In the dialog window below (which pops up as the first dialog for most of the multiple file operation) do one of the following:
*.eaf
) to select
it. It now appears in the rightmost box. Alternatively, you can click on the
annotation file name and click the >>
button.
Repeat this for every annotation file you want to include.
It is also possible to select a complete directory. All
.eaf
files in a selected directory will be
included.
Normally ELAN allows to edit only a single file at a time. There are situations in which it is convenient to edit multiple files at once. The menu item File > Multiple File Processing gives a number of options to do just this. When selecting either of them, you are warned that you should have copies of the files you are going to work on in case you want to restore the files (there is no Undo for multiple file edits).
When you choose this option Create Transcription Files for Media Files..., you see the following dialog.
Options :
To specify a separator in the file name to identify the suffix or prefix, check this option Specify custom affix separator character ('-' and '_' are built in).
Click on Start to create the transcriptions based on the options set.
The option Edit Multiple Files... shows, after clicking Yes in the warning dialog mentioned above, the Multi File Editor. The first thing to do here is to load a domain by clicking Load domain. Loading a domain is the same as for the Scrub Transcriptions... option. To be able to load a domain you must of course have created one beforehand (see How to select multiple files). After loading a domain, the data is shown in the table. In this table you can edit tiers on the Tiers tab and tier types on the Tier Types tab.
To edit a name, annotator or participant of a tier, double click the corresponding table cell or select it and start typing. To change the tier type of a tier, select one from the drop down menu. You can add a tier by clicking Add tier, add a depending tier by clicking Add dependent tierand remove one by clicking Remove tier.
If there are hierarchy inconsistencies (e.g. if a tier in one file does have a parent while a tier with the same name in another file does not) removing tiers is not possible. The button Remove tier is therefore greyed out.
On the Tier Types tab, the name of a tier type can be changed by double
clicking the corresponding table cell in the Type Name column.
Changes made in the Tiers and Types tabs are applied to all the files in the domain after clicking the Save changes to domain files button.
When you choose Scrub Transcriptions..., you first need to specify a new domain or select an existing domain. This option helps you to "clean" the annotation files (*.eaf) of possible tabs or white space characters which are often overlooked by the user but are still saved in the file. To select, create or delete a domain see How to select multiple files.
In the next dialog, you can specify what characters to delete, new line characters, tab characters and/or white space characters, and in what position these characters have to be. Click Start to start the scrubbing process. The progress of the scrubbing is shown in the progress bar.
The option Annotations From Overlaps... for multiple files is the same function as annotations from overlaps in the current open file (see Creating annotations from overlaps ), but applied to a selection of files. The first step allows to select a custom set of files in a file browser or to load a set of files that have been stored as a domain. For loading or creating new domain see How to select multiple files The list of tiers is the sum of all tier names encountered in the selected files. The options in the next steps are the same, clicking the Finish button in the last step the new tier is created and populated with annotations in all files of the domain.
In some projects two or more annotators annotate the recordings and the need exists to assess the level of agreement, in order to be able to improve the training of the annotators and in the end the annotation quality. Several algorithms have been implemented and are being applied in research projects but there doesn’t seem to be consensus on what the best approach is for this type of data (time aligned annotations). The option Calculate inter-annotator reliability... allows to calculate an agreement measure for annotations on multiple tiers in multiple files, providing three different methods based on existing algorithms. Even though multiple tiers can be selected, the comparison is always performed on pairs of tiers. These methods are provided “as is”. Implementation of other algorithms might be added later, if time allows. The three methods are:
Using the Calculate inter-annotator reliability... function requires a few steps to be taken. Some of the steps differ depending on the choices made; other steps are common to all methods. The steps and their availability depending on the choices made are described below.
The following sections will describe each step of the process in more detail.
The first step in the process allows you to select a method of comparing. Depending on what method is selected, the steps after that will differ. The options to choose from are:
This option implements (part of) the Holle & Rein algorithm as described in this publication: Holle, H., & Rein, R. (2014). EasyDIAg: A tool for easy determination of interrater agreement. Behavior Research Methods, August 2014. The manual of EasyDIAg can be consulted for a detailed description and explanation of the algorithm.
This option provides a modified implementation of Fleiss' kappa. The modification concerns (as is the case for the modified Cohen's kappa implementation) the matching of annotations to determine the "subjects" or "units" and the introduction of the "Unmatched" or "Void" category for annotations/events that are not identified by all raters. If the raters only have to apply labels to pre-existing segments, the problem of matching annotations does not exist. Fleiss' kappa works for two or more raters (the other options are limited to two raters).
This is a simplified version of the function that used to be under Tier > Compare annotators.... It calculates a raw agreement value for the segmentation, it doesn't take into account chance agreement and it doesn't compare annotation values. The current implementation only includes in the output the average agreement value for all annotation pairs of each set of tiers (whereas previously the ratio per annotation pair was listed as well).
This will compare the annotations (the segmentations) of 2 annotators using the Staccato algorithm. See this article for more information on the Staccato algorithm: Luecking, A., Ptock, S., & Bergmann, K. (2011). Staccato: Segmentation Agreement Calculator according to Thomann. In E. Efthimiou G. & Kouroupetroglou (Eds.), Proceedings of the 9th International Gesture Workshop: Gestures in Embodied Communication and Human-Computer Interaction (pp. 50-53) .
Once you choose a method, click Next to continue. Note that when the Kappa or Staccato are chosen, the next step will be 'Customize compare method'. Otherwise the next step is 'Document & tier configuration'.
When the modified Cohen's kappa method is chosen, this step allows you to specify the minimal required percentage of overlap. This is the amount of overlap as a percentage of the duration of the longest of the two annotations. The higher the percentage, the more the annotations have to overlap to match.
You can choose to generate and export agreement values per pair of tiers, in addition to the overall values. Since this algorithm compares annotation values as well, it is best to select tiers that share the same (controlled) vocabulary. When done, click Next.
The options for Fleiss' kappa are slightly different; the slider here allows to specify a percentage between 1 and 100. Since there can be any number of raters, the annotation matching algorithm tries to create clusters of as many overlapping annotation as possible, taking into account the required average of the percentages of the overlap and each involved annotation's duration. The figures below try to illustrate the problem.
Figure 1.142. Six raters, two possible clusters of four and six matching annotations, the overlap in light blue
Figure 1.143. Six raters, four of the possible clusters of matching annotations are marked in blue and green
The algorithm gives preference to clusters with more annotations, as long as the required average percentage of overlap is met. If not, a cluster with less annotations is selected. Each annotation can only be part of one cluster.
The Also export matrices checkbox allows to not save the tables of values (see the worked example at Wikipedia), but it is recommended to accept the default.
When you've selected the Staccato algorithm as the compare method, the settings as shown above will appear. You can customize the settings for the Staccato algorithm here. This algorithm takes chance into account by comparing the segmentation with a series of randomly generated segmentations, the Monte Carlo simulation. The nomination length granularity determines how many memory slots for segments of different length will be used internally. For more in-depth information regarding these settings, please see the reference article mentioned before. When done, click Next.
The next step is to configure where the tiers that you want to use for comparison are located and how they should be paired. In the upper part of the dialog, you select the location of the tiers:
In the lower part of the dialog, you can select in what way the pairing of tiers to compare is done:
When done, click Next to continue.
Next, you will select which files you want to use for comparison and how to match tiers.
The screen above shows all possible options. The options available for this step will differ depending on the configuration you made in the previous step. It will not be available when the option 'in current document' together with 'based on manual selection' was chosen in the previous step.
In this last dialog, you will select the tiers used for comparison. The layout will be different, based on what you selected in previous steps. The screen above displays the dialog when you've chosen the option 'In the current document' & 'based on manual selection' in step 3. You can manually select which tiers to compare.
The dialog above will appear when 'based on pre/suffix' in step 3 is chosen. Marking a tier from an annotator will result in a highlighted corresponding tier in the lower part of the dialog. When 'based on same tier name' was chosen, you can only select tier names, corresponding tiers will not be visible in the dialog.
Finally, click Next or Finish to save the output text file to a location on your computer.
The option Annotations From Subtraction... for multiple files is the same function as annotations from subtraction in the current open file (see Create annotation by subtraction ), but applied to a selection of files. The first step allows to select a custom set of files in a file browser or to load a set of files that have been stored as a domain. For loading or creating new domain see How to select multiple files The list of tiers is the sum of all tier names encountered in the selected files. The options in the next steps are the same, clicking the Finish button in the last step the new tier is created and populated with annotations in all files of the domain.
This function Statistics for Multiple Files is similar to annotation statistics for the current file (see Annotations Statistics). The main difference after selecting the files in the domain is that it is possible to select which tiers to include in the calculations. The tables in the tabs do not have the column showing the total annotation duration as a percentage of the media duration but most do have a column for the number of files a certain value (tier or type name etc.) has been encountered in. After changes in the selection of files or in the selection of tiers the Update Statistics button needs to be clicked before the new calculations are started.
Since ELAN 4.7, you are able to do an N-Gram analysis over multiple eaf files. This functionality has been developed by Larwan Berke, you can find an extensive PDF document about this implementation on the third-party resources page of ELAN: https://tla.mpi.nl/tools/tla-tools/elan/thirdparty/
When you first open the N-gram analysis, a new dialog window will pop up that contains the various options for the search and the resulting table showcasing a few statistics.
The first step is to select the search domain, see How to select multiple files. Once that is done a list of tiers seen in the domain will be shown. A note of caution: the code assumes that all files in the domain will contain the same tiers. It then loads the first file in the domain to extract the tiers and display it in the window. Check the tiers you want to analyse.
Next, define the N-gram size in the text box. The software can handle any positive size greater than 1. When set, clicking the “Update Statistics” button will start the search and will calculate the statistics. The annotations are extracted from the files, N-grams created from them, and finally collated into groupings of same N-grams for statistical analysis. When done, you will see a pop-up window with a process report. If there were any errors, they will also be displayed here.
When the search is done, the result table will be displayed in the main window. Some of the columns from the data are visible here, however only a small subset is displayed simultaneously to avoid overcrowding the GUI. The visible columns are: N-gram, Occurrences, Average Duration, Minimal Duration, Maximal Duration, Average Annotation Time, and Average Interval Time.
The first column shows the N-gram. The vertical marker “|” separates the annotations contained in the bigram. For example, if a trigram was selected it would show something similar to “FINISH|READ|BOOK” and so on for larger N-gram sizes.
Finally, in order to see the entire data that was produced it is necessary to export the results into a text file for further processing. This is done by clicking on the “Save” button and a dialog will pop up asking the user where to save the data. It is exported in a CSV-like format (Comma-Separated Values). The CSV file uses tabs “\t” as the delimiters and newlines “\n” as the record separators to avoid ambiguity with the values. A sample row is: “HOLD|IX-1p\t7.9934\t0.348\t0.13754 ...” and contains numerous columns.
Furthermore, it is possible to export the N-grams individually in order to process it separately from ELAN. The data is exported by clicking the “Raw Data” button in the GUI. After supplying the file the data will be exported in the same CSV format as discussed above. For more in-depth information about the N-Gram analysis function and the resulting data, please consult the PDF mentioned earlier on the ELAN website, or consult it here: https://parasol.tamu.edu/dreu2013/Berke/images/DREU_Final_Report.pdf (This link may become outdated at some point).
Apart from exporting part of a clip in a project (see Media clip using script), you can also export multiple clips from the same (or multiple) projects. The clips will be clipped based on the annotation-times on a specified tier.
As with the exporting of a part of a clip, Windows users will need to put a copy of ffmpeg.exe or ffmbc.exe in the program folder of ELAN. ( see Media clip using script) for more info.
To utilize this function, you will need to create a tab-delimited text file first. Go to File > Export Multiple files as... > Tab-delimited text.... (note, this will not work for the single file tab-delimited export, as there is no option to include the videofile-path). Choose a domain, or create a new one (if you want to create multiple clips from only one .eaf, create that .eaf as a new domain). Select the tiers you want to include in the Tab-delimited text file (each annotation on a tier will result in a clip).
Make sure you check the following options:
Under the time column and format options, you will need to check:
The other options have to remain unchecked. Next, click OK and the file will be exported to a text file.
Go to File > Multiple File Processing... > Create Multiple Media Clips.... Choose the exported tab delimited text file you just created, and specify a folder to save the clipped videos to. Click OK to start the process. When done, a process report dialog will appear with information about the clipping process.
Similar to merging tiers within a single project (see Merging tiers), this function allows you to merge tiers in multiple projects. This means the merged tiers will be added to each project you select in the process.
To start, click File > Multiple File Processing... > Merge Tiers.... You will be presented with a dialog in which you either select the eaf files from the file browser, or select files from a domain. When done, select the tiers to use for the merging process.
Next, select the merge criteria, either regardless of the annotation values or according to specified constraints within the annotations of a chosen tier. When checking the option Only process overlapping annotations , ELAN only merges annotations that have the same value. In this case, the values of both annotations are not concatenated, so the created annotation contains the value only once.
In the next step, set a name for the destination tier and decide whether this tier will be a root tier or a child tier. Also select or create a tier type for the new tier.
Lastly, specify the value for the destination tier. You can set a value in a time format, which will put in the specified time values inside the annotation units on the new tier. You can also choose to set a specific value to be filled out into the annotation units. The final choice is to concatenate the values of the annotations from the tiers you have selected for merging.
After clicking Finish, the tiers will be merged and inserted into each eaf you chose at the start. A process report will show an overview of what has been done.
If one or more of the tier types of a transcription are linked to ECV's (External Controlled Vocabularies, see Using an External CV) and annotations have been created using those ECV's, it might be necessary to update existing transcriptions after changes have been made to those ECV's. This function allows to update an entire corpus or a selection of files.
If you choose the option Update Transcriptions for ECVs, you'll see the following dialog:
First a source folder containing the transcription files should be selected, specifying whether or not sub-folders should be processed too. For the destination it is possible to choose to overwrite the existing source files (this should only be done if there are recent back up copies of the files) or to select a folder where the updated transcriptions should be stored. Furthermore it is possible to specify which content language to use; this is only useful if the ECV's are multilingual.
The default behavior of this update action is to change annotation values after changes in an external controlled vocabulary, based on the reference of the annotation to (the id of) an entry in the controlled vocabulary. The Don't change the annotation value... option allows to change this default behavior; instead of updating an annotation value based on a reference to a CV entry, it checks if the annotation value is still in the controlled vocabulary and, if so, updates the reference or, otherwise, removes the reference.
When the files of a corpus are based on a template, the need can arise during the project, to update all files with new tiers, tier types and/or new controlled vocabularies. To achieve this, the template can be updated with the new elements and the changes can then be applied to the files. It is also possible to create a new template with only the new tiers and types etc. and use that one for the updating, because new elements in the template will be added but existing tiers, types and CV's etc. in the .eaf files will not be deleted if they are not (or no longer) in the template.
To start the process of updating a set of files, choose Update Transcriptions with Template... A message will be shown, warning that there is no undo for the changes that are going to be made to the files. Then this dialog will be shown:
The following options are available:
.txt
file
Apart from adding new elements to the existing files, this process also allows to update some properties of existing elements, as long as these changes can't result in data loss. E.g. the Annotator and Participant properties of a Tiercan be updated, but not the Parent Tier. The Tier Type of a Tier can only be changed to a Type with the same overall constraints (Stereotype). The Controlled Vocabulary property of a Tier Type can be changed, but not its Stereotype. Controlled Vocabularies can be converted from internal to external or the other way round. Etc.
ELAN offers the possibility to export multiple annotation files as one file. To do so click on File > Export Multiple Files As... and one of the options.
To export multiple files as toolbox files, click on File > Export Multiple Files As... > Toolbox file(UTF-8).... This process involves 3 steps.
From the drop down list select the tiers to use in the overlaps computation. You can select all the tiers displayed in the list if you click on Select All, or deselect them if you click on Select None. Once you have made your choice for the tiers for which the overlaps should be found, you can select next, this will bring you to the next step.
In this step you can define output settings and the Toolbox options. The option are more clearly defined in Toolbox file(UTF-8)
To export multiple files as FLEx files, click on File > Export Multiple Files As... > FLEx File.... This process involves 4 steps.
In this step you can select a tier type to use for the 'morph-type' tiers. It's also possible to uncheck this, if not needed. From the dialog, you can also map the tier types to the different items, which are listed on top.
In the next dialog, you can specify the element-item tier type and set a language for it. ELAN can try to extract that from a tier name, (if the box is checked) but it is also possible to add (or remove) a value for a language or type. To do so, enter a value ('en' in this example) and click Add. Then, you can select the added value from the drop-down menu under 'language'. You need to set a type and language for every Tier Type Name in order to be able to go to the final step. For more information on the structure of FLEx, see Figure 1.67, “FLEx to ELAN structure”.
To export multiple files as praat textgrid, click on File > Export Multiple Files As... > Praat TextGrid .... This process involves 2 steps. See Praat TextGrid file for more details.
To export multiple files as Tab-delimited Text, click File > Export Multiple Files As... > Tab-delimited Text....
To export multiple files as List of annotations, click File > Export Multiple Files As... > List of annotations....
To export multiple files as Tab-delimited Text, click File > Export Multiple Files As... > List of Words....
There could be situations in which you want to discard or select tiers from
multiple .eaf
files, for instance if you want to present a third
party with a limited number of tiers. To do so, select File > Export
Multiple Files As...
>Selected Tiers as EAF.... In
the first dialog(see How to select multiple files) you can select
the files from which you want to export a selection of tiers.
Once you have selected your files, Export Tiers from Multiple Files dialog appears.
To export, do the following:
Export parent tiers of the selected dependent tiers
automatically
or to Only export dependent tiers if
their parent tiers are selected
.
Save files with original names
of to
Make use of suffixes
. In case of the latter, you can
specify whether to save the files with their original name followed by a
suffix or to save the files with a new base name and followed by a suffix
number.
This function is available via menu File->Export Multiple Files As-> Annotation Overlaps Information....
This function allows the user to select one “reference” tier and multiple other tiers that will be compared (sequentially) with the reference tier. The comparison is done on the level of the annotations.
The following information will be present in the resulting tab-delimited text file:
Column 1- 4:
Begin time | End time | Duration | Reference Tier Name |
These columns will contain information for all annotations of the reference tier. The annotation values are in the column with the tier name as the header, the time info in the first 3 columns.
Next for each “comparing” tier there will be 11 columns, the header of which consists of the tier name and a suffix and the column contains the following information:
1 | Name-ov | 0 or 1, whether there is an overlapping annotation or not (0=no, 1=yes) |
2 | Name-same | 0 or 1, whether the overlapping annotation has the same value. If there are more than one overlapping annotations the value is 0 (0=no, 1=yes) |
3 | Name-ov-dur | The duration of the overlap, the total overlap duration in case of more than one overlapping annotation |
4 | Name-no-ann | The number of overlapping annotations |
5 | Name-value | The value of the overlapping annotation, concatenated, comma separated, in case of multiple overlaps |
6 | Name-bt-to-bt-After | The amount of time from the beginning of the reference annotation to the beginning of the first non |
7 | Name-et-to-bt-After | The amount of time from the end of the reference annotation to the beginning of the first non overlapping annotation |
8 | Name-et-to-et-After | The amount of time from the end of the reference annotation to the end of the first non overlapping annotation |
9 | Name-bt-After | The begin time of the annotation on the comparing tier after the reference annotation |
10 | Name-et-After | The end time of the annotation on the comparing tier after the reference annotation |
11 | Name-value-After | The value of the first annotation on the comparing tier after the reference annotation |
After the header, for each file there will be the following information/data:
All time values are in milliseconds.
Theme© (www.patternvision.com) is an application for detection and analysis of hidden patterns in time (so called t-patterns) in behavioral data. It is possible to export annotations of selected tiers such that they can be imported into a Theme project. A Theme project requires at least two text files:
When exporting, ELAN creates the “vvt.vvt” file and for each transcription file it creates a raw data file, converting each annotation into two records for the raw data, one for the begin time and one for the end time of the annotation.
To start, click File->Export Multiple Files As-> Theme Data Files....
If a file from the selected eaf files is currently opened, you will be presented with a warning. Make sure you have saved your current transcriptions before starting this export process as local changes will be overwritten.
ELAN offers the possibility to import multiple files at once and save them as *.eaf files. To do so click on File > Import Multiple Files As... and one of the options.
To import multiple toolbox files for conversion to *.eaf, click File > Import Multiple Files As... > Toolbox file(UTF-8).... This operation consists of 3 steps.
When the operation has completed, you will be presented with a process report. The multiple *.eaf files are now ready to be used in ELAN.
Multiple TextGrid files created in Praat can be imported and converted to *.eaf files. This process involves 3 steps.
In this dialog, you can choose to include Praat PointTiers and if empty annotations or intervals should be skipped or not.
When the operation has completed, you will be presented with a process report. The multiple *.eaf files are now ready to be used in ELAN.
To import multiple FLExtext files for conversion to *.eaf, click File > Import Multiple Files As... > FLEx file.... This operation consists of 3 steps.
You can select whether to use the 'interlinear-text and 'paragraph' element in FLEx, import the participant information and what the smallest time-alignable element should be: 'phrase' or 'word'. Choose on what level you want to create tier types and set a duration per phrase element (required).
Finally, configure how and where to save your files. You can choose to save with an .XML or .flextext extension, and you can skip files that would result in having no tiers.