In Dutch, Frisian and German dialects, bewildering morphological variation is found in the person-marking system (e.g., the Syntactic Atlas of Dutch Dialects for Dutch and Frisian, and Goossens 1994, Weiss 2005 and Fuss 2005 for German). One type of variation concerns the formation of subject clitics in clauses with inverted word order, i.e. with the subject following the verb. These elements are subsequently extended to other syntactic positions through analogy, via a pathway described by Kathol (2001): originally enclitic elements spread first to the position enclitic to the complementiser, then to verbs in sentences with regular word order, and finally, in a rare number of cases, to verbs in the sentence-final position in subclauses. This paper shows that Kathol’s proposed pathway actually comprises two similar but not identical pathways, i.e. one for specifically enclitic elements, and one for elements occurring both pro- and enclitically. This implies that developments must be modelled as an ‘analogical map’ rather than a pathway, which will be illustrated extensively with data from Dutch, Frisian and German dialects. The map will be shown to apply to more data than Kathol’s original pathway: it not only captures the extension of clitics, but also the spread of many innovative agreement markers and pronouns, and, hence, a range of person marking phenomena, including complementiser agreement (e.g., Carstens 2003, Weiss 2005), clitic doubling (e.g., Fuss 2005) and double agreement (e.g., Zwart 1997). Finally, following Hill (2007), some of the motivations underlying the change and the factors determining their direction are discussed.