Configurations of support include those that exhibit Support-From-Below (cup on table), as well as those involving Mechanical Support (e.g., stamp on envelope, coat on hook). Mature language users show a “division of labor” in the encoding of support, frequently using basic locative expressions (BE on in English) to encode Support-From-Below but lexical verbs (e.g., stick, hang) to encode cases of Mechanical Support. This suggests that Support-From-Below configurations may best represent the core for the category of support, and could be privileged in supporting early mappings to spatial language. We tested this hypothesis by examining spontaneous productions of children younger than 4 years found in the CHILDES corpora. Children used on to encode Support-From-Below more than other types of support configurations. They also showed clear distinctions in how they mapped different verbs (e.g., BE vs. lexical verbs) to Support-From-Below configurations compared to other support configurations. Analysis of parent language suggests that these observed patterns in children’s language cannot be fully explained by input, although a role for input is likely for children’s encoding of Mechanical Support. Thus, a concept of Support-From-Below may serve as a core representation of support, and hence the privileged spatial representation onto which spatial language for support is mapped.