If you wanted to make some money out of natural language processing, an appropriate strategy might be to identify an area of technology that was relatively mature—one where the more fundamental technical problems had been resolved through a significant amount of research activity—and then identify potential applications for that technology.