I have been playing around today with the latest tool from the SIL– it’s called fieldworks and it’s meant to be a replacement, ultimately, for Toolbox/shoebox. It’s a suite of tools for analysing language and anthropology. The language explorer has all of the functions that Toolbox has at the moment. That is, there’s a way to do interlinearisation, and the way to interface text preparation with dictionary preparation. It has live update in the interlinear section — that is, if you make a change in the gloss section, it’s reflected in the interlinear output. In Toolbox, of course, you have to reinterlinearise. There’s a way to import databases from Toolbox, although it’s very clunky. I lost a heap of glosses in the test i did. There doesn’t seem to be a way to import texts at this stage.
The dictionary tools are somewhat similar to Toolbox at present: that is, it’s roughly the same model of dictionary preparation. Data integrity is enforced much more strictly in the fieldworks suite, though, which I think is a good thing. I think it would be very difficult to you end up putting something in a scientific field because you wanted italics for example.
There’s also a way of keeping detailed grammar notes within the same file, which I think is also an advantage over Toolbox. It’s easier to make notes and crossreference different types of information in this platform.
I haven’t really looked at the anthropological sections of the software yet, but it looks superficially pretty similar to the anthropological notetaking structure that was one of the sample files in Toolbox. That is, there’s a list of topics as a suggestion for a notetaking guide.
Another really big advantage of the software is that it can be used by more than one person at once — that is, it can be run over a server. This already makes it a much better option than Toolbox for using in field methods class.
Now onto the things that I don’t like about the software. To be fair to SIL, this is a beta release and it’s clear that there are plans to fix some of these issues. On the other hand, there was never a version of Toolbox released without fairly major bugs of one sort or another. It took me approximately 15 minutes to crash this program. I can’t remember what I was doing but it wasn’t anything extraordinary. In general, the program is incredibly memory intensive and takes up about 1 GB of hard drive space. It also runs rather slowly. This makes it rather frustrating to use.
You can’t link audio or pictures to items in the lexicon, it appears. Perhaps that will change in future updates. I would hope so: linking audio to texts and lexical items would appear to be a pretty basic function these days.
Next, SIL seemed to have continued to advocate that particular dictionary model that is hardwired into Toolbox. That is, a model which places all of the information collected in the entry for a particular Language head word, and not in an English word. This makes it very difficult to create anything other than Language to English dictionaries. Such dictionaries are not optimal for endangered languages, when most of the looking up is done from English to Language, and not the other way around. I would have thought that this could be something that’s changed in a program like fieldworks, which is more dynamic than Toolbox. (By the way, fieldworks is based on XML.)
Another thing that I think it is a big setback in this program compared to Toolbox is how user-friendly it is. To be sure, it’s user unfriendly in a different way from Toolbox: Toolbox takes quite a bit of work to set up, and the structure is somewhat fragile so if you’re in a we are doing it’s quite easy to do something that makes it difficult to work. The fieldworks program is much easier to set up and data entry is much more robust, but I think it requires a much greater level of linguistic sophistication than Toolbox does to get started. For example, in a new dictionary entry you’re presented with a heap of information and categories to choose from. And sure, you can say that you don’t know for any of the particular pieces of information, but I can imagine this being a major hurdle. I received an e-mail recently which, amongst other things, made the point that the software that linguists use shuts out aboriginal people from writing their own dictionaries. There is so much linguistic terminology built in to the software that it requires a great deal of training in order to use it. Now, in this particular case I have disagreed very strongly with the person who wrote this, because I knew in this case that the relevant people did have training and you what they were doing. Moreover, the where the dictionary database was set up was quite user-friendly compared to many. But I think this will be a real issue with fieldworks. It’s designed for a very specific model of working: that of the linguist being in charge and working with another language assistants. I think it’s a mistake to build software that’s less accessible, rather than more accessible.
Furthermore, at this stage some of the defaults seem rather unintuitive. For example, gloss entry fields do not automatically appear in items in the dictionary whether glosses are not filled in.
Finally, fieldworks seems to be presenting a type of language description by numbers. The program comes with a large number of predefined categories which you can just fill in. But they are structured in such a way that I can imagine them being very misleading to someone who didn’t really understand either a) what categories are, or b) how the language really works. And there are a few things which are just wrong. Stative is not a tense!
One more thing: one of the huge advantages of Toolbox was that it did not require large numbers of mouse clicks in order to enter data. It didn’t have drop-down menus and lots of dialog boxes. The new software has considerably more of that which makes data entry quite a bit more time-consuming. I hope in future versions it will be possible to customise keyboard shortcuts or specify the default fields that show up in data entry in the lexicon. Requiring drop-down menus and mouse clicks makes the software less accessible too.