What are they? For major part, they are interesting puzzles.
Let's take the second part for example. You have some sentences in Ancient Greek, and you have translations in English. Your goal is to find which translation is for what sentence.
How would you go on about that? In this case, for example, by taking a look at the sentences. You could count how many times a word appears, and then compare that. And so on.
Yeah. Someone else in the thread said they were more like computational puzzles, which I kind of agree with - they're a lot of just logic - but some background knowledge of etymology is also helpful. For the first puzzle, if you didn't know Brazil might be translated as something like "brazilia", then you would have struggled.
Yes! They can be solved as a puzzle, but should you happen to know some linguistic concepts or be familiar with, out of lack of better words, lesser-known languages like Navajo, Armenian or so on, it helps too.
It's not hard, but some passing knowledge of the languages definitely helps (like the Aragonese table is almost trivial for people who know a Romance language), but especially how linguistic transformations happen
Some surface-level knowledge about Japanese writing systems (ie. the fact that Katakana/Hiragana are fairly regular consonant+vowel syllabries) helps a lot.
How would you go on about that? In this case, for example, by taking a look at the sentences. You could count how many times a word appears, and then compare that. And so on.