Some remarks on Sanders' typology of elliptical coordinations

Starting with Ross (1970), various proposals have been made to classify elliptical coordinations and to characterize different languages according to the types of ellipses which they admit. Sanders (1977) discusses four of these proposals, shows that they are inadequate on various grounds and proposes a fifth typology whose central claim is 'evidently correct', as he states (p. 258). In the following, I shall briefly outline this typology and then show that it is inadequate, too. Since there is only one language 1 know — German — I will take all my examples from this language. Moreover, all examples will be straightforward and easy to be judged. 1. Sanders' typology Sanders assumes that, as a rule, the patterns of coordinate ellipsis may be stated in purely positional terms; it does not matter whether there is a verb or a NP in a specific position — say in the final position of the first conjunct; there are some exceptions to this principle of 'categoryindependence'; they play no role in the typology, and we will not discuss them in what follows. In particular, Sanders says: "There seems to be no empirical basis . . . for distinguishing ... between the ellipsis of verbs in coordinations and the ellipsis of verbal arguments' (p. 248). All our examples given later on are of this kind. Six relevant positions are distinguished. They are labelled A, B, C, D, E, F, respectively, and are best illustrated by (I): (I) In the following general schema [... [A B C]w & [D E F]w ...]w 'ABC' is a preceding conjunct Linguistics 18 (1980). 871-876. 0024-3949/80,0018-0871 J2.00 © Mouton Publishers, The Hague