|No such thing as a small change|
1) Focus on perl's unique selling proposition, which includes:
- the fact that only perl can be used to glue other programming environments together - for example, IPC:Open2 can farm functionality out to other languages and process the results. C and Java can't do that, causing expensive coding whenever something difficult or specialised comes along.
2) 3) I don't know what your scalability issue is.
4) It is perceived as better because most java programmers have never been exposed to solving the things perl can solve and so they don't know any better than to think this. There have also been some benchmarks on the internet that are biased against perl and in favour of java. But there are other benchmarks that make perl look better than java. The reason is that unless the tester has god-like skills, all the test is doing is testing the holes in the tester's knowledge. i.e. if it shows perl as slower it is demonstrating most likely that the test written in perl is naively written - it is not saying anything about perl itself.
5) Java is proprietary, so there is a commercial interest in 'demonstrating' that java is better than perl. So yes, it is reasonable to say that the stories about java being better are rooted in propaganda - there is no huge corporation pumping out perl propaganda by comparison.
I would advise preparing your own comparative analysis with your own benchmark tests that fairly address what you will actually run - perhaps some sample code in perl, java and whatever else that does a small but functionally and technically representative piece of the functionality the business will actually do.
Free your mind