I upvoted your comment to agree with your recommendation of the definitive book on the subject.
To answer the question you pose above about other domains, I have not seen language learning reported in the expert performance literature, but I was a language learner (native speaker of English who was studying Chinese) who perhaps arguably did reach expert level in my acquired language (I passed testing to be a contract Chinese-English interpreter for the United States federal government). It took perhaps 10,000 learning contact hours (many of those hours during a three-year stay overseas after completing my undergraduate degree in Chinese) to reach that level of language proficiency, which was confirmed by other tests. I also had excellent instruction in Chinese with some of the best materials then avaiable, and a lot of supportive help from linguistics that I studied at the time. Some of my advice on language learning
would probably help other learners get the most benefit per hour in their language learning situations that they can.
What I find most interesting about K. Anders Ericsson's research on expert performance is the suggestion that some domains have few or no experts, when experts are defined as persons with statistically reliably superior performance in the domain. The example I recall from one of his research papers is choosing common stocks in which to invest for sustained high returns. Some people beat the market for a while, but most stick-pickers do very little better than simply investing in a diverse basket of stocks chosen without conscious thought.