第406課: Phonology II: Pitch (高低アクセント)   ※Under Revision

The word “accent” refers to two interrelated things. Broadly speaking, “accent” (訛り) is the way one’s speech sounds in relation to other speakers in one’s region – example being having a British accent when speaking English. More technically speaking, “accent” (アクセント) is the combination of phonetic properties imposed over the individual sounds that make up words.

Take for instance the word “debris.” In American English, the “accented” syllable is on the second syllable, whereas in British English, the accented syllable is on the first syllable. From this one word, a lot can be gleaned about how accent systems work across languages. 

i. Accented parts of a word stand out.
ii. Accumulated differences regarding accent placement result in regional accents.

iii. Native speakers are generally aware of how other dialects phonotactically contrast at a subconscious level.

English has a “stress/dynamic” accent system (強弱アクセント) in which syllables differ in how much stress is put on them. Thus, “stress” is the relative emphasis on certain syllables, which is accompanied with increased loudness as well as vowel lengthening.

If, however, accented syllables only differ by pitch alone, the language is said to have a pitch accent (高低アクセント), which is what characterizes Standard Japanese along with the majority of Japanese dialects. 

Terminology Note: 訛り and アクセント both translate to “accent,” but they refer to two different understandings of the word “accent.” 訛り refers to the overall way a person’s speech sounds, which is tied strongly to their background. On the other hand, アクセント directly relates to “pitch accent,” which determines which morae are spoken with a higher pitch. 

Notation Notes: Knowledge of the International Phonetic Alphabet (IPA) is required.

i. Accented morae are in bold.

ii. ↓ indicates a drop in pitch after an accented mora.

iii. ⇒ indicates a shift in accent that has become fully mainstream.

iv. → indicates a currently observed shift in accent.

v. >,<, and ≒ may be used to indicate variation in accent shift but to what frequency one version is more accepted than the other(s). 

Japanese Pitch Accent

Japanese pitch accent is given little attention in textbooks, but it is not out of sheer negligence that resources, including IMABI, do not typically provide accent notation for sentences much less individual words. Reasons for this omission include:

1. Of the 120 million or so native Japanese speakers, those who natively use the normative pitch accent of Standard Japanese, based on the Yamanote Dialect of Tokyo proper, are confined to the greater Tokyo area, and although much of the country still uses the Tokyo-type pitch accent system (~60%), individual words may differ greatly in their assigned accent or lack thereof. Meaning, the system itself defines the patterns which phrases may take, but which pattern is lexically assigned to a given word may vary regionally. 

2. How native speakers perceive standard pitch accent is an oversimplification of what is linguistically true, and posited linguistic theories that seek to accurately describe it also vary in complexity. Typically, as is implied by the name of the system in Japanese – 高低アクセント – every mora is assigned a high (高) or low (低) pitch/tone, but acoustically speaking, not all low and high pitches are respectively equal in sound quality. 

3. As is evident by the leading accent dictionaries published for native speakers such as the NHK日本語発音アクセント辞典, with every edition, numerous edits are made. This is not to say that earlier editions are inherently faulty, but intense research has been shaping how pitch accent ought to be perceived and how it is changing in real time.

4. Most importantly, even among Tokyoites, just as some words die out and new words are formed to represent the living world around us, the accent patterns of words themselves are also subject to change over time. A native Tokyoite in his 20s will not accent words the same as a native Tokyoite in his 20s will 100 years from now. The Tokyo Dialect (東京弁・東京方言), which is the basis of Standard Japanese.

5. The Tokyo Dialect, historically, is at the southern end of the Eastern Japanese branch of the language’s dialect continuum. Most dialects of this branch have what is called a 無アクセント (no accent) system in which essentially all phrases are not accented. Such a simplification of the standard pitch accent is very much underway and will be become ever more prevalent as the 21st century progresses. 

6. Perhaps the crux as to why even Japanese script doesn’t even distinguish pitch is that a non-native speaker can be understood despite incorrect intonation provided the sentence is grammatically sound. In this light, Japanese is not viewed as a tonal language despite individual syllables (morae※) possessing either a low or high tone. This statement holds true despite minimal pairs existing between words which only differ in pitch accent because those pairs only exist in the confines of Standard Japanese and do not take into account intelligible dialectal pronunciations. 

7. If pitch accent were so detrimental, Japanese dialects would not be as intelligible among each other as they currently are, nor would foreign speakers be so easily understood as mistakes would be inevitable. 

To combat all these factors in our formal discussion of pitch, a few ground rules will need to be laid out.

1. Pitch contours are referenced from the NHK発音アクセント新辞典 (2016) and the 新明解アクセント辞典 (2014). 

2. Variant pitch contours are listed by most to least preferred. Those in () are allowed in broadcasting but are still avoided in careful speech. 

3. L (low) tones and H (high) tones are understood as being relative pitch assignments to the surrounding morae. An H in bold indicates an accented mora.

4. “Pitch” as in the tone of a given mora will be used interchangeably with “tone,” and “low tone/pitch” and “high tone/pitch” are notated as L (tone) and H (tone) respectively. 

Terminology Note: The basic unit of Japanese phonology is the “mora” (拍・モーラ) rather than the syllable (音節). A mora is a unit of speech perceived to be equal in length to one another. An independent mora (自立拍) may be composed of a vowel (V) or consonant + vowel (CV), whereas a special mora (特殊拍) relies on surrounding morae to be produced but still count as an individual mora (ex. っ・ん).

How Pitch Helps Distinguish Words But Doesn’t 

A frequently mentioned reason for why proper pitch is important to speech is that it helps distinguish otherwise homophonous phrases. This statement is true in the sense that there are countless minimal pairs made by differences in pitch accent.

For instance, both 生命 (life) and 声明 (proclamation) are pronounced as せいめい, but the former’s pitch is HLLL and the latter’s pitch is LHHL. As an extreme example, the following sentence is often presented:

1. 庭には2羽鶏がいる。
There are two chickens in the courtyard.

Pitch Note: In faster, natural speech, 居る would be pronounced as HH due to the low-pitched mora being sandwiched between two high-pitched morae. Such “pitch flattening” in between segments is a common feature of natural speech. Even within relatively small phrases (whether they be isolated or juxtaposed with other phrases), changes in pitch over the course of a phrase are best understood to be gradual. 

More so than individual words (語) themselves, pitch pattern assignment is best understood by segmenting sentences into 句 (phrases). There are four different pitch accent contours that exist in Standard Japanese: 

  1.  頭高型 Atamadaka-gata (Initial High Tone Accent Pattern):
    Phrase whose accent is on the first mora, meaning the first mora is high and (the) subsequent mora(e) is/are low – HL.
  2.  中高型 Nakadaka-gata (Middle High Tone Accent Pattern):
    Phrase which starts off low, then the pitch is high until the accented mora, at which point the pitch falls for the remainder – LHL.
  3.  尾高型 Odaka-gata (Final High Tone Accent Pattern):
    Phrase which starts off low but the pitch is high until the final accented mora, then there will be a drop in pitch on any affix, usually a particle, that follows – LH(L)
  4.  平板型 Heiban-gata (Flat Accent Pattern):
    An accent-less phrase whose pitch is maintained high even when affixes, usually in the form of particles, follow – LH(H).

Though the ‘standard’ implementation of pitch accent is key to sounding native, there are two scenarios which disprove the necessity of its accurate implementation.

1. The Inability to Distinguish Non-Accent Words from Final-Mora-Accented Words

Consider the nouns 箸 (chopsticks), 橋 (bridge), and 端 (edge). 箸 is HL (頭高型), but 橋 and 端 differ by whether the final mora is the accented mora or not. In the case of 橋, the second mora is accented, whereas for 端 it is not. Thus, 端が is LHH, whereas 橋が is LHL. Although the H tone of 橋 and other 尾高型 words is higher than the H tone of an unaccented mora in a 平板型 word, native speakers normally cannot discern the difference when there is no affix present to contrast the two. 

Though there is an audible difference between 橋が (LHL) and 端が (LHH), it is not as stark as the difference between a word pair that differs by HL vs LH (ex. 朝 (morning) vs 麻 (hemp)). Nonetheless, even in isolation, speakers may pronounce the H tone of an 尾高型 word higher than the H tone of a 平板型 word, but this is not a perceptible difference to native speakers. 

In reality, native speakers do not confuse homophones because speakers are hard wired to pay attention to context clues as opposed to pitch alterations. This does not negate how speakers pronounce these pitch patterns differently, but it is a difference that cannot be heard easily in isolation.

2. The Variability in Accent Across Dialects

As the last mention of other dialects’ pitch accents, basic lexical items in Japanese differing in pitch across dialects certainly doesn’t help build the case of pitch being quintessential to comprehension. 

 Gloss標準語 関西弁

 Valley 谷LH(L) 谷HH
 Really マジHL マジLH

The Four Pitch Contour Patterns in Detail

Determining the pitch accent of a given phrase is an arduous task which requires both knowledge of what the possible contours are and what the likelihood is for a given contour to be assigned. To tackle this multifaceted dilemma, let’s closely familiarize ourselves with what the four accent patterns are. 

■Accent or Flat? 起伏式 vs 平板式

The four accent patterns – 頭高型・中高型・尾高型・平板型 – can be organized according to whether there is undulation (ups-and-downs) in pitch (起伏式) or if the pitch rises and stays flat (平板式). Along with these two larger groups, the following terms greatly enhance understanding of how pitch functions in general.

  1.  Initial Lowering: Of the 起伏式 accent patterns, non-頭高型 words feature what is known as initial lowering in which the first mora is assigned a L tone.  
  2.  Downstep: Immediately after an accented mora, the pitch suddenly drops. 
  3.  Downdrift: Throughout any utterance, the overall pitch register tapers off as the speaker gets closer to the next breath.  
  4.  Accent Nucleus: An alternative name for “accented mora,” which can aid the learner to pinpoint where said accent lies in phrases regardless of length. 
  5.  Pitch Leveling: The reduction/simplification of undulations in pitch observed when combining segments of a phrase and/or sentence together.

□The Accent Nucleus

The accented mora can also be called the “accent nucleus” (アクセント核). Some natural restrictions exist as to which kinds of morae can become an accent nucleus. All examples of special morae (特殊拍) may not be the accented nucleus. 

・The 撥音 (moraic nasal “N”) ―ん.

・The 促音 (moraic obstruent in double consonants) ―っ・ッ. 
・The juxtaposition of two vowels which arose through various sound changes such as the /i/ seen at the end of ON readings which would’ve been at the tail end of what would have been diphthongs in Chinese. Ex. 対 = /tai/.

□The 起伏式 Grouping

Accent patterns that fall under the 起伏式 grouping all demonstrate a particularly H tone mora which serves as the accented mora. Aside from 頭高型 words, there is an initial lowering of pitch to start off, then pitch rises until arriving to the accented mora. At that point, a downstep occurs. 

Accent PatternInitial Lowering?  Example
頭高型X カクテル (cocktail)HLLL
中高型   惨い (merciless)LHL
尾高型 男が (the man subj.)LHHL

※The particle が is an affix; thus, 男’s pitch is 尾高型 as opposed to 中高型.

As for 頭高型 words in mid-phrase, the previous mora – whether it’s assigned H or L – can be likened to initial lowering as their H tone will be obligatorily higher for the sake of being the accented mora, but this difference is practically negligible. This is because any and all prior downsteps along with the downdrift phenomenon will have rendered the mid-phrase H tone L relative to the initial H tone. 

Phrase (Pitches of Individual Morphemes) Natural Pitch Contour
変な女の子HL + LHHLL  ̄ \____         \__

※Because an 頭高型 word began the phrase, it immediately forces a downstep. One thing that must be understood about downsteps, however, is that the abrupt drop in pitch affects the pitch register of the rest of the phrase. This renders the two H-toned morae in 女の子 as L relative to the accented H mora in 変な. Accordingly, the accented tone in 女の子, although never higher than the first accented mora, raises the pitch oh-so enough to trigger yet another downstep. 

□The Lexicon Helps Determine 中高型

Of the 起伏式 patterns, 中高型 phrases will present the greatest challenge for learners to master when the downstep occurs and to know how loud the accented mora will actually be, as the same can be said for the accented mora of any long phrase as demonstrated above – the accented H mora at the end of a long phrase potentially being lower than the first H mora due to downsteps and downdrift. 

Rest assured, because the accented H tone is lexically assigned, there are plenty of rules built into the lexicon that make 中高型 phrases seem far less random. 

Take for instance 麦藁帽子 (wheat straw hat), which is a compound phrase created from three morphemes. The first two morphemes are 頭高型 in isolation. When such words are combined, the resultant accent pattern is most likely 平板型 with 中高型 being a close second. When more (and/or more complicated) morphemes compose a given phrase, the likelihood of it being 中高型 is even higher with the downstep most likely occurring on the first mora of the last morpheme. These scenarios can all be viewed as systematic pitch leveling built into the phraseology of the language.

Pitches of Individual
Pitches of Combined
藁 (straw)/wa↓ɾa/ (HL)麦 (wheat)/↓gi/ (HL)この (this)/kono/ (LH)帽子 (hat)/bo:ɕi/ (LHH) この藁 (this hat)/kono-wa↓ɾa/ (LHHL)麦藁 (wheat straw) /mɯgiwaɾa/ or /mɯgiwa↓ɾa/この麦藁 (this wheat straw)
/kono-mɯgiwaɾa/ (LHHHHH)この麦藁帽子が (this wheat straw hat subj.)/ kono-mɯgiwaɾa-bo↓:ɕi-ga/ (LHHHHHHLLL)

※↓ indicates a downstep. Its absence after lexically H-toned morae indicates that a phrase is 平板式, whereas its presence indicates a 起伏式 contour. Only accented high-tone morae are in bold. This convention is extended throughout this lesson.

□Pitch Leveling Affecting Initial Lowering

At times, as we saw earlier with 居る, pitch leveling occurs when the undulation of a phrase is difficult to maintain in fast, natural speech. This can affect how initial lowering manifests. Generally speaking, phrases with a lexically assigned initial lowering are enunciated this way when sentence-initial, but there are a few conditions which may trigger a pitch leveling that renders this L tone as high as a non-accented H tone. 

Similarly to how an L-toned mora sandwiched between two H-toned morae may become high tone by association, when the second mora of a non-頭高型 word is either a ん or the lengthening half of a long vowel (ー), the tone of the first mora is as high as the H tone of a non-accented mora. 

WordNatural Pitch Contour Lexical Pitch Contour
(Slow/Deliberate Speech)
反応 (reaction)/haNno:/ HHHH(H) LHHH(H)

In deliberately slow speech, such lexically assigned instances of initial lowering  will be clearly enunciated as L tone. Due to this and the fact that it would not be as high as an accented mora, such environmentally not so low L morae are still marked as L in leading pitch dictionaries as entries are meant to be interpreted as single-standing words in isolation in careful speech.

Another phenomenon not overtly noted in pitch dictionaries is how initial lowering is prolonged whenever the second mora is a っ.

 WordNatural Pitch Contour Lexical Pitch Contour
切腹 (seppuku)/seQpɯkɯ/ LLHH(H) LHHH(H)

This is likely due to it being a quasi-mora that although being perceived as equal in length to any other mora is not. However, a second mora being slightly lower than normal is not a hindrance to lexically assigned it a H tone, especially since っ happens to be banned from being an accent nucleus in the first place.  

□The 平板式 Grouping

The 平板型 is the only accent pattern in the 平板式 grouping. This pattern accounts for approximately half of all Japanese words, and this percentage is growing by the decade. 

What is meant by “flat” is that the phrase is accentless with the first mora being low and subsequent mora(e) being H tone, meaning there is no downstep. However, the pitch of such a chain of H tone morae in a “flat” phrase are, in fact, not equally high as one another. This is because of the “downdrift” phenomenon, but imagine this being akin to rolling down a smooth decline rather than falling off a cliff. 

■Binary vs Tertiary Pitch

Earlier we learned how the accented H-toned mora in 尾高型 words is technically higher than the non-accented H-toned mora(e) in 平板型 words when no affix is present. To account for this, it is possible to propose a M (mid) tone and assign it to the less-high H tone in 平板型 words. 

Word Tertiary System Binary System
花 (flower)
鼻 (nose)

Ultimately, the tertiary system hypothesis was abandoned due to subsequent research which indicated that native speakers don’t perceive the distinction even if it is there when certain words are juxtaposed in isolation. 

Oddly enough, the M tone was also used to correspond to any L mora(e) in a 起伏式 word that immediately follows a downstep, assuming that the phrase is short enough for the subsequent morae to not ultimately be lower than an initial lowering if present due downdrift as well as any subsequent downsteps. In this interpretation, M serves to indicate a more natural decline/increase in pitch. 

WordTertiary SystemBinary System
毎朝 (every morning)/maiasa/  HMLL  HLLL
 維持費 (maintenance cost)/iʑiçi/  LHM  LHL

■The Four Pitch Patterns w/ Various Mora Count

Aside from the empty spaces which are impossible due to mora count, the four pitch contour patterns naturally manifest in phrases of various lengths, and that is before investigating how contours interact with each other once phrases are brought together. 

The chart below serves to show isolated words of words of two-six morae in length, but that of course does not exclude the possibility of these patterns being used in larger words. As for one-mora words, only 頭高型 and 平板型 are possible. 

※As will soon be looked at in great detail, the ongoing simplification of Standard Japanese pitch accent has caused the shifting of many words in systematic ways. Of the words listed in this chart alone, two have notably different accent patterns in modern speech than their historically correct assignments.

Ex. 花婿 LHHL ⇒ LHLL

Terminology Note: To distinguish the different possible 中高型, numbers in circles will indicate the mora placement of the accent nucleus. So, 中高型④ would indicate said pattern in which the accent nucleus is on the fourth mora. 

Distribution of Pitch Patterns in Nouns by Mora Count

Many observations can be made about the distribution of the four pitch patterns in nouns by mora count. Most importantly, the ongoing simplification of Standard Japanese pitch, which can be described as the reduction of mid-word undulation of pitch, can be best studied by narrowing one’s view on mora count to see how pitch is changing based on word length. In general, the percentages of 頭高型 and 平板型 are particularly on the rise as their contours are less complex.

One thing that is certain about nouns is that within any given mora count, that the possible contours for said count will always be one greater. This is to account for the 中高型 pattern accounting for any accented mora not on the first or last mora. Therefore, three-mora words display four possible contours: HLL, LHL, LHH(L), or LHH(H). Likewise, four-mora words display five possible contours: HLL, LHHL, LHLL, LHHH(L), or LHHH(H). 

Mora Count拍数Pattern  Tendencies見られる傾向
一拍語 平板型
Approximately 30% of one-mora nouns are unaccented; the more common it is, the more likely it is unaccented.蚊 /ka/ (mosquito)
蛾 /ga/ (moth)気 /ki/ (qi)
血 /tɕi/ (blood)葉 /ha/ (leaf)差 /sa/ (difference)図 /dzɯ/ (illustration)藻 /mo/ (algae)身  /mi/ (body)実 /mi/ (fruit/nut)
一拍語 頭高型H(L)Approximately 70% of one-mora nouns are accented. Uncommon, newly counted, as well as most Sino-Japanese one-mora nouns will be of this pattern. However, these tendencies do not exclude common nouns from being this pattern.巣 /sɯ↓/ (nest)酢 /sɯ↓/ (vinegar)木 /ki↓/ (tree)歯 /ha↓/ (tooth)府 /ɸɯ↓/ (metropolitan prefecture)負 /ɸɯ↓/ (negative)愚 /gɯ↓/ (foolishness)苦 /kɯ↓/ (suffering)目 /me↓/ (eye)非 /çi↓/ (fault)
Approximately 60% of two-mora nouns fall under this pattern. Most uncommon, Sino-Japanese, and/or newly coined nouns are of this pattern. Almost all loanwords are of this pattern.松 /ma↓tsɯ/ (pine tree)雲 /kɯ↓mo/ (cloud)蜘蛛 /kɯ↓mo/  (spider)妻 /tsɯ↓ma/ (wife)罪 /tsɯ↓mi/ (sin)刑 /ke↓:/ (punishment)蝶 /tɕo:/ (butterfly)腸 /tɕo↓:/ (intestine)ケア /ke↓.a/ (care)シェア /ɕe↓.a/ (share)
Approximately 20% of two-mora nouns fall under this pattern. It is rare among Sino-Japanese words but certainly not unheard of. It is even rarer among loanwords.街 /matɕi↓/ (street)鱒 /masɯ↓/ (trout)毒 /dokɯ↓/ (poison)服 /ɸɯ̥kɯ↓/ (clothes)幕 /makɯ↓/ (curtain)島 /ɕima↓/ (island)縞 /ɕima↓/ (stripe)丈 /take↓/ (stature)ヒレ /çiɾe↓/ (filet) 舌 /ɕi̥ta↓/ (tongue)Among native nouns, some words traditionally of this pattern are shifting to either 頭高型 or 平板型. 北 /ki̥ta↓/ ⇒ /ki̥ta/ (north) 鹿 /ɕi̥ka↓/ ⇒  /ɕi̥ka/ (deer)虹 /niʑi/ ⇒ /niʑi/ / (rainbow)熊 /kɯma↓/ → /kɯ↓ma/ (bear)寿司 /sɯɕi↓/ → /↓ɕi/ (sushi)If the vowel of the first mora is devoiced, the accent may fall onto the second mora despite the word being of the 頭高型, thus, rendering it as 尾高型.汽車 /ki↓ɕa/ → /ki̥ɕa↓/ ((steam) train)スト /sɯ↓to/ → /sɯ̥to↓/ (strike)
二拍語平板型LH(H)Approximately 20% of two-mora nouns fall under this pattern. It is rarely seen in Sino-Japanese words and almost unheard of among loanwords.下 /ɕi̥ta/ (down)隙 /sɯ̥ki/ (gap)杉 /sɯŋi/ (cedar)滝 /taki/ (waterfall)客 /kjakɯ/ (customer)宅 /takɯ/ (one’s home)下痢 /geɾi/ (diarrhea)黄身 /kimi/ (yolk)ヒレ /çiɾe/ (file)ゲラ /geɾa/ (galley)
三拍語頭高型HLLApproximately 40% of three-mora nouns. As for Sino-Japanese words and loanwords, they are typically either of this pattern or of the 平板型. Loanwords are especially likely to be of this pattern at 85%+. 兜 /ka↓bɯto/ (helmet/headpiece)辛さ /ka↓ɾasa/ (spiciness)笑顔 /e↓ŋa.o/ (smile)議員 /gi↓iN/ (legislator)関与 /ka↓Njo/ (participation)街路 /ga↓iɾo/ (avenue)宇宙 /u↓tɕɯ:/ (universe)遺族 /i↓zokɯ/ (surviving family)シグマ /ɕi↓ŋɯma/ (sigma)カルタ /ka↓ɾɯta/ (karuta)
三拍語 中高型②LHLApproximately less than 10% of three-mora nouns fall under this pattern, of which many words are turning into 頭高型 or 平板型.心 /koko↓ɾo/ (heart)奇数 /ki̥sɯ↓:/ (odd number)機械 /ki̥ka↓.i/ → /ki̥↓ka.i/ (opportunity)試験 /ɕi̥ke↓N/ → /ɕ↓keN/ (exam)布巾 /ɸɯ̥ki↓N/ →  /ɸɯ̥↓kiN/ (dishcloth)影絵 /kaŋe↓.e/ →   /kaŋe.e/ (shadowgraph)いとこ /ito↓ko/ → /i↓toko/ (cousin)学位 /gakɯ↓.i/ ⇒ /ga↓kɯ.i/ (academic degree)お化け /oba↓ke/ (monster)お茶屋 /otɕa↓ja/ → /otɕaja/ (teahouse)
三拍語尾高型LHH(L)Approximately less than 10% of three-mora nouns fall under this pattern, of which many words are turning into 頭高型 or 平板型.
男 /otoko↓/ (man)切り身 /kiɾimi↓/ (cut/slice)臭味 /kɯ̥sami↓/ → /kɯ̥sami/ (bad smell)豆腐 /to:ɸɯ↓/ → /to:ɸɯ/ (tofu)戦 /ikɯ̥sa↓/ ⇒ /ikɯ̥sa/ (war)林 /hajaɕi↓/ ⇒ /hajaɕi/ (grove)  東 /çiŋaɕi↓/ ⇒ /çiŋaɕi/ (east)会議 /ka.iŋi↓/ (meeting) ⇒   /ka↓.iŋi/区分け /kɯwake↓/ ⇒ /kɯwake/ (demarcation)熊手 /kɯmade↓/ ⇒ /kɯmade/ (rake)
三拍語平板型LHH(H)Approximately 50% of three-mora nouns fall under this pattern.大人 /otona/ (adult)竈 /kamado/ (furnace)岬 /misaki/ (cape)不死身 (ɸɯʑimi/ (immortality)刺激 /ɕiŋeki/ (stimulation)閉鎖 /he:sa/ (closure)眼張 /mebaɾɯ/ (black rockfish)為替 /kawase/ (money order/exchange)副詞 /ɸɯ̥kɯɕi/ (adverb)火薬 /kajakɯ/ (gunpowder)
四拍語頭高型Although relatively more prevalent in Sino-Japanese words and loanwords, less than 10% of four-mora nouns fall under this pattern. Many examples are either becoming 平板型 or 中高型, the latter case being prevalent with compound nouns.洪水 /ko↓:zɯ.i/ → /ko:zɯ.i/ (flood)相続 /so↓:zokɯ/ ⇒ /so:zokɯ/ (inheritance)快楽 /ka↓.iɾakɯ/ → /ka.iɾakɯ/ (pleasure)高禄 /ko↓:ɾokɯ/ ⇒ /ko:ɾokɯ/ (high salary)概念 /ga↓.ineN/ (concept)鉱山 /ko↓:zaN/ (mine)ゴールド /go↓:ɾɯdo/ (gold)恋文 /ko↓.ibɯmi/ ⇒ /ko.ibɯmi/ > /ko.i↓bɯmi/ (love letter)原品 /ge↓NpiN/ ⇒ /geNpiN/ (original article)現品 /geNpiN/ → /geNpi↓N/ (goods in stock)
四拍語中高型②A little more than 10% of four-mora words fall under this pattern with some becoming 平板型. It tends to appear in loanwords and older and/or complex native compound nouns. 小上がり /koa↓ŋaɾi/ (raised tatami-floored seating area)検非違使 /kebi↓.iɕi/ (judicial chief)蔵人 /kɯɾa↓Ndo/ (sake/soy sauce brewer)クリーム /kɯɾi↓:mɯ/ (cream)生娘 /kimɯ↓sɯme/ (innocent young woman)スカーフ /sɯ̥ka↓:ɸɯ/ (scarf)スコップ /sɯ̥ko↓Qpɯ/ (shovel)スズシロ /sɯzɯ↓ɕiɾo/ ⇒  /sɯzɯɕiɾo/ (kind of daikon) ストロー /sɯ̥to↓ɾo:/ (straw)巣離れ /sɯba↓naɾe/ (leaving the nest)
四拍語中高型③Comprised mostly of compound words, approximately less than 10% of four-mora words fall under this pattern, many of which are switching to 平板型. 小売値 /ko.ɯɾi↓ne/ (retail price)海風 /ka.iɸɯ↓:/ ⇒   /ka.iɸɯ:/ (sea breeze)好天 /ko:te↓N/ ⇒ /ko:teN/ (fine weather)荒天 /ko:te↓N/ ⇒ /ko:teN/ (stormy weather)原料 /geNɾyo↓:/ (raw materials)原子炉 /geNɕi↓ɾo/ (nuclear reactor)月面 /getsɯme↓N/ ⇒ /getsɯmeN/ (moon’s surface)雲合い /kɯmo.a↓.i/ → /kɯmo.a.i/ look of the sky)杉板 /sɯŋi.i↓ta/ ⇒ /sɯŋi.ita/ (cedar board)スプレー /sɯ̥pɯɾe↓:/ < /sɯ̥pɯ↓ɾe:/ (spray)
四拍語尾高型Very few extant examples exist, and of those, most show transitioning to 平板型 or sometimes 中高型, especially compounds.芸事 /ge:ŋoto↓/ → /ge:ŋo↓to/ ≒ /ge:ŋoto/ (traditional performing arts)雲脚 /kɯmo.aɕi↓/ ⇒ /kɯmo.aɕi/ (cloud movements)蔵入れ /kɯɾa.iɾe↓/ ⇒  /kɯɾa.iɾe/ (warehousing)気休め /kiyasɯme↓/ ⇒ /kiyasɯme/ (mere consolation)気紛れ /kimagɯɾe↓/ ⇒ /kimagɯɾe/ (whim)陣立て /dʑiNtate↓/ →  /dʑiNtate/ (battle formation)裾除け /sɯsojoke↓/ ⇒ /sɯsojoke/ (underskirt)関取 /seki̥toɾi↓/ ⇒ /seki̥toɾi/ > /seki̥to↓ɾi/ (sekitori)祟り目 /tataɾime↓/ ⇒ /tataɾime/ (the evil eye)茶柱 /tɕabaɕiɾa↓/ ⇒  /tɕaba↓ɕiɾa/ > /tɕabaɕiɾa/ (upright-floating tea stalk)
四拍語平板型The majority of four-mora nouns fall under this category at approximately 70%.  開発 /kaihatsɯ/ (development)賃借 /tɕiNakɯ/ (leasing)潔癖 /keQpeki/ (fastidiousness)原稿 /geNko:/ (manuscript)氷屋 /ko:ɾija/ (ice shop)恋人 /ko.ibito/ (lover)下手物 /getemono/ (low-quality product)訓読み /kɯNjomi/ (KUN reading)グランド /gɯɾaNdo/ (grand/sports ground)Some rare instances of switching over to 頭高型 can be observed.代金 /da.ikiN/ ⇒ /da↓.ikiN/ (charge)商業 /ɕo:ŋjo:/ ⇒ /ɕo↓:ŋjo:/ (commerce)
五拍語頭高型Although extremely rare, this pattern is relatively common among loanwords.赤蜻蛉 /a↓katoNbo/ ⇒ /akato↓Nbo/ (red dragonfly)十九日 /dʑɯ↓:kɯnitɕi/ → /dʑɯ:↓nitɕi/ (the nineteenth)ショッピング /ɕo↓QpiNgɯ/ (shopping)オフサイド /o↓ɸɯsa.ido/ (offside) ダイエット /da↓.ieQto/ (diet)ターミナル /ta↓:minaɾɯ/ (terminal)However, many loanwords are transitioning to 中高型③ as they become nativized just as native and SIno-Japanese nouns are as demonstrated above. ドキュメント /do↓kjɯmeNto/ →  /dokjɯme↓Nto/ (document)  ピーナッツ /pi↓:naQtsɯ/ →   /pi:na↓Qtsɯ/ (peanut)エピソード /e↓piso:do/ → /episo↓:do/ (episode)バガボンド /ba↓gaboNdo/ ⇒ /bagabo↓Ndo/ (vagabond)
五拍語中高型②Accounting for approximately 10% of five-mora nouns but relatively common in compounds in which the first morpheme is composed of just one mora. It is only occasionally seen in loanwords.  無教育 /mɯkjo↓:ikɯ/ (uneducated)女店員 /dʑote↓NiN/ (saleswoman)序奏曲 /dʑso↓:kjokɯ/ (overture)破壊力 /haka↓iɾjokɯ/ (destructive power)手数料 /tesɯ↓:ɾjo:/ (fee/charge)お姫様 /oçi↓mesama/  (princess)お嬢様 /oʑo↓:sama/ (young lady/another’s daughter)お線香 /ose↓Nko:/ (incense stick)お煎餅 /ose↓Nbe:/  (rice cracker)ソリューション /soɾjɯ↓:ɕoN/ (solution)
五拍語中高型③This pattern accounts for 50% of five-mora words and many words are transitioning to this pattern.長話 /naŋaba↓naɕi/ (long talk)長野県 /naŋano↓keN/ (Nagano Prefecture)東京都 /to:kjo↓:to/  (Tokyo Prefecture)城下町 /dʑo:ka↓matɕi/ (castle town)外回り /sotoma↓waɾi/ (perimeter)週刊誌 /ɕɯ:ka↓Nɕi/ (weekly publication)ハイボール /ha.ibo↓:ɾɯ/ (highball)It is also a probable alternative pattern for some 中高型④ words as being slightly easier to pronounce. 

同一視 /do:itsɯ̥↓ɕi/ ⇒ /do:it↓sɯ̥ɕi/ (equating one thing with another)ショベルカー /ɕobeɾɯka↓:/ ⇒  /ɕobeɾɯ↓ka:/ (power shovel)死学問 /ɕiniŋakɯ↓moN/ →  /ɕiniŋa↓kɯmoN/ (useless learning)
Despite how common this pattern is, some of the input that is leading to the increase in 平板型 among five-mora nouns comes from this pattern.随意筋 /dzɯ.i↓ikiN/ ⇒  /dzɯ.iikiN/ (voluntary muscle)消火栓 /ɕo:ka↓seN/ ⇒  /ɕo:kaseN/ (fire hydrant)消化管 /ɕo:ka↓kaN/ →  /ɕo:kakaN/ (alimentary canal)蒸気船 /dʑo:ki↓seN/ ⇒  /dʑo:kiseN/ (steamboat)巡視船 /dʑɯNɕi̥↓seN/ ⇒  /dʑɯNɕi̥seN/ (patrol boat)お人好し /oçi̥to↓joɕi/ ⇒  /oçi̥tojoɕi/ (simple soul)お目通り /omedo↓:ɾi/ ⇒  /omedo:ɾi/ (audience with a high-ranking person)
五拍語中高型④Although only standing at 10% of five-mora words, its percentage among compound words is much higher. 焼却炉 /ɕo:kjakɯ↓ɾo/ (incinerator) タイスコア /ta.isɯ̥ko↓a/ (tie score)大鋸っ屑 /oŋaQkɯ↓zɯ/ (sawdust)竪やの字 /tatejano↓ʑi/ (tateyanoji)
千一夜 /seNitɕi↓ja/ (One Thousand and One Nights)大家然 /ta.ikaze↓N/ (big figure-like)As aforementioned, however, it is losing words to 中高型③, which can then lead to 平板型 taking root.宝船 /takaɾabɯ↓ne/ →  /takaɾa↓bɯne/ (treasure ship)爆撃機 /bakɯŋeki̥↓ki/ ⇒  /bakɯŋe↓ki̥ki/ (bomber)飼い葉桶 /ka.iba.oke/ →  /ka.iba.oke/ (manger)頭数 /atamaka↓zɯ/ ⇒  /atamakazɯ/ > /atama↓kazɯ/ (head count)宝籤 /takaɾaku↓ʑi/ ⇒  /takaɾa↓kuʑi/ (lottery)助け舟 /tasɯ̥kebɯ↓ne/ →  /tasɯ̥ke↓bɯne/ (lifeboat)
五拍語尾高型Historically as well as in modern speech, this category has had few examples with most either transitioning to 中高型②③ or 平板型 at a rapid rate. 半ば過ぎ /nakabasɯŋi↓/ →  /nakabasɯŋi/ (beyond halfway)お正月 /oɕo:ŋatsɯ↓/ →  /oɕo:ŋatsɯ/ (New Year)道具立て /do:ŋɯdate↓/ ⇒  /do:ŋɯdate↓/ (preparation of tools)所帯持ち /ɕota.imotɕi↓/ ⇒  /ɕota↓imotɕi↓/ > /ɕota.i↓motɕi↓/ (married person)紹介所 /ɕo:ka.iʑo↓/ ⇒  /ɕo:ka.iʑo↓/ (agency)女振り /oNnabɯɾi↓/ ⇒  /oNnabɯɾi↓/ (woman’s charm)供え物 /sona.emono↓/ ⇒ /sona.emono/ (offering to the gods)他人様 /taniNsama↓/ <  /taniNsa↓ma/ (outsider)*配電所 /ha.ideNɕo↓/ ⇒ /ha.ideNɕo↓/ (power station)神宮寺 /dʑiNgɯ:ʑi↓/ (Jingu Shrine)

*The traditional reading of this word is よそさま, but たにんさま has all but replaced it.
五拍語平板型This is the second most likely pattern for five-mora nouns, standing at 30%+ and growing.向こう側 /mɯko:ŋawa/ (the other side)同居人 /do:kjyoniN/ (housemate)同音語 /do:oNŋo/ (homophone)千家流 /seNkaɾjɯ:/ (Senke school of tea ceremony)お餞別 /oseNbetsɯ/ (farewell gift)死に所 /shinidokoro/ (fitting place to die) 炊き合わせ /taki.awase/ (simmered assortment)羽交い絞め /haŋa.iʑime/ (Nelson hold)博多織 /hakata.oɾi/ (Hakata cloth)蟇股 /ka.eɾɯmata/ (beam clamp)
六拍語 頭高型Excessively rare among six-mora nouns, many examples are transitioning to either 中高型③ or 中高型④.食いしん坊 /ku↓iɕiNbo:/ ⇒ /ku.iɕi↓Nbo:/ (glutton)東海道 /to↓:ka.ido:/ ⇒  /to:ka↓ido:/ (Tokaido)曽祖母さん /çi↓:ba:saN/ ⇒  /çi:ba↓:saN/ (great-grandmother)バイオレット /ba↓i.oɾeQto/ →  /ba.i.oɾe↓Qto/ (violet)
六拍語中高型②This pattern is excessively rare among six-mora nouns with the following example being the only noticeable example among native words.お巡りさん /oma↓waɾisaN/ (police officer)In the case of loanwords, words whose first syllable in the original language were both stressed and contained consonant clusters had their consonant clusters broken up to fit Japanese phonology, thus disturbing the accent nucleus to what would then be the second mora. Over time, however, the nativization process leads to the accent nucleus transitioning to 中高型③④ or 平板型.ドレッシング /doɾe↓QɕiNŋɯ/ (dressing)トレーニング /toɾe↓:niNŋɯ/ (training)トランポリン /toɾa↓NpoɾiN/ →  /toɾaNpo↓ɾiN/ (trampoline) トレッキング /toɾe↓QkiNŋɯ/ ⇒  /toɾeQkiNŋɯ/ (trecking)
六拍語中高型③This pattern accounts for 35% of six-mora nouns.同級生 /do:kjɯ↓:se:/ (classmate)従業員 /dʑɯ:ŋjo↓:iN/ (employee)障害物 /ɕo:ŋa↓ibɯtsɯ/ (obstacle)蕎麦饅頭 /sobamaNdʑɯ:/ (soba manju)It is also a probable alternative pattern for some 中高型④ words as being slightly easier to pronounce. 
ショーウインドー /ɕo:u↓iNdo:/ > /ɕo:ui↓Ndo:/ (display window)小学生 /ɕo:ŋa↓kɯse:/ < /ɕo:ŋakɯ↓se:/ (elementary student)Despite how common this pattern is, some of the input that is leading to the increase in 平板型 among six-mora nouns comes from this pattern.
水耕法 /su.iko↓:ho:/ ⇒  / su.iko:ho:/ (hydroponics)召喚状 /ɕo:ka↓Ndʑo:/ ⇒  /ɕo:kaNdʑo:/ (subpoena)大腸菌 /da.itɕo↓:kiN/ ⇒  /da.itɕo:kiN/ (colon bacteria)
六拍語中高型④This pattern accounts for 35% of six-mora nouns, and many words from other patterns are transitioning to it. 随意科目 /dzɯ.i.ika↓mokɯ/ (elective course)所得控除 /ɕotokɯ̥ko↓:ʑo/ (tax deduction)蒸気ポンプ /dʑo:ki̥po↓Npɯ/ (steam pump)消化不良 /ɕo:kaɸɯ↓ɾjo:/ (indigestion)遊び相手 /asobi.a↓ite/ (playmate)女心 /oNnaŋo↓koɾo/ (woman’s heart)エスペラント /esɯ̥peɾa↓Nto/ (Esperanto)アップルパイ /aQpɯɾɯ↓pa.i/ (apple pie)シンガポール /ɕiNŋapo↓:ɾɯ/  (Singapore)Perhaps do to covering such a large percentage, some six-mora nouns of this pattern are transitioning to 平板型.黄昏時 /tasoŋaɾe↓doki/ ⇒  /tasoŋaɾedoki/ (twilight)
六拍語中高型⑤This pattern is quite rare but can be found more commonly in compound nouns.傷害致死 /ɕo:ŋa.itɕi↓ɕi/ (manslaughter)敗北主義 /ha.ibokɯ̥ɕɯ↓ŋi/ (defeatism)掃き出し窓 /hakidaɕima↓do/ (sliding glass door)押割り麦 /oɕiwaɾimɯ↓ŋi/ (pressed barley)狸爺 /tanɯkiʑiʑi↓:/ (cunning old man)戦争前 /seNso:ze↓N/ (before WWII) ジングルベル /dʑiNŋɯɾɯbe↓ɾɯ/ (jingle bells)
There are some examples of the accented mora potentially falling a mora sooner – 中高型④ – as a means of making the word easier to pronounce.
朧月夜 /oboɾozɯki↓jo/ →  朧月夜 /oboɾozɯ↓kijo/ (misty moonlit night)ビーフシチュー /bi:ɸɯɕi̥tɕɯ↓:/ ⇒  /bi:ɸɯɕi̥↓tɕɯ:/ (beef stew)
六拍語尾高型This pattern is excessively rare with many being alternatively pronounced as 中高型⑤ or 平板型. It is only likely to appear in compound nouns in which only the last two morae make up the second morpheme.大仏様 /da.ibɯ̥tsɯsama↓/ → /da.ibɯ̥tsɯsa↓ma/  (Daibutsu)確り者 /ɕi̥Qkaɾimono↓/ ⇒  /ɕi̥Qkaɾimono/  (gutsy person)預かり物 /azɯkaɾimono↓/ ⇒  /azɯkaɾimono/ (something left in charge)臆病者 /okɯbjo:mono↓/ ⇒  /okɯbjo:mono/ (coward)
六拍語平板型This pattern acounts for 20% of six-mora nouns.闘牛場 /to:ŋjɯ:ʑo:/ (bullring)配線盤 /ha.iseNbaN/ (plugboard)能動態 /no:do:ta.i/ (active voice)大政治家 /da.ise:ʑika/ (major politician)橙色 /da.i.da.i.iɾo/ (orange color)X線 /eQkɯ̥sɯseN/ (X-ray)思い違い /omo.itɕiŋa.i/ (misconception)お好み焼き /okonomijaki/ (okonomiyaki)沢庵漬け /takɯ(w)aNdzɯke/ (pickled daikon)取り扱い /toɾi.atsɯka.i/ (handling)

Systemic Pitch Accent Rules (To Be Continued…)

Although how words are accented is not entirely predictable and at some level arbitrary, there are in fact many systemic ‘rules’ that can be observed which add a great deal of order to the pitch accent system as a whole.