Writing Japanese with the Latin alphabet is called Romanization. Japanese people use kana and kanji to write Japanese, and they aren't always able to write Romanized Japanese. Romanization is used mainly for non-Japanese. Vowels in Romanized Japanese are based on Latin (Spanish) pronunciation, while consonants are mostly based on English pronunciation. English speakers tend to mispronounce vowels when they read Romanized Japanese.
There are two significant ways of Romanization. One is Kunrei, and the other is Hepburn. The former is established as an international standard (ISO-3602) because it reflects the Japanese sound system well, but the latter is much more commonly used, and all Japanese words found in English are written with it because it's easier for English speakers to read even though it doesn't reflect the Japanese sound system. The hiragana し, for example, is written as "si" in Kunrei Romanization and "shi" in Hepburn Romanization. Its pronunciation being very similar to the English word she, "shi" is easier for English speakers to pronounce it correctly than "si", which might be pronounced like sea. However, writing it as "si" is better to understand how native Japanese speakers treat it because native Japanese speakers categorise it as a syllable with "s" like "sa" and consider it to be different from syllables with "sh" such as "sha". This irregularity comes from a historical pronunciation shift.
I use Kunrei Romanization as well as kana because it helps you to understand Japanese grammar, and I will also list Hepburn Romanization in parentheses when it is different.
Phonemes are the smallest units of the sound system of a language. Actual sounds produced by phonemes are called phones. Phones can vary with dialects and speakers, but the set of phonemes stays the same. For example, the English phoneme /t/ has the following phones:
A letter enclosed by slashes is a phoneme, and one enclosed by brackets is a phone. Even though [ th ], [ t ], and [ ɾ ] are different sounds, they are the same phoneme because English speakers treat them as the same sound unit.
Phone Word Pronunciation [ th ] time [ thaım ] [ t ] style [ staıl ] [ ɾ ] letter [ lɛɾə ]
Understanding phonemes is not very important when you learn Japanese, as long as you can understand Romanized Japanese, because Latin letters well reflect underlying Japanese grammar. The exceptions are the phonemes /N/, /Q/, and /H/, which are not consistently written in Romanization.
1.2.3. Morae and kana
Consonants must be followed by vowels in Japanese. The exceptions are semivowels, which can be inserted between a consonant and a vowel, and the two particular phonemes /N/ and /Q/. This strict rule makes the Japanese syllable structure very simple.
Japanese pronunciation always has either one of the following structures:
The most important rule is that Japanese pronunciation is metronomic - any of the five structures above has the same length of time. Linguists call this minimum beat of pronunciation mora. Morae (plural of mora) are close to syllables, but they are slightly different. A syllable has one and only one sonority peak (a vowel or a vowel-like sound), while /N/ and /Q/ have no vowel-like sound. Perhaps you know Latin also has morae and syllables. Since syllables are not relevant at all in Japanese, some people call Japanese morae syllables to make it easier to understand. I use morae in this site.
- a vowel (including /H/)
- a consonant + a vowel
- a consonant + a semivowel + a vowel
One kana (hiragana or katakana) stands for one mora, not one phoneme. You cannot describe a consonant that is not followed by a vowel, but you don't need it, because Japanese doesn't have such pronunciation.
There are three kinds of accent in world languages: stress accents, pitch accents, and tones. Here accent doesn't mean different varieties of pronunciation, such as in "His English has a Texan accent," but it means a way to distinguish words other than consonants and vowels. English has stress accents, where the strong voice determines accents. For instance, the words subject in "the subject" and in "to subject" have different accents while they have the same consonants and vowels. Japanese has pitch accents, where the high tone of voice determines accents. The strength of voice doesn't matter in Japanese to differentiate words. Chinese has tones, where every syllable has either one of the four tones. Unlike Chinese, wrong accents don't make much trouble in Japanese, so you can skip this section if you want to master kana first.
Let's think about the Japanese word kudamono (fruit) for example. It consists of four morae: "ku", "da", "mo", and "no". Each mora must have either a low pitch or a high pitch because Japanese has pitch accents. In this case, the second mora "da" has a high pitch, and the others have low pitches.
The most important thing for Japanese accents is a boundary between a high-pitch mora and a low-pitch one. For the word kudamono, the boundary between the morae "da" and "mo" is important because the former has a high pitch and the latter has a low pitch. The boundary is called an accent fall, which means a transition from a high-pitch mora to a low-pitch one. There is an important rule: a word has at most one accent fall, and the pitch never rises again in a word if it becomes low. You can determine each mora's pitch in a word if you know where the accent fall is.
In this course, I use the bold face for accent falls. For instance, the accent of the word kudamono will be written as "kudamono" or LHLL, which means its four morae have low, high, low, and low pitches respectively. Note that native Japanese speakers don't always understand this way of notation, because it is highly grammatical. As you know, most people don't understand the grammar of their language.
Having at most one accent fall like a simple word, a compound word has a distinctly different pitch from a mere combination of the pitches of its root words. The accent fall of the last word often remains intact. It is opposite from English, in which the first stress in a compound word is often kept, such as blackboard and darkroom.
1.2.5. Accent falls and pitches
As I have explained, a Japanese word has at most one accent fall. If a word is followed by a postposition, the word's accent fall also affects the pitch of the postposition. Let me give some four-mora words to show accent falls and pitches. I use the postposition ga here, which will be explained later.
As you have learned, the accent fall is a transition from high pitch to low pitch. If the first mora has an accent fall, it has high pitch and the rest have low pitch like this:
Accent fall position No postposition With ga Meaning 1 mi do ri
H L L
1 2 3
mi do ri ga
H L L L
1 2 3 4
green 2 i to ko
L H L
1 2 3
i to ko ga
L [H L
1 2 3 4
cousin 2 ka ta na
L H H
1 2 3
ka ta na ga
L H H L
1 2 3 4
Japanese sword 2 ko do mo
L H H
1 2 3
ko do mo ga
L H H H
1 2 3 4
If the nth mora has an accent fall, the second mora through the nth mora have high pitch and the rest have low pitch like this:
H L ... L 1 2 last
If there is no accent fall, all morae including postpositions except for the first mora have high pitch like this:
L H ... H L ... L 1 2 n n+1 last
The accent fall rule shown here is advanced grammar, so you don't have to memorise it now.
L H ... H 1 2 last