Seeking Scala equivalent for this Haskell snippet

I was going through an old post of Tom Moertel, where he defines a cute function in Haskell, clusterBy, that clusters an aggregate based on a signature function. Here are some examples of its usage from his post ..

*Main> let antwords = words "the tan ant gets some fat"

*Main> clusterBy length antwords
[["the","tan","ant","fat"],["gets","some"]]

*Main> clusterBy head antwords
[["ant"],["fat"],["gets"],["some"],["the","tan"]]

*Main> clusterBy last antwords
[["the","some"],["tan"],["gets"],["ant","fat"]]


The Haskell implementation uses Arrows and is super elegant and concise ..

import Control.Arrow ((&&&))
import qualified Data.Map as M

clusterBy :: Ord b => (a -> b) -> [a] -> [[a]]
clusterBy f = M.elems . M.map reverse . M.fromListWith (++) . map (f &&& return)


I was trying to implement the same in Scala, but could only proceed to the extent of having separate functions clusterByLength, clusterByHead, clusterByLast etc. ..

val l: List[String] = List("the tan ant gets some fat".split(" "): _*)

def clusterByLength = {
  val m = new HashMap[Int, List[String]]
  l.map {x => (x.length, x)}.foreach {v =>
    m.put(v._1, v._2 :: m.get(v._1).getOrElse(List[String]()))
  }
  m.keys.toList.sort((e1, e2) => e1 < e2).map {x => m.get(x).get.reverse}
}

def clusterByHead = {
  val m = new HashMap[String, List[String]]
  l.map {x => (x.substring(0, 1), x)}.foreach {v =>
    m.put(v._1, v._2 :: m.get(v._1).getOrElse(List[String]()))
  }
  m.keys.toList.sort((e1, e2) => e1 < e2).map {x => m.get(x).get.reverse}
}

def clusterByLast = {
  val m = new HashMap[String, List[String]]
  l.map {x => (x.substring(x.length - 1), x)}.foreach {v =>
    m.put(v._1, v._2 :: m.get(v._1).getOrElse(List[String]()))
  }
  m.keys.toList.sort((e1, e2) => e1 < e2).map {x => m.get(x).get.reverse}
}


Is there any way that I can raise the level of abstraction for the above implementation, and have a Scala equivalent of the Haskell clusterBy ?