Talk:Principle of compositionality

From Wikipedia, the free encyclopedia

[edit] Lead paragraph

I've changed the lead paragraph from:

The Principle of Compositionality in linguistics and the philosophy of language is the principle that the meaning of a complex expression is determined by the meanings of its constituent expressions and the rules used to combine them. It is frequently taken to mean that every rule of the syntax should be associated with a rule of the semantics that specifies an operation on the meanings of the constituents combined by the syntactic rule.

to:

The Principle of Compositionality in semantics is the principle that the meaning of a complex expression is determined by the meanings of its constituent expressions and the rules used to combine them. It is frequently taken, for instance in the influential work on the philosophy of language by Donald Davidosn, to mean that every construct of the syntax should be associated by a clause of the T-schema with an operator in the semantics that specifies how the meaning of the whole expression is built from constituents combined by the syntactic rule.

because:

  1. Compositionality is a principle of semantics whether in linguistics or philosophy of language, and it is also important in computer science;
  2. The semantics of a system needn't be just a model, there is also proof-theoretic semantics.
  3. Davidson deserves a mention ---- Charles Stewart 11:20, 8 Dec 2004 (UTC)

[edit] Further Thoughts on Lead Paragraph

I've put the original version of pairing syntactic and semantic operations back in, because the replacement, involve T-schema, is more theory-laden, and not all versions of compositionality that meet the general characterization are readily characterized in terms of the formal tools Davidson advocates. My intent with the original wording was to use the notions of semantic and syntactic opertions pre-theoretically in a way that included both model-theoretic and proof-theoretic semantic systems. If somebody else provide a wording that is more inclusive, I think it should be used. I have retained the discussion of the Davidsonian treatment because it is influential in the philosophical community and probably deserves mention, although I am a bit perplexed by the assertion that Davidson in particular deserves attention for an idea that predates his work by some time, and to which his original theoretical contributions are decidedly modest.

The change from "linguistics and the philosophy of language" to "semantics" is probably a good one. I wonder if perhaps "formal semantics" would be preferable, as compositionality has little to do with many areas of semantic research (for example, most work in lexical semantics).

--q10 16 December 2004

The reason for my introduction of the T-schema is that I wanted something brief in the lead paragraph that gets the idea of compositionality across, without introducing overgeneralisations. Using Davidson's idea of T-schema for natural language certainly achieves this. I think it is not a good idea to give several different analogous explanations of compositionality, just because the lead paragraph should not be a miniarticle.
It makes sense to talk of compositionality in the context of informal semantics, this is often done in connection with the idea that natural language must be compositional because it must be learnable. ---- Charles Stewart 11:18, 17 Dec 2004 (UTC)
I respect this desire, but I think T-schema are a poor choice for this purpose. In particular, they make the material less accessible to people with low jargon familiarity, and indicate (incorrectly) that compositionality is particular to truth-conditional, sechematically definable semantic systems. I think you're right about informality, but I still think it would be nice if we could find a way to point out that (for example) lexical semanticists tend not to care about the issue, while many philosophers who have never done any semantics first-hand are deeply invested in it.
Of course, you're right that the opening paragraph, as it stands now, isn't terribly well-organized. I think the best approach might be to just say "for example..." instead of "it is frequently taken to mean that...", although I still think we should try to dredge up a more accessible example than T-schema. --q10 18 Dec.
Excused me - since I'm using the plural above, I should have said "T-Schemata". Q10 15:24, 19 Dec 2004 (UTC)

[edit] Comments

The word compositionality has no entry in Encyclopedia Britannica, nor in Meriam-Webster. It is used in Britannica Concise Encyclopedia once, but not consistent with this article. Thus the article might be marked clearly as ephemeral.

The principle of compositionality may be useful for simple cases. But consider:

  1. Inside the house, the floors are green.
  2. Inside the house, the birds are blue.
  3. Inside the stomach, the floors are green.

which illustrate that some parts may imply meanings of words used in other parts: a house contains floors, stomachs do not. In programming, instantiation of an object can be seen as having a similar effect: introduction of names of object members. In mathematics use of a partial derivative operator similarly focuses on a particular name of a variable.

As I see it, the principle of compositionality does not hold in these cases, so is it useful? Formally, a frequent structure seems to be

F@C\;- for example house @ floors are green

where @ may be read as applied to. The principle of compositionality may apply, when F does not introduce names of attributes (e.g. floor). In these cases, C may be meaningful in isolation, i.e. have a value. From a mathematical point of view, the formalisation may be twisted slightly so that C becomes a function, which in that field technically 'saves' the principle of compositionality, but this is questionable in fields where functions as values are problematic.

Jørgen Steensgaard 10:06, 27 July 2006 (UTC)