We hypothesize, in the case of tasks represented as structured collections of PaT-Nets, that PaT-Nets at every level above the animation level can effect choices in Text Generation. This hypothesis needs to be refined in terms of which sub-tasks within text generation may be affected.