Depfiles & gems

[Ingrid Willenberg]

Dear Leonid

I have 4 gems in each of my files. At this stage only one of the gems has been coded. when I run CHECK on the files, it says the following about the @bg: line: Tier is not declared in depfile file.

Overall there seem to be several transcription issues which result in the files not passing CHECK. However it's not my dataset and I don't have the resources required to fix everything.

The files are good enough to generate output for FREQ, which is actually all I need. Although I would like to know how to add items to the depfile and I can't seem to find that in the manual.

I also haven't had success trying to use GEM commands on these files. I tried

Ultimately I want to generate FREQs for each of the codes in the data set

Here is a sample of the output for one file:

From file </Users/ingridwillenberg/Dropbox/TOM Project Data/ID test/020304sIW.cha>

Speaker: *CHI:

  2 $ADJ

  2 $NPSP

  1 $NPUSP

  3 $PRONAM

  2 $RPS

  2 $SEQ

When I try  using +d3 to generate an excel file from the freq output, the file contains the types, tokens and TTR but nothing else. Is this where the files not passing CHECK come back to haunt me??

Thanks in advance for your time.

Kind regards,

 

Ingrid

Dr Ingrid A. Willenberg (Ed.D)(Harvard University)

[Brian Macwhinney]

Dear Ingrid,

    Yes, life would be easier if your files passed CHECK.  You can't do a GEM analysis, because your files have @bg instead of @Bg and often they are not correctly matched.  You would find this out if CHECK would run.  To get CHECK to run, you need to add the letter "x" to your tiers such as %xcar or %xref.  You can do this globally, but if you don't want to fix things, then many things will just not work.

   The good news is that FREQ will work, you just have to use the +d2 switch and not +d3 and you want +t%ref or +t%xref.

-- Brian MacWhinney

--
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/CAFhboLGjZXMYz0QOt8bBmvwsCHFqNACEiFdM67WkMt6WwfRZ9A%40mail.gmail.com.
<010206sIW.cha><020402sIW.CHA><020304sIW.cha><030108sIW.CHA><040201sIW.cha>

[Leonid Spektor]

Hi Ingrid,

As Brian has already mentioned it would be best if the data does pass CHECK. However, GEM and FREQ and other commands were designed to work on data files that do not pass CEHCK all the way. One problem with not passing CHECK is that data might not be consistent and when searching for codes on particular dependent tiers or searching for particular GEMS the misspelled tiers or GEM markers will be ignored, thus results might be missing some information. Changing depfile.cut is not recommended.

For your data, for example to create GEM output files for "narrative" GEMS, you can use the command "gem +t@ +t% +snarrative +fnarrative +d *.cha" and then you can run command "freq +t%ref +t%car +d2 *.narrative.cex" to get FREQ Excel results. The "+fnarrative" option in GEM will create output with ".narrative.cex" file extension which will make it easier to identify a particular data set later on.

Leonid.

[Ingrid Willenberg]

Thank you both. With your help I can now access the data I need. I renamed the coding tiers using REPLACE and the checking has been more successful. However, CHECK did not like the use of [c] to mark clauses (not that they were correctly done to begin with). 

Yes, I realise that working with messy files is not ideal but I have zero funding and zero assistance with this and taking time to to fix up 155 files When I will only be needing freqs just does not seem worth the effort.

[Leonid Spektor]

The [c] can be replaced with [^c] to allow CHECK to pass.

Leonid~

On Nov 27, 2020, at 01:05, Ingrid Willenberg <ingridwi...@gmail.com> wrote:



--
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/CAFhboLGPMoJfyZBPQ8ShhCD2ZVK6X9%2BZmoH6whRDvutPVhv3UA%40mail.gmail.com.

[Ingrid Willenberg]

Thanks!

To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/1473141B-9295-46C6-93FD-0C6A4C7C9FEA%40andrew.cmu.edu.