Non-contiguous annotation
Up one level
Non-contiguous annotation
Hi there,
Using controlled vocabularies, is it possible to do non-contiguous annotation in the same tier? For example, if a person says "Is non-contiguous coding ummm possible?" I would want to code that entire utterance as a question, but also code the "um" as a moment of disfluency. I know it's possible to make a second tier where I could mark the time the "ummm" takes as disfluency, but we're looking at a bunch of factors that aren't all mutually exclusive, so to do it that way would require dozens of tiers when we really only need a few separate ones. Is it possible to put "disfluency" (for example) in the same tier as the other type of speech acts we're looking at, highlight the entire utterance and mark it as a question, and then highlight the "ummm" and mark it as a disfluency using controlled vocabularies?
To get a better understanding of why we want this, we want to be able to count how many complete questions we have, as well as how much disfluency we have. To code the first half of the utterance as a question, code the "ummm", and then code the rest of the utterance as a question would mess up the count.
Thank you! Any suggestions would be extremely appreciated!
-Samantha
Re: Non-contiguous annotation
In ELAN it is not possible to have overlapping annotations on any tier. So if you want the whole stretch marked as a Q (question) and part of it as D (disfluency), that cannot be done on one single tier.
If the time alignment of e.g. the disfluency parts are not important and the number of factors you are looking at not too large, you could create a controlled vocabulary with not only the single factors (Q, D) but also the combinations (QD, question with disfluency). You could then use the search functions to filter and count any factor alone as well as any combination of factors. Would this be a workable solution?
-Han
Re: Re: Non-contiguous annotation
Thanks for the reply! We'll end up doing something like this. Appreciated!