Restricting language change through micro‐comparative analysis Jan Don i , Paula Fenger i , Olaf Koeneman j (University of Amsterdam i , Radboud University j ) 1. Introduction A. It is generally assumed that the study of micro‐variation is a synchronic way of studying diachronic change (Kayne 2000; Westergaard, this workshop) B. Although the variation pertaining to subject verb agreement in dialects of Dutch is quite bewildering (cf. SAND atlas, Barbiers et al 2005), four exceptionless generalizations on paradigm structure can be formulated. The observed generalizations pertain to: ‐ The distribution of affixes over the different ‘paradigm slots’; ‐ Inversion morphology (illustrated below). Inversion morphology (Standard Dutch); (1) Jij koop‐t een boek van Brouwers [so called ‘straight word‐order’] You buy‐2 nd a book from Brouwers (2) Van Brouwers koop jij een boek [so called ‘inversion’] From Brouwers buy you a book The observed generalizations are the following: (3) Generalization 1 The affixes associated with 3SG and 3PL contexts in the straight order are never dropped or replaced by another affix in the inversion order, in contrast to affixes associated with 1 st and 2 nd person contexts (singular or plural). (4) Generalization 2 Inversion morphology either uses an affix that already appears somewhere in the straight order (i.e., the inventory of affixes is not extended) or the inversion morpheme is new (i.e. extending the inventory of affixes) but in the latter case the inversion morpheme is invariably null. (5) Generalization 3 Although the affix associated with 3SG can never be dropped in the inversion order, it is dropped without exception in past tense contexts.