parts of speech question

Brian MacWhinney macw at andrew.cmu.edu
Wed Jul 11 17:23:17 EDT 2018


Dear Leigh Ann,
   Because the analysis through MOR+POST+MEGRASP is based on statistical disambiguation, it will occasionally make mistakes of this type.  Depending on the nature of the corpus and speakers, tagging accuracy can get up to about 97%, but it is almost never above 98%.  In English, the biggest source of tagging error involves noun-verb ambiguities such as that with “crack”.  A first sanity check on this involves typing in “mor +xi” and then entering “crack” and you will see that MOR knows that crack can be both a noun and a verb.  You can force the tagger to get the right answer in this particular case, by transcribing the word as “vxcrack” and this will then force the tagger to treat “crack” as a verb.  The good consequence of this is that it would then also help get the right analysis on the %gra line.  However, in the long run, this is a truly ugly solution.  We are exploring other solutions to the noun-verb ambiguity problem, but they will require some extensions of the POSTMORTEM program that would take a couple of weeks.
     If you just want to get your own data set into the best shape, the most straightforward way of dealing with this would be to just correct the %mor line directly.  

—Brian MacWhinney

> On Jul 11, 2018, at 9:48 AM, Leigh Ann Spell <contact at uscaphasialab.com> wrote:
> 
> In a language sample we are analyzing, CHAT/CLAN is counting “crack” as a noun and not a verb in a description of picture showing a boy breaking a window.  The participant is obviously using it as a verb without the morphological marker “boy crack the window” but CHAT/CLAN seems to be coding that as a noun each time she uses it as a verb.  Any ideas about how to make CHAT/CLAN count this as a verb in the analysis?  Thanks!
> Leigh Ann Spell
> 
> -- 
> Aphasia Lab - University of South Carolina
> 
> -- 
> 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+unsubscribe at googlegroups.com <mailto:chibolts+unsubscribe at googlegroups.com>.
> To post to this group, send email to chibolts at googlegroups.com <mailto:chibolts at googlegroups.com>.
> To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/CAMk2RCAwEF8sihXkoKZ---fhiV5Lc4HnqjbOtz49BfYEux1fLQ%40mail.gmail.com <https://groups.google.com/d/msgid/chibolts/CAMk2RCAwEF8sihXkoKZ---fhiV5Lc4HnqjbOtz49BfYEux1fLQ%40mail.gmail.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.

-- 
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+unsubscribe at googlegroups.com.
To post to this group, send email to chibolts at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/C89E9186-041D-4039-8328-9DF210DF8DA7%40andrew.cmu.edu.
For more options, visit https://groups.google.com/d/optout.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/chibolts/attachments/20180711/0be5dc40/attachment.html>


More information about the Chibolts mailing list