CLAN questions - new user

[Brenda Beverly]

To Whom It May Concern:

I am a new user of CHAT/CLAN (long time SALT user). My colleague in China and I are using CHAT/CLAN for transcription and analysis of mothers' engagement behaviors when booksharing with 4 year olds. We may be able to donate the videos and/or transcripts to CHILDES in the future, but in the short term, we are trying to maximize the CHAT/CLAN capabilities. 

I have a few questions: 

1. Are post codes the best CHAT method for our study purposes? After working to learn CHAT , it seems that our study-specific coding of engagement strategies is best accomplished through post codes. So, we have an utterance and then a post code - for example,  [+ CR] which is our abbreviation for Choral Reading, an engagement strategy of the mothers that we are tracking. 

 

2. How could we get separate tiers for when the mothers are reading the text from the children's book versus when the mothers are talking/speaking, not reading? We have ruled out @g. It seems @g is better for 2 different booksharing activities, not these separate types of talk with one booksharing/storybook. We are considering labeling the Participant tier to separate this - @MOT for Mother reading, but maybe @MET for Mother's extra-textual talk. This will likely give us the output we need but it's not truly 2 different participants. In that regard, it feels like we might be missing out on a tier option that would be a better representation of the transcripts. Should we have set this up with a dependent tier using %ETT (extra-textual talk) perhaps? Could we keep the post codes and add this following the MOT utterances that are extra textual? Or do we need to use the codes together, for example:

%ETT : AK (AK = our code for an acknowledgment)

3. Is the freq command the best CLAN program for our purposes? We have successfully run freq to obtain the counts for the postcodes and exported that to excel. 

The program my colleague ran was:

freq @ +[*MOT = s"<+ AK>" 

She ran separate analyses for each of the engagement behaviors - AK for acknowledgement as well as CR Choral Reading etc. 

This is working. I was simply curious if this was the best/right option, especially given my other questions about the use of post codes and tiers. 

4. Lastly, I believe we could run a command that would include all 10 participant transcripts, but I haven't studied or determined how to set that up. Are you able to assist? 

My apologies for these basic questions. We are excited to access CHAT/CLAN for this project and future work. Your support is greatly appreciated!

Brenda Beverly

bbev...@southalabama.edu

251.635.3999 mobile #

[Leonid Spektor]

Hi Brenda,

I can give you my suggestions, but these are definitely not the only choices that you have. Maybe someone else would have better ideas.

1., 2. The post-codes are best way to identify specific utterances from a larger group of utterances. So, for example, if you have *MOT-ET: speaker code for extra-textual talk and *MOR-RE: speaker code for mother reading, then you would have @Participants: MOT Mother header tier to identify one Mother speaker. This way you have only one speaker with different tiers based on what those tiers represent. This way you can use FREQ command to locate just that speaker with the command freq +t*MOT if you want separate results, or freq +t*MOT-RE for just mother reading, or command freq +t*MOT +o3 if you want combined result for all mother's utterances.

The post-codes are the best way to tag utterances, because they will allow you to find other dependent tiers along with speaker tiers. For example, if you have utterances:

*MOT-ET: extra textual talk [+ CR] .

%mor: adj|extra adj|textual n|talk .

%gra: 1|3|MOD 2|3|MOD 3|0|INCROOT 4|3|PUNCT

and you want to run analyzes on %mor: tier with post-code [+ CR], then you would use command freq +s"[+ CR]" +t%mor or if you want only mother's extra-textual talk utterances with post-code [+ CR], then use command freq +t*MOT-ET +s"[+ CR]" +t%mor.

If you use dependent tiers for coding, then it would be more difficult to associate those codes with other tiers of that speaker. It can be done, but it would require two commands 1. KWAL to extract only utterances with that code on dependent tier and next FREQ command to run analyses on those extracted tiers.

If you only want to count codes on specific dependent tiers, then using dependent tiers like %xett : $AK will work. Notice, the dependent tier would have to start with 'x' or it will not pass CHECK command.

3. The command that you have freq @ +t*MOT +s"<+ AK>" is best to count number of occurrences of actual codes. If you want to run analyzes on the words used by the speaker, then you would use command freq @ +t*MOT +s"[+ AK]". This can be combined with other dependent tiers associated with that speaker tier. For example, command freq +s"[+ AK]" +t%mor will analyze words on %mor tier.

4. If you mean to run FREQ on multiple files to get separate results for each file, the you could use *.cha or "File In" button in Commands window. If you want to get combined results for multiple files, then add +u option to the command line.

Leonid.

On Mar 25, 2021, at 11:25, Brenda Beverly <bbev...@southalabama.edu> wrote:

To Whom It May Concern:

I am a new user of CHAT/CLAN (long time SALT user). My colleague in China and I are using CHAT/CLAN for transcription and analysis of mothers' engagement behaviors when book sharing with 4 year olds. We may be able to donate the videos and/or transcripts to CHILDES in the future, but in the short term, we are trying to maximize the CHAT/CLAN capabilities. 

I have a few questions: 

1. Are post codes the best CHAT method for our study purposes? After working to learn CHAT , it seems that our study-specific coding of engagement strategies is best accomplished through post codes. So, we have an utterance and then a post code - for example,  [+ CR] which is our abbreviation for Choral Reading, an engagement strategy of the mothers that we are tracking. 

 

2. How could we get separate tiers for when the mothers are reading the text from the children's book versus when the mothers are talking/speaking, not reading? We have ruled out @g. It seems @g is better for 2 different book sharing activities, not these separate types of talk with one book sharing/storybook. We are considering labeling the Participant tier to separate this - @MOT for Mother reading, but maybe @MET for Mother's extra-textual talk. This will likely give us the output we need but it's not truly 2 different participants. In that regard, it feels like we might be missing out on a tier option that would be a better representation of the transcripts. Should we have set this up with a dependent tier using %ETT (extra-textual talk) perhaps? Could we keep the post codes and add this following the MOT utterances that are extra textual? Or do we need to use the codes together, for example:

%ETT : AK (AK = our code for an acknowledgment)

3. Is the freq command the best CLAN program for our purposes? We have successfully run freq to obtain the counts for the post-codes and exported that to excel. 

The program my colleague ran was:

freq @ +[*MOT = s"<+ AK>" 

She ran separate analyses for each of the engagement behaviors - AK for acknowledgement as well as CR Choral Reading etc. 

This is working. I was simply curious if this was the best/right option, especially given my other questions about the use of post codes and tiers. 

4. Lastly, I believe we could run a command that would include all 10 participant transcripts, but I haven't studied or determined how to set that up. Are you able to assist? 

My apologies for these basic questions. We are excited to access CHAT/CLAN for this project and future work. Your support is greatly appreciated!

Brenda Beverly

bbev...@southalabama.edu

251.635.3999 mobile #

--
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/b74c73f7-f02f-4bd6-8e4e-ab84e066f6efn%40googlegroups.com.

[Brenda Beverly]

Leonid:

Thank you so much for this timely and detailed response to my questions. I am excited to use these suggestions to move forward in a way that will support our research. I cannot thank you enough. 

all the best, 

Brenda

[Janet Bang]

Hi Brenda, 

Adding onto Leonid's suggestions about postcodes, which we found helpful for a similar purpose. Here are some guidelines that we've been using regarding booksharing where in our study we were trying to distinguish when caregivers are reading from the book directly (we refer to as 'recited' speech) or include a mix of recited and spontaneous speech (we refer to as "recited mixed" speech). Any utterances without a postcode we consider their spontaneous and natural speech, and this has been working out ok for us when we want to extract different portions so far. Our decision to distinguish "recited" from "recited mix" was more for descriptive purposes, than for our primary research questions, where we just combine both when we consider reading passages.

Our postcodes:

[+ recit] means they are reciting the text

[+ rmix] means that they are mixing the text with their own deviations from the text or adding to the text

Recited utterances: [+ recit]

Acceptable words for recited speech include speech that is ‘starting’ an utterance to get the child’s attention. In the example below, “look it says” is not part of the text, but is starting the utterance.

 

      *MOT:             look it says Thomas_the_Train went round and round. [+ recit]

 

The following are still accepted as recited:

• If the caregiver reads parts of the recited text and skips around the page, but every word is said as it is written, then give this a recited code.
• Changing “that” for “which” or vice versa is acceptable and not enough for a rmix postcode.
• Contractions or vice versa are acceptable as recited because these do not change the meaning (e.g., the text is “I will be back said the dog” and the caregiver says “I’ll be back said the dog”, or vice versa.)
• Changing the order of words (e.g., white fluffy feathers to fluffy white feathers)

 

 

Recited mixed utterances: [+ rmix]

Any words that are omitted, added or changed within the text should get the [+ rmix] postcode. For example, the text is “and King said he was so happy to see his friends”.

 

         Omitted “so”:

            *MOT:             and King said he was happy to see his friends [+ rmix]

 

         Added language:

            *MOT:             and King said he was so happy to see his good friends [+ rmix]

 

         Changed language: this includes articles (e.g., a vs. the)

            *MOT:             and King said he was so happy to see his friend [+ rmix]

 

           We also count translations of the recited text (this counts for ‘live’ translations, where

the caregiver has a book in one language in front of them that does not have the translated text, so instead is translating it as they read). We will count this here because this language is no longer spontaneous, since what to say is dictated by the text available.

To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/e22e93a1-b3f3-4a4d-9968-408b1fb11ee1n%40googlegroups.com.

[Brenda Beverly]

Hi Janet, 

Thank you for sharing your ideas and example rules/utterances. We have struggled with similar decisions. My doctoral student and I initially considered the mother's engagement strategies to be "extra-textual" or not part of the read story. But, as you indicated, we have moms who are using choral reading with the child for text that has been rehearsed/practiced. That's an engagement strategy using the text, not extra-textual. We also entertained the possibility that some moms would paraphrase the story or skip words/text. We haven't seen that, but it may be because our moms are reading to their 4 year olds. I know I skip/modify text when I'm reading to infants and toddlers. 

I am hesitant to share our coding (i.e., all of the engagement strategies) - simply because we are still refining it for the study's purposes so it looks a bit messy to me. I would be happy to do share as we move forward. Should we be in touch via email in the future to share our coding systems and rules? Are you developing your own and is it published? Or, have you modified from another's work? This line of investigation is relatively new for me and it is led by my doctoral student, Yingshan Huang, from the Fujian University of Traditional Chinese Medicine in Fuzhou, China. She and I will consult about your examples. 

Thank you again. 

Brenda

[Janet Bang]

Hi Brenda, 

We've developed it on our own, but we are still working on data from it. It sounds like you have different questions in mind than we do, but just thought we'd share so you can get a sense of how we're using the CHAT conventions for our analyses with CLAN. Feel free to email me if you have any questions (janet...@sjsu.edu). 

Sure I'd be happy to stay in touch. Good luck!

Janet

To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/a69d5282-0361-4a9d-86a1-a1d01844d9dfn%40googlegroups.com.