Simple and not-so-simple analysis help needed

24 views
Skip to first unread message

Cynthia Core

unread,
Sep 13, 2018, 9:16:57 AM9/13/18
to chib...@googlegroups.com
Hello,

1 - We are analyzing bilingual language samples and want to count the number of utterances assigned to a particular language - only English, for example.

Can I use FREQ for that? Or do I need to generate a MOR tier and then do something else?

2 - We want to count NPs that govern main verbs. I know I can count nouns and I can count main verbs - but how can I delimit the command so I only get nouns that are the heads of NPs that govern main verbs?

Thanks for the help - 
Cynthia

--
Cynthia Core, PhD, CCC-SLP
Associate Professor
Department of Speech and Hearing Science
The George Washington University

Brian MacWhinney

unread,
Sep 14, 2018, 3:36:08 PM9/14/18
to ChiBolts
Yes, FREQ can count this.  First, you would count all utterances and then you would count only the occurrences of the language precode, such as +s”[- spa]”

In order to count NP’s that are the Subjects of Verbs, you would have to use the %gra line and count instances of the code SUBJ.

—Brian

--
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/CAKO0FxaD6hFf49aLfLc0bihy3ZcSichUfy%3DgC2tRX00wuPQ9Og%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

Cynthia Core

unread,
Sep 14, 2018, 3:41:14 PM9/14/18
to chib...@googlegroups.com
Thank you Brian!
I’ve asked a few people but no one seems to know how to use FREQ to count utterances.
I looked around and didn’t fine it either - do I just count the tier prefix MOT and exclude the ones I don’t want?
I apologize for asking such a basic question!
Cynthia


On Friday, September 14, 2018, Brian MacWhinney <ma...@andrew.cmu.edu> wrote:
Yes, FREQ can count this.  First, you would count all utterances and then you would count only the occurrences of the language precode, such as +s”[- spa]”

In order to count NP’s that are the Subjects of Verbs, you would have to use the %gra line and count instances of the code SUBJ.

—Brian

On Sep 13, 2018, at 9:16 AM, Cynthia Core <co...@gwu.edu> wrote:

Hello,

1 - We are analyzing bilingual language samples and want to count the number of utterances assigned to a particular language - only English, for example.

Can I use FREQ for that? Or do I need to generate a MOR tier and then do something else?

2 - We want to count NPs that govern main verbs. I know I can count nouns and I can count main verbs - but how can I delimit the command so I only get nouns that are the heads of NPs that govern main verbs?

Thanks for the help - 
Cynthia

--
Cynthia Core, PhD, CCC-SLP
Associate Professor
Department of Speech and Hearing Science
The George Washington University

--
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+unsubscribe@googlegroups.com.

--
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+unsubscribe@googlegroups.com.

To post to this group, send email to chib...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

Brian MacWhinney

unread,
Sep 14, 2018, 4:12:27 PM9/14/18
to ChiBolts, co...@gwu.edu
Yes, this seems unnecessarily tricky, I agree.  The trick is that you have to add the +y switch, as in 

freq +s"\**" *.cha +y

This looks at all the occurrences of the asterisk, since an asterisk marks every utterance.  The second occurrence of an asterisk separates out speakers.  If you want to merge speakers, you could have 

freq +s"\*%" *.cha +y

And you can add +u if you want a single number for the whole corpus.

--Brian

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.
Reply all
Reply to author
Forward
0 new messages