Question about marking accuracy of past tense use

8 views
Skip to first unread message

Nicole Tracy-Ventura

unread,
Jun 19, 2015, 10:45:17 AM6/19/15
to chibolts
Dear all,

I want to analyze past tense use and plan to code my transcripts for errors but also correct use to quickly calculate accuracy.  I saw in the CHAT manual on pp. 103-104 that there are codes for morphological errors that I could use. Any thoughts on a code to use if the verb is produced correctly or any other suggestions of how best to do this? 

Thank you in advance for any help!

Nicole Tracy-Ventura
University of South Florida 

Brian MacWhinney

unread,
Jun 19, 2015, 11:29:00 AM6/19/15
to chib...@googlegroups.com
Dear Nicole,

     If you have created a good %mor line using MOR, you can run FREQ on the %mor line in this form 

freq +t%mor +s"%|%PAST*" *cha +u +t*CHI

Then you would do a search for errors coded on the main line as past tense errors and subtract.
    
Note that you would have to code both errors of omission and commission.  Also, you may want to change the % to * in the +s switch if you want to see things broken out.

—Brian

--
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/CA%2B3CKJ6KgGaw0%2BY0X226TAejHU_zc8p0E-2adNGbQP22Cmcbkw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

Nicole Tracy-Ventura

unread,
Jun 22, 2015, 10:16:10 AM6/22/15
to chibolts
Dear Brian,

Thank you for the suggestion. We will give that a try since we have run MOR on the files.

Best,
Nicole



Leonid Spektor

unread,
Jun 22, 2015, 11:23:55 AM6/22/15
to chib...@googlegroups.com
Nicole,

There is, also, a more detailed way to search for verbs and error codes associated with them. FREQ has a +sm option that allows this. If you type "freq +sm" in commands window and press return key, then you will get a short listing of all options, short explanation of their meaning and some example of hot to use it. It is also covered in CLAN manual under chapter for FREQ command. Here few examples pertaining to your goal:

To find all past tense words with any kind of error:
    freq  +u +t*CHI +sm-PAST,** +sm&PAST,** *cha

To find all past tense words with only, for example, "neg" error, as in "verb [* neg]":
    freq  +u +t*CHI +sm-PAST,*neg +sm&PAST,*neg *cha

The errors are coded on main speaker tier right after a specific word they refer to.

Leonid.



Nicole Tracy-Ventura

unread,
Jun 22, 2015, 7:41:18 PM6/22/15
to chib...@googlegroups.com
Thank you, Leonid.  We'll look into this option as well.

Nicole


Reply all
Reply to author
Forward
0 new messages