This paper investigates why learning a foreign language locative system is difficult. A developmental view has been taken on in examining the dynamic phenomenon of learning foreign locative terms. Four experiments were conducted. The first three experiments focused on Chinese speakers’ use and learning of English locative prepositions. A supplementary experiment studied English speakers’ use of Chinese locative expressions. In Experiment 1, 20 hand-drawn pictures containing natural scenes were presented to 58 Chinese-English bilingual college students, who depicted the same set of pictures in three ways (Chinese free production, English free production and English multiple choice). General English reading capabilities were also assessed. Experiment 1’s finding indicates that the Chinese speakers’ performance on use of English locative terms cannot be completely accounted for by their general English reading capabilities. In addition, although some of their errors may reflect the cross-cutting relationship between Chinese and English locative terms (e.g. some usages of “on” and “over” being translated into the same Chinese term, “zai.. shang”), native Chinese speakers’ learning patterns of English locative prepositions cannot be explained merely by first language transfer. Cognitive universals may play a fundamental role. Besides, our Chinese subjects’ errors show asymmetrical patterns, in which “on” were more likely to be used when “in” or “over” were needed than the other way around. Experiment 2 and 3 aim to determine the presumed causal relationship (from cognitive universals to the use of locative terms) by directing one’s attention to the spatial relationship between Figure and Ground that are key to selecting English prepositions. In Exp.2, two groups of native Chinese-speaking adults were trained on two pairs of contrasts (“in” versus “on”: N = 18; “over” vs. “on”: N = 17) with pseudo-named artificial novel objects. Two groups of children were trained also on the two pairs of contrasts in Exp.3 (“in” versus “on”: N = 18; “over” vs. “on”: N = 19). It was found that the limited training significantly enhanced our subjects’ performance on depicting natural scenes with English locative terms, except for the child “in” vs. “on” group. In addition, the asymmetrical error patterns observed in Exp.1 with natural scenes were replicated in these two experiments with scenes containing novel objects. The supplementary experiment tested native English speakers with the same set of pictures used in Exp. 1. Eight native English-speaking English-Chinese bilinguals volunteered for this study. Our English subjects’ written response scored fairly well in terms of locative term selection, which mirrors the asymmetrically crosscutting relationship between the tested Chinese and English locative terms. The difficulty of learning Chinese locative expressions, however, arises from sentence structure. Native English-speaking subjects sometimes missed a Chinese locative postposition when it was obligatory in the sentence. Overall results support the role of typicality in spatial relationship being core and initial spatial concept, which is not affected by secondary linguistic factors. First language transfer occurs only when there are semantic or structural ambiguities. Two possible types of prototypical static spatial relationship, simple contact and containment, have been proposed.