May 13, 2006


Last week I looked at mispronunciations and misspellings of the name Mitsubishi, in particular Mitsubushi (the winner in the misspelling bee) and Mitsibushi and Mitsibishi (the runners-up), arguing that the problem presented by Mitsubishi isn't in nativizing the Japanese name, but in remembering and retrieving the name correctly.  Then came the mail (it's always like that here at Language Log Plaza): about actual problems in nativizing words (mostly from Japanese), about Japanese car names and their etymologies, about factors that might have helped boost Mitsubushi to the top of the heap, and about still more manglings of Mitsubishi.  Here's the digest.

Problems in nativization (not all of which I understand).  Bill Poser, who has the cubicle just down the row from mine at LLP, wrote to wonder why English speakers so frequently mispronounce (and sometimes misspell) harakiri (as hari-kari) and karaoke (as karioki). 

Part of this -- the [i] instead of [e] at the end of karaoke -- is easy.  Final unaccented [e] is at best marginally acceptable in English, and is normally "fixed" by raising it to [i]; this sometimes shows up in spelling, in the variant karaoki.  You see raising not just in karaoke, but also in, for instance, karate, in Hare Krishna, and in some borrowings from Italian, like the salami and zucchini (Italian salame and zucchine) that M. I. Amorelli asked about (from Sardinia) on ADS-L back in April.

The second vowel of karaoke -- which is sometimes spelled karioki, in line with its most common pronunciation in English -- is a bit trickier.  This unaccented vowel would be expected to come out as a schwa, giving a sequence of vowels that isn't actually unpronounceable in English (it occurs in supraorbital) but is very rare.  So possibly that [i] is just a fix in the direction of a better unaccented vowel before [o].

[Added later 5/13/06: Aaron Dinkin writes to remind me that (unaccented) schwa generally gets raised to [i] before a vowel, as in Judaism and the three-syllable version of Israel, so karioki is just what you'd expect.  Words like supraorbital and (six-syllable versions of) extraordinary don't show raising, because they're "level 2" morphological combinations (in Kiparsky's terms).]

Harakiri > hari-kari is more puzzling to me.  Something like para-teary seems entirely pronounceable to me; it's just an absurd combination of elements.  As it happens, NOAD2 gives a straightforward nativization of harakiri as the first pronunciation, but then admits that the rhyming pronunciation -- the one I hear from everybody except pedants and people who actually know something about Japanese -- is also possible.  AHD4 goes a step further, and gives hari-kari as an alternative SPELLING for the word.  The correct A A I I spelling outnumbers the rhyming version A I A I about 5 to 1 in raw Google webhits, but we're still talking lots of A I A I spellings. 

Google also turns up some A I I I spellings (with the second vowel anticipating the two I's -- and [i]'s -- that follow), about one-third as frequent as the A I A I.  But this version might have served as an intermediate step from A A I I to A I A I: first anticipation, in A I I I, then an improvement of this into the satisfying rhyming pattern of A I A I.

I know, some of you are thinking that the baseball announcer Harry Caray (1914-98), whose name was pronounced just like hari-kari, must somehow be involved here.  But no, as a quick trip to OED2 shows.  The first OED cites under hara-kiri in fact have the spelling hari-kari, and this is in 1856, 1859, and 1862, surely before Harry Caray's PARENTS were born.  (By the way, H.C. was born Harry Christopher Carabina.)  We don't get "correct" spellings until 1871.  In 1888, we get one of each of these spellings, plus the possibly intermediate version hari-kiri.  So whatever is going on here, it's been going on for a very long time.

Japanese car names.  Several correspondents have pointed out that Mitsubishi is a meaningful compound in Japanese: mitsu 'three' plus hishi 'diamond' (in its voiced variant bishi).  The three diamonds are visible in the company's logo.  This has damn-all to do with the pronunciation or spelling of the name, but it's still entertaining.  (Even cooler is Bill Poser's observation that karaoke is also a compound: kara 'empty', as in karate, literally 'empty hand', plus oke, which is, wonderfully, a borrowing of English orchestra, somewhat truncated.  So orchestra traveled to Japan as oke and then came back inside karaoke.)

One other Japanese car name, Isuzu, gives trouble for English speakers.  Here the vowels are fine, but the S Z sequence is problematic.  As one correspondent pointed out, you'll hear the reversed Izusu (even in some old Isuzu commercials!), and occasionally the assimilated Izuzu.  Here the trick is to explain why Isuzu is troublesome but Suzuki is not.  Probably something to do with the voicing of [s] (when spelled with a single S) between an unaccented vowel and an accented one, as in presume.

Facilitating factors.  Back to Mitsubishi.  Several correspondents have suggested things that might tip the scales in favor of I U U I, over its closest competitors I I U I and I I I I.  The most common suggestion is that bushido, literally 'warrior's way' (bushi + do:) and referring to the code of the samurai, favors bushi (U I) in the second half of the name.  Of course, both I U U I and I I U I have bushi, but I U U I has the extra advantage of preserving the first half, I U, of the original.

More important, I'm doubtful that the word bushido is widely known among English speakers, even though it did make it into both AHD4 and NOAD2.  And even more doubtful that many speakers appreciate that bushi is a significant piece of the word bushido -- though there are Bushido Blade video games, and people who play these games are likely to know bushi meaning 'warrior'.  In any case, I think that the most bushi(do) could have done is helped I U U I a little bit.

One correspondent did suggest the word sushi (which is probably the Japanese word most widely known to speakers of English -- outside of brand names, of course) as a factor facilitating I U U I.  This has some plausibility.

Finally, one ADS-L poster suggested that I I U I should be favored because of the English proper name Mitzi.  It is true that of mitsi, mitsu, mutsi, mutsu, bishi, bishu, bushi, and bushu, only the first is pronounced like a generally recognizable English word (though bushi is not too far from bushy).  There was a general feeling on the ADS-L that Mitzi is now a name of too little currency to have much influence in reshaping word forms.

More manglings.  One further correspondent reported that a pronunciation with two [š]'s -- Mishybishy, as he represented it -- "was quite prevalent in Charlottesville VA a decade ago" and that he'd never heard that version anywhere else.  Here we have an anticipation, in the second syllable, of the [š] in the fourth syllable.  Plus the shift to the I I I I pattern.

It turns out that the spelling Mishibishi gets a modest number of relevant Google webhits, and they seem to come from all over the place, including Australia.  Mishubishi, preserving the original vowel pattern, gets even more.

No doubt there are more.  Well, yes, there are a few occurrences of Misubisi.  And Michubichi.  But it's time for me to put the Mitsubishi file away.  I can barely spell the word myself any more.

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at May 13, 2006 02:53 PM