+xCN switch in KWAL

[jgar...@umd.edu]

Hello,

I was wondering if anyone could clarify how the +xCN switch in KWAL works. I was wondering if I could use this switch to produce only utterances with a certain morpheme length.

 It says under KWAL, in the manual and when I type kwal into the commands window:

+xCN: include only utterances which are C (>, <, =) than N items (w, c, m), "+x=0w" for zero words

I can't find a definition of what "w, c, m" means, but if the +x=0w is for 0 words, then I thought maybe +x=1m could get me all of the 1 morpheme utterances, etc. However, when I run +x=1m it gives me all 1-word, not 1-morpheme utterances. (The output for +x=1m and +x=1w are exactly the same.)

Is the "m" here supposed to be for morphemes? I can't seem to locate a definition. I do see that under TTR, it says "The only new element in this command is +sm;*o%. The +s option tells FREQ to search for and count certain strings. The m tells the program to look at the %mor line. The semicolon says to look at the stem and then the *,o% says to ignore everything else."

Thank you.

Julianne Garbarino

[Leonid Spektor]

Julianne,

You interpreted the +xCN option correctly, the 'm' does stand for morphemes. I have changed KWAL usage to make more clear. The other consideration is that the 'm' option will only work correctly if you have +%mor: tiers in your data files. Last this is that there was a problem with the way KWAL was counting morphemes on %mor tier and as the result it might have missed some utterances in the output.

I have fixed this and if you get new program from dali.talkbank.org/clan/, then the "+x=1m" and "+x=1w" options will produce different results, assume that speaker does actually use multi-morphemes words.

FREQ's "+sm;*,o%" option will interfere with "+xCN" option. If you want to use FREQ with "+sm;*,o%" option, then you should first use KWAL with "+x=1m +d +f" options, for example, and then run FREQ with "+sm;*o%" option on the output of KWAL.

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/636e0c0d-369b-4a99-943d-a094e8c94e22n%40googlegroups.com.

[Julianne Garbarino]

Dear Leonid,

Thank you so much. This switch is very helpful for my project and will allow me to work around not being able to produce a spreadsheet with the MLU of each utterance, as I was trying to do a couple of days ago. 

I've just downloaded the new version of CLAN for Mac and am using it with Mojave. With a relatively short test file, I am still having one slight problem. When I run kwal +d4 +t*CHI +x=1m, some 1-morpheme utterances are not appearing in the results. These are one-word/one-morpheme utterances that just say "yeah." They are appearing when I run kwal +d4 +t*CHI +x=1w. Other one-word/one-morpheme utterances coded as "co" (e.g., "okay") are coming up in the list of one-morpheme utterances I get when I run "kwal +d4 +t*CHI +x=1m".

Please let me know what I can do. I can send you this test file directly if that would be best.

Thank you!

Julianne Garbarino

To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/7D85A781-1227-4157-9AE1-A70AC071B61F%40andrew.cmu.edu.

--

Julianne Garbarino, MS, CCC-SLP
PhD Candidate, Dept. of Hearing and Speech Sciences

[Leonid Spektor]

Julianne,

Please, sent to me directly the test file you are using, the exact command line and indicate the utterances that you are expecting to get with "kwal +d4 +t*CHI +x=1m" command.

Leonid.

To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/CA%2BKHNfku8kd2soFHO8B_hbTphoAfhOsQu71fZRnzbz1KPiB0-g%40mail.gmail.com.