I just watched the “Japanese historical drama horror film” Kuroneko (lots of fun if you like films with samurai and ghosts; this one features Minamoto no Raikō as a character, not to mention the titular black cat), and was struck when a large city gate was shown with the inscription 羅城門, subtitled RAJOMON GATE. “Is that different from the famous Rashomon?” thought I, and immediately investigated. It turns out the answer is “yes and no”; the gate is the same, but it has different names. As Wikipedia explains:
The gate’s name in modern Japanese is Rajōmon. Rajō (羅城) refers to city walls and mon (門) means “gate,” so Rajōmon signifies the main city gate. Originally, this gate was known as Raseimon or Raiseimon, using alternate readings for the kanji in the name. The name Rashōmon, using the kanji 羅生門 (which can also be read Raseimon), was popularized by a noh play Rashōmon (c.1420) written by Kanze Nobumitsu (1435–1516).
The modern name, Rajōmon, uses the original kanji (羅城門 rather than 羅生門) and employs the more common reading for the second character (jō instead of sei).
And if you continue to the article on the Nobumitsu play, you find: “The title is a pun, which involves the Rajōmon (羅城門) outer castle gate but Kanze changed it by using the kanji shō for ‘life’ rather than the original jō for ‘castle’ (note that 羅城門 was originally read raseimon and 生 can also be read as sei).” Complicated! The odd thing is that although the modern name is used on the gate and in the subtitles, when the characters say it out loud it’s clearly Rashōmon rather than Rajōmon.
According to Wikipedia “Japanese phonology”, the j sound is dʑ (voiced alveolar fricative) and there is no sh sound apart from ɕ (voiceless alveolo-palatal sibilant fricative, sounds more like Russian shch to me…).
The intervocalic voicing of consonants (rendaku) is quite variable in different compounds and different regions.
https://en.wikipedia.org/wiki/Rendaku
Compounds with Sino-Japanese forms (as in the case of Ra+sei/sho/jo+mon) usually resist rendaku, but there are many exceptions.
A friend of mine whose family goes by Kaneshiro (Goldcastle) says most of his relatives in Okinawa pronounce the surname as Kinjo (Sino-Japanese).
Wow, that’s weird.
The family name is 金城 ‘gold castle’ but pronounced variably, at least in Okinawa. (I’m now working on a newly installed Win11 PC and had to reinstall my Japanese and Chinese IME options.) I had until now only installed the Polish and Romanian programmer’s keyboard options.
That’s exactly what the Russian щ is. Well, the oldest surviving pronunciation is [ɕt͡ɕ], but [ɕː] is much more common nowadays, and due to the lack of a short counterpart it routinely ends up as just [ɕ].
Maybe you were thinking of the Mandarin x? That is often transcribed as [ɕ], but that pronunciation seems to be limited to southern accents that also don’t retroflex (sh, ch, zh = s, c, z). In the north, including the CCTV accent, x is something that doesn’t have an IPA symbol: a dorso-palatal sibilant, acoustically filling the space between [ɕ], [ç] (ich-Laut) and [sʲ] (Russian сь); cutesy girlish accents approach [sʲ] the closest. The Wade/Giles transcription hs was not completely arbitrary by any means.
The irregularity of rendaku (intervocalic voicing) shows up in placenames all over Japan. Compare 長崎 Nagasaki ‘long-slope’ vs. 岡崎 Okazaki ‘hill-slope’ and 広島 Hiroshima ‘wide-island’ vs. 中島 Nakajima ‘middle-island’. Or 福島 Fukujima city in 富山県 Toyama-ken vs. 福島 Fukushima city in 福島県 Fukushima-ken.