Maybe interfacing to javascript libraries has replaced actual new javascript development and people search for, for instance, "jquery" instead of javascript.
That seems to be the main problems with these metrics. They aren't looking at the popular libraries that make use of the language. If I search "Android OpenGL development" it should count as a query directed at Java. This survey doesn't seem to do that though.
That could very well be the case. Perhaps its position was artificially high as well. The transition from static to dynamic web pages forced a glut of developers to learn it, and hence a lot of searches for answers. Maybe this is just a sign than we're all getting more seasoned in the language on average.