Forum: Bug Reports
Romanian - keyboard diacritics registered as errors
Hi Memory team, With the latest update, diacritics are now required for Romanian words (thanks a mil for fixing this issue!). However, I have encountered an issue with ș and ț - when typing these using the Romanian keyboard, they are marked as errors. It is possible I might have copy-pasted these particular words, or that when adding them to the course I used the on-screen diacritics, but I am not sure. You can check my Romanian course for the word - deferență. For the word containing an ș, șild, I have already used the My answer was correct option to have the answer accepted even when typed directly from the keyboard. Let me know if you need more details. Thanks!
Small update - it seems that ș and ț are displayed twice in the onscreen diacritics menu (Memory menu). For ț, the answer was marked as correct only when using the second ț symbol.
First, thanks for pointing out the duplicate characters - they will be fixed in the next version. I created a course with 2 words: deferență șild I see the problem -- it seems to marked OK when typed with accented letters, but it should say "Watch out for the accents" if I type deferenta for deferență ... so that's a bug. I am looking into this to see if we can fix the accent detection for this language
Actually, I may have misunderstood. When you next get a question like this, could you please hit "Report question" from the play screen in the bottom right?
I reported both terms using the Report button from the Learn screen. Apparently removing the additional diacritics from the special characters menu solved part of the problem - the words are now registered as correct but the accent check screen is triggered even if the words are correctly spelt (all diacritics in place). I can also provide screenshots with the issues if needed. Thanks a mil for looking into this!
Thanks for highlighting this in detail. We will fix this. It seems we have a few really nasty problems with certain characters in various languages that, up until now, have never been recognised 😦 It's high up the list, but a bit trickier than I thought - thanks for pointing it out.
No problem at all - for now, the partial fix works quite nicely and for other cases, using the "My answer was correct" feature as a workaround also works. Thanks for looking into this 😃
Hello, I think I have found the problem finally. In the click-keyboard we had the wrong T for Romanian. We had ţ instead of ț (they look the same in this font, but if you copy and paste them in an editor you will see they don't match), and the same issue for the ș I believe they should have been exactly: ă â î ș ț They are now updated to the above in the next release (later today or tomorrow) on the click-keyboard. But a problem remains -- Romanian courses /questions made using the old keyboard are going to have this problem unless edited. If some questions are typed by keyboard and others with the on-screen, it's going to be a bit mixed up 😦 We could edit them all in batch, but want to be confident that this is definitely the solution before modifying users' courses.
Ok, 90% sure I have fixed this now to the extent it should work for historical courses too rather than having to edit the questions. Have added a "custom compare diacritic normalisation" function when marking answers 😃 So after this, diacritics with VERY similar looking (indistinguishably similar) symbols (but perhaps from other languages or other keyboards) should be marked as EXACT matches. But a t in place of a ț will be an accent check message. I need to be careful with what I add, but have added Romanian formats based on the previously broken keyboard. If that doesn't fix the issue entirely, please let me know and copy and paste them exactly here (e.g. [ğ] [ġ]), as I can easily add more canonical characters, e.g. It won't be live until the next version though, 1 - 2 days. I will update then.
Thanks again for going through all this trouble, much appreciated, especially for a language with so few users 😃 I got an update notification today so I am assuming that the fix from your latest comment is already live. I went through the problematic words and I still get accent check message for [ș] and [ț], both when typing and when using the on-screen characters (click-keyboard). The exact words for which this happened (copy-paste from the edit screen) are [deferenţă] and [şild]. When I paste these in Word, for example, it seems that indeed these are different characters (even if the same font, Calibri, is used) - the one pasted from the course is part of the Symbols Subset called Latin Extended-A, the output on my keyboard is part of Latin Extended-B. Not sure if this info helps in any way though. As I was starting to wonder if my desktop keyboard was the issue, I also tested these on my laptop and iPhone (using Safari and the Romanian keyboard) but with the same result... Please let me know if I can help with any other details.
Hi I'm so sorry, I should have said to save you having to spend time trying to figure out what's wrong. It's not live yet, sadly we were unable to do a hotfix on it as it's not in a batch of other pending updates.... we are updating speaking and typing and the character change conflicts with a few things so couldn't put it live yet. It's going to be a few more days before it's good to go.
No problem at all 😃 Just let me know once the fix is live so that I can re-check.
One of these terms just came up in a practice session today and the keyboard diacritics were accepted by the system with no issues. Closing this as fixed. Thank you!
This post is closed