Actually they have some foundations, componentN functions serve similar role to unapply, so theoretically it should be possible to build good enough pattern matching. But, unfortunately, they seem to slow down language development a lot since release.
It has destructuring in value binding and lambda parameters, as well. Pretty handy when you're dealing with a lot of pairs, triples, and data classes, though not quite as powerful as true pattern-matching.
Kotlin is still a pleasant language to use, and some of the ideas have some good carryover to Scala, with quite a lot less culture shock than going right from Java to Scala.
