Excluding utterances on PAR line

[brielle.stark]

Hi there. I'm trying to exclude utterances on the PAR line tagged with + exc. According to the manual, and to the eval command, I've been doing:

eval +t*PAR -s"[+ exc]" +u *.cha

Am I doing something incorrect? Feedback from CLAN:

Please specify only precodes "[- ...]" with +/-s option.

Thank you very much!

Brie

[Leonid Spektor]

Brie,

EVAL, KIDEVAL and few other commands automatically excludes utterances marked with "[+ exc]" post-code. You do not need to specify it on the command line. The reason you got an error message is because we do not want post-codes used to include or exclude any other utterances by mistake.

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 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/a221cf94-1a72-43b4-97b6-97c9ca4a828e%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[brielle.stark]

Perfect, thanks!

[brielle.stark]

Do I need to specify to exclude [+ exc] with 'freq'?

[Leonid Spektor]

Yes, you need to specify post-codes to make FREQ exclude or include utterances with that post-code.

Leonid.

To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/6d9619d2-8838-4e92-a35b-cf7927683318%40googlegroups.com.

[brielle.stark]

Thanks. I tried as the FREQ command said, with;

freq +d2 *.cha  -s"<+ exc>" -s"<+ per>" -s"<+ jargon>" +t*PAR

eval +t*PAR +u &.cha

when I did this, I got different utterance counts, which makes me think that the FREQ exclusion did not work. 

I appreciate your help, thanks.

[Leonid Spektor]

Post-codes are an exception to regular codes. If you want to exclude data associated with post-codes, then you need these options:

-s"[+ exc]" -s"[+ per]" -s"[+ jargon]"

if you want to exclude post-code's codes themselves, i.e. "[+ exc]", "[+ per]" and "[+ jargon]", which is redundant, because they are excluded by default anyway, then you need  these options:

-s"<+ exc>" -s"<+ per>" -s"<+ jargon>"

Post-codes and pre-codes are treated differently than regular codes. If you are not sure, then I would suggest to just type a command, like FREQ, by itself in commands window and you will get quick description of what those options do. For example, if you type FREQ, then under line "+sS: search for word S in an input file." you will see description of differences between codes and post-codes.

Leonid.

To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/c1acd1b9-38bb-432a-aeee-334807a7fd19%40googlegroups.com.

[brielle.stark]

I'm specifically referencing the *PAR column in the FREQ output as being different from the "Total Utt" column in the Eval. Essentially, the *PAR column in the FREQ output is always bigger than the "Total Utt" column, hence why i was thinking the +exc tiers weren't being excluded.

Thank you for your help.

Brie

[Leonid Spektor]

Brie,

Please email to me directly a sample file along with command lines you are using for FREQ and EVAL that demonstrate this discrepancy and I will try to figure out what is wrong.

Leonid~

To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/ff993112-0e4e-4fa8-bde9-7d8651272257%40googlegroups.com.

[Leonid Spektor]

Brie,

It looks like FREQ counted all speaker tiers in the transcript file and not just the ones that were selected by user. Counting speaker tiers present in the file is not part of FREQ analyses, because tier names are not included in frequency counts or TTR. This function was added to FREQ just for the purpose of counting all speaker tiers in the file. I have changed FREQ to only count speaker tiers that are selected by user to make sure that output results are more consistent with other CLAN commands. I will install new CLAN on the web later today.

Leonid.