Clauses per utterance with EVAL

Mon Jan 7 21:46:26 UTC 2013

Hi Leonid

Thanks so much for looking into this. I re-ran this on a small (French) file just to double check. It's quite possible I've made an obvious error, so I'll take you through my working.

So I ran EVAL on the (.mor.pst.cex) file and it showed me the following:

Utts: 17
Verbs: 14
Clause/Utt: 0.824

To double check this, I ran a search on all the verbs in the file (freq %mor +s@"|v*" C121d*.mor.pst.cex +u) and it tells me the following:

 From file <C:\USERS\KEVIN\DESKTOP\C121dOB.mor.pst.cex>
Speaker: *121:
  1 v:aux|avoir&PRES&3SV
  1 v:aux|être&PRES&3SV
  1 v:exist|être&IMPF&3PV
  2 v:poss|avoir&PRES&3SV
  1 v|arriver-PP-_MASC
  4 v|chercher-PRES-_3SV
  1 v|dormir-PRES-3SV
  2 v|jouer-PRES-_3SV
  1 v|laisser-PRES-_3SV
  1 v|lire-PRES-_3SV
  1 v|marcher-PRES-_3SV
  1 v|perdre-PP-_MASC
  1 v|rentrer-PRES-_3PV
  1 v|trouver-PRES-_3SV
------------------------------
   14  Total number of different item types used
   19  Total number of items (tokens)

As the French Passé Composé (past tense form) is formed of an aux + pp, when you take away the two PP from this list you get 17 tokens. When I count the Vs manually I also get 17. So the 14 verbs result from EVAL would appear to match the total number of different verb types, not tokens?

I hope that makes sense.
All the best
Kevin

--
Dr Kevin McManus
Department of Modern Languages
University of Southampton

tel: +44 (0) 23 8059 3970
http://www.soton.ac.uk/ml/about/staff/km2m10.page

Project website: http://langsnap.soton.ac.uk/

________________________________________
From: chibolts at googlegroups.com [chibolts at googlegroups.com] on behalf of Leonid Spektor [spektor at andrew.cmu.edu]
Sent: 07 January 2013 17:18
To: chibolts at googlegroups.com
Subject: Re: Clauses per utterance with EVAL

Kevin,

        I had to talk to more informed on this subject people than me before I could answer your question and EVAL definitely counts ratio of tokens per utterances. Just to make sure that I am talking about the same thing that you are, here is sample CHAT file:

which will produce clause/utterances result equal to 1. If EVAL was counting verb types or verb types lemmas, then the result would have been equal to 0.5 or 0.25.

        May I ask what leads you to believe that EVAL counts verb types (not tokens)?

Leonid.

On Jan 6, 2013, at 14:43, Mcmanus K. wrote:

> Dear all
>
> I’ve been trying to calculate the average number of clauses per utterance using EVAL. When I run this it seems to work fine, but EVAL appears to use verb types (not tokens) to calculate this.  I’d like to calculate the average number of clauses per utterance using verb tokens (not types). Is there a way I can do this? Or maybe a different command is the answer?
>
> Any feedback or advice would be greatly appreciated.
>
> All the best
> Kevin
>
> --
> Dr Kevin McManus
> Research Fellow in French Applied Linguistics
> Modern Languages
> University of Southampton
>
> tel: +44 (0) 23 8059 3970
> http://www.soton.ac.uk/ml/about/staff/km2m10.page
>
>
> --
> You received this message because you are subscribed to the Google Groups "chibolts" group.
> To post to this group, send email to chibolts at googlegroups.com.
> To unsubscribe from this group, send email to chibolts+unsubscribe at googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>

--
You received this message because you are subscribed to the Google Groups "chibolts" group.
To post to this group, send email to chibolts at googlegroups.com.
To unsubscribe from this group, send email to chibolts+unsubscribe at googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

-- 
You received this message because you are subscribed to the Google Groups "chibolts" group.
To post to this group, send email to chibolts at googlegroups.com.
To unsubscribe from this group, send email to chibolts+unsubscribe at googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.