My hunch as to why most logicians limit the term "tautology" to
sentential logic is that when truth tables are applied to monadic
predicate statements based on some model with some domain of
individuals, one is not building a truth table right off the
quantified statement but (even if undeclared) one is instantiating the
quantified statement such that in the domain {a, b} (x)Px becomes Pa
and Pb. Once that's done, what's actually being analyzed is not a
quantified predicate statement but a statement that aside from the
quantification project could be reduced to a sentential statement.

 Moreover, I think categorical predicate statements can be subjected
to truth table analysis (that's what I infer from Copi, "Symbolic
Logic," p.81) if (x)(Px -> Bx) in the domain {a, b} is instantiated as

(Pa -> Ba) & (Pb -> Bb)

and existential statements like Ex(Px & Bx) are instantiated as

(Pa & Ba) v (Pb & Bb).

Then we could build truth tables for them based on a model. But at
that point our analysis is operating *outside* quantified predicate
statements with statements that could as well be reduced to sentential
letters.

Thanks for your input Chris.

- paul