Chinese relies on lexical tone, English relies on stress, Japanese relies on beat count (mora). These are three separate constraint systems, and none can substitute for another. If the tone in Chinese goes wrong, what is heard isa different character; if the stress in English falls wrong, what is heard isa sentence that has been mispronounced； If the mora count in Japanese is wrong or the pitch accent is reversed, what is heard is a different sentence—— 「雨が晴れて虹の橋が架かり」will be heard as「飴が腫れて2時の端が係」(Takayama Hiroshi, *Melody Method, Expanded Edition*, p.192). ★ The meaning of the whole sentence falls apart, while the singer sings it completely correctly.
★ This skill deals with only one thing: how Japanese mora count, pitch accent, and note-mapping mesh with the music.
For the general principles of rhyme, go to lw-rhyme, for theme go to lw-song-intent, for structure go to lw-structure。
Which sections to read depending on the task
| What you want to do | Read |
|---|---|
| How exactly to count the length of a line of lyrics | ★ §1（Count morae, not kana, not syllables） |
| How many beats the sokuon, hatsuon, chōon, and yōon each occupy, and what notes they take | ★ §1.2 |
| It gets heard as a different word when sung | ★ §2 Pitch Accent + §1.2 consecutive vowels and lengthening |
| How kana land on notes / how to handle one character with multiple notes | ★ §3 Note-Mapping |
| Where a phrase should break and where it must not | §4 Phrasing |
| When the character count and the slots do not match, should you force a change | ★ §5 Jiamari / Jitarazu (intentional) + §6 Four Correction Methods (change the melody side） |
| How Japanese rhymes / why writing it out feels "too Japanese" | §7 |
| Write lyrics first or music first | §8 (correspondence with the terminology in lw-workflow §1) |
| Putting English words into a Japanese song | ★ §9（Iron rule） |
Filling in LYR-SPEC's prosody.* |
§10 |
| Actual singing examples, score-level treatment, and NG/OK comparisons for rhyme and English | ★ reference.md |
Boundaries
| Not handled here | Handled where |
|---|---|
| Chinese inverted characters and composing melody according to the characters | ★ lw-tone-check（You must not substitute the pitch-accent approach for it——see §2.4) |
| The thirteen rhyme categories, level and oblique tones, Chinese character counts and caesurae | ★ lw-mandarin |
| Cantonese tone-matching and the nine tones | ★ lw-cantonese |
| Aligning English word stress with the beat | ★ lw-english(this skill only covers the one item in §9:syllable↔note conversion when English enters a Japanese song） |
| General principles of rhyme: rhyme schemes, surprise factor, deliberate non-rhyming | lw-rhyme(this skill only discusseswhy Japanese rhyme is weak, see §7.1) |
| What this song is trying to say | lw-song-intent |
| Section function, where the hook and title go, line count and balance | lw-structure |
| Imagery, concrete nouns, object writing | lw-imagery |
| Person, point of view, time anchors, negative space | lw-narrative |
| How the melody itself is written, where the highest note is set, the number of bars in the form | Composition and arrangement |
| How those words should be sung(breathy voice, dynamics, articulation, vocal direction) | Voice(recording and vocal direction) |
| Harmony stacking, microphone placement | Voice(recording) |
| Vocal recording, pitch correction, mixing | Voice(recording, mixing) |
1. ★ Mora (beats) — the Japanese counting unit
1.1 Why it is neither a "syllable" nor a "kana"
The Japanese counting unit is the mora (beat), and the default correspondence is one beat to one note. The ideal of 符割り is "one sound, one character" (one note, one kana)(Taniguchi Naohisa, Puro no kyokuzukuri ga wakaru hon, p.58).
First, separate the three ways of counting: the same concept, The number of kana in 「フ・リ・ー」 is 3, the mora count is ★ 3(the long vowel 「ー」occupies one beat by itself）， whereas the number of syllables in the English "free" is ★ 1。
★★ Japanese: 3 beats, English: 1 syllable— this difference is singled out in Puro no kyokuzukuri ga wakaru hon, p.62 as the biggest pitfall: "free" cannot be split into "f-ree" and sung as two notes(a melody note cannot be placed where there is no vowel nucleus); whereas the loanword 「フリー」 is 3 morae and can be split. Likewise, 「フリーダム」 is 5 mora, whereas "freedom" is 2 syllables（free-dom）。
★ Consequence：The "length" of a Japanese line = the mora count, and it isalmost always equal to the number of kana you write, so Japanese is easier to count than English—but only if you have counted the special beats according to the rules in §1.2 below。
★★ The first action of this skill, and the cheapest one： Notate the words as a pure rhythm score following the rhythm of reading them aloud (only durations, no pitches). ★ The words themselves have "lyric time"— reading a sentence aloud naturally produces rhythm and intonation; if you cram the words into a melody based only on the character count, the two kinds of time are bound to clash(Senritsuhō Zōhoban, p.190).
First write down the rhythm of the words, then look at the rhythm of the melody, and everything from §2 to §6 below is the work of aligning these two lines.
Conditions under which this fails: 符割り can merge adjacent kana into one note (the 「ありがとう」 and 「こんにちは」 of §3.2),
in which case the number of notes actually sung is less than the mora count。
★ Count prosody.sections[].syllables You must count "the notes actually sung," not the kana on paper.
1.2 ★★ Special beats: how many beats each occupies, what notes to assign
This section is the table in this skill that most needs to be executed mechanically. Senritsuhō Zōhoban, p.193 calls it単語のリズム (the rhythm inherent in a word), and gives a warning:Ignore the rhythm inherent in the words and singing becomes labored and machine-like, and in extreme cases incomprehensible—— "乗って" in "飛行機に乗って" gets sung in a way that sounds like "ひこきに、のて".
★ The full table of how many morae each special mora occupies and which notes it gets is in reference.md §10。
1.3 ★ Counting units in three languages: one table makes the boundaries clear
| Dimension | Chinese (lw-mandarin） |
English (lw-english） |
★ Japanese (this skill) |
|---|---|---|---|
| Counting unit | Character = syllable, one character one sound | ★ Number of stresses, not the total number of syllables | ★ Mora (beat), neither the number of kana nor the number of syllables |
| Layer where problems occur | Syllable layer——a wrong tone is misheard asanother character | Sentence layer——a misplaced stress changeswhat a sentence refers to | ★ Between words——pitch accent flipped, beats miscounted, phrasing misaligned,the whole sentence gets recut into another string of words |
| Constraining mechanism | Each character carries its owntone pattern | A multisyllabic word carries its ownstress position | ★ Each word carries its ownhigh-low pattern (pitch accent)＋ a string ofbeats |
| Can it be determined programmatically | ★ Yes(wrong characters can be computed) | Not fully automatic | ★ Half possible——the beat count is countable, pitch accent can be looked up in a dictionary,placement requires human judgment |
| What can go in weak positions | Neutral-tone characters, function characters | Function words, unstressed syllables | ★ Particles, unaccented kana（→ §8.3） |
| Highest-note landing point | Open vowels, tone pattern does not clash with the melody | Stressed syllable, open vowel | ★ Cannot be a voiceless sound, cannot be the moraic nasal ん（§1.2） |
★★ Three disciplines： ① Do not look up Japanese words with the "wrong character" approach——Japanese does not have one tone pattern per syllable; it isone high-low pattern per word, looking it up with per-character tone patterns will shatter the contour of the whole word → §2. ② Do not count line length for Japanese words using "number of stresses"——Japanese does not have the English-style compression of "a five-syllable word contributing only one stress", Japanese beats are almost equal, counting beats is counting length. ③ For the section in a Japanese song where English is mixed in, run both systems separately——the Japanese half by morae, the English half by syllable count (§9).
2. ★ イントネーション (pitch accent)
2.1 The rule itself
Every word has an inherent pitch and intensity contour. Principle: the melody's rises, falls, and intensity must match the lyrics' pitch accent. (Takayama Hiroshi, Melody Method, Expanded Edition, p.191–192)
Consequence of mismatch: listeners cannot grasp the meaning by ear alone, and the song becomes one where "the words do not get through"; in extreme casesheard as a completely different sentence(the opening "飴が腫れて…").
Which system to base it on: for pitch accent, follow標準語 (standard language)； ★ Note: 標準語 ≠ Tokyo dialect; if unsure, consultアクセント辞典 (pitch accent dictionary)(『旋律法 増補版』p.192).
Self-check action: take one line of words andsayit out loud (not sing), and draw each word's high-low contour as a line graph; then draw the melody's contour as a line graph;compare the two line graphs' rise/fall directions point by point. ★ The places where they go in opposite directions are the risk points—not every one needs to be changed, butyou need to know where each one is。
2.2 Deliberate deviation is a technique, not a mistake
"Deliberately not following the pitch accent" is an expressive device:it creates a "hook of language" that grabs attention and expresses strong emotion； ★ pushing a particle up to a high pitch can push the emotion up step by step(『旋律法 増補版』p.192).
There are three prerequisites; if one is missing, it becomes an accident： ① the pitch accent of the adjacent words is correct(with the surroundings correct, the deviation can be heard as intentional); ② the overall meaning can still be conveyed； ③ ★ syllables that are easily misheard must give way to the melodic climax—— in other words:if you are going to deviate, deviate on the word that will not cause mishearing.
Failure condition: deviating throughout the whole passage → it is not a hook, it is words that cannot get through. Self-check action: count the number of deliberate deviations in a passage. ★ If there are more than two, restore one first before doing anything else.
2.3 The rise and fall of the ending determines what kind of sentence this is
語尾の上行(sentence-final rising) = question, appeal, undecided, hope; 語尾の下行(sentence-final falling) = answer, conclusion, assertion, resignation. (『旋律法 増補版』p.193.)
★ This is the interface between sections: verse ending rises → chorus comes to "answer";
chorus ending falls → this sectionis closed, and piling on more emotion afterward will seem superfluous.
with lw-narrative 's withholding used in succession: if you want to "stay in the discomfort," thendo not let the last sentence fall。
2.4 ★ The boundary with Chinese 倒字 (this cell must be remembered clearly）
| Chinese 倒字 | Japanese pitch accent conflict | |
|---|---|---|
| What the constraint attaches to | ★ attaches to every character—character tone is phonemic | ★ attaches to every word—it is the high-low pattern within a word |
| What happens if wrong | thatone characteris heard as another character | ★ that string of words is re-segmented, and the whole sentence becomes another sentence |
| How to fix | change the character / change the melodic direction / add filler syllables | change the word / adjust the melodic rise and fall /change the segmentation（§4） |
| Which category it belongs to | lw-tone-check ＋ lw-mandarin |
This skill |
★★ must not substitute for each other: If you use the character-by-character Japanese lookup approach,it will break the pitch pattern of the whole word apart, and you will not find the real problem; if you use the whole-word Japanese approach to check Chinese,it will miss inverted characters in individual characters。 For a line mixing Chinese and Japanese, run both approaches over it.
3. ★ Character distribution (assignment of kana to notes)
3.1 Ideal and reality
The ideal is “one sound, one character”—one note, one kana. (《プロの曲作りが分かる本》p.58)
In reality, in recent years it has been popular to "stuff in more characters to create a sense of speed", and the judgment on this trend is very direct: If the lyricist writes this way unconsciously, the recording session will fall into chaos, and the one who suffers afterward is the lyricist.
★ Reading：Stuffing in more characters is fine, but it must be "you know what you are doing" stuffing— that is 字余り in §5, which has a cost, positional choices, and an effect. Unconscious stuffing is not 字余り, it is failing to count the beats.
3.2 ★★ The same melody will automatically change its character distribution when the lyrics change
This is the most valuable point in this section—the same melody “ソラファファ” set to six four-kana words (《プロの曲作りが分かる本》p.59):
- Akikaze = ★ Standard one sound, one character
- Soyokaze = そ→よ is close to diphthongization,よ is absorbed
- Kitakaze = き is read short,the accent moves to た
- Arigatou ＝ ★ とう two kana merge into one sound
- Sayounara ＝ ★ う is omitted(actually sung as さよなら)
- Konnichiwa ＝ ★ こん merge into one sound
★★ Four kana do not equal four sounds. When Japanese sounds are sung, they areelastic: they absorb, merge sounds, omit, and shorten. When writing and fitting lyrics, you must anticipate this elasticity(《プロの曲作りが分かる本》p.59).
Self-check action (must be done): sing each lineat the tempo you want,
counthow many sounds are sung out, not how many kana are on the page.
★ When the two numbers differ,prosody.sections[].syllables fill in the one that is sung,
and in note , write clearly which two kana merged in sound.
3.3 The best insurance: have the lyricist sing it themselves
★ Have the lyricist sing the lyrics in their own voice and record it and send it over. It does not matter if the recording environment is poor or the voice is quiet. Otherwise this lyric will circulate with the wrong character distribution. (《プロの曲作りが分かる本》p.58–59)
★ Applying it to this library: this corresponds to prosody.sections[].notes and melody_contour in those cells—
the intent of the character distribution must be recorded in some form; a text-only lyric draft cannot be passed on.
3.4 One character, multiple sounds and multiple characters, one sound
| Case | How Japanese handles it | Cost |
|---|---|---|
| Multiple characters, one sound(sound merging) | “とう” and “こん” in §3.2—It happens naturally; there is no need to resist it | ★ The number of lines will be fewer than the number of kana,so you must recount |
| One character, multiple sounds(one kana spans several pitches) | ★ This is the §5jitarazu—stretching one beat and articulating it across multiple pitches | Stretch it too far and the meaning recedes into the background, and the passage takes on an instrumental feel like scat (wordless singing) |
| What to put on the strong beat | ★ Putthe accented beat of a content word; on downbeats and weak positions putparticles or unaccented kana（→ §8.3） | A particle on top of a strong beat → the weight of the whole sentence shifts to a place with no content |
4. Segmentation (phrasing) and breath points
4.1 Rules
Japanese lyrics arewords + particlesconstitutesegmentationstructure.
★ When the segmentation of the lyrics matches the segmentation of the melody, the words are easy to understand. (Melody Method, Expanded Edition, p.197)
Consequences of misalignment: the meaning does not get across, and it may even turn into a different sentence— if the timing of "Kimi ga kuru made matteru" is cut wrong,it cannot be distinguished from "Kimi ga kuruma de matteru"。
★ In this library, line breaks are segmentation：
lw-workflow The LYR-SPEC definition §1.8 explainsLine breaks are not typesetting; they are instructions, and the singer reads the line break as a phrase boundary.
One line of Japanese lyrics must simultaneously be: a segmentation group that makes sense + singable in one breath + matching the number of syllables .
4.2 Deliberate misalignment: three methods of emphasis
Prerequisites (all are indispensable)：the meaning is preserved，pitch accent and strong/weak positions are respected(Melody Method, Expanded Edition, p.198).
| Technique | Effect |
|---|---|
| Break before a particle | Suspends the listener and pushes them up step by step |
| Cut in the middle of a word | Creates tension and pulls attention in |
Three methods of emphasis(example: "Watashi wa koko de"):① Stretch "watashi—" to create tame (held-back tension)； ② Cut "wa-ta-shi" beat by beat；③ Isolate only "wa" unnaturally, suddenly startling the listener awake.
★ The principle in one sentence：Whichever word you want to emphasize, cut it apart more—it will rise to the surface on its own.
(Melody Method, Expanded Edition, p.198; for details of examples see reference.md §2。）
4.3 Full misalignment: singing the whole song with equal notes
⚠ First, correct a misunderstanding that is easy to read into this： Segmentation mismatch is not a rare special technique; it is the norm—— It is inherently impossible to make language and melodic phrases fully match everywhere throughout a whole song, and in practice musical phrases are often given priority over linguistic segmentation （reference.md §2.3）。 ★ So the question is not "can they mismatch" but "to what degree does this passage mismatch".
Method: sing the whole song with equal eighth notes,letting the phrasing of the melody override the segmentation of the language。 Effect: Single-minded / earnest — "Even so, I must sing these words out" (Melody Method, Expanded Edition, p.199).
Two Costs： ① ★ You must preserve the word segmentation through the movement of pitch(ensuring the segmentation of the lyrics through the movement of pitch), otherwise the meaning blurs; ② If overdone, the meaning of the words recedes far into the background and can even become comical.
Self-Check Action: For a passage of uniformly equal notes, read the wordsusing only pitch heightonce through. ★ If you cannot hear where the words divide → change the pitches until they divide, or give up this device.
5. ★★ 字余り / 字足らず — Deliberate Mismatch
The move in Japanese lyrics that most resembles "antisymmetry," directly connected to LYR-SPEC's asymmetry。
5.1 字余り (too many syllables)
Definition: There are more words than the melody can hold → specific words arerapidly bitten out ahead of time。 Effect：A breathless sense of desperate effort and urgency(earnest / pleading) (Melody Method, Expanded Edition, p.203).
★★ The key is not "how many extra syllables" but "where they are squeezed in"(choose one of four; for examples see reference.md §3.1）：
| Squeeze Position | Effect and Cost |
|---|---|
| ★ Beginning of the phrase | ★ Strongest momentum, creating a "pitching forward" thrust; extremely strong impression, most prone to affectation |
| Middle of the phrase | Momentum + emphasizes the word after it; ★ The segmentation is easily misheard |
| At the segmentation boundary | ★ Most conventional and smoothest, butno momentum— equivalent to squeezing for nothing |
| End of the phrase | Like a hurried ending, the words "overflow"; usually a flaw, unless that awkwardness is exactly what you want |
How to Strengthen It：★ Repeat the rhythmic pattern of 字余り throughout the entire phrase, and the feeling of "desperately pushing forward" will be firmly established.
★ Choosing the Device in Reverse: If you wanta sense of rushing, go for 字余り; if you wantsorrow, gloom, go for 字足らず (reference.md , the emotional contrast in §3.3).
⚠ Three Warnings(Melody Method, Expanded Edition, p.205): ① Extremely strong impression, extremely prone to affectation； ② ★ If the word squeezed and emphasized does not resonate, the whole sentence slides toward comedy； ③ ★ If overdone, consecutive equal-value notes push melodiousness into the background, and the melody becomes "like speech"— Conversely,narrative passages can deliberately make use of this。
Self-Check Action: At every instance of 字余り, ask two questions —"Where is it squeezed in?"and"Which word is squeezed out and emphasized?". ★ If you cannot answer the second question → that is not 字余り, it is not having counted the beats clearly; go back to §1.1 and recount.
5.2 字足らず (too few syllables)
Definition: Words don't fill the melody → specific words getstretched out, articulated across multiple pitches。 Effect: that worddraws particular attention, and deep emotions can be sung into the long note (『旋律法 増補版』p.206).
Side effect：★ : stretch it too far and the meaning recedes into the background, giving the passage an instrumental quality like scat (wordless singing).
Self-check actions (hard): for the beat being stretched,check item by item against §1.2—— ★ it must not be a voiceless sound(stretching equals stretching a stretch of breathy voice), ★ it must not be the moraic nasal ん(it can't be sung loudly or held long), ★ best is a true long vowel or an open vowel.
5.3 ★ Connect to this library's asymmetry
asymmetry.short_line fill inthe line with 字足らず(state which word is stretched, across how many pitches);
For reverse usefill inthe line with 字余り(statewhich position in the phrase it's crammed into, §5.1 choose one of four).
★★ This library's discipline holds here as well: every asymmetry must have a stated reason. The convenience of Japanese lies in字余り／字足らず coming with a ready-made set of effect descriptions—— crammed at the start of a phrase is drive, stretched out is deep emotion.Whichever effect you choose, write that word into the field.
6. ★ Four correction methods when the syllable count doesn't match (fix the melody side)
These four are the standard actions for 曲先 (= this library's melody-first, lyrics-later)(『旋律法 増補版』p.201):
| # | Situation | How to fix |
|---|---|---|
| 1 | Number of notes < number of beats(too many words) | ★ Subdivide the note values, or before/after the phraseadd notes |
| 2 | Number of notes > number of beats(too few words) | ★ Merge several note values into a long note |
| 3 | The count matches, but the rhythm doesn't fit | Change the note values / change the rhythm,regroup |
| 4 | ★ The melodic contour doesn't match the pitch accent | While respecting the 分節adjust the pitches and rhythm |
★ Examples of the four cramming positions for 字余り／字足らず are in reference.md §3。
6.1 The "leave room" correction method
★ Don't refine the melody right from the start—leave room for correction, and polish it through dialogue with the words. (『旋律法 増補版』p.202)
In the first draft, fix only the positions of the key words, leaving the rest roughly in place;in the second draft, split/merge notes according to the words； in the third draft, adjust the pitch accent, put the key words at the melody's highest point, and align the beats with the words to finish。 ★ The three-draft method with examples is in reference.md §7。
6.2 ⚠ These four methods have a limit that must be kept
★★ These operations can destroy structural effects like repetition and contrast. If a repetition structure is destroyed to accommodate the words—go back and change the words. (『旋律法 増補版』p.203)
This is the most important sentence on the boundary between this skill and composition/arrangement：
Changing the melody is an action this skill recommends, but changing it to the point of breaking the musical form/structure is overstepping.
★ Judgments at the structural level belong to lw-structure(word paragraphs) and composition/arrangement (musical form); this skill is only responsible for calling a stop here—write a message to music（lw-workflow §6.2）。
Self-check action: after each round of four-method corrections, placethe two passes of the same-named paragraph side by side and look—— ★ Are the places that were originally the same still the same? If they are different, and it is not something you intentionally changed → go back and revise the lyrics.
7. Rhyme and seven-five meter
7.1 Why rhyme (rhyming in Japanese) is weak
Definition：Rhyme = repeating words that contain specific syllables, giving the lyrics a sense of rhythm(Melody Method, Expanded Edition, p.207). It includes two types:
- Repetition of completely identical sounds: sea (umi) / give birth (umi)
- ★ Repetition with the same vowel combination: sea (u-mi) / country (ku-ni) = u-i rhyme
Alliterationplaced at thebeginning，End rhymeplaced at theendof a section; the act of rhyming itself is called 韻を踏む。
★★ Why Japanese rhyme is naturally weaker than Chinese and English（★ This is this library's reading, not the original book's wording): ① Japanese rhymesvowel sequences，consonants do not matter at all—Chinese rhyme requires the entire final to match, English rhyme requires both the vowel and final consonant to match; ② Japanese has onlyfive vowels, and combinations of two or three vowelshave a much lower threshold； ③ ★ The only warning given in the original book is not "it won't rhyme" but "rhyming too long will seem stiff, as if it was arranged in advance"—— It can be seen that the problem with Japanese has never been finding rhymes. Consequence: in Japanese lyrics, "rhyming" is not the main beam of the structure,at most texture; if you try to use rhyme to hold up a paragraph, it will not hold.
★ For the general principles of rhyming (rhyme schemes, surprise level, deliberate non-rhyming), all go to lw-rhyme;
this skill is only responsible for telling youthat in Japanese its leverage is shorter than you think。
7.2 How rhyme pairs with melody
★ At rhyming positions, the rhythm and beat placement are exactly the same, while the movement of pitch is different each time—rhyme naturally condenses the rhythm of the melody.
⚠ Warning：★ Dragging a rhyme too long will seem stiff, as if it was arranged in advance——
Let elements other than rhythm (pitch, etc.) flow and change, to save the sense of freedom back.
★ Usage in rap：Strip pitch from the non-rhyming parts (close to spoken delivery), use leaps in and out for the rhyming parts(flow belongs to lw-rap).
(Melody Method, Expanded Edition, p.207–208; for examples and the practice of "emphasizing rhyme," see reference.md §4。）
7.3 ★★ Seven-five meter and the "Japanese rhythm alarm"
Traditional forms that can be used deliberately：Seven-five meter(alternating seven beats and five beats),haiku 5-7-5、Waka 5-7-5-7-7、 Three-three-seven rhythm、The rhythm of matsuri-bayashi (festival hayashi)—use themto evoke a Japanese flavoris a legitimate technique (Melody Method, Expanded Edition, p.208).
⚠⚠ A reverse warning—this is the real point of this section： ★ These patterns are so familiar to Japanese ears that they slip in on their own before you notice, producingunintentionally very Japanesemelodies. ★ Watch out especially when writing rhythm-driven things like rap— The rhythm of shichigosan and tongue-twister rhythms like "Namamugi Namagome Namatamago"will both pop out unconsciously。
★★ Where this sits in this library：
It is the Japanese version of "pick only the smoothest rhymes"（lw-mandarin §2.8, LYR-LINT item 8).
The same disease: the mold that is smoothest for the language itself—not choosing it actually takes effort.
Self-check action (run as the Japanese substitute for lint item 8): For each line,list the beat counts, and see whether there is a run of 7-5-7-5, 5-7-5, or 3-3-7. ★ If yes → ask "did I do that on purpose?". If you cannot answer, break it up:change the beat count of one lineand that is enough.
8. Shisen/Kyokusen: Two Workflows
8.1 ★ Terminology Correspondence
★ Shisen (しせん) = this library's "lyrics first, then music"; kyokusen (きょくせん) = this library's "music first, then lyrics".
After mapping them,lw-workflow the two stage tables in §2/§3 apply as-is, and this section only adds the differences on the Japanese side.
8.2 Shisen (Lyrics First, Then Music)
Method: write the lyrics first,then conceive the melody while singing the lyrics—— ★ The syllable count, note count, pitch accent, and rhythm of the pitches will naturally align(Melody Method, Expanded Edition, p.200).
Two easier half-shisen methods：
- ★ The whole song does not have to be shisen：Only make the catchphrase part (for example, the start of the chorus) shisen, and it is already very effective;
- ★★ even just predetermining the positions of sokuonis already worthwhile— place "zutto," "motto," "kitto" at the start of phrases, and by using sokuon and chōon to design phrases, you can get a strong expression in which lyrics and music are tightly welded together。
★ This is a convenience unique to Japanese: The biggest fear in Chinese lyrics-first writing is "discovering inverted tones only after the lyrics are finalized" (
lw-workflow§3's biggest trap); Japanese can lock only the skeleton of sokuon/chōon and leave everything else to the melody—— Once the skeleton is locked, the interlocking of lyrics and music is already established, and all the room to revise the lyrics remains.
Self-check action: For Japanese lyrics written lyrics-first, mark two things first:
★ Which beat each line's sokuon falls on、Which beat each line's chōon falls on.
Write these two columns into prosody.sections[].stress_positions(§10) and hand them over as melody constraints; that is far more useful than handing over a syllable-count table.
8.3 Kyokusen (Music First, Then Lyrics): Filling in Lyrics According to Melody Type
General principle: Melody first, lyrics filled in afterward; at the stage when the lyrics come out,operate on and revise the melody according to the pitch accents of the words(using the four methods in §6).
★★ But melody-first does not mean "finish the melody, then think about the lyrics"： Even if you write the melody first, you should also envision in advance the words for key positions such as the opening of the melody and the kime (decisive passage) of the chorus—— ★ These pre-set words do not have to be complete sentences; single words are enough， It is a common and realistic approach for the melody to take shape like "worm-eaten" holes, with words here and there。 Preferentially pick words containing sokuon, chōon, and hatsuon("zutto" "motto" "kitto" "ikōyo") —— The reason is exactly the same as in §8.2:Once these beats are fixed, the interlocking of lyrics and melody is already established. （reference.md §8。）
① Melodies of the repeated-same-pitch type(rap style, often appearing in the A-melo / hirauta):
- It has a definite pitch—relative to the tonality it is usually the tonic or dominant (1st or 5th degree)；
- At the end of a phrase it often jumps down a perfect fourth；
- ★ Where it jumps down, the articulation is necessarily weak → put particles or unaccented kana there, and it sounds clean;
- ★ In the swing rhythm "ta-ta-ta-ta→ta-ka-ta-ka",the "ka" position is likewise suitable for placing unaccented kana。
(Taniguchi Naohisa, Puro no kyokuzukuri ga wakaru hon, p.60.)
② Writing lyrics for the chorus (sabi)：
The formula isto choose highly universal words, words that speak of big things, so that the chorus "sounds like a chorus"—— ★ But if you do not break the formula, nothing interesting will be born.(Same as above)
⚠ "Highly universal words" are exactly the category this library imagery.forbidden wants to blacklist, the conflict is real.
★ Reconciliation of positions：The chorus summarizes the meaning; it does not mean replacing all the concrete nouns with big words(Same as lw-mandarin §6.4）。
invalidating condition: the first half holds only when you genuinely need a "standard product";this library's default stance is on the side of breaking the formula.
9. ★★ English words mixed into Japanese songs: the iron rule
This is the section in Japanese lyric writing where you are most likely to crash, and also the easiest to check.
9.1 The iron rule itself
★★ The biggest pitfall: the number of syllables in the English does not match the number of notes. (Puro no kyokuzukuri ga wakaru hon, p.62)
- free = ★ 1 syllable：★ cannotbe split into "f-ree" and sung as two notes——where there is no vowel nucleus, no melodic note can be placed
- "furī" = ★ 3 beats：★ can be split
- freedom = 2 syllables(free-dom): can place 2 notes / "furīdamu" = 5 beats: can be split
★ One actionable item：Dictionaries mark the syllabification points of English——when writing lyrics, go look them up.
(English's own rules for stress and beat count are in lw-english；this skill only handles this conversion。）
9.2 ★ Translating English words into Japanese ≈ rewriting the entire lyrics
One English word delivers its full meaning in just 1–2 syllables; with the same number of notes, it carries far more information; ★ A literal translation will inevitably not have enough syllables. (『プロの曲作りが分かる本』p.62)
Two ways out：
- ★ 省く力 (the power to cut)—Cut until gaps open up between the characters;
- ★ Or just translate freely (= rethink from zero)。
★★ This and lw-english §6 item 6 are two directions of the same rule:
English→JapaneseNot enough syllables, so cut; Chinese→EnglishSyllables overflow, so rewrite。
Common conclusion: words cannot be translated across languages, only rewritten.
9.3 ★ A control valve unique to Japanese
The same concept, as English, can only occupy 1–2 sounds; as a katakana loanword it can stretch to 3–5. ★ If you need more notes, Japanize it; if you need fewer notes, Englishize it.
⚠ Cost: Katakana English and real English have completely differentlanguage feel—— What you are swapping is register, not just beat count. Before swapping, first ask whether the register of this passage allows it. ★ For NG/OK example comparisons, see reference.md §5。
9.4 When the melody already comes with English scratch words
Scenario: Songs bought from overseas often come with an English guide vocal (仮歌). ★ Forcing one Japanese kana into every note will completely destroy the atmosphere. Four countermeasures(match sounds with a 促音 feel to Japanese words with pauses/beats / match strong consonants with strong consonants /split eighth notes into sixteenth notes to increase the number of notes／ At high, long notes (白玉), just use English）★ In reference.md §6。
フェイク (improvised embellishment): The essence is improvisation; in principleleave it to the singer to freely perform； ⚠ ★ Having a non-native English-speaking singer sing Fake lines mixed with English will not work— If you want that flavor, thendesign the English in advance for this singer, orfirst have him sing syllables that are easy to sing, then replace them with English words that have similar pronunciation (『プロの曲作りが分かる本』p.62–63; for details see reference.md §6.1）。
★ Boundary：How exactly Fake is sung and recordedbelongs toVoice(recording and vocal direction); This skill is only responsible for "which syllables can be handed over for improvisation and which cannot."
10. Correspondence with LYR-SPEC fields
| Field | How to fill in Japanese words |
|---|---|
meta.language |
日文 |
prosody.sections[].syllables |
★★ Fill in the mora count (beat count), and count by "how many sounds are sung"(§1.1, §3.2). For places where sounds are merged, state it in note write it in |
prosody.sections[].stress_positions |
★ For this Japanese cell, fill inthe high/low pattern of pitch accent, andwhich beat the 促音 and 長音 fall on（§8.2） |
prosody.sections[].peak_syllable |
★ It cannot be a voiceless sound, cannot be the moraic nasal ん; preferably a 長音 or an open vowel (§1.2, §5.2) |
prosody.sections[].melody_contour |
andpitch accent、sentence-ending rise/fallWrite the relationship here (§2.1, §2.3) |
prosody.sections[].notes |
○ When there is a pitch sequence, use it to check the line comparison in §2.1 and the deviation of 撥音 from chord tones in §1.2 |
lyrics[].lines |
★ Line breaks = 分節 = breath points(§4.1). Places deliberately misaligned are in note write the reason (§4.2) |
asymmetry.short_line |
★ 字余り／字足らず, specify the squeezed position or the lengthened word (§5.3) |
asymmetry.unrhymed |
Japanese rhyme is inherently weak,deliberately not rhyming must be written more explicitly than in Chinese or English（§7.1） |
rhyme.che |
★ Japanese has no 辙口. In this cell fill invowel combinations(e.g. u-i) + 頭韻／脚韻（§7.1） |
checks.self_audit |
Go through the checklist in §11 item by item |
⚠ Japanese replacements for the two lints：
- Item 7(倒字率／协音违例) ★ does not apply to Japanese—instead run thepitch accent line comparisonfrom §2.1 and thespecial mora check；
- Item 8(辙口 must not use only one or two broad 辙) ★ instead run theseven-five meter alarm：list the mora count of each line, check whether there is a run of 7-5／5-7-5／3-3-7。
★ The other eight items (intent four items, concrete nouns, character count of each section,asymmetry、withholding、pov＋tense_anchor, rhyme unexpectedness, chorus must not repeat verbatim)all apply to Japanese as usual。
11. Final self-check
- ★★ Every linecounted mora, and what is counted isthe sounds as sung, not the kana on paper (§1.1, §3.2)
- ★ 拗音 has not been mostly counted as one beat；促音, 撥音, and 長音 have each been counted as one beat，促音 has been paired with a rest（§1.2）
- ★ a 撥音 ん given its own note is short and weak; a ん falling on a strong-beat long note has left the chord tone（§1.2）
- ★ beats with no sound are paired with short notes；there is no no-sound beneath the highest note and long note（§1.2、§5.2）
- ★ where the same vowel is adjacent, a rest has been inserted or the pitch changed（§1.2）
- ★ beneath every long note there is either a true 長音 or the end of a 分節（§1.2）
- ★★ the pitch accent line has been compared with the melodic line; the reversed placesare all deliberate（§2.1）
- places deliberately deviating from the pitch accentdo not exceed two, andthe surrounding words are correct（§2.2）
- therise and fallat the end of a section matches whether that section is a "question" or an "answer" (§2.3)
- ★ have not used the Chinese 倒字 approach to check Japanese character by character（§2.4）
- ★ Every linehas been sung through once, and the 符割り's merged sounds and omissions have been recorded (§3.2, §3.3)
- what is placed on downward leaps and weak beats is a particle or an unaccented kana, and no particle is placed on a strong beat (§3.4, §8.3)
- ★ Line breaks = 分節; the deliberately misaligned places can be justified, andthe meaning is preserved（§4.1、§4.2）
- In passages of equal notes,the 分節 has been preserved again using pitch（§4.3）
- ★★ Every instance ofji-amari can state "where it's crowded" and "which word is emphasized"（§5.1）
- ★ Every instance ofji-tarazu: the mora that gets stretched can withstand being stretched(not a voiceless sound, not ん) (§5.2)
- ★ After the four correction methods,the repeated structure of same-named paragraphs has not been broken（§6.2）
- ★★ the mora counts have been listed out and checked against shichigosan / 5-7-5 / 3-3-7; if present,it is intentional（§7.3）
- If words come before music,the positions of sokuon and chōon have been written into
stress_positionshanded over（§8.2） - ★ English that has been mixed inhas been checked against a dictionary by syllable count, and no monosyllabic word has been split into two sounds (§9.1)
- ★ The English partwas not translated from Japanese/Chinese（§9.2）
-
rhyme.chewhat is filled in isvowel combinations + alliteration/rhyme, not che-kou (§7.1, §10)
Appendix: Sources
- Main trunk: directly quoted from the lyrics–melody interlocking content of the two original books—— Hiroshi Takayama, *Popular Music Composition no Tame no Senritsu-hō Zōho-ban*, Chapter 7 "Kashi" (p.189–214): the interlocking of the two kinds of time, intonation and standard language, word rhythm (chōon / sokuon / hatsuon / voiceless sounds / consecutive vowels), segmentation and the three emphasis methods, the four syllable-count correction methods and the three-draft method for leaving room, ji-amari / ji-tarazu, rhyme (alliteration / rhyme, vowel-combination rhyme) and the "Japanese rhythm alarm" of shichigosan (→ this skill §7); Naohisa Taniguchi, *Pro no Kyokuzukuri ga Wakaru Hon* 3-6–3-8 (p.58–63): fu-wari and "six words for the same melody," writing lyrics according to melody type, the iron rule of English syllables, feiku and karauta.
- Examples and score-example-level treatment：reference.md, with the respective chapters and page numbers of the two original books indicated—— *Popular Music Composition no Tame no Senritsu-hō Zōho-ban*, Chapter 7 "Kashi," *Pro no Kyokuzukuri ga Wakaru Hon* 3-6–3-8 (Naohisa Taniguchi, book pages 058–063).
- ⚠ Three points of scope that need to be stated： ① yōon counts as one morais the basic rule of mora counting, consistent with the calculation of "free = 3" and "freedom = 5" in the main trunk material, the main trunk material does not separately list yōon, so here it is filled in and marked according to the same counting method; ② The word used in the original book when counting morae is "onsetsu"; this skill uniformly uses mora (haku)，to avoid confusion with the English syllable(§1.1); ③ §7.1 "Why Japanese rhyme is weak" isa reading derived by this library from the material; the original book only gave the one warning that "rhyming too long becomes stiff."
- ⚠ This skill applies only to Japanese.
For Chinese, see
lw-mandarinandlw-tone-check; for Cantonese, seelw-cantonese; for English, seelw-english—— ★ the three sets of constraints cannot substitute for one another（§1.3）。 - All go out of scope: vocal direction and diction, vocal recording, microphone and placement, pitch correction, harmony stacking and vocal mixing → Sound(recording and vocal direction, mixing); melody and form → Composition and Arrangement。