Some functions (such as modified correspondence analysis, GIW, profile-based etc.) are specific for input data that contains variable (i.e. concept two nested linguistic factors) information. Some functions only work if the input is variant-specific, ie. only with one linguistic factor), such as a simple euclidean distance. Other functions may be generic, in that they would work with either kind of input. I have no example for this last type.