accurate utterance and turn counts taking into account linguistic, verbal, and non-verbal contributi

[A Cristia]

Dear Chibolters,

We would like to have separate counts for both utterances and turns, taking into account linguistic, verbal, and non-verbal contributions. For example:

*CHI: 0.

%act: points to a cookie

*MOT: okay .

*CHI: &baba .

*MOT: go for it .

*CHI: baby .

We see in this example:
- 3 turns by the child: 1 non-verbal (pointing), 1 verbal non-linguistic (&baba), and 1 linguistic (baby);
- 2 utterances by the child (because utterances cannot be non-verbal): 1 verbal non-linguistic (&baba), and 1 linguistic (baby)

MLT on the sample above yields 3 utterances and 3 turns. If we MOR it and MLU the output, then we find 1 utterance. Is there a simple way to extract the other subtypes? Or should we add by hand codes such as:

*CHI: 0. [- verbal]

%act: points to a cookie

%trn: $TUR

*MOT: okay .

*CHI: &baba .

*MOT: go for it .

*CHI: baby .

To exclude non-verbal turns from MLT:

> MLT -s"[- verbal]"  @    

mlt -s"[- verbal]" @

Fri Oct 30 09:57:59 2015

mlt (17-Apr-2015) is conducting analyses on:

  ALL speaker tiers

****************************************

From file </Users/alejandrastein/Desktop/prueba.cha>

MLT for Speaker: *CHI:

  MLT (xxx, yyy and www are EXCLUDED from the word counts, but are INCLUDED in utterance counts):

    Number of: utterances = 2, turns = 2, words = 1

Ratio of words over turns = 0.500

Ratio of utterances over turns = 1.000

Ratio of words over utterances = 0.500

To count only non-verbal turns:

> keymap +b$TUR* +t%trn  @   

keymap +b$tur* +t%trn @

Fri Oct 30 09:59:12 2015

keymap (17-Apr-2015) is conducting analyses on:

  ALL speaker tiers

and those speakers' ONLY dependent tiers matching: %TRN;

****************************************

From file </Users/alejandrastein/Desktop/prueba.cha>

Speaker *CHI:

  Key word "$tur" found 1 time

Incidentally, the CLAN manual does refer to non-verbal turns twice; on p. 34:

5.4 MLT Analysis

The third analysis we will perform is to compute MLT (Mean Length of Turn) for both child and mother. Note that, unlike the MLU program, the CLAN program MLT includes the symbols xxx and yyy in all counts. Thus, utterances that consist of only unintelligible vocal material still constitute turns, as do nonverbal turns indicated by the postcode [+ trn] as illustrated in the following example:

*CHI: 0.[+ trn]
%gpx: CHI points to picture in book

We can use a single command to run our complete analysis and put all the results into a single file.

mlt *.cha > allmlt.cex

However, as far as we can see, with or without [+ trn] one gets the same results using the example provided in the manual (*CHI: 0.[+ trn]):

WITH TRN

> mlt  @
mlt @
Fri Nov 13 09:37:51 2015
mlt (09-Sep-2015) is conducting analyses on:
  ALL speaker tiers
****************************************
From file </Users/acristia/Dropbox/ANR_argentina/transcriptions/test.cha>
MLT for Speaker: *CHI:
  MLT (xxx, yyy and www are EXCLUDED from the word counts, but are INCLUDED in utterance counts):
    Number of: utterances = 1, turns = 1, words = 0
    Ratio of words over turns = 0.000
    Ratio of utterances over turns = 1.000
    Ratio of words over utterances = 0.000

WITHOUT TRN (i.e., I removed [+ trn] from 0.[+ trn])

> mlt  @
mlt @
Fri Nov 13 09:38:32 2015
mlt (09-Sep-2015) is conducting analyses on:
  ALL speaker tiers
****************************************
From file </Users/acristia/Dropbox/ANR_argentina/transcriptions/test.cha>
MLT for Speaker: *CHI:
  MLT (xxx, yyy and www are EXCLUDED from the word counts, but are INCLUDED in utterance counts):
    Number of: utterances = 1, turns = 1, words = 0
    Ratio of words over turns = 0.000
    Ratio of utterances over turns = 1.000
    Ratio of words over utterances = 0.000

Thank you all in advance for your help,

Alex Cristia

[Brian MacWhinney]

Dear Alex,

     As far as I can see, MLT is doing what you want.  Running on your six line sample, I get three turns for the child.  It is true that it is not necessary to include the [+ trn] postcode.  I’ll fix that part of the manual.

—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 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/eaddfa14-8f7c-4c55-a030-6e9a427016fd%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[A Cristia]

Dear Brian,

Sorry that my question isn't clear. The hope is to be able to *separate* the different types; i.e. apply one processing scheme and get, for the child:
- 1 non-verbal turn (pointing),

Run a different analysis and get, for the child:
- 1 verbal non-linguistic turn (&baba),

Run a different analysis and get, for the child:
- 1 linguistic turn ("baby")

Regular MLT groups all together, as you point out.

Thank you very much,

Alex

[Brian MacWhinney]

Dear Alex,

     You can create all these different types of analysis through the use of postcodes and the +s or –s switches.

—Brian

To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/d47dbe14-815b-42d3-9c0d-f22a88e73002%40googlegroups.com.

[A Cristia]

Dear all,

I think I found the solution. Posting it here in case someone else ever has the same question.
We want to count separately:

1. Total number of interventions
   

2. Number of verbal interventions 

3. Number of verbal *linguistic* interventions 

4. Number of verbal *non-linguistic* interventions

5. Number of non-verbal interventions

Naturally, A=B+E and B=C+D, so only 3 numbers need to be calculated.

Imagine this example:

1     *CHI: 0.

2     *CHI: &baba .

3     *CHI: &baba .

4     *CHI: &baba .

5     *CHI: bat .

6     *CHI: baby .

We'd like to find: A=6; B=5; C=2, D=3; E=1

mlt  @

Number of: utterances = 6 --> A

mlt +a @

Number of: utterances = 2 --> C
(Incidentally, I believe the manual pp. 118 and 119 is incorrect:"Utterances with no words are excluded, although they can be included using the +a switch." and "This switch causes all utterances to be counted, even if they have no words." versus the in-CLAN explanation: "+a : do not count empty utterances, 0.")

mlt +s"&*" @

Number of: utterances = 3 --> D

And the rest of the numbers are derived from those 3:

B=C+D= 2+3 = 5

E=A-B=6-5=1

-Alex

[Brian MacWhinney]

Dear Alex,

    Yes, you are right.  I made a test file out of your example and all the commands are correct.  I will fix that mistake in the manual. 

 

-- Brian