new %mor line and the CONNL format

[Brian MacWhinney]

Dear Info-CHILDES,

Over the last week, Davida, Andrew and I retrained the POST program to work with the current version of English MOR.  Once this was done, I ran MOR and POST across all of the files in Clinical-MOR, Eng-NA-MOR, and Eng-UK-MOR.  Now, all of these many corpora have new %mor lines that align properly with current versions of DSS, KidEVAL, and the manual.  The last time I did a full run like this was over a year ago.  In the meantime, there have been changes to MOR that improve the detail of analysis for derivational forms and overall accuracy of tagging.  In addition, the new format aligns more accurately with the MEGRASP program.  We are turning our attention next to retraining MEGRASP for the creation of the %gra grammatical dependency tier for all the corpora with %mor lines.  We expect to have that work done in the next week or so.
   
In addition, we have been reconfiguring the system to allow for the usage of the CONNL format which is now a standard for both morphological and syntactic analysis.  This is being done by the creation of a %cnl tier that abstracts away from the details of the %mor line and creates a single atomic part-of-speech tag for each word.  The %cnl line does not replace the %mor line, because the %mor line has a lot more detailed morphological analysis than the %cnl line.  However, the %cnl line provides a cleaner and general interface for training and operation of dependency parsers.

-- Brian MacWhinney

-- 
You received this message because you are subscribed to the Google Groups "Info-CHILDES" group.
To unsubscribe from this group and stop receiving emails from it, send an email to info-childes+unsubscribe at googlegroups.com.
To post to this group, send email to info-childes at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/info-childes/83AEA7AA-3FF2-4802-9245-E85388FEC264%40cmu.edu.
For more options, visit https://groups.google.com/groups/opt_out.