We introduce the task of cross-lingual semantic parsing: mapping content provided in a source language into a meaning representation based on a target language. We present: (1) a meaning representation designed to allow systems to target varying levels of structural complexity (shallow to deep analysis), (2) an evaluation metric to measure the similarity between system output and reference meaning representations, (3) an end-to-end model with a novel copy mechanism that supports intrasentential coreference, and (4) an evaluation dataset where experiments show our model outperforms strong baselines by at least 1.18 F1 score.