On Fri, Jul 3, 2009 at 4:33 AM, Norbert E. Fuchs <<a href="mailto:fuchs@ifi.uzh.ch" target="_blank">fuchs@ifi.uzh.ch</a>> wrote: <div class="gmail_quote"> <blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> Third, variables per se do not have a type. A variable in apposition to a common noun has all the features of this common noun. A bare variable - interpreted as the variable in apposition to "something" - introduces a singular object with undetermined gender. The gender will be determined once the variable is anaphorically referenced by a pronoun.</blockquote> <blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> Fourth, subsuming a common noun to "something" is reasoning that is not done by the ACE parser APE, but for instance by the ACE reasoner RACE that lets you prove that "John has something." follows logically from "John has an apple.". </blockquote></div> I was trying to not to suggest adding a general DL reasoner to the basic ACE package, though now you mention it :-) :-) What I was wondering was whether 'something' might be a special case: if an index i refers to an individual which is of sort something, and another index i refers to something that is a dax, then the individual indexed by i must be a dax, even if we have no idea what a dax is. Thanks Simon