|Texts and Contexts|
Peter S. Baker
For a long time I have found most electronic texts--the ones I know best, at least, Old English texts--to be rather disappointing. It’s wonderfully useful to have texts that can be searched and analyzed in various ways, and hypertext has revolutionized the way we encounter texts; but the text itself has remained pretty much what it’s always been: a stream of letters arranged on page or screen, immutable, its ultimate reality the letter, whether that exists as a binary number in RAM or an inked-in shape on a page. For a long time I’ve been wondering: if the computer is so good at manipulating information, why has the e-text remained so static? The model of the book encourages us to think of the text as something that is, and hypertext encourages us to regard the text in an enriched context. But where is the text that can morph itself--showing us, say, a mix of variant readings? Where is the text that can answer questions about itself? Where is the text that can do as well as be?
About six years ago a few ideas came together for me: I had recently had a visit from an Australian Old English teacher who was putting together a distance-learning course to serve the thinly populated northern region, and I had been reading about a new computer language called Java, which was increasingly being used to deliver programming over the World Wide Web. Java is what’s called an “Object-Oriented” language. You don’t want to hear a technical definition of that term; it’s enough to know that “Object Orientation” is all about erasing the distinction between data (like a text file) and the program that operates on that data. An “Object” is data with behavior, data that does something. So here was the idea: if I could create a text that could be gotten up in a Web browser and could answer questions about itself, that could be the basis for a nifty way to deliver language instruction to students outside the classroom. I learned the Java programming language and set to work. The result is a system that goes under the slightly silly, but, it turns out, rather catchy title “Old English Aerobics.”
This is Genesis 3, one of the texts we often throw at beginning Old English students: it narrates the fall of Adam and Eve. It looks like any text, but the letters you see on the screen are actually graphical representations of what I think of as containers corresponding to grammatical structures such as words and clauses. Each of these containers can hold any kind of information that a computer is able to store. The container for a word, for example, holds the word’s spelling (of course), information about its grammar, such as its part of speech, whether it’s masculine, feminine or neuter, past, present or whatever, and a reference to an entry in an on-line glossary and to the definition in the entry that fits best. To get at that information the student just clicks on a word and chooses from a menu: if we look up the grammar of treowe, we find that it is a common noun, it belongs to what we call the “strong” declension, its gender is neuter, its case is dative, and it is singular. Look up its definition, and we find that it means “tree.” Look up a past-tense verb [forbead] and the glossary will give us a past-tense definition. It can do that because tense is one of the pieces of information stored in the word, which the glossary looks at to decide how to present the definition. The same is true for a comparative adjective [geapre], a plural noun [nytenu], and any other word that can be inflected. Further, when a glossary entry has more than one definition, it can tell us which ones are most appropriate in this place [binnan], because that information is also stored in the word. For clauses, we mainly want to be able to find out the boundaries--another menu item will show us that--and the kind of clause. This text understands only words and clauses, but a revision that I have underway now will also understand phrases, metrical structures, and really any textual feature that one wishes to represent--rhetorical structures, for example. It can already handle fairly arbitrary structures which one can put it for various purposes-making a hyperlink, for example, or having the text read itself aloud to you.
So, how does this go over in the classroom? The real trial will be next year, when I give up the printed textbook I’ve been relying on for twenty years and go almost entirely over to home-grown on-line material, but a preliminary attempt which I made two years ago suggests that students are pretty accepting of this kind of e-text. They do need a printed copy to mark up and carry to class; they will be able to get each text in Portable Document Format and print it out with nice wide margins and plenty of space between lines-a more generous format than you normally get in a textbook. A greater concern for me has been the possible effect of the convenience of the on-line help in a text like this one. In a way, it is the electronic analogue of those editions of Middle English texts with glosses in the margin. They’re easy to read, but you wonder if some students are learning the vocabulary or just reading the glosses. One of my students did in fact just buzz through the text, looking up every word, writing down its definition and calling the result a translation; but that student didn’t do well with a glossary either, and most understood that that method wouldn’t do much to help them learn the language. Next year I’ll be warning my students that they will have to make a transition from e-text to printed text at some point in the term, and, after they’ve done some elementary work, I’ll be giving them e-texts in which many of the common words, which they should have learned, are not glossed. If they’ve forgotten them, they’ll have to look them up the old fashioned way.
This text can communicate with computer programs as well as it can with students. Most of Old English Aerobics consists of exercises like this one [accusative], which presents the student with a text and asks questions about it. To answer the questions correctly, the student has to read the text correctly. What you see in the blue panel is just like the longer text we just looked at, only shorter: we can find out about each word just as we did before [look up Št]. But so can the part of the program that asks the questions: when I wrote this question, which asks the student to click on accusative nouns, I didn’t supply answers, as I’d have to do when building exercises for most language-learning software; I just pointed it towards the text, and it found out the answers for itself. The same is true of most questions, whether they ask the student to click on a word, fill in a blank, or pick from a list. This ability to query the text is a useful feature, because, unlike the natural-language instructions you see on the screen, the question the program reads is in SGML, Standard Generalized Markup Language, and that’s tedious to write. I can take a question I’ve written for one text and apply it easily to another. I don’t even have to know what the answer is; I just have to know an answer is there.
This project is like those web sites with signs that say “always under construction.” The next step will be texts that change in response to student input. Here’s an example. This exercise asks the student to pick the correct subject to go in the blank, agreeing with the highlighted verb. When the student picks the correct answer, it appears in the text: the blank gets filled in. This is very simple stuff, but it may (or so I hope) lead to greater things: ultimately, perhaps, an Old English text that morphs into a translation as the student works with it. That text, I think, would do enough for me.
|TEI markup by Gwen Kern|