Scanned Texts

Jack Lynch

Since the early days of the Web, I've been scanning and preparing electronic texts, mostly but not exclusively of eighteenth-century works.

The quality of E-texts available on the Internet varies tremendously, from the meticulous Rossetti edition prepared by Jerome McGann at Virginia (which will soon be the standard text), through the nearly worthless copies of some works taken anonymously and haphazardly from unnamed editions without textual authority, sometimes missing entire chapters, and often riddled with typos.

Having done my share of preparing electronic texts, I sympathize with those who've released bad ones: the obstacles are considerable. Copyright laws, for instance, restrict the availability of texts on the Net. Of course any copyrighted original work (at the time of writing, that means almost everything published after 1921) is excluded right away. But even those who work in earlier periods can reproduce only editions that have passed out of copyright, effectively excluding most twentieth-century scholarly and critical editions (Fredson Bowers, W. W. Greg, & Co. weren't let loose on texts until after the copyright cut-off). This forces E-text editors to work either from original texts, which are often difficult to obtain and impossible to scan (and may require considerable knowledge of textual criticism), or from out-of-copyright nineteenth-century editions, which are often ill prepared.

Because of these problems, I always recommend caution and skepticism in using almost any electronic text for serious research. To my knowledge, no E-text has yet superseded a hard-copy standard edition (although that will likely change soon). But even an imperfect E-text can be useful for some research purposes (when used in conjunction with a more reliable text), and, perhaps more important, making thousands of texts freely available promises to revolutionize pedagogy.

My motives for making these texts available vary widely. Some texts answer only my own needs for teaching: I want to make hard-to-find texts available cheaply to my students, without the hassle of ordering expensive anthologies or dealing with bulk-packs and reserve lists. These texts are sometimes abridged, modernized, or otherwise modified to suit the needs of my class, with little regard for others.

In other cases, though, I've tried to produce texts that may be useful to others — often things that aren't easily available in modern editions (such as Defoe's "Mrs. Veal" and D'Urfey's songs), or that are available only in expensive anthologies (such as Johnson's review of Soame Jenyns and his Life of Savage).

As the number of such texts grows (thanks to people such as Stuart Curran and Michael Gamer at Penn, who've made many previously rare Romantic texts widely available), instructors will be able to choose their teaching materials without regard for Books in Print at the bookshop and cash in hand among their students. They can put together custom anthologies, either by including pointers to the originals on the Net, or by printing copies and distributing them on paper, perhaps even with custom footnotes.

My own editorial practices have varied from text to text, depending at least partly on the primary use I envision for each one. In a few cases I've typed in texts directly from early editions (as in the case of Johnson's poetry), putting my meager knowledge of textual criticism to use — still, they're not critical editions but diplomatic transcriptions. More often, I've had to choose the best (or most important) nineteenth-century text available — such as Sir Walter Scott's Defoe, the 1825 Oxford Works of Samuel Johnson, or G. B. Hill's Prayers and Meditations in Johnsonian Miscellanies — which I sometimes reproduce verbatim, and sometimes correct or emend (always with a notice of what I've changed).

In any text I prepare, I try to follow these principles, and encourage others to use them as a baseline for responsibly prepared electronic editions:

Having some text is better than having no text: I believe there's a place for E-texts that fall shy of the rigorous standards demanded by a scholarly edition.
Ditto for encoding schemes: a fully marked SGML text is ideal, but (at least now) requires a great deal of effort from both the editor and the reader. Preserve as much of the original format as is convenient, but remember there's room for well-prepared HTML and even plain ASCII texts.
Choose the most authoritative text available, within the restrictions imposed by copyright laws on the one hand and scanning technology on the other.
Always identify the source of the text as clearly as possible.
Spell out all editorial principles, whether they're as simple as silently correcting inverted type and regularizing long s to short, or as extensive as abridging and annotating the text and modernizing accidentals throughout. Pay special attention to features that aren't easily transmitted in the coding scheme: are italics, small capitals, and diacriticals preserved?
Always identify the editor or preparer of the electronic text: if not the person who ran the scanner, the one who is responsible for the final shape of the text.
If this information doesn't appear on every file — such as when the editorial information appears on an electronic table of contents, which points in turn to individual items (stories, poems, essays, &c.) — encourage others to create pointers to this top file, and include a pointer from each individual item back to the table of contents page.
Errors are inevitable. Neither the prospect of financial reward nor the recognition of tenure boards encourages us to spend much time preparing electronic critical editions (a problem that calls for a separate solution). Therefore always encourage, and provide an easy way for, readers to notify you of problems — a "mailto" button on a Web page, for instance, with an address spelled out in full in the text itself.

All these principles can be reduced to a Golden Rule: provide potential readers with as much information about the provenance of the E-text as possible.

Send comments and suggestions to Jack Lynch.