MLU bug

[Leonid Spektor]

 Yuriko,

	New version of mlu works correctly now when +t*... option used. If you want to run analyses on main speaker tiers instead of %mor tiers, then use "-t%mor" option as you would have with previous version of mlu. I am attaching a sample output below to demonstrate that.


You can also tell which tiers are used for analyses by looking at comment(s) between lines:

    "mlu (....) is conducting analyses on:"

and 

    "****************************************". 

In the first example, which analyses %mor tiers, the comment is:   ONLY dependent tiers matching: %MOR;
In the second example, which analyses main speaker tiers, it says: ONLY speaker main tiers matching: *CHI;


> mlu +t*chi adam01.cha
mlu +t*chi adam01.cha
Wed Sep 29 12:39:08 2010
mlu (19-Sep-2010) is conducting analyses on:
  ONLY dependent tiers matching: %MOR;
****************************************
>From file <adam01.cha>
MLU for Speaker: *CHI
  MLU (xxx and yyy are EXCLUDED from the utterance and morpheme counts):
	Number of: utterances = 1232, morphemes = 2644
	Ratio of morphemes over utterances = 2.146
	Standard deviation = 1.067



> mlu +t*chi -t%mor adam01.cha
mlu +t*chi -t%mor adam01.cha
Wed Sep 29 12:39:20 2010
mlu (19-Sep-2010) is conducting analyses on:
  ONLY speaker main tiers matching: *CHI;
****************************************
>From file <adam01.cha>
MLU for Speaker: *CHI
  MLU (xxx and yyy are EXCLUDED from the utterance and morpheme counts):
	Number of: utterances = 1232, morphemes = 2582
	Ratio of morphemes over utterances = 2.096
	Standard deviation = 1.024



Leonid.





On Sep 29, 2010, at 12:15, Yuriko Oshima-Takane wrote:

> Hi Brian,
> 
> Does this mean that we won't be able to ask students to run MLU on the main
> tiers with new version 
> anymore even though we use "legacy" on @Languages headerline?  I ask
> students in my class to morphemicize a sample transcript and run MLU on main
> tiers before they learn how to use MOR (and then calculate MLUs on %mor).
> Students seem to understand better what MOR and MLU program do by trying out
> some morphemicization by themselves.
> 
> Yuriko 
> 
> -----Original Message-----
> From: chibolts at googlegroups.com [mailto:chibolts at googlegroups.com] On Behalf
> Of Brian MacWhinney
> Sent: Wednesday, September 29, 2010 10:46 AM
> To: ChiBolts
> Subject: MLU bug
> 
> Dear ChiBolts,
> 
>   Leonid says that the problem that Bobbi Corrigan discovered was not due
> to the change in the treatment of filled pauses, but to a bug in the older
> version of MLU.  In the older version, if you use the +t option to specify a
> particular speaker to run analyzes on, then the old version of mlu was
> computing MLU off the main speaker tier. By default, without the use of any
> +t option, the old version of MLU was working correctly and was counting on
> %mor tier.  So, the bug has been fixed, but it is a shame we didn't post a
> notice on this.
>   However, it was also good to have a chance to explain the current
> treatment of filled pauses on the %mor line to bring them into proper accord
> with Brown's guidelines.
> 
> -- Brian MacW
> 
> -- 
> You received this message because you are subscribed to the Google Groups
> "chibolts" group.
> To post to this group, send email to chibolts at googlegroups.com.
> To unsubscribe from this group, send email to
> chibolts+unsubscribe at googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/chibolts?hl=en.
> 
> 
> 
> 
> -- 
> You received this message because you are subscribed to the Google Groups "chibolts" group.
> To post to this group, send email to chibolts at googlegroups.com.
> To unsubscribe from this group, send email to chibolts+unsubscribe at googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/chibolts?hl=en.
> 
> 

-- 
You received this message because you are subscribed to the Google Groups "chibolts" group.
To post to this group, send email to chibolts at googlegroups.com.
To unsubscribe from this group, send email to chibolts+unsubscribe at googlegroups.com.
For more options, visit this group at http://groups.google.com/group/chibolts?hl=en.