It is often the case that source software strings are created before a user manual and very often the translation of these two pieces are done separately as well.
Even if you have been diligent about saving all your software strings (and subsequent translations) if those same strings are not worded the exact same way in your manual many consistency issues may occur.
Doing an audit of the source text before translation can save you time, money and a lot of headaches. A source audit or software validation against the documentation is where someone manually checks the text from the manual against the existing strings that are being referenced.
Any discrepancies between the two will be discovered at this stage and can be addressed before translation. Here is something to consider, imagine that the source was not audited beforehand and that the translations were just started. Imagine a translation project of 10 languages, what happens when the translators start to notice discrepancies or errors between the user interface and the manual? The problems multiply by 10 because you are in essence fixing the issues 10 times (once in each language).
If the discrepancies are sorted out before the translation phase begins you will provide the proper conditions for a solid and consistent translation and of course you will have a solid source document as well.
The main concern here is for the end user. Isn’t it frustrating when you read documentation and the instructions refer to a software string or command that doesn’t match what is on screen?