They are, my point is that this only became obvious later. The initial NTRU preprint frames it as a scheme based on modular polynomial arithmetic, rather than anything related to lattices. At the end of section 4 they even explicitly describe it as a “ring based cryptosystem” (contrasted with “group based cryptosystems”). As you say, now we would call it a lattice-based cryptosystem (over an algebraically structured lattice).