Significance testing is poor and confusing way of doing statistics that is used mostly in the humanities. The standard method of estimation is much better at conveying the magnitude of the difference and accuracy and likely repeatability of the experiment. Most academic who use significance testing never understood first year statistics anyway. There is a good argument on this issue here:
http://www.abc.net.au/rn/ockhamsrazor/stories/2011/3333636.htm
I suspect that one of the reasons that Quorum did better in the experiment what that it uses white space rather than semi-colons and brackets. Misplaced semi-colons and brackets are common trivial errors that experienced programmers make. For novices writing in an editor (without any syntax highlighting or compiler errors to pick them up) under exam conditions that would be a major issue. It will be interesting to see their suggested further research with larger sample and languages like Python and C included.
On Perl. Perl was not a carefully designed language. It was intended for for text processing and just grew with the functionality of a shell, sed, awk, REs, bits of OO and whatever else you can think of thrown in. It is large, irregular and requires of lot of learning by rote and practice to master. It is very powerful and efficient at doing what it does but I don't think anyone is going to claim that it is a good language for quickly teaching the concepts of programming to beginners or the disabled. Which is what the authors of the paper are interested in and are comparing.
And if Perl is so wonderful and easy to use then why have they spent over a decade trying to clean up the syntax, reduce the number of ways of doing the same thing and improve the implementation with Perl 6? A neater language with Perl's functionality may have been a good idea but it has taken so long that other languages have been developed to fulfill roles (PHP, Java, Javascript, C#, Python, Ruby, Haskell etc) along with languages designed for modern issues like concurrency and cloud based applications e.g. Go, Opa and Dart.
Perl's main advantage are that there are lot of people who can do stuff quickly in Perl 5 and there is a lot of code available on the net. Perl 6 will now have little to offer because you would have to rewrite everything. It will be easier to back port new functionality where useful into Perl 5 (as is happening e.g with say (which is another illogical bit to have to learn by rote)) than to do that.