Well, "no" is a disagreement marker, then, so I don't think that's broad enough.

There are various terms like "discourse particle" you could use but I'd think that just talking about "yes" and "no" would be enough. Is that a problem?


In my opinion, these are actually pro-forms, like pronouns but for whole sentences. That's not widely used terminology though

In a language like Latin, the answer for "yes" is to repeat the verb in the question:
A: Did you read the book?
B: I read. ['Yes.']

Therefore, I believe that the word yes or no is substituting for a whole sentence. Consider these other proforms:

John walked home then he went to sleep.
John walked home quickly, but Mary less so.
John went home by bus, but Mary did so by train.

Now consider (with the right intonation!):
John won the game, but Mary, no.
John didn't win the game, but Mary, yes.


A simple way to think about this is what part of speech these words are. I can only think that they substitute for full sentences.

I mean, it's certainly a valid hypothesis for Latin and languages that have that feature, but once the structure changes, it's effectively a different phenomena, no?

I'm not sure about that-- it seems like the same structure to me. It is not literal, certainly, in that the event is constrained contextually. One could argue, I think effectively, that the tense on the verb is strictly bound by the tense in the original question. It's anaphoric in that sense.

I suppose when an analysis is forced based in the reigning paradigm of syntax, this is a pretty decent hypothesis.

Haha, not sure I think of it that way. I actually thought of this a long time ago when I was talking to a museum curator about how to label "yes" and "no" in discourse transcripts regarding immigration. I couldn't think of what they should be called. They're probably just boring old "adverbs" in traditional grammar, or we can be lazy and call them "particles", but that doesn't seem insightful to me. They seem to (with polarity) represent entire sentences. So I'll call them pro-sentences. I don't know of anyone else whose made a similar argument, though, so I may be out there on this one

Oh, cool. I hadn't ever come across that

To elaborate, I'm interested in comparing the semantics of 'yes' (and Indo-European 'yes' words) with agreement markers which have a similar functions in other languages. Latin, as stated, has a different strategy as do other Amerindian languages which have words like 'like so' (so-comparative) to mean something like 'yes'.

But under what aegis are both 'yes', repetition strategies, and 'like-so'?

Pragmatic function, I suppose.

Beyond that, there's almost certainly no syntactic category that could unify them all. After all, as you observe, they work very differently.

Quote from: MalFet on May 03, 2014, 09:07:48 PM

though it quickly gets thorny when you start in with second-order semantic configurations.

Can you give an example of a second-order semantic configuration? I've not come across that concept before, and I wanted to understand the full point you were making.

I just mean semantics based on second-order (and higher) logic. Particularly when you start getting into type theory, the utility of understanding yes/no as proforms starts to falter.

A very trivial example:
- Is John a jerk?
- Yes [="John is a jerk"]

- John is a jerk.
- It's mean to say yes [≠It's mean to say that John is a jerk]

Tthe substitution only works with first-order propositions, which has implications for how we understand words like yes and no as functional types. This suggests there's something else going on...something that requires more nuanced attention to discourse-level deixis.

When you start applying substantially different distributional criteria to proforms, your type-level semantics get unbearably ugly very quickly. It's only a slight exaggeration to say that the entirety of Government and Binding was an attempt to deal with this problem.

In other words, postulating the existence of different distributional criteria doesn't fix the problem. It is the problem.