Haskell is king

Haskell has an explicit goal (according to Simon Peyton Jones) of minimizing the pain of strict static typing - of having the fewest possible cases where what you want to express is reasonable but doesn't fit the type system. The useful cases that OOP supports but Haskell 98 didn't are already supported (in a different way) in Haskell via extensions. In fact the modern Haskell typesystem is far superior to anything I know of in the OOP world.

So basically, you just don't need OOP in Haskell. You could do things using Haskell 98 typeclasses that you couldn't do with OOP anyway, and with (stable, widely used) extensions such as higher-rank and existential types you can now do pretty much anything you could with OOP. With multi-parameter typeclasses and type functions, you can go beyond OOP to handle what Java generics can do (probably more, though I'm not that familiar with Java) and a fair chunk of what C++ template classes can. And that's without looking at Template Haskell.

There are some cases where OOP still seems a bit more convenient, but mostly that's a short-term gain that's outweighed by longer-term losses. For example OOP can be a lot more convenient at times because of the pervasive mutability, but Haskell keeps mutability under control for a reason.

HQ9+ has an explicit goal (according to Cliff L. Biffle) of minimizing the pain of writing common example programs - of having the fewest possible cases where what you want to express shows off the features of the language but doesn't fit in one of the primitive instructions. The useful cases that Turing complete languages supported but HQ9+ didn't are already supported (in a different way) in CHIQRSX9+ via extensions. In fact the modern HQ9+ instruction set is far superior to anything I know of in the Turing complete world.

So basically, you just don't need Turing completeness in HQ9+. You could do things using CHIQRSX9+'s X that you couldn't do with Turing completeness anyway, and with (stable, widely used) extensions such as R and S you can now do pretty much anything you could with a TC language. With multi-effect I and C functions, you can go beyond Turing completeness to handle what Java can do (probably more, though I'm not that familiar with Java) and a fair chunk of what C++ template classes can. And that's without looking at X.

There are some cases where Turing completeness still seems a bit more convenient, but mostly that's a short-term gain that's outweighed by longer-term losses. For example TC can be a lot more convenient at times because of the pervasive functionality, but HQ9+ keeps usefulness under control for a reason.