It's technically impossible to get any useful data on overall programming language use: for that to work, you'd have to monitor people's computer activity. At best, you can measure a few indirect numbers, all of which are going to be biased in some way:
- Compiler / tool downloads. A popular language will have a higher download count than an impopular one. The bias, here lies in the fact that number of downloads is not proportional to the amount of usage - a highly hyped compiler that doesn't live up to its reputation may have a huge number of downloads, but little actual use, while a mature, reliable compiler with a large stable user base may have less downloads, but much heavier use.
- Existing software. If more software is written in a certain language, it is likely to be more popular overall. So one could analyze some open source OS and count lines-of-code in various programming languages, maybe even correcting it by package popularity. The biases here are many, though: first of all, the software that makes it into such a repository is typically written by experienced programmers, not entry-level students, so language count is representative for the former category, not the latter. Also, much of that code has a long history, and projects tend to stick with a language once they reach critical mass. Older languages (e.g. C, Perl, shell script) are therefor probably over-represented, compared to newer languages (e.g. Python, C#, Haskell).
- Forum activity. A more popular language is likely to have a more active community, so higher forum activity is an indicator for higher popularity. The problem with this approach is that it is hard to quantify forum activity in a meaningful way, and even harder to decide which forums to include into the survey. Some language communities are centered around a small handful of sites and communication channels, while others are more distributed. Internet forums and mailing lists are archived and duplicated all over the net, people cross-post all the time, and forums get spammed and flooded with irrelevant posts and off-topic discussions.
If you really want to know what 'kids' are using, you are going to have to ask them: find a representative (this is both crucial and hard) group of kids willing to participate, and ask them a few well thought-out questions (or have them fill out a questionnaire). You will still have some bias there, but at least you'll be measuring the thing you want to measure.
Furthermore, popularity is a poor indicator of quality or suitability in this particular situation. As a teacher, low boilerplate and intuitive syntax may be your highest priorities, while 'the kids' are more likely to go by availability, 'cool factor', plain old 'it's what I know', and impressive examples of what you can do with it.
If you want to find a suitable language for beginning programmers, make a list of properties it should have (I'm sure as a programmer you can come up with a lot of these), prioritize them, and find the language that best fits your list.