I'm asking this because there seem to be a few words in my dialect where there have been sound changes that appear random. Such as how "catch" is /kEtS/ while "match" is /m{tS/. Also how "get" is /gIt/ and "bet" is /bEt/.
Can sound changes happen at random?
In many instances, sound changes tend to target commonly used words. For example, I think "catch" and "get" would be much more commonly used than "match" and "bet" in the speech of any native speaker. It's for this reason that for many English people the words "pass" and "mass" don't rhyme, and that for many Americans the words "dog" and "cog" don't rhyme.
I also have [kEtS] for 'catch' a lot of the time. I think for some words these could be older variants that have been maintained in some dialects. I also have [mEok] for 'milk' but [Iok] in 'silk', 'ilk', etc. for no apparent reason. I suspect this is maintained from an Old World dialect, perhaps Scottish or Irish, that isn't reflected in the spelling.
I have ["mI5k] and ["gEt] for "milk" and "get", but I do have ["k_hEtS] for "catch".
I similarly have ["mIMk] and ["gE?] for "milk" and "get", but I have alternation between ["k_hE{tS] and ["k_hEtS], the former showing up in more careful/formal speech, and the latter showing up in less careful/formal speech. This is similar to the alternation between ["k_hE{~:n] and ["k_hE~:n] for "can" here, where the former is the careful/formal realization but the latter realization is by far the more common one in everyday speech. Also, I will sporadically use the pronunciation ["mEMk] for "milk", which is very common here even though I only use it occasionally.
One thing that complicates this, though, is that in unstressed positions the phonemes /E{/ and /E/ can often be only weakly distinguished here, as /E{/ may be just realized as [E] or a very narrow [E{_r] while /E/ may be realized as [E_"], such that there is very little actual audible distinction between the two. This is exacerbated by the phoneme /E/ being lowered in the idiolects of many here, even though my own idiolect preserves the more conservative unlowered realization.
One thing that complicates this, though, is that in unstressed positions the phonemes /E{/ and /E/ can often be only weakly distinguished here, as /E{/ may be just realized as [E] or a very narrow [E{_r] while /E/ may be realized as [E_"], such that there is very little actual audible distinction between the two. This is exacerbated by the phoneme /E/ being lowered in the idiolects of many here, even though my own idiolect preserves the more conservative unlowered realization.