Since I started learning Japanese in Japan, I was always surrounded by the language, even if it was just in the background in coffee shops. So for learners like me, even if we don't pay much attention to pitch, we hear it so much that some of it probably seeps in. Not that I speak with correct pitch, because I know that I don't.

For learners who don't live in Japan, and who don't get that kind of exposure to spoken Japanese, it might be good to be aware of pitch-accent and listen for it a little deliberately.

I also find it interesting that native speakers can go through a video like Steve's and pick out the individual words where the pitch is wrong. I definitely couldn't do that. I wonder if any foreigners develop that clear a sense of pitch in Japanese.

I remember at the start not being able to distinguish the difference at all, but there was a point when the pitch of some words suddenly seemed very obvious. A notable example was the difference between 昨日 and 機能. They're both きのう, but they sound completely different to me. However, that's certainly not true for all words, especially when the change in pitch is on the second half of a long vowel. I don't normally give it much thought when speaking, but do sometimes notice that my pitch is wrong.

As I say, though, I really think that intonation is more important because it expresses a lot of meaning in itself. If the intonation is right then I think the pitch changes fall naturally into place. As was mentioned before, the accent is partly used to to distinguish word breaks, and a natural intonation will automatically have pitch changes in approximately the right locations to facilitate comprehension, even if it isn't correct for individual words. The standard Japanese accent has a naturally falling pitch that rises to mark word breaks. Speaking without that underlying falling pitch is a lot less engaging even with the correct accent on words.

Hi Roan. If you're talking about pitch-accent, then the overall trajectory is down, but individual sentences still run up. In other words, pitch rises until an accent, at which point it drops down. This is the "downstepping" of the standard accent. But as someone is speaking, the pitch is generally getting lower and lower because the pitch keeps dropping at accents, even though it rises between them.

I agree with you about 昨日 and 機能. I checked my intuition about the pitch-accent of those words in my dictionary and I was correct. Maybe I absorbed more than I thought. The fact that my electronic dictionary included pitch-accent markings probably helped because I used to look words up a lot and would have automatically seen their pitch.

When I say that I don't have pitch-accent, I mean that I don't have this "downstepping" in my voice - at least I don't think I do. I don't know if my voice is getting lower overall or not. Anyway, I'm just going to continue trying to concentrate on listening for pitch for the next few weeks and see what happens. I'll be pleasantly surprised if I become more consciously aware of it.

I think I have an interesting experience to share here about pitch-accent (and pronunciation in general).

I’m a French Canadian. I have been in living in Japan for about 15 years (from 1992 to 1994, then from 1997 to today). Of these 15 years, I spent 6 months in Osaka, and the rest here and there around Tokyo. My wife is Japanese (from Yamagata-shi) and we speak only Japanese. I’m not conscious of my pitch-accent when I talk (neither do I when I listen to my wife), and I have never tried to study the pitch-accent.

Quite often when I answer the phone, it takes some time before people finally realize that I’m a foreigner. It’s only when I have to say relatively long or complex things that it becomes evident that I’m not native. For this reason, until recently I thought my Japanese pitch-accent, intonation and pronunciation were similar to the common Japanese way of speaking standard Japanese, at least for short sentences and daily life expressions.

But I had a big surprise recently. A lot of translators who develop pain in their back or shoulders (from bad posture while typing) use a voice recognition software (like Dragon Naturally Speaking) to dictate their translations to the computer. Since my Japanese version of Windows 7 included a free voice recognition utility, I decided to give it a try. This was a few months ago.

Result
That day, I felt frustrated because quite often I had to repeat my voice commands (even very simple ones with only 2 or 3 words) a few times before they were finally recognized by the computer. I thought it was because I used the tiny microphone integrated to the computer, so I thought "Well, the computer will surely understand me when I try with a real microphone".

But then my son (11 years old, native Japanese speaker) came back from school and, looking at the computer from behind my back, started to repeat my voice commands "the proper way". And most of the time, the computer understood his commands the first time (a funny exception or bug was with the command "次"... our voices were never recognized, but it suddenly worked when, joking, I said "Next" !). 爆笑

I don’t know if it’s only a matter of pitch-accent, but I was sort of knocked-out by the machine’s verdict on my spoken Japanese. I don’t really care about it since I feel comfortable in daily conversations, but still, I know that the computer, in contrast to human beings, will never pretend that it understands me when it doesn’t... ;-)

I wonder if other persons here have some experience to share with voice recognition in a foreign language.

This is probably a controversial area, because the explanation I heard was that the the upsteps are actually more rapid that the downsteps, with the voice slowly falling and then rising suddenly on a lexical word break.
For example, in 東京が日本の首都です the voice falls all the way from ト to ガ, but rises sharply at ニ and then falls again to the end of the sentence. The change in pitch has to fit it with this tendency, although when words are said in isolation the accent can be markedly different, particularly for LH words, as there is no longer anything to drive the overall fall.

Also, Japanese people hear a change in pitch even when the vowel is devoiced. I.e. there is no audible change but if one is expected then it is perceived.

I found the web page below quite interesting:
http://hasegawa.berkeley.edu/Accent/accent.html

Basically it says that the whole pitch accent thing is based on Japanese native speaker's perceptions of pronunciations, rather than waveform analysis. Therefore it is not necessarily a reliable model, particularly for a non-native learner of the language.

Pitch-accent is a lexical phenomenon which has to interact with the overall prosody of the phrase. When I say your phrase to myself, it seems that I'm doing what you say. So, what we end up with in the practical world, is lexical pitch-accent interacting with other aspects of prosody. But pitch-accent itself is heard as the fall of pitch from one mora to the next.

The brain often hears what it expects to hear. If it knows that a pitch-change is meant to happen on a devoiced vowel, and the rest of the phrase functions as if that devoiced vowel had experienced its pitch change, then the brain will hear it as if it had happened, whether or not the vowel was pronounced. This sort of things happens a lot in phonetics/phonology. We're not hearing raw data, but rather data that has already been processed by the "language filer" in our brain. To a certain degree, you can say that everything in pronunciation is based on native speaker's perceptions rather than strict data.

To give an English example: the words prince and prints. Probably, many people think these words are pronounced differently despite the sound being (for many people) exactly the same. If we take "prints". This is clearly /prints/ right? Whereas "prince" is /prins/. But is it? The only real difference between "n" and "t" is that "n" is a nasal sound and allows air to escape through the nose, whereas "t" does not. "s" is not a nasal sound, so the nasal passageway needs to be closed when you pronounce it. Many people cut the nasal sound of "n" off early. So when they transition to the "s" they essentially produce a "t" sound even though they're not intending to. They inadvertently pass through "t" on the way from "n" to "s". But people don't hear a "t" because /t/ is not present as a phoneme in that word. The sound of "t" only appears as a result of a bit of laziness in articulation. People hear the "t" in prints because /t/ is present as a phoneme. What we expect to hear greatly conditions what we actually hear.

There's a good book called "An Introduction to Japanese Linguistics" by Natsuko Tsujimura, which is not very technical. It makes interesting reading. But the wikipedia articles are also good.

At any rate, when people start having these detailed discussions about linguistic details, its probably a decent sign that you've wandered into territory which is no longer useful for the average language learner. However, for those with an interest in Japanese Linguistics, let's have at it!

I agree with the point that this discussion in itself is proof that pitch is not necessary for a beginner. If you do a search for academic papers on pitch in Japanese languages, whilst there is consensus that it exists, there is a wide range of opinions on its form and classification. I seen an article that claims it should be classed as an impoverished tonal language, and an other that says that it is closely related to stress in English. Well, it can't be both.

In the article I linked to, it gives similar examples where perception in English differs from reality. E.g. the difference between 'chi' (Greek letter) and 'guy' is actually aspiration rather than voicing. If a foreigner tried to improve their pronunciation by following a native speaker's perception then they will actually end up sounding very strange indeed.

I finished a Japanese pronunciation course a while ago, and I really wanted to continue but time and money didn't allow. Anyway, the main gist of having correct Japanese intonation was as follows:
* Individual words have a set pitch pattern
* This pattern is not used when the word is part of a sentence in normal speech
* Words do still differ, but by the rate at which the pitch falls, rather than rising and falling
* The difference in fall rate is not necessarily related to the pitch pattern in the isolated word

There was quite a bit more to it that that, though. The teacher was really good.
If you're interested, this was the place:
http://www.nihongo-hatsuon.jp/

Hi Roan. I haven't yet checked the first article you linked to, but I will. That 'chi' and 'guy' example is a great. English voiced consonants, for example "g", would be considered unvoiced, for example "k", in many languages. We actually differentiate between aspirated stops (p, t, k) and very lightly voiced stops (b, d, g). If you say the word "stop" and listen to the "t", you'll hear what I mean. It's not aspirated, so if you listen to it, it probably sounds more like a "d" to you, even though it's just an unaspirated "t".

I'll check out your pronunciation link. I'm curious about that. Unfortunately I'm not in Japan right now, but I'm very interested in the approach they take to pronunciation. It's really interesting that it's not just a school aimed at foreigners, but rather than an overall pronunciation/speaking school for Japanese as well who would like to do voice work. I used to wonder if any of those schools had adapted their techniques (whatever they are) for foreigners.

My technical knowledge of Japanese pitch-accent is not deep, and most of it is from quite a while ago, so I can't add too much more to the discussion in a technical sense without doing more refresher reading, but I find it very interesting.

I'm kind of suspicious of the claim that lexical pitch (basically pitch accent) is "not used" in a sentence. My suspicion is that it's there, but it's too difficult to explicitly explain how it interacts with other factors, so you just step back and look at the intonation of the phrase or sentence.

Bortrun:"I'm kind of suspicious of the claim that lexical pitch (basically pitch accent) is "not used" in a sentence".

Ditto. It obviously is.

@Roan,

"I seen an article that claims it should be classed as an impoverished tonal language, and an other that says that it is closely related to stress in English. Well, it can't be both. "

In a sense, it can be. It's lexical in the same way that stress is in English. And it interacts with the overall prosody of the sentence, as English stress does. And it involves (albeit to a limited degree) differentiating words due to their pitch pattern (or tones), as tonal languages do.

As I recall, pitch-accent is not a very common phenomenon. It shows up in some West African language, but pitch-accent there is different than Japanese pitch-accent. I could be wrong, but I don't recall other languages having pitch-accent in the way that Japanese has it.

@Higashi,

I haven't tried one of those programs, but a great deal of fun was made of the fact that Siri couldn't understand the Scottish accent. Lots of funny videos on youtube.

This is just my personal experience, but the Japanese native speakers possibly hear pitches even in English words. When I listen to English, sometimes I am not able to recognize words very well in question sentences, in which individual words sound much flatter than in statements. Basically once we learn a Japanese word with our ears, we naturally remember its pitch accent as well. I am afraid it works even when we learn English words.

Steveさん，「ご自分」は日本語の中でも敬語の使い方になりますが，敬語に関しては日本人でもたいていの人は考えながらしゃべっています。私が思うに，日本人は言葉を口に出す前に，それが失礼でないか無意識にチェックするクセがついていて，もしそうなら適切な言葉を探して言い換えています。言語というより文化の問題でしょうね。

言葉、あるいは言葉使い，敬語も含めて、は正に文化の一部、それに知識だけでは足りない。”なれ”が大切．それには時間と練習が必要です。覚えても、忘れる、単語、文法、と同じだ、と思います．

@Gradius,

That's very interesting. We all hear foreign languages differently because of the properties of our native languages. In English, stressed syllables are usually longer, louder, and slightly higher in pitch. For native English speakers, probably the loudness (or force) of a syllable is the most obvious point, but maybe the slight change in pitch stands out for Japanese speakers. That's very interesting. I'd never thought about that before.

@alexandrec
Not used at all is a bit strong; yes, it's there, but it's by no means the most important aspect of Japanese pronunciation, particularly for so-called standard Japanese. If you read a sentence paying close attention to the pitch of individual words then you won't sound natural. Perhaps you have different goals, but my aim is to sound natural, rather than a textbook definition of what is correct.

As it says in the article I linked to:
'But using such [pitch] markers to speak Japanese creates pronunciations that are worse than a crude synthesizer.'

Ultimately it's how a Japanese native speaker perceives the speech, and if learning pitch in this way works for you then that's great.

Since English words can have different tones/pitches depending on the function in the sentence, the place in the sentence, dialect etc. maybe people who speak a language where tone/pitch is present believe it to be just as important, to convey a different meaning and so on.

Imagine an English teacher reading words off a list. Most likely each word but the last one will have a kind of high-rising terminal (similar to the 2nd tone in Mandarin and maybe the L-H in Japanese). Somewhere I read that it's not uncommon for Chinese/Japanese learners to believe that English words really have tones.

Which reminds me of this thread:
http://www.lingq.com/learn/en/forum/46/14816/

@roan

"If you read a sentence paying close attention to the pitch of individual words then you won't sound natural. Perhaps you have different goals, but my aim is to sound natural, rather than a textbook definition of what is correct."

You won't sound natural the first time, no. But we always have to pay attention to details when we start learning a language or pronouncing things differently or with more minutia, and though we sound unnatural at first, it quickly sounds more natural. I think your argumentation is faulty here. I pay attention to pitch and I don't think I sound unnatural because of it.

"As it says in the article I linked to: 'But using such [pitch] markers to speak Japanese creates pronunciations that are worse than a crude synthesizer.' Ultimately it's how a Japanese native speaker perceives the speech, and if learning pitch in this way works for you then that's great."

I heard that line a few times but it doesn't make sense. Have you ever met anyone who sounds like synthesizer in any language? If you see pitch as a series of do's and mi's, it may sound odd indeed, but we've been saying that intonation also goes up then goes down throughout the sentence -- this natural aspect of the language hasn't evaded any of us. Whoever teaches pitch to their students only to find them sounding like a crude synthesizer hasn't been doing a very good job.

In short, studying pitch hasn't made me sound unnatural or synthesizer-like.

Anyway, it's an interesting subject and I look forward to your next video.

"As it says in the article I linked to:
'But using such [pitch] markers to speak Japanese creates pronunciations that are worse than a crude synthesizer.' "

If you read Japanese deliberately pronouncing high and low according to pitch marking and ignoring all other aspects of intonation, that might happen. But it's hard to imagine someone doing that.

I'm curious to know about what techniques that pronunciation school used with you.

Pitch accent is not important. You shouldn't use it. Tell it to a Japanese person and see what happens.

@okirf What are you suggesting is going to happen?

---

I don't think we are divided on the question of whether pitch accent can be learned, or even that it can be taught (although we haven't demonstrated how that would be done).

However, we disagree on whether it's worth it or not. This implies that we disagree on how much work would be too much work for it to be worth it and what kind of results would warrant the effort. So my question is when and how would it be worth it for you (all) to bother learning pitch?

"So my question is when and how would it be worth it for you (all) to bother learning pitch?"

It's difficult to answer that without knowing how one would go about learning or teaching pitch-accent in Japanese. For myself, I would say that I would have to be told that my pitch-accent is bad, and then I'd have to be shown, or discover, some concrete way to improve it.

But using the word "bother" sort of implies that it's something easy to do and people just aren't bothering to do it.

For the average learner, it's tough to say, but I'd imagine that first they'd need to be convinced that there was something that they could do about it. Of course, you can't address the question of what to do about it until you know that it exists. But even if they know that it exists, what are they supposed to do about it? Perhaps that will be the subject of the future videos.

For some reason this topic is very important to alexandrec, who clearly thinks that pitch accent is a central pillar to speaking correct Japanese. Personally, I cannot see why anybody would think that, although it does certainly exist as a phenomenon and is therefore a part of Japanese linguistics. In Chinese correct tones are required for unambiguous communication, although this is not the case in Japanese.

I find all aspects of Japanese interesting, this included. However, as Botrun says, it is not straightforward, and neither is it clear that the amount of effort is worthwhile, effort could be spent in other areas. Given a a limited amount of time, I would have thought that an investment in learning kanji or vocabulary would pay bigger dividends than learning pitch accent from a dictionary. But that's just my opinion.

The school I went to was one-to-one classes. It has to be really, although the tutor said he did go up to three students. The course covered correct pronunciation of the vowels and consonants and that sort of thing, although a fair bit was listening to examples, saying them back and seeing how the wave form of my voice differed from the wave form of the example and how I should rectify it. The tutor also showed the voice pattern visually using markers to demonstrate the rate of pitch fall in a word or sentence. It was fairly structured and the tutor was excellent.

I enjoyed it and I really wanted to continue, but I haven't had time, and I haven't practised much since it ended. The problem with accent training is that you can't do it just anywhere.

”who clearly thinks that pitch accent is a central pillar to speaking correct Japanese.”

It's certainly a central pillar to speaking correctly like a Japanese person does. However, we had two foreigners named earlier as speaking really natural Japanese with a proper accent. One was Peter Barakan who has lived in Japan for 40 years basically fully immersed. He also works in music, and presumably has good, open ears. I don't know what the other fellow did. That seems to mean that accurate pitch-accent is pretty rare. I'm not convinced that this is entirely because people don't study it.

Look at English. People who learn English in English-speaking countries often pick up stress and natural intonation fairly well. Their vowels and consonants and underlying grammatical constructions are kind of weak (as they're imitating natural, reduced pronunciation without always intuiting what things are being reduced from), but they often get pretty good at stress/intonation - in my experience. People who listen extensively also develop this much more than people who don't. And they rarely deliberately study stress/intonation.

But the same doesn't seem to be true for Japanese. We have people who live in Japan a long time, listen to Japanese extensively, and still do not properly pick up pitch-accent. Is pitch-accent an either/or phenomenon? Is it something you either have or you don't? Or, could we say that someone is pretty good at it, but not like a native? Seeing as we speak in sentences, and not individual words, is pitch-accent something we can even isolate and study in the absence of other elements of intonation?

As for the pronunciation stuff, I get using the wave form examples and whatnot. Those are pretty easy and useful. When I was in school, we used that stuff for phonetic analysis. I always thought it could be handy in pronunciation training. I'm more interested in how they taught pitch-accent and sentence intonation and all those other factors, particularly to a foreigner. Did you feel like the training had a measurable effect on your pronunciation?

One other question to Roan,

You said earlier that you couldn't hear the pitch differences, but then one day the difference between 昨日 and 機能 became obvious. But do you actually hear the pitch difference, or do you just hear the word and recognize which one the speaker is saying? I can hear that these words are different, and they're different when I say them too, but I don't actually consciously register the pitch difference. I just recognize the word. So, at some level, I must be recognizing the pitch pattern even though it's not consciously registering.