Matthews [Matthews, P. H., 1981, “Do languages obey general laws?”, Cambridge, Cambridge University Press] has denied that "languages obey general laws". Natural Morphology takes the opposite position. [...] a sceptical reader should compare the present stage of Natural Morphology with the dismal previous status of morphological naturalness. Instead of vague intuitions on naturalness/expectedness/awkwardness of phenomena that any linguist has who knows more than one language, we tryto define explicitly separate parameters and degree of naturalness; instead of relying on inductive generalizations we derive basic concepts from a tight semiotic system and limit induction to empirical testing, where it is legitimate and necessary; instead of assigning markedness values "ad hoc" or as underived primitives of a markedness model, we try to find extralinguistic explanations; instead of closing our eyes to non-standard data "external evidence", we try to investigate how far our approach works in all conceivable types of language performance;instead of using an undifferentiated notion of naturalness for all levels of abstraction, we try to distinguish (and to relate systematically to each other!) universal markedness, typologica adequacy, and language-specific system adequacy.
- Dressler (2004), Pag. 193