When I learned x-bar theory, there seemed to be an implicit assumption that trees were built top-down, from IP or CP to the VP and its complement, etc. However, as I am learning more about Minimalism (keep in mind I still know very, very little), I see that it posits the opposite direction for tree building. I don't really understand this, because it seems to me that we perceive speech and plan speech, well, from the beginning to the end. I have wondered that if the "bottom up" approach were true, shouldn't we expect to see more speech errors where people stumble on the beginning of the sentence and then finish executing it perfectly, because they have successfully planned the bottom part of the tree/sentence but haven't gotten to or correctly built the top? (Instead of the opposite which usually seems to happen: people start out creating a grammatical sentence and then have an error and the rest becomes incoherent.)
Bear in mind, I do not know a lot about Minimalism and don't really want a Minimalism-centric answer. I am just wondering what general theoretical motivations are there for building directionality (for lack of a better term) -- or if syntacticians disregard this matter completely, why.