Is knowing a statements truth conditions (a) necessary and (b) sufficient for knowing what it means?
The short answer is “No!”
The essay’s title question refers to the idea that the meaning of a “statement” is defined in terms of its truth-conditions. The truth-conditional theory of meaning is principally associated with Donald Davidson(1) who first explicated the modern version. Davidson attempted to carry out for the semantics of natural language what Tarski’s theory of truth(2) for artificial languages tried to do for logic. Tarski attempted to define truth for artificial languages like logic and mathematics in terms of a presupposed notion of meaning. Davidson inverted Tarski’s argument and attempted to define meaning for natural languages in terms of a presupposed notion of truth. In taking this approach, Davidson harks back to Frege and Wittgenstein. Though the concept of truth is central in Frege’s work, he leaves it undefined. From Logical Investigations: “It seems likely that the content of the word ‘true’ is sui generis and undefinable”(3) And from Wittgenstein: “to understand a proposition means to know what is the case if it is true”(4). A more recent advocate of truth-conditional semantics is Jerry Fodor. In The Language of Thought, Fodor documents his premise that “one understands a predicate only if one knows the conditions under which sentences containing it would be true”(5).
Truth-conditional theories of meaning interpret “meaning” in terms of representation. They understand “meaning” in terms of some form of mirroring between statements and actual or possible states of affairs(6). Truth-conditional theories of meaning define the meaning of a given statement in terms of the conditions under which the statement is true. So, to use the standard example, because “snow is white” is true if and only if snow is white, the meaning of “snow is white” is snow is white. For a truth-conditional theory of statement meaning, to know a statement’s meaning it is both necessary and sufficient to know the conditions under which the statement would be true.
The initial appeal of truth-conditional semantics is the intuitive appeal of the notion that if one understands a statement, one also understands the conditions under which the statement would be true. But the question is whether one must know what the statement means before one can know the conditions under which the statement would be true. Or whether knowing what the statement means can be defined in terms of knowing under what conditions the statement would be true.
A number of criticisms of truth-conditional semantics have surfaced to undermine the plausibility of the truth-conditional theory of statement meaning. Because truth-conditional semantics understands meaning in terms of some form of correspondence between statements and states of affairs, it is usually characterised as depending on a prior notion of a Correspondence Theory of Truth. As such, it raises objections from those philosophers who see difficulties with this approach to understanding Truth. A similar source of criticism comes because statements can be meaningful even if they talk about non-existent states of affairs. (“Santa Claus is jolly” is meaningful, even though Santa Claus does not exist.) This seems to require a theory of truth that admits of non-existent states of affairs. As such, it raises objections from those philosophers who are concerned about the ontological commitments that seem to be required. Intuitively, it appears that understanding the meaning of “Santa Claus is jolly” is not the same sort of thing as knowing under what conditions a non-existent entity could be jolly.
Another criticism comes from Scott Soames(7). He argues that in its traditional form, a truth-conditional concept of meaning gives every necessary truth precisely the same meaning. All necessary truths are, by definition, true under all conditions. This means that every necessarily true statement is true in precisely the same conditions. Furthermore, the truth conditions of any random statement are logically equivalent to the conjunction of the conditions that make that statement true and any necessary truth. This means that by truth-conditional semantics, any random statement means the same as its meaning plus a necessary truth. To use the standard example again: if “snow is white” is true iff snow is white, then it is logically necessary that “snow is white” is true iff snow is white and 2+2=4. Therefore “snow is white” means both that snow is white and that 2+2=4. And this process of conjoining necessary truths can proceed indefinitely. That seems intuitively wrong. In singling out just which of the potentially infinite number of conjoined truth-conditions should make up the statement’s meaning, the only available guide is the statements meaning. In other words, understanding the meaning precedes the understanding of the relevant truth-conditions. Which means that knowing the statement’s truth conditions is not necessary for knowing the meaning of the statement.
Closely allied with the criticism from necessary truth, is the criticism from equivalent extensions. Consider the two statements “a zebra has a heart” and “a zebra has a kidney” — a classic philosophical example. In this actual world, it turns out that “creature with a heart” has the same extension as “creature with a kidney”. Everything which is in fact a creature with a heart is also a creature with a kidney, and vice versa. But if evolution had gone differently, or in different possible worlds, there might have been a species of creature which had hearts but no kidneys, or vice versa. Because these two example sentences have isomorphic extensions, they have equivalent truth conditions. They are both true or false in exactly the same circumstances, at least in this world. But it is intuitively obvious that they do not mean the same thing. In other words, again, knowing the statements truth-conditions is not sufficient for knowing the meaning of the statement.
The next criticism of truth-conditional semantics is described well by Brian Loar(8). Consider a master tile layer and a bunch of assistants working to lay a complicated pattern involving different tiles. The crew have developed a simple set of signals. When the master calls “red left”, the assistants know to pass him, say, a red left edge tile. And so forth for any number of differently shaped, or coloured, or patterned tiles. It seems fairly intuitive that the meaning of “s” amongst this small population is “give me a tile of type x” or perhaps, to keep things in the proper declarative mood “The next tile required is a tile of type x”. But the simple signals being used do not have any semantical structure. Simple non-linguistic signals (two lights) seem to have meaningfulness (the redcoats are coming by land) independent of being part of a semantically structured language. The master could just as easily call “left red”. But how do you get to truth-conditions for such simple signals without translating them into a semantically structured language, as I did for “red left” and “two lights”? And how can one do that translation of simple signals into language without presupposing the meaning being translated? What are the truth-conditions for the black-silhouettes that tell you which bathroom is for the ladies, if you cannot first translate the black silhouette into “this way to the bathroom for the ladies”? If such simple non-linguistic signals can have meaning, and can have the same meaning as sentences in a language, the non-truth-conditional method of ascribing meanings must apply for sentences as well. One must know the meaning of the sign before one can know the truth-conditions. Which means that knowing a statement’s truth conditions is not necessary for knowing the meaning of the statement.
Another criticism of the truth-conditional theories of meaning is that it seems to apply only to descriptive (assertive) fact-stating declarative statements. There are many sentences that can be created in any natural language that are not of this kind. Consider questions, commands, exclamations, and (perhaps most important) conditional statements. None of these obviously have truth-conditions, and some might not even be truth-apt. One of the ways that some of the truth-conditional theorists attempt to deal with these non-declarative statements is to posit an underlying logical form that does encode some truth apt statements that have truth conditions. But it can be argued that translating non-declarative statements into their “proper” logical form requires a prior understanding of their meaning — putting meaning prior to truth conditions.
Truth-conditional theories also have similar technical difficulties in dealing with indexical elements and some relational terms of language. Davidson’s own response to this is to treat indexical statements as relativized statements — statements considered as utterances by a speaker at a time. Putting the meaning of “now” or “I” (or any of the other indexicals) into something that truth-condition could apply would seem to demand that they be “translated” out of the initial statement. And the same problem would seem to infect such relational terms as “tall” or “fat”. Without specifying a specific context in which they have some determinate meaning (“taller than Bob” or “fatter than the average American teenager in 2012”), it is difficult to see how to relativize it without presupposing the meaning. In other words, translating an indexical statement or a statement with vague relational terms into a properly relativized statement presupposes an understanding of the meaning of the statement. Once again putting meaning prior to truth-conditions.
The final criticism I will mention is that truth-conditional semantics assumes a pre-existing notion of truth as foundational. There is a lot of philosophical debate (and no general agreement) about the concept of truth. Truth-conditional semantics, it is argued, does not provide an explanation of “meaning”, but merely recasts the problem and moves it over into the front-yard of the debate over the notion of truth. It may be a good thing to know that “der schnee ist weiss” is true if and only if snow is white. And it may provide a good explanation of what “der schnee ist weiss” means. But it does not provide any explanation of what the right side of that equivalence means. There is a distinction that must be drawn between the question “What is the meaning of this sentence?” and “In virtue of what does the sentence have that meaning?” One can argue that truth-conditional semantics provides the meaning of the symbol in quotes (“der schnee ist weiss”) by using words in some language (snow is white). What is needed is a distinct sort of theory — a foundational theory of meaning. What is needed is a theory that explains what it is about the world (or about some person or group) that gives the quoted sentence (“der schnee ist weiss”) the meaning that it has. What the truth-conditional theories of meaning do not provide is an explanation of what it means to say that snow is white (the meta language statement, not the quoted sentence). So, again, knowing the truth-conditions for a statement is not sufficient for knowing the meaning of the statement.
As Michael Dummett (9) has pointed out, to know the meaning of a statement, a person must know the “sense” (the part of the meaning that the person grasps), the “reference” (which indicates what claims about the world are made by the statement), and the “force” (what kind of speech act the statement performs). A theory of conversational implicature (a la Grice(10)) provides what the truth-conditional theories do not — a foundational theory of just what it is in the world that gives the statement ‘snow is white’ (the meta-language statement, not the quoted object language statement) the meaning that it has. And Dummett’s variation on truth-conditional semantics translates the notion of a “truth-condition” into a notion of “assertability condition”(11). By this he explains the notion of linguistic meaning without incurring the criticisms outlined above, and without invoking truth-conditions.
Since knowing a statements truth-conditions is neither necessary nor sufficient for knowing what the statement means, then meaning is not equivalent to the statement’s truth-conditions. But obviously, once one knows what a statement means, one does know the truth-conditions that would render the statement true. And if one is skilled in logic, then knowing the meaning of the statement presumably lets one know what logically follows from that statement. (Although since not many people are skilled in logic, that is perhaps not a good foundation on which to build a theory of meaning.)
More formally — “If M then T”. But since M (meaning) and T (truth-conditions) are not equivalent, this does not imply “If T then M”. Arguing that it does is committing the fallacy of assuming the consequent. So knowing the truth-conditions of a statement is a consequence of knowing the meaning of the statement, not the other way around.
So it is the case that
If A knows that S means that M, then A knows that S is true if and only if T.
But it is not the case that
If A knows that S is true if and only if T, then A knows that S means that M.
Notes & References
(1) Davidson, Donald; “Truth and Meaning” in Synthese, Vol 17, No 3 (Sep, 1967), pp 304-323.
(2) Tarski, Alfred; “The concept of truth in the languages of the deductive sciences” (published in Polish), 1933. Expanded English translation in Tarski’s Logic, Semantics, Metamathematics, Papers from 1923 to 1938, pp. 152–278.
(3) Frege, Gottlob; Logical Investigations (Library of philosophy and logic), Blackwell Publishers, Oxford, England, 1977. ISBN 978-0-631-17190-4. Pg 4.
(4) Wittgenstein, Ludwig; Tractatus Logico-Philosophicus, downloaded from URL=www.gutenberg.org/ebooks/5740. Para 4.024.
(5) Fodor, Jerry; Language of Thought, Harvard University Press, Cambridge, Massachusetts, 1980, ISBN 978-0-674-51030-2.
(6) Lycan, William G.; Philosphy of Language: A Contemporary Introduction, 2nd Edition. Routledge, New York, New York, 2008. ISBN 978-0-203-93000-7. Pg 114.
(7) Soames, Scott; “Truth, Meaning and Understanding” in Philosophical Studies: An International Journal for Philosophy in the Analytic Tradition, Vol. 65, No. 1/2, (Feb 1992), pp:17-35., URL=http://www.jstor.org/stable/4320270
(8) Loar, Brian; “Two theories of Meaning” in Truth and Meaning: Essays in Semantics, Evans, Gareth & McDowell, John (eds.), Oxford University Press, Oxford, England, 1999. ISBN 978-0-198-25007-4. pp 138-161.
(9) Dummett, Michael .A.E.; “What is a Theory of Meaning” in Mind and Language, S. Guttenplan (ed.), Oxford University Press, Oxford, England. 1977, ISBN 978-0-198-75043-7.
(10) Grice, H. Paul; “Meaning” in The Philosophical Review, Vol 66 (1957), pp 377-388. (Reprinted in his Studies in the Way of Words.)
“Utterer’s Meaning, Sentence Meaning, and Word Meaning” in Foundations of Language, Vol 4 (1968), pp 225-242. (Reprinted in his Studies in the Way of Words.)
(11) Dummett, Michael; The Seas of Language, Clarendon Press, Oxford, England. 1993. ISBN 0-198-23621-2.
Evans, Gareth & McDowell, John (eds.); Truth and Meaning: Essays in Semantics, Oxford University Press, Oxford, England, 1999. ISBN 978-0-198-25007-4.
Grice, H. Paul; Studies in the Way of Words, Harvard University Press, Cambridge, Massachusetts, 1991. ISBN 978-0-674-85271-6.
Speaks, Jeff, “Theories of Meaning” in The Stanford Encyclopedia of Philosophy, Edward N. Zalta (ed.), URL=http://plato.stanford.edu/archives/sum2011/entries/meaning/
Tarski, Alfred; Logic, Semantics, Metamathematics, Papers from 1923 to 1938, John Corcoran (ed.), Hackett Publishing Company, Indianapolis, Indianna, 1983. ISBN 978-0-915-14476-1.
Wikipedia contributors; “Truth-conditional Semantics” in Wikipedia, The Free Encyclopedia. URL=http://en.wikipedia.org/w/Truth_conditional_semantics