Yes, quite correct! The vectors are dense representations.

But the features aren't individually interpretable, in practice. For instance, the 'gender' of a word may have it's signal scattered over several features/dimensions of the learned vector.