That's true -- in the U.S., queries are much more commonly associated with the IT field (run a query) than with simply asking a question.
What follows, Barb is not an attempt to prove you wrong, but rather a suggestion that corpora, while immensely valuable, should not be regarded as infallible deities.
I believe what you wrote - I've heard it myself, but the
Corpus of Contemporary American English (COCA) gives 132,844 citations for 'question', 1,187 for query and, for 'run a query'? None. The
[bnc] BNC Simple Search - Using the BNC is equally free of 'run a query' citations.
This goes to confirm, I feel, that present corpora are, as yet, too dependent on the written word.