No woman sleeps. Naive LF: no'(e0, w, e1, e2) & woman'(e1,w) & sleep'(e2, w) & Rexist(e2) Coerce claim of sentence from e2 to e0 thru "no" relation: Rexist (e0) Axiom in KB defining "no" predicate: no(w e1 e2) <--> (A w1,e3,e4) [subst(w, e1, w1, e3) & subst(w, e2, w1, e4) & Rexist(e3) --> ~Rexist(e4)] subst(a,b,c,d): A substitution operator: a plays the same role in b that c plays in d. Here, c and d are instances of a and b, resp. What I'm unhappy about: the e2 argument of "no". I'd prefer no(w,e1) in the naive logical form, with some kind of pragmatic strengthening converting that to no(w,e1&e2). (This is what I did in my paper "Monotone Decreasing Quantifiers" for the example "Few men work", where I strengthened the plural morpheme to the set defined by e1&e2. But here there's no plural morpheme to strengthen. In this case the axiom defining "no" would be no(w,e) <--> (A w1,e1) subst(w,e,w1,e1) --> ~Rexist(e1)