Halliday & Matthiessen (2014: 547):
A fact is thus analogous, as a form of projection, to what we called an ‘act’ as a form of expansion. Each represents the least prototypical form of its own general category; and hence the least differentiated. Whereas there is a clear distinction between expansion and projection in their finite clausal forms – between, say, (projection) he never asked if/whether it was snowing and (expansion) he never came if/when it was snowing – there is only a minimal distinction, and perhaps even blending, between (projection: fact) she liked the snow falling (that the snow was falling) and (expansion: act) she watched the snow falling (as the snow was falling). Seeing that facts and acts come so close together in this way, we can understand how it is that the same scale of interdependency types (parataxis/hypotaxis/ rank shift) is associated with both these logical-semantic relations.