Consider a simple example, provided by Langacker (2000). It is quite obvious, he argues, that the suddenly needed term for a ‘computer device’ mouse was derived from the word's literal ‘rodent’ meaning. It's equally uncontroversial that when first used for the new instantiation, it was through pragmatic inferencing that the new interpretation was derived. Of course, nowadays, mouse is already semantically ambiguous (polysemous) between the two meanings. The computer-device meaning is part of our conventional, grammatical knowledge of the lexeme mouse. Note that it has a different plural form (mouses), and it can serve as the basis for yet a new derived interpretation, as in (1), where clitoris mouse refers to the small, centrally located pointing stick used as a mouse on some laptop computers:

But when/where/how did the change in cognitive status happen for mouse? It did not come about by stipulation. All we can assume is that many speakers started using this innovation because they found it useful, not because they were seeking to change their lexicon. But when did the change actually take place? On the first uses it may not have been easy to even understand (witness B's difficulty in (1)). On the seventh use event of the metaphoric mouse, asks Langacker? We cannot determine a precise moment when the inference leading from ‘rodent mouse’ to ‘computer mouse’ gave way to a conventional, semantic meaning for mouse (see Deutscher, 2005: 152).