I think that - to implement JSONiq well in VXQuery (in a stable, efficient, and well tested way) - there is enough work for 2 GSoC projects.
However, we need separate projects, separate code, and separate evaluations for both projects.
So the projects cannot depend on each other, but they can be complementary.
I think that you could propose 2 projects to GSoC:
One project that would add objects to the data model (and all the functions, parser modifications, rewrite rules, etc. - the proposal would need to lay those out in more detail) and one project that does the same for arrays. That way the projects would complement each other without depending on each other. If one project is not successful, the other project can still complete, VXQuery just wouldn’t get a complete JSONiq implementation in that case.
Maybe Riyafa could create a proposal to add objects and Christina could add one to add arrays?
Of course you could (and are encouraged to) discuss those proposals either with the current community or with each other. But please make sure, that each proposal can be completely implemented independently of the other project. There might be some (small) overlap that would be part of both projects proposals. In that case we (the VXQuery community) would then need to decide how to address that during the coding phase.