It has been shown before that scalar inferences differ in their strength. As an example, imagine the contrast between the cardinal number "three" and the quantifier "some": the scalar inference from "three" to "not four" is in some cases perceived as stronger than that from "some" to "not all". Four diagnoses of strength that underpin this contrast are: 1) the stronger inference of cardinality is harder to eliminate, 2) easier to embed, 3) is learned by children earlier, and 4) is captured faster and easier in online language processing.
We were able to determine that numbers are not the only case of strong scalar implicatures and that further gradations also exist in the area of weaker scalar implicatures. Especially the difference between the simple "or" and the exclusive "either-or" is an example of the latter: We showed that the difference can be demonstrated with adults, but that children do not distinguish between the two possibilities. This does not even happen in languages like Japanese; here the same lexical means is used for both versions (ka`or' vs.ka-ka).