One of the foundational assumptions of Historical Linguistics is Uniformitarianism. We assume that, after the initial period of the evolution of language, Language is going to look the same as a structure, no matter if it’s 5000 years ago or 5000 years from now—because language is determined as a human faculty, and humans have not essentially changed biologically. So long as human brains are the same, and the human vocal tract is the same, language will be more of the same.
In fact, even if you take the human vocal tract out of the equation, language is still pretty much the same. One of the more gratifying conclusions from the study of sign languages is that Cherology is not essentially different to phonology. Which is in fact why we no longer use the term cherology.
Now, if the robots take over, the singularity hits and we get plugged into the Matrix, we all drown, or we all nuke each other, all bets are off. But so long as humans remain identifiably human, and live in recognisably human societies, there’s no reason to think that future languages will look substantially different to our current languages, any more than preliterate languages look substantially different to literary languages.
There’ll be bits that are different, sure. Different fads in discourse organisation. Different semantic fields in vocabulary. Different metaphors. Different extents of linguistic diversity, maybe (though dialects are certainly far from dead; they’re just organised along different distributions now). But there will still be anaphors, and word orders, and sandhi, and coarticulation, and synonyms, and presuppositions, and inflections, and tenses. The core of language as a system will remain recognisable.