Defining utterance

[randall henry eggert]

Christian Nelson requested that I give the citation for Hankamer and Sag.

	J. Hankamer, and I Sag. 1976. Deep and Surface Anaphora.
	Linguistic Inquiry, Vol 7, #3.

I pointed to this article with regard to utterances that span more than
one turn.  The article, as the title implies, concerns anaphora.  The type
of example I had in mind is the following (p. 410).

	Hankamer: Ivan is now going to peel an apple.
	Sag: And Jorge, an orange.

Their argument is that Sag's utterance is only acceptable in conjunction
with Hankamer's utterance, since 'gapping' can only occur if there is a
prior overt expression of what is being gapped. In looking back at their
discussion, I notice that they (like I have done) refer to two different
utterances. In retrospect, in my zeal to make a point, I think made an
error, i.e., this is not really an example of an utterance that spans 2
turns. However, this example supports one of my other points: there is not
a one-to-one correspondence of sentence to utterance. McCawley (1984)
makes a strong case that Hankamer and Sag's example is a single
sentence which switches animators (in Goffman's terms) part way through.
If this is true, and if this is an example of two separate utterances,
then there is not a one-to-one correspondence.

Although I agree with a great deal of what Celso Alvarez Caccamo wrote, I
would like to respond to his response. I will try to be brief.

Caccamo took issue with my comment that a proposition is a "semantic
entity".

>
> May I utter ;-) a minor qualification? A proposition is a "semantic
> entity" only insofar as we understand "semantics" to mean
> "propositional semantics" ;-), or "veritative semantics". It's
> quite a circular argument.

I'm not sure I'd call it a circular argument that I was making.  If
anything, I was proposing that by definition a proposition is a semantic
entity. That does not mean that semantics is limited to dealing with
propositions, since I would be the first to agree that semantics covers a
great deal beyond propositions. What I meant to say is that when
propositions are discussed by linguists, they are considered with relation
to semantics.  I did not mean to say that when linguists discuss semantics
they only consider propositions.

Caccamo also differed with my statement that a sentence is defined
morphosyntactically.

> I differ, again. A sentence HAS semantics, but linguistic
> semantics (that is, meaning = "reference" + "sense" (Hurford &
> Heasley), including for example deixis and, in my book,
> pressupositions. Notice that semantic pressupositions
> escape propositional content, e.g. in "I cheated again in the
> exam", the pressuposition contained in "again" ('I did it at
> least once more earlier') is not a part of the propositional
> content. The proposition only predicates about "I" (the
> speaking subject) having "cheated" "once more", but
> it does not predicate about my having cheated earlier.
>
> Similarly, the utterances "I didn't do anything wrong" and
> "I did nothing wrong" share an identical propositional content
> but they are semantically different, as "nothing" and "not...
> anything" are different linguistic elements/constructions.
>
> Finally, three utterances such as "Read a lot!", "You read a lot"
> and "Do you read a lot?" are enunciations of different sentences
> with identical propositional content, as the three predicate
> the same about your reading a lot. They differ, of course, on sentence
> modality and illocutionary force, but that's not
> propositional meaning.

I agree with all of this.  However, it does not change my point. My point
was that a sentence is defined morphosyntactically.  In English, for
example, a sentence must have a syntactic subject and a verb phrase.  Thus
we get ambient 'it' in sentences like 'it rained'. 'Rained' is not a
sentence in English for syntactic reasons - not for semantic or pragmatic
reasons.  Likewise 'John smart' is not a sentence in standard American
English, not for semantic or pragmatic reasons, but because the syntax
requires a copula.


>
> > My sense
> > is that there is no structural definition of utterance within pragmatics,
> > which is to say that an utterance does not have a one to one
> > correspondence to sentence; an utterance may be as short as a word (or
> > a sound or gesture?) or as long as, well... who knows?
>
> Well, the shortest (and simplest) definition of utterance I
> know of is "Utterance is a sentence-context pair" (by Fillmore?,
> I think). And I would point out that at least in Searle's
> "expressability principle" an utterance is nothing but the
> canonical production of a "sentence", and thus the study of
> speech acts is (theoretically) nothing but the study of
> (prototypical) sentences.

The sentence-context pair definition has been used by quite a few people,
I believe. But not everybody seems to follow it, and I think it will lead
to some problems.  For one problem, c.f. the discussion at the top of this
message.

If you try to follow Searle too closely on this you'll have some real
problems.  Especially when you consider indirect speech acts, which well
might span more than one sentence.  Likewise, if we consider Hancher's
(1979) discussion of cooperative speech acts, such speech acts will be
impossible to equate one-to-one with sentences, utterances, or turns.

>
> Here "sentence" must be understood, of course, as a "sentence-type"
> which underlies the production of a (fragment of) speech. Thus,
> "No!" as an utterance is the pairing of the sentence, e.g.
> "I don't want to" with the context in which the utterance is
> produced.

Personally, I find this practice a bit questionable.  Regardless, it does
not change the fact that the utterance, itself, is not defined
structurally.

I'm sorry if I've bored you all with my response.  I was simply afraid
that in my desire to be brief in my original message I was not clear
enough.

Happiness,
Randy Eggert