Crossposted from https://sopuli.xyz/post/25634723
I wonder how native English speakers do it, but here’s how I approach this problem.
My trick involves using a consistent spelling system for encoding a random letter sequence into a sound which I can memorize. When writing, you just pull those auditory memories, decode the sounds back to the original alphabet salad, and you’re done! Needlessly complicated, but that’s a common theme in English anyway, so it should fit right in.
To make this method work, you need a consistent spelling system, so you could make one up or modify one previously invented for another language. Basically anything more consistent than English should do, so it’s a pretty low bar to clear.
Here are some example words to test this idea with:
- carburetor
- carburettor
- carburetter
Pronounce those letter sequences using that alternate spelling system. It won’t sound like English, but it’s consistent and that’s all we care about at this stage. The end of each word could sound like this:
- [retor]
- [retːor]
- [reter]
In my system, each letter corresponds to a specific sound like e=[e], a=[ɑ] etc. I’ve been thinking of including the Italian c=[tʃ], but you could use other languages too. Feel free to mix and match, as long as you make it consistent.
The idea is that it’s easier to memorize sounds rather than whimsical letter sequences. Once you have those funny sounds in your head, it’s easy to use that same consistent spelling system to convert the sound back to letters.
Once you know that trick, it suddenly becomes a lot easier to spell common words like “island”, “salmon”, “subtle”, or “wednesday. For example “cache” could be stored as [tʃatʃe] in my head. Still haven’t settled on a good way to store the letter c, so I’m open to suggestions.
That’s pretty cool. Tell me more about your system. Is it fully consistent, or only partially?
It's only partial, for things that I could reasonably misspell while writing Portuguese:
I picked those specific readings because they're easy enough to pronounce but they wouldn't otherwise pop up in my dialect of Portuguese. A good chunk of them are actually archaisms, like ⟨c z⟩ used to actually sound like [ts dz] some 700 years ago or so.
Just from that table alone, I can tell that there are many “collisions” (to borrow a cryptography term) where single letter can map to multiple sounds or vice versa. As long as the rules are clear, you don’t necessarily even need spelling bees.
Yup, the standard spelling has a fair bit of collisions. With my system focusing on the multiple spelling → same sound ones, and leaving the rest untouched. It also doesn't solve positional rules (like ⟨ss⟩ vs. single ⟨s⟩ for /s/), as those are predictable.
Predictability is key. As long as thouse double letters don’t misbehave, this system should work just fine.