This page was last updated on June 25th, 2019(UTC) and it is currently August 10th, 2020(UTC).
That means this page is 1 years, 47 days, 2 hours, 2 minutes and 13 seconds old. Please keep that in mind.

Ok, firstly, this is a work in progress, and also I have no certifications or other credentials that people would recommend before teaching. However, this is based on my experience and research into the topic of phonology more at the word level. Please take this with a grain of salt. Also, if you don't know "hiragana," learn that first or you won't even remotely be able to follow this page. Also note, I'm not a good writer, and this is incredibly disorganized. To be fair to myself, I'm not entirely sure that this information even can be organized properly.

Firstly, it needs to be mentioned that Proto-Japanese, like most East-Asian languages was made up of words that were either 1 mora or 2 morae in length. Now, a mora is considered a time unit, but for simplification people refer to them as syllables, which is not accurate, and thus creates huge problems in understanding. If you consider english, the word "banana" is 3 syllables, but it would actually be 4 morae. The first "syllable" is one mora, the second syllable is "heavy" being made of 2 morae, 'cause we slighly elongate the vowel to emphasize the accented part, and the third syllable is also 1 mora. With the exception of combining small letters (except the small つ), each of the 仮名 represents one mora. Now, knowing that, understand that records regarding etymology were not kept nearly as well as English, so you'll find lots of explanations and theories for the origin of words (like with ネコ, which some say is 寝子獣[ねこま], others say 鼠神[ねこま], or even 鼠熊[ねくま]: source here).

Secondly, and for reasons unknown to me, standard Japanese (東京弁 aka 標準語) doesn't like to make the highest pitch 仮名 (also called the 核 or "core," which I'll call it from now on) on the ends of "heavy syllables." This is why you might see some discrepencies between dialects like 関西弁 and standard japanese where words like 日本人 have different cores (in 関西弁, the core is ん, the 3rd mora, while in standard japanese, it's ほ, which is the second mora). This, as a result, has created some scenarios where, I believe, professional linguists trying to study this end up thinking that there's no morphology behind it at all, or come up with incredibly complex explanations for what is going on. However, I noticed alot of these get alot simpler if you use 関西弁 which is not afraid to place the core on the end of a heavy syllable, and, given that largely the center of Japan was 京都市 (look up the individual letters if you doubt me) for most of Japanese history, it seems more reasonable and pragmatic to learn 関西弁 and learn how to translate that into 標準語.

While it's easy to just look at Japanese accenting and say that all words are lexically (meaning, they learned the accent upon hearing the word) determined, that wouldn't leave any room for the complex constructions we see, let alone how in the world Japanese people would be able to learn the pronunciation of words simply by reading them and not hearing them, or even begin to explain why speakers can change how they accent the same word as they tell a story. I will admit that for most people, most words are entirely lexical in their accent, however there is plenty of reason to believe that it's morphological (out of grammatical necessity). I can't even begin to address the 1 and 2 letter 訓読み words in a morphological sense, there is a somewhat emergent pattern when analyzing them. One of the problems, however, is that you can't look strictly at standard japanese, 'cause few actually speak it (even though everyone says they do, reality is a bit different).

For the uninitiated, Japanese has a pitch-accent system. This is in contrast to a stress-accent system, like english. The main difference being, that it's precisely what makes english sound like bombs going off when we speak: in our words, the cores are said not only at a higher pitch, but also more crisp and far more loudly. Japanese people are said to only raise the pitch, but not the loudness. I would argue, however, that japanese people aren't necessarily immune to this in their own language, but rather it's hardly noticeable. Take, for example, the word 現れる (あらわれる). Most people from english speaking countries end up finding this word a royal pain to say, but Japanese people can't conceive of why this is even challenging for anyone to say. The reason being, if we go back to the fact that japanese words are merely compounds of their ancient proto-Japanese words, or they're loan words (at which point Japanese subconsciously try to parse them as if they were proto-Japanese), and if we also learn some of the dirty little secrets of how Japanese grammar is obscured for the sake of "making things easier," we can ascertain that 現れる is actually 現れ (the actual verb, since 一段動詞 require a suffix to be verbs) + る (suffix), and the way to say it is あら + われ + る, then magically the word becomes as easy for the Japanese learner as it is for Japanese people to say, simply by seeing it that way (what we're actually doing is making あ and わ as if they were their own words). Furthermore, if you realize that most cases (especially with 五段動詞) the little つ is actually there to create a pause to preserve the "footing," you realize that all this talk about Japanese not doing footing (which results in sub-accenting) is completely wrong. In fact, Japanese people, while they won't foot some words the same as the next guy, they will indeed say each foot as if it is an individual word (which shouldn't be a surprise, but it's taught to most people learning Japanese that this doesn't occur). It's certainly not as strong as stress accent languages, but we learn very quickly that we can't say feet that are more than 2 or 3 mora (or is it syllables? I really am not sure, and finding out isn't really easy without inventing fake words) long.

So, why bother? Well, it turns out that accenting actually has a practical purpose, and not the usual stated one of simply being to distinguish between to homophones, even though it can be useful for that as well. Basically, it allows us to audibly try to break down complex words into their original components. Japanese is special, in that a word doesn't even have to have a core. In 関西弁, you can have a word that is "low flat" (雨雲) or "high flat," (友達), where the word's pitch either rises slightly at the end (for low) or slightly drops at the end (for high), but is relatively stable. Meanwhile, in standard japanese, the first syllable (yes, not mora, so if the first syllable is heavy, both morae will be low pitch) and the rest are high, until the final mora which will drop slightly like with 関西弁's high flat words. This phenomena also passes on to words that are "not flat," that have a core: the same patterns as flat occurs, except the core will be slightly higher than the average high tone, only to be met with a sharp drop. Now when asking Japanese people (and recording them) from 関西, low flat words are sometimes (though this varies per speaker and per mood of the speaker) I sometimes hear what sounds like the 2nd mora being slightly higher than the first, followed by the 3rd which is as low as the first, with the 4th (or ultimate) being of a much higher pitch (the slight rise at the end). I don't know if this is some sort of unconscious training by the mind to react just in case they want to turn it into an ante-penultimate accent (which would make that mora the high core), I don't know (more on that below). Of course, in addition to all that, it stands out to native speakers like a sore thumb if you can't say a word in a possible accent configuration.

The general pattern seems to be, that if you have 2 high-flat words, you stick them together, and they make another high flat combination. Now, if you have an accented word (山, for example) and an unaccented word (道), the core will fall on the border line between the two words, picking the more prominant word (most likely being the accented word, like with ま of 山道 being the accented core). However, if a word with an accent (let's use 中国, where う is the core in 関西弁) is mixed with an unaccented word (let's use 語), and the unaccented word is the emphasis, the word will become a high flat (中国語). By the way, low-flat accent in 関西弁 seems to manifest as a form of accented, despite creating low-flat constructs when combined, which seems to be closely tied to the fact that any accented word will usually become low-flat in compounds (presumably this process would make it easier to keep the core on word boundaries). This creates the interesting scenario where 雨(めis core in 関西弁 and あ is the core in 標準語) becomes low-flat for the compound (since that's what accented words become at the beginning of a compound) and 雲 (く is core) becomming 雨雲(low flat). As you can see, alot of information ends up lost in standard japanese ("flat" sounds like low-flat, but is ultimate just high flat with an initial low tone on the first syllable), which creates this illusion that there's no way to predict this stuff. However, there are exceptions to this, although very rare: 玉ねぎ (my best guess is that it's like those scenarios where the core disappears [even though it doesn't, here, but rather remains unchanged], since both this case and that case are rather rare).

Lastly, let's take a look at a practical example of how this confusion can be avoided by looking at it first in 関西弁 before switching to standard Japanese. 貧乏's core is the second mora (ん). Now, since standard Japanese doesn't like that second mora of a heavy syllable holding the accent, it becomes the first, instead. Natural speakers will often elongate this to the first syllable as a whole, but it's important to note that this is not necessarily so. This isn't much of a problem, since the word is a compound of two words that are not normally said alone, anyway. However, the morphology becomes really important if we add a suffix, like 神 to it. Now, the way this should work is that you move the core to う, which is the 4th mora, so that it is perceived as 貧乏 + 神 instead of 貧 + 乏神. Once again, standard japanese has a problem, 'cause ぼう is a heavy syllable. Turns out, depending on the speaker, you'll either hear them make ぼ the core, or you'll hear が be the core. NHK's accent dictionary says that both are appropriate. Naturally, both are really unnatural 'cause they're violating the function of accenting to help you tear words apart back into their original words. As a result, suddenly びんぼがみ, with ぼ being the core, becomes the common pronunciation for most of Japan. The same issue happens with 貧乏人: びんぼ人. Yet, at the same time, this shortening of 乏 is counter to the Japanese rules of pronunciation. Just google compounds of 貧乏+suffix as well as びんぼ+suffix and just compare the search result numbers (obviously japan prefers to type properly, but the fact that the numbers are always at or above 50% should be telling).

If you google "japanese accent preserving suffix" you'll likely find a PDF (My guess is that this will likely not work at some point) that makes this complex case, but then if you look at it from 関西弁, suddenly it's not so mysterious. On page 9, one would think that the examples marked "(22)" are confusing it. However, 助詞 is high-flat, i think in 関西弁, while 動詞, 名詞, and 副詞 i have recordings showing to be low-flat. I assume 数詞 and 冠詞 are as well, so then when you see 前置詞 in the next example that seems confusing, you can clearly put together that rather than seeing it as 前置+詞, you should instead be seeing it as 前+置詞. This is further clearer with 助動詞, where even in 関西弁 the accent is on ど, not う, as one would imagine, which is why the whole construction seems like there's something magical going on. However, 接続詞 was also mentioned as an example, which should've given it away (then again, he marked the core to be ぞ, when NHK marks it as く [to be fair, the く ends up devoiced, so the core must move left]); the reason for this accent, by the way, is that 接続+詞 is the construction. Presumably, we can then conclude that 詞 is accented, and that if you have 3 morphemes, low flat words need to have a core so that they can be dissected properly.

Get your own web kitty here!
©Copyright 2010-2020. All rights reserved.