The addkey function causes problems if there are
addkeys with spaces in them.
I am using a corpus which includes the tags VS and VE
for two functions of the same verb-form. In a handful
of places this has not been fully disambiguated so the
layout is <w tag="VS VE">. There are also plenty of
instances of <w tag="VS"> and <w tag="VE">.
When I look on the addkey list with "any" ticked, I
don't see VE VS as an entry, but rather multiple
instances both of VE and of VS on the list. However,
when ANY one of these is highlighted, what appears in
the corner of the dialogue is VE VS VE if VS is
selected, and VS VE VS if VE is selected. These 3-tag
strings don't occur anywhere in the corpus. See
attached JPEG.
Any query for any VS or VE causes the client to crash
with a Runtime error! dialog.
(This happens with some other tags pairs too.)
If at all possible, it would be useful if Xaira could
handle these addkeys by treating "VS VE" etc. as a
single unitary addkey disjoint from VE and VS, or by
reckoning inclusively so that a query for VS
finds "VS" and "VS VE", and a query for VE finds "VE"
and "VS VE". This would be handy not just for handling
incompletely disambiguated data, but also for tagsets
where each analysis is designed to be composed of a
set of separate tags.
If this is not possible, then it would be useful if
the indexer gave some alert about the presence of
invalid addkeys.
screenshot of problem