Different %mor lines for same transcript

11 views
Skip to first unread message

Brielle Stark

unread,
Sep 12, 2018, 9:39:29 AM9/12/18
to chib...@googlegroups.com
Hi all, 

I've attached two identical transcripts, one done with a slightly older clan version and one done with the most recent clan version. Mor was run identically on both. The participant tiers are the same but I notice the mor tier is different in the &uh looks .. line, the penultimate line in the transcript. Since one is tagged as a cop and the other a verb, the two transcripts, albeit identical, present with different propositional density scores. This isn't a problem, just trying to sort out the reason! 

Thank you!

Brie 

--
Brielle Stark, PhD
Assistant Professor, Speech and Hearing Sciences
Indiana University

1045.BW1.secondrater.cha
POLAR_1045_BW1_chat.cha

Brian MacWhinney

unread,
Sep 12, 2018, 9:53:29 AM9/12/18
to ChiBolts, briell...@gmail.com
Dear Brie,
    The verb "looks" is ambiguous between v|look-3S and cop|look-3S.  In this case, the current version of MOR made the wrong choice.  However, we now have added a facility to MOR that uses the prepost.cut file to make a better choice.  The relevant rule chooses the verb listing before a preposition.  So, the most current version of MOR will get this right.
     More generally, tagging for part of speech in English will always make a few errors.  Our goal is to continually reduce the error rate and the prepost.cut facility helps in that regard.

--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/CAEs2yTryB%3DGypWE32qjUj%3DFB0nSy58FzkoG%2BK7cdfX0_o1Babw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.
<1045.BW1.secondrater.cha><POLAR_1045_BW1_chat.cha>

Reply all
Reply to author
Forward
0 new messages