Modeling multiple many-to-many relationships based around a language component

Power BI has been built with the thought that brute force should be its main strength. Also, the compression algorithms make sure that milions and millions of rows are handled gracefully with speed. Secondly, many-to-many relationships in Power BI have been designed to solve one particular problem: that of granularity. You don't seem to have it in your model, so your relationships are simply wrong. Granularity issue raises when one table needs to join to another on a higher level than its own grain. A classic many-to-many relationship can only be handled, in Power BI as well as in MD cubes and data warehouses, through bridge tables. ONLY. Also, if you start creating relationships directly between fact tables, then be prepared that you'll get into big trouble sooner than might think. If you want to stay sane and have an easy time creating simple and fast DAX, please build a model that is a proper star-schema. If you forgo this advice, all hell will unleash.

You've been warned 🙂