Twenty-something years ago, when I was just starting my professional career, I'd probably agree with Paul and say "Go for a universal naming conventions!". :)
Today my answer is different. Coming up with a decent naming conventions is a challenging task even when you are dealing with one programming language. Having universal conventions for multiple languages/frameworks is a noble goal and excellent brain teaser, but in general I don't think it's feasible. The syntactic and lexical rules of different languages/frameworks are way too diverse to make "one size fit all". Besides, the set of the employed languages/frameworks can change at any time, and appropriate adjustment of your "universal conventions" could turn out to be prohibitively expensive or prohibitively untimely.
So I would recommend to focus primarily on solid and decent conventions for each language/framework (or for each group of languages/frameworks that are sufficiently similar - like C and C++, for instance). One nice outcome of this approach is that the projects involving only one language would have to adhere to a set of conventions that does not look "foreign".
Having said all that, I believe that some conventions can be common for all the involved languages/frameworks without becoming too "foreign". Moreover, such thing as vocabulary should be common for entire project. This is mostly about semantic and lexical rules, of course; but even some syntactic rules can be universal, too. For example:
Lexical rules for entity names (e.g. "always use singular nouns").
Lexical rules for acronyms (e.g. "use only wide-known acronyms").
Syntactic rules for composite names (e.g. "always use general-then-specific sequence").
The examples above are oversimplified, of course. But I hope the idea is clear: whatever conventions can be common should be common, but it's always going to be just some subset of all the conventions that you need.
GetFoo()in C++, while it's common to seegetFoo()in actionscript.