Searching on several tiers?
Up one level
Searching on several tiers?
Hi, is it possible to do a search on several tiers? For example, I code gestures. One tier gives the type of gesture, for instance "representational". A second tier gives the semantic function of the gesture, for instance redundant or supplement. Can I have the number of the representational gestures which are redundant? Moreover a third tier gives the distance between the gesture and the associated clause, for instance D = 0, D= 1. Can I have the number of representational gestures, redundant and D = 0?<br>Thanks,<br>Audrey<br>
Re: Searching on several tiers?
Yes, that is possible. There are two options: search in a single file (the file that you have opened), accessible via Search->Find (& Replace), and search in multiple file, accessible via Search->Structured Search Multiple eaf.
The procedure for the first option, the single file search, is described in the manual here:
http://www.lat-mpi.eu/tools/elan/manual/ch07s01.html
You can construct the query by adding "Constraints". In your case you can start with gesture type tier, add a constraint in which you define the semantic function you're interested in and in a third constraint you can have the distance. As long as you are mainly interested in the number of hits, this variant will do.
The procedure for the multiple file search here:
http://www.lat-mpi.eu/tools/elan/manual/ch07s05s03.html
Here you can construct the query in "layers", one layer for each tier. In your case 3 layers or rows and one column in the "Multiple Layer" tab should do. Exporting all information in the hits to tab-delimited text is possible here, so including the values of the annotations on the semantic function and distance tiers.
The examples in the manual will hopefully clarify the procedure.
-Han
Re: Re: Searching on several tiers?
Thanks a lot for your help,<br>Audrey<br>
Re: Re: Searching on several tiers?
hi, another problem is appeared. I have made a test with one transcription. I have coded the transcription with the four tiers : nature, relation, sychronicity and distance.<br>I can do a search on theses four tiers : 2 columns, 4 layers (the first "nature" with a strict mention "Représentationnelle" and the others with the code ".+" because I want to have only the representationnal gestures with all the values in the three others layers). But when I click on "save hit statistics", the results are false... the results says that in the transcription there are 27 occurences whereas in reality there are only 9 representationnal gestures in the transcription file... I don't understand why! Can you help me one more time? Thanks, Audrey<br>
Re: Re: Re: Searching on several tiers?
it seems that the problem is the option ticked: I ticked "no constraint" and my results were multiplied by 3 (I have 27 occurences instead of 9)... but if I ticked "fully aligned" the results is good but when I open my statistic files I have only 6 occurences instead of 9... :-/ I'm waiting your help!<br>
Re: Re: Re: Re: Searching on several tiers?
Are you saying that in the result list in ELAN you see 9 hits but when you export and open the file the total number of occurences is 6 hits? That would be srange.
In order to understand the situation better it might be good to have your file(s) with screenshots of your settings. Maybe there is an explanation for the result, maybe there is a bug in the export, difficult to say.
If you want you can send files to han.sloetjes AT mpi.nl
-Han
Re: Re: Re: Re: Re: Searching on several tiers?
yes, I know... it's very strange. I have sent you a mail with screenshots. Thanks a lot, Audrey<br>
Re: Re: Re: Re: Re: Re: Searching on several tiers?
From the examples it appears that difference in the exported result can be explained by the "Separate hit count per hit value" option in the "Save hit statistics" window. With this option selected, each unique hit value is only once in the output with a number in the "Occurrences" column indicating how often the hit occurred. The sum of the values in that column is, correctly, 9.
-Han