Melody-writing methodology - pitch-motion mechanics, note duration and rhythm, modes and pentatonic scales, motivic development, phrase construction, lyric-melody fit, and whole-song climax design. Use when the user wants to write a melody (verse/chorus), wants to improve a bland melody, needs motivic development techniques, or wants to understand how melody interacts with harmony/lyrics.
Melody-writing methodology - pitch-motion mechanics, note duration and rhythm, modes and pentatonic scales, motivic development, phrase construction, lyric-melody fit, and whole-song climax design. Use when the user wants to write a melody (verse/chorus), wants to improve a bland melody, needs motivic development techniques, or wants to understand how melody interacts with harmony/lyrics.
Japanese-Style Melody Writing (Melodic Method)
Source: the complete system of Hiroshi Takayama, Popyurā Ongaku Sakkyoku no Tame no Senritsu-hō, Expanded Edition (ポピュラー音楽作曲のための旋律法 増補版, Rittor Music).
This file reorganizes the book into a practical reference by compositional task. The large lookup tables (seven types of non-chord tones, motive-construction table, eight degrees of cadential openness, form formula collection, situation-based reverse-lookup skeleton, etc.) are in reference.md in this directory.
0. Core Concept: A Melody = One Tone Moving Through Time
A melody is not "multiple tones arranged in space" but one tone moving through time (imagine a floating ball moving up and down).
Listeners connect earlier and later tones by memory and retroactively update each tone's role (starting point → passing point → arrival point) — only upon hearing the 3rd tone is the 1st tone reinterpreted as "the start of an ascent."
Every tone sits in a force field (場, ba) made of four forces:
Gravity of register: high notes are tense — moving upward requires effort; low notes are settled — moving downward feels like releasing tension;
Force of the mode: in-scale tones are stable, out-of-scale tones are pulled back;
Force of harmony: consonance with the chord is stable, conflict makes the tone want to move — this is precisely the source of forward momentum;
Force of beat and rhythm pattern: on-grid positions are stable, positions off the rhythmic grid are unstable.
Two additional sources of expressive motion:
Everyday-experience associations: speech intonation (イントネーション), the physical load of bodily movement — stepwise motion = small gestures, hence smooth; large leaps = big gestures, hence effortful; repeated low notes = the strong vibration of a large instrument, hence a sense of heat.
In-song experience (designable by the composer): an unusual use of a tone can be accepted empathetically if it is prepared in advance within the song (letting listeners accumulate the relevant experience first); an ordinary use of a tone can become fresh through contextual relationships. This is the key to resolving the "banal vs. self-indulgent" dilemma.
Segmentation (分節, bunsetsu): the same string of tones, segmented differently, = a different melodic experience. When writing a melody you must constantly anticipate "where and how the listener will segment." The units cut out by segmentation are phrases (フレーズ).
Hierarchization of motion: tone → motion → phrase → motion-of-⟨tonal motions⟩ → large-scale structure.
Stacking and enlarging tension (tension↑ → more tension↑ → expansion↑) creates sustained expression and large forms;
The final release acts on everything that came before, producing an intense sense of relief (安堵感).
Terminology convention (strictly distinguished in the book): pitch = the height of a tone itself; interval = the distance between two tones.
1. Expressive Rules of Pitch Motion
Direction (向き)
Motion
Expression
Ascending
Tension, exaltation; phrase-final ascent = question, tossing out, addressing the other
Hesitant wavering (逡巡, shunjun — dithering back and forth), neither ascent nor descent
For phrases of three or more tones, overall direction is judged by the relative height of the first and last tones; if the local motion arriving at the final tone opposes the overall direction, the sense of overall direction is weakened.
Segment long phrases first, judge direction section by section, then assess the composite effect.
The harmonic background can rewrite the meaning of direction: if an ascent lands on a stable tone by harmonic mechanics (e.g., G7→C landing on the tonic), the "questioning feel" is replaced by a "sense of resolution." To emphasize question/tossing-out, remove the elements that block it (don't land on a stable tone).
A slow ascent with restrained volume and tension can express "longing gazing into the distance, pure yearning" rather than tension.
Bodily analogy: ascent = stretching up to look far away (伸び上がり), descent = bending down to look closely (かがみ込む).
Width (幅)
Stepwise motion (順次進行, junji shinkō):
Smooth and flowing, without obstruction; ascending/descending along the scale = a smooth increase/decrease of tension, extremely common;
Adding auxiliary tones a 2nd above/below the existing tones as ornamentation = increasing motion while keeping the contour, more florid.
Semitones:
The narrowest interval, expressing subtle nuance and delicate tenderness (デリケート); a melody containing semitones = voluptuous beauty (艶やかさ) and depth;
Overuse (especially chromatic semitones with accidentals) → neurotic, morbid, grotesque (グロテスク) — must be restrained;
"How to use semitones well" is a major requirement of melodic sensibility (one of the two great sources of melodicity, alongside pentatonic density).
Leaps (跳躍進行, chōyaku shinkō):
Dramatic, large emotional swings; the position of the leap is emphasized (whichever word the leap lands on, that word stands out);
A single leap should stay within a 6th; a perfect octave, however, is actually easy to sing and easy to hear, and is usable;
Too many leaps still make a melody hard to sing and unnatural — use with restraint.
When pitch stays still, the listener's attention shifts to rhythm → combine syncopation and varied durations = percussive / speech-like (語り, katari — narration) expression.
Width × speed:
The narrower the interval, the easier to sing and the more it suits any tempo; the wider, the more it needs slow tempo and preparation;
A large leap needs sufficient duration at its start or end (or both) to sound natural and singable; rapid continuous alternation between high and low notes stiffens most easily.
2. Preparing and Landing Leaps
Upward leaps are easy, downward leaps are hard (moving while adding force is controllable; motions that release force are hard to stabilize).
Two remedies for a wide downward leap:
The landing point naturally carries an accent (アクセント) — anticipate it and use it deliberately;
Lengthen the starting tone fully + diminuendo (ディミヌエンド) — build up emotion and tension at the start before leaping.
Among narrow leaps, downward is actually easier (force release) — a "short high note leaping down" at the start of a phrase to create a phrase-head accent is a common device (the 「いつでも」「どこでも」 type, matching word intonation).
Tame (タメ, windup/holding back): before a leap, place 1 to several tones moving in the opposite direction (usually stepwise), like winding up before a big gesture; it can also be placed mid-leap.
Josō (助走, run-up): before a leap, place 1 to several stepwise tones in the same direction; the run-up tones may lie inside or outside the leap interval.
Landing cushion: after arriving at the leap, follow with 1 to several stepwise tones in the same direction, softening the landing impact — especially effective for downward leaps.
Rebound (reaction): after a large leap, stepwise motion in the opposite direction = a landing rebound; that rebound step can double as the tame for the next leap, forming a chain of "leap → rebound(windup) → leap → rebound(windup) → leap."
Splitting a 7th: a 7th leap is most commonly split with an intermediary tone into 3rd+5th or 4th+4th (stepping-stone style, 飛び石, in two steps).
Hard limit on same-direction consecutive leaps: leaps of a 4th or larger may continue in the same direction at most 2 times, and the total interval between the two ends must stay within 1 octave.
Compliant examples: 3rd+5th = 7th, 4th+4th = 7th;
Violations: 5th+5th = 9th, 4th+4th+4th = 10th — unnatural and hard to sing.
A melody with many large leaps tends to sound instrumental and mechanical; but a large leap at slow tempo with unhurried preparation can draw an expansive, open (伸び伸び) melodic line.
Leaps larger than an octave strongly sever the melody (they cannot be segmented as one phrase); reverse use: separate registers widely at segmentation boundaries → a pseudo-two-voice melody (a low-register motive and a high-register motive in dialogue).
The value of "unnatural" motion (all require restraint — overuse kills the effect):
3. Range Setting and the Human Voice
Basic Rules of Range
First set the melody's central range at roughly 1 octave, and place the center of each phrase within it — otherwise the melody becomes diffuse, or even sounds like a mix of multiple voices.
The whole song's actual range is about 1.5 octaves, at most 2 octaves (which happens to be the usable range of most voices and instruments).
Contained within 1 octave = a narrow-interval melody, intimate and easy to sing;
Spanning 2 octaves = strongly dynamic but hard to sing.
Effects of a phrase's position within the range:
Center shifted high = bright, full of tension; a tone breaking above the range ceiling spikes in tension; extreme highs = an anguished (苦しげ) expression;
Center shifted low = calm and relaxed; falling below the floor = heavy (重々しい); extreme lows = powerless slackness.
The same motion has different effects at different positions in the range:
An ascent in the middle of the range = moderate tension; an ascent toward the top, breaking out of the range = tension conspicuously amplified;
An ascent climbing from a low position back to the middle = the tension increase is offset by "low notes turning soft," and may even act as relaxation;
The same phrase-final ascent: low register = a light-hearted question; high register = an earnest plea, even an interrogation (詰問).
Pitch character is relative: the same middle C is heard as a "high note" in a low-register melody and a "low note" in a high-register melody.
The Human Voice (声の音域)
Male voices are about 1 octave lower than female voices; by convention male voice parts are notated 1 octave higher than sounding. In pop practice: male voices rarely use the low register and often use the high register; female voices rarely use the extreme high register and commonly use the low register.
Chest voice (チェストボイス, ≈ jigoe/natural voice):
The root of pop-music reality (リアリティ) and intimacy; the character of pitch motion is expressed most intensely here;
The strained quality of the high register can express earnestness (the urgency of wanting to convey something).
Middle voice (ミドルボイス):
Half-closed vocal folds, extending the range upward; the timbre is bright and open = a positive, projecting strong expression;
The same high phrase: sung in chest voice = strained and imploring (苦しげ); sung in middle voice = expansive, gazing into the distance — register choice changes the melody's character;
Whether a male singer has a middle voice, and where his passaggio (換声点, kanseiten — register break) lies, is a major precondition determining range and key (キー) — confirm it first.
Falsetto (ファルセット/裏声):
Treat it as a different instrument from chest voice: pure and delicate, suited to refined, tender (デリケート) expression;
Limitations: few overtones → diction slightly blurred; dynamics hard to vary → unsuited to intimate confiding and large emotional swings;
Flipping into falsetto from chest voice = the desperation (懸命さ) of reaching a high note even at the cost of the voice breaking → intense longing, ultimate setsunasa.
For singers with an awkward passaggio: design registers phrase by phrase (this phrase in chest voice, that one in falsetto/middle voice), so no single phrase crosses the break.
Set the melody's basic octave in the middle of the usable extreme range — leaving headroom while keeping the option of deliberately touching the limits for expression (a strained high note conveying depth of feeling).
The shōwa (唱和, inner sing-along) effect:
Listeners hum along with the melody in their heads — after being induced into entrainment by accessible repetition, a subsequent high note still feels "high and tense" to the listener even if the singer produces it effortlessly;
A sudden high note at the song's opening or a phrase's start gives no time to entrain → it is externalized and appreciated as display (a sense of virtuosity);
When writing, be explicit about whether a passage will be "sung along with in entrainment" or "appreciated from outside," and actively induce the intended mode.
4. Duration and Cadence
Usable Range of Note Values
Common note values: 16th note to whole note; tuplets are normally only 1-beat or 2-beat triplets (32nd notes, quintuplets, etc. appear only occasionally to cram in lyrics).
Principle: pop music must let listeners stay entrained — note-value combinations must not make the listener lose the beat.
For a whole song in triplet feel, don't write triplets everywhere; renotate in compound meters 6/8, 9/8, 12/8.
First examine character at a moderate tempo around BPM ≈ 100, where "note value ≈ note length," then consider the shifts at extreme tempos.
Long Notes and Short Notes
Long notes = sustained force = tension; expression can be shaped inside the note (vibrato ビブラート, crescendo/decrescendo) → climaxes commonly place a long note in the high register (the maximum-intensity expression of tension).
Consecutive long notes + little pitch motion = sustained tension; but sustaining too long turns slack and boring — insert leaps, or weave short notes between the long ones to revive it.
Long notes + large leaps combined = smooths out large pitch motion while amplifying the sense of scale (スケールの大きさ).
Each long note carries strong expression: sung forte = power and conviction; sung piano = gradually fading, hanging by a breath.
Short notes = light, agile, lively; sung forcefully = sharp, brusquely flinging away (突き放す). Short notes in the low register are offset by heaviness (重い).
Short notes + wide leaps are hard to sing and come with built-in dynamic swells — usable intentionally as accent/load; a large leap still needs "motion-preparation time" before it.
Narrow intervals × short notes = melody approaching speech (語り/セリフ) (news-anchor speech rate ≈ 8th notes at BPM 120, which does not sound fast).
Continuous short notes across a wide range = light rolling motion; kept up too long = neurotic, obsessive.
Mixing long and short: long notes = principal (center of gravity / natural accent), short notes = subordinate — regardless of beat position.
Common cycle: short notes approach (アプローチ) a high long note → tension → a low short note releases → approach again…;
Extreme compression = alternating long-short-long-short skip feel (スキップ) (common in 6/8, 12/8) — handy but prone to yotei-chōwa (予定調和, predictable resolution / cliché predictability); beware banality.
Duration Treatment at Cadences
Lengthening the cadence tone = motion stops; two follow-ups: add one short note to release the tension, or use a rest for de-tensioning (脱力).
Duration alone is not enough to close a cadence: if harmony and mode (scale degree) do not cadence at the same time, the sense of ending remains half-finished — usable in reverse: let the chorus settle while the whole song stays unresolved, leaving a "words left unsaid" feeling.
Large cadences (section-final / song-final):
Ritardando + decrescendo bleed off energy; a "notated ritardando" = writing each note before the cadence progressively longer, or inserting one bar of 2/4;
Variant: stretch the note before the cadence fully (tame) then end on a short note — tension held to the last instant, then released in a flash;
Further: the melody omits the cadence tone, handing it to the accompaniment — a dreamlike afterglow like breath cut short;
Reverse: no slowing, no lengthening, close with an accented short note = a decisive, clean, strong will (can be reinforced with an additional accented short note).
Rests are part of the melody:
A powerful segmentation tool — at the end of a long phrase, don't fill the full value with a long note; place a rest to make the segmentation explicit yourself (or use a breath mark ブレス or a slur スラー to frame the phrase);
A short rest inside a phrase = de-tensioning, functioning like a low short note;
Wherever the lyric contains a sokuon (っ, geminate stop), a short rest necessarily results.
5. Beat Position and Syncopation
Beat Position of Phrase Openings and Closings
Beat position
Opening effect
Closing effect
On the strong beat
Straightforward and sincere; but easily monotonous, and sounding simultaneously with the accompaniment bass, the opening note is easily buried
Straightforward but stiff
After the strong beat
Led in by the background, prompted to speak; singing on the groove, coyly withholding (もったいぶる)
Ending as if hurried along, words trailing off
Before the strong beat (anacrusis/pickup, アウフタクト)
The melody leads the whole song; actively addressing someone, strong emotion and clear will
The melody leads the background to a close
A song opening with no accompaniment + flat note values easily falls into a meterless state (beats cannot be grouped) — beware; it can also be used deliberately as an unstable, mysterious introduction.
Mid-song meter changes: a section-level change = dramatic but risks splitting into two songs (keep unity with the same melodic figure); locally inserting one bar of 2/4 = a tame before the chorus that launches a pickup — a stock device; at any meter change, anticipate what the drums and bass will do.
Syncopation (シンコペーション) and Anticipation (アンティシペーション)
Syncopation = a weak-beat (or offbeat) note tied into the following strong beat → that note becomes "strong-beat-ized" → the next step is dragged later = a tame feel, unhurried lag.
Anticipation = a strong-beat note played early on the preceding weak beat → the progression arrives early = urgency, restlessness, a driving rush.
Precondition for both: the listener already clearly perceives the beat — used a cappella or at the song's opening where the beat is not yet established, they are not only ineffective but disrupt the sense of meter.
A melody glued to the beat throughout = plain, earnest and forceful, but stiffly foursquare (四角四面), and since the accompaniment strikes on the beat, the melody doesn't stand out; moderately detaching from the beat (offset the opening + timely syncopation) → the melody floats up off the beat, free and expansive (伸び伸び).
Coordinating with the Rhythm Pattern (リズムパターン)
Write the melody with a rhythm pattern already implied inside it: while composing, play back an 8-beat/16-beat groove in your head and let the melodic motion move within that groove (ノリ) — avoid "forcing a melody into a ready-made rhythm pattern."
Three relationships with the backbeat (the snare on beats 2 and 4 in an 8-beat):
Have the melody rest at the backbeat positions → a call-and-response (掛け合い) with the backbeat;
It need not be strictly vacated; consciously yielding to it is already effective;
Reverse: melodic strong accents in unison (ユニゾン) with the backbeat — you can partially overlap and let the backbeat show through where the melody rests.
8-beat melodies often move on the offbeats (ウラ) of quarter notes; 16-beat melodies on the offbeats of 8th notes → syncopation/anticipation arises naturally.
Ura-nori / omote-nori (ウラノリ／オモテノリ): a groove consciousness taking the offbeat/onbeat as the starting point of each gesture (start → resolution); neither is superior, but the feel differs:
The groove determines which words get focus (in ura-nori, words starting on offbeats become focal) — to emphasize a word, adjust note placement according to the groove.
Bounce (バウンス): offbeat notes shift later, closer to the next onbeat (≒ onbeat notes lengthen); the slower the tempo, the bigger the bounce; fixed at triplet values = shuffle (シャッフル); in a bounced song the melody can use triplets without incongruity.
The rhythm pattern belongs to the accompaniment: leave gaps in the melody so the pattern shows through the gaps (a three-dimensional sound); don't trace/duplicate the rhythm pattern with the melody — redundant, cluttered, mutually canceling.
The Motion of Rhythm (リズム) Itself
Rhythm = grouped note values + accent placement (different accent placement = a different rhythm); high notes naturally act as accents — you can build rhythm by placing high notes alone; swapping on/off positions (ウラオモテ) radically changes a rhythm's character.
Plain repetition: stable but rigid; enlarging the repeated unit to 2+ bars reduces the sense of formality; the fewer note-value types, the stronger the repetition effect and the heavier the formality.
Displaced repetition: repeating a 1.5-beat (dotted quarter) unit in 4/4; repeating a 2-beat unit in 3/4 = hemiola (ヘミオラ) — the displacement generates tension.
Deformed repetition: overall compression = impatient, pressingly earnest; overall expansion = unhurried and settled; lengthening via anticipation = paradoxically more urgent.
Partial repetition: leisurely echoing the phrase ending after a gap = echo (エコー) — introspection, lingering regret (心残り), murmuring to oneself; progressively shortening and beat-compressing the repeated part = tatamikakeru (たたみかける, pressing in step by step) — urgency.
Contrast: rhythmic shapes must have vivid characters to be recognized; the most legible = short-value groups vs. long-value groups; a latent contrast with the order flipped prevents boredom without seeming arbitrary.
Rhythm with unclear segmentation = long-winded; segmentation that is too self-evident = dull — accumulate tension with "hooked" note values and beat positions, then release at a plain-spoken position (a sense of opening, 開放感).
6. Modes and Pentatonic Scales
Major Mode / Minor Mode (長旋法／短旋法)
A mode = selecting tones from the 12 + assigning roles (tonic, etc.); the key is determined by the tonic's position, not by the mode type; the same set of tones with a different tonic = a different mode (all white keys can be C major or A minor).
Major-mode mechanics: the leading tone (導音, dō-on, degree vii) resolving up a semitone to the tonic is strongest > dominant (v) > subdominant (iv);
Mechanics ≠ compulsion — the leading tone moving by arpeggio into another chord tone is also very common; the major mode overall has few constraints and free motion;
An overly strong functional feel (leading tone → tonic, tracing dominant-to-tonic harmony) = prim and stiff (school-anthem feel / punk's blunt shouting);
Major key ≠ cheerful: graduation songs and other works of sorrow and nostalgia are mostly in major.
The three forms of the minor mode:
Natural minor scale: no leading tone; motion is free but melodies easily go flat and featureless;
Harmonic minor scale: vii raised a semitone as leading tone (extremely strong pull to the tonic); an augmented 2nd arises between vi–vii — avoid connecting them directly;
Melodic minor scale: ascending, both vi and vii are raised (eliminating the augmented 2nd; the raised vi strongly tends toward the leading tone); when vii is not acting as leading tone it is not raised and moves toward vi (descending form).
Standard practice: proceed freely in natural minor, and locally mix in harmonic (or melodic) minor at section boundaries to "punctuate" — strong effect but easily predicted and contrived-sounding (わざとらしい); use sparingly.
Beyond the leading tone, minor has several other semitone locations → well suited to emotional (情緒的) melodies; minor key ≠ dark (much hard rock and many jazz standards are in minor).
Church modes = grab the characteristic tone (each differs from the major/minor mode by exactly 1 tone; full table in reference.md):
Dorian (ドリア) = major 6th; Phrygian (フリギア) = minor 2nd; Lydian (リディア) = augmented 4th; Mixolydian (ミクソリディア) = minor 7th;
Place the characteristic tone at the right moments in the melody + support it with chords containing the characteristic tone; never using the characteristic tone = the mode is imperceptible, wasted;
Discrimination: a melody with lowered vii is usually "major mode + blues pentatonic," not Mixolydian.
Pentatonic Scales (ペンタトニック) — a Major Source of J-Pop Melodicity
Pentatonic scales have no leading tone and cannot form a dominant chord → cannot establish a key by themselves; the mechanical core = "neighbor-tone tendency":
Tones tend to move to adjacent scale tones (adjacency comes in 2nds and 3rds; 2nds are easier); leaping to a farther tone requires force, and after the leap the tendency is to reverse direction.
Four typical figures ("pentatonic-sounding"):
Move to a neighbor and fold back;
Move to a neighbor and continue in the same direction;
First "hook" (引っかける) to the neighbor in the opposite direction, then leap back two scale positions (= the pentatonic version of tame; the variant whose first step is a 2nd is the most common);
Overshoot past the target neighbor to the far side, then fold back.
Three figures that destroy the pentatonic feel:
Frequently leaping to tones 3+ scale positions away (one such leap followed by a return to normal is fine);
Leaping 2+ positions and then continuing in the same direction;
Consecutive leaps or repeated same-direction leaps.
Of these, arpeggio-like (broken-chord) motion is the most destructive — but arpeggio leaps across phrase boundaries are fine.
Same-direction stepwise motion ≤ 3 steps: 4 steps sweeps the entire scale = a characterless scale-run that merely announces "I am pentatonic."
Mode shifting (転旋法, ten-senpō):
Two pentatonics a fifth apart differ by only 1 tone → easy to switch, commonly used, enriching the pentatonic's small tone supply;
Switching between major/minor mode ⇄ pentatonic (usually without a key-signature change): a pentatonic opening (intimate, singable) → splice into stepwise scale motion (semitone momentum) as complements; locally mixing the leading tone into a pentatonic to fit the harmony — freely adjusting distance between "pentatonicism" and "functional harmony."
The pentatonic tonic may differ from the key tonic (e.g., C major with an E-minor pentatonic = floating, modern color) — a standard way to avoid pentatonic flatness; any major/minor pentatonic built on a scale degree of the key is usable (if it contains an out-of-key tone, drop that tone).
Pentatonics are highly independent of harmony: tones that look like appoggiaturas/passing tones need no resolution and may leap away freely; the same harmony fits many pentatonic phrases and the same pentatonic phrase fits many progressions — "coexisting well with the harmony while acting independently and freely" is the essence of pentatonic melody.
Blue Notes (ブルーノート)
Definition: temporary semitone lowerings of degrees iii, v, vii in major (the "melancholy tones") — not a fixed change to minor-scale tones, but:
Lowered temporarily, or deliberately sung flat with the "true tone" as premise;
The typical usage is a before/after contrast of "original tone → blue note," or ornamental oscillation between the two.
The effect is strongest when the harmony does not contain the blue note (the semitone collides with the chord's diatonic 3rd): blue iii against the 3rd of I, blue vii against the 3rd of V — a head-on collision, the fiercest blues color.
Often packaged as the blue-note pentatonic (ブルーノートペンタトニック) (a minor-pentatonic shape containing ♭iii and ♭vii) used over major; blue v does not join the pentatonic and almost always resolves by semitone step up/down (used as a passing/ornamental tone).
To "dye" the whole song blue: add chords such as ♭III, ♭VI, ♭VII, Vm, I7, IV7 (usable even when the melody has no blue notes; this is a non-functional usage carrying no harmonic function).
7. The Relationship Between Harmony and Melody
Chord Tones and Non-Chord Tones
Chord tones (和音内音): stable, straightforward, plain; but easily buried in the chord, and inevitably leap-heavy, lacking smoothness. Stability ranking: root > 5th > 3rd.
A purely chord-tone melody easily runs parallel to the harmony and gets swallowed by it; but when the harmony moves frequently or contains local modulations, it can actually highlight the harmonic interest.
Non-chord tones (和音外音): unstable, tense, hard to bury; their urge to resolve (basic form = moving by 2nd up or down to a chord tone) is the greatest engine of melodic momentum.
Tension control: non-chord tones a semitone from a chord tone are strongest (hard and fierce, to be used with care); the longer the value and the stronger the beat, the greater the tension.
In pentatonic/blues contexts, modal force takes priority and non-chord tones need not resolve.
Non-chord tones may be used consecutively; other chord tones may be interposed mid-resolution; unresolved / irregularly resolved non-chord tones = subtle feelings, an ineffable afterglow, inorganic expression (also a resource, but too long or too strong harms the sense of harmony).
Definitions and resolutions of the seven types of non-chord tones (passing tone / neighbor tone / appoggiatura / anticipation / escape tone / suspension / suspension-appoggiatura) → reference.md Table 1.
The more chord tones a chord has (triad → 7th → 9th → 11th → 13th) → the fewer non-chord tones remain → the freer the usable tones, but the blander each tone's relationship to the chord, losing the delicious mechanics of non-chord tones.
The chord-tone/non-chord-tone status of the same melody changes with the chord symbol (under C / CMaj7 / CMaj7(9), the identities of B and D differ) → writing a melody requires simultaneously imagining the chord sounding with it;
The whole song needs a plan for "how many chord tones, and where and why the count changes" — a 9th chord suddenly appearing in a triad-dominated song = the forces at that spot lurch, unbalancing the structure.
Melody within Functional Harmony (機能和声)
Function groups and progression rules (T→anything; D must return to T, D→S forbidden; S = buffer between T and D) → reference.md Table 2.
Stock progressions at the instant of the chord change: leading tone → tonic (semitone up), degrees iv/vi → down a 2nd, dominant degree → hold or up a 4th — following them = a strong sectional/cadential feel; within the dominant chord itself, arpeggios and steps are free.
The melody should avoid implying "forbidden progressions" (e.g., a melody strongly implying V→IV) — done deliberately, it yields a non-tonal, modal effect.
Yotei-chōwa alert: a melody traveling with functional harmony (especially tones over the dominant chord) is extremely easy to predict — break textbook resolutions with suspensions, appoggiaturas, and anticipations.
A section may cadence directly on the dominant chord, and the next section need not start from the tonic chord (a strong sectional break permits harmonic discontinuity, e.g., after a G7 section, restarting straight from F).
Melody within Non-Functional Harmony
Modes without a leading tone (natural minor, pentatonic) often pair with non-functional harmony: chords planted directly against melody tones, parallel motion, harmony that doesn't progress and serves only as rhythmic accent, chains of 4ths (lacking the dominant→tonic thrust).
Weak momentum → songs easily go flat; countermeasures: a combined tactic of rhythmic-density changes, register leaps, pentatonic ⇄ scale switching, appoggiatura resolutions, and locally borrowing V7→I to punctuate sections.
Mutual Independence of Melody and Harmony
Melodic motion and harmonic motion need not coincide: melody stalls while harmony moves, they move in opposite directions, the same melody over different harmonies — all are sources of three-dimensional depth (exploit them actively; they are not errors).
A melody that traces (なぞる) the harmony = flat and monotonous (occasionally useful for a unified sense of power).
At least one of the bass (lowest voice) and the melody should leave the root: melody constantly coinciding with the root = stable but sonically impoverished; change the chord's bass note (inversion) or have the melody avoid the root.
Chords First or Melody First
The trap of fixing a complex progression first: the melody degenerates into "walking along the harmony"; harmonic rhythm becomes formulaic (one chord per bar, dominant at the end of 8 bars); transplanting a genre's typical progression ≠ obtaining that genre (characteristic progressions exist bound to rhythm, melody, and position in the form).
Practice path: freely invent melodies over 1–2 chords → over a 3–4-chord loop progression (play the loop until your body memorizes it, then "hold the harmony in mind while inventing melody") → once fluent, invent even complex progressions in relation to the melody.
Conclusion: melody and harmony are parallel temporal structures that should be conceived simultaneously and interlocked tightly — neither takes precedence.
8. Fitting Lyrics and Melody (歌詞と旋律)
Intonation (イントネーション)
Principle: the melody's rises, falls, and stresses match the word's pitch accent — otherwise meaning becomes hard to catch or is heard as a different word (「雨が晴れて」 "the rain clears" → 「飴が腫れて」 "the candy swells").
Pitch accent follows standard Japanese (標準語) (standard ≠ Tokyo dialect; when unsure, consult an accent dictionary アクセント辞典).
Deliberately defying the accent = a linguistic hook, expression of large emotion — but adjacent words must have correct accent to preserve the sentence's meaning; setting particles on high notes can build emotion in rising steps.
Chōon (長音, ー, long vowel): give it a long note value (or re-articulate for emphasis); stretching the wrong place changes meaning (「彼が好き」 "I like him" → 「カレーが好き」 "I like curry") — align the stretched note with the bunsetsu (word-phrase) boundary.
Sokuon (促音, っ, geminate stop): express with a rest; or lengthen the preceding vowel, then cut it off short and strong (「はしーぃって」).
Hatsuon (撥音, ん, moraic nasal): merge into the previous syllable, or give it a short, weak independent note; a long strong-beat 「ん」 is hard to sing and easily buried (unless you make it float by taking it off the chord tones).
Devoiced vowels (無声音) (devoicing of u/i after t/k/f/s/p, e.g., the く of 「たかく」): no perceptible pitch — must get a short note, otherwise the phrase's musical meaning changes (the effect of figure repetition is lost).
Consecutive identical vowels (「こと＋おもってる」): insert a rest between the two notes, or change pitch, to prevent them merging into one long tone that swallows the second vowel.
Segmentation and Emphasis
When lyric bunsetsu and melodic segmentation coincide, comprehension is easy; misalignment changes meaning (「君が来るまで待ってる」 "I'll wait until you come" → 「君が車で待ってる」 "you're waiting in the car").
Uses of deliberate misalignment: breaking before a particle to suspend interest; cutting mid-word to create tension (「わ・た・し」 syllable-by-syllable insistence); stretching the phrase head as tame to induce expectation — use only where word meaning survives.
Letting the musical phrase override linguistic segmentation = a comprehension burden on the listener → reverse-used as the obsession and earnestness of "singing it out even so"; all-equal 8th notes + preserving word boundaries by pitch motion alone = a single-minded (一途) devotion.
Lyrics-First and Music-First (詞先・曲先)
Shisen (詞先, lyrics-first): writing only the catchphrase at the head of the sabi (サビ, chorus/hook) lyrics-first is already effective; even merely reserving positions for sokuon words (「ずっと」「もっと」「きっと」) is effective. Give the words to be highlighted long values, high notes, and kime positions.
Four repairs for kyokusen (曲先, music-first):
More syllables than notes → split note values or add notes before/after;
Fewer syllables than notes → merge note values;
Counts match but rhythm doesn't fit → change values/rhythm;
Contour conflicts with pitch accent → adjust pitches while minding segmentation.
Don't finish the melody to the last detail in advance — keep room for revision, and iterate back and forth with the lyrics; take care not to destroy repetition/contrast structures to accommodate words — when necessary, go back and change the lyrics.
Ji-amari / Ji-tarazu (deliberate mismatches)
Ji-amari (字余り, too many syllables): words crammed into short values, patter-like → a breathless desperation; compressing the notes at the phrase head = a forward-toppling surge; repeating that rhythm = step-by-step advancing earnestness. Overdone, it becomes strings of equal values, a speech-like melody; if the emphasized word earns no sympathy it slides into comedy.
Ji-tarazu (字足らず, too few syllables): words stretched, articulated over multiple pitches → arrests attention, pouring in deep feeling; stretched too far = word meaning dissolves into scat (スキャット, meaningless vocalise).
Rhyme and Rhetoric
Rhyme (韻): alliteration/end rhyme; the most natural setting keeps the same rhythm and beat position at rhyme points while varying pitch motion (rhyme automatically brings rhythmic neatness; varied pitch keeps flow and avoids stiffness); identical melodic figures at rhyme points = strong formality; rap = near-speech in non-rhyming parts + leaps in and out at the rhymes.
Beware shichigo-chō (七五調, 7-5 meter), san-san-nana byōshi (三三七拍子, 3-3-7 clapping meter) and other Japanese stock rhythms slipping in unconsciously (over-familiar to Japanese listeners; check especially when writing rap).
Parallelism (並置): sequences of like words/phrases → melodic rhythm repeats with pitch climbing stepwise to drive the climax.
Inversion (倒置): be explicit about "which word this inversion emphasizes" → charge up with a rest-tame before the key word, or delay the key word by one bar; a catchphrase can be paired with unison (ユニゾン) accompaniment.
The conjunction between lines steers the melodic trajectory:
そして/だから (additive/causal) → the melody continues and repeats in understated narration; or receives it grandly, lifting the register into a "cause-and-effect climax";
だけど (adversative) → change the melody's character, a large sectional break (narrow-range stasis vs. wide-range leaps in opposition); or connect blandly with the same character = an objective, bird's-eye view;
ところで (juxtaposition) → still keep some relationship between the two lines (same character = implying a bird's-eye view; rising vs. falling response = implying cause and effect).
9. From Motive to Phrase to Section: Principles of Construction
The Hierarchy of Terms
Partial motive (部分動機) (cannot stand alone but has a distinguishable feature; ≈ a word; labeled a1, a2, with ' for variants)
→ Motive (動機, dōki / モティーフ): the smallest melodic unit that can stand alone, at least 1 bar (shorter, and its beat position cannot be judged independently); ≈ a bunsetsu; labeled a, b; on repetition it usually inherits the rhythmic profile and repeats with variation.
→ Small phrase (小楽節, shō-gakusetsu): usually 2 motives; labeled A1, A2.
→ Phrase/period (楽節, gakusetsu): most often 8 bars, 4 bars also common (other bar counts are mostly additions to or omissions from 8 or 4); a whole unified by one mood (ムード); labeled A, B.
→ Setsu (節): the strophe corresponding to verse 1 / verse 2 (strophic form, 有節形式).
Different motives may share partial motives (e.g., a'' = a1' + b2) — an important means of structural unity.
The character of identical material is determined by context: the same motive is a small climax after a static phrase, but a calming answer after a soaring phrase.
The Six Causes of Segmentation (区切り), by Clarity
Long values and rests (clearest);
Interval leaps (severing pitch continuity);
Repetition of a motion pattern (each repeated unit forms its own segment);
A clear cadence (a definite ending = strong segmentation);
A clear beginning (retroactive recognition: only when the next phrase's start is recognized is the previous segment confirmed as ended);
Segmentations of 2, 4, and 8 bars are easiest to grasp; odd-bar groupings (3+3+2, etc.) are hard to perceive.
Strong segmentation (tonic + strong beat + tonic chord + long note/rest) = a sense of fresh start — good for opening a new section and reinforcing repetition/contrast; but it chops the song in half, so use cautiously anywhere except section ends.
Weak segmentation = smooth continuity, sustained atmosphere; can blur structure or create a sense of stasis.
The Four Principles of Construction
Response (応答, ōtō): given two segments, the latter is heard as a reaction to the former (the most basic engine; need not be a literal Q&A).
Like phrases responding = echo type; contrasting phrases responding = continuation type; the "answer" may even come first (the stable form first, the question after);
Paired (ペア) responses are most stable but yotei-chōwa; juzu-tsunagi (数珠つなぎ, bead-chaining) chained extension develops richly but risks sprawl;
Responses run in parallel at multiple levels of motive/phrase (a motive responds both to the immediately preceding motive and to the whole previous phrase).
Repetition and contrast (反復と対照): a response of like character = repetition; of unlike character = contrast.
Each element is judged separately: rhythm repeated + pitch contrasted (or vice versa) is the most common intermediate state;
Repeat at most twice; from the 3rd consecutive occurrence of the same phrase, listener interest drains away — variation/development is mandatory;
Weak repetition (few shared elements, heavy variation, e.g., repeating only the rhythm pattern) escapes the "boring by the 3rd time" rule — even 4+ repetitions stay fresh;
Conditions for contrast: the preceding phrase's character is vivid enough + the following phrase's character is opposite; multiple elements reversed simultaneously = strong contrast; different-but-similar characters = a contrast-free "unified mood" (usable in quiet sections);
Tension accumulated by multiple ascents can be resolved at once by one sufficiently large descent; deliberately answering with a small descent = energy not fully spent, passing lingering regret (心残り) on to what follows;
Summary: melody = a craft woven from repetition and contrast at various strengths — distribute the two to produce "continuation feel" and "sectional feel."
Recurrence (再現, saigen): repeating a phrase at a distance of a full phrase-unit or more — "the protagonist returns from a journey"; the same phrase seems changed in character after time has passed.
Preconditions: the original phrase was well memorized (repeated earlier) + the middle section is substantial ("the journey was full enough");
Setting the Climax (クライマックス)
Position: usually at 2/3–3/4 of the segment, before the phrase's resolving tone (matching the golden ratio and ki-shō-ten-ketsu); the head-climax type also exists.
Size scales with the containing unit, nesting layer by layer: a motive has its apex tone → a phrase's climax = one particular motive → a setsu's climax = one phrase, and that phrase is called the sabi; within the sabi, the climax again falls in the latter part per phrase-level rules.
Half the effect of a climax depends on how it is wound down afterwards: when a climax feels thin, first revise the endings of the motives/phrases (add a descending close after the apex, magnify the phrase's resolution, rework the ending to carve a clear apex) — often that alone rescues the section.
Purpose: preventing shiri-subomi (尻すぼみ, petering out / anticlimax) and flatness; even a sabi that opens high must be checked for a sufficient climax in its second half and a substantial close (avoiding a dragon's head with a snake's tail, 竜頭蛇尾).
The Principle of Section Count (full table in reference.md, Table 7)
One-section construction = the repetition principle: [conclusion](−conclusion…); dance-music riffs, counting songs; repetition seeping into the unconscious = heightened corporeality (a repeated coda chant triggering sing-along).
Two-section construction = the response principle (most basic): [introduction]−[conclusion]; climax in the latter section or just before it.
Three-section construction, two types:
Type 1a a−a'−b (outset−development−resolution): the second section is a varied intensification of the first; stable;
Type 1b a−b−c (outset−another viewpoint−integrated resolution): this is the A−B−sabi construction most common in Japanese pop;
Type 2 a−b−a' (ternary form) = the recurrence principle: the climax is in the middle section b; b must be more emotional, differing in mood from its surroundings; the final section is usually varied to gain finality.
Four or more sections = compounds of two/three-section forms; using all-different motives risks sprawl → share partial motives; expanding three into four = repeat the first or last section (repeating the middle is rare).
Temporal principles: ki-shō-ten-ketsu (起承転結) (ki = introduction, shō = repetition and stabilization, ten = large contrast and climax, ketsu = conclusion) and jo-ha-kyū (序破急) (slow intro → a climax section that articulates time → crisp close) — both are typical paradigms, not absolute rules.
The two master questions of construction design: how many sections, and in which one the climax sits; through what ladder of characters the climax is reached. Universal device: give weight to the opening (出だし) of each phrase, and use it as the material of contrast.
10. Song Forms and Whole-Song Assembly
The Two Great Setsu Forms
Verse–chorus form (ヴァース＝コーラス形式 / メロ＝サビ形式):
Verse (ヴァース / hira-uta 平歌 / A-melo Aメロ) = introduction and narration: typically narrow range + short values, katari (speech-like); a distinctive hooky figure can also open;
Chorus (コーラス / sabi サビ) = the setsu's climax: the catchphrase type of short repeated motives (common in rock) or the long-lined utai-age (soaring delivery) type (the Japanese-style sabi, ballads);
Pillars = the contrast principle (the gap between verse and chorus = explosive release) + internal repetition;
Extensions: A−A' −B (repeated verse); A−(bridge ブリッジ)−C (the short Western connective section, raising register while heating up); A−B (B-melo Bメロ)−C (sabi) (the J-Pop staple: a second verse of different character); dance type A×n−B.
Ternary form: A (melo)−B (sabi)−A (A') — sabi in the middle, final section a recurrence; B unfolds A's content more emotionally / from a wider vantage; returning to A = deep realization and feeling; extension A−A'−B−A'' = textbook ki-shō-ten-ketsu.
Patterns of expansion from A into sabi: narrow-interval katari → wide-interval expansiveness; little motion → syncopated busy rhythm; low register → high register; short chords → long chords.
Workflow: make the sabi (C) and the introduction (A) first, and the B (middle section) last.
Endings of Setsu and Phrases (節や楽節の終わり方)
How a phrase ends directly determines how the next phrase begins and the logic between them; how the setsu ends governs the whole song's aftertaste (後味).
Closing (閉じて終わる) = full cadence on a stable tone: the statement is finished; the next section "starts anew." Opening (開いて終わる) = ending on an unstable tone: leaving resonance and lingering feeling (余韻); the next section "continues speaking."
The eight-degree openness sequence (three axes: tonic → 3rd → leading tone × tonic chord → non-tonic chord → dominant chord × strong beat → weak beat via suspension/anticipation) → reference.md Table 5.
Checklist of closing devices: append a closing phrase (終結フレーズ); repeat the final 2 bars (driving the point home, 念押し); repetition + augmented values (extremely strong cadence but easily shitsukoi/overbearing); deliberately append an incomplete cadential phrase for afterglow; neither melody nor harmony cadences (maximum afterglow); harmony-only cadence ("never quite finishing the sentence"); cut the cadence and charge straight into the next phrase (prevents dragging but sounds impatient).
Kime (キメ, decisive tag): a flag-waving short phrase (~2 bars) appended to the sabi ending, often with the accompaniment stopping dead or a signature rhythm, crisply singing out the title/keyword; if built from a variant of the sabi's motive, unity is strong; a kime grown to phrase length = effectively a double chorus. When the song lacks meri-hari (メリハリ, dynamic contrast/edge), consider adding a kime.
Assembling the Whole Song
Norm: 2–3 setsu, 3–5 minutes; slow songs reduce setsu count, fast songs increase it; the only criterion = listener satisfaction.
Omit phrases in the latter half to defeat prediction: verse 2 no longer repeats A (A-A'-B-C → A'-B-C); verse 3 starts from the sabi run-up only (B-C).
Repeat the sabi in the last setsu as the whole-song climax (A-B-C-C, even C-C-C-C); optionally with a modulation lifting the register.
Atama-sabi (頭サビ, chorus-first opening): open directly with the sabi — verbatim = a splashy opening; thin accompaniment + slow tempo + rubato = a katari-style introduction.
Pre-verse introduction: a recitation-like quiet opening (mid register + nearly static pitch + simple rhythm + barely moving harmony) or its opposite, a high-flying opening (leaps + high register).
Ōsabi (大サビ, the added bridge between repeated setsu): a transitional section inserted between setsu repetitions — its character ≈ an introspective second B-melo, not a bigger climax; often feels half-time, low murmured asides; it must be followed by the sabi (after the calm, the sabi lands more indelibly). Not obligatory — an ōsabi added out of habit actually saps momentum.
Coda (コーダ), four types: the grand-finale broad-melody type (= one more sabi at the song's end) / the pre-coda modulation type (receiving the whole song and sealing it) / the rubato·rit. calming introspective type / the A-melo recurrence type (slower tempo, thin accompaniment, repeating the opening material = a profound close by the recurrence principle).
Intro (前奏) = the instrumental version of the introduction (usually borrowing A's or the sabi's harmony with a new melody; when the song already has a katari pre-verse or atama-sabi, the intro is often minimal or absent); deliberately no intro: a pickup melody entering on bare voice = raw and immediate (生々しい).
Interlude (間奏) = the Second Introduction (verse 2's intro); interludes grow longer toward the song's latter half; often also an instrumental repetition of the setsu's final phrase.
Outro (後奏) = the instrumental version of the coda; after it, another sung coda can still follow. Vamp (ヴァンプ) = a section presenting only the rhythm pattern, no melody.
Designing the empty space: the melody need not carry everything — where the melody rests, listeners' attention turns to the accompaniment's rhythm and harmony; the build-up before the sabi can be delegated to the interlude/accompaniment; thinking about "the parts not sung" naturally clarifies what arrangement is needed.
11. Avoiding Cliché (予定調和)
Yotei-chōwa (予定調和, predictable resolution / cliché predictability) = the listener reads the next move, everything unfolds as expected — the problem melody writing must guard against most.
Checklist of high-risk patterns:
The same phrase repeated 3+ times without change;
A motive shifted in parallel along the chord progression (by the 3rd chord the listener has read the whole thing);
Formulaic harmonic rhythm: one chord per bar, dominant at the end of 8 bars;
Verse 2 copying verse 1's A−A' opening (the listener instantly predicts B−C must follow);
The long-short-long-short skip rhythm maintained throughout;
Textbook leading tone → tonic and iv → iii resolutions played straight.
Countermeasures:
The 3rd repetition must vary; in verse 2 omit the opening repetition (A-A'-B-C → A-B-C);
Exploit expectation and turn the tables: first induce the listener's prediction, then at the predicted point betray it decisively (大きく裏切る) — staging the climax with surprise;
Stretch the melodic line: a long arc not chopped up by motives (use anticipations to bind the rhythmic shapes, dive deep → the rebound ascent becomes the climax → a strong semitone appoggiatura as a second center of gravity → an unhurried close) — a long line = a sense of direction + a relay of varied motion devices, not mere absence of breaks (which would be dara-dara, dragging);
Keep materials to a minimum: any new phrase must clarify "what differs from before, what continues" (part contrast + part repetition); songs with less material are more comprehensible and more moving;
Break stock resolutions with suspensions/appoggiaturas/anticipations;
Between "banal" (copying stock devices — reassuring but worthless) and "unintelligible" (straying so far from shared forms it becomes a wall), support originality (オリジナリティ) with strong logic internal to the work.
Distortion + answer: all-smooth (スムーズ) = boring; you need an ibitsu (いびつ, warped/distorted) spot to stir the waters — but the distortion must be answered (a moving resolution after a run of unstable tones; an accent-defying leap = the endorsement of strong emotion), otherwise it is merely a defect.
Earning trust: once earlier ingenuity convinces the listener "this song is worth listening to," even a flat phrase will be waited out patiently for "the good answer coming later."
The ultimate standard: a 3-minute song consumes 3 minutes of the listener's life — .
12. The Creative Workflow (from Spark to Full Song)
Conceive the Whole First
Section-by-section accretion ("waiting for the next good phrase to descend") jams up as constraints pile layer on layer: the second section is constrained by the first, the third doubly so… classic symptom = a gripping opening + a fizzled ending.
The first fragment that surfaces is a good thing (it is the very motive force of composing), but soon afterwards answer: where does this song go? What development toward what climax? What is the theme? What about the introduction?
The four items of conception:
Musical: is there a use of tones "only this song has" (exceptionally fluid / exceptionally soft / one exquisite point of unease)?
Content: what is sung, what drama (ドラマ) — the subject must be something everyone has experienced (empathizable), the telling must be one-of-a-kind (extraordinary); private imagery must be translated through public associations (a mayfly = evanescence works; an insect nobody knows = hitori-yogari 独り善がり, self-indulgence);
Overall mood (全体のムード): ultimately cashed out in concrete elements — tempo, rhythm pattern, key and mode, degree of functional-harmony usage, degree of tension-note stacking, register and voice quality, whether drums/rhythm section are used;
Extraordinary yet empathizable.
When the song drifts from the initial conception mid-writing, trust your musical instinct and go back to revise the conception — "the process of constructing" and "the process of the content being polished" run simultaneously; that is the delight (醍醐味) of composing.
Plot (プロット) Thinking — Narrative Arrangement
Beyond fixing the theme, decide from what standpoint and in what order to tell it: is heartbreak stated directly, or set off against past happiness? Revealed at the start or only at the end? First person / general statement / role-play?
The essence of song form = "a plan for deriving the sabi": a wide-range, high-sung sabi heard in isolation = contrived (作りものっぽい); as the result of layered preparation = enormous emotion. The reverse strategy (the abrupt allure of an atama-sabi, returning to narration after the sabi) is equally valid.
Practical Procedure (Appendix 01; motive-construction table in reference.md, Table 6)
Make partial motives — start from any one of these entry points (when the whole is too much to hold, pick one and focus):
Centered on one pitch (repeated notes + rhythm/syncopation; add tones above/below to fit the accent) — unaffected, honest, intimate;
Approaching (アプローチ) a target tone: scalewise approach (1–2 short neighbor lead-ins) = fluent and forthright; chord-tone leap approach = forceful; multiple approaches (from above and below) = roundabout but rich;
2 partial motives → a motive: repetition (varied repetition is the norm) / contrast (invert the single most characteristic element) / continuation (shiritori linking is smoothest).
2 motives → a small phrase: the same three options; note that the final motive is usually not repeated verbatim (to defeat prediction).
Small phrases → a phrase (楽節):
First fix the cadence plan: setsu-final phrase = a strong cadence with motion fully stopped; mid-setsu phrase = a weak cadence keeping expectation; when the next section's start is unmistakable, no cadence is needed;
Set the climax at 2/3–3/4;
Choose the motive construction: a a a b (repetition + a swapped final motive carrying the cadence), a a b a (ternary recurrence, suits the sabi), a a b c (ki-shō-ten-ketsu type), a b a b (catchphrase shuttling), a b a c (a hooks + c is the grand climax), a b c d (needs shiritori / a rhythm pattern as safety net), a b c a (final recurrence closing).
Phrases → a setsu: A / A−B / A−B−C / A−B−A constructions; for expansion, prefer "phrase repetition" over adding new material (too many elements → the road to the sabi meanders, focus dissipates); when meri-hari is lacking, add a kime.
Setsu → the full song: sing it 2–3 times through; the opening approach (intro / bare-voice no-intro / atama-sabi); interludes (growing longer in the latter half); ōsabi or not (judge whether it is truly needed); three ending options (cadence right at the setsu's end + repeated sabi / returning to the opening material in recurrence / a newly added coda).
Timing of Words and Music
Ideal = conceived simultaneously; reality = like "whole ↔ part," words and music polish each other in shuttling passes (行きつ戻りつ).
Even music-first, pre-place words at key positions (the melody's opening, the sabi's kime spot; single words suffice) — the melody taking shape "worm-eaten" around embedded words is a common and realistic method.
Lyrics-first: first sort which passage is the A-melo and which the sabi, and give the words to be highlighted long values, high notes, kime positions.
13. From Humming to Notation (Transcription and Reverse Engineering)
The essence of notation: re-visualizing the background (beat/key/harmony) that was already sounding in your head when the melody was born — not an external analysis; the endpoint of training = "conceiving melody and accompaniment simultaneously."
Determining beat placement:
Sing the first ~2 bars while clapping and counting → establish the basic beat and "what note value equals one beat";
Find the next strong-beat location, and verify the meter hypothesis with a pre-count (プリカウント);
Don't assume "phrase start = beat 1" — pickups (アウフタクト) and offbeat (ウラ) openings are common;
When the same rhythm pattern recurs, it very likely falls on the same beat positions (an important proofreading clue for transcription);
"The meter changed midway" usually means you over- or under-counted 1 beat; the vast majority of songs are in 4 or 3 (otherwise 6/12).
Determining the key:
Play a "landing," settled tone right after the phrase = the tonic;
From the tonic, play the major/minor mode and see which fits; then verify with the tonic chord (major triad settles = major key);
All white keys ≠ C major / A minor; theoretical aids = the principles of avoiding augmented/diminished intervals + fewest accidentals in the key signature;
Beware: minor-key temporarily raised vi/vii, enharmonic spellings (ミ = ファ♭ etc.), and the shadow of blue notes (iii/v/vii lowered a semitone + melancholy expression).
Four steps of reverse-engineering the harmony (detailed decision table in reference.md, Table 11):
Fix the beginning and end first: song/section endings are almost always the tonic chord I; the opening = I or a reverse-cycle (逆循環) start (beginning from the 2nd or 3rd chord of a "loop progression starting on the tonic"; common major-key openings II/III/IV/VI; in pickup songs, the I falls on the first full bar after the pickup); section endings often settle on V, the dominant of the next section's opening chord;
Find the bass note on strong beats (chord changes mostly fall on strong beats; discovering a "weak-beat chord change" → suspect instead that the melody's beat placement was transcribed wrong);
On the bass note, stack every-other scale tone in 3rds to form the chord; if it doesn't fit → consider inversions (転回形) (bass = 3rd/5th/7th; highly likely where the bass moves stepwise);
Refine chord-tone/non-chord-tone judgments: phrase-final tones ≈ chord tones; opening tones not necessarily (often appoggiaturas/suspension-appoggiaturas); the starting tone of a leap of a 4th or more = a chord tone; a long tone across a chord change = anticipation or suspension; consecutive steps = passing/neighbor tones; consecutive leaps = arpeggio (all chord tones). Note that pentatonic melodies are independent of harmony, so this method applies only partially.
14. Quick Diagnostic Table (check here when the melody has problems)
Symptom
Check and prescription
Melody bland and boring
All steps / chord tones / on-beat? → add non-chord tones (appoggiaturas/suspensions), syncopation, leaps + tame; check whether pentatonic density and semitone usage stay unvaried
Every next move predictable (yotei-chōwa)
Repeated 2+ times unchanged? Motive shifting parallel with the chords? → vary the 3rd time, omit repetitions, or induce-then-betray for the climax
Sprawling, nothing to hold onto
Too many materials, all-new motives in a row? → share partial motives, keep repeated elements in new phrases; pull the range into ~1 octave; segment in 2/4/8 bars
Strong start, weak finish (尻すぼみ)
Climax position and wind-down? → put the climax at 2/3–3/4, revise how the motives/phrases end, reinforce the climax in the sabi's second half
Hard to sing
Leap > a 6th? Same-direction leaps > 2 or total span > an octave? No duration at either end of a big leap? Crossing the passaggio? Augmented 2nd? → split the 7th (3+5/4+4), add tame/run-up/landing, assign registers per phrase
Melody buried by the accompaniment
Opening on the strong beat together with the bass? Melody tracing the rhythm pattern? Glued to the beat throughout? → offset the opening, yield to / answer the backbeat, leave gaps
Lyrics unintelligible / misheard
Accent defied? Segmentation misaligned? Devoiced vowel on a long note? Identical vowels in a row? → correct to standard-Japanese accent, segment with rests, give devoiced syllables short notes
Sounds like an instruction manual / school anthem
Leading tone → tonic and other functional moves too blunt? → break them with suspensions/appoggiaturas, mix in pentatonics, avoid the strongest cadence (tonic + I + strong beat) at phrase ends
Pentatonic melody has no personality
Only walking the easy neighbors? → mode-shift (転旋法), offset the pentatonic tonic from the key tonic, insert big leaps/arpeggios to dilute, add kobushi semitones
Ending lacks finality
Duration stopped but mode/harmony didn't? → land on tonic + I + leading-tone approach = strongest cadence; for afterglow do the reverse (see the eight degrees of openness)
Whole song monotonous
Every section starting from the tonic chord? Interludes all the same length? → start some section on a reverse cycle, omit phrases in the latter half, repeat the sabi in the last setsu + modulate
An unprepared large leap = a sudden emotional outburst (filling in intermediary tones makes it smooth but destroys the explosive force);
Not resolving a leap in the opposite direction but continuing the same way = a special sense of setsunai (切ない, aching bittersweet heartache);
Consecutive leaps with thin internal connection = disembodied lightness, an inorganic pop feel;
Displaced semitones with no perceptible scale affiliation = artificiality, detached irony;
Augmented intervals (増音程, zō-ontei) = a non-Western feel.
Master formula: embed local unnaturalness as a "hook" (引っかかり), then finally resolve it in a natural way.
Weakness: only 5 tones invite cliché, and no semitones means emotional scarcity — you can write in fixed kobushi (こぶし, melismatic pitch inflection)-style semitone turns (overuse becomes garrulous).
Diluting the pentatonic feel: insert large leaps or arpeggio-like motion. In practice, regulate "pentatonic density" phrase by phrase (labeling each phrase as ~pentatonic (strong)/(weak)/intermediate); even when not using a pentatonic, always ask how pentatonic the current melody feels.
Pentatonic cadences: the tonic = the strongest cadence tone, dominant next; arriving via a pentatonic neighbor or an ascending 4th is stronger; the precondition for a strong cadence = the pentatonic tonic coincides with the key tonic; otherwise only soft cadences are available.
Carries the persuasive closure of fu ni ochiru (腑に落ちる, sudden deep understanding) → best placed at the end (A-x-x-A); recurrences elsewhere also tighten the material and give one phrase multiple meanings.
Continuation and shiritori (継続としりとり, word-chain linking): the last 1–several tones of a phrase reused as the next phrase's opening → smooth connection by the continuation principle, segmenting without severing.
The more tones shared, the more it sounds like repetition (shiritori at partial-motive scale is heard as repetition);
Weak adhesion — long sections built on shiritori alone sprawl; combine with repetition/contrast.
every moment must be full, and finishing it should feel like having lived through a good experience
Escaping C-major fixation: add key signatures one at a time (G/F → E/D minor), targeting 3 sharps and 3 flats each (the seven tonics C, D, E♭, F, G, A, B♭ fully covered; phrases in any register become writable).
Building a harmony track first in the DAW = good practice approximating "simultaneous conception," but use a simple loop progression and practice letting the melody move freely — a complex progression reduces the melody to "describing the chords."
