Japanese Japan (ja-JP)
This documentation was updated on December 17, 2023.
Creating grammars
The following subsections describe key issues for working with grammar documents in the Japanese language.
Grammar file encoding
Nuance has full internal Unicode support. Create your grammars using Shift-JIS, EUC-JP, or UTF-8. For example, your grammar header might be:
<?xml version=‘1.0’ encoding=‘Shift-JIS’?> <grammar xml:lang=“ja-JP” version=“1.0” root=“test”>
alphanum_lc built-in grammar
The alphanum_lc built-in grammar recognizes a connected string of up to 20 digits and lower case alphabetic characters. For example, this grammar could be used to recognize a product code or order number.
Valid characters are the letters a-z. The pronunciation of the letter `z’ as the English-style “zed” is recognized, but the American-style “zii” is not, and would probably be misrecognized as the letter “c”. Digits are 0-9: for 0, callers can say ぜろ, れい, まる ; for 4, only よん is allowed; and for 9, きゅう is allowed. The grammar does not support し for 4, or く for 9.
Note: The alphanum_lc built-in grammar replaces the alphanum built-in grammar.
alphanum built-in grammar
The alphanum built-in grammar recognizes a connected string of up to 20 digits and alphabetic characters. For example, this grammar could be used to recognize a product code or order number.
Valid characters are the letters a-z. The pronunciation of the letter `z’ as the English-style “zed” is recognized, but the American-style “zii” is not, and would probably be misrecognized as the letter “c”. Digits are 0-9: for 0, callers can say ぜろ, れい, まる ; for 4, only よん is allowed; and for 9, きゅう is allowed. The grammar does not support し for 4, or く for 9.
boolean built-in grammar
The boolean grammar collects an affirmative or negative response.
Properties
The y and n parameters let you associate any two touchtone buttons as synonyms for yes and no.
| Parameter | Description |
|---|---|
| y | Desired DTMF digit to be equivalent to はい (default = 1) |
| n | Desired DTMF digit to be equivalent to いいえ (default = 2) |
Examples
| Caller says… | MEANING key |
|---|---|
| はい | true |
| いいえ | false |
ccexpdate built-in grammar
The ccexpdate grammar understands the expiration date on a credit card. Expiration dates are usually a month and a year, and are often embossed on a credit card in the form “mm/yy.” The grammar recognizes variations on the date, for example, " ニ セン ゴ ネン ジュウ ニ ガツ" , “ジュウ ニ ゼロ ゴ”, and “ジュウ ニ ガツ ニ セン ゴ ネン” etc.
Some credit cards are stamped with a day of the month as well as the month and year; the ccexpdate grammar recognizes these dates as well. However, the only day of the month it recognizes is the last day of a given month. For example, “ニ セン ゴ ネン ジュウ ニ ガツ サン ジュウ イチ ニチ.” The grammar does not check for leap years: both February 28 and February 29 are recognized, regardless of the given yea
creditcard built-in grammar
The creditcard grammar understands a caller saying a credit card number, optionally preceding the number with the credit card name, or the words “カアド バンゴウ” or " カアド." For example, a caller can say, “ビザ カアド ヨンゼロ イチﾉ,” " カアド バンゴウ ヨン ゼロ イチﾉ," or “ヨン ゼロ イチﾉ.”
currency built-in grammar
The currency grammar collects currency amounts using Yen.
The caller must speak the word 円 (えん) at the end of the monetary amount.
To speak fractional amounts, the caller must speak the number of Yen, followed by a decimal point (てん) and then each digit of the fraction one-at-a-time (natural numbers are not allowed for the fraction; for example に ぜろ is allowed but not にじゅう). In summary, the caller must speak using this format: <Natural number for Yen> てん <digits for 銭（せん）> 円（えん）
| MEANING | Contains a string in this form: JPY main_unit_amount.subunit_amount If the caller omits the main unit or subunit amount, then that field is zero. The string contains a leading zero if the subunit amount is collected without the main unit. |
|---|---|
| SWI_literal | Contains the exact text that was recognized. |
Examples
The following examples show various currency amounts spoken by callers with the resulting values placed into the MEANING key.
| Caller says | MEANING |
|---|---|
| ０円（ぜろえん, れいえん） | JPY0.00 |
| １円（いちえん） 1.00円（いちてんぜろぜろえん） | JPY1.00 |
| ２円（にえん） | JPY2.00 |
| 2.01円（にてんれいいちえん） | JPY2.01 |
| 2.10円（にてんいちれいえん） | JPY2.10 |
| 12.20円（じゅうにてんにえん） | JPY12.20 |
| 1998円（せんきゅうひゃくきゅうじゅうはちえん） | JPY1998.00 |
| 999999998円（きゅうおくきゅうせんきゅうひゃくきゅうじゅうきゅうまんきゅうせん きゅうひゃくきゅうじゅうはちえん） | JPY999999998.00 |
| 14円（じゅうよんえん, じゅうよえん） | JPY14.00 |
| 20円（にじゅうえん） | JPY20.00 |
| 20.10円（にじゅうてんいちえん, にじゅうてんいちれいえん） | JPY`20.10 |
| 0.20円（れいてんにえん、れいてんにぜろえん） | JPY0.20 |
| 1.24円（いってんによんえん） | JPY1.24 |
| 1000000円（ひゃくまんえん） | JPY1000000.00 |
| 1000000.25円（ひゃくまんてんにごえん） This example only works if the maxallowed parameter is set higher than the default. | JPY1000000.25 |
Here are examples of utterances that do not parse when spoken by callers:
| Caller says | Reason for not being recognized |
|---|---|
| 2円1銭（にえんいっせん） | The grammar does not support the format: <yen> 円（えん） <seng> 銭（せん）. The caller would have to say “2．01円（にてんぜろいちえん）” for the intended meaning of 2.01. |
| 2.20円（にてんにじゅうえん） | Each digit must be spoken after the decimal point (speaking fractions as natural numbers is not allowed). |
| 201円（にぜろいちえん） | The grammar does not support a sequence of digits for the main unit amount; the caller must speak the whole number naturally. |
Properties examples
If you set granularityallowed to 10, the phrase " 15.10円（じゅうごてんいちれいえん）" is not allowed (because 15.10 is not a multiple of 10).
If you set minallowed=0, maxallowed=10, and granularityallowed=5, then only 0円（ぜろえん, れいえん）, 5円（ごえん）, and 10円（じゅうえん） can be recognized.
date built-in grammar
The date grammar accepts a date spoken in any of several formats.
Recognized phrases include:
6 月4日(ろくがつよっか)
2001年6月4日(にせんいちねんろくがつよっか)
平成13年6月4日(へいせいじゅうさんねんろくがつよっか)
4日(よっか)
6月4日月曜日(ろくがつよっかげつようび)
The grammar also accepts 昨日(きのう), 今日(きょう), and 明日(あした), which return values of -1, 0, and +1 respectively into the MEANING key.
Note that to be understood, at least the day of the month must be present. Phrases like 次の水曜日(つぎのすいようび) and 2001年6月(にせんいちねんろくがつ) are not understood.
Return keys/values
In addition to the standard date return keys (MEANING, SWI_literal, CENTURY, TWO_DIGIT_YEAR, YEAR, MONTH, DAY, and WEEKDAY), the Japanese date grammar returns a Western/Japanese style flag:
| JW | Set to ‘W’ if the caller spoke a western style date; set to ‘J’ if the caller spoke a Japanese style date. |
|---|
Examples
All keys are set whenever known. For example, the phrase 1911年12月4日水曜日 (せんきゅうひゃくじゅういちねんじゅうにがつよっかすいようび) sets the following keys:
| MEANING | 19111204 |
|---|---|
| WEEKDAY | 4 |
| YEAR | 1911 |
| CENTURY | 19 |
| TWO_DIGIT_YEAR | 11 |
| MONTH | 12 |
| JW | W |
The following examples show various dates spoken by callers (written in Kanji/Kana) with the resulting values placed into the MEANING key.
| Caller says | MEANING |
|---|---|
| 今日(きょう) | 0 |
| 明日(あした, みょうにち) | +1 |
| 明後日(あさって) | +2 |
| 昨日(きのう) | -1 |
| 一昨日(おととい) | -2 |
| 1911年の12月4日水曜日(せんきゅうひゃくじゅういちねんのじゅうにがつよっかすいよう び) | 19111204 |
| 12月4日の水曜日(じゅうにがつよっかのすいようび) 12月4日水曜日(じゅうにがつよんにちすいようび) | ????1204 |
| 1日(ついたち) | ??????01 |
| 7日(なのか) | ??????07 |
| 31日(さんじゅういちにち) | ??????31 |
| 1月１日(いちがつついたち) | ????0101 |
| 12月1日(じゅうにがつついたち) | ????1201 |
| 12月31日(じゅうにがつさんじゅういちにち) | ????1231 |
| 1月27日の水曜日(いちがつにじゅうしちにちのすいようび) | ????0127 |
| 平成9年の1月1日水曜日(へいせいきゅうねんのいちがつついたちすいようび) 1997年1月1日(せんきゅうひゃくきゅうじゅうななねんいちがつついたち) 1997年の1月1日水曜日(せんきゅうきゃくきゅうじゅうななねんのいちがつついたちすいよ うび) | 19970101 |
| 1999年12月31日(せんきゅうひゃくきゅうじゅうきゅうねんじゅうにがつさんじゅういち にち) | 19991231 |
| 2097年1月1日(にせんきゅうじゅうななねんいちがつついたち) | 20970101 |
| 平成12年1月1日(へいせいじゅうにねんいちがつついたち) | 20000101 |
Here are examples of utterances that are not handled by the grammar:
| Caller says | Reason for not being recognized |
|---|---|
| 水曜日(すいようび) | Too ambiguous. |
| 昭和65年12月4日(しょうわろくじゅうごねんじゅうにがつよっか) | There is no year 65 in shouwa. |
| 12 4 (じゅうによん) | Too ambiguous; does not contain the day. |
| 1日(いちにち) | The caller should use the more common ついたち. |
| 12月32日(じゅうにがつさんじゅうににち) | No month has 32 days. |
| 2年の6月4日(にねんのろくがつよっか) | Needs an emperor era or western date. |
| 2297年1月1日(にせんにひゃくきゅうじゅうななねんいちがつついたち) | The value is above the maximum date of 21991231. |
| 1997 12月13(せんきゅうひゃくきゅうじゅうななじゅうにがつじゅうさん) | Needs ねん for year. |
| 1850年6 6(せんはっぴゃくごじゅうねんろくろく) | Needs “month” and “day” words. |
Properties
The range of possible dates is different depending whether the caller speaks Japanese or Western style:
| Style | Earliest | Latest |
|---|---|---|
| Western | 18000101 | 21991231 |
| Japanese | 18670101（明治元年１月１日） | 20501231（平成６１年１２月３１日） |
Extensions to VoiceXML specification
The VoiceXML specification does not specify how to handle ambiguous months (for example, “Monday the 4th”). The implementation allows the ambiguity by returning questions marks (?) for ambiguous information in both speech and DTMF grammars.
Japanese style years are supported for:
明治(めいじ)
大正(たいしょう)
昭和(しょうわ)
平成(へいせい)
For Western dates, the grammar allows a prefix せいれき, for example せいれき2001年7月4日(せいれきにせんいちねんしちがつよっか) .
digits built-in grammar
Valid digits are 零一二三四五六七八九.
The digit 零 can be pronounced as ゼロ, レイ, or マル. The grammar does not support し for 4, or く for 9.
number built-in grammar
The number grammar recognizes whole numeric numbers (the caller must not speak the individual digits).
Examples
| Caller says | MEANING key |
|---|---|
| 123（ひゃくにじゅうさん） | 123 |
| 25（にじゅうご） | 25 |
| 12345（いちまんにせんさんびゃくよんじゅうご） | 12345 |
| -4（まいなすよん） | -4 |
| 14.56（じゅうよんてんごろく） | 14.56 |
phone built-in grammar
The phone grammar collects telephone numbers (landline and mobile) including 3-digit, 10-digit, and 11-digit numbers.
The caller must speak each digit one at a time (no natural numbers).
Examples
The following examples show various telephone numbers spoken by callers (written in Kanji/Kana) with the resulting values placed into the MEANING key.
| Caller says | MEANING |
|---|---|
| 117（いちいちなな） | 117 |
| 177（いちなななな） | 177 |
| 932（きゅうさんに） | 932 |
| 0512345678(ぜろごいちにさんよんごろくななはち) | 0512345678 |
| 0742-03-1234（ぜろななよんにのぜろさんのいちにさんよん） | 0742031234 |
| 05-1234-5678内線155（ぜろごのいちにさんよんのごろくななはちないせんいちご ご） | 0512345678x155 |
| 0336677811内線1（ぜろさんさんろくろくななななはちいちいちないせんいち） | 0336677811x1 |
| 0336677811内線24（ぜろさんさんろくろくななななはちいちいちないせんによん） | 0336677811x24 |
| 0336677811内線番号107（ぜろさんさんろくろくななななはちいちいちないせんばん ごういちれいなな） | 0336677811x107 |
| 0336677811内線1074（ぜろさんさんろくろくななななはちいちいちないせんいちぜ ろななよん） | 0336677811x1074 |
Here are examples of utterances that are not handled by the grammar:
| Caller says | Reason for not being recognized |
|---|---|
| 011（ぜろいちいち） | 3-digit numbers must not begin with zero. |
| 0012345678（ぜろぜろいちにさんよんごろくななはち） | No numbers can begin with double zero. |
| 022の743の999（ぜろににのななよんさんのきゅうきゅうきゅう） | 9-digit numbers are not allowed. |
| 0229の63の6825（ぜろににきゅうのろくじゅうさんのろくはちにご） | Natural numbers are not allowed. |
| 022の743の456253（ぜろににのななよんさんのよんごろくにごさん） | 12-digit numbers are not allowed. |
| 177内線1番（いちななななないせんいちばん） | 3-digit numbers (177x1) do not have extensions. |
| 03366778111234（ぜろさんさんろくろくななななはちいちいちいちにさんし） | The caller must say “ないせん” or “ないせん バンゴウ” before the extension. |
| 0336677811内線番号12345（ぜろさんさんろくろくななななはちいちいちないせん ばんごういちにさんよんご） | The maximum length of an extension is 4 digits. |
| 0336677811内線番号12（ぜろさんさんろくろくななななはちいちいちないせんばんご うじゅうに） | Natural numbers are not allowed in extensions. |
postcode built-in grammar
The postcode grammar collects Japanese 3-digit and 7-digit postal codes.
There is almost no validity checking on the postal code; most any sequence of 3 and 7 digits is allowed (see example below for a disallowed sequence).
Examples
The following examples show postal codes spoken by callers (written in Kanji/Kana) with the resulting values placed into the MEANING key.
| Caller says | MEANING |
|---|---|
| 1670021（いちろくななれいれいにいち） | 1670021 |
| 905（きゅうぜろご） | 905 |
| 532の0000（ごさんにのぜろぜろぜろぜろ） | 5320000 |
Here is an example of an utterance not handled by the grammar:
| Caller says | Reason for not being recognized |
|---|---|
| 000の9999（ぜろぜろぜろのきゅうきゅうきゅうきゅう） | Postal codes cannot begin with triple zero (for example, 0009999 is not allowed). |
time built-in grammar
The time grammar recognizes a time of day. Recognized phrases include:
| Times spoken in… | Example |
|---|---|
| 12-hour format | ごぜんごじ |
| 24-hour format | にじゅうさんじじゅうごふん |
| Qualified times | ごじごろ |
Examples
All keys are set whenever known. For example, the phrase 6時59分（ろくじごじゅうきゅうふん） sets the following keys:
| MEANING | 0659? |
|---|---|
| HOUR | 06 |
| MINUTE | 59 |
| AMPM | ? |
| QUALIFIER | exact |
The following examples show various clock times spoken by callers (written in Kanji/Kana) with the resulting values placed into the MEANING and QUALIFIER keys. (Not shown are the values of the HOUR, MINUTE, and AMPM keys.)
| Caller says | MEANING | QUALIFIER |
|---|---|---|
| 1時15分（いちじじゅうごふん） | 0115? | exact |
| 1時1分（いちじいっぷん） | 0101? | exact |
| 1時15分頃（いちじじゅうごふんごろ） | 0115? | approx |
| およそ1時15分（およそいちじじゅうごふん） | 0115? | approx |
| 午後1時15分（ごごいちじじゅうごふん） | 0115p | exact |
| 午前1時15分（ごぜんいちじじゅうごふん） | 0115a | exact |
| 朝6時（あさろくじ） | 0600a | exact |
| 午後6時（ごごろくじ） | 0600p | exact |
| 夕方6時（ゆうがたろくじ） | 0600p | exact |
| 夜の6時（よるのろくじ） | 0600p | exact |
| 夜12時半（よるじゅうにじはん） | 1230a | exact |
| 昼12時15分前（ひるじゅうにじじゅうごふんまえ） | 1145a | exact |
| 24時（にじゅうよじ） | 2400h | exact |
| 23時15分（にじゅうさんじじゅうごふん） | 2315h | exact |
| 正午（しょうご） | 1200p | exact |
| 昼12時（ひるじゅうにじ） | 1200p | exact |
| 昼の12時（ひるのじゅうにじ） | 1200p | exact |
| 12時5分過ぎ（じゅうにじごふんすぎ） | 1205? | exact |
| 昼の12時5分過ぎ（ひるのじゅうにじごふんすぎ） | 1205p | exact |
| 深夜（しんや） | 1200a | exact |
| 昼の12時5分前（ひるのじゅうにじごふんまえ） | 1155a | exact |
Here are examples of utterances that are not handled by the grammar:
| Caller says | Reason for not being recognized |
|---|---|
| 1時04分（いちじれいよんぷん） | The grammar does not support digit sequences for the minute field. |
| 1時62分（いちじろくじゅうにふん） | 62 minutes is not valid. |
| 0800時（ぜろはっぴゃくじ） | The grammar does not support military-style numbers. |
| 5時前（ごじまえ） | Because of the ambiguity in Japanese between “10 to 5” and “before 10 past five” ( 5時10分前（ごじじっぷんまえ）), the grammar does not allow the phrases “before 5” or “after 5.” |
Vocabulary items and pronunciations
This chapter describes considerations for vocabularies and their pronunciations in Japanese (ja-JP).
Japanese pronunciations
This section provides detailed reference information to help create pronunciation dictionaries. It is intended for people who have sufficient knowledge of the Japanese language. It provides information about transcription and pronunciation.
The Japanese phoneme system
The Japanese phoneme system can be divided into two groups:
- Consonants
- Vowels
Furthermore, it is possible to distinguish six different types of Japanese consonants:
- Plosives
- Fricatives
- Affricates
- Nasals
- Liquids
- Glides
The Japanese spelling is very regular. This means the relationship between spelling (grapheme) and sound (phoneme) is easy to define since the orthography is very regular and correlates very well with pronunciation. Nevertheless, there are some pronunciation variants, occurring in phonetic contexts different from one another. They will be explained further on in Multiple pronunciations (variants).
Except for glides, consonants have palatalized counterparts. Palatalized consonants are always represented in SAMPA with /y/ for the secondary articulation.
Within the vowel group, a further distinction can be made between long and short vowels.
Japanese symbol set grouped by phoneme classes
The following table shows all phonemes used in Japanese transcriptions. They are listed according to their phoneme classes with their SAMPA, IPA, and various Japanese representations.
| Phoneme class | SAMPA | IPA | Examples of usage Kana/Kanji - Hiragana - Katakana - Pronunciation |
|---|---|---|---|
| Consonants | Plosives | b | b |
| by | bʲ | 病気 | びょうき |
| d | d | 電話 | でんわ |
| dy | dʲ | ||
| g | g | 額 | がく |
| gy | gʲ | 逆 | ぎゃく |
| k | k | 書く | かく |
| kk | さっき | ||
| ky | kʲ | 客 | きゃく |
| p | p | ||
| pp | 河童 | かっぱ | |
| py | pʲ | ||
| t | t | てこ | |
| tt | 蹴った | けった | |
| Fricatives | f | ɸ | 富士 |
| h | h/ç | 履く | はく |
| hy | hʲ | 百 | ひゃく |
| s | s | 総理 | そうり |
| S | ʃ | 勝利 | しょうり |
| ss | あっさり | ||
| SS | 必死 | ひっし | |
| z | z/dz | 草履 樹 | ぞうり じゅ |
| Affricates | ts | ts | 通気 |
| tS | ʧ | 注記 | ちゅうき |
| dZ | ʤ | 銃器 | じゅうき |
| Nasals | m | m | 幕 |
| my | mʲ | 脈 | みゃく |
| n | n | 猫 | ねこ |
| ny | nʲ | 如意 | にょい |
| N | m/n/ŋ | ||
| Liquids | 4 | ɾ/ɻ | 六 |
| 4y | ɾʲ/ɻʲ | 旅行 | りょこう |
| Glides | y | j | 夢 |
| w | w | 綿 | わた |
| Vowels | Single Vowels | a | a |
| aa | a: | ||
| E | e | 干支 | えと |
| EE | e: | 英語 | ええと えいご |
| I | i | 石 | いし |
| II | i: | いいえ | |
| M | ɯ | 筋 | すじ |
| MM | ɯ: | 数字 | すうじ |
| O | o | 沖 | おき |
| OO | o: | 大きい | おおきい |
Japanese consonants
The standard Japanese consonant system is considered to have:
- Fourteen plosives (including a glottal coda)
- Eight fricatives
- Three affricates
- Five nasals
- Two liquids
- Two glides
Plosives
There are six voiced and eight voiceless plosives in Japanese, which can be arranged in pairs as shown here, according to voicing and palatalization.
Unpalatalized
| Voiced | Voiceless | |
|---|---|---|
| Kana/ Kanji | Hiragana | Katakana |
| /b/ | 帽子 | ぼうし |
| /d/ | 電話 | でんわ |
| /g/ | 額 | がく |
| /pp/ | 河童 | かっぱ |
| /tt/ | 蹴った | けった |
| /kk/ | さっき |
Palatalized
| Voiced | Voiceless | |
|---|---|---|
| Kana/ Kanji | Hiragana | Katakana |
| /by/ | 病気 | びょうき |
| /dy/ | ||
| /gy/ | 逆 | ぎゃく |
Note that the phoneme /dy/ only occurs in foreign words, and has no voiceless counterpart.
Geminates
Japanese has double consonants, called geminates. The first part of double consonants is defined as a glottal coda in the Japanese SAMPA transcriptions, and is uniformly represented “small <tsu>” in the language’s orthography. The geminate consonants have a distinctive function to contrast with the lengths of single consonants. For example, words with geminates and without geminates make differences of meanings.
| Pronunciation | Meaning | Kana/Kanji | Hiragana | Katakana |
|---|---|---|---|---|
| kakki /kakkI/ | vigor | 活気 | かっき | カッキ |
| kaki /kakI/ | oyster | 柿 | かき | カキ |
Therefore, in Japanese, the glottal coda is phonemically independent. However, its sound is dependent on the following consonant.
Note that the glottal coda cannot appear in the word-initial position.
Fricatives
There are eight fricatives in Japanese, one voiced and seven voiceless:
| Voiced | Voiceless |
|---|---|
| Kana/ Kanji | Hiragana |
| /z/ | 草履 樹 |
Affricates
In Japanese there are three affricates, /ts/, /tS/, and /dZ/. More precisely, the voiced dental affricate /dz/ appears before the vowel /M/, while the fricative /z/ never appears in that environment. This means that /dz/ is an allophone of the phoneme /z/ in a complementary distribution. In the Japanese Symbol set no distinction is made between /dz/ and /z/ in order to reduce the number of phonemes.
| Voiced | Voiceless |
|---|---|
| Kana/Kanji | Hiragana |
| /ts/ | 通気 |
| /tS/ | 注記 |
Nasals
There are five nasals in Japanese, /m/, /my/, /n/, /ny/, and /N/.
| Kana/Kanji | Hiragana | Katakana | Pronunciation | |
|---|---|---|---|---|
| /m/ | 幕 | まく | マク | /makM/ |
| /my/ | 脈 | みゃく | ミャク | /myakM/ |
| /n/ | 猫 | ねこ | ネコ | /nEkO/ |
| /ny/ | 如意 | にょい | ニョイ | /nyOI/ |
| /N/ | パン | /paN/ |
Note that the phoneme /N/ is a nasal with place features underspecified. It assimilates in place of articulation to the following consonant. For example:
| Romaji | SAMPA | IPA | Kana/Kanji | Hiragana | Katakana |
|---|---|---|---|---|---|
| tongbo | /tONbO/ | tombo | 蜻蛉 | とんぼ | トンボ |
| kingsa | /kINsa/ | kinsa | 僅差 | きんさ | キンサ |
| gangko | /gaNkO/ | gaŋko | 頑固 | がんこ | ガンコ |
The placeless /N/ never appears in the syllable-initial position.
Liquids
There are two liquids in Japanese, /4/ and /4y/. They are pronounced with the tongue-tip flapping the post-alveolar region.
| Kana/Kanji | Hiragana | Katakana | Pronunciation | |
|---|---|---|---|---|
| /4/ | 六 | ろく | ロク | /4OkM/ |
| /4y/ | 旅行 | りょこう | リョコウ | /4yOkOO/ |
Glides
There are two glides in Japanese, /y/ and /w/.
| Kana/Kanji | Hiragana | Katakana | Pronunciation | |
|---|---|---|---|---|
| /y/ | 夢 | ゆめ | ユメ | /yMmE/ |
| /w/ | 綿 | わた | ワタ | /wata/ |
Japanese vowels
Long and short vowels
Japanese vowels can be divided into two groups according to their length of articulation: short and long. There is a clear distinction between short and long vowels in Japanese, the long vowels being approximately twice as long as the short vowels. All the short vowels have long vocalic counterparts, which are represented with a double of the vowels.
There are five short and five long vowels in Japanese:
| Short | Long | ||
|---|---|---|---|
| Kana/Kanji | Hiragana | Katakana | Pron |
| /I/ | 石 | いし | イシ |
| /E/ | 干支 | えと | エト |
| /a/ | 後 | あと | アト |
| /O/ | 沖 | おき | オキ |
| /M/ | 筋 | すじ | スジ |
Note that the Japanese high back vowel /M/ is unrounded.
Specific pronunciation transcription methods
Transcription of the nasal coda /N/
The Japanese nasal coda, whether it undergoes assimilation, is always represented by the Japanese SAMPA symbol /N/:
| Romaji | SAMPA | IPA | Kana/Kanji | Hiragana | Katakana | |
|---|---|---|---|---|---|---|
| /Nb/ | tongbo | /tONbO/ | tombo | 蜻蛉 | とんぼ | トンボ |
| /Ns/ | kingsa | /kINsa/ | kinsa | 僅差 | きんさ | キンサ |
| /Nk/ | gangko | /gaNkO/ | gaŋko | 頑固 | がんこ | ガンコ |
Transcription of geminates
Only plosive and fricative geminates are pronounced in Japanese. Gemination arises mostly from verb inflection, Sino-Japanese compounds, and loanwords. For example:
| Romaji | Pronunciation | Kana/Kanji | Hiragana | Katakana | |
|---|---|---|---|---|---|
| utta | /Mtta/ | 打った | うった | ウッタ | verb inflection of the past tensed form |
| gakkou | /gakkOO/ | 学校 | がっこう | ガッコウ | Sino-Japanese compound noun |
| ippong | /IppON/ | 一本 | いっぽん | イッポン | Sino-Japanese quantification |
| baggu | /baggM/ | バッグ | loanword |
The two exceptions are for geminate /s/ and /S/. For these:
| Geminate | Non-geminate |
|---|---|
| Pron | Meaning |
| /kaSSa/ | pulley |
| /assa4I/ | bland |
Pronunciation of foreign words
To transcribe foreign words, you must use the Japanese SAMPA symbols. Every language has a different phoneme inventory, so you may have problems in covering each and every sound. You can simply choose the closest Japanese SAMPA symbol. For example, the English ‘{’ can be transcribed as /E4/.
Note that for a true multilingual application, words from multiple language packs (e.g., ja-JP and en-US) can be activated at the same time (in the same grammar, or in parallel grammars.
Vowel insertion
Japanese has strict properties of the Consonant-Vowel syllable structure. When loanwords do not fit its template, vowel insertion occurs for adjustment. For example, the alveolar plosives in coda undergo insertion of /O/, and others mostly undergo insertion of /M/:
| coat | /kOOtO/ |
|---|---|
| friend | /fM4ENdO/ |
| school | /sMkMM4M/ |
| milk | /mI4MkM/ |
Gemination
Monosyllabic loanwords with lax vowels undergo gemination of the following consonants. For example:
| bag | /baggM/ |
|---|---|
| kit | /kIttO/ |
Multiple pronunciations (variants)
The type of pronunciation used in SAMPA and the Japanese system dictionary conforms to the standard non-regional Japanese pronunciation. Since it is possible to have more than one pronunciation for a word by using pronunciation variants, it may be difficult to determine how many pronunciation variants should be created.
The general rule is: Variants should only be created if the pronunciation differs in more than one phoneme. Minor systematic variants such as the devoicing vowels or the nasalized velar plosive, can usually be reflected in the training material for the phonemes, and need not be covered by pronunciation variants. If such a word causes recognition errors, the creation of pronunciation variants may help to solve the problem.
Below, both major and minor pronunciation variants of the Japanese language are listed.
Particles
There are two particles, which are pronounced in a different way from the conventional reading of the Japanese syllabary.
| Romaji | SAMPA | Meaning | Kana/Kanji | Hiragana | Katakana |
|---|---|---|---|---|---|
| ha | /wa/ | topic marker | は | ハ | |
| he | /E/ | directional marker | へ | ヘ |
Devoicing of the short high vowels /I/ and /M/
Devoicing of the unaccented short high vowels occurs in the voiceless environments: between voiceless consonants, or in the word-final positions following voiceless consonants as in susuki or desu . However, there is no need to account for the devoicing in your transcriptions because this is done automatically inside the recognizer’s acoustic models.
Nasalization of the voiced velar plosive /g/
The unaccented voiced velar plosive is nasalized word-internally, as in kagami, where the internal voiced velar plosive is pronounced as a nasalized sound. (The use of nasalization is declining in contemporary speech.) There is no need to account for nasalization in your transcriptions because this is done automatically inside the recognizer’s acoustic models. Thus, the transcription for kagami:
| Romaji | SAMPA | Kana/Kanji | Hiragana | Katakana |
|---|---|---|---|---|
| kagami | /kagamI/ | 鏡 | かがみ | カガミ |
The Japanese symbol set in alphabetical order:
| SAMPA | IPA | Examples of usage |
|---|---|---|
| Kana/Kanji | Hiragana | Katakana |
| 4 | ɾ/ɻ | 六 |
| 4y | ɾʲ/ɻʲ | 旅行 |
| a | a | 後 |
| aa | a: | |
| b | b | 帽子 |
| by | bʲ | 病気 |
| d | d | 電話 |
| dy | dʲ | |
| dZ | ʤ | 銃器 |
| E | e | 干支 |
| EE | e: | 英語 |
| f | ɸ | 富士 |
| g | g | 額 |
| gy | gʲ | 逆 |
| h | h/ç | 履く |
| hy | hʲ | 百 |
| I | i | 石 |
| II | i: | |
| k | k | 書く |
| kk | ||
| ky | kʲ | 客 |
| m | m | 幕 |
| M | ɯ | 筋 |
| MM | ɯ: | 数字 |
| my | mʲ | 脈 |
| n | n | 猫 |
| N | m/n/ŋ | |
| ny | nʲ | 如意 |
| O | o | 沖 |
| OO | o: | 大きい |
| p | p | |
| pp | 河童 | |
| py | pʲ | |
| s | s | 総理 |
| S | ʃ | 勝利 |
| ss | ||
| SS | 必死 | |
| t | t | |
| ts | ts | 通気 |
| tS | ʧ | 注記 |
| tt | 蹴った | |
| w | w | 綿 |
| y | j | 夢 |
| z | z/dz | 草履 樹 |
Automatic pronunciation module
The automatic pronunciation module is provided to pronounce words that are not in any dictionary and composed of hiragana, katakana and kanji characters.
The automatic pronunciation module supports all hiragana and katakana but the support of kanji is limited to the following set of characters:
- 1945 characters of Jōyō kanji which are kanji characters announced officially by the Japanese Ministry of Education
- 805 out of 983 characters of Jinmeiyō kanji which are kanji characters used for idioms and names
- 379 characters of Hyōgai kanji which are kanji characters outside the two major lists above
A complete list of supported characters can be provided upon request.
The parameter swirec_max_auto_prons controls the maximum number of automatically generated pronunciations per word. By default this parameter is set to 1 and increase it allows for more variant pronunciations to be generated.
Feedback
Was this page helpful?
Glad to hear it! Please tell us how we can improve.
Sorry to hear that. Please tell us how we can improve.