Applying Language Technology methods to historical archives is a challenge for NLP-related research: data is heterogeneous (image /text), often incomplete (e.g. OCR errors), multilingual within one document (historic documents with Latin or/and classical Greek paragraphs) and difficult to structure (paragraphs, titles, pages are somewhat different in historical texts). Corpus-based methods, nowadays standard in NLP research, often cannot be applied as the necessary large training data is missing. This was the focus the second edition of Language technology for Digital Humanities in Central and (South-)Eastern Europe workshop. In the 2019 International Year of Indigenous Languages this edition expands also to Middle East and North Africa.