Language presents a specifically problematic but nonetheless central issue for accounts of human evolution. It does not fossilize and there is no agreement on what a precursor would look like. One main issue is the difficulty of studying how symbolic language affects cognition, given that all healthy humans develop language. However, one avenue is ape language research. Cognitive experiments with symbolically competent bonobos, compared with bonobos without competency in human symbolic communication, show that symbols do affect the primate brain. These results provide empirical support for social niche construction theory as the best explanatory framework for language evolution. The research further supports the argument that symbols, rather than grammar, is what theories of language evolution should seek to explain first. The ancient philosophical question of language and thought is also discussed as the research with symbolically competent bonobos provides data that can speaks to what their relationship might be.