than the emerging linguistic analysis, the latter suggesting that it would become possible to explain in more detail the workings of the literary text and its rhetorical or other effects. Thus, for the first time, there was the opportunity to explain linguistically a range of literary features including the musical effects in poetry and the narrative complexities of modernist novels.

We will demonstrate the stylistic analysis of such different text-types throughout the rest of the book, but for now, let us consider one illustrative example of how linguistics might inform literary analysis. A principle that all stylisticians subscribe to is that meaning in language comes about through the linguistic choices that a writer makes (either consciously or unconsciously).⁴ The following short extract from James Fenton's poem ‘A German Requiem’ illustrates this:

The second line of this extract stands out because of Fenton's use of the verb ‘forget’, which seems unusual in this context. Prototypically, we might expect ‘remember’ in its place. This is because the phrase ‘old times’ has a positive semantic prosody – that is, it collocates most often with positive concepts. We will discuss the concepts of collocation and semantic prosody in Chapter 7, but briefly, collocation refers to the fact that particular words frequently occur within close proximity to a restricted set of words, and semantic prosody refers to the meanings that delexicalised (i.e. relatively ‘empty’) words take on as a result of collocational patterns. Since ‘old times’ has a positive semantic prosody, it therefore seems appropriate for people to want to get together and remember rather than forget them. Moreover, ‘forgetting’ is not normally considered to be a conscious act, so it is odd to imagine a situation where a group of people would gather together in order to collectively and actively forget something – on the face of it a logical impossibility. We might describe the verb, technically, as changing its transitivity pattern, being used to describe an intentional action instead of an unintentional one, or supervention. (Transitivity will also be discussed more fully in Chapter 3.)

The word ‘forget’, then, is foregrounded – it stands out as not being what we would normally expect. Fenton is essentially playing with our preconceptions and the effect is to focus our attention on the absurdity of actively trying to forget something – an action which takes on a much greater significance when we realise that ‘A German Requiem’ is a poem about post-war Germany and its efforts to come to terms with its immediate past. The surface-level interpretation here is something that would no doubt be picked up on by a careful ‘literary’ reading, but the linguistic detail added by a stylistic analysis enables us to see more clearly where the literary effect is coming from.

Further evidence of how a linguistic approach can explicate an interpretation of a literary text can be seen if we consider Fowler's (1977) notion of mind style. Basically, mind style refers to the idiosyncratic world view of a character or narrator, exhibited through the linguistic structure of a text. Leech and Short (1981, 2007) explain the concept through an extract from Faulkner's The Sound and the Fury. The narrator of the extract below is Benjy, a young man with the mental age of a child, whose unusual view of the world is reflected through the linguistic choices he makes when describing the game of golf he is watching:

We can notice, for example, that Benjy does not use any of the lexical items associated with the game of golf; for instance, he uses table rather than tee. Furthermore, he uses the transitive verb hit, which we would normally expect to have an object, as though it is intransitive and does not need an object. This, according to Leech and Short (2007: 166), reflects Benjy's lack of understanding of the concept of cause and effect. The hitting that he witnesses is, for him, to no purpose; there is, literally, no object to the men's activity.

The reasons why such explanations of literary effects appealed to early stylisticians were various. It meant that some attention could be given to the question of whether ‘literariness’ was indeed a linguistic phenomenon, and to what extent features that were to be found in poetry were also to be found in other texts, such as advertisements, sermons, speeches of all kinds and so on. For example, foregrounding, of the sort described in example 1, can also be found in adverts and other kinds of non-literary writing. Example 3 is from a retirement announcement in a Huddersfield University staff newsletter. The final word has been blanked out here, and it is probably fair to say that most people, if asked, would guess that this last word is likely to be ‘family’ or ‘children’ or such like.

The reason that most people would choose a noun relating to family members to fit the blank slot in example 3 is again because of common collocational patterns. There is a restricted set of words that we would normally expect to follow ‘wife and’. In actual fact, the word in the original announcement was ‘caravan’, giving rise to a humorous effect,⁵ largely because the word ‘caravan’ is foregrounded as a result of not conforming to our expectations for this frame. This deviant collocation appears to put ‘caravan’ on a par with children and other family members, leading us to make inferences about the sort of person who would care as much for a caravan as for members of his family (depending on the cultural schemata of the reader, the word ‘caravan’ may also be imbued with certain negative connotations, perhaps deriving from long-remembered childhood holidays…). It turns out, then, that the kinds of features common to literary texts turn up just as much in non-literary texts too, though one thing we have not done here is to consider the literary merit or value of such language use. Such questions of aesthetics have normally not been the main concern of stylisticians, but it is not completely unknown for linguistic definitions of literary value to be debated too (see, for example van Peer et al. in press).

Much recent work in literary stylistics has been concerned with the reader's role in the creation of textual meaning, and this has led to an explosion of cognitive approaches which have taken the view that stylistics can help to explain the ways in which readers put together a meaning from a text. Some of these approaches have contributed to our understanding of how readers conceptualise the world described by a literary text, and this has led to the development of text world theory (described fully in Chapter 6) into a rich account of this process of imagining hypothetical situations and events in reaction to textual triggers. Hoover (2004a), for example, shows how changing just one or two of the words in a piece of writing can radically affect our interpretation of that text. In example 4, the words Mr and Miss are the only additions to this extract from William Golding's novel about prehistoric man, The Inheritors:

Hoover makes the point that adding these honorifics to the text greatly restricts the way in which the reader can form an image of the text world. Mr and Miss are titles which have no place in a prehistoric world, and their presence in the text therefore prevents the reader from generating the kind of text world that is triggered in the original text. Instead, Hoover suggests, the titles are likely to trigger the reader to generate an image of a foreign or colonial fictional world (note that this is especially likely if the reader has a Western European background). The full account of text world creation that stylistics can provide is a further indication of the value of stylistics for contributing to our understanding of how readers make sense of texts.

Other stylistic approaches which take a cognitive view have included the use of schema theory (from psychology) to explain certain kinds of responses to literary (and other) works. Some researchers (e.g. Cook 1994) have claimed that one of the distinctive features of literary works relates to the effect that they have on the perceptions of the reader. To the extent that they change the reader's ‘schemata’ – or standard ways of understanding the world – they are more or less ‘literary’ in effect. This viewpoint has been debated (see, for example, Jeffries 2001 and Semino 2001), but it points to one of the useful and productive features of stylistic study, which is the ability to advance thinking about literary value as well as processes of reading and negotiating meaning.

1.2.2 Language, rhetoric and power

The explanatory power of stylistics can also help us to understand in more depth the ways in which the style of texts can help to influence the perceptions of readers in more everyday situations, such as listening to political speeches, responding to advertisements and so on. One example, from a teenage magazine (Jump), is discussed in Jeffries (2007b):

Here, the writer of the article manages to create a ‘new’ semantic opposition which is a more specific version of the superordinate opposition between normal and abnormal, ubiquitous in such magazines and likely to have a strong influence on the young girls who read them. Jeffries comments on the structure which makes this ideology possible:

This distancing from abnormality is achieved by the negation of a case which is pumped up by a hyperbolic and negative description of earthy types. The constructed opposite of the normal male (who doesn't like such women) and the abnormal male who does is compounded by the constructed opposition between experiencing life and eco-terrorism, the latter in some sense being a lack of living in the terms of this writer.

(Jeffries 2007b: 114)

Stylistics has evolved a detailed linguistic account of the kinds of persuasive techniques which are more generally covered by classical rhetoric. The use of stylistics for these purposes enables scholars to approach the explicitly persuasive aspects of style as linguistic phenomena, with the similarities between these rhetorical techniques and literary style also constituting comparative data for each other, since the tools available for the analysis of both these effects are essentially the same.

In addition to providing a more technical account, then, of persuasive textual features, stylistics, aided by insights from other fields such as pragmatics and discourse analysis, may also provide us with an account of the more implicitly manipulative uses of language. The development of critical discourse analysis, though not a product of stylistics alone, can nevertheless be seen as one of the sub-disciplines that has a family resemblance to stylistics, particularly in its positioning of the text at the centre of its concerns. This field is concerned with how the texts that surround us may subtly and sometimes even deliberately influence our political, social and even consumerist outlooks. For instance, Fairclough (1992), in an analysis of an extract from a university prospectus, shows how the language of the text constructs students as consumers and higher education as a commodity to be advertised in the same way as any other marketable product. In relation to his chosen text, he makes the point that the phrase ‘You will need’, which precedes a graphical representation of the university's entry requirements, shifts the emphasis away from the university as an authoritative gatekeeper, and instead constructs the student as a powerful consumer rather than a powerless applicant. This shift in emphasis is also realised through the university's entry requirements being outlined in a graphic, which, according to Fairclough, marginalises the entry conditions so that they are ‘construed as matter of fact which no-one is responsible for’ (Fairclough 1992: 214). Thus, the linguistic structure of Fairclough's example text reflects the increasing ‘marketisation’ of higher education.

The ‘exposing’ of such potentially insidious uses of language has been one of the most radical uses of the techniques of stylistic and critical discourse analysis, though it is instructive to see that it is at the level of interpretation, in context, that these more political considerations enter the discussion. The analytical techniques, varied as they are, are all available to the stylistician (see, for example, Simpson's 1993 use of CDA techniques for literary analysis), whatever the main purpose of the research project in question.

In summary, then, we need stylistics because much of our lives is negotiated through language, and though this language is well-described in structural terms by descriptive linguistics, and in contextual terms by such disciplines as discourse analysis and pragmatics, there remain insights about textual meaning that are addressed more effectively by a discipline which arose from literary studies, took on the apparatus of linguistics, and with the text at its core, became a powerful discipline in its own right. How, for instance, do we understand the hypothetical world in a sci-fi novel? Why does a particular line of poetry seem to move not only one reader, but a number of readers? Is there something subtly persuasive about Tony Blair's speeches, and how do insurance companies attract (and keep) our attention? All of these questions can be addressed through stylistic analysis.

1.3 The scope of stylistics

This section will introduce some of the parameters of stylistics which define its scope. In some cases, the topics explored here will be discussed in more detail later in the book. The following sub-sections, therefore, will delineate the boundaries of stylistics with reference to the kind of texts that it studies, the theories that it draws upon and the methodologies that are available to those working in this field.

1.3.1 Range of texts

In section 1.1 we made a distinction between literary and non-literary texts, both of which are studied by stylisticians. Though the origins of stylistics are in the literary field, and many stylisticians even today consider literature their field of study, it rapidly became clear that the techniques of analysis being developed in this hybrid discipline were as applicable to non-literary texts as to literary ones.

There is, therefore, in principle, no restriction whatsoever on the kinds of text that may be subjected to stylistic analysis. However, there are both historical and also practical reasons why there has been more emphasis on the literary aspects of style in the past, and also on the written language in preference to the spoken. Now that recording techniques have made the capture and transcription of spoken texts more accessible, we may find that stylistics concerns itself with the full range of linguistic usage. Certainly, there have been recent examples of stylisticians turning their hand to look at the language of spoken conversation (McIntyre et al. 2004), advertisements (Short and Wen Zhong 1997, Jeffries 2007b), humour (Simpson 2004) and film (Simpson and Montgomery 1995).

This latter development in the direction of multimodality, of course, is not restricted to stylistics, but is also reflected in the move from literary theory to ‘cultural theory’ and the increasing interest that all such commentators are taking in not only linguistic, but also visual communication of all kinds.

1.3.2 Range of theories

Stylistics, as we shall see throughout this book, is eclectic in its use of theory, though it originated in literary theories of formalism and took on the theory of structuralism as developed by Saussure (1959 [1916]) in the early twentieth century. What these theories together provided was the descriptive apparatus (such as grammatical and lexical terminology and categories) which would enable scholars to pinpoint the precise techniques of construction that writers were using in order to demonstrate the linguistic basis of well-known literary effects, particularly those which were foregrounded. The focus on the actual language of the text which is epitomised by these theories is still present in some stylistic practice, and demonstrates that stylistics does not originate from an author-based view of textual meaning in the same way that, for example, some areas of literary studies did.

In time, stylistics responded to the developing of new theories of language, based more on contextual factors in the case of pragmatics and discourse analysis and on cognitive factors in the case of generative grammar and cognitive linguistics. With the ever-present aim of explaining textual meaning and effects, it was able to use the insights provided by all of these theories to support new analytical processes and provide new insights into the style of texts and their reception by a range of potential audiences. Thus, there are now stylisticians working alongside psychologists (see, for example, Sanford et al. 2004 and Schram and Steen 2001) to establish some of the processes by which readers respond to linguistic style. There are stylisticians working in critical discourse analysis (e.g. Fowler 1991, Mullany 2004), with theories of social exploitation and manipulation at the heart of their approach. There are also those working with computational and statistical theories (e.g. Culpeper 2002, Hoover 2004b), who draw literary and linguistic conclusions from the computer analysis of large quantities of data.

A recent set of developments in cognitive stylistics have also drawn on theories that are seen by some as beyond the scope of linguistics, such as psychology and philosophy, but have provided useful insights and models for analysing what is going on in the processing of texts by readers. These theories include schema theory, possible worlds theory and theories of figure and ground. Chapters 5 and 6 examine the interface between the analysis of style and such theories of the human mind, and exemplify theories of how texts can build up mental pictures of their topics in the mind of the reader.

1.3.3 Range of methods

In addition to drawing upon a wide range of theories about the nature of language and particularly the nature of reading, stylistics is eclectic in its use of methodologies. It would be true to say that theories, such as those mentioned above, produce possible models of what the language or text is like, and that these models also tend to dictate the methods to be used to analyse them. However, there is normally some choice of method to be made, even when a theory and a model have already been chosen. For example, if a stylistician wanted to find out whether the vocabulary of Shakespeare is really much wider than the vocabulary of Ben Jonson, the model of a vocabulary range as the number of different lexemes used by each of the authors will dictate a corpus-based (probably computerised) methodology in which statistical analysis will be paramount. It would be possible, in theory, to choose a different model for measuring vocabulary, such as the range of different senses of common words that are used, but it is still likely that the results would be statistical.

This brings us to the most important methodological distinction in all research, that of the difference between qualitative and quantitative methods. Traditionally, most stylistics has been qualitative, though some slightly separate offshoots, such as stylometrics⁶ (the study of authorship attribution), have been more quantitative in method. In recent years, with the development of easily accessible and powerful computer software, there has been renewed interest in quantitative study in stylistics; indeed, there is even a developing sub-discipline increasingly referred to as corpus stylistics.

It is probably relatively easy to recognise what we mean by quantitative study, since it clearly involves the statistical analysis of elements from large quantities of data, in order to test the significance of numerical findings. Thus, one might compare the incidence of high-frequency function words (such as pronouns and determiners) by different characters in Jane Austen's novels, as Burrows (1987) does, and discover that Austen's characters have their own unique stylistic ‘fingerprint’. Once statistically significant differences are found, the literary and stylistic questions of what these differences mean can be discussed.

The question of what constitutes qualitative study is less easy to define. There are as many ways of carrying out qualitative study as there are people and texts, though we will see in section 1.6 that there are certain guiding principles of stylistics which constrain the range of possibilities a little. For now, we need to consider a few examples of what might be included under the heading of qualitative study. An analysis of a single poem, for the sake of analysing that poem alone, might well be qualitative, particularly if the poem is fairly short. It would make no claims about texts other than the poem itself, though it may well indicate potentially fruitful directions for future study. It would be possible to count the features of a short poem, but no statistical significance is likely to be provable with so few cases.

The advantage of qualitative study is that there is the possibility of taking many more of the contextual factors into account, and this means that one is likely to use a different range of tools. Take, for example, the reporting of the so-called ‘war on terror’ which has been a regular feature of news reporting since 11 September 2001. Whilst it would be possible to collect every single occurrence of this phrase in news reporting since that date and subject these to computer-based analysis, it is also possible to choose, for example, a few texts, possibly reflecting different attitudes or concentrating on a single incident, and scrutinise them in detail, using some tools of analysis which may not be amenable to automatic searching (see, for example, Montgomery 2004). The result of this analysis would be some insights into the texts themselves and, depending on how they were chosen (e.g. random or structured selection), may also have implications for data beyond the data analysed, which could be tested by others or at a future date.

1.4 Aims of stylistic analysis

The aims of stylistic analysis are varied and reflect the rich range of approaches taken to all sorts of text-types and genres. This means that it is not easy to narrowly define what the aims of all stylistic analysis will be. However, in order for the reader to grasp some of this range, we will attempt here to rationalise the practice of stylistic research under two main headings, reflecting the distinction between what is sometimes called ‘bottom-up’ versus ‘top-down’ approaches. Before we look at these two broad approaches in detail, it should be made clear that this distinction cuts across the one made in the last section, between quantitative and qualitative research. There is a tendency for quantitative research to be more inductive (bottom-up) and for qualitative research to align with deductive (top-down) approaches, but this is no more than a tendency and the researcher embarking on a project should consider carefully the particular combination of approaches that s/he will take.

1.4.1 Starting with data

The task that stylisticians often set themselves is to analyse texts and draw from their analysis any patterning or features that they find there, without necessarily wanting to generalise these findings beyond their data. This ‘inductive’ approach (see section 1.6) suits the material of stylistics very well, and has perhaps been more widespread in the past than the alternative (see 1.4.2).

It is probably worth pointing out that for many stylisticians, individual examples of various features (such as a particular metaphor or symbolic use of grammatical structure) will be all the results that are needed to fulfil the research aim of her/his project. Thus, a scholar might wish to use a stylistic approach to discovering some of the techniques that Wordsworth's Prelude seems to make use of in conjuring up the mountains of the Lake District or the excitement of skating on a frozen lake in winter. This will require a close analysis of the passages concerned, followed by an interpretation of the technical features that are discovered there.

To take a different scale of example, let us consider a hypothetical study of the style of a contemporary novelist, such as Margaret Atwood or Julian Barnes. Depending on the precise research questions being posed, we may want to consider the whole of the author's output, or a section of it (according to dates maybe) or possibly to sample it either randomly or with some structured sample in mind (see section 1.5). Whatever the precise focus of the research, the impetus is to find out from the data what patterns or features of language make it distinctive or make it work in particular ways for the reader. This could involve a computer-based search for lexical or grammatical features, or informant-testing for reader reactions, or a qualitative close analysis of selected passages. Whatever the decisions about the precise data set and the method of analysis, the data sits at the centre of the research project.

Readers will be able to imagine many other research projects of this nature, dealing with all kinds of data, and using different methodologies. Some examples to demonstrate the range of work that could still be considered to be data-driven include Jeffries's (2000) analysis of deictic positioning in Carol Ann Duffy's poetry, Semino's (2002) study of linguistic variation in poetry and McIntyre's (2005) investigation of mind style in Alan Bennett's play The Lady in the Van. Note that underlying this type of analysis there are assumptions about the nature of language and theories that help to explain it. The outcomes of such research, too, might well include some development of the theories and models of language, in response to the analysis. However, in general terms, we may still wish to characterise some stylistic research as fundamentally asking questions about the data. This was called ‘literary stylistics’ in section 1.1, but we can now see that the data-driven type of stylistics may also be asking questions of non-literary data.

1.4.2 Starting with theories

Whilst many stylistic research projects are concerned with questions about texts, using theories and models developed in other fields as convenient, there are also approaches which might be seen as more deductive (see section 1.6) and theory-driven than these. Such a ‘top-down’ approach would probably involve asking general questions about the nature of textual meaning, possibly including questions about the process of reading, and the reader's role in negotiating textual meaning. The part that textual data plays in this kind of project is secondary, and serves the general aim of pushing forward the theoretical view of how language works. For example, a stylistician may be interested in how metaphor works in literature from the point of view of the reader, and decide to do some informant-testing with particular examples of metaphors from literary sources, to establish how they are received by readers (see Steen 1994 for an example of this kind of study). The aim of the research, in this case, is to find out more about how language works in general, though incidentally there will also be some outcomes which shed light on the workings of particular metaphors. Note that the amounts of data (see section 1.5) are relatively small for this kind of project, but this is not necessarily the case in theory-driven projects, which can necessitate large amounts of data where the research aim requires it. It would, therefore, be possible to devise a theory-based project whose aim was to demonstrate that, for example, words which regularly collocate with certain other kinds of words may carry with them an association with that collocational tendency. This idea is at the root of Louw's work on semantic prosodies (Louw 1993) for instance, and he uses examples from the poetry of Philip Larkin to demonstrate the theory in action. Thus, he is able to show, for instance, that the poem by Larkin which begins ‘Days are where we live’ evokes a kind of negativity because, perhaps surprisingly, the words ‘days are’ tend to collocate with negative words in a significant majority of cases found in large computer corpora.

What is important to note here is that ultimately it is not necessarily the case that stylistic research is only either data-driven or theory-driven. Though many studies will begin from one of these positions, it is very often the case that both impulses are served by a single piece of research.

1.5 Data in stylistics

It may sound like a truism, but we should start by establishing that stylistics cannot happen without some language to analyse, or ‘data’. The reason that this is important is that some sub-disciplines in linguistics, and particularly (but not only) general and theoretical linguistics, can at times function without any particular collection of data upon which to carry out analysis. Thus, the developments of some grammatical theories, approaches to language acquisition and so on, have progressed at times with only the smallest number of illustrative examples, and these may even be invented by the researcher, rather than being in any sense ‘naturally occurring data’. This is not to criticise such endeavours, since they have often produced very large steps forward in our thinking about language, and the kinds of ‘thought experiments’ carried out by philosophers of language in particular produced theories of language which underpin much of linguistics as we know it today.⁷

However, stylistics is different. It is fundamentally attached to data in a way that other branches of linguistics are not. The central concern of stylistics is with the style of particular texts, whether they are representative of a genre, an author, or themselves alone. The context in which they are produced and received cannot be ignored, as we shall see, as this affects their ‘meaning’ in a range of subtle ways. But the unavoidable basis of all stylistics remains the text itself, and the linguistic choices that have been made (albeit unconsciously) to arrive at a particular form of words.

If data is at the heart of stylistics, then, we need to establish what kind of data, and make other decisions about the range and scope of the kinds of texts that this sub-discipline can encompass. This section addresses some of the important questions that stylisticians need to answer. This is intended to delineate the fundamentals of stylistics enough for students to be able to begin to do stylistic analysis of the kind outlined in Chapters 2 to 5. We will then revisit some of these methodological considerations in Chapter 7, when the aims and methods of stylistics are investigated in more detail.

1.5.1 Literary and non-literary texts

In part, we have already considered (see section 1.1) one of the most important questions for stylistics, this being the extent to which stylistics is concerned with only literary texts. The simple answer, as we have stated already, is that it isn’t, though there are many stylisticians who only work with literary texts, and a few who think stylistics ought to be so restricted, in the tradition of classical rhetoric and poetics. In the main, though, there is a recognition that however we might try to define literature as distinct in some way from ‘ordinary’ language, the same kinds of stylistic phenomena turn up in all sorts of text, and it is thus difficult to make a clear distinction on linguistic grounds. Whilst there may be no absolute linguistic distinction between literary and non-literary works, there is clearly a functional and social difference between, for example, a novel and a gas bill, or, to make the texts more obviously similar, between a short story and a ‘real-life’ story told by one neighbour to another. Some stylisticians are particularly interested, for good reasons, in the linguistic detail of literary works, because their findings may inform a literary response to the work by, for example, giving some indication of why readers may respond in some way to particular effects, or how well-known effects are actually achieved in linguistic terms. We have already seen some examples of this, and will see more in later chapters.

Whilst we are considering the literary end of stylistics, we should address the question of what is sometimes labelled ‘high’ literature, in other words ‘great’ literature, versus the popular end of the spectrum. These two kinds of literary work are often thought of as being situated at either end of some kind of cline on which can be found the more elite, statusful, canonical novels, plays and poetry and the popular, low-status work of fiction, drama or verse. It is not necessarily as simple as a single range, of course, since much of the division of literature into such categories depends more on social judgements than linguistic ones, and there may be different opinions about where on the scale a particular work belongs in relation to others. Does, for example, a novel by Stephen King or a poem by Dorothy Parker belong on the popular end of the range compared with a ‘chick-lit’ novel like Bridget Jones’s Diary (Fielding 1999) or the poetry of Wendy Cope or Pam Ayres? The latter example is particularly interesting, given that Cope's poetry is funny, formal (rhyming and metrical) and deals with everyday subject-matter. Parker's and Ayres's verse is, likewise, light-hearted, formal, metrical and funny. Cope is published by what might be considered a ‘serious’ poetry publisher (Faber and Faber) whereas Ayres is seen more as an entertainer and a comedian, as evidenced by the fact that her verse is published by the BBC.

The last paragraph implies that there is no real difference between the work of writers who have very different status in literary circles. However, we have not analysed the poems of Cope and Ayres stylistically, and have only the most superficial impression of their similarities in subject-matter and form. Were we to do so, we may find that there are indeed some stylistic differences which may account for their different status, but it may also be that there is no linguistic difference, and only a difference of presentation and context that accounts for their respective reputations. The way that a stylistician may approach questions of high versus popular literature, then, might be to take texts from genres and authors which have different social values, and consider to what extent there is indeed some linguistic indication of the differences between them. This valuing of a certain aesthetic would be context-bound, in the sense that it would have its origins in the society in which it accrues such aesthetic value, and there is one type of stylistics that is indeed interested in charting the linguistic aspects of different aesthetic judgements in this way. However, stylisticians are not only interested in what might be seen as ‘great literature’ in certain societies. They are also interested in popular literature for its own sake, and might, for example, be found investigating what it is that makes popular romance novels, such as those published by Mills and Boon, so distinctive in their style that they form a sub-genre on linguistic grounds, as well as on the grounds of plot, characterisation and so on.

We have established, then, that the data considered by stylisticians may be ‘literature’ in very broad terms, including popular fiction as well as canonical and high-status works. We have also suggested that non-literary texts may be of as much interest to stylistics as literary ones. One simple reason why this is the case is related to stylistic techniques themselves. When early stylisticians attempted to define literature in linguistic terms (see, for example, Havránek 1964 and Mukařovsky 1964), it became clear that, on the one hand, many of the so-called literary ‘devices’ (such as punning or metaphor) were present in other, non-literary, texts such as advertisements and that, on the other hand, any style of language could occur in literary works, depending on the subject-matter. For instance, cognitive metaphor theory (discussed fully in Chapter 5) has shown how metaphor is present in all discourse types since metaphor is one of the primary ways in which we conceptualise our experience of the world around us. As an example, the conceptual metaphor LIFE IS A JOURNEY is a structuring device that underlies our experiences of daily life and the way in which we talk about these. Common phrases that make use of this conceptual metaphor include ‘I feel like I’m going nowhere’, ‘I’m at a crossroads in my life’ and ‘I don't know which path to take’. The underlying conceptual metaphor, then, is one which is found in many different kinds of texts and spoken language, and is clearly not restricted to literary texts.

This is not to say that the presence of certain conceptual metaphors (for example, novel metaphors) wouldn't have potentially different effects in literary contexts, something which we will come to, but it certainly belies any notion that the actual words of literary texts are in any sense different from other texts. The question of the effects of texts is one that stylistics has recently begun to grapple with in many different ways, as we shall demonstrate in later chapters. Here, it is worth pointing out that the stylistic analysis of non-literary texts may use the same tools, but may be concerned with a different set of questions in terms of the effect or interpretation of the texts themselves. Thus, critical discourse analysis and other stylistic approaches in a similar vein may be ultimately concerned with the persuasive or manipulative effect of advertisements, newspaper reports or political speeches, whereas a literary stylistic analysis may be concerned with the literary value or interpretative effect of a literary work. This will not prevent them from using similar basic tools of analysis.

So far, we have discussed – and rejected – the possible narrowing of stylistic data to ‘high literary’ texts and have also rejected the notion of stylistics being only concerned with literary works in general. We also need to consider, given this opening up of data to all texts, whether spoken as well as written data might be the subject of stylistic analysis. Increasingly, this is the case (see for example Lambrou 2003). It may be worth pointing out first of all that the question of spoken or written medium cuts across the literary/non-literary division, since much ‘literature’ may be spoken, most obviously in the performance element of drama, but also in performance poetry, or even broadcast fiction on radio or television. Likewise, of course, much non-literary text is written, though there is also a great deal of spoken language in our everyday lives.

In principle, then, the question of whether texts are spoken or written does not affect their eligibility for stylistic analysis. In practice, much stylistic analysis has tended to concentrate on the written language, or a transcribed version of the spoken language, so that the same kinds of issues (lexis, grammar etc.) are normally considered more than, for example, intonational or other spoken phenomena.

1.5.2 Defining the data

Having established that there is, in principle, no restriction on the kinds of linguistic data that can be subjected to stylistic analysis, we have to consider what kinds of decisions need to be made when embarking on stylistic study. As we will see in section 1.5.4, these questions of defining the data for analysis arise out of the research questions that we might wish to ask, and also from the methods and theories that underlie our approach. Nevertheless, there are some issues of data definition that can be addressed in general and separately from the particular study or analysis.

The first question to consider is ‘How much data is needed?’ This question may be answered fairly automatically if, for example, the analysis is pre-determined as relating to a single poem or a pair of contrasting poems. Beyond such minimal data sets, there will always be questions of amount to be answered. If we consider the case of a single novel – for example, Great Expectations by Charles Dickens – any researcher setting out to study this novel stylistically would need to make a decision about whether every word in the novel was to be analysed, and if so whether this would involve a computer-based analysis. An example of such an approach would be Hardy's (2003, 2004, 2005) stylistic analyses of the work of Flannery O’Connor. An alternative, using more traditional and qualitative methods, would be to choose a set of extracts to analyse. This could be a randomly-selected set, to represent the novel as a whole, or it might rather be a set of selected extracts that are expected to contrast in some way, perhaps reflecting the different phases of Pip's life or relationships.

The decisions taken about how much data to analyse are likely to be in response to, and also to limit, the kinds of analysis that will be possible. A large-scale study, involving many thousands of words, makes for more statistical data and less contextually-based analysis. At different times these different approaches may serve the overall purpose equally well. What is most important, in all stylistic research, is to be clear about the data to be included, and to be clear about why this data set is appropriate. Thus, we may be aiming to find out why all of Alan Bennett's writing is recognisable as having been written by him, whether it is a play, a diary or a lecture. We will therefore need to make sure that the data includes some of each of these genres, and also that there is enough data to see any overall stylistic patterns that may be present. This means that the quantity of data is large, and will necessitate an approach that can cope with large numbers of words, such as corpus stylistics. Alternatively, extracts from across Bennett's works could be randomly selected and analysed in more detail and with regard to context. Whichever of these approaches is taken, the important thing is to define the process of deciding how much data to analyse, how it is selected and what it consists of. This is mainly so that the stylistic analysis that results can, if required, be checked against the data by other scholars, or compared with other studies of a similar nature or with similar aims.

If the data for study is not already well-defined (e.g. a whole poem), then the question of selection involves making decisions that are often relatively arbitrary. Nevertheless, it is important to try to use some kind of rationale to guide these decisions. For example, the choice of advertisements to study may involve a single day's (twenty-four hours’) worth of TV advertising, to investigate style in TV advertising generally and at a single point in time (see, for instance, Leech's seminal 1966 book on the stylistics of advertising). Alternatively, one might choose to concentrate on the advertising of particular products, for example cars versus make-up, which may reflect stereotypically gendered approaches to selling. If one is studying a novel, the choice of whether to choose ten 1,000-word extracts or two 500-word extracts may depend on the tools to be used and the purpose for which this analysis is to be carried out. The more different tools of analysis that will be applied to a text, the longer it will take to complete the analysis and the longer the written-up account of the analysis will be. Depending on whether the analysis is an undergraduate essay, a PhD thesis or a journal article, the amount of data and length of extracts will vary. This practical aspect of data definition is unavoidable, and not problematic, as long as there are clear reasons given for the choices made in relation to the research questions that are being asked.

1.5.3 Authorial versus genre style

In discussing the choosing of data, we have already mentioned in passing the question of exactly which ‘style’ the researcher is trying to analyse. This question relates, of course, to the main aim of the research and the research questions it produces, as we shall see in section 1.5.4. Here, we will simply exemplify the kinds of different style which could be the focus of stylistic study.

Clearly, the style of a single text (advertisement, letter, poem, novel, play) may be of interest in itself, or we may wish to look at a particular phase in an author's output (e.g. by poetry collection or groups of fictional works by date). Broadening out, we may wish to consider the whole output of an author, or the works of a group of writers who have some common theme or other connection, for example, absurdist writers, Romantic poets or writers of instruction manuals. In all cases, of course, we may wish not only to analyse a set of data for its patterns of similarity, but we may wish to construct a data set with inbuilt contrasts, to compare the sub-sets with each other.

Beyond the individual writer, and groups of writers, we may wish to consider the analysis of a whole genre, such as contemporary poetry in general or the style of Acts of Parliament. As the potential pool of data gets larger, questions of sampling become more significant. It is probably also the case that there is a sense in which the most common, and possibly also the most successful, stylistic analysis to date has looked at relatively limited and homogeneous sets of data. This position may well be changing in the light of recent developments in corpus stylistics, though it remains the case that the larger the data set, the less one is able to conclude in terms of distinctive style. With non-literary works (e.g. a gas bill) it is likely that the research will be looking for patterns across the whole of that text-type. Staying with the non-literary, then, we may wish to study the whole of a particular genre, such as begging letters, or planning objections and so on. In attempting to draw out the common stylistic factors in such genres, we will need to decide how to sample the available material, and this will take us back to considerations of quantity that were discussed earlier.

In addition to the genre and text-type approach to data selection, one may also have a different kind of focus, relating more to the topic of the texts than their genre. Thus, we may ask the question of what, if any, stylistic tendencies there are in texts on the internet relating to Christianity, or we may ask how a current political question (such as the proper funding of political parties) is treated across a range of text-types, including news reporting and political speeches. This will have an effect on the range, quantity and type of texts that make up the data for a particular study (see Chapter 7 for examples of this kind of work).

Finally, the focus of a stylistic study may be more theoretical than any that we have mentioned so far. This focus could be, for example, the question of how widespread the use of metaphors is in literary and other works, and whether these metaphors can be categorised in certain ways, irrespective of their context. Similarly, there could be a focus on opposition-creation across a range of text-types, including literary and non-literary texts, to see what general case can be made for contextual construction of new and unconventional opposites. Another example may be the development of the theory of mind style, and the use of a range of texts to provide a broad set of data from which to draw generalisations about how mind style is reflected in language (see, for example, Baker and McEnery 2005, Gregoriou 2006). In practice, stylistic studies are often a combination of these approaches, and may well provide evidence for a general theory at the same time as focusing on a topic or theme, and limiting the data to a particular genre or other set of texts.

1.5.4 Matching data to research questions

We have seen already that the question of what data to choose is related to the questions that you wish to ask. Most research projects have an overall aim, which is made more specific as a set of research questions to be answered. This set of questions will in some sense determine what data is needed. Let us take, for example, the question of the rather unusual style of the Irish poet Mebdh McGuckian. One possible way of approaching the question of what makes her style distinctive is to start with a single poem, and consider what foregrounded stylistic features occur in this text. The analysis could be followed up by a series of similar analyses, using the findings of the first analysis to inform the tools to be used in subsequent analyses. This process is rather like a set of ever-widening nets that catch ever more of McGuckian's poetry and, consequently, style. In theory the process would not be complete until all of her poems had been considered, though in practice some kind of sample might suffice. The question of whether such analysis truly captures McGuckian's uniqueness is harder to establish, since the task of comparing her with all known poets in terms of the particular combination of stylistic features that have been found would be unmanageable, and, indeed, unnecessary, since questions of evaluation are not necessarily the purview of stylistics.

At a different level, the stylistician may wish to try to define linguistically the stream-of-consciousness style often attributed to Virginia Woolf, among others. In this case, one might decide to narrow down the set of analytical tools one is using, to focus on this research question, but it will also be important to define the range of material that will be analysed. Woolf's novels could be randomly searched for 500-word extracts, to give a representative sample, but one might also justify choosing particular passages where the effects of changing consciousnesses are particularly striking, in order to analyse an impressionistically ‘typical’ set of data, and to produce a description of the range of techniques being used (free indirect style, for example – see Fludernik 1993), before testing these findings on further data.

As we have seen, the questions that arise when a stylistician chooses her/his data are many and varied. As we shall attempt to make clear in the next section, the underlying principles of the discipline may provide a guiding hand in answering them.

1.6 Principles of stylistics

As we have described, stylistics evolved out of the literary and linguistic developments of the twentieth century and it has developed so many strands and sub-fields that the sheer variety and exuberance of the discipline as a whole is hard to pin down to a set of procedures, theories or methodologies. Some fields of linguistics have developed a very clear and agreed set of standard practices, based on a consensus about current theories and models; the field of conversation analysis would be one such example. This has not so far happened with stylistics, perhaps mainly because of the enormous range of practices which seem to shelter under the label. Nevertheless, there are certain principles by which most, if not all, stylistics operates, and these will be the subject of this section.

1.6.1 Stylistics as text-based

As we saw in section 1.5, stylistics is very much based on textual data (though, arguably, the notion of what constitutes a text is becoming ever broader; see, for example, Fairclough's 1992 definition of a text as the written or spoken ‘product’ of discourse). In a sense, there can be no such thing as stylistics unless texts are being analysed, and although some recent developments in cognitive stylistics have hypothesised principally about the processes that a reader engages in when s/he is reading a text, ultimately the task of the stylistician remains that of working out what effect is achieved by particular texts, whether the research aims are phrased as a cognitive, a linguistic or a literary question.

This adherence to the centrality of the text has earned stylistics some criticism, not least from literary theorists (for example, Fish 1981) who have claimed that it is only a slightly more technical version of ‘close reading’, which is less fashionable than it used to be, because it seems to imply that literary works have single meanings, and that these can be ‘teased out’ by a set of analytical procedures. In fact, though early stylistics did indeed claim to provide a more rigorous set of tools for analysing literary language in detail, there has not been a time since the early twentieth century when stylistics wasn't changing and responding to new linguistic and literary theories and models. Thus, in the early twenty-first century, we have a range of practices subsumed under stylistics, including cognitive stylistics and corpus stylistics as well as stylistics based on functional and discourse-analytical approaches to language. At their most extreme, they vary considerably, but it is probably fair to say that they all remain text-based in the sense that they are trying to explain something about the operation and effect of particular texts.

1.6.2 Objectivity and empiricism

The second set of principles, which stylistics shares with other sub-disciplines in linguistics, derives from principles in the natural sciences, which have been variously taken on board by the human and social sciences as evidence of their status as reputable academic disciplines. Whilst literary studies has never made any attempt to be recognised as comparable with the sciences, linguistics certainly started out as a reaction against the notion that human communication couldn't be studied in a scientific way. There had, of course, been other attempts to be objective about studying literature, most notably I. A. Richards’s argument that literary criticism is essentially a branch of psychology, dealing with the states of mind induced by art (Richards 2001 [1924]; see also West forthcoming). It was on this scientific basis that Richards created a new school of practical criticism devoid of subjective emotionalism. This approach led to many developments in our understanding of how language works, and though stylistics itself straddles both literary and linguistic attitudes to scientific principles, it nevertheless owes a great deal to linguistics.

The first principle that we should consider is the importance of rigour in stylistic analysis. Probably the most general of the scientific principles, rigour refers to the way that research is carried out and written up, so that, whilst they may not share the same conclusions, other scholars can easily see the consistency and clarity in the work that has been done. To give an example, stylistic research which takes samples from larger bodies of data must be clear about the way in which the sampling was done, and why. This is the only basis upon which later commentators or reviewers can comment upon the sampling that has been done. Similarly, analytical tools must be clearly stated, whether or not they are a generally agreed set, and even if they are controversially applied in a new or different way. The important thing is to be transparent about what has been done, what methods have been used, and why. Whether quantitative or qualitative, it is vital that analytical tools are applied consistently, to alleviate the risk of bias or unconscious slanting of the analysis toward favoured outcomes.

This latter point, the reason for consistency, leads us to the question of objectivity and replicability in stylistics. Stylistics has on occasion been criticised over its claims to analytical objectivity (see Fish 1981 and MacKay 1996, for example). Whilst it is true that stylistics aims to avoid purely subjective commentary on texts, objectivity in stylistic analysis needs some discussion, since the notion of objectivity as a concept is often misunderstood (see Short et al.'s 1998 response to MacKay 1996 for a discussion of this). Objectivity in stylistic analysis does not mean making impassive comments on the meaning of a text without regard to context or ideology. As Simpson (1993: 3) points out, few stylisticians would claim to do this. Short and van Peer (1999) outline what being an objective analyst means when they explain that:

In trying to be objective, one tries to be (a) clear, detailed and open (so that one's position is unambiguous), and (b) ready to change one's mind if the evidence or a subsequent counter-argument demands it.

(Short and van Peer 1999: 273)

The aim, then, is to be as objective as possible, given the constraints of the subject-matter and the difficulties of distinguishing the analyst from the subjective reader. There are, of course, techniques for minimising the effect of the personal subjectivity of the individual analyst. These include the replicating of research by other analysts, and this can be built into projects so that more than one researcher applies the analytical tools to any single piece of data.⁸ At one extreme, there are tools which can employ computer-based techniques for making the analysis and these rely on the surface form of the text. Such corpus studies are perhaps the most clearly ‘objective’ in their operation, though as in any analytical procedure, the choice of categories and models may influence certain outcomes of the analysis. What corpus studies demonstrate most clearly is something that is true of all stylistic study, and this is the separation, in principle if not in practice, of the analytical and interpretative processes. As we will explain in section 1.6.4, interpretation is another of the important principles of stylistics. Here it is worth noting that objectivity is easier to approximate to at the analytical than the interpretative stage, though both are vital to stylistics.

The penultimate principle that we will discuss in this section, and which is also a scientific principle, is that of empiricism. The requirement of an empirical approach to study is that all claims should be based on observation or experience. In other words, the strictly empirical approach would insist upon an inductive method (see section 1.4.1) of working whereby the patterns observed in large quantities of data would be the only available outcomes of the study and no generalisations or predictions beyond these outcomes would be permissible. In practice, some branches of stylistics are more inductive (e.g. corpus stylistics) and some more deductive (e.g. critical stylistics) in method, and even within these sub-fields, many studies will in fact use both approaches to answering the research questions. Thus, many stylistic projects may analyse a set of texts, or extracts from texts, and describe what is found empirically by this method, whilst also making predictions from this basis about the likely occurrence of similar features and effects in other data. Note that the deductive method represented by these generalisations and predictions leads us to another requirement of research in stylistics, namely that it should be falsifiable. This requirement means that any claims must be clear enough to be able to be challenged by other researchers, either by replication of the original work, or by the application of the findings to new data. The requirement for falsifiability, then, takes us back to the need for rigour and objectivity, which together enable other scholars to challenge the findings of the original researcher(s). Chapman (2006) puts it this way:

What is important to a deductive hypothesis, indeed what makes it a valid scientific statement at all, is not that it can be supported by data, but rather that it can be falsified. That is, the hypothesis must make specific predictions that can be tested against data and that can in principle be proved wrong.

(Chapman 2006: 20)

The notion of falsifiability is important in stylistics (see, for example, van Peer 1980, Hogenraad et al. 1997). As we have seen, stylistics models itself on the scientific principles that underpinned the origins of linguistics, but we will also see that the reality of stylistic analysis – and some would argue any linguistic analysis – is that striving to be scientific is limited to some extent by the nature of the data that is being investigated and that eclecticism (see section 1.6.3) is not only necessary but acceptable.

This section has introduced a range of scientific terms which relate to each other in complex ways. One way of seeing the relationship between these concepts is to see the drive towards objectivity as the overriding scientific principle which is (partially) achieved by practical measures such as making sure that stylistic work is replicable and falsifiable. These two guiding principles may be seen as the core values of a rigorous approach to research in any field, including stylistics.

1.6.3 Stylistics as eclectic and open

As we have seen in relation to induction and deduction, stylistics is not a discipline to be constrained to one particular theoretical viewpoint or methodology, though some individual researchers may do so, and some of the sub-disciplines espouse one approach more than others. As a whole, though, stylistics’ main strength has been to remain open to new theories of language and literature, and to evolve by incorporating these new insights into its practice. This eclecticism is not only a source of strength, but is a principled position that is taken by many researchers in the field. Though the hard sciences may, in principle, be aiming at the ultimate theory of everything, in practice most scientific theories are recognised as supplying only part of the explanation of how the physical world works. These theories provide models of the data which inevitably concentrate on those aspects of a complex system that are under investigation, and to this extent they simplify the picture for the purposes of understanding and explaining one particular strand of the whole field. Likewise, in most stylistic endeavours, there is a recognition that the act of theorising and modelling textual meaning and effects is bound to be partial (in the non-judgemental sense of ‘incomplete’), or simplifying, and that this is not only practically expedient in order to make progress, but methodologically sound too, since no theory that is just as complex as the data it describes will manage to explain anything.

The result of this eclecticism in stylistics is the explosion of new theories and methods in recent years, drawing from cognitive linguistics, corpus linguistics, critical discourse analysis and functional linguistics. In addition, the practice of text analysis still depends extensively on those models of language which developed in the formal linguistics of the early twentieth century and gave us the detailed methods of description and analysis that are exemplified in Chapter 2.

Stylistics, then, takes on the new, but does not necessarily throw out the old in doing so. As Chapman (2006) says:

Theorists in all fields generally accept that they are unlikely to come up with the definitive account of their subject matter that will be proved to be correct and will be universally accepted.

(Chapman 2006: 22)

One way to think of the question of theory/theories is to realise that we do not abandon theories wholesale when they are proved wrong in some small way. We either adapt them or we add another theory to explain those aspects of the world which our original theory didn't cover. By these means, scientists and social scientists add to human understanding by accretion rather than revolution.

1.6.4 Choice, analysis and interpretation

The final set of principles upon which stylistics depends is more specific to linguistics than some of the general, scientific principles we have been investigating. The principles outlined in this section underlie most of what we would call stylistics, though some stylistic approaches emphasise one or the other more emphatically.

Some of the earliest discussions of stylistics emphasised that the choices available to writers (or, in our case, speakers too) were evidence that there was something we could call ‘style’ which was separate and separable from the normal concerns of linguistics (morphology, syntax, lexis and so on):

the idea of style implies that words on a page might have been different, or differently arranged, without a corresponding difference in substance. Another writer would have said it another way.

(Ohmann 1970: 264)

Thus, it was argued that there are many different ways of saying essentially the same thing, and that this element of choice over how to say something was the proper subject of study for stylistics.

Developments in our understanding of human language have made this apparently straightforward claim seem rather naive in recent years. It is, for example, not clear that in a very formal situation, such as a job interview, we have any clear choice in ways of expressing ourselves. So, although in some abstract sense saying ‘I’m bleeding ecstatic about this job’ is semantically equivalent to saying ‘I’m very enthusiastic about this job’, there is only a theoretical choice, since the former would have quite an adverse effect on one's chance of being appointed. Stylistics, as we shall see in the following chapters, has grown, with other linguistic sub-disciplines, into a contextualising discipline, with the awareness of social and other factors that characterise all actual uses of language. Nevertheless, at its core there remains a consciousness of the importance of linguistic choice (see Leech and Short 2007: 9–31), though one constrained by all sorts of non-linguistic factors.

The other principle, which is perhaps more generally applicable across the social sciences, and linguistics in particular, is the importance of the duality inherent in the relationship between analysis and interpretation. Whilst even the hard sciences would probably agree that one can distinguish the results of, say, an experiment into the efficacy of a particular drug from its interpretation as a contributor to human health in the future, the distinction between analysis and interpretation in stylistics is both vital and yet not as complete a separation as in some endeavours. The choice of what data to study, which tools of analysis to use and what research questions we are trying to answer is often dictated by our overall desire to explain something about interpretation, whether that be the literary effect of a poem or the manipulative effect of political rhetoric. These motivations for our research may help us to choose which analytical methods to use, but once the analysis is embarked upon, the techniques should not differ, whatever the larger motivation. For example, one might use the same analysis of modality in a passage of a novel as in a newspaper article, and the results will be presented in similar ways. However, the research is not satisfactorily completed until we draw some conclusions from these results which relate to the meaning of the text in its context. Thus, a novel showing little sign of modality may well be interpreted as demonstrating the typical style of the hard-bitten detective novel, whilst the lack of modality in a news report may be interpreted as over-confidence in the details and sources of a news story. It is the combination of rigorous analysis and contextual interpretation that makes stylistics the very rich field that it is.

1.7 The structure of the book

This book is structured in part to reflect the historical development of stylistics as a discipline. Chapter 2 focuses on the structural properties of language and how these can be manipulated to produce foregrounding effects. Here we concentrate on the psychological principles underpinning foregrounding theory, in many ways the cornerstone of stylistic analysis, and how this begins to take account of the ways in which readers engage with texts. In Chapters 3 and 4 we focus on the notions of discourse and context, and the frameworks that have been developed in order to take account of these concepts in stylistic analysis. Such approaches were developed as attempts to counter the limitations of those purely structural approaches to style that were common in the early days of stylistics. Chapters 5 and 6 concentrate on the relatively recent abundance of work in the area of text cognition. Here we outline the development of cognitive approaches to stylistic analysis, and explain how such approaches can complement those frameworks described in Chapters 2, 3 and 4. Following these chapters, in Chapter 7 we return to issues of methodology in stylistics, and outline, by exemplifying with particular stylistic studies, the variety of methodological approaches that can be adopted in stylistic analysis, and the contexts in which these might be employed. Finally, in the light of what we have outlined throughout the book, in Chapter 8 we consider the place of stylistics within the discipline of linguistics, and how stylistics might develop in the future.

Exercises

In this introductory chapter we have attempted to sketch out the main aims and principles of stylistics, along with a flavour of its historical development and some indicative analyses. We have not, at this stage, fully introduced the linguistic tools with which to go about doing stylistics, and so it would be rather unfair of us to set some analytical exercises here. Nevertheless, there are a number of questions that it would be worthwhile to consider at this stage. Keep your responses to these questions in mind as you continue through the book.

Exercise 1.1 We have emphasised in this chapter the interdisciplinary nature of stylistics. If you are approaching stylistics with a background in literature, consider the following questions:

• How do you currently go about carrying out an analysis of a literary text? Write down the stages you go through, and as you continue reading this book, see if what you currently do differs in any way from what is common within stylistics.
• Is the kind of analysis you currently do satisfactory? (i.e. Does it answer all the questions you might have about your text?)
• Is there anything that you find difficult about the study of literature? Make a note of any issues in order to see later on whether stylistics can assist with addressing these.

If, on the other hand, you are approaching stylistics from a background in language study, think about the questions below:

• What constitutes data in the area of language study that you are most used to? (For example, is it spoken language, written text, intuitive examples? Why is this?)
• How do you proceed when investigating a particular issue in the area of language study with which you are most familiar? (e.g. What comes first – research question or data?)
• Which of the questions that arise in the study of the data you most frequently use do you find difficult to answer?

If your background is in another discipline, think about what specifically has prompted you to read this book. What ‘answers’ are you looking for and how do you envisage that stylistics might help?

Figure 1.1 Stylistics and the intersection of other disciplines

Exercise 1.2 Stylistics, as we have said, is by its nature interdisciplinary. As you progress through this book, think about what particular aspects of the various disciplines that stylistics has been influenced by intersect. One way of thinking about this is to consider the overlap between stylistics and other disciplines as being represented in a Venn diagram (Venn 1880), as in figure 1.1.

Stylistics might be seen as incorporating the overlapping elements of numerous disciplines. What specifically are these overlapping elements?

Further reading

John Douthwaite's Towards a Linguistic Theory of Foregrounding (Edizioni dell’Orso, 2000) incorporates a detailed history of stylistics in its introduction, which is useful for setting the practice of stylistic analysis in its historical context. Some of the key texts from the early days of stylistics are the edited collections, Style in Language (Sebeok 1960), Essays on Style in Language (Fowler 1966) and Linguistics and Literary Style (Freeman 1970). Roman Jakobson's (1960) paper, ‘Closing statement: linguistics and poetics’, is considered a seminal work in terms of defining the remit of stylistics and can be found in Sebeok's (1960) collection. It is worth returning to these texts once you have read more of this book. The more you learn about the practical aspects of stylistics, the easier it will be to follow the arguments outlined in the texts listed above.