My team is using a custom Hunspell-format dictionary (dic & aff files) to do spellchecking of transcripts in ELAN. We have two questions:
Is it possible to see the underlined typos in Transcription mode? This would make it easy to identify spelling mistakes in real time, as opposed to switching back to Annotation mode and clicking on the Text window to review them.
In Annotation mode, inside the Text window, typos are underlined in red but only after a small delay. We assume ELAN is searching for typos in the background, and this is causing the delay. Our real problem is that once we find a typo and double-click on the annotation to correct the spelling, all of the red underlines disappear, as if the process is being refreshed and ELAN goes searching for typos all over again. This adds considerable time to the process of fixing spelling mistakes. Is there a workaround?
Thank you for the feedback. Concerning your points:
That would make sense indeed. I don’t know what it would take in terms of development time (the current spell checking features are based on external code contributions), we’ll have to look into that.If this isn’t on the wish list already, I’ll add it.
The checking and underlining of typos is indeed performed in a background thread. Like other annotation viewers, the Text viewer has its own characteristics and requirements, making it relatively slow in updating after a change in annotations on the selected tier. Currently the whole view, including the different types of highlighters, is rebuild from scratch after a change in the annotations. In principle that wouldn’t always be necessary, but I can understand (given the data structure backing the current view) this choice was made over implementing a partial update. From a user perspective, there is no workaround for this problem, as far as I can see. From a developer perspective, I believe it will be a considerable effort to improve this performance (based on a quick look at the code).
I’m currently looking into spell-checking transcriptions in ELAN too.
Has anything happened with regards to adding spell checking to the transcription mode? Are you planning on adding it?
I’ve integrated the German Hunspell-format dictionary (dic & aff files) for spell-checking our transcriptions and am encountering issues with the spell-check results. Several words that I find in the dictionary file are marked as incorrect while some words that aren’t in the dictionary are not marked. Also, some words are only partially marked or the red squiggle is interrupted within a word. Is this an ELAN bug or somehow related to the dictionary? I’m using ELAN 6.8 on Windows 10.
Regarding spell checking in transcription mode, it has been added in the wish list to be implemented but we cannot guarantee the timelines.
Regarding the issues you are encountering with spell-check results, it has been pointed out earlier and found that highlighting is indeed incorrect for ‘special’ characters.
It was to do with java version and regular expressions. We have fixed the issue. It will be realised in the new version of ELAN.
