Abstract words (e.g. freedom, truth) differ from concrete ones (e.g. table) because linguistic experience is more crucial for their acquisition and representation. In previous work we proposed that the linguistic input is more pivotal to learn abstract than concrete concepts, because the members of the first are less perceptually similar and more heterogeneous. During abstract concepts processing we would therefore activate language, either because we re-enact previous acquisition modality or because we use inner speech to master their complexity. Here I propose that, because abstract words evoke previous linguistic experience to a larger extent than concrete ones, they are more affected by linguistic relativity and by the differences between spoken languages. To substantiate this proposal, recent studies with words of different abstractness level are taken into account and reviewed, showing that the weaker the environmental constraints and the more abstract the words are, the more their meaning differs across languages.