To add to this, voicing is also a way for Japanese words to become more “coherent”, the same way you write “dislike-chopsticks” as one combined noun, and not “dislike chopsticks”.

Someone downvoted this, but the poster is correct, so there was absolutely no reason for downvoting.

Rendaku, i.e. the voicing of the initial consonant, happens in the native Japanese words (i.e. not in the Japanese words of Chinese origin), in most cases when they are a part of a compound word and they are not the initial word. This serves indeed to distinguish a sequence of unrelated words from a compound word.

There are exceptions when rendaku does not happen, but typically whenever a word like hashi becomes a part of a compound word it will be voiced to -bashi.

"H" is a special case among the consonants, because in old Japanese it was pronounced as "p", which is why it is voiced as "b". Later, in initial positions the pronunciation was changed to "f" and even later the pronunciation was changed to "h". The "f" pronunciation has been retained only before "u", like in Fuji. In non initial positions, the original "p" has become later "v" and even later "w".

These pronunciation changes happened after the creation of the hiragana and katakana syllabaries, so they were not reflected in writing. The orthographic reform that was forced after WWII has brought the written form of the words closer to the pronunciation, e.g. by writing consistently "w" where it is pronounced so. Before WWII, many words written now with "-wa-" were still written with "-ha-", a spelling that has been preserved now only in the particle "wa" (like the spelling corresponding to the old pronunciation "wo" has been preserved for the particle "o").

While the Japanese orthographic reform had some positive effects, in simplifying a little the Japanese writing, it also had the effect that for someone who knows only the modern written Japanese it is difficult to read the Japanese books published before WWII, where many different kanji are used and also their hiragana transcriptions are different.

I assume that this was actually an effect intended by the American occupation forces, as a similar policy was applied by the Russians in all the territories of the Soviet Union (except the Baltic countries), where they forced the native populations to change their writing systems to the Cyrillic alphabet, in order to make difficult for the younger generations to read anything dating from before the Russian occupation.

> The "f" pronunciation has been retained only before "u", like in Fuji.

Well, there is a convention that syllables starting with h- are spelled with f- (in foreign transcription) if the following vowel is -u. There's not much difference in the pronunciation itself; maybe there was more of one when the spelling convention was set.

At least in the recent past and probably also today in some Japanese dialects, the "f" pronunciation must have been retained before "u".

For example, in some Okinawan dialects the "f" pronunciation has been retained before all vowels.

Because of this, after Okinawa was occupied by Japan in the last quarter of the 19th century, the Japanese used "fu" before vowels, to transcribe the Okinawan pronunciation. For instance, the Okinawan syllable "ha" (pronounced "fa") was transcribed by the Japanese as "fua", because writing it like "ha" would have resulted in a too different pronunciation.

So at least by that time "fu" must have been still perceived as clearly different from "ha", "hi", "he" and "ho".

> At least in the recent past and probably also today in some Japanese dialects, the "f" pronunciation must have been retained before "u".

I wasn't disputing that as to the recent past.

I searched up some Japanese-language videos on youtube as a followup, and I can report:

A noticeable "f" is present before "u" in many cases. (I found it in the words "tofu" and "daifuku", plus the obvious English loanwords "soft", "firm", and "waffuru". My best guess as to the vowel following "f" is "u" for "soft" and "a" for "firm".)

But, not consistently. You don't have to pronounce the syllable that way. (Observed also in "tofu" and "daifuku".)

The nature of my low-effort search precludes any statements about dialectal variation. I wouldn't want to claim that the syllable onsets are "clearly different" to modern speakers today. But (1) the option to have an "f" is still present in -u syllables, and (2) the existence of common loanwords where the foreign sound is recognized is, if anything, going to serve to strengthen awareness of the hypothetical difference.

I agree with you.

I was explaining the historical pronunciation, because without knowing it, for the English speakers there are many puzzling things related to the syllables starting with "h", e.g. why "hashi" is voiced to "-bashi", why hiragana "huji" is transcribed to Latin "Fuji", why the particle "wa" is written "ha" in hiragana, why the capital city of Okinawa, which is now written "Naha" (because now the traditional Okinawan pronunciation does not matter any more) can be found in older texts written as "Nafua", why "Yawara" (the original native name of what is now called "jiu-jitsu", through translation into Sino-Japanese) was written in hiragana as "yahara" in the old books, and so on.

As you have mentioned, modern Japanese frequently uses "fu" before vowels or in final position, in transcribing the words borrowed from English or other languages, to mark the consonant "f", which otherwise does not exist in Japanese, and in these borrowed words it is more likely to be pronounced as English "f".