Revisiting 32: Exception-triggered alternation

The phenomenon, from my 8/25 posting “Exception-triggered alternation”, involves:

Two contrarily opposed states — good vs. bad [in a joke routine], forestressed vs. afterstressed [in the accentuation of N + N compounds] — for some phenomenon, with the choice between them determined by context, the choice flipping between the two as the context narrows more and more (with each flip, the contextually more specific choice overrides the contextually more general one).

“More specific overrides more general” is a familiar principle, known by many names; what the statement above emphasizes is that this overriding can cascade, through a number of iterations.

Now, as an addition to these two examples Larry Horn (Laurence Horn of Yale Univ.) offers another, from formal semantics: Sobel sequences. Here I’ll turn the floor over to Larry, for a guest posting on them. As background, the cover of David Lewis’s Counterfactuals book (in the paperback edition, much more visually exciting than the severe cover of the edition I once had):

In Larry’s words, from here on out:

Arnold’s characterization of the Archie Campbell and Simpsons joke routines in Exhibit A, which as he points out carries over to the (superficially quite different) nominal compounds of Exhibit B, is as follows: “At each twist, flip, or zig, Campbell would introduce new information that makes the situation more specific, which then overrides the good/bad judgment of the previous stage…” This reminds me of the hallowed Elsewhere Condition (a.k.a. Subset Principle) in phonology and morphology, which states that a more specific rule overrides a more general rule in their shared domain. This principle is invoked to explain how womans is blocked by women, stealer by thief, and more good by better. Different versions of the principle were formulated in the 1980s and ’90s by Anderson, Kiparsky, Aronoff, and…yes, Zwicky, but it can be traced back 2500 years to the great Sanskrit grammarian Panini.

More specifically, the routines in the comic sketches reminded me of the Sobel sequence, first identified by Howard Sobel (“Utilitarianisms, simple and general”, Inquiry 13 (1970): 394-449) and popularized (inasmuch as anything in the semantics of counterfactual conditionals can be said to achieve popularity) by David Lewis in his book Counterfactuals (Harvard University Press, 1973), pp. 10-19. A classic example of a Sobel sequence (or “alternating sequence”, as Lewis also calls them) is that in (1):

(1) a. If Otto had come, it would have been a lively party.

(1) b. …but if Otto and Anna had come, it would have been a dreary [= non-lively] party.

(1) c. …but if Waldo had come as well, it would have been a lively party.

And so on.  Or, translated into the format of the Archie Campbell routine:

(1′) a. Otto has come to the party. [Good news!]

(1′) b. But wait — Otto and Anna have come to the party. [Bad news!]

(1′) c. But hold on — Otto and Anna and Waldo have come. [Good news!]

Such sequences are of particular interest because of the problems they pose for some of the otherwise tempting semantic theories of counterfactual conditionals. Perhaps a more colorful sequence based on the counterfactual in the first sentence of Lewis’s 1973 book will render the situation more vivid (or at least more suitable to illustration by a cartoonist).

(2) a. If kangaroos had no tails they would topple over.

(2) b. …but if kangaroos had no tails and used crutches they wouldn’t topple over.

(2) c. …but if kangaroos had no tails used crutches but the crutches were made of rubber they would topple over.

Two points that may be worth pointing out about Sobel sequences:

— (i) as we saw in (1′), such alternating sequences are by no means limited to counterfactual conditionals —

(3) a. If I let go of this pencil it will fall to the floor. [True]

(3) b. If I let go of this pencil but first tie a helium balloon to it, it will fall to the floor. [False]

(3) c. If I let go of this pencil and first tie a helium balloon to it and weight it down with a dumbbell, it will fall to the floor. [True]

— or even limited to conditionals at all. Generics allow the same alternating intuitions:

(4) a. Working class voters support Trump. [True]

(4) b. Working class voters who are black support Trump. [False]

(4) c. Working class voters who are black and served in the military support Trump. [True?]

(4) d. Working class voters who are black and female and served in the military support Trump. [False]

— (ii) these sequences are typically irreversible, as was first pointed out by Kai von Fintel (“Counterfactuals in a dynamic context”, in M. Kenstowicz (ed.), Ken Hale — A Life in Language, 123-152. Cambridge, MA: MIT Press, 2001): (# indicates oddness.)

(5) a. If kangaroos had no tails they would topple over but if kangaroos had no tails and used crutches they would topple over.

(5) b. #If kangaroos had no tails and used crutches they wouldn’t topple over but if kangaroos had no tails they would topple over.(# indicates oddness)

(6) a. If Otto comes to the party it’ll be great but if Otto and Anna come it’ll be awful.

(6) b. #If Otto and Anna come to the party it’ll be awful but if Otto comes it’ll be great.

In (5b) and (6b), we’re trying to move from the specific condition or rule to the general one, which turns out to be much harder than moving from the general to the specific as in (5a) and (6a). (Of course, (6b) is fine if the last condition is given as if just Otto comes or if only Otto comes.)

So why does “If Otto comes to the party it’ll be great” sound fine when asserted on its own in a context like (6a) (or (1a)) but not once Anna’s accompanying him is brought up as a possibility as in (6b)? One way to explain what’s going on is by invoking a general rule proposed by the philosopher Sarah Moss (“On the pragmatics of counterfactuals”, Noûs 46 (2012): 561-86):

If a speaker cannot rule out a possibility made salient by some utterance, then it is irresponsible of her to assert a proposition incompatible with this possibility. (Moss 2012: 568)

You can ignore a possible state of affairs that would threaten your generalization as long as it hasn’t been expressed out loud, but it’s hard to unhear what you’ve just heard.

 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s


%d bloggers like this: