Wikigrammar

Yawning jaguar This time next year, if my maniacal plans succeed and if my suffering students don’t garrotte me with a microphone cable in self-defence, you’ll be looking at the most complete reference grammar of Guaraní ever written in English.

Our university has what may be one of the last two-semester field methods courses left in the world, and it’s probably going to be the last time we offer it that way before we have to chop it down to one semester like everybody else. So I was determined not to waste this last opportunity to do some in-depth work on a language, without getting bitten by mosquitos or having to beg the government for funding.

Of course, life couldn’t let me have it easy this year.

I’ve taught a few field methods courses before, usually with about ten students in them, which is just about the limit of what’s pedagogically good. I vaguely remember having 15 students the year we studied Dakota — “vaguely remember” because it was an experience my mind has been trying hard to forget, except for the part where I vowed never to do that again. (The 15 students, not Dakota.)

Through some horrid confluence of circumstances that will need to be thoroughly investigated in order to make sure it never happens again, the enrolment in this year’s course ended up at 25.

Ten was borderline. Fifteen was traumatic. Twenty-five is so far away from the pedagogically ideal size for a field methods course that you can’t see Earth from there. Fortunately, at the very least, the acting dean came through with some extra money to hire more consultants.

So there’s an upside too. Like my elders always used to say: When life gives you lemons and way too many students and lots of time with speakers of a fascinating language, make lemonade and a factory for writing a reference grammar. (I paraphrase.)

So we’ve got very enthusiastic speakers of Guaraní. We’ve got two dozen unbelievably motivated and talented students working on the language. We’ve got eight months. We’ve got ethics board approval for the whole course, so we can actually publish anything that comes out of it. And we will publish. I think I’ve set things up so that that’s practically unavoidable.

In past years when I’ve taught field methods, one of the ways I tried to get the students to pool their work was by having them photocopy the fieldnotes from their individual sessions for each other. It wasn’t very successful. They were always half-hearted about the copying — doing it very late, if ever, and usually only under threat of losing marks. And I have a sneaking suspicion that I wouldn’t need more than one hand to count the sum total of minutes that every student across all the years ever spent reading another student’s fieldnotes.

Well, what was mostly pointless in previous years would have been totally unworkable this time around. Making 25 copies of all their notes would bankrupt most of the students and kill a small forest. And at least one of the students is blind and wouldn’t find a stack of photocopies terribly helpful.

So instead the students will be sharing their notes and descriptions with each other on a website running MediaWiki, the same system that runs Wikipedia.

I’ve divided them into groups of three, since scheduling only eight extra sessions a week with the speakers is less impossible than scheduling 25. Every week or two, each group will get a topic area they have to try to figure out. (“Group 3, figure out as much as you can about imperatives. Group 4, you’ve got yes/no questions…”) The students have to plan the sentences they want to elicit during their sessions. They have to type their fieldnotes immediately after their sessions and upload them to the wiki. They have to upload the resulting description assignment to the wiki as soon as it’s done.

We haven’t got this far in the course yet, but in my fantasies what happens next is: groups read each other’s descriptions, give feedback, maybe offer counterexamples or even better example sentences that they’ve found in their own sessions, ask questions that the original writers never thought of, spurring them to probe even more deeply during their next session. (You know: all the stuff that I’d be doing myself if there weren’t 25 of them.) After a couple of rounds of revisions and follow-up sessions on their grammatical topic, voilà, we’ve got a solid section of a reference grammar. We can copy it over into our growing draft of the reference grammar, link it to all the other relevant sections, and tinker with it for the rest of the year as the need arises.

I’ve tried to design the wiki so that the things I hold as virtues in grammar writing are dead easy to do, while bad things are nigh on impossible. Every point or argument you make should be backed up by data, with tons of example sentences from texts or your fieldnotes, overtly cross-referenced to the texts or your fieldnotes. In the wiki, displaying a sentence as an example in one of your grammatical descriptions is as simple as using the incantation {{eg|191kr003}}, where 191kr003 is the number that the wiki assigned to that sentence when you uploaded your fieldnotes. Everything else is done automatically — from example numbering, to showing the speaker’s initials and a link to the notes, to interlinear glosses that automatically wrap to the next line if they don’t fit inside the margins, with every small-caps grammatical abbreviation linked to its official definition. It’s beautiful, if I do say so myself. On the other hand, doing sleazy things, like inventing a sentence out of thin air and trying to pass it off as data, is as hard as trying to accomplish all of the formatting mentioned above using only Wikipedia markup — good luck with that.

We’re also recording everything — every class, every group session — and those recordings will end up in the descriptions. Students have to create a Praat textgrid for the recording of their session, marking out the intervals where every sentence from their fieldnotes can be found. My computer automatically chops the marked intervals into little sound files, uploads them to the wiki, and links them to the wiki’s record of that sentence. As a result, every single word and sentence used as an example in our final grammar will have a little icon beside it allowing you to hear the original speaker saying it.[1]

It seems kind of a shame to flatten all of that into a stack of paper at the end of the year. I kind of like the idea of the resulting reference grammar becoming a permanent website — suitably sanitized and de-historicized so that you all won’t get to see how bad our transcriptions are during week 3. I’d also like to make something a little more accessible to the actual speakers of the language in Paraguay than a media-intense English website in another continent. I’m not sure what that should be, but I’m pretty sure it’s not going to be one of those maroon minivan-sized Mouton books that cost more than the average Guaraní speaker earns in a month.

It’s still too early to say how it’s going to work out. A few of the students freaked out at first over all the technical stuff I was expecting them to do. But after their first practice assignment, and halfway through their work after their first real group sessions, I think they’re realizing it’s not impossible. Most of them seem to be getting near the point where the mechanics become second nature and fade into the background, letting them concentrate on the linguistics. At least they haven’t strangled me yet.

So there’s an increasingly good chance we’ll all survive the year with not only our sanity, but also some real research to show for it.[2]

 


  1. I’ve seen a few online grammar sketches that have lots of audio links, but I don’t recall ever seeing one with consistent audio links for every scrap of data. Maybe we’ll be the first.
  2. I wasn’t quite this optimistic a couple of weeks ago. I was pretty exhausted from the summer of hacking PHP, Javascript, CSS, and MediaWiki’s twisted template system,[3] and barely recovered from the month of oxygen deprivation that was my stay at this year’s Linguistics Institute in Colorado. But technical challenges are a lot like recreation to a geek. Far worse were the months of banging my head against a wall trying to solve the bureaucratic problems of getting a wiki running at the university. I’d started pestering Computer Services in May trying to find out how to go about it or even who to ask.[4] I got our Faculty of Arts computer guru involved in the pestering, in the belief that maybe they’d be less likely to ignore him than a mere prof. (Note to self: Computer Services is egalitarian in their ability to ignore.) And here it was, a week after class had already started, we weren’t the slightest bit closer to having a running wiki than we were four months earlier. Amazing Arts computer guru finally slapped Ubuntu onto an old PC, made me the root user on it, and left the beast whirring away permanently on a shelf in his office, plugged into his last free Ethernet port. Not ideal, but it’s working well enough.
  3. Hell, I had to learn PHP to do the bulk of the hacking — and, oh boy, the hideosity of that bastard child of Perl and bad 90s markup was just as agonizing as I thought it was going to be.
  4. I didn’t even need them to do any actual work. I was willing to run the whole MediaWiki system myself out of my home directory on the university’s personal web-page server, as long as they were willing to raise my disk quota to something useful from its current value of approximately 73 bits.

[Creative Commons licensed Wikimedia Commons photo of a yawning jaguar is by Marcus Obal.] Jaguar is one of the words English has borrowed from the Tupi-Guarani family.

This entry was posted in Linguistics and tagged , . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *

*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>