> the sound change doesn't have to be otherwise attested. Somewhat counterintuitively, historical evidence shows that most sound changes are regular! Which mea

> the sound change doesn't have to be otherwise attested. Somewhat counterintuitively, historical evidence shows that most sound changes are regular! Which means they affect all words satisfying the same condition in the same way [0]. Irregular sound changes still occur, but if you look at the history of a language, regular sound change suffices to explain almost all developments. Focussing on the history of English (which is very well documented, e.g. [1]) shows this well. Consonantal changes tend to be very regular: for instance, final /ŋɡ/→/ŋ/ in most dialects. English has an unusually complex vowel system, so vocalic changes tend to be somewhat more chaotic, but still show regularity: e.g. /a/ regularly turned into either /æ/ or /ɑ/. (See also @canjobear’s comment, which makes the same point in a nice way: https://news.ycombinator.com/item?id=36758576 https://news.ycombinator.com/item?id=36758576) [0] This is the ‘Neogrammarian hypothesis’: https://en.wikipedia.org/wiki/Neogrammarian https://en.wikipedia.org/wiki/Neogrammarian [1] https://en.wikipedia.org/wiki/Phonological_history_of_English https://en.wikipedia.org/wiki/Phonological_history_of_Englis...