Bilingual MOR not running in CLANc

[Rachel Romeo]

Hi all,

I am working with a bilingual English & Spanish corpora, and I'm using the newest version of CLANc (May 2022). When I run MOR on the second language, it runs fine with no errors, but the output files are not showing the %mor and %gra tiers for utterances that are fully in the L2 (e.g., lines with [- spa] for an English-dominant file). Additionally, individual words tagged with @s are getting tagged as expected, but it seems to not be happy with full utterances in the L2. 

For clarity's sake:

I've got my mor grammar set to spa. Command:

mor +s"[- eng]" EnglishDominantFile.cha

Will give me output like this:

*MOT: [- spa] no pero no pegas +//.

######No dependent tiers given

*MOT: mira@s this is a little piggie .
%mor: L2|mira pro:dem|this cop|be&3S det:art|a adj|little n|pig-DIM .
%gra: 1|3|LINK 2|3|SUBJ 3|0|ROOT 4|6|DET 5|6|MOD 6|3|PRED 7|3|PUNCT

######seems to do fine with individual words

Am I doing something wrong? 

Thanks!

Rachel 

--

Rachel R. Romeo, PhD, CCC-SLP

Assistant Professor 

Department of Human Development and Quantitative Methodology

Department of Hearing and Speech Sciences, by courtesy

Program in Neuroscience and Cognitive Science

University of Maryland College Park

education.umd.edu/leadlab

Phone: 301-405-2809

Pronouns: she/her/hers

[Leonid Spektor]

Hi Rachel,

I think the problem is with +s option. The +s"[- eng]" option tells MOR to run only on utterances that have "[- eng]" pre-code.

If your file is English-dominant, then to run MOR on only English utterances you need an option -s"[- spa]", if you want MOR to run only on Spanish utterances, then you need an option +s"[- spa]". In English-dominant data file there should not be any "[- eng]" pre-codes. If your data file has both "[- eng]" and "[- spa]" pre-codes, then +s"[- spa]" option will tell MOR to run only on Spanish utterances and +s"[- eng]" will tell MOR to run only on English utterances, but it is best not to mix those two pre-codes in the same data file.

If this still doesn't work, then please email one of your data files to me at spe...@cmu.edu.

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 on the web visit https://groups.google.com/d/msgid/chibolts/CALbyt0eCFDwSMT7hJzFDxRoTyZ6Ge2X2LuVK_rBOz5SdZDm%3Dsg%40mail.gmail.com.

[Brian Macwhinney]

Dear Rachel,
I would have to look at your input file and the commands you are using. More generally, when giving bug reports, please try to follow the procedure in section 6.8 of the CLAN manual.

— Brian MacWhinney

[Brian Macwhinney]

Dear Rachel,

Yes, this is all correct. Section 16.1 of the CHAT manual on code-switching also explains this.

— Brian

> To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/8E4369C4-22A1-4264-821A-0924496F6BA1%40andrew.cmu.edu.

[Rachel Romeo]

Thanks all! Our files were transcribed correctly, and I had simply swapped the + for - (got tripped up by the page break in the MOR manual). Works perfectly now, thanks!

To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/692E0A3F-9AFF-4803-89D8-30FC30E79A6E%40cmu.edu.