Japanese Pitch-Accent Minimal Pairs: The Drill List You Must Hear

Japanese pitch-accent minimal pairs are word pairs with the same kana spelling and mora count. They differ in where the High-to-Low pitch drop sits, so they carry different meanings.¹ If you have met the four Tokyo patterns and the 0/1/2/3 notation, this article is the drill catalog those pattern pages point to.

Overview

A pitch-accent minimal pair shows that pitch in Tokyo Japanese is contrastive: two words use the same kana, but differ only in the position of a single H→L fall, or in whether that fall is absent.¹ The Tokyo system is constrained by culminativity, meaning every word carries at most one drop. The contrast in a minimal pair is therefore always about where that single drop is, or whether it exists at all.¹

This catalog stays within the Tokyo NHK standard throughout. When dictionaries disagree on a headword, the disagreement is flagged in place rather than smoothed over. The regional-variation article covers the Kansai and Keihan inversions that affect several of these pairs.

What counts as a minimal pair here

A true pitch-accent minimal pair is two words with the same segmental form and mora count whose only contrast is the location of the H→L fall: Tokyo Japanese "distinguishes /haꜜsi/ 'chopsticks' from /hasiꜜ/ 'bridge' and /hasi/ 'edge'."¹

A near-minimal pair, as used in this drill list, matches the sound string and the mora count, but one member falls outside the strict Tokyo-NHK frame. Most often, this is because dictionaries disagree on its accent class. 柿 versus 牡蠣 is the standard case: NHK 2016, Daijirin, and Shin-Meikai 5 all place 柿 at [0] heiban and 牡蠣 at [1] atamadaka, but Wikipedia's pitch-accent article gives the opposite polarity.²³⁴⁵

Tokyo standard, regional variation deferred

Everything in this article follows the NHK 2016 Tokyo standard. Several of these pairs (はし, かき, the verb 切る) invert in Kansai and Keihan. That detail belongs to the regional-variation article, not this one. The JLPT and national broadcast media remain Tokyo-anchored, which is why Tokyo is the right starting point.⁶⁷

How to read each drill entry

Every drill entry gives the same six pieces of information: the kanji, the kana reading, the accent number (0–3), the surface H/L contour over each mora, the contour with the particle が attached, and a one-line tip for telling the words apart.

The accent number names "the mora after which the H→L drop occurs"; 0 marks the no-drop pattern.¹ The surface contour is the F0 pattern, or pitch track, after Tokyo's initial-rise rule has applied: "the first two syllables in a word bear a LH tonal sequence, sometimes known as initial lowering or initial rise, unless the first syllable is accented."¹

The particle column helps you avoid the citation-form trap: "monomoraic and bimoraic finally-accented words (e.g., /kaki/ 'fence') and unaccented words (e.g., /kaki/ 'persimmon') are phonetically very similar, if not identical, when they appear in isolation."¹ Adding the particle makes the difference audible.

Form, notation, and the particle-attachment test

Pattern recap in one table

The four Tokyo patterns each produce a distinctive surface contour and a distinctive behavior on a following particle. The table below gives a compact recap. The four pattern articles cover each one in depth.

Pattern	Accent number	Word-internal contour	Particle behavior
Atamadaka (頭高型)	[1]	H L L L	Particle stays L1
Nakadaka (中高型)	[2] / [3] / … (drop inside)	L H … Hꜜ L … L	Particle stays L15
Odaka (尾高型)	accent = mora count	L H … H	Drop lands ON the particle5
Heiban (平板型)	[0]	L H H … H	Particle stays H15

The four-pattern inventory itself is from Kawahara's 2015 survey: "four distinct accent patterns: accent can fall on any of the n-th syllables, and there can additionally be an unaccented word."¹

Why the particle matters: heiban vs odaka

The citation-form trap is the main reason every drill row in this article includes a particle column. Kawahara states it directly: "the distinction between finally-accented words (e.g., /kaki/ 'fence') and unaccented words (e.g., /kaki/ 'persimmon') are phonetically very similar, if not identical, when they appear in isolation (Vance 1995; Warner 1997)."¹

Wikipedia gives the same point for はし: "the sequence hashi plus the subject-marker ga can be accented on the first syllable or the second, or be flat/accentless: háshiga 'chopsticks', hashíga 'bridge', or hashiga 'edge'."⁵

For odaka the drop lands on the particle; for heiban the particle stays H.⁵ Two-mora words drilled alone (橋 and 端, 鼻 and 花, 柿 and 牡蠣) collapse into near-identical LH contours; only the particle pulls them apart.

Drill the particle, not the bare word

The most common minimal-pair drilling mistake is to repeat 橋 and 端 alone, in citation form, and call it ear training. Both surface as LH in isolation. Always say 橋が LHꜜL versus 端が LHH; the contrast is on the particle, not in the noun.¹⁵

Sources used for accent values

Every accent integer in this catalog follows the NHK 2016 accent dictionary (NHK16 in Wiktionary's acc_ref field),⁶ cross-checked against Daijirin (DJR)⁸ and the Shin-Meikai 日本語アクセント辞典 5th and 8th editions (SMK5 / SMK8).⁹ For historical claims, Kindaichi's 1981 Meikai Akusento Jiten table, reproduced at sci.lang.japan FAQ, is the secondary reference. It is cited explicitly where it contradicts NHK 2016.²¹⁰

OJAD (the Online Japanese Accent Dictionary, from the University of Tokyo's Minematsu and Saito laboratories) is the recommended audio source for every headword below: "look up Tokyo dialect accent for approximately 9,000 nouns" and "examine accent patterns for roughly 3,500 predicates across multiple conjugation forms."¹¹

The core drill catalog: high-frequency pairs

These are the pairs every serious learner should be able to hear before claiming a working ear for Tokyo pitch. They are organized one headword family at a time, with a per-mora contour and a particle row for each member.

はし: 箸 / 橋 / 端

This is the textbook three-way contrast and the cleanest demonstration of the pattern roles in a single kana string.

Word	Reading	Accent	Citation contour	With が
箸 (chopsticks)	はし	[1] atamadaka	HꜜL	HꜜL L12
橋 (bridge)	はし	[2] odaka	LH	LHꜜL13
端 (edge)	はし	[0] heiban	LH	LHH14

Wikipedia's verbatim Tokyo summary lines up exactly: "háshiga 'chopsticks', hashíga 'bridge', or hashiga 'edge'."⁵ Kindaichi's 1981 table preserves the same layout: "箸 (chopsticks) HL / 橋 (bridge) LH(L) / 端 (edge) LHH."²¹⁰

One-line drill tip: chopsticks fall first, the bridge falls after the particle, the edge never falls.

Example contrast in sentence frame:

箸はしが長ながい。¹²
"The chopsticks are long."

橋はしが長ながい。¹³
"The bridge is long."

端はしが長ながい。¹⁴
"The edge is long."

A subtle point is worth saying out loud: 箸 and 端 both leave the particle L, so the particle alone does not separate them. Only the citation-form contour (HꜜL versus LH) does. The pair the particle test actually resolves is 橋 versus 端.¹

あめ: 雨 / 飴

This is the canonical two-way contrast, and probably the single most-quoted minimal pair in Japanese-learning material.

Word	Reading	Accent	Citation contour	With が
雨 (rain)	あめ	[1] atamadaka	HꜜL	HꜜL L15
飴 (candy)	あめ	[0] heiban	LH	LHH16

Wikipedia's scalar-pitch table places them as 飴 L-H and 雨 HꜜL, matching NHK 2016 exactly.⁵

One-line drill tip: rain falls (HꜜL); candy stays sweet and flat (LHH on a particle).

雨あめが降ふる。¹⁵
"Rain falls."

飴あめが好すき。¹⁶
"I like candy."

かみ: 神 / 紙 / 髪

A three-headword family that is often mis-taught as a tidy three-way pitch contrast. Under NHK 2016 it is not. The contrast is two-way: 神 stands alone, and 紙 and 髪 are matched.

Word	Reading	Accent	Citation contour	With が
神 (god)	かみ	[1] atamadaka	HꜜL	HꜜL L17
紙 (paper)	かみ	[2] odaka	LH	LHꜜL18
髪 (hair)	かみ	[2] odaka	LH	LHꜜL19

In citation form, 紙 and 髪 are isotonic, meaning they have the same pitch contour: both surface as LH. The particle test puts both at LHꜜL. Under NHK 2016 they are not a pitch-accent minimal pair; only context distinguishes them.¹⁷¹⁸¹⁹ The only member of the family that contrasts through citation-form HꜜL is 神.

One-line drill tip (corrected against the version that floats around learner pages): god falls on the first mora; paper and hair both drop only when a particle joins.

神かみが怒おこる。¹⁷
"The gods are angry."

紙かみが足たりない。¹⁸
"There's not enough paper."

髪かみが長ながい。¹⁹
"Her hair is long."

かき: 牡蠣 / 柿 / 垣 / 下記

This four-way near-minimal family sits at the intersection of two recurring problems: the citation-form trap and dictionary disagreement.

Word	Reading	Accent	Citation contour	With が
牡蠣 (oyster)	かき	[1] atamadaka	HꜜL	HꜜL L3
柿 (persimmon)	かき	[0] heiban	LH	LHH4
垣 (fence, hedge)	かき	[2] odaka	LH	LHꜜL20
下記 (the following)	かき	[1] atamadaka	HꜜL	HꜜL L21

The Kindaichi 1981 table reproduces the same Tokyo layout: "牡蠣 (oyster) HL / 垣 (fence) LH(L) / 柿 (persimmon) LHH."²¹⁰

Sources disagree on 柿 versus 牡蠣

NHK 2016, Daijirin, Shin-Meikai 5, and the Kindaichi 1981 table all assign 柿 [0] heiban and 牡蠣 [1] atamadaka. The English Wikipedia Japanese pitch accent article gives the opposite polarity (柿 odaka / 牡蠣 nakadaka, treating 牡蠣 as 3 morae), and some third-party learner pages independently report the inverse. This catalog follows NHK 2016. If a dictionary or app shows the opposite, it is a documented disagreement, not a typo. The Kansai realization is yet another story and belongs to the regional-variation article.²⁵³⁴

One-line drill tip: oysters drop on bite one (HꜜL); persimmons stay flat even with the particle (LHH).

牡蠣かきが好すきだ。³
"I like oysters."

柿かきが赤あかい。⁴
"The persimmons are red."

いま: 今 / 居間

This is a clean two-way contrast. It drills the atamadaka-versus-odaka distinction without the dictionary noise of the previous family.

Word	Reading	Accent	Citation contour	With が
今 (now)	いま	[1] atamadaka	HꜜL	HꜜL L22
居間 (living room)	いま	[2] odaka	LH	LHꜜL23

Kindaichi 1981 confirms the same Tokyo polarity: "今 (now) HL / 居間 (living room) LH(L)."²¹⁰

One-line drill tip: right now drops at the head; the living room stays flat until the particle drops it.

今いまが大事だいじ。²²
"Now is what matters."

居間いまが広ひろい。²³
"The living room is spacious."

High-yield near-minimal pairs worth adding to the drill

Beyond the core catalog, the following pairs appear often enough in N4–N3 reading and listening to earn space in the drill rotation. Several of them show the limits of pitch as a clue: some pairs are isotonic in Tokyo, so only context separates them.

朝 / 麻

Word	Reading	Accent	Citation contour	With が
朝 (morning)	あさ	[1] atamadaka	HꜜL	HꜜL L24
麻 (hemp, flax)	あさ	[2] odaka	LH	LHꜜL25

One-line drill tip: morning falls right away; hemp only drops on the particle.

朝あさが早はやい。²⁴
"The morning comes early."

厚い / 暑い / 熱い

The あつい triplet has the same sound string across all three adjectives, but it is not a clean pitch contrast. That fact is itself the lesson.

Word	Reading	Accent (NHK)	Accent (DJR / SMK)	Citation contour
厚い (thick)	あつい	[0] heiban	[2] nakadaka	LHH / LHꜜL26
暑い (hot weather)	あつい	[2] nakadaka	[2] nakadaka	LHꜜL27
熱い (hot to touch)	あつい	[2] nakadaka	[2] nakadaka	LHꜜL28

Under NHK 2016, 暑い and 熱い are isotonic (both nakadaka [2]), so only context distinguishes them. The only candidate for an accent contrast is 厚い versus 暑い/熱い. Even that contrast disappears under Daijirin and Shin-Meikai, which assign 厚い to [2] as well.²⁶²⁷²⁸

Pitch will not save you on the あつい triplet

Treating 暑い and 熱い as a pitch contrast is a false drill: under every major Tokyo dictionary they are both nakadaka [2]. The words are disambiguated by the kanji, the context, and sometimes the collocation (暑い夏 versus 熱いお茶). The 厚い versus 暑い/熱い contrast is real under NHK 2016 only; Daijirin and Shin-Meikai collapse all three to [2].²⁶²⁷²⁸

One-line drill tip: thick is flat under NHK; the two "hot"s are not actually a pitch contrast; the kanji is.

鮭 / 酒

Word	Reading	Accent	Citation contour	With が
鮭 (salmon)	さけ	[1] atamadaka	HꜜL	HꜜL L29
酒 (alcohol, rice wine)	さけ	[0] heiban	LH	LHH30

The しゃけ alternative reading for 鮭 is also atamadaka [1].²⁹ One historical note: Kindaichi's 1981 Meikai Akusento Jiten tags 酒 as odaka LH(L) rather than heiban; NHK 2016 and Daijirin have moved to heiban [0], and this catalog follows NHK 2016.²³⁰

Sources disagree on 酒 across 35 years

The Kindaichi 1981 table places 酒 at odaka; NHK 2016 and Daijirin place it at heiban [0]. This is a documented 35-year accent shift in the dictionary record, not a typo on either side. Pick one source for a study session and stick with it; for broadcaster-anchored work, NHK 2016 is the defensible default.²³⁰

One-line drill tip: salmon falls fast; alcohol stays flat; pitch keeps the dinner straight.

鮭さけが美味おいしい。²⁹
"The salmon is delicious."

酒さけが好すきだ。³⁰
"I like sake."

切る / 着る

Word	Reading	Accent (Tokyo)	Citation contour
切る (to cut, godan)	きる	[1] atamadaka	HꜜL31
着る (to wear, ichidan)	きる	[0] heiban	LH32

The lexical contrast is clean: 切る drops on the first mora, while 着る is unaccented. The contrast persists in the polite form (切ります HꜜLLL vs 着ます LHꜜL under Kawahara's conjugation rules).¹³¹³² The deeper conjugational story, or how each verb class carries accent through every form, belongs to the verb-and-adjective pitch article.

Wiktionary records that 切る inverts to heiban [0] in Keihan, which is the textbook Tokyo–Kansai accent inversion for verbs.³¹

One-line drill tip: cut drops; wear stays flat; they were never the same accent.

切きる音おとが聞きこえる。³¹
"I can hear a cutting sound."

着きる服ふくがない。³²
"There's nothing to wear."

花 / 鼻

Word	Reading	Accent	Citation contour	With が
花 (flower)	はな	[2] odaka	LH	LHꜜL33
鼻 (nose)	はな	[0] heiban	LH	LHH34

This is the textbook particle-test pair: 花 and 鼻 are isotonic in citation form (both LH), and only の or が separates them: 花が LHꜜL versus 鼻が LHH.⁵³³³⁴ The Osaka variant of 花 shifts to [1]; the regional-variation article handles that detail.³³

One-line drill tip: nose stays flat even on the particle; flower drops when the particle hits.

花はなが咲さく。³³
"The flowers bloom."

鼻はなが高たかい。³⁴
"His nose is high (i.e., he's proud)."

How to drill these pairs

A catalog without a workflow is just a list. The protocol below turns the rows above into a usable training loop. Treat it as practical guidance: it draws on standard L2 phonology teaching rather than on a single cited source.

The perception-first sequence

Perception comes before production. The five-step protocol takes one pair at a time and trains the ear before the mouth gets involved.

Step 1: Blind listen. Play a single pair five times in random order without labels. Score each play. The goal is to find your baseline before any visual cues interfere.

Step 2: Labeled listen. Replay the same pair with the kanji and accent number visible. The ear is now binding the contour to the label, not guessing.

Step 3: Shadow. Speak along with the audio at native speed. OJAD's Suzuki-kun is the recommended controllable source: "an interactive component called 'Suzuki-kun' (韻律読み上げチュータ), a prosody tutoring system that estimates word accent and intonation patterns for input text."¹¹

Step 4: Self-record. Produce the pair into a phone or any voice recorder. Do not skip this step. Recording reveals drift that the live ear often misses.

Step 5: Overlay and audit. Open both tracks in a spectrogram tool (Praat, or any free alternative) and check whether the H→L drop in your recording lands on the same mora as the native one. Kawahara's criterion is the audit point: accent is "an abrupt H(igh)-L(ow) fall in F0," and the goal is a sharp, correctly positioned drop.¹

Where to get reliable audio

OJAD is the primary recommendation: "approximately 9,000 nouns" and "roughly 3,500 predicates" with native male and female audio, plus Suzuki-kun for arbitrary text.¹¹ The NHK 2016 dictionary is the print reference behind most of the accent integers used in this article.⁶

Forvo supplies crowd-sourced native recordings. These are useful when OJAD's controlled audio sounds too lab-like for your ear.³⁵ JPitch operates over roughly 160,000 entries with a minimal-pair filter. It is the natural follow-on tool once you have internalized the curated short list above.³⁶

OJAD's Suzuki-kun for arbitrary sentences

The drill rows above all sit at the citation-form-plus-particle level, but real listening happens in full sentences. Suzuki-kun will render any text pasted into it with predicted accent and intonation curves. That lets the drill extend from word pairs to the surrounding sentence frames without leaving a single tool.¹¹

One caution worth saying plainly: YouTube creator tracks vary widely in pitch accuracy, and many speakers' realizations are influenced by their first language. Use them for context and listening volume, not as a primary accent reference. OJAD and NHK 2016 are the anchors.

A weekly drill schedule

A workable pace is ten minutes a day, three pairs at a time. Move to the next set after you reach a perception score of about 90 percent on the current set. Perception often lags production by months; that asymmetry is normal and not a sign the drill is failing. The goal at N4 is to start hearing the contrast on the dozen or so highest-frequency pairs above, not to grind through a 160,000-entry tool.

Good to know

"Pitch accent" is not "tone"

Japanese is a pitch-accent language, not a tone language. Kawahara puts the constraint precisely: Japanese "lexically uses only two levels of tonal heights (High and Low, and not, for example, Mid)" and culminativity restricts a word to "at most one accentual HL fall."¹ In Mandarin, every syllable carries its own tonal contour. In Tokyo Japanese, the whole word carries at most one drop, located by the lexical accent. The drill is find the drop, not memorize a melody.¹⁵

Citation-form blindness on two-mora words

Two-mora words in odaka and heiban sound identical in isolation (both as LH), and the contrast only appears on a following particle. Kawahara cites this directly: "monomoraic and bimoraic finally-accented words (e.g., /kaki/ 'fence') and unaccented words (e.g., /kaki/ 'persimmon') are phonetically very similar, if not identical, when they appear in isolation (Vance 1995; Warner 1997)."¹

The wrong drill repeats 橋 and 端 alone, in citation form, hoping the difference will appear. Both come out LH, so the learner trains on a non-contrast. The right drill anchors the particle:

橋はしが長ながい。¹³
"The bridge is long."

The drop on が is the test.

Dictionaries disagree more than learners expect

Several headwords in the catalog above carry documented disagreement between major dictionaries. 厚い is acc=0 heiban in NHK 2016 but acc=2 nakadaka in Daijirin and Shin-Meikai.²⁶ 柿 and 牡蠣 are [0] heiban and [1] atamadaka in NHK 2016, Daijirin, and Shin-Meikai 5, but the opposite polarity appears on Wikipedia.²³⁴⁵ 雲 is [1] atamadaka in Daijirin and NHK, while some teaching materials give [0] heiban.³⁷ 酒 was odaka in Kindaichi 1981 and is heiban in NHK 2016 and Daijirin.²³⁰

The rule for a learner is simple: pick one accent dictionary and stick with it for the duration of a study session. NHK 2016 is the broadcaster standard and the most defensible default. Flipping between sources mid-drill turns honest dictionary disagreement into manufactured confusion.

Regional realizations and the Tokyo frame

This article follows the NHK 2016 Tokyo standard from start to finish. Kansai inverts several of the same pairs:

Word	Tokyo	Kansai7
はし (bridge)	L-H(-L)	H-L
はし (chopsticks)	H-L	L-H
はし (edge)	L-H(-H)	H-H
日本 (Japan)	L-H-L	H-L-L
二本 (two sticks)	H-L-L	L-L-H

Wikipedia's summary is blunt: "The pitch accent in Kansai dialect is very different from the standard Tokyo accent, so non-Kansai Japanese can recognize Kansai people easily from that alone."⁷ Verb-internal inversions of the same type appear in 切る (Tokyo [1] atamadaka / Keihan [0] heiban).³¹ The regional-variation article covers the full Keihan and Kansai system. In this catalog, Tokyo is the only frame.

雲 versus 蜘蛛 is not a Tokyo minimal pair

The popular tip "the cloud floats flat, the spider falls" reflects either a non-NHK source or a Kansai-influenced reading. Under NHK 2016 and Daijirin, 雲 and 蜘蛛 are both atamadaka [1]. They are isotonic, so context distinguishes them, not pitch.³⁷³⁸ The pair is included here as a documented isotonic-homophone case so a learner does not waste drilling time on a non-contrast.

Where overlearning hurts

Most everyday Japanese words have no drop at all. Kawahara notes the heavy lexical skew toward unaccented (heiban) words in the Tokyo lexicon, and that "the distinction between finally-accented words ... and unaccented words ... are phonetically very similar, if not identical, when they appear in isolation."¹ The dozen pairs above carry the highest teaching value. Further drilling with a 160,000-entry tool like JPitch is optional, not required, at the N4 level.³⁶

See also

• Japanese Compound-Word Pitch Accent: How Two Words Combine into One Accent Pattern
• Minimal-Pair Production Drills in Japanese: Train Your Mouth to Say the Difference
• Japanese Pronunciation Drills: A Daily 5-Minute Protocol with Minimal Pairs, Shadowing, and Record-and-Compare
• Should You Learn Pitch Accent? An Honest Cost-Benefit Analysis
• Long vs. Short Vowels in Japanese: The Distinction Beginners Miss

References

Footnotes

1. Kawahara, Shigeto. "The phonology of Japanese accent." In Handbook of Japanese Phonetics and Phonology, edited by Haruo Kubozono, 445–492. De Gruyter Mouton, 2015. https://user.keio.ac.jp/~kawahara/pdf/HandbookAccentPublished.pdf ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸ ↩¹⁹ ↩²⁰

2. 金田一春彦 (ed.). 明解アクセント辞典, 三省堂, 1981. Page 12. Cited at sci.lang.japan FAQ as the source of the minimal-pair table reproduced under "The core drill catalog" below. https://www.sljfaq.org/afaq/pitch-accent.html ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰

3. Wiktionary. "牡蠣." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E7%89%A1%E8%A0%A3 ↩ ↩² ↩³ ↩⁴ ↩⁵

4. Wiktionary. "柿." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E6%9F%BF ↩ ↩² ↩³ ↩⁴ ↩⁵

5. Wikipedia contributors. "Japanese pitch accent." https://en.wikipedia.org/wiki/Japanese_pitch_accent ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³

6. NHK放送文化研究所 (ed.). NHK日本語発音アクセント新辞典 (NHK Pronunciation Accent New Dictionary, 2016 edition; abbreviated NHK16 in Wiktionary pronunciation templates). NHK出版, 2016. ISBN 978-4-14-011345-5. Reviewed in 国際交流基金『日本語教育通信　本ばこ』, https://www.jpf.go.jp/j/project/japanese/teach/tsushin/bookshelf/201610.html ↩ ↩² ↩³

7. Wikipedia contributors. "Kansai dialect." Pitch-accent section. https://en.wikipedia.org/wiki/Kansai_dialect ↩ ↩² ↩³

8. 松村明 ほか (eds.). 大辞林 (Daijirin; abbreviated DJR in Wiktionary pronunciation templates). 三省堂. Wiktionary's acc_ref=DJR cites this dictionary's accent annotation. Entry-level accent values reproduced here are sourced via Wiktionary's per-headword {{ja-pron}} template, which cites DJR alongside NHK and SMK. ↩

9. 金田一春彦, 秋永一枝 (eds.). 新明解日本語アクセント辞典 (Shin Meikai Nihongo Akusento Jiten; abbreviated SMK / SMK5 / SMK8 in Wiktionary pronunciation templates). 三省堂. The 5th edition (SMK5) and 8th edition (SMK8) are cited by Wiktionary as acc_ref=SMK5 and acc_ref=SMK8 respectively. ↩

10. Bullock, Ben. "What is Japanese pitch accent?" sci.lang.japan FAQ. https://www.sljfaq.org/afaq/pitch-accent.html ↩ ↩² ↩³ ↩⁴

11. Online Japanese Accent Dictionary (OJAD). Minematsu Laboratory and Saito Laboratory, Graduate School of Engineering, The University of Tokyo. https://www.gavo.t.u-tokyo.ac.jp/ojad/ ↩ ↩² ↩³ ↩⁴

12. Wiktionary. "箸." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E7%AE%B8 ↩ ↩²

13. Wiktionary. "橋." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E6%A9%8B ↩ ↩² ↩³

14. Wiktionary. "端." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E7%AB%AF ↩ ↩²

15. Wiktionary. "雨." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E9%9B%A8 ↩ ↩²

16. Wiktionary. "飴." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E9%A3%B4 ↩ ↩²

17. Wiktionary. "神." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E7%A5%9E ↩ ↩² ↩³

18. Wiktionary. "紙." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E7%B4%99 ↩ ↩² ↩³

19. Wiktionary. "髪." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E9%AB%AA ↩ ↩² ↩³

20. Wiktionary. "垣." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E5%9E%A3 ↩

21. Wiktionary. "下記." Pronunciation block. https://en.wiktionary.org/wiki/%E4%B8%8B%E8%A8%98 ↩

22. Wiktionary. "今." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E4%BB%8A ↩ ↩²

23. Wiktionary. "居間." Pronunciation block. https://en.wiktionary.org/wiki/%E5%B1%85%E9%96%93 ↩ ↩²

24. Wiktionary. "朝." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E6%9C%9D ↩ ↩²

25. Wiktionary. "麻." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E9%BA%BB ↩

26. Wiktionary. "厚い." Pronunciation block. https://en.wiktionary.org/wiki/%E5%8E%9A%E3%81%84 ↩ ↩² ↩³ ↩⁴

27. Wiktionary. "暑い." Pronunciation block. https://en.wiktionary.org/wiki/%E6%9A%91%E3%81%84 ↩ ↩² ↩³

28. Wiktionary. "熱い." Pronunciation block. https://en.wiktionary.org/wiki/%E7%86%B1%E3%81%84 ↩ ↩² ↩³

29. Wiktionary. "鮭." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E9%AE%AD ↩ ↩² ↩³

30. Wiktionary. "酒." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E9%85%92 ↩ ↩² ↩³ ↩⁴ ↩⁵

31. Wiktionary. "切る." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E5%88%87%E3%82%8B ↩ ↩² ↩³ ↩⁴ ↩⁵

32. Wiktionary. "着る." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E7%9D%80%E3%82%8B ↩ ↩² ↩³

33. Wiktionary. "花." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E8%8A%B1 ↩ ↩² ↩³ ↩⁴

34. Wiktionary. "鼻." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E9%BC%BB ↩ ↩² ↩³

35. Forvo. Crowd-sourced pronunciation database with native-speaker recordings. https://forvo.com/ ↩

36. JPitch: Japanese Pitch Accent Dictionary. Audio + minimal-pair filter over ~160,000 entries. https://www.jpitch.org/ ↩ ↩²

37. Wiktionary. "雲." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E9%9B%B2 ↩ ↩²

38. Wiktionary. "蜘蛛." Pronunciation block, Etymology 1. https://en.wiktionary.org/wiki/%E8%9C%98%E8%9B%9B ↩