We examine the diachronics of a New Passive construction in Icelandic and use Yang's model of language learning and change to explain its rapid rise. The New Passive has been spreading at the expense of a Canonical Passive in the recent past 50 years. Applying empirical measurements from the IcePaHC corpus, we show that our model can be used to account for the spread of the New Passive and the rate of change. The model also has implications for the actuation of the change.