I spotted this in comp.lang.smalltalk. The link is to the original post - Niall Ross answers:
One can make the point that this is the only data there is. Noone has published any data telling a contrary story (that I know of). Making a decision based on scanty evidence is wiser that making a decision against what little evidence there is.
Lacking more data, one can sometimes use big name quotes, e.g.
- Kent Beck (Mr XP) said (at ESUG 1999 in Ghent) that, "I program at a third of my Smalltalk speed in Java."
- Bran Selic (RealTime UML guru, built the tool that is now the RT part of Rational Rose) said (at UML 2000 in York) that, "I was sorry to see Smalltalk lose out to Java because Smalltalk was the only language in which I noticed a real difference in my productivity".
As both these people make money from selling books that Java programmers buy, providing tools that Java programmers use and attending conferences that Java programmers go to, they have no commercial motive for the above remarks.
Of course, if your manager's reaction to the name 'Kent Beck' is to say, "Who?", then this won't achieve much.
the Gartner Group numbers are for initial Java projects, where the lack of experience is the major risk, and you will face (to a lesser extent - because Smalltalk is simpler than Java) a similar risk when doing your first Smalltalk Project
Quotes from people saying that Smalltalk is the ideal language for teaching and learning OO can be found easily and might address any management concerns. Be sure to get a couple of good books and a Smalltalk-aware coach to make the initial change.