The paper aims to explore the semantic and syntactical environment of the constructions habban+PP and hæfdon+PP based on the data collected from Old English and Middle English corpora of prose texts. The main emphasis is made on mechanisms which contributed to the transition of the semantics of hæfdon+PP from resultative to anterior. The starting point is a statement, which has been demonstrated on an ample material in a series of studies, that the construction in question possessed a state-resultant semantics and was used to express a state of the subject which simultaneously played the role of agent. Applying the scheme, proposed by B. Heine (Heine, B. 2002), of how a linguistic expression acquires a new meaning, first, bridging contexts were identified where interpretation of both state-resultant and anterior semantics was available, and secondly, switch contexts were found where the target (new) meaning was the only one possible. The latter contexts are represented by temporal clauses in Old English and counterfactual ones in Middle English. The analysis shows that the outlined scenario can be applied to the development of the past perfect, which marks pure taxis, rather than to the present perfect.