The University of Edinburgh participated in the WMT19 Shared Task on News\nTranslation in six language directions: English-to-Gujarati,\nGujarati-to-English, English-to-Chinese, Chinese-to-English, German-to-English,\nand English-to-Czech. For all translation directions, we created or used\nback-translations of monolingual data in the target language as additional\nsynthetic training data. For English-Gujarati, we also explored semi-supervised\nMT with cross-lingual language model pre-training, and translation pivoting\nthrough Hindi. For translation to and from Chinese, we investigated\ncharacter-based tokenisation vs. sub-word segmentation of Chinese text. For\nGerman-to-English, we studied the impact of vast amounts of back-translated\ntraining data on translation quality, gaining a few additional insights over\nEdunov et al. (2018). For English-to-Czech, we compared different\npre-processing and tokenisation regimes.\n