Just like other languages including English, German has a variety of possessive attribute constructions. All of are synonymous NPs, but belong to vastly different varieties of the language.

a, with a postnominal possessive attribute in the genitive, represents the current standard variety. #b, with a prenominal genitive attribute, instantiates a construction schema which is still current with a set of nouns, but which a colloquial variety would have used with the given noun phrase several hundred years ago. The variant #c, finally, has the possessive attribute in the dative and in prenominal position, followed by the possessive pronoun fitting it. This has been in the colloquial language and in the dialects for several hundred years, but has never made its way into standard German. It is this construction whose origin is at stake here. Its development is represented by the sequence of rows in .

At the origin, there is a construction with a malefactive adjunct in the dative (traditionally called dativus incommodi). In a, this occupies the first position in the clause. The malefactive adjunct implicates a possessive relation between the referent of its NP, the possessor – here, the boy – and another participant which undergoes the situation, the possessum – the trousers in . In such a construction, the determiner on the possessum may be either an article or the possessive pronoun. The further course of things presupposes the latter option.

The malefactive adjunct precedes the verb in a, so it is not contiguous with the subject NP, which proves that it is no subconstituent of the latter. If anything else occupies the clause-initial position, the adjunct follows the verb immediately and thus directly precedes the subject NP (#b). Now on the one hand, the construction implicates a possessive relation between this NP and the malefactive adjunct. On the other hand, the language already has a prenominal possessive attribute (b), which now serves as an analogical model (c). On the basis of this model, the construction of b is reanalyzed so that the combination of the possessor NP and the possessum NP forms one syntagma (#d). As usual, this reanalysis is not apparent in the expression of d. It becomes indubitable in variants like e, which illustrates a well-established test for constituency in German syntax.

The sequence of c must be one constituent in e because in an independent declarative clause, the finite verb obligatorily occupies the second syntactic position (more on this in a later section).

The result of this reanalysis is the prenominal possessive attribute in the dative, with an intermediate possessive pronoun, as in c. Apart from the dative case, which is cross-linguistically not so frequent in this construction, the possessive NP in which the possessive pronoun on the possessum functions as an index to the possessor is well-established as a type.

Like other cases of reanalysis, this one shows that a reanalysis is based on an analogy with an existent construction. If no such model were available, there would be no motivation for reanalyzing a construction. A final point to note is that this is an instance of the variety of reanalysis which involves no grammaticalization.