Disclaimer: I won't listen to the podcast, because I am boycotting any medium that is not text.

Language learning may have extremely low ROI in general but extremely high in special cases. E.g. I would not be surprised by finding that people learning the language of the foreign country they live in increases their subjective wellbeing. Or if people want to work as translators. Or they are investors and are specialising in a region not speaking English as their main language.

This almost seems like a fallacy. I might call it "homogenity bias" or "mistaking the average for the whole" just to find out that is already known under a different name and well documented.