Open-source machine translation has become a viable alternative to Google Translate or DeepL for quantitative text and #TextAsData#Pythongithub.com/UKPLab/EasyNMT)

But would it negatively affect measurement reliability if you used an open-source instead of a commercial translation model in your quantitative text analysis? Our analyses focusing on LDA topic modeling and Transformer classifier fine-tuning suggest that the answer is No! 📊