Coordinating tiers with CLAN

Wed May 14 03:33:38 UTC 2003

Thanks. In my case, it doesn't look powerful enough, as my coding tiers for
grammatical and pragmatic features only contain as many words (by which you
mean typographical units separated by spaces?) as there are deictics on the
main line. It looks as though I might have to go through the whole lot
inserting dummy symbols on the coding tiers (e.g. xx) for all the other
words. However, I'll give it a go when I get going on research again in
about a month.

Charles Watkins.
-----Message d'origine-----
De : Javier <lpxao at psychology.nottingham.ac.uk>
À : Info-ChiBolts at mail.talkbank.org <Info-ChiBolts at mail.talkbank.org>
Date : lundi 12 mai 2003 21:52
Objet : Re: Coordinating tiers with CLAN

"MODREP" code associates dependent tiers to the main one, but I'm not sure
whether it'd be helpful for you or not.,

Let me give an example:

If you want to know the number of times that 'la' was introduced incorrectly
you can pool them out with:

freq +t%mor +s"**det|la" *.cha

Or you can create a new file with kwal, including the tiers you want to keep
for ulterior analyses with the delimiter +o. The example shown bellow would
include all the tokens of "la", wither correct or not, plus the %err lines
associated to them, and getting rid of %syn tiers:

kwal +t%mor +s"*|la" +o%err *.cha

Alternatively, you can associate dependent tiers to the main one by using
"modrep".

For instance, if you want to know how often "M" appears with "the" and how
often with "la", you can use this command:

modrep +b*%syn +c*CHI +o"*M" *.cha

This command is not very powerful, though. You need a perfectly coded
transcript (with exactly the same number of words in all the tiers), and it
just gives you the words and numbers, but you cannot include anything else.

Hope this helps,

Javier Aguado Orea

School of Psychology
University of Nottingham
NG7 2RD
UK

on 13/5/03 10:36 pm, Charles Watkins at charles.watkins at wanadoo.fr wrote:

> I am glad this has been raised again: I will shortly be getting to a stage
> in my research into deixis in bilingual children where it would be
extremely
> convenient to be able to match grammatical characteristics of a given
> deictic (coded on one dependant tier) with its pragmatic characteristics
> (coded on a separate tier) and link both to the deictic itself (flagged on
> the main tier). Like Patti, I am anticipating problems in the frequent
cases
> of utterances with more than one deictic.
>
> I raised this question a couple of years or so ago on Chibolts and got
some
> answers which didn't really address my concerns: this I realised was
because
> my question had not been formulated with sufficient clarity. Indeed, the
> issue is complicated, and might possibly be specific to cross-language
> studies, such as ours are - perhaps Patti would be kind enough just to
> confirm I have interpreted her question correctly, and that we both have
the
> same concern.
>
> Again thanks for raising it : the research season (i.e. end of teaching
and
> examining year) is about to start here (France), and I am dusting off the
> files and trying to remember where I was... I look foward to being able to
> forge ahead in the summer holidays if we get this one straightened out.
>
> Charles Watkins
>
>
>
>
> -----Message d'origine-----
> De : Patti Spinner <pattispinner at hotmail.com>
> À : Info-ChiBolts at mail.talkbank.org <Info-ChiBolts at mail.talkbank.org>
> Date : lundi 12 mai 2003 14:51
> Objet : Coordinating tiers with CLAN
>
>
>> Hello,
>>
>> I'm using CHAT and CLAN for the first time, and I'm trying to figure out
> how
>> best to code transcripts of second-language production for use of
>> determiners.  I want to code separate tiers for %mor, %syn, and %err.  A
>> problem seems to arise when there are multiple determiners in a speaker's
>> turn.  If I code two or more instances on one tier, then how will the
>> program know which %mor coding belongs with which %err coding (i.e.,
which
>> determiner)?  I need to be able to retrieve a single determiner's %err,
> %mor
>> and %syn information simultaneously to check for correlations, etc.
>>
>> I have considered simply making duplicates of each turn that has multiple
>> determiners and then coding only one determiner for each copy, but this
>> doesn't seem like the best way.  Does anyone have a suggestion for coding
>> and/or analyzing the data so that this is not necessary?
>>
>> Thanks!
>> Patti Spinner
>> University of Pittsburgh
>>
>> _________________________________________________________________
>> Hotmail - Absolut kostenfrei! Der weltweit größte E-Mail-Anbieter im
Netz:
>> http://www.msn.de/hotmail
>>
>>
>
>
>