(a) that it is indeed reasonable to compare grammatical ellipses to the general process of anaphora; (b) that it is also reasonable not to see aIl cases of ellipsis as a unified phenomenon; (c) but that there are a number of counter-examples to the mixed hypotheses (of the Hankamer and Sag type). On the other and, the criteria put forward by J.-e. Milner (Ordres et raisons de langue, Seuil, 1982) for the noun/ pronoun relations seem to open up the possibility of a new mixed hypothesis which has a further advantage: it generalizes to intonational anaphora, which 1 have called prosodic binding in previous work.