What does the semantics of an imperative consist in?

The natural analogue of truth-conditions for imperatives seem to be compliance conditions, which a conduct may or may not satisfy. Just like declaratives partially specify a state of affairs, then, imperatives would partially specify a conduct.

However, this view is not quite right. For instance, while (1a) has more stringent compliance conditions than (1b), from (1a) one cannot infer (1b).

1a. Post this letter!
1b. Post this letter or burn it!

In our proposal, what an imperative partially specifies is not a definite conduct, but rather a set of options. In this view, (1a) is not stronger than (1b): for, unlike the former, the latter implies that burning the letter is consistent with the relevant set of options.

 

References:

Maria Aloni and Ivano Ciardelli, A logical account of free-choice imperatives, 2013. The dynamic, inquisitive, and visionary life of ϕ, ?ϕ, and ◊ϕ. A festschrift for Jeroen Groenendijk, Martin Stokhof, and Frank Veltman.