CLB analysis lost after ABBR in grammar checker

[snomos]

echo kvæð. | ./tools/grammarcheckers/modes/trace-faogram-release3-mwe-split.mode 
"<kvæð.>"
	"kvæð" ABBR Gram/IAbbr N Abbr <W:0.0>
;	"." CLB <W:0.0> "<.>"
;		"kvæð" ABBR Gram/IAbbr N Abbr <W:0.0> "<kvæð>" REMOVE:1709:longest-match
;	"." CLB <W:0.0> "<.>"
;		"kvæða" V Imp Sg <W:0.0> "<kvæð>" REMOVE:1709:longest-match
:\n

The correct analysis should have been the second one, which would have given a separate .+CLB analysis after cg-mwesplit.

Please see SME and other Sami Languages for working examples.

[Trondtr]

The output is still the same, and the "Please see SME..." is a bit too cryptic for us not-so-mwe-versed. Where should I look, @snomos ?

[snomos]

In the tools/tokenisers/mwe-dis.cg3 file. It is a disambiguation problem: is the CLB full stop part of a lemma (e.g. an abbreviation) that does NOT end a sentence, or is it a regular sentence delimiter, as it is in this case. The mwe-dis.cg3 file is the place to do this kind of analysis, and when done, the next step (mwe-split) will reformat the cohort stream to reflect the disambiguated tokenisation.