search

UAlbertaALTLab / plains-cree-fsts

Mirror of the source code for the Plains Cree morphological analyzer/generator.
https://github.com/giellalt/lang-crk
Other
2 stars 1 forks source link

Descriptive analyzer should accept "ni-ki-pimohtanan" #23

Open eddieantonio opened 4 years ago

eddieantonio commented 4 years ago

It should have these results:

ni-ki-pimohtanan    pimotêw+V+TA+Ind+Prt+1Pl+3SgO   0.000000
ni-ki-pimohtanan    pimohtêw+V+AI+Ind+Prt+1Pl   0.000000
aarppe commented 4 years ago

Using the current FST machinery and making use of a new morpheme boundary after the person prefixes, we would want to invoke the +Err/Orth feature when a hyphen is used at that juncture. So, the analysis results would be rather:

ni-ki-pimohtanan    pimotêw+V+TA+Ind+Prt+1Pl+3SgO+Err/Orth  0.000000
ni-ki-pimohtanan    pimohtêw+V+AI+Ind+Prt+1Pl+Err/Orth  0.000000

This means that the above spellings are recognized by the descriptive analyzer but not the normative analyzer, and also that they are not generated (by the normative generator).