Dear all,
for an analysis of spoken language I have an ELAN-transcription with the following tiers: a speaker tier, a tokenized tier of the speaker tier and a “noun”-tier. The speaker-tier is an independent tier. The tokenized tier is subordinate to the speaker tier and the “noun”-tier is subordinate to the tokenized tier. The subordinate tiers have a linguistic type with Time Subdivision.
My objective is to have an annotation for each noun on the “noun”-tier. For most annotations it works fine.
Now I have encountered an annotation which seems to have a time-alignment-problem within the annotation itself. For the first and the last few words of the utterance the annotations on the noun tier work just fine. But if I try to enter an annotation for one of the words in the middle of the utterance in the noun tier, there is an error message that a time-aligned annotation could not be produced because the superordinate annotation is not completely time-aligned.
I am puzzled that this is only the case for some annotations and for some words within those annotations.
The puzzling utterance in question is represented by one continuous annotation on the speaker tier which has been tokenized (default-setting) in the tokenized tier. I am currently working with ELAN version 5.0.0 beta.
Is this a bug or did I make some mistake in the order of tiers or the tokenizing?
Thank you in advance! Appreciate any kind of feedback.
It is difficult to tell based on your description, but it could be that some annotations on the token tier are not (yet) fully aligned. If you use the Tokenize Tier function and the destination tier is a Time Subdivision tier, all annotations on the token tier are not fully aligned; this is shown in the user interface by the shorter vertical lines between those annotations (as compared to e.g. the annotations on the speaker tier). The Figure 2.101 on this page shows these shorter lines. When you manually change the alignment of these annotations the vertical lines will be longer.
As long as the token annotation is not fully aligned (left and right boundary) it is not possible to create a dependent annotation on a Time Subdivision dependent tier (the noun tier in this case).
If my understanding is correct it would mean that you have been aligning some token annotations, for which you could then add noun annotations. If not, I’m misunderstanding the situation.
I would guess that you could use a Symbolic Association tier type for the Noun tier; you probably will add only one noun annotation per token? If you would use that type, I guess it is possible to add depending annotations for the ones you can’t now.
Thank you very much! The problem is indeed that the tokenized annotations were not time-aligned (all annotations have short vertical lines between them).
I changed the tier to an independent one without Time Subdivision and now it works for all the annotations that I couldn´t edit before.
Problem solved. Thanks again!
Ok, good if it works.
Alternatively you could tokenize to a Symbolic Subdivision dependent tier; you would still have the dependency relationship between the tiers and you can add noun annotation immediately without having to worry about time alignment of token annotations.
