peer reviewed ; Perhaps the most puzzling of the Dutch vestigial genitives is the partitive genitive with indefinite pronouns and numerals, as in (1) and (3). What is most noteworthy is its stubborn resilience in the face of deflexion, even though other partitive genitives have long given way to alternative s-less constructions, such as close appositions, and even though such an alternative is occasionally attested for the partitive genitive at issue as well, as shown in (2) and (4) (Van der Horst 2008: 1624-1625). This has led numerous researchers to assume a new, grammatical function for this –s suffix, yielding a vivid – if mostly theoretical – discussion (Abney 1987, Broekhuis Strang 1996, Kester 1996, Van Marle 1996, Hoeksema 1998, Booij 2010: 223-228, Broekhuis 2013). On the basis of a more systematic analysis of corpus data, we call into question the purely grammatical nature of this suffix. (1) iets leuk-s something fun-GEN ‘something fun’ (2) iets leuk something fun ‘something fun’ (3) veel interessant-s much interesting-GEN ‘a lot of interesting things’ (4) veel interessant much interesting ‘a lot of interesting things’ We try to uncover the function of the –s suffix by looking for contexts in which the presence or absence of the genitive –s can be predicted. More concretely, we use logistic regression to unearth the variables which determine the alternation between the variants with and without –s suffix. We assess both (socio/region)lectal variables and grammatical variables, such as regional variety, register, the type of quantifier and adjective, the length of the adjective, etc., as well as possible interaction effects. We also include a random factor Phrase, to take into account the different phrases (measured in types) in our dataset. The results of our analysis strongly suggest that the function of the -s suffix is multifactorial in nature, and cannot reside solely in the grammar. Our fine-grained analysis allows us to give more precise underpinnings for the assumed regional split in the ...