The company is Toyota, but the family name of the founder is Toyoda. Why the difference? Bill Poser discusses it at the Log; after citing an implausible theory about stroke count, he says:

Another explanation is that Toyota served to dissociate the motor vehicle company from farming, which advanced the company’s goal of presenting itself as innovative and high-tech. A third is that voiced sounds like [d] are considered to be “murky” while voiceless sounds like [t] are considered “clear”. Finally, it may be that the aesthetics of the logo played a role.

He shows alternate versions of the logo, and I have to agree that the one without the dakuten looks better, which is not to say that I believe that version. A useful comment by Gene Buckley says:

An important fact is discussed in the linked BBC article, and is implied by the link to rendaku on Wikipedia by Dan, but it might be useful to make it explicit on this page. The written form 豊田 can be read Toyo-da, with voicing of the initial consonant in the second morpheme, or as Toyo-ta, without this voicing. (Other family names have similar alternate forms, such as 山崎 as Yama-saki and Yama-zaki.) In fact, the pronunciation Toyota is more common as a family name, according to Japanese, Chinese, and Korean surnames and how to read them (W. Hadamitzky, 1998). It’s hard to imagine that the greater currency of this alternate pronunciation of 豊田 played no role in the choice of the company name. The katakana spelling adopted for the company name removes the ambiguity in the pronunciation of the second Chinese character.

(Please ignore the unseemly squabbling about national flapping in the early comments.)


  1. The explanation commonly heard in Japan is that the stroke count both in hiragana and katakana without the dakuten is the propitious (in Asia) number of 8.
    とよた トヨタ
    I don’t know much about probability, but the likelihood of a word having the same stroke count in both hiragana and katakana would seem vanishingly low, which probably also contributes to the propitiousness of this case.
    Now, you might wondering, “Stroke count? Who cares?” Well, stroke count is one of the key factors taken into consideration when choosing a baby’s name and the characters for the name (last name + first name for boys, but first name only for girls, because their last name will change (by law) after marriage), so this explanation is the kind of thing that results in much head-nodding in Japan, apocryphal or not.

  2. Yes, but who ever heard of stroke count for kana (apart from this theory)?

  3. Philip Spaelti says

    “I don’t know much about probability, but the likelihood of a word having the same stroke count in both hiragana and katakana would seem vanishingly low,”
    Indeed you don’t seem to know much about probability. Almost all kana have 2 or 3 strokes (a few have 1 or 4). So the chances that any given word will have the same count in both hiragana and katakana are in fact extremely good.
    Also from these numbers you can see that 3 kana words are extremely likely to have a count of 8.

  4. Almost all kana have 2 or 3 strokes (a few have 1 or 4). … Also from these numbers you can see that 3 kana words are extremely likely to have a count of 8.
    I fear that on this evidence what you say about Marc’s comment:

    Indeed you don’t seem to know much about probability.

    applies equally to you.
    For simplicity, let’s consider only kana with 2 or 3 strokes, since you say that almost all have this number of strokes. Any three kana words display one of the following patterns, disregarding word order (I represent each word by the number of strokes in it):

    2 2 2 = 6 strokes total
    2 2 3 = 7 strokes total
    2 3 3 = 8 strokes total
    3 3 3 = 9 strokes total

    On the assumption that kana words with 2 and with 3 strokes occur with similar frequencies (among all those having 2 or 3 strokes), we see that there is a 25% probability of three words having a total of 8 strokes. That’s hardly “extremely likely”. The probability of getting heads in a coin toss is greater.

  5. Yes, but who ever heard of stroke count for kana (apart from this theory)?
    Wow, everybody’s in a bad mood here today, huh? The dust-up that got erased must’ve been really bad.
    There’s no reason kana would be excluded for the calculation. Even Latin letters are counted. (More examples here here here here…) Lots of girls have hiragana-only names. Katakana-only girls’ names were a fad in the early decades of the 20th century. Here’s a partial list (from a stroke-count evaluation/divination website): only famous people whose last name is in the a-gyo. There’s quite a few examples of people with kana-only first names in that list: 安立スハル 彩乃かなみ 綾野まさる 綾小路きみまろ あびる優
    Now, a lot of these are probably geimei (stage names), but geimei are extremely closely vetted for propitiousness by the agencies taking on the (hopefully future) stars.
    And who does the vetting? Well, there’s people (basically fortune-tellers) who you can pay to do it. It used to be done without question (and you’d pay crazy amounts, too — $500+) when a baby was born, when a company was started, etc. These days there’s books which are filled with charts (it also has to do with when the baby’s born, I think), so a lot of people just buy those. In the case of my children, my wife’s grandmother did all the calculations (and my son’s name is Julius, so his was calculated in Latin letters and in katakana).
    Here’s one company that gives professional advice on corporation naming. There’s lots of these.
    Oh in fact even punctuation is counted: the period at the end of Morning Musume’s name: TBS系の番組「うたばん」で、「。」について聞かれた時にメンバーのほとんどは不要と答えた。しかし、中澤裕子の誕生日(26歳)を前に1999年6月13日に放送されたフジテレビ系の番組「ハッピーバースデー!」での安斎勝洋による姓名判断で「『。』が付いたことで画数が23画となり、理想的なグループ名となった。」と言われた後、肯定的な発言をしている。「。」のない22画ではグループが分裂すると言われているため、後に別の番組でリーダー(当時)の中澤が「『。』は大事。」と語っている。
    (Basically, the stroke count is better with the period at the end.)
    Anyway, it’s a huge thing in Japan.
    Philip, that’s a good point, but remember – the coincidental stroke count isn’t the point. It’s the fact that not only do they coincide, but they coincide at 8 that’s important in this case. Still, I think you’re forgetting that the probability involved here requires that a kana character has to have the same stroke count in both hiragana and katakana variants. All kana could be either 2 or 3 strokes, but if the stroke count doesn’t match for the same sound, the probability of any given word having the same stroke count in hiragana and katakana remains pretty low.
    As I say, though, I don’t know much about probability, so I’d be more than willing to take a look at some examples. You say the odds are extremely good, so it shouldn’t be that hard to come up with a convincingly substantial list.
    Also, if everyone could tone down the snipy-ness a bit, it would make the discussion a lot more pleasant. One of the things that I like about this blog is that people are nice to each other, unlike the rest of the intertubes.

  6. I pronounce it as “Toyoda” anyway, and I’m sure many other Americans do the same.

  7. Yeah, it’s been interesting to hear American newscasters and reporters trying differentiate between Toyota and Toyoda.
    (By the way, Grumbly Stu, your message came in while I was writing mine, so the snipey-ness comment isn’t directed at you.)

  8. Wow, everybody’s in a bad mood here today, huh?
    I certainly didn’t mean my comment to read as snippy, more as jovial/inquisitive. Thanks for clarifying the situation!

  9. if everyone could tone down the snipy-ness a bit
    I sometimes find it opportune to give snipy-ness a good whack, in the most pleasant fashion.
    But you’re right, people are civil at Hat’s. Occasionally a slipper-biting pseud turns up, but such people generally retire in a huff within a few months.

  10. LH: my bad. I misread your comment, then.

  11. Propitiousness odds of 25% are pretty good, on my view. They’re not “vanishingly low”, but not “extremely likely” either.

  12. Grumbly Stu, if you’re going to ignore the order you have to count 223 and 233 three times each. The probability is 3/8 = 37.5%, not 1/4 = 25%. Still not extremely likely, though.

  13. if you’re going to ignore the order you have to count 223 and 233 three times each
    Oops, you’re right and I’m wrong. It’s back to the urns for me. It was just obvious that the probability was neither “vanishingly low” not “extremely” high.

  14. Here is the correct argument, for anyone who is interested:
    Any three kana words have one of the following patterns (I represent each word by the number of strokes in it):

    2 2 2 = 6 strokes total
    2 2 3 , 2 3 2, 3 2 2 = 7 strokes total
    2 3 3 , 3 2 3, 3 3 2 = 8 strokes total
    3 3 3 = 9 strokes total

    There are 8 possible different outcomes. They all have equal probability, on our assumption. Three outcomes have a stroke sum of 8. So the probability of getting a word with a stroke sum of 8 is 3/8.

  15. Actually, a lot of the kana have one stroke, and some four. Also, names vary from two to five characters.
    Those factors reduce the probability even more, don’t they?
    I take back the “vaninishingly small” (I just wanted to use the word “vanishingly” ;P), but the odds are still low.

  16. God, it gives me the creeps to read Language Log. They’re like a bunch of football hooligans. And the commenters aren’t much better.

  17. If I remember rightly, the town of Toyoda in Aichi prefecture was named after Toyota, and retains the nigori.

  18. It depends on what you mean exactly, numerically, by “a lot of the kana”, and “vary from two to five characters”. I showed you how to do the math in principle. If you don’t know the numerical proportions of words with different numbers of strokes, there’s not much point in speculating at all.
    On Philip’s assumptions, we’re at around 37 % propitiousness. If you guys have very different impressions of the actual numerical proportions, then the math results will be very different. Either you stick with gut feelings, or trade them in for some simple statistics. Aren’t there any stroke-count frequency tables at any of those divinatory websites ?

  19. Well, according to English Wikipedia, the city’s name is Toyota. And yet, I seem to remember it being called Toyoda when I went there….
    A quick check of Japanese Wikipedia reveals that:
    “濁音で「とよだ」と誤読される事があるが、正しくは清音で「とよた」である。” (It is sometimes mistakenly read with dakuon as “Toyoda”, but correctly should be “Toyota” with seion.) I was obviously mixing with the wrong people….

  20. There was a director at the company I worked at whose last name was 小谷 (Otani) but who pronounced it Odani. I found out when I made his card with “Odani” on it (because that’s the way he said it) and it came back with a correction. I asked him and he said it was Otani, but he just found it easier to say Odani.
    I don’t know if this is dialect (in Toyama dialect 自転車 – jitensha – bicycle is pronounced jidensha, so there might be something there), but to write your name one way and pronounce it another — that was a first.

  21. Grumbly Stu: point taken. I really do suck at anything to do with numbers, so I’ll bow out of the conversation. I will just point out that the probabilities you have given are the probabilities of any single word having that stroke count, but what we’re talking about is the probability of a word having that stroke count in both katakana and hiragana. When you add in the fact that Japanese words vary in length beyond the range given your calculations, I think it’s safe to say the likelihood of a word’s hiragana *and* katakana stroke count equalling 8 is a lot lower than 37%. At most half that, is what my gut tells me.

  22. When I hear Japanese say 自転車 jitensha it always sounds like jidensha, and for a long time I thought that’s what they were saying. Later I realised that the Japanese /t/ (non-aspirated) in this phonological context sounds like a /d/ in English.

  23. Going by the counts in Marc’s first link, in hiragana (restricting ourselves to modern characters without diacritics, for now, & excluding を), there are:
    5 1-stroke kana
    10 2-stroke kana
    20 3-stroke kana
    8 4-stroke kana
    2 5-stroke kana
    To get 8 we need:
    1 2 5, 1 5 2, 2 1 5, 2 5 1, 5 1 2, 5 2 1 : 6 × (5/45 × 10/45 × 2/45)
    1 3 4, 1 4 3, 3 1 4, 3 4 1, 4 1 3, 4 3 1 : 6 × (5/45 × 20/45 × 8/45)
    2 2 4, 2 4 2, 4 2 2 : 3 x (10/45 × 10/45 × 8/45)
    2 3 3, 3 2 3, 3 3 2 : 3 × (10/45 × 20/45 × 20/45)
    = ~0.217, unless I’ve made a mistake somewhere in that.

  24. They’re like a bunch of football hooligans.
    Where on earth are you from?
    We’re from England

  25. probability of three words having a total of 8 strokes
    This isn’t a strictly a probability problem. The best way to keep track of the possible combinations would be with a truth table, ignoring the outputs.
    Going by the counts in Marc’s first link
    So 20 out of 35 of the characters have 3 strokes. Whether this probability is “high” or not is a matter of interpretation (or whether it would be high after taking the other factors into account). If you were getting on a airplane with 35 seats and you knew that only 20 of them would eject upon reaching a certain altitude, would that be “not extremely likely”? Or compare it with, say, the probability of rhyming the word orange.

  26. Okay, so it’s 22%, but then you have to add (subtract? multiply?) the probability of this eight-stroke word having eight strokes in both katakana *and* hiragana. Gut feeling, but it’s got to be around 2%. And then if you remove the non-name words from this (since we’re talking about a name), I think it’s safe to say that the likelihood of a name having exactly 8 strokes in both katakana and hiragana is pretty damn low.

  27. You pick the hiragana word with the auspicious number of strokes, since that’s the hardest to control for; the corresponding kana probability is the only thing you are worried about. If it’s say, one in three, you only have to try three hiragana words before you hit one that’s auspicious in kana.

  28. @bathrobe: The city of Toyota in Aichi where Toyota is headquartered has the same pronunciation as the company (and is written using the characters 豊田).

  29. Being in a business where exact workding can mean the difference between failure and success, I can understand the need to change the “D” to “T.”

  30. Sloppy workding has always worked for me. As long as it dings, it works, as far as I’m concerned.

  31. Strokes may be answer, but my ‘tort’ As they were advertizing in Saxon, it would be so symmetrical to say ” A Toyota ” in your drive way, so when looking in thy rear view mirror you read “atoyot a

Speak Your Mind