Feles wrote:clawgrip wrote:Nevertheless, there are certain compounds that have undergone the change. The ending -hara often changes to -wara because of this rule, e.g. Fujiwara, Yoshiwara. Also 八幡 can be Yahata or Yawata.
Are there any cases when the ending happens after /ɴ/? What then?
/ɴ/ usually turns /h/ into /b/ or /p/:
何 /naɴ/ 'what' + 本 /hoɴ/ 'long, thin object' = 何本 /naɴboɴ/ 'how many (long, thin objects)'
金 /kiɴ/ 'gold' + 髪 /hatu/ 'hair' = 金髪 /kiɴpatu/ 'blond hair'
My intuition tells me the choice between /p/ and /b/ has to do with the preceding vowel quality, but I haven't researched it.
四 /joɴ/ 'four' is a notable exception, typically leaving the /h/ unchanged:
四本 /joɴhoɴ/ 'four long, thin objects'
Feles wrote:Where did the geminates come from?
Originally geminates come from compounding. /tu/ will reliably become Q as in finlay's post above when it comes immediately before an unvoiced stop or fricative. In front of /h/, it will cause a geminate /pp/
発 /hatu/ 'emit; release' + 車 /ʃa/ 'wheeled vehicle' = 発車 /haʃʃa/ 'vehicle departing'
物 /butu/ 'thing' + 価 /ka/ 'value' = 物価 /bukka/ 'prices; cost of living'
合 /ɡatu/ 'combine; fit together' + 併 /he:/ 'combine' = 合併 /ɡappe:/ 'merger; annexation'
/ku/ will also fairly reliably form geminates when it comes before /k/, e.g.
特 /toku/ 'special' + 許 /kjo/ 'permission' = 特許 /tokkjo/ 'patent'
白 /haku/ 'white' + 血 /ketu/ 'blood' + 病 /bjo:/ 'disease' = 白血病 /hakketubjo:/ 'leukemia'
There is a resistance to this in official word use when compounds are further compounded, e.g.
水族 /suizoku/ 'sea life' + 館 /kaɴ/ 'hall' = 水族館 /suizokukaɴ/ 'aquarium (i.e. a building for fish viewing, not a single tank)'
洗濯 /seɴtaku/ 'laundry' + 機 /ki/ 'machine' = 洗濯機 /seɴtakuki/ 'washing machine'
However, in informal speech, these are invariably pronounced /suizokkaɴ/ and /seɴtakki/.
Certain basic words also have irregular geminate-forming properties:
六 /roku/ 'six' ends in /ku/ but will form geminates with any stop;
日 /ni(ti)/ 'sun; day; Japan' and 八 /hati/ 'eight' will form geminates just like morphemes with final /tu/
Sometimes a geminate may spontaneously occur to retain the mora-length of a word when another sound is lost for some reason, e.g.
火 /hi/ 'fire' + 男 /otoko/ 'man' = ひょっとこ /hjottoko/ '
hyottoko, a traditional mask character' (reduction of /i/ to /j/ meant a loss of moraic /i/, causing subsequent gemination of /t/)
真 /ma/ 'true' + 赤 /aka/ 'red' = 真赤 /makka/ 'bright red' (loss of one moraic /a/ resulted in gemination of /k/).
In contemporary Japanese, geminates are also common in onomatopoeic expressions, such as びっくり /bikkuri/ 'surprised' or がっかり /ɡakkari/ 'dejected'. They also occur frequently in English loanwords in place of an English short vowel, cf.:
ビート /bi:to/ 'beat'
ビット /bitto/ 'bit'
The common geminates
-tta/-tte in verb conjugations come from contractions of older grammatical forms combining with /ɸ/, /t/, and /r/
Middle Japanese 思ふ /omoɸu/ + たり /tari/ = 思ひたり /omoɸitari/ -> */omoɸtari/ -> omotta
Middle Japanese 勝つ /katu/ + たり /tari/ = 勝ちたり /katitari/ -> */kattari/ -> /katta/
MIddle Japanese 取る /toru/ + て /te/ = 取りて /torite/ -> */torte/ -> /totte/
* This is theoretical. I don't know exactly how the sound changes and grammatical changes relate to each other chronologically, or if forms like */torte/ actually ever existed at all.
Feles wrote:Also I had a personal theorem regarding the vowels, dunno whether to show it.
Share your ideas.
P.S. I'm not sure if I'm supposed to write /naɴboɴ/ or /namboɴ/ here. I figured /naɴboɴ/ was more transparent, and the resulting sound would be predictable anyway based on allophony rules.