A Protocol for Making Backlist Volumes Respectable

A guest post by Gemma Coole, Publisher, CoolMain Press

In only two years, CoolMain Press, a micropublisher, has published eighteen superb books, all of them international bestsellers, including six previously published backlist books republished as ebooks. Among the other recognitions are prizes to our authors for literary quality and threats of lawsuits for truth-telling that no one else would risk.

Unfortunately, working with such long-established authors as Andre Jute Andrew McCoy means that many of the attractive volumes from their backlists that we want to republish are not available in any kind of electronic form. We must use optical character recognition to render printed books back to editable texts.

You might say, “But writers like Andre and Andrew, both from the Secker & Warburg stable, were surely superbly edited before their books were published.” Yes, they were. But optical character recognition is only 99% accurate, and that’s at best, with a truly clean, aligned scan of a page set in a font that the particular OCR you’re using just lurfs. I would love to see that fortunate concatenation of circumstances just once in my life! In practice the best and most expensive OCR (Adobe’s Acrobat) is 90% accurate. Think about it. OCR gets 10 or more out of every hundred characters and punctuation marks wrong, and of course it also gets any speck of stray ink or dirt on the page wrong, classifying it as a spurious character or, worse, an anomaly to be individually attended to.

The right way to proceed with the mess that OCR leaves one, is first to reduce the uncertainties to the maximum extent by enthusiastic use of the spellchecker built into all good word processors. Then reduce all consecutive multiple spaces to a single space. That’s the easy part but it can waste an inordinate amount of time to overlook it.

What’s left is then the most intransigent of the OCR errors, including some that are very difficult to spot unless each page is extremely carefully examined at good magnification.  Particular examples that my own eye just glides over — until the book is published, when they become knives in my heart — are reversed quotation marks, ticks for quotation marks instead of proper curliques, stops and commas  switched around at random, exclamation points turned into the digit 1, and hyphens exchanged for en and em rules. All these things matter.

After the spellcheck, it usually makes sense to find all hyphens and inspect each individually to ascertain which should remain, which are superfluous hyphens picked up by the OCR from the line-ends of the previous setting and turned “hard” when no hyphen is required, and which should be em and en rules to show broken thoughts or subsidiary material.

Next we need, for each book individually, to work out a set of algorithms which define the order in which quotation marks will be regularized. By this I mean that quotation marks should have all occurrences of straight ticks replaced by curliques, that they should all point in the right direction, and that, wherever possible, British single quotation marks should be exchanged for American double quotation marks.

This raises another question, apostrophes. In a solid pre-edited book from a classically-educated author and a leading literary publisher, one doesn’t need to look for spelling errors, grammatical solecisms, and the infuriating carelessness with apostrophes (tomato’s — aaargh!) one finds in the “books” that some conglomerates and too many indies put out. But correctly applied apostrophes can trip you up thousands of times in a single novel if you’re converting from single quotation marks to double.

So the algorithm you work out must in the first instance change all the opening quotes for each sentence which is dialogue, and observation of the text you’re working with will quickly show that there are two forms to search for, paragraph-quotation to open paragraphs and space-quotation after intervening description, in both cases right-facing curliques. Similarly, at the end of any unit of dialogue, there will be a stop, a comma, an exclamation, a question mark, or a rule, followed by left-facing curliques. You will now be left with some self-standing quotation marks which can be found as space-quotation and individually attended to. Anything else is an apostrophe or an error requiring editorial judgement.

Next, attend to the apostophes. Search for them one by one from the beginning of the volume and when you find an error, correct it for the entire volume by search and replace, then find the next error, until no more errors are found. At this point, you have reduced the errors in the OCR to the maximum extent possible by semi-automated means.

You still have a bit of a mess but at least the editor will not waste hours and days on what you can do with a computer in a few hours or even minutes if you’re organized or lucky.

Of course, at CoolMain, a tiny publisher, we would never have been able to publish so many hefty books in such a small period of time if it were not for the volunteer editors — currently Diane, Lisa, Lynne, and Sarah, with at least a dozen more who’ve helped on one or more books — working for the love of literature in Andre’s Editorial Menagerie on Goodreads.

I understand why the big publishers hate backlist books. It’s the cost of OCR. That makes me all the more proud of what CoolMain has achieved with very limited financial resources and the goodwill of the outstanding volunteers.

If you see a remaining error in a CoolMain book, please write to us: info at coolmainpress with the commercial extension. We will make the correction at the next edition.

***
Gemma Coole was a merchant banker until she realized her life’s ambition to become a publisher when she founded CoolMain Press with Bill Tremain. Among CoolMain’s bestselling, prizewinning authors  are Andre Jute, Dakota Franklin and Andrew McCoy.