Is there a difference between these three concepts: sentences, utterances, propositions. This is the first topic tackled in a series of handouts on the practice of formal semantics in a generative grammar (created by Chris Potts at U Mass). Of course, you already know the answer the handout will give: yes, beyond the shadow of a doubt.
So, what are they? Well, maybe utterances are easiest. They are physical events, located in space and time (as events like to be). They involve two “participants” – an agent who produces a linguistic object and that linguistic object itself. Good so far. What’s a sentence? It’s some abstract entity “produced by the grammar,” though divorced from the notion of a speaker (or writer, or whatever) and from meaning. Okay, good, good. And a proposition? Well, it’s basically some sort of “idea” that can be (quoting Potts) “specified with language.”
How can we tell they’re different? We show that they don’t map one-to-one with each other. Some of these are easy. For instance, a single sentence (read ‘grammatical entity produced by a speaker’s grammar’) can be used in multiple utterances. For instance, the linguistic entity,
[I didn't quite catch how many traces there were between your main verb and the direct object]
can be used on multiple occasions, i.e., for many different utterances. Great, so one sentence corresponds to many utterances. Same goes for propositions: the proposition Chomsky fooled all of you! can be expressed by either “Chomsky fooled all of you!” or “I fooled all of you” (spoken by Chomsky), and so on. And a single utterance can contain multiple sentences. I guess. I mean, it depends on your definition of utterance. But let’s be nice and consider that “I like him: he’s nice” contains both sentences [I like him] and [he's nice]. Great. More? Okay, there’s lots more, but let’s get to the fun part.
“A single utterance can correspond to more than one proposition,” Chris writes. Example given: “It’s cold in here” corresponds to both propositions It’s cold in here and Someone should close the window. Okay…wait. What was a proposition, again? Let’s say someone says the above utterance, and then someone responds: “It IS cold in here.” So, did that “correspond to” (Chris’ words) the propositions: It is cold in here, I agree with you that it is cold in here, Someone should close the window, I agree with you that someone should close the window. It’s not quite like he’s saying that perlocutionary effects are included in propositions, but it’s close.
How about this: someone asks someone apparently paralyzed if they can speak, and they respond verbally: “It’s cold in here.” Along with the two propositions established above, is another “proposition” that “corresponds” to this utterance Yes, I can speak you heartless bastard, or something like that?
It’s very tricky. I, when reading this handout, thought this:
Does this mean that the proposition someone should close the window can correspond to the sentences [someone should close the window], [it is cold in here], [I don't like open windows], …
Well, this depends: what’s the connection between sentences and utterances, and between these two and propositions? If a “proposition” is some pragmatic entity, including “assertions” made by the speaker to effect a perlocutionary effect, and if a sentence is an entity of the syntax module, then there should be no direct connection – the production of an utterance should mediate these two things. This seems clear to me now, but unfortunately Chris Potts’ handout makes things a bit confusing by using similar vector graphics to illustrate the relationships between these entities. It also requires an unpacking of various verbs:
- A single utterance can correspond to more than one proposition
- A single utterance can correspond to more than one sentence
- A single sentence can be used in multiple utterances
- A single sentence can convey more than one proposition
- A single proposition can be expressed by more than one sentence
- A single proposition can be expressed by more than one utterance
So utterances map to propositions and sentences, and sentences are made use of for the production of utterances and the communication of ideas/propositions, and propositions the the byproducts (are expressed by) of utterances and propositions. Of all of these, I’m dubious but willing to accept that a single utterance can contain multiple sentences, as long as you tell me what mechanism turns sentences into utterances, and in particular what conjoins them that is not syntactic. However, I (and others learning this stuff) seem to not to be comfortable with the idea of a sentence expressing propositions, let alone multiple propositions, if that’s what utterances are for.
And now for something slightly different, where I’m going to try to say something other than “language is based in human experience.” Let’s consider what a sentence is. A sentence is something “linguistic,” (Potts’ word), which we took to mean “produced by a grammar.” How can we tell what is produced by the grammar and what is not (and what is a sentence and what is not)? We need speakers to determine the rules of the grammar, for only speakers and trained analysts can determine the rules of a language (oh, and children as well, and also algorithms trained on corpora, but hey, look at that giant moose!). The resulting grammar will be the result of humans figuring out the grammaticality of sentences. But how does one determine the grammaticality of a sentence? You have to test if you could actually say it, that is, imagine using it. Fine, but that’s not a sentence; it’s an utterance. Maybe you can smack your mind around the room until it believes in a speakerless sentence, produced by a speakerless set of rules. And when you mind starts to point towards the next room over, where linguists are muttering to themselves as they imagine trying to say “It was the store that I went to the bank and,” well, you just give it another good thrashing.
What I’m saying is, I don’t know what I’m saying. I understand the need to differentiate between “ideas expressed,” “act of expressing ideas,” and “means of expression, linguistic (and otherwise?).” I just don’t have the right drugs around to keep them separated in my head.