Literary sources favoured
How in practice did the lexicographers determine the range and nature of the sources evidenced in the final publication? What conclusions about language and writers is it therefore legitimate to draw from OED quotations?

One of the first tasks of Coleridge (confirmed as editor in 1859) and his successor, Furnivall, was to make out a list of books and assign volumes to individual readers (volunteers were not in short supply and within two years of Trench's original lectures more than 100 readers had offered their services). Exclusion of certain sources began straightaway. The Philological Society's Proposal (1859: 3) declared that 'all English books' were to be admitted as 'authorities' (i.e. acceptable sources) 'except such as are devoted to purely scientific subjects, as treatises on electricity, mathematics, &c., and works written subsequently to the Reformation for the purpose of illustrating provincial dialects'.

We can see from the early lists and appeals issued and re-issued over the next few years (now preserved in the OED archives; see specimens photographed in our Historical documents section) that the sources recommended for perusal by the lexicographers were predominantly literary. They were divided into three periods, marked by the date of the 'first printed English New Testament' (i.e. Tyndale's in 1526) and the death of Milton (1674).

Identifying vocabulary for inclusion
To help them identify which words should be noted, contributors were asked to consult three 'Bases of Comparison':

1. the words printed in the Glossarial Index to the Printed English Literature of the Thirteenth Century compiled by Herbert Coleridge (Coleridge 1859)[1]
2. concordances to the Bible and Shakespeare  
3. an index of words in the works of Edmund Burke

and to provide 'a quotation for every word, phrase, idiom, &c.' in their source which was unrecorded in the relevant 'Basis' (Proposal 1859: 5-6, 8).

The Burke Index, if it ever existed, was never produced in separate form. As Herbert Coleridge explained two years later, it was 'found advisable to abandon this plan'; it was replaced with a substitute 'Basis for Comparison' issued in three parts, 'formed by extracting for each letter a number of words large enough to serve as a foundation, from the writings of Dryden, Wordsworth, and Tennyson, and then adding to this substratum all, or nearly all the contributions for this Period already in the Editor's hands' (Coleridge 1861: 4; Preliminary Notice).[2] (The special attention paid to these three authors at the outset resulted in significant quotation from Dryden and Tennyson in the Dictionary when it was eventually published, but not so many quotations from Wordsworth: see Top sources for Dryden and Tennyson, and the graph below for Tennyson and Wordsworth.)

The Proposal (1859: 5-6) had acknowledged that the reliance on these three 'Bases', i.e. word lists and concordances from specific sources, was not ideal, and in the event it turned out to be most unfortunate, as it dissuaded readers from recording usual words, whose subsequent documentation consumed vast amounts of valuable editorial time (see further Issues & problems in our pages on Reading and readers).

'Principal writers'
Contributors who preferred to work on nineteenth-century literature were asked to analyse carefully 'the works of any of the principal writers, extracting all remarkable words, and all passages which contain definitions or explanations...Wordsworth, Scott, Coleridge, Southey, Tennyson, Ruskin, Macaulay, and Froude may be mentioned as pre-eminently important' (Proposal 1859: 6).

This list of names directs prospective Dictionary readers to sources then commonly recognized by the educated middle classes, without defensiveness, embarrassment, or anxiety, as canonical - not just for English literature, but for the English language in its entirety (the distinction was not recognized as a meaningful one).[3]

The list also makes it clear that writers who use language in markedly idiosyncratic ways were not to be excluded. Here Walter Scott is the egregious example, and the graph below shows how extraordinarily productive his oeuvre turned out to be in yielding quotations for OED.

Eventual quotation totals for each of the nineteenth-century authors specified in Philological Society's Proposal


(This graph should be compared  with that at Top sources: Scott is the second most quoted individual author in the OED after Shakespeare and towers above many other distinguished authorities for the English language (as represented in OED); as we note elsewhere he was also an astonishingly popular author whose works were printed in enormous quantities. We have yet to analyse more fully the character of the words instanced from his work, but Scott seems to have been a rich source of several different types of usage – dialect and regional vocabulary (cailleach, dinmont), revivals from Middle English and Scots, where Scott is the only example cited for two centuries or so (e.g. bruckle, dindle), as well as archaisms (dern), learned or facetious hapax legomena (ambagitory) and nonce-words (debind), together with hosts of 'ordinary' usages. More generally, the variation in numbers of quotations for these authors is notable. Does it reflect the linguistic qualities of their work or the reading and quoting preferences of the lexicographers and their volunteers?)

Literary bias?
Do these instructions for the original readers, and the sorts of sources that were read for the Dictionary, mean that, as Dennis Taylor has observed, 'the OED's reliance on literary quotations is problematic because it skews the representative character of the sampling' (1993: 6)?

The answer is probably yes. Not only did the initial instructions to readers emphasise literary as against other sorts of sources, but the contributors who chose (and were in an economic position to be able) to volunteer their services were likely to be enthusiastic literary readers (see Knowles 2000). In addition, as outlined in Literature and the nation, there was strong external pressure to quote from literary sources.

For more see Literary vs. other sources below.


Footnotes
[1] Coleridge wrote in his Preface that 'The present publication may be considered as the foundation-stone of the Historical and Literary portion of the Philological Society's proposed English Dictionary', explaining that 'the raw material of the Dictionary, the words and authorities, are being brought together by a number of independent collectors, for whom it is consequently necessary to provide some common standard of comparison, whereby each may ascertain what he is to extract, and what to reject, from the author, or work, he has undertaken. This standard for works of earlier date than 1526 is furnished by the following pages, which contain an alphabetical inventory of every word found in the printed English literature of the 13th century' (Coleridge 1859: iii). His Index is based on around 30 Middle English works or collections previously published; many more were subsequently available in print after Furnivall had founded the EETS.
[2] The first new 'Basis' for the third period, a list of words beginning A-D, compiled by Herbert Coleridge shortly before his untimely death, was produced in February 1861; the remaining two were compiled by Furnivall, the list for E-L appearing in April 1861, and that for M-Z in March 1862. In the Preliminary Notice to Part II, Furnivall records that 'words since furnished by contributors...and by Mr. Rossiter's Index to Burke's Works' had been added to the lists. See Historical Introduction: ix-x, and Coleridge 1861, Coleridge and Furnivall 1861, Coleridge and Furnival 1862. We thank Peter Gilliver for tracking down the reference to Burke's Index and elucidating which 'Basis' was which.
[3] See OED s.v. literature 3. a.: '...the body of writings produced in a particular country or period...'