[developers] Morphology 2
bart.cramer at gmail.com
Fri Sep 18 14:30:38 CEST 2009
Not sure whether this mail turns out to be a request for help or a bug
In short, the symptoms are: the morphological rules can be flopped with the
rest of the grammar, but when the .tab file points to a morphological rule,
PET gives a warning (at load time) that that rule doesn't exist. In LKB, all
More specific, I have the following code snippets:
(as you can see, I would like to do the regular morphology via the irregs
mapping as well)
irregs-file := "lexicon_static.tab".
;; status values that marks rules
rule-status-values := rule.
lexrule-status-values := lex-rule.
;; suppress results of regular morphology rules if irregular form exists
:begin :instance :status lex-rule.
%suffix (bla bla)
lex-infl-rule-verb-npd & [ SYNSEM.LOCAL.CAT.HEAD infl-verb-npd-pr-1-sg ].
stehe IR-VERB-NPD-PR-1-SG stehen
The interesting thing is: it seems to flop well. The stderr shows:
(...) including `rules_infl.tdl'...(...)
Also, this is what is present in the .grm file:
[bcramer at malabo cheetah]$ strings german_static.grm | grep -i -C 3 ir-verb
suffix (bla bla)
And this is what cheap says, eventually:
[bcramer at malabo cheetah]$ echo ich stehe | cheap -t german_static
reading `pet/german_static.set'... including `pet/cm.set'...
loading `german_static.grm' (Cheetah (0.13)) [ disabling unification
quickcheck ] [ disabling subsumption quickcheck ] Ignoring entry with
unknown rule `ir-verb-npd-pr-1-sg' in irregular forms
3569 types in 0.65 s
no lexicon entries for:
Did we forget any important setting? Or does this really look like a bug?
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the developers