I spent the last month or so working very intensively on the Bardi grammar and getting it ready for submission to the publishers. Now that I have time to breathe a bit, I thought I’d write down some reflections on this process, in case they are useful to others.
First, a little background. This grammar is based in part on my dissertation, which was partly theoretical, partly-historical, partly descriptive. It concentrated on the verb morphology, with historical and synchronic phonology, and a brief sketch of other important features of the language. The reference grammar, however, is a standard reference-grammar type of book, so there was a fair amount of editing and a lot of additional writing required. I don’t know exactly how many extra pages I wrote in the last month to 5 weeks, and I think I probably don’t want to know.
I’ve been working on this grammar for a long time – since about 2001, and I’ve been working on the language (on and off) since 1999. That made some things possible that otherwise wouldn’t have been possible. I had a pretty good idea, for example, about what I needed to work out and what I knew but just needed to write down. I also had a lot of notes and drafts of sections.
I can’t really recommend writing the way I did this, which was to collect data, then write intensively, then collect some more data, then write a few articles, then write really intensively for a month or so just *before* going to the field to get more data! Here are some things that are absolutely essential for writing:
- A place to keep notes about what needs to be included, questions to solve, ideas to test, etc. There’s a huge amount of organisation required for something like this.
- An efficient corpus searching mechanism. I strongly recommend learning GREP or some other regular expressions searching. In the end, I put all my texts into a single text file and searched for examples with TextWrangler. Toolbox has a corpus concordance function but this ended up being more flexible, because I could cut and paste pretty much directly into the LaTeX grammar file.
- A good backup system. I used Dropbox, which was synching on up to four computers, and also emailed the files to myself and my husband several times. I fully approve of paranoia in such circumstances.
- Time-aligned transcripts. I wish I had more time-aligned transcriptions for Bardi. I have a lot for Yan-nhangu, where I started working through Elan from the start, but for Bardi, I was working from a large corpus of texts which had been transcribed on paper to start with. They are not necessarily quite as easy to work with as large textual corpora but having intonational information is invaluable.
The intellectual part of the grammar writing was the easy part. That was helped, I imagine, by having jointly taught a class on Australian languages last semester with Erich Round, where I got to straighten out a bunch of thoughts on a bunch of topics, and to have a bunch of very smart people try to break my analyses. There’s plenty more to say about the language, but I’m reasonably confident that what’s in the grammar doesn’t misrepresent things. It’s incomplete, but I’m reasonably happy with the analysis.
Having said that, this was not an easy grammar to write. It was also a real trip down memory lane. I have a lot of good memories of past fieldwork, but it’s also pretty clear now that there aren’t very many people with really fluent Bardi, which makes writing the grammar quite a responsibility. There are things that I asked the elders who have now passed away, and since some of the other current speakers have different intuitions, there are things that I know about the language on the basis of some badly documented and quite ephemeral evidence. Then there are all the things that I didn’t realise I knew until random questions came up the course of other work. That all needs documenting; I owe it to the elders.