@rgriscom and @SarahDopierala pointed this one out to me today in our writing group:
There are a lot of useful ideas here about an interlinked web grammar, where the texts, dictionary, and grammatical information are interlinked.
This is really well done, even if it’s not entirely complete. I have to motor, just now, but I have set this post to “wiki” mode so anyone can edit it. Maybe we can collaborate to make some observations about the site.
Also hey, I didn’t realize that @SarahDopierala contributed to this project! Mad props!
I like how linguistic content in the grammar is linked to the lexicon (and in the future perhaps also the texts). Here, for example, clicking on the word té ‘exist’ opens the lexical item in the dictionary:
There also appear to be plans for a database export function (as TEI):
And a mockup for a search function:
More cool stuff:
A built-in citation system. Notice that the reference in the URL is quite specific, to §3.3:
There is embedded audio here (awesome!), but it takes a couple clicks to get to:
First you click more and then audio. It seems to me this would be worth surfacing at the top level, because it is a great addition to the corpus. On the other hand, this is clearly an in-development feature because the other toggleable options aren’t implemented yet.
Search User Interface Design
A couple more thoughts on the search interface that @rgriscom mentioned:
It’s easy to look at something like this at a first glance and complain <( but it doesn’t work! ) but let’s appreciate what is actually here: a lot of work in designing the search interface. This is an area where we desperately need more work in language documentation, because it’s very rare to see user search interfaces that can work across a whole corpus with specificity. Designing such a search interface is a non-trivial part of the process, because even if you have the fanciest search capabilities in the world, if users can’t figure out how to use it…
It’s interesting to go through to drop-downs here and see what kinds of searches Jonas is imagining. It looks like the idea would be to be able to search at pretty much every level of analysis over subsets of the corpus. Of these, search phone versus phoneme strike me as the trickiest to implement, because as strings, they are often identical. Perhaps representing everything with a phonemic as well as phonetic tier could make this possible.
Another cool feature: more than one orthography, e.g., ojíbẹ̀rẹ̀ vs odʒíbɛ̀rɛ̀. The orthographic description suggests that these orthographies are one-to-one — I wonder if if the variant spellings are produced automatically?