%gra tier treats comma as subject

35 views
Skip to first unread message

Zhuo Chen

unread,
Jan 24, 2014, 4:22:51 PM1/24/14
to chib...@googlegroups.com
Dear CHIBOLTS,

I found that when a comma is followed by a verb, %gra tier consistently tags the comma as a subject. 

I am using the most updated Valian corpus. I have attached a picture of a sentence tagging. As you can see, the comma is tagged as a subject.

Thanks,
Zhuo
wrong_tagging.tiff

Brian MacWhinney

unread,
Jan 25, 2014, 5:17:38 PM1/25/14
to ChiBolts, Zhuo Chen
Dear Zhuo,

    I definitely see the problem.  It will take some work to repair this, because I will need to retrain the MEGRASP parser.  Please give me about a week.  Thanks for reporting this problem.

Best regards,

—Brian MacWhinney

--
You received this message because you are subscribed to the Google Groups "chibolts" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chibolts+u...@googlegroups.com.
To post to this group, send email to chib...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/319000ef-3ba9-4809-878f-81aaf3652d63%40googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
<wrong_tagging.tiff>

Zhuo Chen

unread,
Mar 19, 2014, 3:09:16 PM3/19/14
to chib...@googlegroups.com, Zhuo Chen
Dear Chibolts,

How is the retraining going?

Best,
Zhuo

Brian MacWhinney

unread,
Mar 20, 2014, 10:15:57 AM3/20/14
to ChiBolts
Last week, I finished retraining and created a new %gra line for all the English files.  

— Brian MacWhinney

Zhuo Chen

unread,
Mar 20, 2014, 11:01:10 AM3/20/14
to chib...@googlegroups.com
Hooray!
Thank you!

Marwa R.

unread,
Mar 21, 2014, 1:51:32 PM3/21/14
to chib...@googlegroups.com
This is great news! I looked online and I think the files on the browsable database has been updated, but not the ones under 'Transcripts' (http://childes.psy.cmu.edu/data/). Would it be possible to update these files for the  English folder as well? Also, where can one access the recently retrained parser (MEGRASP)? It would help me a lot to get those updated files and the most recently trained parser.

Thanks in advance!
Marwa R.
Indiana University


On Thursday, March 20, 2014 10:15:57 AM UTC-4, Brian MacWhinney wrote:

Zhuo Chen

unread,
Mar 21, 2014, 2:10:09 PM3/21/14
to chib...@googlegroups.com
Same question here. 

Thanks,
Zhuo

Brian MacWhinney

unread,
Mar 21, 2014, 2:23:10 PM3/21/14
to ChiBolts, Marwa R.
It’s fixed now.  So many links to update!

—Brian

Zhuo Chen

unread,
Mar 21, 2014, 7:49:55 PM3/21/14
to chib...@googlegroups.com, Marwa R.
Thank you very much!!!

Zhuo Chen

unread,
Apr 7, 2014, 1:00:32 PM4/7/14
to chib...@googlegroups.com, Marwa R.
Dear Brian,

Have you updated the files in Eng-UK-MOR? I found that files in it still have %xgra instead of %gra.

Thanks,
Zhuo 


On Friday, March 21, 2014 2:23:10 PM UTC-4, Brian MacWhinney wrote:

Brian MacWhinney

unread,
Apr 7, 2014, 1:21:25 PM4/7/14
to ChiBolts, Zhuo Chen, Marwa R.
Yes, I stopped at the North American data.  I am making some minor changes to MOR right now.  After that, I will run MOR, POST and GRASP again on the UK files.  Hopefully two more days.

—Brian

Reply all
Reply to author
Forward
0 new messages