Answering a Question With a Question

People naively say it’s rude to answer a question with a question. This betrays a widespread but wrong conception of human language use.

The November 20 Dilbert strip illustrates nicely. The boss says stand-up desks will be purchased for those who want them. Wally, perennial slacker and couch potato, responds: “Literally the only good thing about this job is that I can do it while sitting down.”

“How did you get to this meeting?” the boss asks.

And Wally says: “Your chair doesn’t have wheels?”

I said “response” rather than “answer” advisedly. Wally doesn’t answer the question. Questions have meanings that determine sets of answers. For a question like “How did you get to this meeting?” there are indefinitely many answers, all of them descriptions of ways of getting to the meeting (walked, crawled, used a scooter, flew through the air, etc.). Wally does not give an answer, and certainly not what the boss was probably expecting (“I walked”). But for once, Wally is not being uncooperative. What’s going on is that he seems to have run through a chain of unspoken reasoning something like the following:

1. The boss has asked me to specify my mode of travel to this meeting room.
2. Everyone (I assume) seeks to minimize energy output.
3. Walking consumes more energy than scooting along the corridor on a wheeled office chair.
4. The boss would surely know this, and would (by premises 2 and 3) use that mode of travel.
5. But it seems (in virtue of 1) that he does not use it.
6. By 2 and 3 again, he would certainly use it if he had a wheeled office chair and knew its capabilities.
7. The most obvious explanation for his not knowing about the capabilities of wheeled chairs is that he does not have one.

Wally thus has grounds for inferring that the boss uses a chair without wheels, but only via an abduction. He needs confirmation. So he asks a question, phrasing as a special type of clause: the confirmation-seeking declarative clause with rising intonation that has been called a rising declarative by the linguist Christine Gunlogson. These are specialized confirmation-requesting questions.

In asking his question, Wally reveals he is tentatively assuming that the boss uses a nonwheeled chair. This would not be relevant unless it were somehow connected to the boss’s question about transportation modes. There is a way to find a relevant connection: A wheeled chair would clearly provide in principle a means of getting down the corridor to this meeting room without standing up. So the boss has grounds for assuming that “I scudded down the corridor on my wheeled chair” must be the answer that Wally did not give.

The boss’s inference, if he draws it, depends on making an effort to find relevance in the question. Wally’s utterance suggests he has inferred that the boss did not travel in that way, hence perhaps cannot do so; thus Wally has (it seems) hypothesized that it is because of the lack of the right chair type. That in turn implies not only that Wally may have used his wheeled chair to get to the room, but also that he would expect anybody else to do the same thing, and he is wondering why the boss did not, and has conjectured a reason, and wants confirmation that his conjecture is correct.

I am well aware of how tedious it is to spell out all of this sub-rosa inferencing and assumption-making. But understanding that it goes on is crucial to understanding how human languages are actually used. We do not always answer questions directly. We do not always use declarative clauses to convey statements. We present each other with utterances of true (or sometimes false) statements, and with questions (sometimes rhetorical), as moves in a kind of subconscious inferencing game where the moves are aimed at capturing relevance. Quite often we prevent actual life from being as slow and tedious as the account I just gave of Wally’s response by shortcutting mere statement: Rather than convey a fact by formulating a clause that states it, we say something that will trigger the addressee into performing an inference that yields a statement of the fact as a consequence. A question will often serve that purpose just as well as a statement.

We are touching here on the subdiscipline of linguistics known as pragmatics. As argued in Sperber and Wilson’s brilliant book Relevance, it is through reasoning of the kind laid out above that we humans typically operate linguistically. Language use would be complex enough even if we did it literally and explicitly, but instead we do it indirectly, relying on lightning-fast inferencing and mind-reading. We engage in such reasoning virtually all the time when we speak or write or listen or read.