2 The theory and philosophy of stylistics
The theories of stylistics
A chapter on the theory of stylistics should describe ‘that department of an art or technical subject which consists in the knowledge or statement of the facts on which it depends, or of its principles or methods, as distinguished from the practice of it’ (OED definition of theory). What, then, are the facts or knowledge, or the principles or methods, which underpin and justify the practices of stylistics? These are the questions, I believe, that an enquiry into the theory of stylistics should address. Accordingly, I will not discuss style itself, as a concept or theory (see Lang Reference Lang1987; Nagy Reference Nagy2005), nor attempt a full survey of stylistic history and practice (Nørgaard et al. Reference Nørgaard, Busse and Montoro2010; Wales Reference Wales2011), nor comment on theoretical studies of specific aspects of style (e.g. those on poetic rhythm and metre from Attridge Reference Attridge and Thomières2005; Cureton Reference Cureton2001; Fabb et al. Reference Fabb, Halle and Piera2008). Nor do I address the theories of stylistics, or applications of linguistics in the study of verbal art that have developed even in closely related traditions such as those of French or German academia, let alone more distant ones such as in India or China. Rather, I discuss a much smaller field: those principles and assumptions that are most frequently invoked as the grounds for doing what is done, when academics using western traditions of linguistic analysis undertake stylistic analyses of literary texts written in English.
It is worth asserting at the outset that stylistics does have a theory – or that different forms of stylistics reflect different underlying theories, clearly articulated or otherwise. In large degree the theories stylisticians at least tacitly invoke are theories of language, which they inherit from the particular kind of linguistics (systemic-functional, corpus, cognitive, etc.) they chiefly employ. But like other specialist fields that occupy an intermediate position between larger disciplines (such as discourse analysis, and more especially critical discourse analysis; or forensic linguistics) stylistics has sometimes been described as all practice and no coherent theory. It is reassuring then to find that Peter Barry (Reference Barry2009: 203) considers this question in his popular overview of literary and cultural theory aimed at undergraduates, and decides that for all its enjoyable practical methods it is not untheoretical. Moreover he evidently regards it as theoretical enough to warrant a full chapter to itself. Likewise Terry Eagleton’s widely used Literary Theory: An Introduction (Reference Eagleton1983) has a long chapter on structuralism and semiotics, much of it devoted to stylistics (it also begins with the Russian Formalists, and literary language – which is where quite a few stylisticians began their own theorising).
What Eagleton warns about literary theory applies equally to stylistics theory. Much as he says of literary theory, we can say of stylistics theory that it is a metadiscourse, an interrogation of a cultural and academic practice; it is not just another way of ‘doing stylistics’, but an attempt to consider the assumptions of those ways of doing stylistics. And, as Eagleton warns, theory ‘tends to suspect that much of what is said [within the discipline or practice] is question-begging’ (Reference Eagleton1983: viii). By way now of situating stylistics and its theory in relation to impressionistic literary commentary, we can turn to a few of Eagleton’s comments on poems and their readers in another of his books, and see what questions these beg.
Putting impressionism in its place (before or after stylistic analysis)
At the outset of his primer for the literarily perplexed, How to Read a Poem (Reference Eagleton2007), Eagleton regrets the lack of close reading by students today:
What gets left out is the literariness of the work. Most students can say things like ‘the moon imagery recurs in the third verse, adding to the sense of solitude’, but not many of them can say things like ‘the poem’s strident tone is at odds with its shambling syntax’.
Evidently in Eagleton’s view it would be a good thing if students could talk about the stridency or otherwise of a poem’s tone, and about shambling syntax (he doesn’t here say why, and stylisticians are among those least likely to challenge him on this). But what are stridency of tone and shambling syntax? ‘Shambling’ and ‘strident’ are, for apprentices, too contentual, impressionistic and unverifiable as ways of talking about syntax and tone. So Eagleton needs to explain to his student readers what he means by shambling syntax, stridency and so on, in clear and coherent ways. He is a brilliant writer and critic, so he stands a chance of doing so. A rapid search of How to Read a Poem, however, suggests that Eagleton never returns to the terms ‘shambling’ and ‘strident’ after page 3, whereas he himself continually alludes to the ‘imagery’ in the texts he analyses, which contrasts sharply with the practice of John Lennard (Reference Lennard2006: xv) in another influential primer, who finds the term image so vague or varied in sense as to be best avoided altogether. In fairness it should be noted that Eagleton (Reference Eagleton2007) has a brilliant closing section on imagery (Reference Eagleton2007: 138–42), despite concluding that the theory of imagery ‘is in something of a mess’ (Reference Eagleton2007:141).
Eagleton’s discussion of syntax in poems (Reference Eagleton2007: 121–4) is more expressive than informative. The opening point, that ‘A good many poetic effects are achieved through syntax’, might suggest a lengthier treatment than the two-page commentary on three extracts found there. But it is clear from that commentary that Eagleton simply isn’t equipped to offer a lengthier and more explanatory treatment. Of the grammar of the opening lines of Edward Thomas’s ‘Old Man’ (the title is the name of a herb) he writes:
The jagged, knotted syntax struggles to unpack the poet’s constantly swerving thoughts about the plant he is contemplating. As it does so, its hesitations, stops and starts and doublings-back act out something of the convolutions and self-qualifications of his response to the herb.
Enjoyable as this is to read, even its author seems to realise that it is not explanatory. Soon he is complaining that a Yeats poem ‘almost deliberately provokes us into belletristic waffle’, and evidently could not resist: ‘The last line, with its artful change of key, is a kind of final flourish to this masterly performance, with its look-no-hands bravura’ (Eagleton Reference Eagleton2007: 123).
It is stylistics that undertakes to be precise, analytical and verifiable about the grammar that underlies and creates the literariness effects which in turn induce readers to reach for such complex evaluative terms as shambling, strident, alienated, terse, passionate, placid and so on. Stylisticians should not be dismissive of these powerfully synthesising and summarising evaluative terms – their existence makes stylistics possible, and wide agreement that a particular passage is strident or shambling or restless helps allay anxieties about conflictingly various textual intepretation. They have an important place in literary reading, but one that stylistics endeavours to keep distinct from that of analysis.
Nevertheless the piling up of value-laden judgements, with limited further explanation, can oppress the reader. As early as page 4 of his poetry primer, Eagleton has reached interim conclusions about the opening sentence of a famous Auden poem:
The tone of the piece is urbane but not hard-boiled. It is civilised, but not camp or overbred, as some of Auden’s later poetry.
The problem with this is that readers who fail to register stridency of tone are unlikely to be able to say (let alone understand) that another poem’s tone is urbane, but neither hard-boiled nor camp nor overbred. So one must learn all these discriminating evaluations – and it seems that the chief way Eagleton will teach these labels is by using them in his own commentaries. In the best outcome, if you read enough Eagleton commentaries on poems and if these are consistent in their usage of these labels and if you encounter no contrasting uses of them by other critics, you may in time grasp what he means by a hard-boiled tone and a camp one. Even this uncertain understanding will not proceed smoothly, or without loss to self, if as is likely you are already familiar with these labels in various other contexts, and wonder if they are more charged than explanatory in the present text-analytical context.
In brief, Eagleton’s guide is actually an autobiographical report – How I read poems – with innumerable subtle insights on poems’ effects but few explanations. Stylistics believes such explanations are needed (‘show your working’, as the maths exams say), and are possible, and sets out to build them. Not without difficulty. But we stylisticians are sobered by the thought that while we may hope to be more explanatory than Eagleton, we will never produce an introduction half as entertaining as his.
Knowledge/facts
Here are some of the things that most stylisticians take as foundational knowledge or facts:
i. That as David Lodge famously said about novelists at the opening of Language of Fiction (Reference Lodge1966: ix) but could be said in large part about all writers and speakers, literary or not, everything they do is done in and through language. Language is central to the meaning, or effects on readers (it is not clear that the two can be rigorously distinguished), of texts. In a world as full of texts spoken and written as ours, a rich reflexive study of all aspects of speech and writing is guaranteed. Hence our departments and institutions devoted to speech communication, speech and language therapy, literary criticism, linguistics, languages, media studies, creative writing, journalism, literary studies, and so on. And these are just the institutions concerned centrally with speaking and writing almost as ends in themselves, without sharp constraints on content, as distinct from disciplines like law and history in which writing plays a large role.
ii. That among all the linguistic communicative activities practised by a culture, a special place has arisen, especially for those whose living conditions allow them several hours per week of ‘free’ or leisure time, for reading and talking about literature. At one time in western culture much of that reading would have been scriptural or religious (and not quite conceived as ‘leisure’ activity), but that time is largely past. In the twenty-first century, and notwithstanding local wars and recessions, more members of western countries have more opportunities to spend their leisure time in more ways than ever before; and still a significant proportion read novels, go to plays, sample the odd poem, join reading groups, and flock to film adaptations of memorable stories.
iii. That literary texts can be as short and sweet as a two-line haiku, can be devoid of overt metaphor or other figures, lacking in patterning of any kind, childlike in vocabulary and syntax, but rarely are. Much more often and typically (but again, not definitionally, as necessary conditions) literary texts are complex and sophisticated creations in such terms as those just mentioned; and this is no surprise since (along with statutes and diplomatic agreements and love letters) they are among the most crafted and re-drafted texts in the culture, and usually composed by experts, desperate to get it right. If the literary author is an expert writer like the legal or diplomatic draughtsperson, they also often have the anguish and personal investment of the inexpert author of a love letter.
iv. That all linguistic communications within a linguistic community draw on that community’s knowledge of linguistic forms, structures and effects as encountered in a host of situations. Since literary texts are such exceptionally considered and designed uses (or exploitations, or floutings) of those forms, structures and effects, we are justified in attending to them very closely in accounting for what those texts mean, to readers.
v. That stylisticians’ consideration of the craft and design of texts and their shaping of reader response assumes the author has made multiple critical choices in the composing: they have chosen these forms and contents in preference to others (and, sometimes, relatively easily specified alternatives), and the particular choices are effective in ways that the alternatives would not have been.
vi. That the meanings readers derive from a text are not sourced in the text alone; the text is not an icon or island, despite its linguistic boundedness. Commentary is radically incomplete, without a consideration of the role of intertextuality and the reader’s recognition of echoed or alluded-to texts and events, in the reading of a poem. Similarly, a reader’s familiarity with different genres and their conventions, and with literary history more generally (or their unfamiliarity with all these), will clearly affect their response and interpretation. In Simpson’s terms (Reference Simpson, Bex, Burke and Stockwell2000: 3) stylistics ‘acknowledges that utterances (literary or otherwise) are produced in a time, a place, and in a cultural and cognitive context’. These wider contexts interact with the text, so that a grammatical study of the latter always needs to be integrated with consideration of the former. Still, a grammar of the unvarying language of the text is a good deal more feasible than a grammar of its infinitely variable contexts of reception, which is why stylisticians concentrate on the former and approach the latter mostly by way of the former.
vii. That stylisticians mostly select literary texts because they are highly valued and highly crafted, deemed worthy of being the focus of university degree studies, for example. At the heart of these texts’ power, we believe, is the exceptionally sophisticated deployment of linguistic resources within them, linguistic resources which scholars only incompletely understand. Additionally, even enthusiastic readers, such as committed members of reading groups or students taking literature degrees, may struggle to find a way of talking about these texts that is comprehensible to others and advances their own insight into the writing. For these reasons – to advance our understanding of literary linguistic phenomena primarily, and to share this understanding with readers of literature (a very large constituency, potentially) – stylisticians select texts or extracts as examples, test cases and forms of linguistic challenge. They do so, focusing on the text’s craft and often original design, very much as musicologists and art critics develop systematic analyses of symphonies and paintings. And, like the latter, stylisticians’ labours cannot come to an end: not only because the next decade’s readership of Shakespeare’s Sonnet 118 or audience for Bach’s St John Passion may respond to these differently from earlier ones, but also because in a decade or two the theory and practices of linguistics and musicology that might be brought to bear may be different from those extant at present.
viii. That comprehensive stylistic analysis of any text is impossible; selectivity and sampling are always involved and these in turn cannot be finally rooted in high principle or an abstract theoretical position, but in the low pragmatics of finitudes of time, energy and interest. Analogously, there is no rule or theory to tell us how long to stand looking at Rembrandt’s The Night Watch in the Rijksmuseum or how many times is ‘sufficient’ for listening to Beethoven’s Grosse Fuge quartet. So stylisticians’ claims about the virtues of their analyses’ greater precision and detail in the description of textual effects have to be made with care: the objection that an analysis has too much detail is not always easy to rebut, although it may be rebuttable in light of the particular purposes it serves. A close-up study of a picture’s surface may make it impossible to take in the picture as a whole; but if that study has a different purpose from whole-picture apprehension, such as understanding the deployment of pixels or brushstrokes, then it is justified. Generally, stylisticians avoid assuming a sharp separation between the ‘lay’ readers’ reading and their own technical analyses, since the technical analyses derive their strongest rationale from being explanatory accounts of the effects that a text’s texture works upon the ordinary reader.
ix. That stylistics has proven invaluable to students of literature, thanks to its clear, accurate, detailed, systematic and illuminating scrutiny of the language of literary texts. Stylistics has been performing that service for many decades; a few of the earliest and still useful studies include Leech (Reference Leech1969), Nowottny (Reference Nowottny1962) and Cluysenaar (Reference Cluysenaar1976); numerous publications since then have continued this work. As Lambrou and Stockwell (Reference Lambrou and Stockwell2007: 4) note, over and above the detailed grammatical descriptions based on explicit criteria afforded by stylistics, doing stylistics (or thinking about texts stylistically) can produce ‘startling, pleasurable and perspective-changing moments in reading’. It can also prompt startling adjustments to the way you conceptualise reading more generally, and writing, and language.
x. Most stylistic analyses negotiate covertly if not overtly certain core assumptions about literary text. These include the idea that meanings will be ‘more integrated’ in literary texts than elsewhere (or at least that to read a text as literature is to expect a full and rich integration); that reader attention will be guided by textual patterns – noticeably repeated forms or the ‘gap’ of a noticeably absent form – to segments which are especially important in the creating of effects (i.e. foregrounded); and that there may be a degree of ‘fit’ between the foregrounded forms and the perceived meanings or effects so that the form may be called iconic. Iconicity in text, using language that is otherwise typically symbolic, is itself remarkable, if not an attempted impossibility. Language is symbolic in that the addressee’s attention will usually turn from the signs used to the ideas and entities they represent; but when it is partly iconic, the signs seem partly to embody or perform the ideas and entities expressed, and the addressee is less impelled to turn their attention away from or beyond the form. Iconic text invites the reader/addressee to focus on the text for its own sake, to gloss – approximately – one of Jakobson’s ideas in his famous article of Reference Jakobson and Sebeok1960 (see Toolan Reference Toolan2010 for a commentary on Jakobson and equivalence, and Toolan Reference Toolan and Jones2011 for some proposals regarding repetition in literature).
xi. That, treated segregationally, feature by feature, a literary text is not radically different from any other linguistic act, spoken or written: it may or may not have more metaphors, more – or more complex – modality, more deixis, more transitivity, more blends, more text-worlds, more iconicity, and so on, than any given non-literary text. A short poem may have fewer of all the aforementioned items than, say, a magazine advertisement promoting a brand of perfume but the former may still, taken as a whole, remain much more intellectually interesting. That is because, treated as the textual core of an integrated communicational event, the literary text is different, and aims to be. As our culture suggests to us, the successful poem requires a level of creativity and inspiration far beyond that of the successful advertisement; because verbally gifted as copy-writers are, those gifts are slight in comparison with the poet’s; because while both texts will have been worked on, the kinds of revision and critique that attend a poem – long after its publication as well as before its completion – tend to far outreach those of the advertisement. Most importantly of all, for many in the culture, the literary text is ‘religious’ in ways that the advertisement or other verbal form is not. Here I use the term ‘religious’ to denote a host of concepts, with some family resemblance, frequently invoked when the powers and purposes of literature in modern society are alluded to: an exceptional insight into specific (invented) individuals, and by extension into the lives of real individuals, including the reader; explanations of the coherence or incoherence of human existence; verbal pictures that ‘strike through the mask’ of banality to probe hidden motivations; ‘deep’ analysis; indications of the meanings of things, including human life or glimpses of a transcendent reality or consolation.
Stylistics as grammar
Against the above background, stylisticians operate in relation to a number of further principles and difficulties. Firstly, I will argue that stylistic explanation of the relations between textual form and literary function is a specialist grammatical commentary. Accordingly, a theory of literary stylistics is at core a theory of grammar. In both these claims, I intend the term ‘grammar’ to be capaciously understood, so as to include synoptic description of the prominent licensed patterns of phrasal and clausal structuring (sequencing) of words in the language generally (see Sampson Reference Sampson2007 on grammar as the network of beaten tracks through the landscape that is a language), but also to reach much further than this to include genre- and register-sensitive norms and patternings, and departures therefrom. If the bases of the more interpretive categories that a particular stylistic study invokes (text-worlds, deictic shifts, cognitive blends, empathetic alignment, etc.) are not derived from a grammar of text (widely conceived), then the study is not a stylistic one.
By way of exploring the idea that stylistic studies are contributions towards a grammar of literary texts, I want to examine in some detail a stylistic analysis published in 1970, thus more than 40 years ago, by the distinguished linguist James P. Thorne. The essential steps in Thorne’s argument are, I contend, still typical of most stylistic studies today, even though descriptive frameworks and technological affordances are greatly changed, because it attempts to trace interpretive impressions back to sources in the grammar that ‘generated’ the text. Thorne writes: ‘What the impressionistic terms of stylistics are impressions of are types of grammatical structures’ (Reference Thorne and Lyons1970: 188). Thorne was a generative linguist, and he sees an affinity between stylistics and generative grammar since both are essentially mentalistic: ‘In both cases the most important data are responses relating to what is intuitively known about language structures’ (Reference Thorne and Lyons1970: 44). Earlier grammars’ failure to animate stylistics are said to stem from linguists’ attention being restricted almost entirely to directly observable structural facts in or about language, and their neglect of mentalistic responses, and ‘what is intuitively known [by poets and readers] about language structure’ (Reference Thorne and Lyons1970: 188) and deep structure aspects rather than only ‘those structural facts which can be directly related to what is observable in language’ (Reference Thorne and Lyons1970: 189). Thorne’s championing of mentalistic generative grammar has interesting echoes in more recent emphases on the psychology of the reader or on the cognitive underpinnings of linguistic choices, discussed later.
Such a distinction creates a clear mission for the stylistician: that of uncovering and bringing to attention various deep structure facts or characteristics that give the surface text the style it has and which we readers respond to. In this respect, the mission is a sophisticated variant of the more general idea that stylistics is a consciousness-raising exercise, bringing to the conscious attention of the reader those underlying or overlooked linguistic phenomena of the text that are instrumental in its subtle achievement of meanings and effects (effects that the less tutored reader may sense they are deriving, but whose source they cannot quite pinpoint). Much the same position is to be found in an article by Ronald Carter (Reference Carter and Carter1982) on Hemingway’s story ‘Cat in the Rain’, an exemplary demonstration of the stylistic approach and frequently used in teaching for that reason. Much as Thorne did, Carter argues that his (our) interpretive intuitions about the opening paragraphs of a literary text such as the Hemingway story ‘are to a large extent conditioned by linguistic patterning’ – patternings that the linguist or stylistician is well placed to identify and describe.
Foregrounding and iconic aptness
At this point a crucial question becomes: which patterns? How do we know which patterns are the major ones involved in prompting our interpretive intuitions, rather than being only minor ones? Carter’s answer is ‘those that are most striking’, meaning here ‘striking to the analyst’. This was the essence of Thorne’s answer too (see below): he was ‘struck’ by the frequency of I VP-ed and I VP-ed structures, and relied on absence of reader demurral. (Strangeness, or defamiliarisation, has long been a touchstone of formalist criticism.) The conventional metaphor of ‘strikingness’ is very prominent in this phase of the stylistic analysis; it is a variant of more abstract formulations in terms of ‘foregroundedness’ or ‘prominence’ or ‘markedness’, but they amount to the same move in the argument, which is subjective at source, even if the analyst can persuade their readers of the reasonableness of their identification of what is striking or foregrounded. Slightly more detached or abstract are formulations in terms of what is statistically or situationally prominent (a speech in which every other word is like), or disproportionate in frequency (a paragraph in which every verb is in the progressive), or situationally non-congruent (an insurance policy which uses slang). Thirty years later, Jeffries and McIntyre (Reference Jeffries and McIntyre2010) go through quite similar steps as Carter or Thorne before them with their first brief demonstration of stylistic method: they select the word forget, from lines in a James Fenton poem that run How comforting it is, once or twice a year, / To get together and forget the old times as ‘unusual’ (Reference Carter, McIntyre and Busse2010: 5) and foregrounded. They appeal to intuitions and collocational associations to justify this singling out, and then hazard an explanation: Fenton’s marked choice makes us focus our attention on the absurdity of actively trying to forget something, they suggest.
If the subjectivity of foregrounding selections has been a weakness in stylistic studies, it is one that cognitive poetics (Stockwell Reference Stockwell2002) may help to address. The kinds of phenomena that cognitive poetics centrally draws on in its explanations include ideas about figure and ground, prototypicality, deixis or perspective, script and schema, metaphor, text-worlds and mental spaces. A great many of these are accounts of human forms of attention, and of selective and discriminating attention, in our making sense of the world and its signs. They address the fact that whatever object or scene we contemplate, there are always a great many more differences we could attend to than, in all the circumstances, it is useful or appropriate for us to attend to. Any signifying complex, including a literary text, may make prominent (or foreground) some features rather than others, to direct the recipient’s attention. So a calculation as to what kinds of attention to which distinguishable phenomena is merited (or relevant) is always going on, and cognitive poetics undertakes to explain some of the fundamental principles, many of them universals (underpinning non-literature cultures as much as literate ones) rather than language- or culture-specific, underpinning the sense-making faculties of human beings in their embodied existence. These principles may in time put stylisticians’ decisions as to the striking patterns and foregrounding in a text on a much firmer footing.
A Spitzerian circle or spiral is sometimes said to operate: a text is read, and from that initial reading, first (verbal) impressions and responses (usually from the stylistician him- or herself, and/or from a sample of other readers) emerge; the stylistician returns to the text to identify the prominent (foregrounded) patterns or deviations, and seeks to describe and explain them in linguistic (perhaps cognitive-linguistic) detail. The analyst attempts both to identify what it is in the text’s patterns (its form) that induces the impressions first noted, and to integrate these effects into an evaluation and interpretation that others might accept. These form-interpretation relations will rarely be statable as absolute free-standing rules, but a generalisable descriptive grammar of literary texts remains the long-term goal of stylistics. Attempts may be made to confirm the plausibility of the form-interpretation conjunctive statements via controlled testing of reader-subjects; but even when it is not, the stylistician’s very act of presenting or publishing their analysis puts it in a public domain where every reader or listener can assay its claims. In passing, it is noteworthy that Thorne attributes ‘impressionistic terms’ (complex, terse, etc.) to stylistics, the implication being that linguists or grammarians can usefully unpack these impressionistic categories. Only a very few years later it became customary for stylisticians to attribute the vague, subjective and impressionistic terms to literary critics and readers, with the stylistician brought in, as expert, to explain what lay behind such terms in relation to a given text. An interesting shift in job demarcations and the characterisations of experts and amateurs evidently occurred – but the underlying steps in the analytical proceeding are unchanged.
There is one important way in which Thorne’s ambitions differ from those of most stylistics today, and from my suggestion that stylistic analyses are contributions towards a grammar of literary texts. On the basis of what appear to be specific deviant rules of composition in single poems by Donne and Roethke and others, Thorne speculates that every poem has its own distinct dialect, so ‘the task that faces the reader [of a poem] is in some ways like that of learning a new language (or dialect)’ (Thorne Reference Thorne and Lyons1970: 194). Stylisticians today tend to assume that a language is much less sharply determinate and bounded, and would lean far less heavily on the idea that poems are dialectally distinct from the circumambient language.
Thorne on Chandler
How in practice does Thorne proceed? His most fully worked example in the Reference Thorne and Lyons1970 article is a grammatical analysis of a virtuoso passage in Raymond Chandler’s crime novel The Lady in the Lake (Reference Chandler1943). The first-person narrator is Philip Marlowe, a private detective:
An elegant handwriting, like the elegant hand that wrote it. I pushed it to one side and had another drink. I began to feel a little less savage. I pushed things around on the desk. My hands felt thick and hot and awkward. I ran a finger across the corner of the desk and looked at the streak made by the wiping off of the dust. I looked at the dust on my finger and wiped that off. I looked at my watch. I looked at the wall. I looked at nothing.
I put the liquor bottle away and went over to the washbowl to rinse the glass out. When I had done that I washed my hands and bathed my face in cold water and looked at it. The flush was gone from the left cheek, but it looked a little swollen. Not very much, but enough to make me tighten up again. I brushed my hair and looked at the gray in it. There was getting to be plenty of gray in it. The face under the hair had a sick look. I didn’t like the face at all.
I went back to the desk and read Miss Fromsett’s note again. I smoothed it out on the glass and sniffed it and smoothed it out some more and folded it and put it in my coat pocket.
I sat very still and listened to the evening grow quiet outside the open windows. And very slowly I grew quiet with it.
Thorne begins by suggesting that for much of the passage the underlying grammatical structure is highly repetitive, with in rough terms a I VP-ed and I VP-ed pattern (two conjoined clauses each with I as their subject and the following verb phrase in the past tense: I pushed it to one side and (I) had another drink). The pattern is only slightly masked in surface structure by the deletion of the repeated first-person pronoun in the second conjunct. Thorne claims the ‘highly repetitive style’ is instrumental in creating ‘the mood of aimless, nervous agitation the passage conveys’ (Reference Thorne and Lyons1970: 191); in contrast, the final sentence stands apart, he says, announcing a change of mood in part by being grammatically different. The grammatical exceptionality he highlights (and implies contributes to the mood change) is the fact that only the final sentence has an initial ‘And’, followed shortly by an overt first-person subject pronoun. Other final-sentence differences we might note include:
1. Only the final sentence thematises a manner adverbial, very slowly, prior to mention of the grammatical subject.
2. Very slowly clearly pairs, positionally and somewhat semantically with very still; but there are differences/modulations too: very still is adjectival, and stative, very slowly is adverbial and dynamic or developmental.
3. The final sentence ‘copies’ its verb from the previous sentence, but with morphological change from infinitive to past tense (grow > grew).
4. There is a clear echoic pattern where the evening grow quiet pairs with I grew quiet; plus the intimation, partly projected by the sentence-initial cohesive conjunction And, that the evening’s growing quiet causes the I-figure to grow quiet too. The with it construction hints at this too: compare I grew arrogant with success > ‘my success caused me to become arrogant’.
Such a commentary is typical of stylistic analysis in itself, and in being repeatedly open to challenge. To begin with the main claim, that the repetitive I VP-ed and I VP-ed structure is instrumental to ‘the mood of aimless, nervous agitation the passage conveys’, it is hard to see how that specific grammatical structure is causative of readers inferring that mood, as distinct from saying that on this occasion, in this matrix of conditions, it does not clash with the ‘aimless agitation’ message. One can easily invent I VP-ed and I VP-ed sentences that carry no mood of aimless agitation (I started the engine and steered the plane onto the runway; I engaged the enemy and delivered three short bursts of deadly fire). Even the preliminary observation that I VP-ed and I VP-ed sentences are dominant in the passage has to be qualified: in the first paragraph, for instance, three sentences are of that type, but seven are not; in the second paragraph, again three fit that structure while five do not. These facts have consequences for the five subsequent claims about the contrasting fourth paragraph: with respect to the first of these, for example, while it is clear that only in the final sentence is there an initial And, it is not true to say that only this sentence lacks a medial and (sentences 1 and 3 in paragraph 1 are the first of many without medial and).
The very possibility of corrective commentary of the kind just offered is what entitles stylisticians to argue that parts of the stylistic analytical procedure are inspectable, replicable, testable and falsifiable. The replicable–falsifiable opportunities apply chiefly at the level of grammatical description. When it comes to the more interpretive stage (‘this frequently repeated I VP-ed and I VP-ed structure helps create the mood or effect of aimless agitation’), falsifiability is scarcely possible, because the ‘aimless agitation’ predicated of the passage has not itself been established on independent and inspectable grounds. Perhaps one can appeal to lexical evidence and presuppositions in the passage itself: the early mention of ‘beginning to feel less savage’ and the late mention of ‘growing quiet’, suggest that at first the narrator was in an angry, dangerous mood, and that the passage shows (rather than baldly tells) how he gradually calms down.
Returning to the issue touched on earlier, regarding causation vs correlation: stylisticians are not really content with claiming that identified pattern A ‘helps create’ claimed effect or meaning Y. They are equally uncomfortable with the suggestion that they are not doing anything fundamentally different from literary analysts after all (see Attridge Reference Attridge, Fabb, Attridge, Durant and MacCabe1996: 44–5, who characterises Jakobson’s ‘empirical’ studies as ultimately aiming, like the critic, to persuade the reader). The idea that they are participants in a discourse, persuading with evidence rather than pinpointing definitive explanations, troubles them. Working on some of the most chameleon and elusive materials, still they crave scientific truth, and closure. They would much rather attain what lawyers sometimes call the ‘but for’ causation standard: but for the presence of this pattern A, meaning/effect Y would not have been created (or, at least, would not have been satisfactorily or fully created). In practice, however, the ‘but for’ causation standard is very difficult to meet, in textual commentary – unless there are underlying connections between the explanandum effect and explanans pattern from the outset.
Another foundational point to consider, in relation to the most clearly testable and falsifiable part of stylistic analysis, where grammatical description is undertaken, is that here too matters may be contested, and agreement about categories and identifications may not always be as shared and ‘common ground’ as is assumed, despite the emphases on what is empirical, quantifiable, objectively present. In the Chandler passage, for example, Thorne and I appear to have accepted that the graphological sentence is a stable and agreed-upon unit of analysis (since Thorne is making a point about the striking or foregrounded high frequency of graphological sentences containing a I VP-ed and I VP-ed structure). Another analyst might take a different line. They might point to the fact that some of the graphological sentences are not well formed by the standards of formal writing (e.g. Not very much, but enough to make me tighten up again), and might suggest that Thorne and I are using far too superficial a notion of sentence-hood. They might then propose and apply a more abstract definition of the sentence category, and this might lead them to treat the final paragraph as underlyingly just one sentence, not two:
I sat very still and listened to the evening grow quiet outside the open windows, and very slowly I grew quiet with it.
In this alternative stylistic analysis, several of Thorne’s and my claims about the contrastiveness of the final paragraph (e.g. about sentence-initial And, and the creative departure from the I VP-ed and I VP-ed structure) would be nullified. Thus although stylistics makes appeal to the agreed and stable public descriptions of linguistics, much of that description is not uncontroversial. Many linguistic categories and their membership are matters of potential disagreement, and not even basic categories (‘sentence’, ‘word’, to say nothing of ‘modal verb’, ‘behavioural process’, etc.) can be invoked without considering what conception of them the analyst has in mind. In the same way and for the same reasons some circumspection is needed when stylisticians claim that their studies are expressed in a common language, since the latter is more a rhetorical trope than a demonstrated entity (every analytical tradition claims to have its own shared but specialist language variety).
Stylistic practice and the return of the reader
None of the above means that the categories of sentence or word or modal verb or passive voice are arbitrary, or meaningless beyond the descriptive system in which they arise; it only means that these categories come with a rich history and, although used in accounts of the texture of texts, are themselves metonyms of complex accounts of the workings of language. That is why some of the grander flourishes of refutation contained in Stanley Fish’s (Reference Fish and Chatman1973) famous critique of stylistics misrepresent what stylisticians do – at least, they misrepresent the vast majority of stylisticians, who eschewed a purely structuralist or generativist and semantics-free grammar, a point to which I will shortly return.
In an assessment of stylistics papers of the 1960s, Fish complained that repeatedly the analysts made a leap from description of forms to attribution of value (i.e. meaning) that was arbitrary and (contrary to more recent stylistic conventional wisdom) unfalsifiable. Some analysts, he concedes, are more subtle than others, but all at base are saying ‘here’s a formal feature’ (e.g. reduced relative clauses) ‘and here’s what it means’ – by implication, what it invariably means. In the scenario Fish uncovers, the form must invariably carry the specific value or meaning because an alternative position threatens to end in vacuity: if the analyst says the form has meaning A in the present text, but may create meaning/effect B in the next text in its different context, meaning/effect C in a further text in yet another context, and so on, then no generalisable or constrained explanation at all has been achieved. The 1960s stylistician’s direct and invariant form-meaning (or form-value) pairing, Fish suggests, is unwarranted, arbitrary, and ‘a game that’s too easy to play’ (Fish Reference Fish and Chatman1973: 100), since the meaning or value arrived at is typically a version of the intuition or pre-analytic impression the stylistician began with. He declares:
While the distinctions one can make with the grammar are minute and infinite, they are also meaningless, for they refer to nothing except the categories of the system that produced them, categories which are themselves unrelated to anything outside their circle except by an arbitrary act of assertion.
There is also a small irony in Fish’s use of the indefinite reference one in the first quoted line: he no doubt had in mind the linguist or grammarian. But if we were to treat the one here as referring to the poet (or even the reader), then the stylistician’s point is that the small distinctions a poet can make with a grammar are emphatically not meaningless. In any event, in sentences like that above it is clear that Fish’s target is the kind of structuralist–distributionalist and generative grammars that flourished in the 1950s and 1960s which, consonant with his criticism, often claimed as a virtue that they made no appeal to meaning-bearing categories outside the system they postulated. But stylisticians since the late 1960s have almost never adopted and applied such a grammar; rather, they have tended only to use those grammars and linguistic descriptions that include commentary on the meaning-bearing or semiotic functions of structures at their core. And where one description or approach seems inadequate to the complex task – as is typically the case with works as complex as literary ones – stylisticians have been enthusiastically eclectic, as many commentators have noted. Jeffries and McIntyre (Reference Jeffries and McIntyre2010: 4), for example, report that stylisticians draw on both context-free formal descriptions and the contextualised linguistic descriptions found in versions of pragmatics and sociolinguistics while also recently leaning heavily on either cognitive or corpus linguistics and occasionally both together.
Some of Fish’s chastisement nevertheless was justified, and in fact stylisticians from the 1970s onwards moved quite rapidly to make clear how much more guarded their claims would be, as to the tendency, all other things being equal and not without scope for exceptions, of particular forms to correlate with or contribute towards the achievement of particular meaning effects. They also acknowledged and addressed the fact that context can powerfully alter the effects of linguistic forms. Fish (Reference Fish1980) called for an ‘affective stylistics’, in which fuller attention was paid to the sequential processing of the text that the reader necessarily enacts, and a number of stylisticians and literary analysts in the 1980s attempted forms of ‘reader response criticism’; but this has been developed far more fully in recent years, with the growth of cognitive poetic and empirical studies of reading.
A fairly standard characterisation of stylistics is the following:
Stylistics is the principled, systematic and rigorous analysis of texts (mainly literary), using linguistic description.
It is arguable that one premise of such a stylistics is that something like a mirror-image of its self-characterisation applies to its target or only begetter, the work of verbal or literary art:
Verbal art is the principled, systematic and rigorous creation (synthesis?) of texts (mainly literary), using tacit linguistic knowledge.
In seeking to analyse a poem into its main meaning-bearing verbal parts, stylistics is a re-tracing of some of the compositional steps taken by the author. Stylisticians spend so much time and effort on testing and revising their hypotheses about the key linguistic features or compositional steps because they believe that great ingenuity and care and discrimination has gone into the original work of literary making. Stylisticians are among the greatest respecters of literary authors, but they by no means wish to treat what those authors do as a sublime or sacred mystery, beyond analysis. On the contrary, in being the application and development of grammar (widely conceived) to literary or other texts, stylistics is explicitly intent on demystification of literature: it believes that, for all their literariness, literary texts are amenable to analysis and explanation just like any other texts. But – a balance of considerations, again – most stylisticians at least also recognise that certain literary texts may have an exceptional power for their readers. This freshness of literature, sourced in the freshness of its linguistic texture is well captured in this one-sentence summary by Simpson: ‘To do stylistics is to explore language, and, more specifically, to explore creativity in language use’ (Simpson Reference Simpson, Bex, Burke and Stockwell2000: 3).
Falsifiability and standards of proof
The crucial scientific characteristic of stylistics is often said to be falsifiability: genuinely stylistic claims are claims relating to the language of a text (or class of texts) that might with effort be proven to be wrong or need refinement. But falsifiability involves a prior clear and agreed articulation of the precise criteria underpinning any claim. If, for example, analysts agree as to what an English modal verb is by reference to various criteria (invariant in form or unmarked for tense; leftmost in the verb phrase; requiring, prior to ellipsis, an infinitive form of a lexical verb or have or be to their right; attracting negation; etc.) then a claim that the first sentence of this paragraph contains a modal verb (might) while the second sentence does not is a falsifiable claim. But in practice, since stylistics typically involves a hermeneutic circle or spiral of grammatical description and textual interpretation, it crucially brings in commentary that is not falsifiable. Many stylisticians have discussed kinds of free indirect speech and thought in narratives, for example, and have drawn up a useful list of typical indicative characteristics which, if present, usually suggest that sentences have modulated from narrator-voiced narrative to the covert voicing of a character’s internal reflections that is free indirect thought (FIT). But those characteristics are not necessary conditions, and FIT may be strongly suspected even in their absence. Thus, towards the end of Joyce’s ‘The Dead’, when the narrative text runs as follows –
He stretched himself cautiously along under the sheets and lay down beside his wife. One by one, they were all becoming shades. Better pass boldly into that other world, in the full glory of some passion, than fade and wither dismally with age.
– most stylisticians will want to say that sentences 2 and 3 report Gabriel’s free indirect thoughts, while sentence 1, purely narrative, emphatically does not. But no easily stated criteria underpin that analytical judgement (rather, an awareness of other such narrative-to-FIT modulations earlier in this story and other Dubliners stories, and an awareness of Gabriel’s thwarted and humbled mood guide the analyst, including an awareness that the full glory of some passion is very much the kind of phrase that Gabriel is capable of). So the FIT classification here is not strictly falsifiable, and alternative interpretations which attribute the pronouncements in sentences 2 and 3 to Joyce or a narrator cannot be proven wrong.
The essence of falsifiability is not that something is testable, but that it can be proven wrong and that efforts are made to prove it wrong. Relatively few stylistic studies centre on such efforts – perhaps because of the interest in pedagogy, where it is natural to emphasise positive proposals. In practice, then, stylisticians do typically write as advocates for the linguistic analysis they present, arguing for its relevance to the target text, and to readers’ readings of that text. Even where they focus directly on falsifiability, and have made clear what their analytical categories are (the difficulty noted earlier), they grapple with texts affected by a greater variety of confounding variables than those in any other linguistic sub-discipline. Or it may be that difficulties with disproving the significance of any isolated textual feature of the context-embedded literary reading experience highlight, more vividly than other kinds of linguistic analysis, that the difficulties are actually general in linguistics.
Since scientific falsifiability may be unavailable, it is often useful to set stylistic claims and arguments by the less absolute legal standards of proof or certainty: not ‘beyond a reasonable doubt’ but ‘on the balance of probabilities’. The more ambitious kinds of stylistic analysis often involve this lower standard: they aim to persuade others that, on balance, it is more likely than not that these linguistic phenomena are the necessary cause of that effect. But that is a low level of certainty, and this is where the ‘systematicity and rigour’ come in: it is not only possible but necessary that the qualified confidence of such stylistic claims be put to further tests, with the aim of raising the standard of proof from ‘more likely than not’ to ‘quite sure’.
Disciplinary maturity
Stylistics has never been objective, definitive and rigorous along the lines of experimental sciences. It couldn’t be – not simply because literary texts are variously contextualised and variously interpreted, but because language is endlessly potentially variable in forms and functions, languages are not finally codes, and human beings are not machines for processing those not-codes (Harris Reference Harris1981, Reference Harris1987; Toolan Reference Toolan1996). But it takes a long time to turn the linguistics tanker and its accompanying ferryboat, stylistics; not necessarily a 180-degree turn, but enough for linguists to recognise that progress in the brain sciences and in linguistics has been mostly incremental rather than revolutionary, that alongside the normative pressures on public language which ensure a degree of convergence and code-like iteration there is always variation and change, that change is not the same thing as improvement or progress, and that while some aspects of human life are susceptible to improvement it is doubtful that others are. There is a better understanding now of human physiology and illnesses or of meteorology than in 1600, but is there better understanding of poetry, or language? The physical and medical sciences progress, but do the humanities? If they did, we should be able to say that today’s poets and playwrights write better than Shakespeare and his contemporaries, but such a claim seems wrong or absurd; and in the area of language study we should find people learning languages much faster than four centuries ago, but they don’t seem to. Lack of demonstrable ‘definitive’ progress in an area of activity (playwriting, research into language teaching, stylistics) is not, however, an argument for abandoning the activity but for conceptualising it in different terms, where ‘objective progress’ is not mistakenly assumed to be the chief concern. This is why I have always argued that while stylistics aims at being explicit, methodical, using agreed categories and argumentation it is not itself a ‘method’ but a way of describing and explaining the linguistics of texts and their meanings for readers (Toolan Reference Toolan1990: 28 and passim). Doing stylistics is primarily participating in an ongoing discourse that reflects on the multiple functions of language in our cultures and seeks to revise and renew our shared understanding of language forms and functions.
It may once have been assumed that stylisticians needed to wait for a sufficiently mature and flexible linguistics to emerge, ready for application; in recent years some stylisticians have tired of this client status. The kinds of complexity of language-related phenomena that stylisticians increasingly recognise as playing a part in the creation of a work’s meanings and effects have led them to review their various source linguistic descriptions and models more critically and questioningly. And where the proffered linguistic model is found to be a poor fit for the complex linguistic-analytical demands of the literary text, stylisticians are more willing than in the past to propose amendments and supplements to the received model.
Do competent readers (listeners, playgoers, etc.) need stylistics? We all like to be needed, so there’s a temptation for stylisticians to answer in the eager affirmative, seeing that responding in the negative only encourages those who would shove us to the cultural and academic margin. A more plausible reply is that readers may or may not find stylistics useful, just as music-lovers may but do not invariably find musicologists helpful. And even if some, many, or even all lay music-lovers found musicology of no use to them, this would be no compelling argument for abandoning such studies. Cultures need to understand their own practices, and not just do them. So a culture that puts great store by complexly composed imaginative fiction (prose, drama or poetry) in language, such great store that they canonise and anthologise it, adapt it into non-verbal artforms, and often spend a good deal of their free time reading it and talking about it, is sure to develop evolving reflexive discourses contributory to the understanding of that literature. Stylistics is the most language-focused of those discourses; its emergence was guaranteed in advance, with the emergence of artistic writing.
