The way i understood the GP is that Japanese has the benefit of having its own writing system so you won't get so easily lured into mapping sounds to back your own (assumedly latin script based) language. Use that as an opportunity and avoid the Latin alphabet altogether.
When learning French you don't have that opportunity so you're going to keep having to remind yourself that not a single character in a French word sounds like what it does on your native language.
I've seen this in action when learning Finnish, a long time ago, in a class full of expats. Most Germans had serious trouble with the ä, which sounds entirely different in Finnish than in German. They would consistently pronounce it like in German, which made them sound ridiculous. The rest of the class, whose native language alphabet didn't contain the ä, got their pronunciation much closer to the real thing. I can only imagine how powerful this is if not a single character, but the entire script, is different from your native language.
I'm not sure about Finnish but learning German and Czech as an English native the "uncanny valley" of familiar consonants (and familiar vowels with funky headgear) was only a stumbling block for the first few weeks or so in my experience.
Every now and then I'll mess up when reading out loud a "c" ("ts" in Czech) if I'm not paying attention, or writing "ch" instead of "č" - but that's pretty rare.
It's not the letters that mess me up, but the "false friends" - words that look similar to some in English and that can be a pretty big deal. I'm gonna write something in weird pseudo-code, because I can't express it cleanly in English.
German.ja == English.yes
Czech.já == English.me
So if you learned both Czech/German as a second language you can have moments of "wait, wtf!?" here and there. Also:
English.no == Czech.ne
English.yes == Czech.ano (shortened to "no")
So a Czech person agreeing to go for a beer with you might say "no" meaning "yes". There's some other good ones in Czech actually - the word "poluce" looks like "pollution" but actually means "emission of semen in sleep".
If it sounds like I'm complaining then I'm not - I love all this weirdness :)
* = actually reflection and cases make this more complicated still - could also be mého/mému/mém/mým/své/svého/svému/svém/svým depending on context
"Recorded since the 1340s, as "discharge of semen other than during sex", later, "desecration, defilement" (1382), from Late Latin pollutio (“defilement”)"
Other half is Hungarian, she often uses "Not" in place of "No", "Did you make it to the school on time?" "Not".
Turns out that in Hungarian No/Not == Nem so when she translates it in her head she uses No/Not interchangeably and it's flip a coin on what comes out, I find it endearingly cute, "Did you remember the milk?" "Not!".
I still confuse her with "aye" as an affirmative yes (I'm from Yorkshire in England and we still use aye and to a lesser extent 'nay' as an emphatic no).
When learning French you don't have that opportunity so you're going to keep having to remind yourself that not a single character in a French word sounds like what it does on your native language.
I've seen this in action when learning Finnish, a long time ago, in a class full of expats. Most Germans had serious trouble with the ä, which sounds entirely different in Finnish than in German. They would consistently pronounce it like in German, which made them sound ridiculous. The rest of the class, whose native language alphabet didn't contain the ä, got their pronunciation much closer to the real thing. I can only imagine how powerful this is if not a single character, but the entire script, is different from your native language.