CureDolly Wrote:One thing that worries me about not doing some specific vocal study is that the mind has a horrible tendency to post-process what it hears into what it finds more familiar. This is especially a problem with mora-timing because European-language ears naturally parse what they hear into syllables rather than morae.

I feel that's not entirely accurate. I come from a syllable-timed language and as expected the only real problem mora-timing in Japanese induces is with long vowels, and even if the scale of this phenomenon is in the microseconds, it's something you can train to ear if you know it exists. But imho the biggest problem in listenning Japanese is what another user called the "slurring" and what yudandaiteki partly identified as the "whispered mora". Basically Japanese may sound like "ksttr mondnda ksawan" but natives don't have any problem with it because they rely on the prosodic curve (suprasegmental) to decypher it (among other things). If anything this phenomenon is more akin to stress-timing than mora-timing.

CureDolly Wrote:One thing that worries me about not doing some specific vocal study is that the mind has a horrible tendency to post-process what it hears into what it finds more familiar.

This is very true and why relying solely on "picking it up" as an overall strategy is a gamble at best. It's not uncommon to hear fluent foreigners with accents worse than beginners (though there are many other factors behind that).

One thing I couldn't hear or reproduce even after years of listening was pitch accent. I got frustrated when I couldn't say ヤギ after about 10 attempts last year and decided to do something about it. A mixture of shadowing sentences with accent codes and occasional help from my g/f has made a big difference, and I feel I'm reliably able to reproduce a given pattern and have mastered at least a couple of thousand words. The problem is all the other thousands of words I've learned without the accents...

Erati Wrote:I come from a syllable-timed language and as expected the only real problem mora-timing in Japanese induces is with long vowels, and even if the scale of this phenomenon is in the microseconds, it's something you can train to ear if you know it exists.

I found that ん was more of a mora-timing issue for me, e.g. making 先 in 先生 into a single-syllable "sen". Since "if you know it exists" can be a big if, I didn't realise until somebody pointed it out.

CureDolly Wrote:すごいね！

I really like your way of shadowing, Helena-san. When I read that I feel pretty convinced that you have developed a very good accent. I don't think it is a method everyone can use. We all work differently. I find shadowing very difficult and couldn't manage it at all until I discovered a particular method and then developed it into something I call "harmonizing". Frankly, compared to your work it is very wooden, but it's the best this poor doll can manage. I mean if I had been made a century ago I probably would have been wooden myself.

I listen to native audio and watch native material a lot. In fact I don't use any non-Japanese media at all. Not a 我慢 thing. That's what I enjoy. I never found much that I really liked that wasn't Japanese even back when I was stuck with English subtitles. I also read Japanese books and play Japanese visual novels. My Japanese 3DS won't even play non-Japanese games. I don't mind region-locking so long as I'm locked into the right region!

I don't sing though. I should, shouldn't I?

One thing that worries me about not doing some specific vocal study is that the mind has a horrible tendency to post-process what it hears into what it finds more familiar. This is especially a problem with mora-timing because European-language ears naturally parse what they hear into syllables rather than morae.

Having said that, I believe you are throwing yourself into the spirit of Japanese in a way that means you are probably catching the rhythm naturally and getting it into your bloodstream. I absolutely believe that organic is best wherever you can do it. 羨ましい.

I am thinking about looking for a pretend Japanese parent to shadow!

Thanks. I'm really not perfect at shadowing. I may miss some chunks of what is being said but the main thing as I keep saying is to never worry. You can listen again and try again if you want, and with my pretend parent, I sometimes shadow the same videos 10 times or more. But you're right we do all work differently. What exactly is this "harmonising" technique?

I gathered that you were using a lot of native material, my comment about native material early on was more trying to diagnose yudantaiteki's problem. I would say the same thing about the 3ds region locking, I really want to pick up a Japanese model. Transferring all that reading in games and books into reading aloud will probably help you. I pretty much always read Japanese aloud.

In terms of post-processing.... Firstly, I don't know what mora timing is. As I said, I never tried to focus on what was different about Japanese, just how to make my speech the same as Japanese. I do find that Japanese has it's own rhythm, that's what I try and replicate. I think post processing should slowly stop happening if you continue to surround yourself with Japanese audio and if you do regular shadowing/reading aloud/singing/rapping practice, because the Japanese rhythm will also become familiar. If I do post process something, then the word will come out Anglicised when I say it and because I am listening for the correct Japanese sound, I recognise that problem and try and repeat the word until it sounds Japanese.

jimeux Wrote:This is very true and why relying solely on "picking it up" as an overall strategy is a gamble at best. It's not uncommon to hear fluent foreigners with accents worse than beginners (though there are many other factors behind that).

If you are talking about the idea that if you study without trying to get good pronounciation at all now but reckon that when you go to Japan you'll pick it up, yes, you could call that a gamble. However, if you are talking about my suggestion of practising in a more natural non-technical way then I could not disagree more. It is anything but a gamble. I am making a concious effort to improve my pronunciation. Saying that is a gamble is like saying that playing guitar songs every day and mastering them with the aim of learning to play guitar well is a gamble compared to reading about and practising your scales every day until you know every scale. The former gets you real results that you can perform with, the latter a string of perfectly technical sounds that you will struggle to put together naturally. There are many famous musicians who couldn't read music. Those people who enjoy music theory are very rarely celebrated. Learning to pronounce by practising intensively but not technically is completely legitimate and requires effort, and it will reap results.

In terms of fluent foreigners with accents worse than beginners, I agree that there are many many many more factors behind that. One is that those from poorer countries like India or countries that force their children abroad like China, sometimes are learning English out of duty and therefore do not really want to master it, so are quite happy when they can simply communicate, no matter how bad their accent is.

Those communities are also quite insular, and will speak their own language whenever and wherever they can and often even import their own TV programs, so although they need to function in English and therefore achieve some fluency, since they are not doing what I am saying, i.e. surrounding themselves in English so they get used to it and stop post-processing it, and actually working on their accent, they are not going to improve their accent. (I go to a school that is pretty much half Chinese people, and unless they have something to say to you directly, they literally speak right across you in Chinese all the time, yet they all appear fluent when speaking English. Needless to say none of them have impressive accents though.)

Another problem is, that while we learning Japanese in English-speaking countries are often forced to do a lot of work ourselves, and therefore must work with native media and diagnose I own accents, those people in other countries are all taught English in a classroom and rely on their teachers. Many of them will not have native teachers and so will actually directly pick up an accent from their teacher. And if they do have access to native speakers, they may be doing a lot of speaking before they have time to get used to the flow of the language or practise their accent and so will get an accent ingrained in their English early on. Same goes for people who plant themselves in the country with only basic skills.

Lastly, English people are quite used to hearing badly accented English, and non-native speakers will also pick up on how little a perfect accent matters or at least will never encounter great problems because of their accent so will be relatively oblivious of it (I've had a couple of foreign friends who were convinced they didn't have an accent; they're English was usually the best out of all these groups I've mentioned, and their accent was not thick but very noticeable), so even if they are keen on learning English well, the need in their mind to improve their accent disappears.

That was waaaaaay too long.

CureDolly Wrote:That is, the unvoiced plosives - T, K and P - in English are pronounced with aspiration (described as a small puff of air). As I understand it this is accompanied by a fractional delay between the plosive and the following vowel.

In practice that means that Japanese people hear (and transliterate), for example, English "ka" as キャ. So Candy→キャンディー.

I'm fairly certain that aspiration isn't the reason for this transliteration - if it were, then "gap" wouldn't be rendered as ギャップ. I believe this has more to do with the vowel rather then the consonant - the "a" in "candy" represents a more fronted vowel than the "a" in "father", and this might make the preceding consonant sound more palatalized to speakers of some languages. It certainly does for us Russian speakers, since the sound [æ] in Russian exists only as an allophone of /a/ after [j] and palatalized ("soft") consonants. The Japanese also transliterate German words with ü preceded by a consonant using palatalized consonants (Zürich - チューリヒ, München - ミュンヘン), even though palatalisation isn't a source of phonemic contrast in German.

All of that is to say that, while weakening the aspiration of your plosives is important for developing native-like pronunciation, it's not really crucial for basic phonemic contrasts. As long as you keep your vowels clear and make a noticeable distinction between voiced and voiceless consonants the amount of aspiration shouldn't be too much of an impediment for intelligibility.

jimeux Wrote:I found that ん was more of a mora-timing issue for me, e.g. making 先 in 先生 into a single-syllable "sen". Since "if you know it exists" can be a big if, I didn't realise until somebody pointed it out.

Yes. Since ん is a hybrid entity I guess I was reluctant to treat it like a purely mora related issue. For example in words like 観光 I find more practical to focus on the ŋ sound (with indirect mora consequences), but for words like 先生 focusing on the length is probably a better approach.

@Helena4

My comment wasn't about your approach, which seems much more involved and reasonable than "just pick it up". I'm not sure about the guitar analogy though. I'm not suggesting people learn phonology (scales and music theory). I just think it's better to have some basic knowledge about sounds and practise them a little to gain some awareness (watch a YouTube video about pinch harmonics and squeal some out). Unless you have a knack for discovering guitar techniques yourself, I think initial awareness is actually a more natural approach for music.

I was alluding to advice I've heard from others about not actively studying pronunciation at all, but simply listening a lot and learning it completely passively. Similar arguments are made for grammar, too. Some people seem to get results like that, but you never know what you're missing with such an approach, which is where the gambling aspect comes in.

Personally, I generally find I'm not one of those who gets results without developing awareness first. Pitch is a painful example of this. Awareness can be as easy as just reading a short paragraph about how a sound or grammar structure is formed, so I've never seen it as a significant trade-off.

I have confidence in an approach like Helena-san's, because it really involves getting Japanese into the blood. "Just listen a lot (while still fundamentally living and thinking in English)" is not likely to work unless one has a kind of linguistic "perfect pitch".

But I still take Jimeux-san's point. There are some things most of us can't hear until we are aware of them. And a few things we can't hear even when we are aware of them and try very hard to hear them. In my case unvoiced plosives in particular.

But things one can hear if one listens with an informed ear, like the moraic value of ん or of unvoiced morae, in my case have been really worth learning. Jimeux-san no iutoori: it doesn't have to be an in-depth phonological study, just reading a paragraph can make a big difference here. I might have gone through life thinking 先生 was two syllables and pronouncing it as such however often I heard it. But then my ear is far from perfect.

@Helena-san: Out of interest, can you hear the difference between the K-sound in か and the K-sound in cap/cup? Can you hear the difference between the d-sound in ど and that in door? (apparently they are different: the English one is not fully voiced and the Japanese is, though I am not sure how far Japanese people can hear this. They can hear the aspirated plosives, while we tend not to be able to hear their unaspirated ones). Do you think it matters? Not a rhetorical question. If one's timing and general pronunciation are near-perfect, a slight cute foreign "lisp" may not be a bad thing.

@John-san. I have recorded my voice on other occasions, though not for this specific purpose. The problem here is that I can't hear the difference even when Japanese speakers are saying them correctly, so I don't see how it could help in this particular case.

EDIT: Sorry, I forgot to reply about "harmonizing", Helena-san. It is pretty 情けない compared to what you do. I am planning to write an article on it though as some people have the same problems as me. First let me confess that I am not ball-jointed. My body is fairly stiff and my personality tends to reflect that, being stiff, shy and awkward, especially around 人間. I find "live" shadowing very difficult and had more or less given up on it.

Then someone suggested using anki recorded sentences to shadow and I built up a technique based around this in which I "harmonize" with a female speaker (it should be a speaker of one's own gender) on one sentence at a time. Often many times. I concentrate on various aspects, particularly mora-timing. I try to make sure I am aware of each mora. I work on, well, whatever comes up in the sentence.

At first a lot of Japanese sentences seem like 早口言葉 and one learns to get one's tongue around them in exact pace with the speaker. I find using the Amenbo no uta (a Japanese voice training exercise. I wrote an article on this too) first helpful. If one can do Amenbo at full speed it makes fast Japanese speech shadowing psychologically slow down (among other benefits).

The whole subjective speed thing is really interesting. At times one realizes "You are taking this sentence too fast, slow down". And actually that is nonsense because you are doing it at the same speed as the speaker on the "fast" and "slow" attempts, but it is something about relaxing and letting the morae flow properly. I am sure the phonologists here might be able to analyze what is really happening, but the important thing for me is that it gets subtle, hard-to-conceptualize elements of timing into the blood.

The important thing really, though is harmony. When you have the timing exactly right it "chimes" and you have a lovely voice combination. It helps to get Japanese with its sounds and rhythms into one's heart. But it is only a poor shadow of what you do with all that life and human emotion.

@CureDolly

I like your praise of my approach, thanks.

Yeah I can hear the difference definitely, and mixing up those sounds is definitely something that would trigger me to realise I am saying a word wrong. I'm speaking as a Brit here, so I'm not sure how it compares to American, but I find the roof of the mouth sounds in Japanese more muddy in Japanese and more sharp in English, and that the か sound is just further back in my mouth, making it also sort of muddy cos it's sort of coming from my throat. As I said, I'm not technical at all so all I can say is that I find it "muddy"... not really a proper description, right?

Don't knock your technique, it sounds good! Putting it in anki means you get a lot of repetition like I said and you don't miss anything! It's really hard not to miss a lot when shadowing like I am, and I definitely miss bits (try to get them the second time around).

"Muddy" interestingly is exactly how Japanese describes voiced consonants (the ones made by giving them a tenten). In fact that tenten is more technically called a 濁点 or muddy-mark. 濁り, which uses the same kanji means both literally "muddy" and consonant-voicing no koto.

I think you are using the term muddy in a very similar way =less "sharp".

You have a very good ear I think, and you don't need to be technical if you can hear. Technical analysis is a wooden leg for the lame-of-hearing (of whom, I am afraid, I am one). Or more accurately it is like training wheels and should be dropped once one gets to where you already are. After all its aim is to achieve by very artificial means what every Japanese ten-year-old is doing without thinking about it.

To clarify, I do not use my harmonizing deck as "Anki" at all. For this purpose (only) Anki just works as a receptacle for a bank of sentences and a randomizer/one-button replayer. The whole easy/hard, pass/fail mechanism is irrelevant here (except you can use it to send other-gender sentences to the back of the line). For my part I don't even try to "early-retire" the simplest sentences, like はい on its own. They are a good chance to practice that simple word and get it just right.

Further back in the mouth is interesting. I seem to feel GA further back than KA (and if you devoice GA it becomes か). Thank you. I will go with your instinct there. Muddier is interesting too. I find when I try to devoice plosives it is hard to make them clear enough.

For others wrestling with this and related problems, I have discovered that the relevant part of the book Yudantaiteki-san kindly mentioned is available online.

http://yalepress.yale.edu/yupbooks/langu..._intro.pdf

From this (among quite a bit of other brief but valuable information in the notes to each basic sound) it seems that Japanese T-sound is indeed made with the tongue against the teeth (which would make it a little more TH-like and "muddier", I would think).

PS: playing around with T on the teeth, what I think I am finding is that it is very TH-like if you aspirate it. The secret to T on the teeth not sounding at all like TH is to keep it very puff-free. I think this is a very important lead on how to get た、て、と correct.

が is muddier and further back than か yes . か is just further back and muddier than the English ca/cu-type sounds. And as for what I said about Japanese being muddier than English, the roof of the mouth sounds and が are definitely muddier to me, but in terms of other sounds it's sort of complicated. For instance か seems muddier, but it is more "a" sound that produces that muddiness than the crack of the "k". The crack of the "k" is still sharp but a bit further back.

Thank you so much Helena-san. Your non-technical analysis is really helpful.

All right, here is a Dolly Method in pre-beta that I am submitting to all you wise folks for peer review.

Once you know that た、て、と are pronounced with the tongue against the teeth, the whole problem becomes more intuitive and manageable.

You just need to practice た、て、と on the teeth. If it sounds a bit like TH, then you are Anglo-aspirating. If you instinctively try to solve the TH problem by moving the tongue back from the teeth (and I find this is what the mouth "wants" to do at first) you are almost certainly still Anglo-aspirating because you are trying to solve the problem without de-aspirating by moving the tongue.

Just make sure the tongue is on the teeth and work to get a clear T sound. Be aware that the way to get it clear is to avoid all "puff". Also a lighter and quicker touch seems to help.

What do we think, minasama?

PS : if the product of this experiment sounds a little D-like to your ear, provided it isn't actually D it may not be a bad thing. It seems to be an established fact that while pronouncing unvoiced plosives without proper aspiration in English does not inherently affect intelligibility (the difference is not phonemic), it does increase the chances of their being mistaken for their voiced equivalents.