Last update: 2008-03-19
This class meeting will be an attempt to get a grip on the discourse strategy that appositive expressions embody. I'll provide some descriptive generalizations and a dynamic logic, but these will take us only so far. I'm looking for suggestions. I'm game to experiment. Some inspiration from Craige Roberts (personal communication):
The main point for me is that I have come to strongly suspect that proper understanding and use of dynamic interpretation will permit one to account for/calculate conventional implicatures without a separate semantic dimension per se. Easy to say, I know, and far more challenging to really work out, but I think it would be worthwhile to try.
Yes! Here goes, for the subcase of appositives.
Most of my examples will involve nominal appositives (1) and nonrestrictive relatives (2), but we should feel free to let the discussion range more widely if that seems fruitful.
(1) Lucille Gorman, an 84-year-old Chicago housewife, has become amazingly immune to stock-market jolts. [Treebank]
(2) uh, she starts a new job tomorrow, which should take her out of the house about four days a week. [Switchboard]
(3) Mr Shepherd, a civil, cautious lawyer, who [...] would rather have the disagreeable prompted by anybody else, excused himself from offering the slightest hint [...] [Persuasion]
These constructions are extremely common in written discourse (van Delden and Gomez 2002), and they are well-represented in speech as well, particularly sentential appositive relatives like (2). So speakers are evidently compelled to put them to use. Why? In particular, what do appositives bring to an utterance that distinguishes them from sentence-sequences like (4)–(6)?
(4) Lucille Gorman is an 84-year-old Chicago housewife. She has become amazingly immune to stock-market jolts.
(5) She starts a new job tomorrow. It/That should take her out of the house about four days a week.
(6) Mr Shepherd was a civil, cautious lawyer. He would rather have the disagreeable prompted by anybody else. He excused himself from offering the slightest hint [...]
The details of intra-sentential semantic composition provide reasons for distinguishing apposition from sentence-sequencing (Potts 2005), but these facts don't tell us much about the pragmatics. Dynamic semantic theories are more helpful here. I'll run through all the tests I know of, involving anaphora, ellipsis resolution, and dynamic tests. The result of that work is at least one clear generalization:
(7) Appositives are updated at the point at which they are encountered in the string, moving left-to-right.
This contrasts with, for example, a view on which appositive updates happen before the main-clause (a sort of forced, seamless, accommodation) or after (perhaps because they are "secondary").
Generalization (7) is useful, especially when considered alongside what we know about appositive semantic composition. What's more — and this is really encouraging — dynamic semantics has progressed to the point where we can fairly readily develop theories that embody these descriptive generalizations. (I've not worked much in this area; I'm largely a consumer.)
I'll present one such logic. It is a slightly simplified version of one developed by Groenendijk et al. (1996), but enriched with the prominence-driven approach to anaphora of Dekker (1994) and Bittner (2001, 2003). Its dynamics are subsentential enough to break an example like (8) down into three basic updates and mediate the links between them anaphorically (Sells 1985; del Gobbo 2004):
(8) Lance, a cyclist, is training.
(9) x = lance ; cyclist(x) ; training(x)
I hope to explore this logic in class. To that end, I've written a small Web application that interprets its formulae in a simple intensional model like the one in (10).
(10) Intensional PLA with Apposition
| Intensional model | |||
|---|---|---|---|
| w1 | w2 | w3 | w4 |
| 1, 2, 3 | 1, 2, 3 | 1, 2, 3 | 1, 2, 3 |
The numbers have their expected numerical properties (prime, even, ...) as well as the non-necessary properties depicted here (large, green, ...). We begin with all four worlds as possibilities and no discourse information, and then we begin updating:
(11) Something is large. It is red. Something is small. It is green.
[], w1 [], w2 [], w3 [], w4 |
EX large(X) |
[1], w1 [2], w1 [3], w1 [3], w2 [1], w3 [2], w3 [3], w3 [2], w4 |
red(p0) |
[2], w1 [3], w2 [2], w3 [3], w3 |
EY small(Y) |
[3, 1], w2 [3, 2], w2 |
green(p0) |
[3, 2], w2 |
Existential statements extend the sequences (empty at the start) in every way that is consistent with their scope. Thus, after EX large(X), there is a new column, filled with large things. The pronoun p0 picks up on it. The new existential introduces a new column, and p0 now picks up that new column. (p1 now picks up on the column behind it, and so forth for all pN defined for the current state.) As discourse information accumulates, the set of possible worlds is reduced. In this example, we even determine that we are in world w2: the unique world containing a large red thing and a small green thing.
In addition to the logical operators of Groenendijk 1996, there is also a simplified additive particle + (meant to behave like a simplified also). This is useful for probing the order in which content is updated:
(12) Something is red. Three is red too.
[], w1 [], w2 [], w3 [], w4 |
EX red(X) |
[2], w1 [1], w2 [3], w2 [2], w3 [3], w3 [1], w4 [3], w4 |
+red(3) |
[1], w2 [3], w2 [2], w3 [3], w3 [1], w4 [3], w4 |
(13) #Three is red too. Something is red.
[], w1 [], w2 [], w3 [], w4 |
(+red(3) & EX red(X)) |
additive presupposition of + unsatisfied
[], w1 [], w2 [], w3 [], w4 |
And, of course, there is a binary appositive operator -, defined so that (A - B) updates first with A and then with a discourse-level update of B. (Nouwen 2007 develops related ideas.) These formulae are equivalent to conjunction in unembedded cases:
(14) Two, a red number, is large.
[], w1 [], w2 [], w3 [], w4 |
((EX (X = 2) - red(p0)) & large(p0)) |
[2], w1 [2], w3 |
(15) Two is a red number, and it is large.
[], w1 [], w2 [], w3 [], w4 |
((EX (X = 2) & red(p0)) & large(p0)) |
[2], w1 [2], w3 |
However, under operators like negation and (test) modality, the differences are striking:
(16) Two is not both red and large.
[], w1 [], w2 [], w3 [], w4 |
~(red(2) & large(2)) |
[], w2 [], w4 |
(17) Two, a large number, is not red.
[], w1 [], w2 [], w3 [], w4 |
~(red(2) - large(2)) |
[], w4 |
This is the widest-scope property of appositives, implemented so as to respect the ways in which appositives do and don't support discourse anaphora and presuppositionality.
In class, we'll pause at this point to play with the Web tool. It's my hope that the properties of the logic shine through despite the limited lexicon and artificial syntax. Here are two quick examples showing that appositives support the sort of anaphora that additive particles determine:
(18) One, which is small, is red. Three is small too.
[], w1 [], w2 [], w3 [], w4 |
((EX (X = 1) - small(p0)) & red(p0)) |
[1], w2 [1], w4 |
+small(3) |
[1], w4
|
(19) Two is large. A green thing, which is large too, ...
[], w1 [], w2 [], w3 [], w4 |
large(2) |
[], w1 [], w3 [], w4 |
(EX green(X) - +large(p0)) |
[1], w1 [3], w1 [1], w3 [2], w4 |
We can't stop here, though. The truth is that the logical dynamics still doesn't make enough distinctions between pairs like (1) and (4), repeated here:
(20) Lucille Gorman, an 84-year-old Chicago housewife, has become amazingly immune to stock-market jolts. [Treebank]
(21) Lucille Gorman is an 84-year-old Chicago housewife. She has become amazingly immune to stock-market jolts.
These lead to all the same information states. This is illustrated in (14) and (15) above, in fact. The order of updates is different, and this means that the sequences of states is different. However, despite the advertising, most dynamic theories don't make significant use of update-order. It seems that we still don't have a grip on why speakers choose appositives over other options.
I'm still working on how to resolve this, but my current strategy is to accept that appositives and sentence-sequences often are equivalent, in terms of their truth conditions and in terms of the general information flow. I will look instead to the pragmatics. I think the crucial factual consideration is the asymmetrical relationship between appositives and the stuff that anchors them, and I think the crucial formal insights lie waiting in the information-theoretic ideas of van Rooy 2003 (and related papers). I'll have more to say about this in class. I'm still trying to work out the details.
Bittner, Maria. 2001. Surface composition as bridging. Journal of Semantics 18(2):127-177.
Bittner, Maria. 2003. Word order and incremental update. In Proceedings from CLS 39. Chicago: Chicago Linguistic Society.
Büring, Daniel. 1998. Identity, modality, and the candidate behind the wall. In Devon Strolovitch and Aaron Lawson, eds., Proceedings from SALT 8, 36-54. Ithaca, NY: CLC Publications.
Dekker, Paul. 1994. Predicate logic with anaphora. In Lynn Santelmann and Mandy Harvey, eds., Proceedings from SALT 9, 79-95. Ithaca, NY: CLC Publications.
van Delden, Sebastian and Fernando Gomez. 2002. Combining finite state automata and a greedy learning algorithm to determine the syntactic Role of commas. In Proceedings of the 14th IEEE International Conference on Tools with Artificial Intelligence, 293-301.
Del Gobbo, Francesca. 2003. Appositives at the Interface. Ph.D. thesis, UC Irvine.
Groenendijk, Jeroen, Martin Stokhof, and Frank Veltman. 1996. Coreference and modality. In Shalom Lappin, ed., The Handbook of Contemporary Semantic Theory ,179-213. Oxford: Blackwell Publishers.
Nouwen, Rick. 2007. On appositives and dynamic binding. Journal of Reserch on Language and Computation 5(1):87--102.
Potts, Christopher. 2005. The Logic of Conventional Implicatures. Oxford University Press.
Sells, Peter. 1985. Anaphora with which. In Jeffrey Goldberg, Susannah MacKaye, and Michael T. Wescoat, eds., Proceedings of WCCFL 4, 299-313. Stanford, CA: The Stanford Linguistics Association.
van Rooy, Robert. 2003. Questioning to resolve decision problems. Linguistics and Philosophy 26(6):727-763.