This has been obsessing me for quite a while. While 十三 is normally pronounced ジュウサン, it is often pronounced ジュウゾウ in people's names and also pronounced ジュウソウ in names of places. I am wondering how these pronunciations come into place. Are there any relevant historical perspectives? I don't see exotic alternative pronunciations for say 12 and 14, and it is not the pronunciation of traditional Japanese numbers either (which seems to be トサ), so what is making 13 so special?
Odd readings of 三: looking back in the history
I've read here and there that researchers think that the Chinese-derived reading さん was originally borrowed as さむ. This is based partly on the reconstructed Middle Chinese reading of
/sɑm/, and partly on the fact that Old Japanese (the stage of the language when most kanji were borrowed) didn't have any ん yet. In fact, the kana ん comes from a hentaigana form of む, deriving from kanji 无.
However, む on the end has been unstable historically, and tends to evolve phonologically in a few different ways.
- It becomes ん, as in archaic / formal constructions like 言わ[ん]【●】とする, from 言わ[む]【●】とする, where 言わむ is the classical volitional.
- It becomes う, as in 言わ[う]【●】, the precursor to modern volitional 言おう.
- In rarer cases, the
/m/in む shifted to a
/b/sound to become ぶ -- and we see that
/b/alternation in a number of words and placenames, like さびしい ↔ さみしい, or the name of the "Tama River" in Tokyo, which seems to become "Taba" at some points along its course.
Looking again at 三, the ancient さ[む]【●】 reading shifted regularly to さ[ん]【●】 in most instances -- but that final む took the other routes and became ぶ or う in a few places, notably in names.
For instance, there's the not-uncommon male given name さ[ぶ]【●】ろう, spelled in kanji as 三郎. We would ordinarily expect this to be さ[ん]【●】ろう instead, but here, ancient さ[む]【●】 became さ[ぶ]【●】, giving us the modern さぶろう reading.
In your example, 十三 would ordinarily be じゅうさ[ん]【●】, as you note. However, in some names, the ancient さ[む]【●】 became さ[う]【●】 instead. That would give us じゅうさう, to which we add rendaku, the voicing of an initial consonant in compounds, to yield じゅう[ざ]【●】う.
But what about the vowels?
Through normal vowel shifts, many cases of
/au/ in older forms of Japanese (pronounced like ow! in English) monophthongized or flattened out to became
/ɔː/ (pronounced like awwww in English).
/ɔː/ vowel sound, not-quite-あ-not-quite-お, was still distinct four hundred years ago, which we can see in various entries in the 1603 Portuguese-Japanese dictionary, Vocabulario da Lingoa de Iapam, a.k.a. the 日葡【にっぽ】辞書【じしょ】. Click through to the link, expand the page, and look at the two entries above the yellow-highlighted one -- Còxocu for 紅色 ("Curenaino iro", i.e. 紅の色), and Cǒxocu for 好色 ("Irogonomi", i.e. 色好み). Regular お is just plain o in the dictionary. Long おお or おう is ò. And the oddball ǒ is used to indicate that
/ɔː/ sound that arose from the flattening of
/au/. And sure enough, we can see in many monolingual Japanese dictionaries that 紅色 has always been spelled in kana as [こ]【●】[う]【●】しょく, aligning with the ò in the Nippo Jisho, while 好色 used to be spelled in kana as [か]【●】[う]【●】しょく, pointing towards
/ɔː/ and aligning with the ǒ in the Nippo Jisho. These old kana spellings were made obsolete by spelling reforms in the early-to-mid 20th century, when there was a push to get the then-archaic spelling conventions to align more closely with everyday pronunciation.
So between 1603 and today, all of those
/ɔː/ sounds further shifted to become the long
/oː/ sounds we're used to in modern Japanese, producing the long
/oː/ sound we hear in じゅうぞう.