I agree with you entirely on this. I remember coming across Hylton’s text years ago and finding it valuable at the time but, of course, things have moved on very considerably since then. It’s really a piece of Internet history now — and good that it’s still around *as* a historical document. Most people should, I think, understand that what you’re all doing at MIT these days is light years away from such outmoded projects (granted that they were pathbreaking in their time).

We’ve had some good exchanges on here about  editorial changes and the nature of the text as evidence (and how we might use it).

I’ll be interested to see what new things you’re cooking up at MIT these days!

Andy

It’s very interesting to get your thoughts as someone who is involved in the electronic Variorum. I’d be keen to know, at some stage, how you feel the project might change the way in which we look at all of these issues.

There are several problems. Scholars like Jonathan, of course, need some kind of stable, acceptable dataset that they can work from. Editors once believed that it would be possible to provide something like this. But, of course, it is an illusion: editions simply beget editions, and the beat goes on. Quite apart from the — absolutely spot on, in my view — point that you make about Shakespeare’s mind simply being an unreachable destination, there is also the problem that scholarship moves forward and today’s textual conclusions will fall out of favour tomorrow. Wells & Taylor’s Oxford edition was wonderfully provocative and sparked a great number of valuable and interesting debates, but many of its key textual propositions have not entirely stood the test of time.

What electronic textuality should give us is the possibility of presenting texts that are productively flexible and texts that are adaptable to different kinds of work. This is, really, where we should ideally be headed with digital Shakespeare 2.0.

Anyway: many thanks again, Alan.

Andy

On the other hand, to say it doesn’t matter which text we use (Martin’s “Nameless Shakespeare” argument) also strikes me as the fetishizing of simplicity — using the term “data” to paint over the aspects of texts that aren’t computationally tractable, so that we can promote the complexity of our computational methods instead. (David Golumbia’s got a fascinating new book on what he calls “computationalism” — written by a programmer, too.)

To say that the choice of texts doesn’t matter also risks the fallacy of reasoning from conclusions, in that if the texts did matter, it would mean text analysis scholars have a lot more work ahead of them than they might like. Not everyone working in that field commits this fallacy, but I think it motivates a lot of their resistance to the materialist argument. The materialists probably fall into their own version of this fallacy in some ways, too — it’s a very human response.

As you point out, Andrew, evidence works differently in the humanities than in the quantitative social and natural sciences. The idea of statistical significance is valid in the sciences because one carbon atom is essentially the same as another; it’s usually context that makes the difference. By contrast, literary texts, metaphors, narratives, genre markers, etc. may have family resemblances (as Michael and Jonathan’s  article elegantly demonstrates) but we tend to value them for their particularity. It’s not so much a sonnet as this sonnet. (One might invoke the type/token distinction from linguistics in opposition here, but I’ve read too many Random Cloud articles to be able to think of texts as tokens which matter only in their connection to abstract types.) As Anthony Appiah once said, for the humanities the universal is in the service of the particular. Whatever some may claim for Shakespeare’s universality, the act of choosing to analyze Shakespeare or any other canonical author is deeply particular. I think that’s why the choice of texts matters, and why your fatality argument about the Moby is valid.

But dismissing statistical approaches isn’t valid, either, and we need ways forward. Jonathan and Michael’s article offers one very promising avenue, in that they use statistical methods not as ways of answering questions or verifying qualitative hypotheses — which would get us nowhere — but rather to ask questions and improve them. I see their approach as embracing hermeneutics rather than trying to replace it, as many older approaches in stylistics have done.

Another way forward is to reframe the debate slightly, so that we’re not simply wrangling about which edited text is the most suitable for statistical analyses. Like Martin, I don’t see that debate leading anywhere useful, at least framed in those terms. Rather, the point of facing up to the Moby’s shortcomings is to bridge between formalist and historicist approaches (as others have been doing under the banner of historical formalism). Traditional textual critics would examine variants for the purpose of establishing what Shakespeare actually wrote. Fair enough, but I think Shakespeare’s mind is an unreachable destination. A better reason to talk about the histories of texts is McKenzie’s idea notion of sociology. Many agents, not just the author, are embedded in their histories, and it goes beyond which words editors have changed. It’s not a question of which text is the most authoritative, but whether or not computational approaches close the door on history. If the computationally tractable aspects of texts become the only kind of evidence that matters, we stand to lose a great deal. (There I go arguing from consequences…

On that note, another way forward would then be a hybrid approach that enabled distant readings in combination with historical particularity. Michael and Jonathan’s paper hints at what this kind of hybrid approach looks like. I like Martin’s term “scalable reading” in his comments to their paper — it suggests an interpretive method that doesn’t require me to stop being “textually-minded” (as Andrew puts it) in order to use digital tools. Those of us on the Electronic New Variorum Shakespeare have been working toward such an approach in our visualizations, in which it’s the affordance of zooming in and out from the particular to the general that matters, not staying at one level exclusively (close or distant reading). There’s a lot more work to do, but it’s good that these questions be debated in the open, not swept aside as they’ve been so often in the past.

Let me take up your point about ’150 years of improvement’ or ’150 years of difference’. I think there are a few points to be made here:

1. To what extent do the rules of ‘statistical insignificance’ really apply in this area? Let’s say, for argument sake, that we discovered . . . I don’t know, that half of Ophelia’s speeches should actually be assigned to Gertrude. Statistically, that would be a very small change in the text as a whole, but it might radically alter the way in which we read the text, no? Statistical norms work convincingly in broadly scientific research (or in things, topically, like opinion polls), but they probably work less well in literary work.

2. Yes: we’ve had 150 years of difference, and some of the differences haven’t endured. So, for example, McKerrow and the New Bibliographers had a high degree of faith in their ability to tell the difference between ‘foul papers’ and ‘prompt books’ and a lot of changes have been introduced into the text on that basis. But Werstine and Long have called those very categories into question. The same, of course, is true of Memorial Reconstruction, Hand D and various other discoveries of the New Bibliographers.

However, there have been discoveries that are incontrovertible and that have stood the test of time. The example that I give in the review is a good one I think: the Pavier quartos were all published in 1619, and not on the various false dates given on some of their title pages. And this makes a difference.

It doesn’t help your work of course, given the kind of information processing that you are doing, but you may have seen that I’ve suggested to James Harriman-Smith of Open Shakespeare that one solution to the Globe problem might be effectively to crowd source textual annotations which would provide a sense of how the Globe would look now, in the wake of the textual discoveries that have been made since Clark and Wright’s day. I suspect that textually-minded scholars would be happier with a text of that kind, rather than a straight offering of the Globe.

What I really like about your comment, though, is that it prompts us to think about what we mean by ‘difference’. As you say: an editor might trumpet the fact that his/her edition offers the latest, greatest, most up-to-date text, but, unless we are talking about something like revisionism, how much difference are we seeing, in real, measurable terms? And also, of course, headline-grabbing changes like revisionism, new Shakespeare poems, etc., tend not to stand the test of time anyway.
Andy