Ok, so the really proper way would be to import the data as a MullikenPopulation object, but then you would have to include some information about the populations too. So a hacky way that would work easier would be to use a Forces object anyway - at least this will allow you to see the arrows, as you drop the Forces object on the Viewer.
The proper way to do this is to make a LabFloor plugin, but for testing, you could just make a Python script that reads you file, makes a Forces object, and nlsave-s it - then you can import the NC file in VNL.
To make a new Forces object:
from NanoLanguage import *
# Read and parse data file, store x,y,z coordinates as Nx3 numpy.array (Angstrom unit assumed)
coordinates = []
# Read and parse data file, store "forces" as Nx3 numpy.array, no unit
my_forces = []
# Make a NanoLanguage Forces object
forces = Forces.__new__(Forces)
forces._Forces__forces = my_forces*eV/Angstrom
forces._Forces__forces_origin = coordinates*Angstrom
nlsave("file.nc", forces)