If a field's term vectors are stored, will phrase queries against it be faster?
For example, if a field has "term_vector": "with_positions", will "match_phrase" queries take advantage of those, and execute faster than if the field had "term_vector": "no" (the default)?
I'm thinking they might, if phrase queries are just AND match queries followed by position validation. But I have no idea if phrase queries are implemented that way, or if it'd even make sense in practice. Part of my motivation here is to just understand how phrase queries work.