Evolution is normally seen as a time sequence based on fossil records over millions of years. Unfortunately we cannot find historical evidence on brain capabilities such as for language since the brain does not petrify or fossilize. Even if it did, it might be hard to determine whether and how such a fossilized brain would support language.
Another approach is to look for DNA structures that clearly indicate language capabilities and functions. Unfortunately no such structures have been identified as yet.
A third approach is through an examination of presently existing species to piece together a possible time sequence. We follow this approach by analyzing vertebrates as information processors with legs. We conjecture about the information processing required to support observed behaviours. Unfortunately, we find it difficult to understand the organic 'hardware' that we find in today's species, and we are unable to obtain a blueprint of information processing architectures from the fossil and DNA information. In consequence, a bottom up approach is blocked. We therefore focus on a top down approach that examines high level requirements for information flow and transformation without worrying about the specifics of how the information might be processed and transmitted.