Syntactic analysis of search queries is important for a variety of information-retrieval tasks; however, the lack of annotated data makes training query analysis models difficult. We propose a simple, efficient procedure in which part-of-speech tags are transferred from retrieval-result snippets to queries at training time.