differences in MLU between browser and CLAN

[Jennifer Ganger]

Hello again,

Some of my students pointed out today that they are getting different MLU results when they run it within the browser versus in CLAN. The effect seems to be widespread--not just one corpus. They noticed discrepancies with the Tardif corpus at first but then found more.

Taking Eve (Brown corpus) file 020000a in the browser as an example, the command

mlu +t*CHI  020000a.cha yields:

From file <childes/Eng-NA/Brown/Eve/020000a.cha> MLU for Speaker: *CHI: MLU (xxx, yyy and www are EXCLUDED from the utterance and morpheme counts): 

Number of: utterances = 424, morphemes = 3687 

Ratio of morphemes over utterances = 8.696 

Standard deviation = 5.953

That can't be correct. 

In downloaded transcripts using CLAN, the same command yields:

From file <C:\talkbank\clan\Brown\Eve\020000a.cha>

MLU for Speaker: *CHI:
  MLU (xxx, yyy and www are EXCLUDED from the utterance and morpheme counts):
Number of: utterances = 424, morphemes = 1468
Ratio of morphemes over utterances = 3.462
Standard deviation = 1.975

Any advice would be appreciated.

Thanks,

Jenny

[Leonid Spektor]

Hi Jenny,

Browser CLAN is over 5 years old. It is not longer compatible with the data format. Until people in charge of the web things here at CMU update it to the latest version, please tell your students to rely on downloaded transcripts using CLAN only for accurate result. 

Leonid.

--
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 view this discussion visit https://groups.google.com/d/msgid/chibolts/cafe4c39-c9f5-44d6-aae3-3d547b810828n%40googlegroups.com.

[Jennifer Ganger]

Ok, thanks. 

On Nov 18, 2025, at 5:25 PM, Leonid Spektor <spe...@andrew.cmu.edu> wrote:

Hi Jenny,

You received this message because you are subscribed to a topic in the Google Groups "chibolts" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/chibolts/PYaO-3L2CPo/unsubscribe.
To unsubscribe from this group and all its topics, send an email to chibolts+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/chibolts/294F4F51-1045-4F50-A70C-4798DAFA2D84%40andrew.cmu.edu.