See my additional explanation above (I edited my question with further information).

Ok I managed to do so.One example of the problem encountered is in Set2 --> Set2P1 - subquestion 14.I imported the subquestion 14 (among the 60 others subquestion by pasting them) and this one disappeared. I've heard from a coworker that it's not the only example but we tried different modifications to fix this. This is the only example that we know for sure to be here.Thanks again for all 

In Set2 > Set2P1 there is no subquestion with the code "14", so I assume you are talking about "89014" (éduquer)?

Are you sure? I did a couple of tests and "éduquer" always showed. Are you sure that the subquestion isn't there? There are so many, you can easily overlook it.

What???
How can this be?
In the original study, this exact word does not show...

By the way, you start this question with the subquestion 89011 instead of 89001 and then you skip item 11 later.

Due to this error, you only have 59 subquestions, despite your last subquestion code suggesting that there should be 60 (89060).

I assume you are aware of this?

Once again that's super strange. in my questionnaire it doesn't appear this way.
Honnestly i'm getting a little crazy with this plateform

I switched off randomization for the subquestions to better test (but it also showed when it was on):

 

Here you see it. Tested on LS 3 Version 3.27.18+210921

 

I don't think the problem is the platform...

I have to admit that I’m really confused. When I programmed the sub questions they were not labeled that way.
And more : on the developper section for subquestions, the number 89014 does not appear.
I’ll gladly admit that I’m no LS expert but the disappearance of a subquestion in the developer mode, then it’s reappearance when duplicated in another survey or when in preview mode.
Perhaps you can explain, for preventing further stressful experiences.
This could be related to the version I’m using?
Anyways thanks for everything

If you say you did not enter these strange codes initially, so you will have copied something or clicked twice when entering the subquestions (maybe in quick-add)
Something like that.
Now there are double codes and LimeSurvey renames to these 89012 codes or whatever.
The best to get rid of it. Remove the question and create it again .

Joffm

I don't know if it is related to the version you are using, but it could be. We would need to look at the release notes for all versions after your version to see if a bug that fits the description has been fixed. But your version is now over 12 months old, there could be a potentially lot of versions in between and a lot of bugs being fixed. This is not a bug I had ever heard of, but that doesn't mean anything, it could be introduced in one version and immediately being fixed in the next one.

Sometimes, the survey is corrupted (for whatever reason), did you do an integrity check? (Configuration --> Check data integrity - You might not have the rights to do so, if you are on a university installation, for example).

You can have a look at the survey logic file to see if you can see something strange and if the item is shown there.

And sometimes, if a survey is corrupted, it helps to export and import it again. Give it a try.

Or like Joffm said, try to create the question from scratch and see what happens.