This discussion also suggests the possibility that presuppositional phenomena are
not homogeneous. If the hypothesis presented here is correct, then the presuppositions of,
say, change of state sentences are derived very differently from the presuppositions
generated by even, too, and again, which plausibly have a conventional source. On the
other hand, this picture of (some) presuppositions as conversationally generated inferences
suggests the possibility of unifying the treatment of these presuppositional data with that of
other types of “preferred readings,” such as those illustrated in Section 2.