Utterance-level information
16 views
Skip to first unread message
Bridget Rennard
Sep 1, 2021, 4:28:44 PM9/1/21
to chibolts
Hello,
Do you know if there is a command in CLAN to calculate utterance-level information? For example, instead of calculating the MLU for a speaker for the entire transcription, looking at the number of morphemes in each utterance line in the transcription.
Thank you!
Brian Macwhinney
Sep 1, 2021, 5:20:14 PM9/1/21
to ChiBolts
Dear Bridget,
The answer to this depends a lot on which of the 25 or so CLAN commands you wish to use. It also depends on what you mean by “utterance-level”. In the case of MLU, perhaps you mean that you want to compute MLU on the basis of the main tier, rather than the %mor tier. Perhaps the transcript you have only has a main tier. In that case, you would add the -t%mor code to work on the main tier only. Or perhaps you really want each turn separately, and then you would use the -u switch. Or perhaps you just want to merge across all speakers. In that case you would use the +t* switch. If you are working with some other program, there would be still other possible configurations and ways of interpreting what you might want.
Best,
— Brian MacWhinney
Teresa Heinz Professor of Cognitive Psychology,
Language Technologies and Modern Languages, CMU
> --
> 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 on the web visit https://groups.google.com/d/msgid/chibolts/ba46bc74-406e-4e82-b07c-a9c9ff9c27dbn%40googlegroups.com.
The answer to this depends a lot on which of the 25 or so CLAN commands you wish to use. It also depends on what you mean by “utterance-level”. In the case of MLU, perhaps you mean that you want to compute MLU on the basis of the main tier, rather than the %mor tier. Perhaps the transcript you have only has a main tier. In that case, you would add the -t%mor code to work on the main tier only. Or perhaps you really want each turn separately, and then you would use the -u switch. Or perhaps you just want to merge across all speakers. In that case you would use the +t* switch. If you are working with some other program, there would be still other possible configurations and ways of interpreting what you might want.
Best,
— Brian MacWhinney
Teresa Heinz Professor of Cognitive Psychology,
Language Technologies and Modern Languages, CMU
> 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 on the web visit https://groups.google.com/d/msgid/chibolts/ba46bc74-406e-4e82-b07c-a9c9ff9c27dbn%40googlegroups.com.
Bridget Rennard
Sep 15, 2021, 4:45:42 PM9/15/21
to chibolts
Hi,
Thank you for the response! To clarify, I'm looking at MLU and/or number of morphemes per line for the two speakers (Mom and child) that are in the transcriptions I'm working with. The conversational turn option with the -u switch gave me some interesting information, but I am hoping to get more specific information for each line apart from conversational turn. For example, something like this:
*MOM: what is that ? (3 morphemes)
*CHI: ball . (1 morpheme)
Does that make sense? Based on the codes you described I'm not sure if any of them would give this type of information.
Thank you,
Bridget
Reply all
Reply to author
Forward
0 new messages