As a language researcher, one of my frustrations is that we don't have a lot of data. That's not quite true -- a lot of good research is done by using computers to parse text on the Internet. The problem is that's not great for studying what language *means*, because all you have is a bunch of words, not their meanings. But studying meaning is slow: you need to sit down and look at a lot of sentences and decide what they mean.
To put it another way: One of my main areas of research is pronouns. The meaning of a pronoun depends on context ("she" refers to somebody different depending on the sentence it is in). But much of the work in linguistics and psychology has focused on really just around one thousand sentence contexts out of the infinite number of possible contexts.
One of the ways I've been getting around this is by running experiments on the Internet. We get thousands of participants and so are able to test large numbers of sentences, contexts, words, etc. -- things we couldn't do in a traditional laboratory because there just aren't enough participants to go around. If you are interested in helping out, please visit http://gameswithwords.org. We design these experiments to be fun and short (most run around 5 minutes, except for the somewhat longer "Mind Reading Quotient", which is a set of tasks involving "reading between the lines" and figuring out what somebody is really trying to say). For some of the experiments, like "Mind Reading Quotient" and "Ignore That!", you can find out (some of) your results immediately at the end of the experiment. When the results of these experiments are published, they appear on the associated blog (http://gameswithwords.fieldofscience.com) and are sent out via a mailing list you can subscribe to at the site.
I think that some people at this forum will be interested. To the rest of you, I apologize for the blatant self-promotion.