If a translator asks you many clarifying questions about your code or application, they're not being annoying – they want to do their best job.
Say you have a furniture store and you have built a nice inventory management application. You are getting the application localised and a translator wants to know the meaning of 'table'. Are they an idiot? Surely not. It depends on the format of the localisation file, but context is super important in translation. In worst cases, the file contents may be alphabetised and therefore they've lost all context. In other cases, the string IDs may be something generic like
How about now? How about now?
How would you translate 'table' into any other language you know? Or, if you're not a polyglot, you can think of this in terms of definitions: how would you define 'table'? (Click the paragraph/arrow)
How about now?
How about now?
See? Even though the words in English are identical, they carry completely different meanings and will likely be translated using different words in another language. If this type of context information is not available through surrounding words, string IDs or comments, the translator is basically just flipping a coin. More often than not, a multi-sided coin (if that were a thing). Sometimes they can make an educated guess but due to the concise nature of software texts, even that can easily go wrong. You, the developer, be unambiguous – or helpful and understanding – instead.
By the way, if for example those above two meanings of 'table' co-exist in an application, you'll need to handle that carefully in your translation memory. Luckily, you can add multiple translations for one source text!