I am trying generation of this sentence in LKB from its tree:
tassa-u-vuq iglu
that.is-COP-IND.3SG house
`It is a house.`
And I get the following:
tassa iglu is also correct, but what surprises me is the -vuq iglu. I have checked that it uses exactly the same rules as tassa-u-vuq iglu:
I have also checked that directly typing -vuq iglu in “Parse > Parse Input …” in LKB does not yield any results, so it looks like somehow some parts related to STEM may not be working correctly only in generation.
Is this related to token mapping? For anyone who might be interested in checking it themselves, the version of grammar can be downloaded using the following:
P.S. There is sometimes a possibility that the newlines of the files in the repository are contaminated with dos-style newlines; in that case, please let me know.