This paper motivates the typological framework dubbed “dynamic functional typology” in terms of its utility in the description and analysis of classifiers and gender marking, which, we contend, have been mishandled by leading researchers in the field. The advantage of functional typology over the form-based, holistic typology presupposed by the WALS survey of numeral-classifier languages is demonstrated by its ability to bring to light a large-scale methodological limitation inherent in the past treatments of classifier/gender marking. In particular, it exposes numerous classifier/gender-marked constructions that do not accommodate a head noun functioning as a gender-agreement controller (Corbett 1991), or for a classifier to individuate the nominal referent for the counting purpose (Greenberg 1974) or to categorize the referent of the head noun (Allen 1977, Aikhenvald 2019). In place of the traditional analysis of classifier/gender-marked forms taking their modification use (e.g., Mandarin [yī zhǐ] ɡǒu ‘[one CLF] dog’, Spanish [una] casa ‘[one.F] house’) as a starting point of analysis, we take a radically different nominalization approach that analyzes such classifier/gender-marked forms as [yī zhǐ] and [una] in their own right, as nominalized constructions denoting respectively “one animal-class thing” and “one feminine-class thing”, independently from the head noun, which may not exist. A need to dynamicize functional typology arises in accounting for both crosslinguistic variation and language-internal crossconstructional/ functional variation of classifier/gender-marking. We offer a two-dimensional structural/functional hierarchy that provides a fine-grained comparative framework that both constrains the synchronic distribution patterns and predicts historical developments of classifier/gender marking across constructions, functions, and languages.