it is convention and logic: anything near an absolute minimum or maximum can be ceiling or floor

I would add to Ulf-Dietrich's comment that in some data it is clear to see that you have a floor and/or ceiling effect. For example, assume some survey in which the theoretical max is 100 points, but you find that the data never show a true score close to that (let's say that the maximum true score is 90 pt.). Thus, you can argue that this survey has a ceiling effect because the true maximum scores never appear to coincide with the theoretical max of 100 points.