Pronunciation II: Consonants
日本語の音韻体系②・子音
Although Japanese has fewer sounds than English, its consonant inventory includes sounds not found in English. Mastering these unique consonant sounds may be challenging at first, but learning their correct pronunciation will significantly improve your Japanese speech over time.
Japanese consonants can be grouped into seven types:
- Unvoiced Consonants
- Voiced Consonants
- Nasal Consonants
- Palatal Consonants
- Liquid Consonant
- Semi-Vowels
- Moraic Consonants
Symbol Notation: // denotes pronunciations from the perspective of phonemes, which are the sounds of the language perceived by speakers as being distinct. In contrast, [] indicates pronunciation from a phonological standpoint. For example, if a sound is represented as /X/, its allophones1 (variations) are then denoted by [X1], [X2], etc.
Unvoiced Consonants
無声子音
In the throat are the vocal cords, which do not vibrate when an unvoiced consonant is produced. In English, unvoiced consonants are also accompanied by aspiration—a puff of air released when the sound is made. To feel this effect, place your hand in front of your mouth and say the word “king.” The burst of air you feel is aspiration. In Japanese, unvoiced consonants are produced with far less aspiration, but aspiration is still notably present.
/k/: The Japanese /k/ is made by pressing the back of the tongue against the soft palate. It sounds like the English /k/ but with less aspiration.
| Kado (corner) | Kagi (keys) | Kao (face) | Kaisha (company) | Kaidan (stairs) |
| Kazoku (family) | Kamera (camera) | Kawa (river) | Kōhii (Coffee) | Kotoshi (this year) |
/s/: The Japanese /s/ is the same as its English counterpart. However, when paired with the vowel /i/, it is pronounced as [sh].
| Sakana (fish) | Sukoshi (little) | Sentaku (laundry) | Hasu (lotus) | Sa (difference) |
| Sampo (stroll) | Sono (garden) | Su (nest/vinegar) | Kaosu (chaos) | Saka (slope) |
/sh/: The Japanese /sh/2 is made by having the middle of the tongue bent and raised toward the hard palate. Many English native learners misperceive it as /s/, but Chinese native learners will find it more familiar.
| Shima (island) | Shumi (hobby) | Shio (salt) | Shefu (chef) | Shitagi (underwear) |
| Shakai (society) | Shokuji (meal) | Shatsu (shirt) | Shitsumon (question) | Shiai (match) |
/t/: The Japanese /t/ is made by placing the blade of the tongue behind the upper teeth. When followed by the vowel /u/, it is pronounced as [ts]. The “t” is by no means silent. When followed by the vowel /i/, /t/ is pronounced as [ch].
| Ta (rice field) | Tamago (egg) | Te (hand) | Tōkyō (Tokyo) | Tatoeba (for example) |
| Tabako (tobacco) | Taifū (typhoon) | Tani (valley) | Tanoshimi (fun) | Tatami (tatami) |
| Tsuma (wife) | Tsuba (spit) | Tsukue (desk) | Tsuyoi (strong) | Tsuyu (rainy season) |
| Tsuchi (ground) | Aisutī (iced tea) | Pātii (party) | Chīmu (team) | Maikurosofuto Tīmuzu (Microsoft Teams) |
The mora /ti/ does not occur in native words, but it does occur in recently borrowed loanwords. Loanwords of longer standing exhibit [chi] instead. Even for words which do contain /ti/, many speakers still pronounce /ti/ as [chi].
/ch/: The Japanese /ch/3 is made by placing the blade of the tongue behind the alveolar ridge4. This difference in articulation is difficult for English natives to mimic. However, it will be familiar to Chinese and Korean native learners.
| Chūi (caution) | Chi (blood) | Chūshajō (parking lot) | Chizu (map) | Ocha (tea) |
| Chiri (geography) | Chikatetsu (subway) | Chekku (check(ing)) | Chotto (a little) | Tsumi (sin) |
/h/: The Japanese /h/ is identical to its English counterpart. It becomes [f] when followed by the vowel /u/. When followed by the vowel /i/, it sounds like the h-sound in the word “hue.”
| Hachi (eight) | Hito (person) | Higashi (east) | Hontō (really) | Hambāgā (hamburger) |
| Heya (room) | Hidari (left) | Hoteru (hotel) | Hikōki (airplane) | Hikari (light) |
/f/: The Japanese /f/ is not the same as its English counterpart. Instead of placing the front teeth against the bottom lip, the lips are compressed, and air is blown through them. Remember, no teeth!
| Fukuzatsu (complicated) | Fuku (clothes) | Forō (follow) | Fikushon (fiction) | Futon (futon) |
| Futsū (usual) | Furui (old) | Fōku (fork) | Finrando (Finland) | Fakkusu (fax) |
/p/: The Japanese /p/ is like its English counterpart but with less aspiration.
| Piano (piano) | Pokemon (Pokémon) | Supūn (spoon) | Purin (pudding) |
| Poketto (pocket) | Purezento (present) | Kapibara (capybara) | Purintā (printer) |
Voiced Consonants
有声子音
Unlike unvoiced consonants, voiced consonants are produced with vibration of the vocal folds. Unvoiced-voiced consonant pairs are common. For example, /d/ is the voiced counterpart of /t/. In Japanese, voiced consonants are pronounced with noticeably stronger vocal fold vibration compared to English.
| Unvoiced Counterpart | Voiced Counterpart |
| /k/ | /g/ |
| /s/ | /z/~[dz] |
| /sh/ | /j/ |
| /t/ | /d/ |
| [ts] | [dz] |
| /ch/ | [dj] |
| /h/ | /b/ |
/g/: In Standard Japanese, the pronunciation of /g/ depends on its position in a word. Word-initial /g/ is pronounced as the [g] in the English word “go,” but word-medial /g/ is often pronounced as [ŋ (ng)], which is the consonant heard at the end of the English word “sing.” If you find the [ŋ] pronunciation too difficult to produce in between vowels, you can still substitute it with [g] and be understood. In fact, many native speakers pronounce /g/ solely as [g].
| Gaikoku (foreign country) | Gakusei (student) | Gozen (A.M.) | Gohan (cooked rice) | Kage (shadow) |
| Gaikokujin (foreigner) | Gitā (guitar) | Gogo (P.M.) | Daigaku (college) | Fugu (pufferfish) |
Pronunciation Tip: To get used to pronouncing /g/ as [ng], take the word iga (burr). Start by saying “ing-ah.” Retry by making sure the first mora is just /i/, then pronounce [ŋa] as one mora.
/z/~[dz]: /z/ is the voiced counterpart of /s/, and [dz] is the voiced counterpart of [ts]. However, these two sounds are blurred by most speakers5. To conceptualize how [dz] sounds, think of the word “kids.” Additionally, when /z/ is followed by /i/, it becomes [j] as in “judge.”
| Kazu (number) | Zutsū (headache) | Zen (Zen) | Zō (elephant) | Zeikin (tax) |
| Kaze (wind/cold) | Shiji (instruction) | Tsu(d)zuki (continuance) | Zōo (Hatred) | Zaseki (seat) |
/j/~[dj]: /j/6 is the voiced counterpart of /sh/ and [dj] is the voiced counterpart of /ch/. However, these two sounds are blurred by most speakers7.
| Joshu (assistant) | Jitensha (bicycle) | Jettoki (jet aircraft) | Joō (queen) | Ji (character) |
| Jikan (time) | Jidōsha (vehicle) | Jesuchā (gesture) | Jōzu (skillful) | Chi(d)jimu (to shrink) |
/d/: The Japanese /d/ is made by placing the blade of the tongue behind the upper teeth. It becomes [(d)j] when followed by /i/ and [(d)z] when followed by /u/.
| Dēto (a date) | Doa (door) | Te(d)zukuri (handmade) | Shio(d)zuke (salting) | Denki (electricity) |
| Doku (poison) | Dizunī (Disney) | Hana(d)ji (nosebleed) | Hada (skin) | Dansei (male) |
In recently borrowed loanwords, /di/ can occur. Loanwords of longer standing exhibit [(d)j]—e.g. rajio (radio).
/b/: Acoustically, /b/ is the voiced counterpart of /p/, but in Japanese it is treated as the voiced counterpart of /h/.
| Batā (butter) | Baka (idiot) | Terebi (TV) | Bōeki (trade) | Bangō (number) |
| Basho (place) | Kaba (hippo) | Benri (convenient) | Fuben (inconvenient) | Butaniku (pork) |
Nasal Consonants
鼻音
Nasal consonants are pronounced by both vibrating the vocal folds and passing air through the nose.
/n/: Made with the blade of the tongue on the back of the upper teeth with /a/, /e/, and /o/, behind the alveolar ridge of the mouth with /i/ (like in news), and behind the teeth with /u/ (like in noon).
| Niwa (garden) | Neko (cat) | Neru (to sleep) | Nōto (notebook) | Netsu (fever) |
| Nekutai (necktie) | Nomu (to drink) | Noru (to ride) | Nioi (smell) | Nodo (throat) |
/m/: Pronounced by bringing the lips together in the same way as in English.
| Miso (miso) | Minato (harbor) | Musuko (son) | Mushi (bug) | Machi (town) |
| Manga (manga) | Mukashi (olden days) | Musume (daughter) | Mori (forest) | Michi (road) |
Liquid Consonant
流音
Liquid consonants are sounds like /l/ and /r/. In most languages, there are at least two liquid consonants. In Japanese, there is only one, but its pronunciation is unique yet varied.
/r/: It is typically pronounced as a flap, which sounds like the “t” in the word “water” in American English. When word-initial, /r/ sounds similar to /d/, but the tongue only taps the alveolar ridge without prolonged contact. Also, the Japanese /r/ and /d/ are not pronounced in the same location of the mouth.
The variant pronunciation [l] is very common, and it is useful when learners cannot produce the flap pronunciation. In coarse speech, it is even pronounced as a trill.
| Rakuen (paradise) | Renga (brick) | Rekishi (history) | Ari (ant) | Kuru (to come) |
| Rei (zero) | Roku (six) | Repōto (report) | Ura (reverse side) | Suru (to do) |
Semi-Vowels
半母音
Semi-vowels resemble vowels in that the tongue does not touch any part of the mouth. However, the tongue moves to partially constrict airflow, similarly to how consonants are articulated.
/y/: For /y/, the tongue is raised close to the hard palate, creating a narrow passage for air to flow through. This sound is very similar to its English counterpart, though it is slightly more tense in Japanese. Traditionally, /y/ pairs only with the vowels /a/, /u/, and /o/, but it also appears with /e/ in loanwords.
| Yubi (finger) | Yakusoku (promise) | Yon (four) | Yasumi (rest) | Yasashī (nice) |
| Yowai (weak) | Yomu (to read) | Yama (mountain) | Yasui (cheap) | Yēru (Yale) |
/w/: The Japanese /w/ sound is closer to the Japanese /u/ than to the English /w/. It is produced by compressing the lips rather than protruding them outward. In modern Japanese, /w/ is typically paired only with /a/, though some speakers still use it with /o/. In recent loanwords, however, /w/ can appear with every vowel except /u/.
| Wana (trap) | Warui (bad) | Awa (bubble) | Weitoresu (Waitress) | Wani (crocodilian) |
| Wakai (young) | Watashi (I) | Wikipedia (Wikipedia) | Wēruzu (Wales) | Webusaito (website) |
Palatalized Consonants
拗音
Palatalized consonants are made by placing the body of the tongue touch against the hard palate of the mouth. These consonants are typically restricted to the vowels /a/, /u/, and /o/8, and they are all created with the help of the consonant /y9/ merging with a preceding consonant.
As is the case for all these consonants, never insert the vowel /i/ between the base consonant and the palatalized glide that follows, as this will most likely change the word.
/ky/:
| Kyaku (customer) | Kyō (today) | Kyōshitsu (classroom) | Kyonen (last year) | Jōkyō (situation) |
| Kyōto (Kyoto) | Kyū (nine) | Kyōdai (sibling) | Jukyō (Confucianism) | Kyōkai (church) |
/gy/: In word-medial position, /gy/ may alternatively be pronounced as [ŋy].
| Gyaku (opposite) | Gyakutai (abuse) | Wagyū (wagyu) | Gyūdon (gyudon) | Gyogyō (fishing) |
| Gyūniku (beef) | Gyōji (event) | Gyūsha (cow shed) | Gyōza (potstickers) | Gyōgi (manners) |
/sy/ > [sh]; /zy/ > [j]: The combination of /s/ and /y/ results in [sh]. Likewise, the combination of /z/ and /y/ results in [j].
| Shōyu (soy sauce) | Shokudō (cafeteria) | Shū (week) | Jaguchi (faucet) | Shashin (picture) |
| Shukudai (homework) | Jū (ten/gun) | Jiko (accident) | Ha’isha (dentist) | Shako (carport) |
/ty/ > [ch]; /dy/ > [(d)j]: In native words, the combination of /t/ and /y/ results in [ch]. Likewise, the combination of /d/ and /y/ results in [(d)j]. Due to an influx of loanwords, however, /ty/ and /dy/ do occasionally occur.
| Kōcha (black tea) | Chero (cello) | Cheko (Czech) | Cheju (Jeju) | Chairo (brown) |
| Chōsa (investigation) | Chesu (chess) | Dyuetto (duet) | Dyuo (duo) | Tyuruku (Turkic) |
/ny/: This consonant is only found in onomatopoeic expressions in native vocabulary, but it is very common in Sino-Japanese vocabulary.
| Nyā (meow) | Gyūnyū (milk) | Nyūshu (obtaining) | Nyūyoku (bathing) | Nyūryoku (input) |
| Nyūsu (news) | Nyōbō (wife) | Nyō (urine) | Nyūjō (admission) | Nyūnen (careful) |
/hy/: This consonant is rare in native vocabulary, but it is very common in Sino-Japanese vocabulary.
| Hyaku (100) | Hyōga (glacier) | Hyōka (evaluation) | Hyōban (reputation) | Hyōjun (standard) |
| Hyō (vote) | Hyūga (Hyuga) | Hyōgen (expression) | Hyōmen (surface) | Hyōzan (iceberg) |
/by/: This consonant only occurs in onomatopoeic expressions in native vocabulary. Otherwise, it occurs in a handful of productive Sino-Japanese roots.
| Byōki (illness) | Byū (view) | Byōin (hospital) | Byō (second) | Byōnin (sick person) |
| Byōsha (depiction) | Gobyū (fallacy) | Byōdō (equality) | Gabyō (tack) | Byōshō (sickbed) |
/py/: This consonant only occurs in onomatopoeic expressions in native vocabulary. Otherwise, it is found in only a handful of loanwords.
| Pyūma (puma) | Pyū (swoosh) | Pyon’yan (Pyongyang) | Pyonpyon (hop-hop) |
/my/: This consonant is rare in native vocabulary, and its use is limited even in Sino-Japanese words and loanwords.
| Myaku (pulse) | Myūjikaru (musical) | Myōji (surname) | Bimyō (subtle) |
| Myō (weird) | Myunhen (Munich) | Kimyō (peculiar) | Myūtsū (Mewtwo) |
/ry/: This consonant is the hardest for English native speakers to pronounce. While /ry/ is uncommon in native Japanese vocabulary, it is frequently used in Sino-Japanese words.
| Ryū (dragon) | Ryaku (abbreviation) | Ryōri (cooking) | Ryōshin (parents) |
| Ryō (dorm/quantity) | Ryokō (travel) | Ryokan (Japanese inn) | Ryōshi (fisherman) |
Long Consonants
長子音
All non-voiced consonants, except /h/, can be “doubled” in native vocabulary. Japanese speakers perceive long consonants as two equal-length morae10. To represent a long consonant in romanization, simply double the first letter of the consonant. For ease of reading, the only exception is /ch/, which is spelled as “tch” when lengthened.
/kk/:
| Makka (bright red) | Yokka (four days) | Sakka (author) | Sakkā (soccer) | Sakki (moment ago) |
| Kokka (nation) | Mikka (three days) | Nikki (diary) | Shikkari (firmly) | Sukkari (completely) |
/ss/:
| Kassōro (runway) | Issai (entirety) | Hassō (conception) | Dassen (derailment) |
| Zassō (weeds) | Dassō (escape) | Tassuru (to reach) | Kassai (applause) |
/ssh/:
| Zasshi (magazine) | Asshuku (compression) | Nesshin (zealous) | Hasshō (outbreak) |
| Isshun (moment) | Kasshoku (dark brown) | Hassha (departure) | Dasshutsu (break-out) |
/tt/, /tts/:
| Batta (grasshopper) | Settei (settings) | Kottōhin (antique) | Chotto (a little) |
| Kettei (decision) | Settai (entertainment) | Shittsui (forfeiture) | Motto (more) |
/tch/:
| Shutchō (business trip) | Matcha (matcha) | Pitchā (pitcher) | Hatchū (ordering) |
| Satchūzai (insecticide) | Katchū (armor) | Matchi (match) | Matcho (macho) |
/pp/:
| Shippai (failure) | Happyō (announcement) | Shippo (tail) | Seppuku (harakiri) |
| Shuppatsu (departure) | Happa (leaves) | Kappō (Japanese cuisine) | Appare (bravo) |
When loanwords are included, /h/ as well as a select number of voiced consonants can also be doubled. However, even though they remain spelled as doubled voiced consonants, they most often become unvoiced when spoken.
| Bahha (Bach) | Beddo (bed) | Baggu (bag) | Bajji (badge) |
The Moraic Nasal
撥音
Japanese has a unique voiced consonant called the “moraic nasal” (/ɴ/), which counts as its own mora. Although it is typically transcribed as “n” in romanization, its pronunciation varies depending on the surrounding sounds.
In utterance-final position, at the very least, it is traditionally viewed as a uvular nasal consonant. As this classification suggests, /ɴ/ is produced with the uvula (a teardrop-shaped flap of tissue that hangs down in the back of the throat. When pronouncing /ɴ/ in this position, the mouth constricts as air passes through the nose, similarly to when producing the consonant sound /n/. However, the exact positioning of the tongue aligns with that of the preceding vowel.
From this description alone, what is clear about /ɴ/ is that it assimilates with surrounding sounds. Assimilation is the process by which a sound is altered to become more similar to a neighboring sound. /ɴ/ has the following allophones.
Pronounced as [m]: When /ɴ/ occurs before a /p/, /b/, or /m/, it becomes [m]. This results in double /mm/.
| Sampo (walk) | Kampeki (perfect) | Tammatsu (device) | Sammyaku (mountain range) |
| Shimpai (worry) | Kampai (cheers) | Sembei (rice cracker) | Shimpan (refereeing) |
Pronounced as [n]: When /ɴ/ occurs before /t/, /d/, /n/, /r/, it becomes [n]. This results in double /nn/.
| Kanri (management) | Minna (everyone) | Kantō (Kanto) | Sentaku (laundry/choice) |
| Shinri (mentality) | Shinrai (faith) | Tennō (emperor) | Handan (judgment) |
Pronounced as [ŋ]: When /ɴ/ occurs before /k/ and /g/, it assimilates into [ŋ]. Because /n/ is pronounced the same way in English under these circumstances, [ŋ] will be spelled as “n” in romanization for simplicity.
| Shinka (evolution) | Kankaku (feeling) | Sanka (participation) | Tango (word) |
| Kingyo (goldfish) | Kangae (idea) | Kango (Sino-Japanese word) | Bangō (number) |
Pronounced as [ny]: When /ɴ/ is before /ch/ or [dj], it is pronounced in the same place of the mouth as these consonants, resulting in [ny]. The “y” indicates palatal articulation. To avoid confusion with the palatal consonant /ny/, this pronunciation is spelled as “n” in the words below.
| Kanja (patient) | Tanchiki (detector) | Kanjō (emotion) | Shinchū (brass) |
| Tanjōbi (birthday) | Kanji (Chinese characters) | Sanchō (summit) | Shinchō (height) |
Pronounced as [ũ]: When before vowels, /y/, /w/, /s/, /sh/, /z/, /h/, and /f/, /N/ sounds like a nasalized vowel. Typically, it sounds like a nasal [ũ], which is how it is spelled in the words below.
| Taũ’i (unit) | Koũwaku (perplexity) | Deũsha (train) |
| Kaũzei (tariff) | Kiũyū (finance) | Kaũsai (Kansai) |
Pronounced as [ɴ]: /ɴ/’s default pronunciation, seen in utterance-final position.
| Nihon (Japan) | Hon (book) | En (yen) | Ton (ton) | Ten (point/heaven) |
| Kan (can) | Kin (gold) | Gin (silver) | Hen (strange) | Gan (cancer) |
Although the allophones of /ɴ/ are predictable, there is considerable flexibility in their distribution. For example, some speakers pronounce /ɴ/ predominantly as [n]. In singing and historical settings, it is unilaterally pronounced as [m]. In intentionally slowed speech, its uvular pronunciation [ɴ] becomes more prominent.
- A “phoneme” is what constitutes a “speech sound” as understood by the speakers of a language. In actuality, a phoneme can have multiple pronunciations. These pronunciations often correspond to separate “phonemes” in other languages. However, in the context of a given language, multiple pronunciations of a phoneme are treated as one of the same thing, and these pronunciations are referred to as “allophones. ↩︎
- /sh/ began as an allophone of /s/ when followed by /i/ or /j/. Due to the influx of loanwords, the sound can now be seen paired with any vowel. This has led to its reclassification as a phoneme in its own right. ↩︎
- /ch/ began as an allophone of /t/ when followed by /i/ or /j/. Due to the influx of loanwords, the sound can now be seen paired with any vowel. This led to its reclassification as a phoneme in its own right. ↩︎
- The gum-covered ridge of bone behind the teeth in the upper jaw. ↩︎
- The pronunciation [dz] is common after pauses and the moraic nasal /ɴ/. In all other environments, [z] and [dz] are interchangeable. ↩︎
- /j/ began as an allophone of /z/ when followed by /i/ or /j/. Due to the influx of loanwords, the sound can now be seen paired with any vowel. This led to its reclassification as a phoneme in its own right. ↩︎
- The pronunciation [dj] is common after pauses and the moraic nasal /ɴ/. In all other environments, [j] and [dj] are interchangeable. ↩︎
- Just as is the case with the consonant /y/, palatalized consonants are only used with /e/ in loanwords. ↩︎
- For simplicity, the IPA (International Phonetic Alphabet) is not utilized to represent palatalization, but the IPA notation for it is j. ↩︎
- It is not always the case that a long consonant is twice as long as its short counterpart. However, there must always be an audible difference in both length and intensity of the consonant. ↩︎