Research on language acquisition is often genuinely cute.
I have some related anecdote on this. When my nephew was learning to talk (back then he was, like, 1~2yo? He's now 16), I recorded and transcribed some things that he said. Here's a few of them:
Orthographic | Adult pronunciation | His pronunciation | Gloss |
---|---|---|---|
chocolate | [ʃo.ko.'lä.te] | [ku.'wä.te] | chocolate |
vovó | [vo.'vɔ] | [bu.'bɔ] | grandma |
Amon | [ä'mõ] | [mu'mõ] | my cat's name |
dodói | [do.'dɔɪ̯] | [du.'dɔɪ̯] | boo-boo, hurtsie |
mexerica | [mi.ʃi.'ɾi.kɐ] | [mi.'ji.kä] | mandarin orange |
Look at the pattern - pre-stressed vowels get raised. The reason why my nephew was doing this in Portuguese is basically the same as why Orla (from the text) is using [χ] (the "guttural ck") in her English, because even as the child is learning to talk, they're already picking up features from the local variety. And that pattern where the vowels get closed before the stress is common place for Sulista Portuguese speakers (check how "mexerica" is pronounced, with [i] instead of [e]), just like Scouse English conditionally renders coda /k/ as [ç x χ].