I'm not sure how many people actually do this or not, but I'm trying really hard to get more natural Japanese intonations so I listen really closely to the voice example.

When I reproduce this to my native Japanese boyfriend or when he hears the voice example, he tells me that the accent is weird. I was wondering if anyone could tell me what part of the country the voice actress is from?

Being a native Japanese myself, I noticed that too when I was playing around with Japanese Skritter (I actually signed up for learning Chinese). It's not that she has an accent, but the pitch, which plays a very important role in Japanese pronunciation, is neglected. The pitch doesn't only affect the pronunciation of the words but also the pronunciation of following particles.

For example, 橋 and 箸 both are written はし in kana. All Skritter does is to look at the kana and uses the same pitch for every word with the same kana. But in fact, 橋 is pronounced は(low)し(high) and 箸 is pronounced は(high)し(low). This is why it sounds weird to natives.

To be honest, I think the users of Japanese Skritter are better off without the audio samples. It's better to have no example than actively learn the wrong pronunciation.

You can find more info on Japanese pitch here:
http://sp.cis.iwate-u.ac.jp/sp/lesson/j/doc/accent.html

P.S. The reason your boyfriend thought she has an accent is that the pitches of certain words differ from region to region.

P.P.S. I'm from Tokyo and have a standard Tokyo accent, for what it's worth.

I had the same experience many times with the pitches in Skritter audio. My boyfriend is from Tokyo with a standard accent, and his pronunciation of the words is quite different from Skritter's in more than a few cases. He also claims not to understand about 5 - 10% of the words when I play them to him due to the pitch being wrong. As such, I gave up with the audio since the only useful thing about having it is not reliable enough to copy!

Textbooks hardly ever teach pitch, presumably because up to intermediate level it doesn't make a lot of difference in people's ability to communicate. But they should teach it because it is the only thing that differentiates a lot of compound words that have exactly the same hiragana. For an English equivalent, imagine someone thinking that 'light housekeeping' and 'lighthouse keeping' had the same pronunciation.

文化中級日本語 has a little pull out vocab booklet that has pitch markings indicated. Perhaps Skritter could eventually include markings like that around the hiragana if it is trying to go into the business of providing pronunciation models?

Also, NHK produces both a dictionary and audio CD set of 'standard' pitches that all TV broadcasters need to learn. Could Skritter just license their audio files?

James

Although I am new to Skritter, I am very much enjoying it. For my Internet based Japanese study, I use this website to supplement the vocabulary and audio sentence drills I do over at iKnow.jp

I have to say, I noticed that the audio over here at Skritter was a little off as well. Although, the audio samples from iKnow.jp are of a very high quality, so I may just be spoiled.

I am _very_ happy with the Skritter service so far, though! Since I'm not using it as my sole study resource, the strange Japanese pronunciation isn't a big deal to me. It would be nice, though, if fixing the audio was somewhere on the road map for the future.

Anyway, keep up the great work Skritter devs! I am very excited to see how Skritter's Japanese support will mature over the years to come.

In my opinion, you can't be spoiled enough in this regard. Only the very best and most accurate audio samples are good enough for studying pronunciation of a foreign language.

Ah, I thought we got a pretty good speaker, but then none of us are natives so it's hard to tell. We did the recording ourselves, finding a native speaker who lived nearby who was willing to sit in front of a microphone for a few hours. I was thinking of doing another batch in the future to expand our sound resources, but it sounds like we'd be better off investing in redoing what we have first with a more standard speaker and being sure to differentiate between different pronunciations of the same kana. I'll keep this forum post in mind when we get to that. Thanks for your guys' input!

And thanks for your kind input, tshack! :) You won't have to wait for years to see some big improvements Japanese side, I've got a handful of things to do in the coming year that will really bring the Japanese side forward. Hope you like them!

Ahh, it's the pitches! Ok I'll keep using my boyfriend for the time being as an example and hope I don't sound too manly :)

Good luck Scott on the improvements!

Scott, the speaker is fine. She doesn't have a non-standard accent (at least you wouldn't notice as long as she only reads single words out loud). You don't have to redo the whole recording, just those of the words that have the wrong pitch now!

By the way, here is a nice dictionary with pitch information:

http://www.excite.co.jp/dictionary/japanese/

If you look closely, you'll find a tiny numeral with the word you look up. These are the pitch types which are described here:

http://sp.cis.iwate-u.ac.jp/sp/lesson/j/doc/accent.html

including any particle that might follow.
Remember: particles only have a high pitch if they follow a word that belongs to type 0.

0: LHHH... like in 「日本語（の）」、「私（が）」、「魚（を）」
1: HLLL...　like in 「箸（に）」、「朝（が）」、「どこ（に）」
2: LHLLL... like in 「橋（を）」、「ありがとう」、「高い」
3: LHHLLL... like in 「先生」、「ひらがな」、「カタカナ」
4: LHHHLLL... like in 「忙しい」
5: LHHHHLLL... etc.

If you want to incorporate pitch typographically it needs to be kept visual so the user can just see what the pitch pattern is without having to learn things like types etc.

The 文化中級日本語 pull-out vocab book puts little 『 　」symbols throughout words to indicate where pitch starts, and where pitch changes up and down. This has the benefits of both being a) instantly understandable and b) incorporated without distracting from the words themselves. Do have a look at it.

If you wanted to give an intern something to do for a few days, it would be good to include a visual representation of pitch for users who are visual learners. (It may sound odd to learn pitch visually, but just think about people who play the piano -- some can play things by ear, others have to play from a score.)

And why not train pitches as you train tones in the Chinese version - they are just as important.

Anyway, very happy this is being taken seriously. It is great how much you guys listen to your userbase.

Ah okay, thanks for the clarification, Schnabelhund. Those links are handy indeed. Hmm, even so unless we can have a very quick way of figuring out which ones need to be corrected, it might be more efficient to start over. It takes very little time to say a word, it would have to take even less time, per word, to identify all those where we need another go. I suppose I could write a script to go through our dictionary and identify all readings with multiple writings and just redo those. That would probably let us keep a good portion of our existing audio set, and the sounds wouldn't have to be manually sorted.

I hadn't thought of that, Jason. I'm a piano player myself and do a lot better learning by sheet music :) Besides this vocab book, are the 『,」 symbols used by others as well to indicate pitch? Ideally we would use whatever is considered standard, the more well known the better. I'm not sure how feasible it would be to do this, but it's certainly something we can look into for the future.

As I've always said, Skritter wouldn't be half the site it is without input from the users!

I wouldn't say there is any particular standard that I have noticed, not least because most books just ignore the issue.

And, in any case, I'd personally prefer it if you just invented your own method that fits the software, rather than copy your strategy from books which have medium-specific constraints, or from excite which has approached categorisation from a programmer's perspective rather than a user's perspective.

Basically, as long as I can see how the word is pronounced without having to put my stylus down or click anything or check anything then I don't really care what it looks like. Perhaps better than the markers in the book I mentioned would be tiny little coloured up-and-down arrows between the hiragana, or a stepped line graph superimposed over the word, or perhaps even literally moving hiragana slightly lower and some slightly higher over a faint baseline... all sorts of fun ways to do it.

But it would be a super feature and one that would definitely see you jump into the world of tomorrow in terms of teaching trends.

James (although I have nothing against the name Jason)

@scott yeah, those symbols are used in IPA to denote intonation. See http://en.wikipedia.org/wiki/International_Phonetic_Alphabet#Suprasegmentals

@James S. the problem is that I'm also James. I think there's another one of us on here who goes by James rather than Jim or Jimmy etc.

James W

I don't accept that the solution to the problem of having more than one James is to unilaterally rename one of them Jason.

What should we do? Revolvers at dawn?

James S

Oh boy, I'm getting excited! :)
The pitch could form an own studying category, much like the tone prompts in Chinese. Skritterers could get rid of the pitch problems that make that "Western" accent many people have.
I totally agree with James Sharp; all the feedback is not going to waste here!

@schabelhund I study Chinese, not Japanese, and have found dictation practice very helpful. That is, you hear a recording and have to determine the tones and pinyin. I study them both at the same time, but I've often thought it would have been better if I had separated out the tones so I could focus on them better.

One of the most impressive language learners I've run across is an Italian named Luca who has learned quite a few languages; in the languages I can give an informed opinion on, which are English and Spanish, his accent is impeccable. He has written about his technique and he heavily stresses the importance of prosody and intonantion - in all languages, not just in Chinese - for those who want to develop a native-like accent:

http://thepolyglotdream.com/

@JamesS Agreed, "Jason" is simply unacceptable. I have two good friends who always go by Jimmy. If someone mixes us up, although I laugh it off, I am secretly enraged and put the person on my bad list.

James W

@jww1066 Yeah, we have dictations every week in our Chinese class. We have to write down the hanzi and pinyin with tones. I find it very helpful, too.

I guess I should check out your link; thanks!

Hi, I've just discovered Skritter, and wanted to use it for two purposes:
First, learn to write kanji correctly and rapidly, as I already know them but need more writing practice.
Secondly, learn all the JLPT 1 vocabulary, and correctly pronounce them !

So I've really been amazed when I heard the strange pronunciation ! I thought that I eventually couldn't use Skritter, but if you really change the recordings to correct accents, I think I will go on.
Maybe it's too much, but it would be a good thing if you could record all the kanji prounciations, and make them available by a click.

By the way, do you plan to make JLPT 1 list ?

JLPT 1 is on my list; there were basically a bunch of characters in the list that didn't have words to go with them for study, and I wanted to figure out how to do that first. Now I've got the method, I just need to carry it out. Shouldn't take too long!

Great !
I can use JLPT 2 list to get used to Skritter, then the JLPT 1 list will be out, that's perfect !

Skritter would really be the perfect app if the pitches were correct (even if utterance isn't your priority, I think that pronouncing words helps to remember them - anyway as your app can be a means to learn pitches, it would be a waste not to use it for that).
But serious as you seem to be, I guess it's just a question of time, and that correct pitches will soon be recorded.

I wouldn't say soon. But it's on the list. When it will happen is anyone's guess; as a developer it's pretty difficult to estimate when something will actually get done. We've always got more to do, so it's a question of priorities.

I'm responding to say two things.

First, I'm all for any changes made to Skritter to incorporate Japanese pitch. I've been studying for a couple of years now, and even though I'm enrolled in a night class, I can only say that I'm aware there are pitch differences but have no idea what they are.

Second, my name is Jason, which I agree is unacceptable, but I'd also add that renaming one of the numerous Jameses to Jason would not eliminate any confusion. I would suggest one of you take the name, "Slartibartfast."