FP languages have their own patterns. Some of them can be plausibly cast as "fixing weaknesses" in them, too.

Multiparadigm languages end up with lots of patterns you need to learn, because while multiparadigm languages may support many approaches to a given problem, generally there are definitely better and worse approaches, specific to the language in question, and between the number of options, the subtly of their implementation details and how those interact with the problem, and the specific preferences of the language itself (because all multiparadigm languages still have preferences), it can take a multiparadigm language community a decade to work out the best pattern for a given task. The existence of many options inevitably expands the number of wrong options, or if you prefer, "distinctly suboptimal" options, as well.