updates in CHAT
Brian MacWhinney
In response to work that Johannes Wagner and Lone Laursen are
doing on a gold standard corpus for Danish CA -CHAT transcription, we
have modified three things in CHAT and I am interested in getting
reactions to these changes. Before going into the details, let me
emphasize that Lone and Johannes are trying to avoid breaking up
TCUs. Because non-CHA CHAT tends to break up TCUs, it is important to
provide ways of avoiding TCU breakup for CA transcription.
1. Earlier, we had introduced the triple wavy mark as the indicator
of a continuation of a TCU across an interruption from another
speaker. See line 9 at http://talkbank.org/CABank/codes.html. This
is not the mark of basic speaker-internal latching, but rather of TCU
continuation. However,initially we used the same triple wavy symbol
at the end of the first segment of the TCU and the beginning of the
second segment. This is not good for computational analysis, so we
now added a plus before the triple wavy to mark the second case.
2. It is often the case that a single TCU includes a variety of
phrases that are marked with one of the six final contours gien in
rows 3-7 at http://talkbank.org/CABank/codes.html. However, in order
to make it clear to the computer that these marks are not the ends of
the TCU, we are adding a comma after them when they are not TCU final.
3. It is often the case that a speaker will continually mark
acknowledgement, assent, or coparticipation with short forms such as
"ja" "mhmm" or "uhhuh" that punctuate a longer TCU by the other
speaker. Interrupting the ongoing TCU overlp marking and new lines
for this can be cumbersome. So, we have added a new code to CHAT that
allows for in-line marking of these acknowledgments. The form is
&*SAM=yeah, where SAM is the code name for a speaker such as SAM or,
if the code was just S, then it would be &*S=yeah. This code is
placed directly after the word with which it overlaps.
I would be interested in comments regarding any of these additions to
CHAT. Many thanks.
-- Brian MacWhinney
Mike Forrester
Just a couple of queries:
1. Do you have any examples of what the ouput will look like when one
has CHA>CA'd such a file (or rather section with the new TCU symbols
embedded).
2. It might seem to some ethno/CA people that the conventions are now
extending quite considerably from the initial 'Jeffersonian' set.
This is not necessarily a bad thing but might put some traditionalists
off using computer supported transcription. Also, and just a thought,
the co-participation indicator development makes me think that the
business of initial transcription is becoming more interdependent with
interpretation. For example, Schegloff's scheme for repair
organization might treat some of these 'continuers' as initiations for
repair. I'm not sure, so will look forward to seeing how people might
use the whole range of the new symbols.
regards for now and thanks for new version of CLAN
Mike
2009/4/8 Brian MacWhinney <ma...@cmu.edu>:
Brian MacWhinney
Johannes, Lone, Franklin, and I are working to see if we can avoid
some of these
complications. When we manage to do this, we will post the outcome,
along with examples.
I think we all agree that the co-participation indicator is unlikely
to be picked up by
CA people. However, as you say, the best thing would be to get some
examples
out so we could look at the details.
People can always continue to use the Jeffersonian set within what
we call Heritage mode.
For the non-Heritage mode the goal is to try to move things forward
computationally.
Also, an interesting side-effect of some of these changes
is that non-CA people will find it increasingly easy to use CA
markings, since there is
no longer any computational separation between CA and CHAT.
-- Brian