My wife and I are in the design process of a card game.
CodeQuartets (Codename) lets you play "Top Trumps / Ace Trumps" ("Supertrumpf" in german) and of course standard quartets.
If you´re not familar with these games, you can take a look here:

http://en.wikipedia.org/wiki/Ace_Trumps

However, CodeQuartets is not about cars, aircrafts or boats.
It´s about programming languages.

Currently every card has a set of data. To give you an example, here is the data set of C#:

class C# : C

class C# : C
{
private int yearOfBirth = 2001;
//
private int keywords = 76;
//
private float platforms = ;
//
private int helloWorldCount = 93;
//
private int rank = 7;
}

The rank is based on data from http://lang-index.sourceforge.net/

As it turns out, a lot of languages are theoretically available on every platform, so "float platforms" becomes infinite. This is bad for a balanced top trumps game, we need data which differs for nearly every language. On top of that the property must be suited for research.

So here comes my question:

Do you have any suggestions for some more properties?

Edit
To give you a small teaser, here is a very early, work-in-progress design of a card:

Maybe

private bool isIntepreted = 1;

or

private bool getsCompiled = 0;

... depending on which you prefer.

Doesn't differ enough, because, well it's a bool. Nobody will choose this value to battle, because the chance is high to end up in a tie. We need a property which differs a lot from language to language, such as the char count of the hello world programm.

Number of Classes / Functions (or "Size" to be more general) in the Standard-Libary. Most Languages come with such a thing. LiteC has built-in, specific functions, C has its stdlib, Java the Java-Class-Libary, C# the .NET-Class Liabry, and so on.
It differs on many Languages. Of course, the problem could be how to measure it (classes contains often more logic than single function, and what is part of the "official" bundled libary?). And to get to values could be also difficult...

But maybe an option for you ("popularity" can also just be estimated)

This was suggested by MasterQ32, too.
Comes, for example, "Plankalkül" with a standard lib? (rethorical question - read: it´s really hard to get these values)

But it´s not impossible either...

Needs virtual machine?

Doesn´t work with a top trumps game, because the answer is either yes or no.
We need a value which differs a lot from language to language.

This would be the same as if you write on a car top trumps card game "has sunroof?". It makes no sense, because the chance of a tie is very high.

I´m going to investigate the "standard lib" idea further.
But please keep your suggestions coming, they help me to look at the game from different directions.

The problem you are going to have, and already have, is that it's hard to put a language in numbers. Platforms for example is hardly a good measure, if any at all, because a language isn't tied to a platform. It might work better on some platform, but the compiler and runtime is certainly portable to other platforms. Take Quipper for example, a language that is based on Haskell, which is a language for quantum computers, which don't exist yet (despite whatever D-Wave is claiming), yet it also runs on x86 based computers, just exponential slower.
There are lots of booleans though, for example one might ask about the languages type system, GC support, multithreading capabilities, wether it's based on the lambda calculus and so on. But you are going to have a really hard time to put that into comparable numbers.

PS: Sorry if that sounds a bit harsh and doesn't really contribute to the topic, I was just writing my thoughts down.
Edit: You might want to get the current rank of the language from the TIOBE index though, which takes some more measures into account.

You summed up my thoughts pretty well.

Also thanks for pointing out Quipper. After reading about Plankalkül, my wife asked me if there are languages that are too heavy for today´s computers.
It´s pretty exciting that my wife actually bothers about programming languages...

Edit: I know the TIOBE index already and I´ll keep an eye on it. What I like about the tool on soundforge is that I can define my own set of languages.

I suppose there are a few things that you can put on there, though they aren't any good metrics (unlike for example horse power with cars which give you at least some kind of information):
- Size of the standard library (which has the downside of languages which provide a huge and extensive library themselves are getting bigger numbers, while languages merely providing a wrapper over other things get lower numbers, without providing less functionality)
- # external dependencies (that's highly platform depended and a bogus number anyways)
- Time to execute xyz (which requires you to write a test program which is fair for all languages. Good luck with that, Erlang for example is useless for most tasks, but shines when you try to write a Jabber server or a telephony system)
- Number of page in the language spec (which C++11 with over 1100 pages would win by far, though not every language is standardized and provides a spec, PHP or Perl for example)
- Number of lines to express the language Yacc grammar (not sure if you can find one for every language. C++ is going to be hard, because the standard deliberately doesn't provide one for a few things since they can't be expressed in any way)

Stuff like that... But yeah, you see the problem with that I guess.