You may be familiar with the concept of sentence-level pitch changes in English; for example when you are asking a question, you end the sentence with a rising pitch to indicate that it is indeed a question. Japanese also has sentence-level pitch changes, but more relevantly to this question, it has word-level pitch changes.
In the standard (Tokyo) dialect, word pitch accents are realized by something called a "downstep". The pitch of a word increases until the downstep, at which point the pitch drops.
Downsteps happen strictly between kana (the linguistics term being "mora"), not in the middle of a kana.
These pitches can sometimes be used to distinguish words with the same spelling. The canonical example is はし which is three-way ambiguous between 箸, 橋 and 端.
These words can be disambiguated in speech as follows:
- 箸 (chopsticks) is はꜜし, namely the first kana must be pronounced high to facilitate the following drop
- 橋 (bridge) is はしꜜ, the first kana pronounced low, the second high to facilitate the drop before the following kana (e.g., a particle in the sentence) which would be pronounced low.
- 端 (edge) is はし, or "accentless", meaning that all characters are said at around the same pitch (including any following kana).
"Binary" (LHL) Notation
You may see an alternate "binary" notation for pitch accents which is composed of a series of "L"s and "H"s. For example, you would notate the はし words as follows:
- 箸 (chopsticks) is HL(L)
- 橋 (bridge) is LH(L)
- 端 (edge) is LH(H)
However this notation is a little extraneous when it comes to succinctly marking the pitch accent of the word in the Tokyo dialect, because it always follows the pattern: start low (unless the downstep is right after the first kana), be high until the downstep, then stay low.
When discussing other dialects, sometimes more than just downsteps are required to analyze what is going on.
To find pitch accents for Japanese words, I'd say the best online resource is 大辞林.
For example, the 大辞林 entry for 箸 says はし with a subscript "1" to its right.
It uses yet another notation, where the number works as follows:
- if the number is 0, it is an "accentless" word, i.e., does not have a downstep
- if there is no number, the downstep is placed after the last kana
- if it is any other number, the downstep is placed after that kana (hence 1 for 箸)
There is a nice picture relating this number notation to the binary notation provided by 三省堂, along with a number of example words.
In the end, I'd say the downstep notation is the most succinct and expository notation when trying to notate standard Japanese pitch accent, but learning these other notations is useful to understand material about pitch accent.