We have a problem with the Vietnamese language in the dropdown menu field. Indeed our questionnaire was made in English and we translated it into Vietnamese. However, entering words in Vietnamese for search does not work for the dropdown menu questions. Moreover, our tablets are well up to date and are in the Vietnamese language.
Replicate the issue on the demo server, then tell me how to see it.
If the issue doesn’t occur on the demo server, check the differences between your server and the demo, such as the version of the software.
In the 1st picture, I was able to write the word “mì” to search food that contains this keyword.
Now I change the keyword to “ngô” and the result is bad. First, I am not able to write the correct word (2nd picture). Secondly, I don’t know why but the word I want to type is limited to two letters like “ng” and I can’t add “ô”. It means foods that contain the last keyword are not displayed.
In Viet picture, I am able to write in Vietnamese and it is the same in the webtester. So in text question and webtester, Vietnamese works well and the keyboard is showing that language too.
But in dropdown menu, I remarked that the keyboard is indicating English while the language is in Vietnamese.
How can we solve this problem since I tested in demo and the main servers and still have the same issues?
I see several possible problems in your description and I am not sure which one you mean of them, or perhaps yet a another one.
the questionnaire you are working with contains content with characters not present in the Vietnamese alphabet (link) thus you will have troubles typing them in with Vietnamese keyboard. (I am not familiar with Vietnamese alphabet nor with any transformations that the letters may undergo in combinations etc). If you want to be finding the options listed, you need to be able to type exactly those characters as contained in the options. My understanding is that these variations are various tones, but I have no idea how to enter them.
The language of the input itself switches to English when switching between questions (the language title you see on the spacebar key on the virtual keyboard in the screenshot). Strange, but should not be a deal breaker, instruct the interviewers to switch to the Vietnamese input language again, if wrong language is selected.
Impossibility to enter a particular key - not Survey Solutions driven. If you enter a combinations of characters not contained in the search list, no items will be suggested. But we don’t limit what you are typing. Correspondingly the inability to enter a particular letter is coming from the input method/keyboard.
Please note that the problem is likely related to the input method (the specific keyboard being used) and not to the Survey Solutions Interviewer App itself. Correspondingly, please try other input methods. I see some variation when I install Google’s keyboard (Gboard), though you may scout for other alternatives or recommendations of users.
Perhaps other users from Vietnam could jump in with clarifications.
@martinb was dealing with diacritics in a different language but in similar situation and perhaps can contribute.
@sergiy , would there be any interest in a feature that made combo box filtering query without accents?
In a separate area–filtering data tables on websites–the same problem arises. If a tables contains Zürich, then querying Zurich will yield no results. Even if the underlying code doesn’t help, perhaps it code could serve as inspiration. See more here and here (in filtering/type-based/). Not sure if C# has a drop-in replacement for JS’s normalize(), which appears to be doing the heavy lifting for neutralizing accents in this package. Not sure easy/hard it would be to reproduce the dictionary for neutralizing diacritics.
This is a problem that interests me a lot, since it crops up regularly in the Francophone countries that I support.
Apologies if I’ve already posted on this. If so, I couldn’t find that post on quick search.
@arthurshaw2002 , the component you are referring to is just a wrapper as it doesn’t do any heavy-lifting, but rather utilizes the i18n API to do it:
Please note, that I see a problem with their approach. If you search for “A A” it will match “A”, because normalizer adds both original and normalized/sanitized texts, causing false matches in some (admittedly rare) cases.