I think the StackOver flow data is highly skewed towards newer languages. Any basic questions about older languages would have been asked and answered long ago along with there being tons of other documentation and tutorials available. Older languages will simply have fewer new questions asked. So I think counting StackOverflow tags is a flawed measure. I think if they measured page views based on tags that would be much more accurate.

People at work tend to ask highly specific questions to solve very specific problems. While hobbyists can find all the information they need without having to ask questions on SO because there are so many tutorials and questions that have already been answered. So it could be that even though SO data shows an increase in work questions vs. newbie questions, it could just be because more and more answers to basic perl questions are easier to find and there's no need to ask them.