We have been helping running computer clubs for about 20 years and have documented the things that work and don't. Creativity and ownership are key. Simply hacking Java code will go nowhere. Have them build games where they can also build their own 2D/3D artwork. Use tools like AgentSheets and AgentCubes that include powerful 2D/23D authoring and end-user debugging tools to motivate them and help with the programming. Otherwise, as you already see, you will quickly loose your audience.
Here is some research data: http://scalablegamedesign.cs.colorado.edu/gamewiki/images/4/44/Will_It_Stick-submit_CR.pdf
AgentCubes in action: http://www.youtube.com/watch?feature=player_embedded&v=2GWcb3aG2w0
Better Idea: Forget computer clubs! If your goal is to expose students to computer science then the computer club idea goes nowhere. This is not an opinion. We have the data. You will get few girls, rarely any minority students and the overall percentage of students participating is dismal. Try the Scalable Game Design curriculum http://scalablegamedesign.cs.colorado.edu/ You can have your teachers do this and expose nearly 100% of the students at just about any middle schools or high school. With this strategy we get ~300 students per school and year instead of the ~15 computer club ones.
One can argue about the value of localized languages. For instance the localization of Pascal as educational programming language into German and French turned out to be a big flop. Of course GUI components of some IDE probably need to be localized. AgentSheets is localizable and has been localized in a number of languages but not Dutch. Even more important is the localization of tutorials. You can find some here: http://scalablegamedesign.cs.colorado.edu/wiki/Frogger_Design
There is a mix of native localization (good) and Google translated ones (not so good).
The other point is that perhaps your daughter just is not excited about the making simple 2D animations and would like to make complete 3D games including 3D characters that she can create? You may want to give AgentCubes a chance: http://www.youtube.com/watch?v=2GWcb3aG2w0&feature=player_embedded
This runs as desktop application or in browsers via HTML5 (no Flash no Java) site even on cheap $200 Chromebooks.
AgentCubes/Inflatable Icons allows you do create 3D shapes very quickly with no 3D modeling background. Paint images in 2D and turn into 3D. You indicated that you are struggling with 3D tools such as Blender and even Sketchup. I guess I don't know what kinds of shapes and what kind of quality of 3D shapes you have in mind. We have been exploring for some time why many people have problems using these kinds of tools. The short answer is that these tools are aimed at typically professional 3D designers or, more generally, at people with a lot of time at their hands to learn an interface with a steep learning curve. If your goal is to produce 3D shapes of the Pixar level quality then there just is not way around these kinds of tools. If, on the other hand, you just need to build very simple shapes that you can produce in, say, a couple of minutes, and maybe print that on a 3D printer then perhaps Inflatable Icons may do the trick.
A benchmark with Inflatable Icons was that if it takes more than a minute to explain how to make a 3D shape it is too complex. The idea is to make casual 3D tools. We have tested this with many kids and it works great. The short version of the concept is that practically all 3D tools work on the "First Shape then Paint" while we have flipped this around to be "First Paint then Shape" You are basically drawing a 2D image first using a Photoshop-like editor. Then you use tools including inflation to turn then 2D image into a 3D shape.
video: http://www.youtube.com/watch?v=2GWcb3aG2w0&feature=player_embedded
sample Inflatable Icons (you can even edit them in the browser): http://scalablegamedesign.cs.colorado.edu/arcade/
What this country needs is a good five dollar plasma weapon.