A good way to search for answers relative to the language-mind riddle might be departing from language, and posing the following hypothesis: if the human language faculty is constrained by ontological knowledge (some kind of ‘pre-knowledge’) then it is quite likely that this same ontology (or parts thereof) will be constraining other subsystems of human cognition. Under this view, we then ask: 1) which subsystems of human cognition are easily comparable with language (one valid answer is spatial cognition); 2) what is universally shared between–in this case–the way in which we elaborate space non-verbally and the way in which we talk about it; and 3) is there a way to reconcile language specific categorization of space and (universal) conceptual knowledge?

In this paper, by proposing an answer to questions 1) and 2) we also try to contribute to a positive answer to the third question. Departing from a crosslinguistic analysis of the ‘on’–‘in’range of prepositional spatial usages (Bowerman and Pederson 1992, 2003), we a) try to describe and explain the central aspects of crosslinguistic variation in the ‘on’–‘in’semantic (prepositional) usage; b) propose the elements and principles identified as being at the core of crosslinguistic prepositional variation and also the main causes bringing about language specificity; and c) explore what seem to be some (cognitively based) universal constraints on language specificity. The result of the study is that it is possible to logically systematize this ‘on’–‘in’range, in terms of different (but systematic, non arbitrary!) combinatorial patterns of features from only three domains: DIMENSIONALITY (points, axes, volumes), ATTACHMENT (contact; presence/absence/quantity of) and ORIENTATION (90/180 angle, directionality).