Hi,
Yes and yes, but will have to code your way through, it's not plug and play.
Input language:
If your target language is 'western' (ie, it splits words with spaces, simple inflections) maybe you can just ignore the POS tagging in the templates and work your way with the tokens. Ie, just ignore everything and write query templates happy-go-lucky.
If you need better support for more complex queries or a more complex language then you are going to have replace the english preprocessing (tokenization and pos tagging) with your own. Look at the code for this part, it's not complex or long.
Output language:
It looks like you could get a subset of GQL working with quepy. Quepy uses an internal intermediate representation and at the last time it compiles the output. Check out how the output is built for SPARQL and MQL as examples. You will have to write your own module for GQL.
Send us a pull request, cheers!