Obviously, there is a difference between modern humans and all other animals. We have the ability to easily learn how to speak in ways that communicate complex and subtle observations to others of our species who have learned our common language. The obvious proof of this is the human ability to warn our friends of dangers which are invisible to them. That is the key forcing factor which has the ability to become heritable, because if we can verbally warn our friends of life-threatening dangers we can later interbreed with them and their progeny. If we and they lack that ability to warn of dangers, they and we are obviously at greater risk of being taken out of the gene pool, and that ability to communicate will not be included in our offspring’s genetic code.
The question then becomes what are the necessary precursors to learning this ability and encoding it into our DNA? Our ancestors of say one million years ago had large brains, and the Neanderthals of half a million years ago even very large brains, and it would seem reasonable to assume that they had the capacity to learn a large number of words. But their expressions would in the beginning be single referent words, without a syntax for linking them together into what we would now call simple sentences.
In the earlier eras, there would be simple verbal warnings, saying “look out” there is danger, and these kinds of words are common even to birds. Some primates are known to have warning words for a dozen or so unique dangers. Single syllable warnings such as cat, snake, eagle, wolf, would be essential to not being killed and eaten by those animals. As pre-humans evolved and were searching for edible foods, such warnings would be applied to potentially edible plant foods. And these too would have a value judgment applied to them. Single syllable terms for food items such as fruit or root for edibles and yetch and puke for inedible things.
There is a positive-friendly quality to some of these things and a negative-danger aspect to others, but these qualities would still be indicated by single words, and possibly single syllable words. But, about this time in the development of a vocabulary, there comes the ability to apply these positive-negative concepts to other people and to potential mates. Mate selection has been going on from near the beginning of sex, but with the ability to identify the “good” and “bad” mates with words communicated among early people there is a new layer added to the selection process, and that brings into the mate choice the potential for this verbal ability to appear and to be selected for. Natural selection has developed into sexual selection of living ideal healthy survivors, and that with the advent of words is refined into verbal selection for qualities beyond simple adaptation to the existing environment. The ability to use words better than other potential mates becomes a selective factor that can be carried into the group gene pool.
When the ability to speak, even to speak single words, becomes a selective factor, then the ability to link two words together becomes possible for the selection process to advance. With that ability, it becomes possible to say snake-tree, cat-ground, good-guy, and the selection process can take off and select for all of the qualities that distinguish modern humans from all other creatures.
Humans became human the moment they could put two words together.