Wednesday 6 August 2014

Identifying Eliza Haywood's Sources

I have just read Douglas Duhaime’s essay on “Digital Approaches to Intertextuality: The Case of Eliza Haywood,” which was published on his blog on 3 January this year (see here; for a profile of Duhaime, see here and here). I am in danger of developing a tick if I say the essay was “thought provoking,” but it was. Very.

Duhaime is interested in “the ways in which writers borrow language and ideas from other writers.” He has used his considerable computer skills to write a plagiarism-detection script, which he has used/tested/developed on Haywood's Betsy Thoughtless. As a result of his analysis, Duhaime is now able to suggest a source for almost all of Haywood’s quotations and paraphrases not previously identified by editors.

Like all editor of Haywood’s works, Christine Blouch was frequently forced to admit defeat with the statement "Source unidentified." And like all editors of Haywood’s works, I am sure she wished she had had access to Duhaime’s skills and his plagiarism-detection script before having to make such an admission in her Broadview edition of Betsy Thoughtless. Having edited a number of eighteenth-century texts myself, I know how much work is represented by these two words—"Source unidentified"—often much more work than is represented by a neat and succinct reference to the Bible, a Classical author, or a Canonical writer.

Having "set out to uncover the materials that informed Haywood's work," Duhaime used an API to search Literature Online's text-base. (An API is an Application Programming Interface. Duhaime used Python's Selenium package to create a script to “programatically investigate” the text-base. See here and here for details.) It emerged that “in many cases,” Haywood's unidentified literary borrowings were fairly easy to identify—though the source was occasionally mis-identified by Haywood. In this way Duhaime identifies quotations from Nathaniel Lee, Alexander the Great (1677), John Dryden, The Spanish Friar (1681), William Congreve, Love for Love (1695) and The Mourning Bride (1697).

Of course, since these works are easy to find on EEBO, ECCO, Google Books and the Internet Archive, there really is no difficulty doing this sort of search online without an API, although it is obviously more time-consuming. No matter how dodgy the OCR of each individual copy digitised, the multitude of copies and editions available online means that just about any short search string (any short sequence of words) is likely to lead you to the source of a quotation. I have done this myself, many times, with interesting results.

(In June of 2012 I did a post on a quotation repeatedly used and mis-identified by Haywood (see here), which I concluded with the words “in the Pickering & Chatto edition of Ab.64 Epistle for the Ladies, edited by Alexander Pettit and Christine Blouch, the verse is not identified … [this] edition was published in 2000, back in the before time, the long-long-ago. When the internet was young and you had to read whole books to find a quote, or miss one, as the case may be. Oh how far we have come in only a decade …)

So the really interesting part of Duhaime’s essay is not so much the “many cases” in which the source was easy to locate, but the discovery of many other “instances of intertextuality in Haywood's writing” where a given quotation—usually multiple lines of verse—“seems to derive from multiple sources.” This is where the API appears to really come into its own. By quickly chopping up any given Haywood quotation into segments, searching for those segments, and presenting the findings in a convenient spreadsheet, Duhaime’s API makes it possible to establish that Haywood “often combined lines from disparate literary works in order to forge her own ideas” (italics mine).

Duhaime argues that three of the following five lines, for example, are cobbled together from three sources:

  Pleas'd with destruction, proud to be undone,
  With open arms I to my ruin run,
  And sought the mischiefs I was bid to shun
;
  Tempted that shame a virgin ought to dread,
  And had not the excuse of being betray.

[1] Richard Blackmore, “Advice to the Poets” (1718): “Let them this gen'rous Resolution own, / That they are pleas'd and proud to be undone”; [2] Mary Wortley Montagu, “The Basset Table” (1716): “I know the bite, yet to my ruin run, / And see the folly which I cannot shun”; [3] Aaron Hill, “The Excursion of Fancy: A Pindaric Ode” (1753): “Let us throw down this load of doubt, with which no race is won: / And, swift, to easier conquests, lighter, run, / The way, which reason is not bid to shun!

Although this is not—contrary to Duhaime’s claim—a clear case of Haywood “combin[ing] lines from disparate literary works,” like Lego blocks stacked one upon another, it is a particularly intriguing selection of works which Haywood appears to echo in her poetry. The third echo, from Hill, is perhaps the most interesting of all because it seems far less likely to me to be a source text than the Montagu text (which is much closer to Haywood's actual lines). Since Hill's influence on Haywood is well-established a critic who found the Hill echo first/alone could be forgiven for looking no further and use the echo to suppport the existing interpretation of Haywood as being heavily-influenced by Hill. Having multiple echo-texts simultaneously presented as possibilities, it is less likely that a critic will be side-tracked by a single, convenient partial-match, like that from Hill.

Duhaime’s second example is slightly more convincing:

  When puzzling doubts the anxious bosom seize,
  To know the worst is some degree of ease.

[1] Joseph Mitchell, “Poems on Several Grave and Important Subjects”: “When puzling Doubts invade my Breast, / And I am cloath'd in Shades of Night . . . "; [2] Davild Mallet, Eurydice (1731): “When others too / are miserable, not to know the worst / is some degree of bliss.”

Here, “invade my Breast” becomes “the anxious bosom seize,” and “degree of bliss” becomes “degree of ease”—the sole changes made being necessary to make the couplet rhyme.

* * * * *

Although I think the really interesting part of Duhaime’s essay is not the “many cases” in which the source was easy to locate, the fact is one could feed every snippet of verse quoted by Haywood into Duhaime’s API and produce a spreadsheet of Hawood’s sources, ranked by author, title and number of quotations, total number lines etc. Such a list would offer an intriguing insight into Haywood’s own “reading” and the major influences on her writing.

(I put “reading” in inverted commas because it is already clear that Haywood frequently quotes plays and often seems to do so from memory, suggesting that she spent a lot of time at the theatre—a suggestion supported by her own plays, acting, and her two volumes of drama criticism—The Dramatic Historiographer (1735) and A Companion to the Theatre (1747).)

If a spreadsheet of Hawood’s confirmed sources were used in conjunction with a spreadsheet of the works Haywood appears to echo, it may be possible that something like a reasonable or defensible judgement could be made between competing texts (in cases where there are two equally-possible origin-texts of similar constructions). That is, if a corpus-wide analysis of Haywood's sources shows a multitude of quotations from author A and none from author B, then it is more likely that Haywood is paraphrasing the former than the latter. (Though the example above with Montagu and Hill suggest that, even here, each case must be judged on its merits.)

I hope Duhaime decides to expand his work on Haywood and to publish his findings: I, for one, would love to have them on hand when editing her works.