It's also worth noting that most of PL research does not focus on simply designing new languages. For work that does propose new language features/constructs, it is typically done within the context of a small lambda-calculus/ML-style language to allow the authors to easily explain the implications of and prove properties about the feature. If the authors do take the next step, they'll often add the feature to an existing language (Java/C/Scheme) rather than design a new one from scratch. It's up to practitioners to take the most useful of these "idea-nuggets" and include them in new languages. Most academic researchers view their job as generating these new nuggets, not building and supporting tools based on them for wider consumption.

Work on new type systems is a big part of the field, however these type systems aren't always intended to be deployed as part of languages. Rather, they can often be viewed as bug-finding program analyses. Indeed, program analysis (including type inference/checking) is a much more common research topic than "a new language for X".

I think many in the field agree that more work should be done on language usability, although it is a hard topic to research. The PLATEAU workshops (on the Evaluation and Usability of Programming Languages and Tools) were efforts in this direction. The Software Engineering field is much more focused on usability and would be the most likely publication target for PL usability research. (The top PL conferences focus on theoretical concerns (POPL, ICFP) and implementation/systems (PLDI, ASPLOS)). Although usability studies are rare, the PL field has gotten tougher about requiring strong arguments about the usefulness of research: the evaluation and motivation sections of papers really need to be convincing.