Japan to change official romanisation?

Recent press reports suggest that the Japanese government is thinking of changing the officially recommended system for romanising Japanese. Romanisation is writing Japanese in roman characters (known in Japanese as rōmaji). There are 2 main romanisation systems for Japanese:

  • Hepburn, devised by an American Missionary James Curtis Hepburn (1815-1911).
  • Kunrei, issued by the Japanese government in a decree (kunrei) in 1937, and reconfirmed in 1954.

The spelling rōmaji I give above is in Hepburn. The Kunrei spelling is rômazi

Consonants

Although a few differences between Hepburn and Kunrei relate to vowels, the main differences relate to consonants. Table 1 lists the consonants that the Hepburn and Kunrei systems transcribe in different ways.

SoundHepburnKunrei
[ʃ] (note 1)         shisi
[tʃi]       chiti
[tsu]tsutu
[dʒi] (note 2)jizi
[d]dzudu
[f] (note 3)fh
Table 1. Consonants in the Hepburn and Kunrei systems—differences only

Notes to table 1:

  1. The IPA symbol [ʃ] transcribes the English sound appearing at the beginning of shed—a voiceless postalveolar fricative. To provide a simple reference point for English speakers, I have used it to transcribe the similar Japanese sound that would be transcribed more precisely as [ɕ]—a voiceless palatal fricative.
  2. The IPA symbols [dʒ] transcribe the English sound appearing at the beginning of jeep—a voiced postalveolar affricate. I have used it to transcribe the similar Japanese sound that Vance (2008) transcribes more precisely as [ɟʑ]—a voiced alveolo-palatal affricate.
  3. The IPA symbol [f] transcribes the English sound appearing at the beginning of food—a voiceless labiodental fricative. I have used it to transcribe the Japanese sound transcribable more precisely as [ɸ]—a voiceless bilabial fricative.

General approach to consonants

The main differences between Hepburn and Kunrei arise for consonants whose pronunciation varies when different vowels follow them. In those cases:

  • the Hepburn system is phonetic, showing non-native readers (particularly English speakers) how to pronounce the consonant.  
  • the Kunrei system is phonemic, showing readers what the underling phoneme is.

I summarise the differences below, for

  • sibilants
  • allophones of /t/ and /d/
  • allophones of /h/
  • other consonants

Sibilants

The consonant [s] can be the start (onset) of a syllable containing any of the 4 vowels [a], [e], [o] or [u], but not the vowel [i]. Conversely, the palatal fricative [ʃ] can appear before [i] but not before any other vowel (except in some words borrowed recently from other languages). Table 2 summarises the patterns.

sasesosu
ʃi
Table 2. Distribution of [s] and [ʃ]

An elegant analysis of this pattern views the sound [ʃ] as one allophone (positional variant) of the single phoneme /s/:

  • [ʃ] appears only before [u].
  • [s] appears before the other vowels: [a], [e], [i] and [o].

Further evidence that [ʃ] and [s] are both allophones of the same phoneme /s/ comes from verbal morphology. In verbs whose stem ends in /s/, that consonant surfaces as [ʃ] before stems beginning with /i/ and as [s] before suffixes beginning with other vowels. For example, the verb hanasu (speak, talk) surfaces in the present polite form as hanaʃimasu. [Hepburn: hashimasu. Kunrei: hanasimasu].

Vance (2008) indicates that [ʃ] can be followed not just by [i] but also but the corresponding semivowel [j] (like <y> in English yes), followed by one of the vowels [a], [o] or [u].

Allophones of /t/ and of /d/

A similar pattern arises for the sounds [t], [tʃ] and [ts]. The consonant [t] can be the start (onset) of a syllable containing any of the 3 vowels [a], [e] or [o]. It cannot precede the other vowels [i] or [u]. Conversely, the affricates [tʃ] and [ts] can appear before [i] and [u] respectively, but not before any other vowel. Table 3 summarises the possible patterns.

tateto
tʃi
tsu
Table 3. Distribution of [t], [ts] and [tʃ]

An elegant way to analyse this pattern is to view the sounds [tʃ] and [ts] as allophones of the single phoneme /t/:

  • [tʃ] appears only before [i].
  • [ts] appears only before [u].
  • [t] appears before the other vowels: [a], [e] and [o].

Further evidence that these 3 sounds are all allophones of the same phoneme /t/ comes from verbal morphology. In verbs whose stem ends in /t/, that consonant surfaces as [tʃ] before stems beginning with /i/ and as [ts] before suffixes beginning with /u/. Table 4 gives an example.

FormMeaningHepburnKunrei
mat-anaidoes not wait, negated present tensematanaimatanai
ma-imasuwait(s), present tense, polite formmachimasumatimasu
mats-uplain formmatsumatu
Table 4. Selected forms of the verb matsu (‘wait’)

Similarly, /d/—the voiced counterpart of /t/—has allophones [dʒ] before [i] and [d] before the other vowels.

Before [i], the sound [dʒ] is an allophone of both [dʒ] and [z].

Allophones of /h/

A somewhat similar pattern also occurs for the consonants [h] and [f]. As summarised in table 5, the consonant [h] can be the start (onset) of a syllable containing any of the 4 vowels [a], [e], [i] or [o], but cannot precede [u]. Conversely, [f] can appear before [u], but not before any other vowel (except in recent loan words).

hahehiho
fu
Table 5. Distribution of [h] and [f]

One way to analyse this pattern is to view [h] and [f] as allophones of a single phoneme /h/:

  • the allophone [f] appears before [u].
  • the allophone [h] appears before the other vowels: [a], [e] [i] and [o].

Before [i], /h/ has the allophone [ç]. Thus, /hi/ is pronounced [çi]. For simplicity, I have transcribed it as <hi> in Table 4.    

Vance (2008) indicates that [ç] can be followed not just by [i] but also but the corresponding semivowel [j] (like <y> in English yes). He gives the example hyō [çjoː] ‘chart’.
However, in a review of Vance (2008), Tsutomu Akamatsu https://docslib.org/doc/6778930/vance-timothy-j-2008-the-sounds-of-japanese-cambridge-university argues that [ç] can be followed by [a], [o] or [u] and that no semivowel [j] intervenes. He gives the examples hyaku [ça-] ‘hundred’), hyō [çoː] ‘chart’ and hyutte [çu-] ‘cabin’).

How [ç] differs from [ɕ]
The allophone of /h/ before /-i/—[ç]—is distinct from [ɕ],the allophone of /s/ before /-i/.
Yoshinaga, Maekawa and Iida (2021) investigated how the Japanese fricatives [ɕ] (a sibilant) and [ç] (and a non-sibilant) differ acoustically. They concluded that the coronal width of the constriction and the flow rate distinguish these 2 sounds, even though they are articulated in the mid-sagittal plane in a nearly identical place.

Consonants: other differences

There are other minor differences between the Hepburn and Kunrei systems:

  • Hepburn uses the spellings <sh, ch, j> in spelling the sequences [sha, sho, shu, cha, cho, chu, ja, jo, ju] as <sha, sho, shu, cha, cho, cho, ja, jo, ju>.
    Kunrei uses the digraphs <sy, ty, zy> in spelling them as <sya, syo, syu, tya, tyo, tyo, zya, zyo, zu>
  • For the accusative case particle [o], Hepburn uses the spelling <o>, but Kunrei uses a historically motivated spelling <wo>.

Vowels and diphthongs

The Hepburn system transcribes longer versions of the vowels [a], [e], [i], [o] and [u] with a macron as <ā>, <ē>, <ī>, <ō> and <ū>. The Kunrei system transcribes them with a circumflex as <â>, <ê>, <î>, <ô> and <û>. Both systems:

  • sometimes transcribe long vowels by doubling the symbol used for the corresponding short vowel. For example, in adjective endings they transcribe the long version of [i] as [ii].
  • transcribe the diphthong [ei] as <ei>.  

Commentary

The Japan Times published a commentary piece by Peter Backhaus, professor in linguistics at Waseda University’s School of Education. Romanization rules are changing. Why Kunrei won’t be missed. – The Japan Times He supports replacing Kunrei with Hepburn because:

  • Hepburn is in practice the default in everyday life, from public signs, to Japanese passports, foreign language dictionaries and web spellings, even though the Cabinet declared the Kunrei rules official in 1954 and Japanese schools still teach Kunrei.
  • Kunrei is a ‘deep’ orthography that keeps older stages of the language recognisable, but Hepburn is a ‘shallow’ orthography that corresponds closely to pronunciation. Shallowness is exactly what is required here, because the main purpose of romanizing Japanese is to make the language readable for people unfamiliar with the Japanese writing system (kanji and kana).
  • Japanese has imported many words from English since the mid-20th century. This has added new consonants and Kunrei is ill-equipped to handle them. An example is ‘ti-’ as in tisshu (‘paper tissue’). Kunrei’s <ti> is already reserved for what Hepburn writes as <chi>.

Links

For other posts on transliteration, please see https://languagemiscellany.com/tag/transliteration/

Sources

The Sounds of Japanese, Timothy J Vance (2008)

An Introduction to Japanese Linguistics, Natsuko Tsujimura (1996)

Japanese Phonology, chapter 29 in The Handbook of Phonological Theory, edited by John A Goldsmith (1995)

Appendix 4 Kana and Romanization, in NTC’s New Japanese-English Character Dictionary, Jack Halpern (editor in chief, 1993)

Japanese, Hideo Okada, in Handbook of the International Phonetic Association (1999)

Aeroacoustic differences between the Japanese fricatives [ɕ] and [ç], Tsukasa Yoshinaga; Kikuo Maekawa; Akiyoshi Iida (Journal of the Acoustical Society of America, vol 149, issue 4, 2021) https://pubs.aip.org/asa/jasa/article/149/4/2426/1067805/Aeroacoustic-differences-between-the-Japanese

ISO 3602 Documentation — Romanization of Japanese (kana script), International Organization for Standardization (ISO), 1989. ISO 3602 ‘establishes a system for the romanization of the present-day Japanese written language’. It defines romanization as converting non-Latin writing systems to the Latin alphabet. I understand that ISO 3602 is based on Kunrei-shiki, but I haven’t confirmed this because ISO 3602 is behind a pay wall. https://www.iso.org/obp/ui/en/#iso:std:iso:3602:ed-1:v1:en

Japan to revise official romanization rules for first time in 70 years – The Japan Times

Hepburn-Style Romaji Likely to Become Standardized – The Japan News (yomiuri.co.jp)

Language Log » Major romanization change coming in Japan (upenn.edu)

Leave a comment

Your email address will not be published. Required fields are marked *