In the latest issue:

Botanic Macaroni

Steven Shapin

What made the Vikings tick?

Tom Shippey

In the Lab

Rupert Beale

Will there be a Brexit deal?

Anand Menon

Short Cuts: Under New Management

Rory Scothorne


Bridget Alsdorf

Sarah Moss

Blake Morrison

Poem: ‘Country Music’

Ange Mlinko

On the Trail of Garibaldi

Tim Parks

Art Lessons

Peter Campbell

You’ll like it when you get there

Tom Crewe

Early Kermode

Stefan Collini

‘The Vanishing Half’

Joanna Biggs

At the Movies: ‘The Truth’

Michael Wood

The Suitcase: Part Two

Frances Stonor Saunders

Poem: ‘Siri U’

Jorie Graham

Diary: Getting into Esports

John Lanchester

Jane Austen’s Word ProcessMarilyn Butler
Vol. 9 No. 12 · 25 June 1987

Jane Austen’s Word Process

Marilyn Butler

3744 words
Computation into Criticism: A Study of Jane Austen’s Novels and an Experiment in Method 
by J.F Burrows.
Oxford, 245 pp., £25, February 1987, 0 19 812856 8
Show More
Show More

Why put the novels of Jane Austen onto a computer? The first thing that strikes you about Computation into Criticism is what it says about its Australian author’s dedication, or obsessiveness, or just plain nerve. Most literary research is cheap, and indeed looks very cheap as long as the cost of maintaining libraries is not counted in. John Burrows’s project of putting a dozen novels onto a computer was plainly from the first going to prove expensive. When one begins to cost Burrows’s travel, subsistence overseas, and time, together with computer-time, programmer-time and secretarial time, each of his 211 pages of text and 34 pages of statistical appendices comes to represent a sizeable public investment.

He has been supported in his native Australia by his own university of Newcastle, New South Wales, and by the central government, through the Australian Research Grants Scheme. The Fellows of St John’s College, Cambridge, have three times been his hosts, and he has used computing facilities, or received expert advice on computing, at both Cambridge and Oxford. He must have had to explain to each of his benefactors precisely why a book at once on computing and on Jane Austen was worth paying for. It’s interesting to imagine what went through the minds of the members of the committees concerned. There was, of course, something up-to-the-minute, the smell of New Blood, about a project that offered to apply technology to English Literature. Whether it seemed that technologists would be humanised, or aesthetes set to learn about computing, this one certainly looked educational. Some people must have relished the sublimely unobvious choice of Jane Austen as a target for modernisation. But they also had to face the risks. What if the finished book stayed on the shelves: too literate for the number-crunchers and too numerate for the literati?

The diversity of the specialisms Burrows had to reconcile is graphically stated in his title and subtitle – as fine a balancing act as one of Pope’s couplets. First and last, emphatic and aggressive, the pincers of ‘computation’ and ‘experiment in method’; in the middle, defensive or safely entrenched, ‘criticism’ and the reassuringly particular promise to study Jane Austen’s novels. Burrows, an exact writer, means just what he promises – one book that speaks two languages. As far as an amateur can tell, his book succeeds as a model exercise in computing and in statistics. It is certainly a model account of such an exercise. Literary readers are shown, with exemplary lucidity, how statistics can inform issues of style, and transform our understanding of the representation of character. Surprisingly general conclusions are arrived at by marshalling minute particulars. From among the tables and graphs there emerges the most accomplished ‘close reading’ to date of Jane Austen’s dialogue, and the most stylish book written on Austen since Mary Lascelles’s Jane Austen and her Art in 1939.

The fact that Burrows writes like Austen herself is splendid. But other people may write well, and not get the much smaller sums they ask for, so it’s worth considering why those who put our money into this project seem so triumphantly vindicated. English literature is an academic discipline that has grown ever more numerous and ever more dispersed, now that so many of the world’s populations learn English as a second language. To add substantially to its methodologies or to its theoretical understanding is a challenge, but also a problem, especially if you live far from the big academic centres in America and Britain. It is a major headache in Australia (and presumably throughout the southern and eastern hemispheres) to get the latest books, which will arrive selectively, slowly, and at twice the price asked in the country of origin. Had Burrows proposed, say, a sensitive new reading of Wordsworth, or a robust new application of Foucault to the 18th-century novel, or a helpful new commentary on Bakhtin or Deleuze, his scrutineers might reasonably have wondered, not just if the book was worth writing, but if Newcastle, NSW was a good place to write it.

At any one time, a large percentage of the professional books appearing in America or Britain represent small adjustments, or applications, of an immediately current idea, and in such cases speed, currency, the air of sharing a new vocabulary and of belonging to a club, must be an asset. Though it’s impossible for most books of this type to look individually distinguished, they represent the mainstream, and thus collectively are not useless. But Burrows’s project stands apart from all of this, its method ensuring its independence. His discussions only rarely introduce the views of others, and scarcely ever on issues of critical interpretation or evaluation. He generates his own evidence, and single-mindedly addresses it. If his book had failed, it would have looked eccentric or marginal, but since it is in general extremely persuasive, it derives further strength from its autonomy. Among its other ways of being exemplary, it demonstrates how to execute a research project in the humanities so that fellow professionals anywhere will take account of it.

The opening words are those of Pride and Prejudice, but the content puts one more in mind of Austen’s contemporary Thomas Malthus, whose Essay on Population sets off equally stylishly to display the elegancies of arithmetic: ‘It is a truth not generally acknowledged that, in most discussions of works of English fiction, we proceed as if a third, two-fifths, a half of our material were not really there. For Jane Austen, that third, two-fifths or a half comprises the twenty, thirty or fifty most common words of her literary vocabulary ... Eight personal pronouns, six auxiliary verb-forms, five prepositions, three conjunctions, two adverbs, the definite and indefinite articles, and four other words (“to”, “that”, “for” and “all”) ... almost always find their place ... among the thirty most common words of each novel.’ And, he adds, of the other novels he has examined. Then he unveils his first surprise: that he means to found a study of Austen’s novels on no other evidence than statistical analysis of the relative frequency of these very common words.

It may turn out that rival experts, perhaps whole sub-departments, have been working for years on the same lines as Burrows. But the rest of us, the literary generalists, have indeed tended to leave these words, the ‘harmless drudges’ of language, as Burrows calls them, to the grammarians or the literary sleuths. It has long been known that the incidence of even minor words differs from author to author. So characteristic of a style are they that in cases of doubtful authorship such as interpolations or forgeries, the patterns have been used as the equivalent in writing of a fingerprint. Their usefulness arises precisely because they are generally beneath the notice of the literary critic, and thus presumed to be beneath the notice of writers too, unconsciously chosen, hard to change, and hard to imitate.

Burrows challenges the view that such words are unconsciously chosen, and that they do not carry meaning. He is primarily concerned, not with Austen’s language when she writes in her own voice, but with the dialogue she gives her characters. By far the greatest part of his book consists of analyses of the speeches of her 47 major characters, those who speak more than two thousand words apiece, plus Eleanor Tilney, who ranks next with 1938 words. One of the first of his findings, and the most basic, is that Austen is consistently able to discriminate between the idiolects of her characters, even to their use of ‘that’ and ‘in’.

Jane Austen's major characters (word-types 1-30 of her dialogue).

Jane Austen's major characters (word-types 1-30 of her dialogue).

In addition to the six Austen novels and the unfinished fragment ‘Sanditon’, Burrows has put onto his computer as controls a small group of novels by others. They are Henry James’s The Awkward Age, E.M. Forster’s Howards End and Virginia Woolf’s The Waves, together with two modern attempts to imitate Austen’s Regency English, Georgette Heyer’s Frederica and the continuation of ‘Sanditon’ by Another Lady. He conducts some tests comparing the idiolects of the leading characters in each novel in his sample with the commonest words in the author’s narration, setting out the results in tables, and in some elegant and expressive graphs. (This is, obviously, no orthodox-looking literary text.)

It turns out that authors remain verbally true to themselves even when putting speech into the mouths of invented people. Miss Bates and Henry Tilney don’t sound like one another, yet they share the Austen speech-patterns, and would be linguistically out of place in The Awkward Age or Howards End. It follows that the dialogue of a novel, properly recorded, can be used as a test of authorship. But Burrows is more interested in the range of voices each of his novelists achieves, for he implies that the greater the range, the greater the writer’s care and skill. He finds the speech habits of Jane Austen’s characters more divergent than those of Forster, and considerably more divergent than those of James. Woolf’s six ‘speakers’ in The Waves come closer yet to one another and to the author’s narration. Georgette Heyer discriminates less interestingly than the major writers in the sample: she uses basically only two types of character, the adult and the infantile.

Austen emerges as very much the heroine of this book, a writer who is even more precisely attentive to language than we thought. So fully has she mastered the idiolects of her characters that she can even show them developing. At least, Burrows considers these modifications of the norm intentional, an aspect of characterisation. It is a case which works well enough for the five most changeable speakers in his table, Marianne Dashwood, Mr Knightley, Fanny, Emma and Henry Crawford, not so well for the next best learners, Mrs Norris and Mrs Bennet. Once again, other authors fail to match Austen’s remarkable diversity. Forster’s Margaret Schlegel ‘develops’ considerably more than other figures in her world, but less than 14 characters of Austen’s.

Mostly, however, it is by comparing Austen characters with one another, often with another figure in the same novel, that Burrows arrives at his most interesting discriminations. It is predictable but still satisfying that the sociable Admiral Croft of Persuasion uses ‘we’, ‘us’ and ‘our’ more than anyone else in Austen’s world (24.58 incidences per 1000 words), and the egocentric Lady Catherine de Burgh of Pride and Prejudice less than anyone else (2.13 per 1000 words). Equally, we know how utterly opposed in behaviour are the conforming Mr Collins and the wayward Lydia Bennet, but it takes a neat diagram correlating their word-types to prove that their differences are also finely verbal. Burrows’s commentary on the diagram wittily sums up his findings:

Chiefly in self-gratifying anecdotes, Lydia uses ‘was’ at eight times as high a rate as Collins. With his honoured patroness ever foremost in his talk, if not quite foremost in his mind, Collins uses ‘her’ at almost eight times as high a rate as Lydia, who has little attention to spare for the rest of the female population. With a tribe of plump sentences to support, he even uses the conjunction ’that’ at more than three times the rate she requires for her syntactic starvelings. The prosaic ‘at’, the refutatory ‘but’, and the merely accumulative ‘and’ are the only words of a connective function to fall on her side of the diagonal line.

It’s remarkable how a good diagram can at once break down a complex picture into its constituent details, and indicate how it may be simplified and generalised. In the diagram Burrows plots where all the major Austen characters stand linguistically, in relation to one another and to the narrative voice of their author. Imagine Austen herself standing on the middle of the bottom line, with Emma nearest to her, the sententious, half-realised Mrs Gardiner furthest away at the top. Reading from right to left, we get a hierarchical ordering of another kind, a sorting-out determined by the characters’ syntax. ‘In moving across the map from Harriet Smith to Collins,’ says Burrows, ‘we see a gradual transformation from garrulousness and intellectual indiscipline, through a middle area of civil and articulate speech-habits, to formality and dignity, and onward to pomposity.’ Ironically, Fanny Price’s childishness of speech causes her to be grouped here with Austen’s sizeable band of (mainly female) fools and vulgarians, including Mrs Jennings, Isabella Thorpe and Miss Bates. The diagram sums up Burrows’s Austen, an artist remarkably in command of her medium. Whether by labour or by intuition, she controls her creatures and their language, down to the lowest and least of its parts.

If there is to be argument about whether he is right in his representation of her language, it could focus on the diagram. Her diversity is well represented by what is at first glance the random scattering of names. Gradually, however, Burrows teaches us to see the strongly marked patterns. Behind Emma, the most Austen-like of Austen’s speakers, the characters spread out in a shape (Burrows’s analogy) like an angel’s wings. On the whole the men are behind her right shoulder, more formal and authoritarian in tone than she. On her left are the women, uncertain, ungrammatical, irrational, impressionistic. Her notion of character thus broadly maintains stereotypes of masculine and feminine, though it also allows a considerable number of exceptions on both sides. As for the other common structuring principle, class, Austen may seem to evade our questions by not allowing the wholly uneducated to speak. (The silence of Robert Martin in Emma, the character nearest to a working man, saves her from having to represent accented speech.) Yet on Emma’s left in the diagram we have all Austen’s vulgar characters, as well as her women – many are, in fact, the same people. It does look as though Austen’s characters and their speech-patterns are ordered on systemically hierarchical principles, male preferred to female and politeness to vulgarity. Or is it conceivable that Burrows’s focus upon the 30 or 60 commonest words, grammatical building-blocks, itself creates or exaggerates Austen’s commitment to logic and correctness? What if the 30 commonest words don’t represent the effects conveyed by language as a whole, let alone the total imaginative impact of a novel, its web of ideas and feelings? Could Burrows’s interpretation of his diagrams show him imposing too specific a pattern, and too much pattern, on Austen?

Some at least of Burrows’s findings will plainly not be welcomed by anyone who shares the type of objection to the 19th-century realist novel first raised by Roland Barthes’s observations about Balzac. Critics in this tradition find fault precisely with the techniques Burrows admires, strongly individualised characterisation and purportedly natural dialogue. They reject the idea that either characterisation or dialogue accurately represents personality or speech: these are merely conventions which reflect certain notions, challenged even in their own day, in favour of economic individualism and of educated speech. On page 94 Burrows seems to think he will have confounded critics of the Barthes school by proving that Jane Austen really does insist on the individuality of all her characters, and on the good grammar of those she most approves of. On the contrary, he will have confirmed their worst suspicions about her and about the kind of middle-class novel she wrote.

Burrows’s neat quotations from thinkers he admires, such as Gregory Bateson, are among the pleasures of his text. But, though his bluff, cantankerous persona is generally attractive, a kind of guarantee of his independence, he is really too cavalier about the opinions of those he doesn’t agree with. He gives only three paragraphs at this point in the middle of his book to outlining, and dismissing, what Barthes, Marx, Derrida and their followers might have to contribute to the notion of character: it is one of his rare perfunctory passages. Momentarily he allows himself to wonder about ‘genuine individuality’ and where it may come from – rather than about the more relevant charge, that it is arbitrary for novelists to make individuals seem so very unalike. ‘Some Marxist criticism’, he observes, with uncharacteristic vagueness, has tried to deal with the possibility that character-differentiation reflects social conditioning, but ‘in such terms as exclude all but gross, long-lasting influences like social rank and sexual category’, and these terms, Burrows claims, his evidence ‘transcends and refines’. Of course his evidence does nothing of the kind. His book confines itself to the artificial structuring of character within a literary artefact, the novel. A firm distinction has to be made, and oddly Burrows never pauses to make it, between the individuality we commonly find in real-life people and the much more marked distinctness or eccentricity with which characters are represented in many (but not all) novels.

When does strongly individualised character, expressed in dialogue, become a distinguishing feature of the novel? Minor characters become more vivacious, eccentric, linguistically distinctive as the 18th century wears on. Their individuality distinguishes them from the stereotypes – bragging soldiers, comic or cowardly or faithful servants – inherited from drama. The perception that society consists of a host of individuals, alike and notionally equal if only in their diversity, resembles the intellectual universe of Adam Smith: it seems to imply a degree of optimism about our freedom and potential. Novelists go on portraying the social panorama through minor characters’ diversity until this dispersed, atomistic emphasis is superseded by a generalising one. The sociological or biological or psychological essentialisms of Marx, Darwin and Freud belong to the same era as the gradual toning-down of ‘character’. Novelists, who represent individuals, and write for individual readers, quickly catch the disturbing implications of the great 19th-century systems of thought. Jane Austen thus falls in the heyday of individualistic fiction, near the middle of a line which includes Smollett and Burney before her, Scott as her contemporary, Dickens, Thackeray, Trollope and the younger George Eliot as her successors. Burrows’s demonstrations that her characters are more distinctive than those of James or Woolf may merely confirm that by 1900 the concept of character is generally in decline.*

But if this is so, he may not have chosen the best five controls. In any case, it was surely a mistake to squander his resources on Georgette Heyer and Another Lady. It was less unreasonable to go for James, Forster and Woolf, novelists who resemble Austen in using speaking characters who come from one class, and number as many women as men. But to choose only 20th-century novels as controls is to imply that historical change has little significance apart from minor adjustments to linguistic usage. Because it plainly does matter more than this, Burrows, while showing the way, has not yet established what is Austen’s sort of dialogue, still less what is the dialogue of fiction.

A question that remains open, for Burrows – as one profoundly hopes – to tackle in a sequel, is how far Austen was an innovator with dialogue. Though a diversitarian compared with early 20th-century novelists, how would she rate among the generations writing just before and after she first published, in 1811? In the novel of her day the outstanding innovation was considered at the time to be the introduction of dialect, supported by an attempt to render regional and lower-class accents phonetically. Austen does not introduce dialect speech, unlike many other genteel women contemporaries, such as Edgeworth, Morgan, Hamilton and Ferrier. If she had, Burrows’s task with the computer would have been more difficult. But her refraining from doing so makes her, in at least this respect, less diverse than her most significant contemporaries, including the most celebrated, Scott.

It would be foolish to try to guess what one of Burrows’s meticulous comparisons between an Austen novel and Edgeworth’s Belinda might reveal. How contemporaries actually read the different delineators of character is matter for another type of research – among reviews and letters. Austen’s first readers surely associated Smollett, Burney, Edgeworth and Scott more readily with diversity of character and of speech. Her tendency to represent women and vulgarians as foolish differs pointedly from Edgeworth’s counter-cultural example of making women masterful and articulate, and Irish peasants either shrewd or keen to learn. Scott, Tory or not, was even more inclined than Edgeworth to distribute wisdom among the middle and lower orders: dialect apart, the characters who sound most like their author include Baillie Nicol Jarvie and the pedlar Edie Ochiltree. In poetry, Wordsworth’s respect for the insights of seven-year-olds and vagrants raised the whole issue of the natural Tightness of the social hierarchy, to the irritation of his reviewer Francis Jeffrey and his friend Coleridge. If Burrows had chosen to compare Austen with any of these, would she still seem to stand out as a full-hearted individualist?

Burrows takes as a classic sentence one that illustrates how richly and diversely Austen’s characterisation works, Harriet Smith’s pronouncement in Emma: ‘I have now quite determined, and really almost made up my mind – to refuse Mr Martin.’ As Burrows reflects, it is a sentence that sends very different messages to different readers. One is that Harriet is still unhappy about the decision towards which Emma has been pushing her. Another is that she does not understand the word ‘determined’, which she has presumably picked up from the more educated Emma. A third is that this is the way that Harriet Smith, and no one else, uses language. The last of these messages is the one Burrows himself hears most willingly, the one that leads us with least complication to appreciate Jane Austen’s art. He has proved Harriet Smith’s uniqueness, and in a style which shows conclusively that technology may go hand in hand with wit and exuberant inventiveness. But where really has he left the more literary reader, the one naturally inclined to hear in this speech message one or two? A heartening humanistic faith in everyone’s individual value isn’t really Austen’s import, not at least if we believe Burrows’s examination of the diagram. There Austen’s shadow falls on her characters, the wings of a guardian angel with disconcerting notions of what we are all to be judged by. Despite Burrows’s own celebratory tones, I suspect that Harriet’s uniqueness, and Austen’s art, may emerge from this exercise less cheery than they seemed before.

Send Letters To:

The Editor
London Review of Books,
28 Little Russell Street
London, WC1A 2HN

Please include name, address, and a telephone number.


Vol. 9 No. 14 · 23 July 1987

SIR: Naive in the ways of English scholars these days, I ask for reassurance: Marilyn Butler is joking, isn’t she, in her review of J.F. Burrows’s Computation into Criticism (LRB, 25 June)? I mean, the whole thing is surely a gigantic spoof – postponed, perhaps, from a cancelled April Fools’ issue because no other entry could match its parodic subtleties or the accuracy with which it hits the targets of rationalisation (‘it’s worth considering why those who put our money into this project seem so triumphantly vindicated’) and truism (‘Surprisingly general conclusions are arrived at by marshalling minute particulars’)? Otherwise, what am I to believe? That J.F. Burrows really ‘proves’ the development of Jane Austen’s characters through a computational analysis of their dialogue? Hence Harriet Smith’s ‘uniqueness’? But I proved that in 1970 when I was a second-year undergraduate at Liverpool and, late with an essay, was reduced under pressure to word-counting and indexing as a substitute for hard thought. Such ‘proofs’ abounded. So she can’t be serious. I may add that tutors were tolerant then and allowed occasional backsliding – I even picked up a B+ (+?) for my efforts. I dread to think how savagely, albeit wittily, Professor Butler would deal with someone like me now. At least, I suppose she would …

Alma Cullen

send letters to

The Editor
London Review of Books
28 Little Russell Street
London, WC1A 2HN

Please include name, address and a telephone number

Read anywhere with the London Review of Books app, available now from the App Store for Apple devices, Google Play for Android devices and Amazon for your Kindle Fire.

Read More

Sign up to our newsletter

For highlights from the latest issue, our archive and the blog, as well as news, events and exclusive promotions.

Newsletter Preferences