Natural language with sentential pronouns

February 28, 2005 5:51 PM

Are there any natural languages that allow pronouns to appear in sentential position?

Specifically, I'm looking for a language that could say something to the effect of: "If John said something, then it," where "it" refers to the proposition that John said.

Traditional first-order logic would not be happy with this kind of sentence. (S)(JohnSaid("S") -> S) doesn't work, because in the first instance of the variable you're quantifying over names, whereas in the second you're quantifying over sentences. Yet it wouldn't surprise me if there were some language that could say something like "If John said something, then it." Is there?

posted by painquale to religion & philosophy (10 comments total)

Why is this a problem? "For all x, if John said x, then x"
posted by crunchburger at 6:00 PM on February 28, 2005

Well, "x" isn't a pronoun. You wouldn't say "No matter what it is, if John said it, it.". I wouldn't, anyway.
posted by kenko at 6:42 PM on February 28, 2005

crunchburger: The problem is that you're treating X as both a sentence and a name of a sentence. Note that in the first instance you're passing X as an argument to a predicate, and in the second instance you're using X as a sentence with a truth value. Tarski (one of the granddaddies of logic) used T-biconditionals ("Snow is white" is true iff snow is white) to formally define truth, and he had to invent model theory and truth hierarchies to get around this problem.

I'm looking for help more from linguists here... not logicians....
posted by painquale at 6:53 PM on February 28, 2005

I've never heard of one, though that certainly doesn't mean it doesn't exist. I'm no typologist. Have you considered asking on Linguist-L?
posted by redfoxtail at 8:35 PM on February 28, 2005

painquale:

Do you mean something like "If John said [the sky is blue], then [the sky is blue]," where the second instance of [the sky is blue] can be replaced by some kind of dummy element?

English kind of has something like that: "[The sky is blue] if John said [so]" where [so] can replace the sentence [the sky is blue]. But I'd hardly call 'so' a pronoun.

Is this helpful at all, or totally up the wrong tree (Chomskian pun not intended)? Can you give an example of what you think such a sentence would look like in this hypothetical language?
posted by greatgefilte at 8:58 PM on February 28, 2005

Not sure if this is what you're looking for, but regarding problems in translating between natural languages and FOPC, there is quite a bit of discussion about the donkey sentence (i.e., Every farmer who owns donkey beats it). Or are you looking for the specific form that you mention?
posted by brool at 9:36 PM on February 28, 2005

"John said 'white men can't jump.' True dat."
posted by zanni at 12:01 AM on March 1, 2005

redfoxtail: That sounds like a very good idea! If I don't get any responses here that immediately help me out, then I'll sign up for their mailing list and ask them. (I checked out your web page by the way... you sound like you're doing really cool work).

brool: The donkey sentence is closely related with my concern because both deal with anaphora. Maybe I should be a little more explicit: I'm seeking this particular form to help with T-biconditionals and the definition of truth. I want to say:

(S) (S is true < -> S)

using a new kind of quantifier that binds variables in both sentential and nominal positions; grammar dictates whether S is to be instantiated as a sentence or a name of that sentence (this is not my theory, by the way). I want to know if there are any natural languages that do something that looks like this. The latter half of the biconditional can't name the sentence, so it would have to anaphorically refer to the sentence by means of a pronoun. In English, this would seem to look like something like: a sentence is true iff it. But obviously that's not an English sentence. I don't necessarily need a sentence about truth; any sentence that has a pronoun standing in sentential position will make me happy.

greatgefilte: Hmmmm. What you've given me isn't exactly what I want -- for one, "so" is in nominal position, so it's standing in for the name of a sentence, not a sentence. But you've given me a few ideas. Thanks!
posted by painquale at 12:10 AM on March 1, 2005

This is a tall order. I can't think of base grammar models in the IE or afro-asiatic languages that would work. Perhaps in some agglutinating language, this would be feasible-- some affix would insinuate the veracity of the direct object.

Of course, the most likely scenario in agglutinating languages is that the second clause because unnecessary-- an affix makes the veracity implicit in the conditional: "If John said it [+ affix confiming veracity]."

So we can narrow your search. If you really want to find an example, it's probably going to be from Africa or Oceania or maybe, maybe SE Asia. Other languages that I can't rule out (listed increasingly from "least likely" to "who knows":

Sino-Tibetian Family
Finno-Ugric Family
Turkic/Mongolian Family
Georgian
Basque

Families and isolates where I'm fairly certain you won't find this:
IE
Amerind (so agglutinative that complex sentences are reduced to a couple words)
Afro-Asiatic
Japanese
Korean
posted by Mayor Curley at 5:49 AM on March 1, 2005

afaik you won't find a real pronoun making up an entire root clause (including the consequent of a conditional). You will, however, find ordinary English pronouns and demonstratives in places where full clauses can also appear:

(1) John believes that Mary went to the party.

(2) John doesn't believe that!

In this case, if (2) is spoken after (1), "that" clearly means something like "that Mary went to the party". You can also do something similar with "it", and possibly "this".

Also, one could imagine that the word "yes" is the kind of root-clause anaphor you're looking for.

(3) Is John going to the party?

(4) yes.

(5) If Mary comes, then yes.

"Yes" in (4) and (5) clearly means something like "John is going to the party."
posted by advil at 8:52 AM on March 1, 2005