This paper builds on a previous proposal that the gradual emergence of syntax in language evolution was engaged in a feedback loop with the effects of Human Self-Domestication (HSD), both processes contributing to enhanced connectivity in the cortico-striatal networks, which is the mechanism for suppressing reactive physical aggression, the hallmark of HSD, but also the mechanism of cross-modality, relevant for metaphoricity and syntax more generally. In this paper, we aim to bridge the gap between these neurobiological changes and the changes facilitated, or even promoted, by the gradual complexification of grammars. We propose that increased cross-modality associated with the brain modifications described above would have enabled and supported, more specifically, a feedback loop between categorization abilities relevant for vocabulary building and the gradual emergence of syntactic structure, including the core combinatorial operation in natural languages, such as Merge. In brief, an enhanced categorization ability not only brings about more distinct categories, but also a critical number of tokens in each category necessary for Merge to take off in a systematic and productive fashion; in turn, the benefits of expressive capabilities brought about by productive Merge encourage more items to be categorized, and more categories to be formed, thus further potentiating categorization abilities, and with it, syntax again. The need to amass a critical number of words before breaking into syntax is clearly demonstrated in language acquisition studies. It is also clear that without breaking into syntax there is no advancement in language or categorization abilities. Animal communication abilities, including those of trained animals, seem to stop short of this critical number, which, in our analysis, provides a rationale for why they do not develop productive syntax, or human-like categorization abilities. In addition to evidence from linguistics, language acquisition, and comparative studies, we also build on evidence from biology, neuroscience, paleoanthropology, and clinical linguistics.