I read that the cloth that painters and chefs wear, the one now called "apron", used to be called "napron". But then because of rebracketing, "a napron" became "an apron".
But my question is, why was this new word form ? Why was "an apron" more popular than "a napron", but for example, "a napkin" didn't become "an apkin" or "an alley" didn't become "a nalley"?
Thank you
The commenters are right that in linguistics it's often hard to say why exactly, but they're wrong to pretend there are no factors that influence the probability of rebracketing and reanalysis in general.
(There's a huge space between not 100% predictable and 0% predictable.)
For example: the frequency of the word, the frequency of the word in spoken language vs written language, the frequency of the word in certain phrases, the frequency of the subwords, the frequency of the subwords in the reanalysis, the frequency of related words...
This tends to be because of loans and roots that are opaque, either because they are forgotten, like the ick in ickname, or never understood, like the nap in naperon or the al in alcohol.
We can contrast this with the tens of thousands of other words that start with n- in English. For example, verbs are unlikely to occur mostly only after the indefinite article a/an. A noun like nonsense is unlikely to be rebracketed to onsense if speakers make the association between the initial subword and no, not, nought, none, non- etc.
There is chance that napkin could have been rebracketed, but napkin was formed from nape and -kin in English i.e. by English speakers at a time when English speakers understood what a nape is.
If I had to name one theory: language model. This is typical for overfitting to the model, or when there's no data for the input, or very skewed data for the input. For example, if you build an automatic speech recognition system, this problem can happen, and it's definitely non-random.
