I see this as a tool-based solution rather than a management based solution. You are already using resource files, and
We have also separate files per purpose, every language have few files (like with dialogues or options). With something around 5 languages and few thousand records per language it really isn't easy to track.
So your problem (as I see it) is to determine when resource texts are changed, and what work must be done to fix up all the resources for the other languages.
My first thought was to create a program to extract all resource texts to a flat file with fields like
...and then sort it, and difference the current flat file with a previous version. Now you can see the changed resources and manually fix them up.
Of course, new inputs from your translators will show up in the diff's and will need to be ignored.
But your situation (100 employees) strikes me as an opportunity to create a much better tool that would enhance software development in your next projects. Here is a more complex tool concept:
Create a program to extract all resource texts to a database table with fields like
The program would add to this table with changed resources - IE, if the language and Text is the same, the row would not be updated. If the language/text is different than the current row, add that resource as a new row.
Thus the table would hold all versions of the resource by date. The From-Resource-File-Flag would indicate that the source of this row is from a developer change to a resource file.
Because the table is in SQL, you would be able to execute queries. You would be able to identify changes by developers that would need to be translated. You would also be able to locate resources that are not translated - for example, there is an English and French version but no Spanish version.
But the best part is that a separate tool (or website) would now be possible so that you can apportion out translation changes without too much human oversight. A translator would sign on and see a list of texts that need to be translated; the translation would be added to the database and then propagated back to the resource file with another program.
As I said, more of a tool based solution. A management based solution might fail if people are not diligent. There would be more testing to ensure that all resources were translated (since if a resource is not translated, the resource text would be output in the wrong language).