3.02.2006

i hate model theoretic semantics

there are a lot of things syntax can't account for without reference to semantics. so in syntax class, we're learning that each 'linguistic expression' is an ordered triple consisting of a pronunciation, a syntactic structure, and a semantic meaning. simple! then we just add them together in some certain way, and we can explain how larger linguistic expressions are formed.

some notes on semantics from class today:
This is pretty uncontroversial in the field today, Model Theoretic Semantics. Well, i shouldn't say that; everything is controversial. It's uncontroversial in all reasonable linguistics. [laughter]

we will treat meaning as if it is an object in the world that we talk ABOUT.

meaning is the semantic grammar that maps an expression to a model theoretic object. [little picture of a dotted line between the two.]

sentences are assigned a value of 1 or 0; true or false. there are some who argue that there are actually an infinite range of values, but whether that actually gets you anywhere...

knowing the meaning of a sentences is knowing what conditions it would take to make it true.

meaning is a function that inputs all possible worlds and returns a value of 1 in worlds where the sentence is true.

how poetic.

my rage is limitless.

10 comments:

Seb said...

Holy shit. That's so backwards.

That theory of semantics is going to be first against the wall when the revolution comes.

Cristi said...

it rules with an iron fist. It's all they teach you in linguistics classes, and it's all we learned in philosophy of language...

Seb said...

But what is the alternative?

Cristi said...

aha! that's the million dollar question...

give me about 10 years.

i'm half kidding. I really believe poststructuralism has the right idea about language, and i'd like to see some of the foundational assumptions of model theory swept away to make room for an approach that gives language more credit for DOING things in the world.

pretty vague, huh?

Marc André Bélanger said...

I never actually had courses on MTS, and I've had more than a few courses on semantics. Thing is, I studied in a neck of the woods that wasn't in love with "mainstream" linguistics. Most of what we saw was inspired by what is called (I was going to write "what is known as", but I can't really say that it is actually known) the Psychomechanics of language.

The name may be farfetched, but the theory had interesting points (and less interesting ones, but what can you do). The main point of this theory is a bottom-up approach to language: sentences begin with words, not syntactic structures.

Cristi said...

thanks for the comment, Marc!

i'd be interested to hear more about this. the only possibly similar thing i've encountered was an introduction to neural net modeling where scientists proposed that syntax should be built up based on the clusters of neurons that fired for a particular word and then how they synchronized. I'm not sure if that's the kind of thing you mean.

our prof. assured us that MTS was the basis of almost all current semantics, but who knows what she knows...

my most basic complaint with MTS is that it starts with 'every sentence is true or false' and then bases its semantics and syntax on figuring out how smaller units must build up within that framework. push aside presupposition, vagueness, questions, implicature ......

Marc André Bélanger said...

I've been wanting to post an overview of this theory for quite a while now. I could tell you to read the first chapter of my thesis but I'm not too sure it is that good an introduction.

One of the key points of the theory is that words have what is called "potential meaning" that is actualized in discourse. The way I see this potential meaning is somewhat like thoses strange attractors of chaos theory: the boundaries are not clearly defined, neither is the core, but the various actual meaning observed all seem to gravitate around this core and within these boundaries.

The potential meaning coalesces when the word is made incident (linked to) other words, and with the context. There is no question about the truth of the sentence, or other such things. Also syntax is somewhat seen as some sort of by-product of the features of the words, of what they need to be made whole.

Cristi said...

tried to read your thesis marc, but it's in french! my apologies...

i'm clearly a fan of the non-truth conditional bent of your theory, and i also like the syntax-as-afterthought thing. I think the reason linguists love truth so much is that it gives them something coherent to work toward and check against. like 'we know this should be true, and then we can test how to get there'. i find this REAL problematic.

but i wonder about this 'potential meaning' thing. are these the representations in the speaker's head? are they a feature of the mysterious language-at-large? where are they located?

Marc André Bélanger said...

Only the first few pages are in French, from page 5 (v), it gets more understandable.

'we know this should be true, and then we can test how to get there' Yeah. That's actually a big problem in many sciences. I think many find that comforting.

The potential meaning would be in the speaker's head, arrived at from his observation. That helps explain language change: the language we speak is ever so slightly different from our parents', because it is not cleanly transmitted.

Anonymous said...

One way of thinking about MTS is that it provides a mathematically precise but *non-constructive* theory of entailment. E.g. A entails B, according to the theory, if every model making A true also makes B true.

Nonconstructive because the class of relevant models is infinite, and the models themselves may also be of infinite size.

Entailment is important enough so that any theory of it at all should be welcome; in principle, non-model theoretic approaches are possible (and envisioned in the 70s by Katz, Jackendoff and others), but in practice, nothing useful happened. This might however be starting to change with the recent development of Proof Theory (Pollard & Lappin's hyperintensional semantics, etc)

- Avery Andrews

Related Posts with Thumbnails