you get used to it, it took me a year and a half of mistakes and 30 second thinking but now I do the 3 group base to 4 group base and yen to dollars as fast as I used to do binary to decimal to hex back in university

I do it the same as everything else - without converting in my head.
If you think of a number in English and then try to switch it over (Just like if you think of a word and then try to switch it over) it won`t work well.

I do have to pause for a second when hitting things over 億... But I think that`s more because the number itself is so big and I rarely use it in normal life.

It`s kind of strange because 100万 and a million are the same thing... But they exist as separate entities in my mind.

I guess it's because I'm such a numbers guy: undergrad degree was in abstract math.

It's my one major linguistic flaw. I don't know why. I guess because numerical symbols and base systems are just so inextricably linked to words in my head—I was doing two-digit multiplication in first grade (thanks, Dad!), so maybe it's that long history of connecting sounds with symbols.

If they're written in Japanese with kanji, I've got no problems. But when I see 1.000.000 I think "OK, 10.000 is man, so 100.000 is juuman, so 1.000.000 is hyakuman!" and finally I can vocalize it.

Then again, maybe it's because I rarely have had to say anything greater than a few hundred in Japanese ever. I really should make a program that generates flashcards of Japanese numbers and see if I can train to say them fast.

This isn`t quite as bad as you think... Well, for 1,000,000 it is, but for larger numbers it seems quite normal for people to sort of count backward before saying the number. When checking numbers, it seems like everyone (at least around me) will do a sort of whispered 千、万、十万、百万 and so on under their breath before actually reading the number.

I wonder if it's because Japanese notates with . base-1000 but notes with kanji base-10000.

In other words, you put a . every three digits, but a new kanji comes into play every four.

1.000 man = 1.000 x 10.000 = 100.000.000
So in one, it's a "thousand" or something, but the other it's a "hundred" of something.

What I mean is that every three digits you have a dot, but every four digits you'd have man, oku, chou, etc.

Whereas in many western languages, you have them in harmony. Every three digits is a ., and every three digits you introduce thousand, million, billion, trillion, etc.

I think people struggle in English only past 100 billion, and that's only because almost never do you even have to use the word "trillion."