Language stability is not a technical decision. It is a contract between the language author and the users.
The author advertise a given version as more or less stable. The less stable a language is, the more changes the author can make. Each user interested by the language can decide if he wants to invest time in it to learn new features or develop applications that might be broken by next month's update.
Using an unstable language can be interesting because you are interested by a new concept, or you want to help by giving your feedback. If you are a business, you might prefer to wait for a technology to be more stable before investing your time in it.You care more about stuff like time to market, and user experience.
So this is a communication & trust issue. Look at the rust language development. They are crystal clear about what they are changing and what they are keeping. When they want to delay a decision about a given feature, they use what they call a feature gate. On the other side, the angular team faced a lot of anger over their 2.0 announcement because the changes were larger than expected.
Even libraries author have to communicate about the stability of their apis. Pretty much any technology used by other people have to strike a balance between stability and perfection. A car maker cannot change the position of the pedals, and a laptop designer won't invent a new keyboard layout for the same reason: you are not helping your users if you cannot make a decision about the way they will use your product.