The canon

The poets and novelists of the Victorian literary canon were favourite quotation sources for OED1. This page briefly recapitulates issues concerning OED1’s identification of a canon, then discusses in more detail the views and practice of R. W. Burchfield, editor of the 1972-86 OED Supplement, under the sub-headings Burchfield’s Supplement and Language and literature.1 A shorter section follows on OED Online and OED3.

See further our overview of Literary sources.

OED1 and 1933 Supplement

The Victorian and Edwardian lexicographers who created the first edition of the OED were relatively confident of and unembarrassed about the existence of a culturally homogeneous and consensually identified literary canon (as discussed further at Literature and the nation; see also Top sources and ‘Principal writers’ in Initial practice on assembling OED1). This reflected the prevailing assumption that literature was the background of language, so that understanding the history of the language entailed examining the usage of great writers of the past.

By contrast, it is clear to us nowadays that the concept of a literary canon is in itself problematic, while identifying a body of writers who might be thought to constitute such a literary canon is a tricky undertaking. We would have to consider, for example, a balance (however determined) of female with male authors, or native British with Commonwealth or other post-colonial writers, or with other writers writing in English. Moreover, after the birth of linguistics as a new academic discipline at the beginning of the 20th century, we would question the view that literary writing is in some way the epitome of the language – though you can find this view implicitly present in early descriptive linguists such as Jespersen, who is happy to refer to literary sources (Chaucer, or Shelley, Byron, Tennyson, Dickens, Swinburne etc.) to illustrate his analyses of language.2

The chief editor of OED1, James Murray, displays some interesting signs that he did not regard literary usage as paramount in the way that his contemporaries did. Thus he told the Philological Society in 1884:

The only rule was to take the best quotation you had for the word you had to illustrate, and not be so silly as to choose a poor quotation because it had a big name tacked on to it.

Murray 1884b: 372 (italics added)

Contrast the view of OUP publisher Kenneth Sisam, who had an important influence on the first Supplement of 1933, writing to G. G. Loane on 9 March 1932: ‘I should have thought any words in R. L. Stevenson’s Letters which were not purely freakish or native would be at least worth consideration, for an important author must always have a preference’.3

Despite this comment, literary sources were not an especially strong feature of the first Supplement of 1933 – and R. W. Burchfield, the editor of the second Supplement, tells us that Sisam ‘expressed contempt for the poetical vocabulary of poets like Blunden and de la Mare’ (who seem to have been instanced not on account of their singular characteristics but only insofar as they represented contemporary poetry generally; Burchfield 1984: 117).

Burchfield’s Supplement

However, when R. W. Burchfield became editor of the second Supplement in 1957 he took the view that the OED should continue its role as ‘a literary instrument’, providing ‘a “key”‘ not only ‘to the central core of the language’ but also ‘to the great literature of the English-speaking world of the period since about 1800’.4

Burchfield shows no sense, at any stage in his career, that defining ‘great literature’, especially of the contemporary period, might be contentious. In the Introduction to the first volume of the Supplement, he instanced the vocabulary of Kipling, Yeats, James Joyce, Dylan Thomas, Beckett, Sinclair Lewis (for the term Babbitt), Lewis Carroll, T. S. Eliot, D. H. Lawrence, Roy Fuller, C. P. Snow, Auden, de la Mare, and Tolkien as self-evidently deserving inclusion in the OED, making no attempt to explain his choice of these writers (OEDS1 1972: xiv). Nevertheless, a hostile review of his third Supplement volume, by Roy Harris, seems to have made him aware that identifying a literary canon, and quoting from its contents in order to illustrate the history and development of the English language, raised linguistic if not cultural questions (Harris 1982). Instead of analysing the relationship between literature and the lexicon and arguing for his own position, however, Burchfield responded with self-justification and counter-attack in the Preface to his fourth volume:

The period of preparation of OEDS [i.e. Burchfield’s Supplement] has coincided with the arrival of new linguistic (and especially structuralist) attitudes. What I have elsewhere called ‘linguistic burial parties’ have appeared, that is scholars with shovels intent on burying the linguistic past and most of the literary past and present. I refer to those who believe that synchronic means ‘theoretically sound’ and diachronic ‘theoretically suspect’. It is theoretically sound, so the argument of the synchronicists runs, to construct contrastive sentences or other laboratory-invented examples which draw attention to this or that element of lexis, and to do only that. I profoundly believe that such procedures, leading descriptive scholars never to quote from the language of even our greatest living writers, leave one looking at a language with one’s eyes partly blindfolded.

OEDS4 (1986): x-xi

He continued, ‘A small measure of autobiography is necessary’, and described how his own research on the late 12th-century text Ormulum, and his teaching of the language of ‘writers like Spenser, Shakespeare, Milton, Dickens, and many others’ to Oxford undergraduates, had demonstrated to him ‘that the entire vocabulary of all the main literary, philosophical, religious, etc., works [since 1100] had been included in the OED’.

If the OED ‘had room’ for eccentric usage of the past, he concluded,

it could, and must, admit the vocabulary of Edith Sitwell and Wystan Auden. Of course, the structuralists and other scholars at one or two removes from the work of Ferdinand de Saussure could not see this, and they probably never will. But OEDS, like its parent work, has been hospitable, almost from the beginning, to the special vocabulary, including the once-off uses, of writers like T. S. Eliot, Virginia Woolf, D. H. Lawrence, and others.

OEDS4 (1986): x-xi.

As these remarks make clear, in differentiating his lexicographical method from that of present-day linguistics, Burchfield appealed to standards of judgement (his personal history, ‘our greatest living writers’) which are not objective or fully articulated. 

Language and literature

In the same Preface, Burchfield explains that his quotation from the special usage of canonical literary authors forms ‘only a tiny fraction of the vocabulary presented here’ and has not disturbed the ‘balance’ of his record of the language. Elsewhere he refers to such quotations as ‘golden specks’ in the Dictionary as a whole, and describes how he smuggled them into the Supplement against the inclinations both of his staff – ‘my staff (I don’t know about anyone else’s) have a genuine horror of poets. I love poetry and poetical use has been poured into the Supplement, because it is my own preference compared with that of my colleagues’ – and of his ‘publishing overlords within OUP’ (Burchfield 1989: 12; 1980: 282).

The view that the distinctive uses of literary writers are on the one hand the backbone of the language and must be faithfully recorded by the OED if it is to explain the history and development of English, but are on the other hand ‘golden specks’ which must be swamped by a welter of other entries so that they are not too visible, is not entirely coherent. As Murray’s co-editor Henry Bradley recognized in 1904, some writers – he instanced Lydgate, Malory, Caxton, Spenser, Shakespeare, Pope, Johnson, and Walter Scott, among others – have had a marked influence on language, ‘either because of their boldness in the introduction of new words and new senses of words, and the extent to which their innovations have found acceptance, or because their writings have afforded abundant material for literary allusion’. Others, by contrast, have not – or at least, not in clearly demonstrable ways – for example (Bradley suggests, contentiously) Chaucer, Milton, and Carlyle.5

All these writers are generously quoted by the OED, to the delight and satisfaction of generations of readers who value the Dictionary as a treasure-house of the writings of major contributors to the literary canon. But OED’s quotations from such sources tell us as much about the values of the lexicographers and their readers as about the language itself: as it now appears, some writers – late 16th-century, or male, for example – were favoured over others – 18th-century, or female – for cultural, not linguistic, reasons (see material under Period coverage).

Burchfield emphasizes many times his fondness for inclusion of the hapax legomena and eccentric usages of literary writers – Beckett’s athambia, Joyce’s impotentising, Woolf’s scrolloping, Edith Sitwell’s Martha-coloured – ‘the result of a personal memory…As a child, I had a nursery maid called Martha, who always wore a…gown…exactly the colour of a scabious’, Hopkins’s unleaving, etc.6

He says:

I can best illustrate my own attitude towards literary English, and its preciosities, in the following manner. I have been as much concerned to record the unparalleled intransitive use of the verb unleave (‘to lose or shed leaves’) in G. M. Hopkins’s line:

Margaret, are you grieving/Over Goldengrove unleaving’ (‘Spring and Fall’, ll. 1-2)

as Murray was to record Milton’s unparalleled use of the word unlibidinous:

But in those hearts/Love unlibidinous reign’d (Paradise Lost, Bk V l. 449)

or Langland’s unparalleled use of unleese ‘to unfasten’:

Seruiantz…nau3t for loue of owre lord vnlese here lippes onis (Piers Plowman, B-Text, Pro 213)

Burchfield 1989: 173-4.

But Burchfield never tackles the paradox at the heart of the relationship between literary language and language more generally. As countless scholars, and writers themselves, have investigated, literary writers often choose to express themselves by deviating from ‘ordinary language’ rather than merely exemplifying it.

One clear disparity in Burchfield’s identification of a literary canon of 20th-century authors deserving of quotation in the OED is his different treatment of male and female authors. While the male-authored canon evidenced in his Supplement – and carried over without change into OED2 – is fairly conventional (e.g. G. B. Shaw, Kipling, Joyce, Lawrence, with c. 1,500-2,000 quotations each), his favourite female authors, who are quoted in far fewer numbers (c. 450-480 quotations each), are the three detective writers Agatha Christie, Dorothy Sayers, and Ngaio Marsh (see 20c fe/male sources). Virginia Woolf and Edith Sitwell have respectively 237 and 215 quotations in the Supplement.

OED Online and OED3

It is hard to assess how OED3 is in practice adjusting their predecessors’ de facto identification of a canon. As we saw in the discussion of OED3 in Policy and practice, the new edition is expanding its coverage of ‘many non-literary texts’, aware of criticism of past editions of OED for undue literary bias. But the results cannot as yet be clearly identified and evaluated. The current version of the Dictionary at OED Online does not permit users to identify which entries have been revised for OED3 (other than by looking up individual entries one by one), or to compare revised entries with their unrevised equivalents in OED2, or to identify in any other way new patterns of quotation.7

Searches of several handfuls of canonical authors in OED2 and OED3 made by EOED project before 2010, when the OED2 database was still available on the OED Online website, did appear to show that some such writers were being targeted for special treatment in the early stages of revision – that is, their quotation tally was rising sharply. For example, by 2010, Austen’s quotations had gone up by over half (from 1,044 in OED2 to 1,607 in OED Online, an increase of 563), and Joyce’s by over a third (from 1,749 to 2,311, an increase of 562).

Over the last nine years, however, this enthusiastic excerpting has tailed off, amounting to just over 100 additional quotations apiece for these two authors: as of August 2019, Austen’s total is 1,733 and Joyce’s 2,472. These are nevertheless large figures for individual writers. Looking through the individual quotations concerned, many (though not all) for unremarkable uses of language, it is difficult to understand the special attention given them as other than culturally determined. For example, so far we haven’t come across any comparable increase for individual non-literary authors.

The other available recourse for those interested in OED3’s treatment of the canon is OED Online’s own list of 1,000 most-quoted sources. This changes every quarter and is based on the current mixture of unrevised with revised entries. It shows unequivocally that – apart from newspapers and periodicals, from which tens of thousands of new quotations have been added – OED1’s canonical bias continues to dominate the Dictionary’s quotations from individual sources. See screenshot and discussion at Policy and practice and compare the pages under Top Sources.

Last updated on 8 October 2019


  1. Material here draws on Brewer 2007b.
  2. See e.g. the first chapter of Jespersen’s seminal Growth and Structure of the English Language, originally published in 1905, a work in which he warmly acknowledges his debt to the OED.
  3. OED archives: Misc/393/54 (italics added).
  4. Burchfield uses the term ‘literary instrument’ in a report to OUP on the progress of the second Supplement in 1962 (Brewer 2007b: 165); the second statement was made in 1980, after the appearance of the first two volumes in 1972 and 1976 (Burchfield 1980: 279).
  5. In acknowledging that ‘it might well be expected that in any notice of the literary Makers of English a large place must be given to Chaucer’, but that ‘it is singularly difficult to prove this by definite examples’, Bradley 1904: 215-40 (226) anticipates Cannon 1998. The point in both cases is that reading more widely in Middle English texts earlier than Chaucer may show that his usage was less singular than might appear; Cannon uses the evidence of MED to demonstrate this. Carlyle’s extraordinarily eccentric vocabulary, quoted some 6,240 times, is often represented by OED as hapax legomena; but at least one of his terms, gigman and its derivatives, penetrated general usage (as OED’s definition – written by Bradley – helpfully describes: ‘one whose respectability is measured by his keeping a gig; a narrow-minded person belonging to the middle class, who views “respectability” as the chief concern of life, a “Philistine”‘).
  6. On scrolloping see Brewer 2009d.
  7. See Re-launched OED Online.