A few weeks back, I gave this talk at KU's Hall Center for the Humanities. What fun!

From Rayna Kalas, Frame, Glass, Verse: The Technology of Poetic Invention in the English Renaissance (Cornell UP, 2007), 16. This reads to me like a germ of all the great work on materiality, poetry, making, language, historicity, etc. etc. done in the 16 years since this was published.

An epigram published in 1608, and I don't quite know what to think about it:

A Scholer newly entred marriage life,
Following his studdie did offend his wife,
Because when she his company expected,
By bookish busines she was still neglected:
Comming vnto his studdy, Lord (quoth she)
Can papers cause you loue them more then mee:
I would I were transform'd into a Booke
That your affection might vpon me looke,
But in my wish, withall be it decreed,
I would be such a Booke you loue to reede,
Husband (quoth she) which books form: should I take,
Marry (said hee) were best an Almanacke,
The reason wherefore I doe wish thee so,
Is, euery yeare wee haue a new you knowe.

I'm not sure if this is something about which one should boast, but I am an advanced user of the Early English Books Online (EEBO) platform. My scholarly work depends, to a rather drastic extent, on the 130,000ish scanned books and 66,000ish transcribed image sets that make up the EEBO and EEBO-TCP. I have given entire papers (and, you could argue, written entire books) based on what I found while screwing around on EEBO.

Recently, ProQuest, the owner/manager of EEBO, has moved away from the longstanding EEBO interface (with all its frustrating charms and deceptively powerful search capabilities) and has implemented the corpus/images into the standard ProQuest interface. There are some good things about this change, which I won't dwell on here. Instead, I want to illustrate one enormous, cataclysmic, frankly awful limitation to the new interface: you cannot easily browse particular results.

The old (and soon to be abandoned) EEBO interface involved rather Byzantine search protocols with  major virtues. Let's say I wanted to look for the word "love," which in early modern English could be spelled various ways, including "loue," "looue," and "lufe." I'd simply search with Variant spellings and forms selected:

Then I scroll through the results until I see something I want to explore further. In this case, I chose Fedele and Fortunio:

So far, so good. I now click on the little "full text" icon, and find the blue link for "First Hit"--that is, the first instance of "love" (IN ANY SPELLING) in the text.

I can then scroll through all the instances of "love" (in whatever spelling or form) in this text.

It's extremely clumsy, and awkward, and a bit web 1.0, but it's also very powerful. In my own research, this feature has allowed me to scroll across thousands of instances of particular words, phrases, and forms. The new ProQuest interface, by contrast, does not allow this kind of scrolling. Or rather, it makes it much more difficult. (I acknowledge the possibility that I've simply missed some feature of the search tool, despite spending hours reading the documentation and test searching.)

Let's say, again, that I want to search all the forms of "love." ProQuest searches those variants by default, and  allows you to "show additional terms," though you aren't allowed to see the variant spellings, which I find odd. The third image below shows the list of variants searched with "love." (I pause to observe that ProQuest's one-size-fits-all interface also seems to think it can tell me which document is most "relevant" to my search for "love.")

OK, ProQuest, fine. FINE. Setting aside that this searching actually lacks the very granularity and precision that made the old EEBO interface so useful for my purposes, I'll accept this so far. But then, once I locate Fedele and Forunio (ranked #29 in relevance??!!), I want to scroll through the results for "love." Clicking on the promised "full text," I arrive at a page with highlighted search results.

I can, it seems, hide the highlighting. And I can, it seems, scroll through the full text looking for highlighted results. ProQuest seems to think this will suffice, and that if users want to scroll through all the search hits, they can simply use a "search" command in their browser (command + f). But what term do I search for? My original search was for "love." If I search this full-text for "love," I get no hits on this page, BECAUSE "LOVE" WASN'T SPELLED THAT WAY IN 1585.

If you too are a frequent searcher of EEBO, then you'll know why this is a major problem. Rather than a system that aligns the indexical nature of a full-text resource with the variation endemic to early modern English (as the old EEBO interface offered), we have a new system that, in its algorithmic concealment of variation, actually limits the usefulness of the database. To read across instances of words, phrases, or linguistic forms (what Dan Shore calls "plural reading" and what I have called "computational philology"), users must now actually scroll through entire EEBO texts looking for highlighted terms.

Moreover, I have done a detailed comparison of the two interfaces to highlight just how much time the new one will cost (a great irony, considering ProQuest's rhetoric of efficiency). Myself, I have spent the last 18 months collecting thousands of examples from EEBO for my book project. I shudder to think how long it would have taken me to find the same number of examples if I'd had to scroll through one full text at a time looking for highlighted hits.