Extensional (provisional) definition of "phenomenon"
Things to look at:
- Grammarian generated lists
- CSLI test suite classification scheme
- TSNLP (too small)
- GOLD ontology
How many levels of classification?
use cases
- Cross-grammar comparison (difficult, requires defensible lists)
- Grammar exploration (more lenient) -- even just having the lists will be interesting
- Constructing list of phenomena for phenomenon labels on constraints
One way to manage the granularity issue is to cap the number of phenomena that can be listed.
"cover a phenomenon": doesn't entail that you cover everything about it (disclaimer at the top)
Task:
- Each grammar can do up to 100 phenomenon labels for this first step.
- Attempt to capture all that is interesting in those 100 phenomenon labels rather than just going to 100 and stopping.
- Use non-language-specific terminology if possible; if reference to a morpheme of the language is unavoidable, provide a one-sentence explanation of the phenomenon.
- Don't lump distinct phenomena into a single label based solely on shared morphemes.
- Include one-two examples sentences that illustrate the phenomenon
- If you have time, please provide translations into English for the examples.
- If you are enthusiastic, consider adding IGT.
Motivation:
Antske --- keeping track of what's in each release, giving organized view of what's covered without having to go into detail (would make dissertation too long), want it for other grammars.
Dan --- altruistic only; eventual possibility of tying these labels to objects in the grammar would be interesting but we're not doing that yet.
Getting it to happen:
Emily creates list for Wambaya and sends it around with invitation for others to contribute/post same by some specific time. Use CSLI file from Dan as inspiration and also send it along. Do this soon so that it doesn't seem too big of a task. [Make sure Wambaya demo is actually running.] Put on DELPH-IN wiki in a similar structure to the MatrixMrsTestSuite pages. Link from grammar catalogue too.