A major challenge of research on non-English machine reading for question\nanswering (QA) is the lack of annotated datasets. In this paper, we present\nGermanQuAD, a dataset of 13,722 extractive question/answer pairs. To improve\nthe reproducibility of the dataset creation approach and foster QA research on\nother languages, we summarize lessons learned and evaluate reformulation of\nquestion/answer pairs as a way to speed up the annotation process. An\nextractive QA model trained on GermanQuAD significantly outperforms\nmultilingual models and also shows that machine-translated training data cannot\nfully substitute hand-annotated training data in the target language. Finally,\nwe demonstrate the wide range of applications of GermanQuAD by adapting it to\nGermanDPR, a training dataset for dense passage retrieval (DPR), and train and\nevaluate the first non-English DPR model.\n