Requires linguist check: lib/constants/MD-pos.json

[fbanados]

The file MD-pos.json contains heuristics to guess the POS annotation for MD entries. However, they may not be as precise as they can. e.j. yahaw and ahpô used to map to Part, while they could map to IPJ and IPC, respectively. I've made some candidate changes, but I believe a linguist should check (and perhaps extend) this set of mappings to get more precise info in the importjson.

[aarppe]

I have a fuzzy recollection of creating this equivalence file ages ago - the problem is that wrt FST it is underspecified, and we might need to add in the actual lexical (inflectional) classes manually anyhow, in the MD_LexCat field that we have for MD.

In this and other respects, I've been tweaking a script that tries to copy over the stem and lexical category fields from CW to MD, always when available, so that we are not reliant on Arok's CW having that information (e.g. for generating FSTs), if he would happen to change his mind. The script for that is:
crk/bin/add-cw_lemma-stem-and-lexcat-from-cw-2-md.sh and the resultant file: crk/generated/Maskwacis_altlab2.tsv - the latter will need some manual editing before replacing the ALTLab version of MD in crk/dicts/.

[fbanados]

An alternative would be to parse the POS directly from the FST analysis that is stored in MD_LexCat when available.