The person placing the text into the bubbles made the exact same mistake. The text was pulled in the correct order, but placed with panels 2 & 3 swapped.
1 2 3 4 5 6 7 8 <- Text pulled
1 2 5 3 4 6 7 8 <- Placed in bubble
If the person pulling the text got mixed up, but the person placing it was correct, then the three bubbles would have been swapped the other direction.
1 2 5 3 4 6 7 8 <- Text Pulled
1 2 3 4 5 6 7 8 <- Placed in bubble
Hilariously, if the puller and placer are
both wrong, then everything gets placed correctly.
1 2 5 3 4 6 7 8 <- Text Pulled
1 2 5 3 4 6 7 8 <- Placed in bubble
However... this can result in translation errors when one character is speaking across multiple bubbles. Since Japanese sentence structure is different from English, it's best to translate everything at once in the proper order. If you translate one bubble at a time you'll likely make some very weird mistakes.
Not a real example, but... if we cut the English sentence "My dog climbed a tree." into two bubbles, and treat them independent of each other....
(My Dog) - (climbed a tree.) <- What the bubbles say.
(Yo dawg!) - (I climbed a tree.) <- How you could interpret them.
