Skip to content
Coordinates of ThoughtHome of Philosophy
Join
Reading Group

Reading group: Tarski’s “The Semantic Conception of Truth” (1944) in three sittings

Tomasz Szewczak Fellow
· 6 replies
TruthAnalyticIndianAlfred TarskiAristotle#convention-t#object-language#reading-group#tarski
Follow

Tarski's technical work on truth is hard going, but in 1944 he wrote a non-technical paper for Philosophy and Phenomenological Research that explains the ideas for philosophers. It is clear, polemical in places, and only thirty-odd pages. I propose we read it in three sittings.

Text. Widely reprinted and easy to find in anthologies on truth. If your copy has numbered sections, use them for reference.

Plan.

  1. Sitting one: the problem and Convention T. What Tarski wants from a definition of truth: that it be materially adequate and formally correct. The famous schema: "snow is white" is true if and only if snow is white. Note how he connects this to Aristotle's formula.
  2. Sitting two: the liar and the hierarchy of languages. Why a language that contains its own truth predicate, and the usual logic, is inconsistent. The distinction between object language and metalanguage, and why natural languages are a problem.
  3. Sitting three: the polemical remarks. His replies to objections, including his claim that the semantic conception is philosophically neutral: you can accept it and remain a realist, an idealist, an empiricist or a metaphysician.

Post as you go. I work in many-valued logic, so I will probably be the person asking what happens if we drop bivalence.

LikeReply
Follow6 replies

Join the conversation: members reply here, and joining is free. Reading is always free.

Join free to reply Sign in

6 replies

  1. Oskar Nyberg

    Fellow

    Sitting one. What struck me is how modest the requirement is. Convention T doesn't define truth; it is a test any definition must pass: it must imply all the instances of the schema. Tarski is explicit that he is not offering a new concept, only trying to capture the classical one precisely. People sometimes describe this as deflationary. Reading it again, I think Tarski would have been puzzled by that.

    Helpful · 1
  2. Derek Hollis

    Fellow

    Sitting two. The object language and metalanguage distinction is exactly the distinction between a program and the language in which you state what the program must do. You cannot fully specify a system's correctness inside the system itself; you need a richer language outside it. Tarski's undefinability result has always felt to me like an engineering truth before it is a philosophical one.

    Helpful · 2
  3. Govinda Prasad Upadhyaya

    Fellow

    Sitting two, an old parallel. The Sanskrit grammarians already needed a way to talk about words as words. Pāṇini has a rule saying that, in his grammar, a word refers to its own form unless it is a technical term. So "agni" in a rule means the word agni, not fire. That is the use and mention distinction, stated as a rule of metalanguage, more than two thousand years ago. Pāṇini did not face the liar, but he saw that a language about language needs its own conventions.

    Helpful · 3
  4. Pieter van Dijkhuizen

    Fellow

    A naive question from sitting two. Why can't English just have its own truth predicate, since it obviously does? I say "that's true" every day and nothing explodes.

    1. Tomasz Szewczak

      Fellow

      Pieter, Tarski's answer is that English does have one, and that is exactly why, in its full generality, it is inconsistent: you can construct the liar ("this sentence is not true") and, with classical logic, derive a contradiction. Everyday use survives because we rarely build such sentences. Since Tarski, people have tried to have a language contain its own truth predicate by changing the logic, for instance allowing truth-value gaps (Kripke's fixed-point approach) or, more radically, accepting some contradictions. That is where most of my own reading lives, so I will stop before I write an essay.

      Govinda, the Pāṇini rule is wonderful, I had no idea.

      Helpful · 1
  5. Tomasz Szewczak

    Fellow

    Sitting three. The neutrality claim is the part I most want to discuss. Tarski says his conception commits you to no epistemology or metaphysics. That is either a strength (everyone can use it) or a sign it doesn't tell us what truth is (as Oskar suggested). I lean to the first. A definition that every side can share is what lets them disagree about the same thing.

    Helpful · 2