Sentence Relations and Truth 109
2 We assume here a simple correspondence theory of truth; see Grayling (1982) for a
discussion of this and other theories of truth. For many semanticists employing truth
conditions to investigate meaning it is in fact the proposition expressed by a sentence
(uttered in a particular context) that, depending on the facts of the world, may be true
or false. In Chapter 10 we review the proposal that this truth-evaluable proposition is the
essential part of the meaning of sentences and therefore the object of study in semantics.
For simplicity, in the present discussion we continue to talk of sentences being true
or false.
3 Logicians sometimes distinguish between two types of what we are here calling coun-
terfactuals: subjunctive conditionals, which set up a hypothetical situation in the
antecedent, as in If Liverpool were to win the championship, he’d be a happy man;and
counterfactual conditionals, where the antecedent is implied to be false, as in If Liv-
erpool had won the championship, he would have been a happy man. For the rest of this book,
we will use the term counterfactual as a cover term for both types. See Lewis (1973)
and Haack (1978) for discussion.
4 Including for example Leibniz (1981), Kant (1993), Quine (1953), Carnap (1956), and
Kripke (1980).
5 Another denition of necessary truth uses the notion of possible worlds, due originally
to Leibniz. Possible worlds in the work of, for example, Lewis (1973, 1986), is a notion
used to reect the way speakers use language to do more than describe the world as it is.
Speakers can, for example, hypothesize situations different from reality, as in counter-
factuals like If Ireland was a Caribbean island, we’d all be drinking rum. Such situations
that are not asserted as real are called possible worlds, the idea being that the world where
Ireland is a Caribbean island is linguistically set up as a possible world, not the actual
world. One denition of necessary truth uses this notion as follows: A statement is neces-
sarily true if it is true in all possible worlds. However, since the constraints on setting up
hypothetical worlds and their possibilities of difference from the real world are far from
easy to ascertain, such a denition needs some work to establish. See Grayling (1982:
43–95) for introductory discussion and Kripke (1971), Lewis (1973) and the papers in
Loux (1979) for more detailed discussion. We come back to this idea of possible worlds
again in chapters 5 and 10.
6 This idea, often known as concept containment, derives from Leibniz. See the papers
in Jolley (1995) for discussion.
7 An anonymous reviewer has suggested that an example like Whales are mammals brings
out the difference between necessary and aprioritruth. Following Kripke, this sentence
is a necessary truth, but it is not an aprioritruth for our hypothetical speaker who thinks
that whales are sh. Similarly Wa t e r i s H
2
O might be a reasonable candidate for a neces-
sary truth but might only be learned by experimental inquiry and thus be a posteriori.
8 This assumes that we rule out self-reference to avoid paradoxes. For example by choosing
to instantiate Y as “are not Xs,” we would get the necessarily false statement All Xs that
are not Xs are Xs.
9 We discuss a formal approach to these lexical relations, meaning postulates, in chap-
ter 10.
10 Another, more strictly logical way of describing this entailment relation is to say that p
entails q when an argument that takes p as a premise and q as a conclusion must be
valid, for example the argument:
The anarchist assassinated the emperor.
∴ The emperor died.
is valid.
11 Since this relation is clearly similar to the bi-conditional connective described earlier,
we could give a logical denition of synonymy as in: p and q are synonymous when the
expression p ≡ q is always true.