Hello, when adding new participants (Tier >>> Add new participant), not all the original settings of the tier are preserved. Although the general set-up of the tier (tier name, linguistic type, content language etc.) is properly copied, any settings made via the “More Options” button, such as the font type and the font colour, are not copied into the tiers for the new participants. When making such changes in the tier (i.e., through More Options), the choice is given to make these tier settings for “all tiers with the same tier type”, “all dependent tiers” and “all tiers with the same participant”, but not “the same tier for all other participants”. In other words, all these tier settings need to be manually changed for each additional participants for each tier. Is that correct? This is quite cumbersome, especially if you have many different transcription and translation tiers using different fonts and font colours. Cheers, Tim.
Hi Tim,
If I understand correctly you address two different, though related, issues here:
- visualization/rendering settings of tiers are not copied when using the “Add New Participant” function. I can reproduce this, I would have to look into the code to see if this has not been implemented or if there is a bug in the implementation. These additional rendering settings are not stored in the eaf but in the preference pfsx file, copying these settings requires a separate step, it might well be that this has never been implemented.
- when applying rendering settings for a tier (through More Options), 3 choices are offered for applying the same settings to other tiers. You propose to add a 4th option which, if I’m correct, would come down to finding all tiers with the same name except for a prefix/suffix. I think the assumption behind the 3 options was, that “tiers with the same type” would cover the situation you describe, namely that “the same tier for other participants” will probably have the same tier type. Obviously this doesn’t always have to be the case. From an implementation point of view “the same tier for other participants” is special because it would require additional UI elements to specify if the tier name matching should be based on prefix or suffix etc.
Correct me if I’m wrong.
Dear Han,
Yes, your analysis is correct, except for me, as a user, I do not see those two issues as ‘different’: just like I want to assign a certain content language to a specific tier (for example, a free translation tier in zh - Chinese on tier type ‘phrase-item’, just ignoring already the fact that FLEx code zh does not exist in ELAN but is zho there, which we discussed and worked around earlier), I also want to assign a specific font to that specific tier (for example, SimSun) and a certain colour (for example, 999900). I want that content language, font and colour to be the same for all participants that I add (A, B, C, …) for that specific Chinese free translation tier. Whereas the content language is automatically copied when adding new participants, the font and font colour are not: I have to set these two manually again for the Chinese free translation tier for each additional participant. I, specifically, do not want this font and font colour to be the same for all “tiers with the same type”, because I have other tiers with the same type ‘phrase-item’ (e.g., a free translation in Hindi hi, i.e., hin) that I do not want to have in the SimSun font and font colour 999900, but in the Mangal font and font colour 9900CC. In addition, I would preferably want these specific font settings to appear in the .etf template file, so that any new recording that I open as a new .eaf project (with a recording and the template) will already have those font settings for each participant.
It seems I am bugging you with really uncommon requests, which in a way is worrying me, because it makes me think I do something completely wrong, and that there must be a much easier way to deal with multiple languages and multiple writing systems in ELAN and in the ELAN to FLEx conversion: I can’t imagine that other people working with such a set-up have had to manually adjust the font settings for each tier for each participant for each new recording over and over again?
Your views are highly appreciated.
Thanks, Tim.
Hi Tim,
First off, it’s clear that from a user’s perspective there’s no difference between the various tier properties and that should be leading. I was just ‘thinking aloud’ what the reason could be for the difference you reported (I haven’t found the time yet to look into that).
I don’t know how in general other users handle ELAN and FLEx conversion/roundtripping, but I know that there are two ELAN-FLEx workflow descriptions on our third-party resources page. Maybe you already had a look at those, otherwise you might want to do so.
Attempting to help out based on current ELAN functionality, I’m trying to get an idea of your current workflow. Am I right that you start out in ELAN, then go to FLEx and later back to ELAN? And if you add new participants in ELAN is that before or after export to FLEx?
Either case, I would guess that if you create a complete ELAN template, with all possible participants and fonts and colors, it should be possible to apply those settings (again) after a (re)import into ELAN. Possibly via Update Transcription with Template, or maybe via Preferences->Import Preferences… (don’t know if this copies the fonts and colors) or even by copying the template’s pfsx file and renaming it to the same name as the new ELAN file. Again, I may be wrong.
Concerning the ‘phrase-item’ tier type example you give, wouldn’t that be solved if, when importing from FLEx, you would select to create a new tier type for each item type and language? If you would do that once and create or modify your ELAN template based on that result, would that be a way around manually changing those settings over and over again?
Best,
Han
Dear Han,
Yes, I have consulted the existing ELAN to FLEx workflows, and also consulted with several people who have worked with multiple participants, multiple languages and multiple writing systems before. I am actually working on an integrated workflow incorporating these three variables, which is why I am so tenacious in getting this all sorted out, rather than just getting it done for my own purpose and then letting it be (the trial-and-error-till-it-works-for-me approach).
I actually set up the vernacular and analysis writing systems and a basic set-up in FLEx, but I start with a recording and with setting up the types and tiers (including multiple participants) corresponding to this FLEx set-up in ELAN. Only after the initial segmentation and draft transcription and translation in ELAN, I make the conversion to FLEx for the subsequent annotation.
If I understand you correctly, after creating a complete ELAN template and preferences file for the first recording, I can apply those same settings effortlessly to subsequent recordings by using the preferences file (not just the template file). I presume, then, that it is just a matter of inconvenience, that to create this initial template and preferences file, I would need to change the font settings for each tier manually? Let me give this a try, and see whether it works.
Best, Tim.
Well, I think that should work (update with template), but I didn’t check it yet. Yes, please give it a try (I’ll do so too next week).
Best,
Han
Dear Han,
Working on it.
Another question: Where is information regarding / the settings of “Create Annotations on dependent tiers” stored? I mean, if I enable tiers dependent on the parent tier to be annotated in this way once, I would expect this setting to be maintained automatically for subsequent times as well (new recordings etc.). However, I have noticed that this does not work, i.e., this setting is not saved in the preferences or template file, unless I have actually made annotations in a certain tier. So, every time when I open a new recording, adding the template file, and the preferences file, I would still have to indicate on which dependent tiers I want to make annotations manually again, through the option “Create annotations on dependent tiers”. Is that a correct observation?
Thanks, Tim.
Hi, I think that’s correct: when referring to the option from the Tier menu “Create Annotations on Dependent Tiers”, this is a single, manual action that can be performed at any given time (e.g. after having finished segmentation on a parent tier), but it is not a preference setting.
There is a related preference setting in the Editing tab of the Edit->Preferences->Edit Preferences window, labeled “Create new annotations on dependent tiers when a new annotation is created”, which is a global, document independent setting.
Finally, for a single, existing annotation that has become the active annotation, there is the option Annotation->Create Depending Annotations.
-Han
Hi Han,
That’s a pity. There are a lot of non-trivial choices regarding the settings to be made, which need to be explicitly stated to be reproducible. But then, it is just part and parcel of the non-standard approach I am taking here 
Cheers, Tim.
Hi Tim,
I wasn’t quite sure before if I did understand the issue correctly and I’m now convinced I didn’t.
Apparently the global preference setting doesn’t do what you aim to do, so I’m now thinking that maybe the question was if the user choices in the different steps of the “Create Annotations on Dependent Tiers” window would be or could be saved and restored the next this function is applied (either in the same document or in any document).
I think this is done for configuration windows of some functions but not for others. The reason or this difference is not always clear (accept for the fact it is quite a bit of implementation work).
-Han
Hi Han,
Like I said, it’s a pity that this needs to be done manually, but I can understand this is not a pressing issue for most (if not all) other users.
Another question. It concerns a font, that has the name Qomolangma-Uchen Sarchen. Note the hyphen in the name. All Tibetan fonts from the Qomolangma series have this hyphen.
I make a first conversion from ELAN to FLEx, which contains no font information, because I have added that font information to the More Options function in Tier Attributes, and these settings are not copied.
Subsequently, in the first FLEx to ELAN conversion, I add the analysis and vernacular language settings, and there is nothing out of the ordinary in the flextext file:
I then make a few changes to the ELAN file, but notably I do not change the language settings. The modifications within ELAN are saved in the file with the extension flextext.001, which is in the same folder as the .flextext file that I had imported from FLEx. The language settings in that flextext.001 file are expressed as follows:
I then export a .flextext file from ELAN, in which the following seems to have happened: Because of the hyphen (-) in the font name Qomolangma-Uchen Sarchen of the analysis language bo, the text behind the hyphen, i.e. Uchen Sarchen, has taken the place what in other (vernacular) font names is occupied by the value “true”:
It doesn’t give an issue in ELAN, but it does give a problem when exporting from ELAN to FLEx again, as it can’t be opened in FLEx because of this.
My understanding is, that it gives a loading error message because of the fact that the Boolean attribute field for vernacular language has a name “Uchen Sarchen” rather than just a value true. In fact, in this case, there should not be an attribute field at all, because bo is an analysis language and not a vernacular language, so there is no reason for the value true to be there. This value true only appears in vernacular languages, not in analysis languages.
After recovering the problem with help from SIL, I was able to edit the flextext file and remove the error:
I am not sure where, when and how the problem came into being. But it seems to me that the reason is the hyphen in the font name? Or do I miss something?
Thanks, Tim.
Hi Tim,
Thanks for reporting this. Your analysis is correct, this is a bug in the ELAN export function.
We’ll try to fix this for the next release.
Thanks again,
Han