macw at cmu.edu
Fri Nov 15 19:56:20 UTC 2002
It appears that the new version of CLAN solved both of Christian's
problems (FREQ crash and capitals inside French words). However, Christian
also called my attention to another issue. This is the fact that, with the
advent of XML, the 00depadd.cut facility is no longer a possible way to
extend CHAT. This is a tough issue. I really like the idea of user
customizability. However, the idea underlying the concept of XML is that,
with a well articulated Schema, data are understandable by many programs.
The idea of 00depadd.cut goes against this philosophy. However, it turns
out that the things that people are doing with 00depadd.cut can be
accommodated in the new XML world in three ways:
1. User-defined main tiers can be simply entered as @Comment. So the field
@Medical Treatment: placebo
Can be changed to
@Comment: Medical treatment is placebo
I have made many changes of this type already in the database. No
information is lost, although it is not quite as pretty.
2. User-definied coding tiers can be entered using names such as %co1,
%co2, %co3. This works OK now with the new version of CHECK, but it is
clearly much uglier. A better approach is to add new tiers to the CHAT
Schema. To do this, it is best to discuss these with me. If they seem of
general interest to other researchers, we should add them to the Schema.
3. User-defined special form markers. I think it was a bad idea to allow
for these. If there are special forms that were not properly defined, we
should add them to the Schema.
So, in general, if the current Schema is not matching your needs and if the
above solutions do not work, please contact me directly and I will try to
expand the Schema. In any case, we can no longer use 00depadd.cut with XML.
Please note that the files currently in the database all pass CHECK and also
work fine in XML without the presence of any 00depadd.cut files.
More information about the Chibolts