Interpreting Motion Expressions in Route Instructions Using Two Projection-Based Spatial Models