Arnim von Stechow, Sveta Krasikova, & Doris Penka. “Anankastic Conditionals”. ms.
We propose a new analysis for anankastic conditionals, which are exemplified by the following pattern due to Kjell-Johan Sæbø
a. You have to take the A train if you want to go to Harlem.
b. If you don’t take the A train you can’t go to Harlem.
c. To go to Harlem you have to take the A train.
All these are claimed to be truth conditionally equivalent. The hardest problem is a compositional analysis of (a): what is the role played by “want” in the antecedent? Our proposal starts from the observation that (a) is elliptical. Its overt forms is (a’):
a’. If you want to go to Harlem, you have to take the A train to go to Harlem.
The consequent in (a’) contains no “want” and is in fact (c). We propose a counterfactual analysis for (c): the sentence is true if you take the A train in the nearest worlds where you go to Harlem. The want-antecedent in (a) adds a felicity condtion: the sentence can be uttered appropriately only in contexts whose conversational background is compatible with the proposition that you want to go to Harlem. As far as we can see, this analysis solves all the problems that have been discussed in the literature about anankastic conditionals.
The draft contains some responses to the paper on the same topic that I wrote with Sabine Iatridou and to Janneke Huitink’s recent talk on the topic. I suppose it really is time for us to revisit the topic and to respond to this new work.
Well, I have two initial questions, if you do not mind.
{A} Firstly, I want to understand why (I) below would be a counterfactual in a natural language?
For me a counterfactual sentence related to (1) would be something like (2):
But if the necessity of taking a certain train is a fact, inasmuch as the need of air and water, why would it be something contrary to the fact?
For me, what seems to be case is that first there is a presupposition-filtering mechanism at work. But this is not something limited to counterfactuals in human languages.
Perhaps it is just my ignorance on the subject, in which case I be your pardon.
{B} Secondly, the frequent use of paraphrases can in some cases be misleading in the study of human languages. And word or constituent order is a very important factor to be taken into consideration. Human languages are not commutative. Thus, I am not sure whether in all languages a sentence like
is really perfectly equivalent to:
I can think of pair of (counter-)examples in English:
The semantic and pragmatic differences between a and b above clearly depend on the order of the clauses.
For the syntax-semantics interface there are problems too. If one considers binding relations, then it is clear that the paraphrasing will not produce perfect equivalents:
Maybe it is just my little grasp of what Stechow and ali mean, but I would like to see some answers to these questions.
December 2nd, 2004, at 11:56 pm #PS: I have added three other questions about their assumptions in my blog.
December 4th, 2004, at 2:10 pm #Still persisting in the question of whether the conditionals in question are real counterfactuals, let us just add one extra set of tests. In Kai von Fintel’s (2001) article on Counterfactuals, which can be found in his papers archive, it is explained that among the properties of counterfactuals are failures of some monotone inferences. Let us quote two of the examples given:
Ok, now let us see some sentences like the ones discussed above and apply the tests:
In the scenario where it is known that two businessmen can only negotiate if they meet in the southern Brazilian city of New Hamburg, the pair below is intuitively ok to me:
Thenceforth, in the examples examined above the anankastic conditionals do not behave as counterfactuals.
December 7th, 2004, at 2:59 am #If anyone thinks I am wrong, please correct me.
Erratum:
December 7th, 2004, at 5:02 am #