6. Scripts, spelling and romanization
Central Okinawan is written in kana, in mixed kanji and kana, and in the Latin alphabet. A spelling has to be read within its convention. In particular, two texts may write the same word differently, and one spelling may leave a contrast in pronunciation unmarked. This chapter uses Central Okinawan readings and treats the spellings of historical texts and other varieties according to their own conventions. (Miyara 2010, pp. 55–84)
Kana and kanji in Okinawan texts
Mixed writing can use kanji for lexical material and kana for particles, endings and portions of words. Kana can also represent a word or an entire passage without kanji. Hiragana commonly supplies the kana in mixed prose. Katakana can supply pronunciations in a dictionary or teaching text, mark a word out typographically, or serve as the main script of a transcription. A shift between hiragana and katakana can therefore be a change of presentation alone.
A kanji spelling identifies a word or part of a word but can leave its pronunciation implicit. 沖縄, for instance, can have a Central Okinawan reading in Okinawan prose and a Japanese reading in Japanese prose. A string of shared characters does not by itself identify the language. In mixed Okinawan writing, the kana portions can also have Okinawan values. Reading the kanji as Okinawan while applying Japanese readings to every kana sequence will give incorrect results in forms such as 手ぃー.
The following constructed entries illustrate three ways of presenting a reading. Romanization in the examples doubles a vowel for length. It records a glottal contrast only where the example identifies one.
| Mixed or kanji form | Hiragana | Katakana | Romanization | Gloss | Translation |
|---|---|---|---|---|---|
| 沖縄 | うちなー | ウチナー | uchinaa | Okinawa | ‘Okinawa’ (constructed) |
| 沖縄口 | うちなーぐち | ウチナーグチ | uchinaaguchi | Okinawa-speech | ‘the Okinawan language’ (constructed) |
| 手ぃー | てぃー | ティー | tii | hand | ‘hand’ (constructed) |
A reading printed beside or above kanji can supply the complete Okinawan form. Keep that reading distinct from the kanji string when citing the text. The reading may cover the whole compound, as it does when うちなーぐち is supplied for 沖縄口. The matching unit is then the compound. Allocating one kana or one mora to each kanji would misrepresent the spelling.
Kana normally represents vowel or consonant–vowel units. A full-size kana and a following small kana can jointly represent one such unit. The independent moraic nasal, the first part of a geminate, and vowel length have their own spelling devices. In the table, each short vowel and its length-marked counterpart is shown in both kana scripts. These are sign values in the convention that uses ー for length.
| Vowel | Hiragana, short | Hiragana, long | Katakana, short | Katakana, long | Romanization |
|---|---|---|---|---|---|
| a | あ | あー | ア | アー | a, aa |
| i | い | いー | イ | イー | i, ii |
| u | う | うー | ウ | ウー | u, uu |
| e | え | えー | エ | エー | e, ee |
| o | お | おー | オ | オー | o, oo |
The independent vowel signs in this table give vowel quality. In a convention that distinguishes initial glottal closure, the onset must be read from the additional notation or from the convention’s assignment of the vowel signs. The table alone supplies neither that assignment nor the pitch of a word. (Miyara 2010, pp. 55–84)
Historical spellings and contemporary conventions
Older Okinawan writing includes kana and mixed writing. A literary text can be encountered in manuscript spelling, in an edition using modern character forms, or in a reading prepared for present-day readers. These versions can differ in the use of kanji, the choice of kana and the notation of length or consonants. When an edition prints both a text and a reading, the spelling in the text is evidence for the written form and the reading is evidence for how the editor interprets it. The two belong on separate lines in a gloss if both are relevant.
The Latin tradition also has historical layers. Chamberlain’s description uses the name Luchuan and represents the language in Latin letters for an English-reading audience. Its letter values belong to that work. To compare one of its printed forms with a later transcription, first identify the sound values each author assigns and then compare the forms at that level. Changing the spelling of a historical transcription to a modern one produces an editorial conversion. (Chamberlain 1895, p. 1)
Reading older and literary spellings
Historical kana values are determined by the date and conventions of the text and by its intended reading. For example, an older kana choice can be retained as a feature of the written tradition after the pronunciation that motivated it has changed. Conversely, a present-day reading may contain a digraph, small kana or length mark that has no counterpart in the spelling of the base text. Even when a modern Central Okinawan form is known, its pronunciation by itself cannot determine which historical spelling an editor will print.
The distinction between spelling and pronunciation also matters for morphology. A kanji written across several forms of a word can keep the lexical element recognizable while the following kana change with inflection. Within a compound, a reading can give the pronunciation of the combined form, including a consonant alternation. The spelling 沖縄口 shows the two lexical elements of うちなーぐち. Its kana reading supplies ぐち guchi in the compound. The compound provides evidence for that reading there. Other compounds require their own readings.
Applying a contemporary convention
A contemporary kana convention makes choices about which contrasts to show and which signs to use. Three consequential choices are whether to distinguish glottal onset, whether to use small kana for combinations absent from ordinary Japanese kana spelling, and how to show long vowels. A spelling may be precise about length and still leave glottal onset implicit. Kana readings supplied in a mixed text can follow the same rules as a kana-only text.
Use the spelling rules of the edition or institution where they apply. In an unconverted quotation, retain a source’s kana size, length notation and use of kanji. In a converted reading, apply the chosen rules consistently to the entire string. The examples here use one practical set of kana choices for illustration. The cited descriptions do not specify a later spelling reform in force across Central Okinawan publications. (Chamberlain 1895, p. 1) (Miyara 2010, pp. 55–84)
Writing glottal and consonantal contrasts
In Central Okinawan, the onset of a vowel-initial word or a word beginning with a glide can distinguish lexical forms. A phonological transcription can mark glottal closure with ʔ. In a less explicit kana spelling the distinction may have to be recovered from the word and its context. The absence of a special kana sign in such a spelling is compatible with either reading. The values of any small kana or other special signs used for the distinction must come from that text’s spelling key. (Miyara 2010, pp. 55–84)
| Kana in a spelling that leaves onset implicit | Contrastive transcription | Gloss | Translation |
|---|---|---|---|
| やー | yaa | house | ‘house’ (constructed) |
| やー | ʔyaa | 2SG | ‘you’, singular (constructed) |
The length is visible in both forms. The difference represented by ʔ is at the beginning of the word, before the glide. Keep that information if it is present in the transcription. The bare kana forms on their own yield the same sequence of characters.
Small っ or ッ needs particular care when a source assigns it more than one function. Before a consonant within a word, the ordinary geminate spelling uses it for the first mora of the long consonant. A source using small tsu as an onset sign can assign it a glottal value in another position. In that case, copy the position of the sign and use the source’s rule to transcribe it. Position and convention together determine the reading.
Consonants represented by digraphs
Several Central Okinawan consonant–vowel combinations can be written with a full-size kana followed by a small vowel kana. The small sign helps specify the vowel of the whole combination. For example, the ぃ in てぃ belongs to the unit ti. In てぃー, the length mark extends i, the vowel of that unit. In a convention using these digraphs, a full-size second kana represents a different written sequence.
The four signs below form the voiced and voiceless pairs for ti, di, tu, and du in the illustrated convention. Dakuten marks the voiced member. Ordinary kana signs with the corresponding vowels are also shown because they can represent different consonants in this same convention.
| Romanization | Hiragana | Katakana | Kana to distinguish in this convention | Romanization of those kana |
|---|---|---|---|---|
| ti | てぃ | ティ | ち / チ | chi |
| di | でぃ | ディ | じ / ジ | ji |
| tu | とぅ | トゥ | つ / ツ | tsu |
| du | どぅ | ドゥ | ず / ズ | zu |
Digraphs with small ゃ, ゅ, ょ and their katakana counterparts also combine with the preceding kana. For example, ちゅ represents chu in the examples here. The small sign does not add a separate yu mora. Similarly, ふぇ represents fe, as in にふぇー nifee. A small ゎ or ヮ after a velar kana can mark a labialized combination. In the examples, くゎ gives kwa and ぐゎ gives gwa. In conventions sharing these values, keeping the small size distinguishes a digraph from a sequence with a full-size second kana. The kana assignments illustrate ways of writing Central Okinawan consonant and vowel combinations. (Miyara 2010, pp. 55–84)
| Form | Romanization | Gloss | Translation |
|---|---|---|---|
| てぃー | tii | hand | ‘hand’ (constructed) |
| ちゅー | chuu | today | ‘today’ (constructed) |
| にふぇーでーびる | nifee deebiru | thanks | ‘thank you’ (constructed) |
| くゎっちー | kwacchii | feast | ‘a feast’ (constructed) |
| わらびぐゎー | warabigwaa | child-DIM | ‘a little child’ (constructed) |
In くゎっちー, くゎ represents the labialized onset and its vowel, っ marks the first part of the geminate, and ちー supplies the remaining consonant and the long vowel. Each sign has a distinct role even though two of the signs are small. The accent and the exact phonetic realization of a consonant require a separate transcription if they are relevant.
Writing vowel length and moraic consonants
Vowel length contributes to the mora count. In a spelling with ー, the mark repeats the vowel of the preceding kana unit. The unit can be a single sign, as in なー, or a digraph, as in てぃー. The same mark is used after either script. Length may also be shown by a vowel kana or left to the reader in a particular older text or transcription. A second vowel kana can represent length or a sequence of vowels depending on the convention and the reading of the word. (Miyara 2010, pp. 55–84)
The spelling of a suffix or of a combination involving a particle can reflect the vowel length of the resulting form. A topic-marked word printed with ー can thus have a long vowel in its written surface form even though the particle is not printed separately. When citing the form, keep the long vowel in the romanization and identify the grammatical contribution in the gloss.
| Form | Romanization | Gloss | Translation | Length in the spelling |
|---|---|---|---|---|
| うちなー | uchinaa | Okinawa | ‘Okinawa’ (constructed) | なー has long a |
| てぃー | tii | hand | ‘hand’ (constructed) | てぃー has long i |
| ちゅー | chuu | today | ‘today’ (constructed) | ちゅー has long u |
| いっぺー | ippee | very | ‘very’ (constructed) | ぺー has long e |
| まっちょーん | macchoon | wait.PROG | ‘is waiting’ (constructed) | ちょー has long o |
In the illustrated convention, ん or ン writes the moraic nasal. It can occur at the end of a word, as in まっちょーん, or before another consonant, as in うちなーんちゅ. A phonological transcription can use N for it. The example romanization uses n, so reading that transcription may require distinguishing a moraic n from n at the beginning of a consonant–vowel unit. A following consonant can affect how the nasal is pronounced without requiring a change of nasal kana. (Miyara 2010, pp. 55–84)
Small っ or ッ, in its geminate function, occupies a mora before the following consonant. Romanization represents the geminate within the consonant spelling. The chapter uses pp in ippee and cch in macchoon. Some romanizations use tch for the latter kind of geminate. The geminate and the following vowel length are independent. In いっぺー the spelling shows both a geminate p and long e.
The following counts apply to the constructed examples under the indicated readings. A hyphen separates moras in the analysis column. It is absent from the ordinary kana spelling.
| Form | Romanization | Mora analysis | Gloss | Translation |
|---|---|---|---|---|
| てぃー | tii | ti-i | hand | ‘hand’ (constructed) |
| いっぺー | ippee | i-p-pe-e | very | ‘very’ (constructed) |
| くゎっちー | kwacchii | kwa-c-chi-i | feast | ‘a feast’ (constructed) |
| うちなーんちゅ | uchinaanchu | u-chi-na-a-N-chu | Okinawa-person | ‘an Okinawan person’ (constructed) |
The analysis makes explicit where counting Unicode characters would give a different result. The two signs in てぃ supply one mora, ー supplies the second, and the two signs of ちゅ in the last example supply one mora. The nasal kana immediately before ちゅ supplies a mora of its own.
Word and particle boundaries in different conventions
Kana and mixed prose can run words and particles together on the page. Spaces may instead be introduced in a teaching text, a romanization, or an interlinear analysis. A space in such material follows an editorial decision about the size of the unit being displayed. A printer can group a noun with its following particle while an interlinear gloss displays the particle independently. Neither choice changes the construction.
The following layouts illustrate the distinction. GEN is genitive, TOP is topic and POL is polite. The spaces in the reading and gloss make the grammatical groupings visible.
| Continuous form | Romanized reading | Gloss | Translation |
|---|---|---|---|
| うちなーぬくとぅば | uchinaa nu kutuba | Okinawa GEN language | ‘the language of Okinawa’ (constructed) |
| 沖縄口 | uchinaaguchi | Okinawa-speech | ‘the Okinawan language’ (constructed) |
| わんねーうちなーんちゅやいびーん | wannee uchinaanchu yaibiin | 1SG.TOP Okinawan.person COP.POL | ‘I am Okinawan’ (constructed) |
In the first form, ぬ nu contributes a grammatical relation. It can be recognized within the continuous kana string and shown as a separate unit in the gloss. The second form contains a lexical compound, and its two kanji units do not call for an inserted particle. In the third form, the topic is expressed in わんねー wannee. The romanized reading preserves the form as written. The gloss recovers the topic function without adding a separate particle to that reading. (Miyara 2010, pp. 55–84)
There are three practical places to put boundary information. Ordinary spelling supplies the spaces or other marks actually printed. A reading can add spaces to make units easier to recognize. An analysis can additionally use a hyphen between morphs or an equals sign at a clitic boundary if it defines those signs. For example, an analysis could display uchinaa=nu where the source has うちなーぬ and a spaced reading has uchinaa nu. Each version can be traced to the same string.
The script changes in mixed writing are useful clues but give an incomplete account of boundaries. A kanji may be followed by kana that belong to the same word, as in 手ぃー. A compound can contain adjacent kanji, and several consecutive kana can contain words and particles. A word’s boundary therefore depends on its lexical and grammatical analysis as well as the spelling.
Spacing around an abbreviation, number, personal name or borrowed expression can also be typographic. Interpreting such spaces as grammatical divisions requires independent evidence from the text’s convention. In a reading intended to explain pronunciation, preserve a digraph, a moraic consonant and a written long vowel within their own word even where the printed form uses a different choice of script.
Romanization and source transcription
Romanization can serve as a reading of a kana or mixed-script form. It can also serve as an independent notation of pronunciation. These uses coincide where the spelling supplies the sound distinction and the romanization carries it over. A source can go further by indicating onset, pitch or phonetic detail that the written word does not show. Chamberlain’s spelling and a later linguistic transcription both use Latin letters, so the sound values have to be established for each source before their forms can be compared. (Chamberlain 1895, p. 1) (Miyara 2010, pp. 55–84)
Reading the romanized examples
The table defines the conventions for the features illustrated in this chapter. The examples use practical spelling unless an explicitly contrastive transcription is given.
| Item | Representation | Condition |
|---|---|---|
| Short vowel | a, i, u, e, o | One vowel letter for a short vowel |
| Long vowel | aa, ii, uu, ee, oo | Two matching letters for a long vowel |
| Glottal onset | ʔ | Included in contrastive transcription when the onset is identified |
| Glide written with y | y | The glide in forms such as yaa, including after ʔ |
| Coronal stops in the illustrated digraphs | ti, di, tu, du | Read the digraph as a unit with the stated vowel |
Consonant spellings in ち, じ, つ, ず | ch, j, ts, z | The corresponding romanized vowels follow in the reading |
| Labialized velar in the illustrated digraphs | kw, gw | The w belongs to the onset of the same unit |
| Moraic nasal | n or N | n in practical readings and N where moraic status is shown explicitly |
| Geminate | Doubled consonant spelling, with cch for geminate ch | Applied to the consonant represented by the following kana unit |
A source using macrons may write a single marked vowel where practical romanization doubles it. Conversion is appropriate when the mark signifies length in that source. A source using IPA can use ː. The following forms are constructed renderings of the same readings. The IPA column gives only broad segmental information.
| Kana form | Practical romanization | Macron rendering | Broad IPA rendering | Gloss | Translation |
|---|---|---|---|---|---|
| てぃー | tii | tī | [tiː] | hand | ‘hand’ (constructed) |
| ちゅー | chuu | chū | [tɕuː] | today | ‘today’ (constructed) |
| いっぺー | ippee | ippē | [ipːeː] | very | ‘very’ (constructed) |
A letter can have different values in different systems. y writes the glide in the practical forms here, whereas [j] represents that glide in IPA. Practical j belongs to the spelling exemplified by じ. Likewise a system using c or č can give it a value represented with ch in this chapter. A transcription’s apostrophe can have a sound value or serve as a separator. To convert one, consult its position and the source’s assignment rather than treating it as interchangeable with ʔ.
A Roman-letter reading of a kanji string needs lexical information. For instance, the mapping from 沖縄口 to uchinaaguchi uses the established reading of the word. Conversely, uchinaaguchi can be written in kana or mixed script. A reading in Latin letters loses that choice unless the original form is recorded alongside it.
Transliteration and representation of sounds
For a reversible transliteration, the distinctions to preserve are the ones in the written form. Small size, script, the voicing mark, and the spelling of a long vowel can all matter, as can the original word divisions. For a phonological transcription, the distinctions to preserve are those established for the spoken form. Thus two kana forms with different script or length notation can have the same phonological transcription, while a single ambiguous kana spelling can have two transcriptions if it represents two different words.
An interlinear citation can include a source form, a separate romanized reading, a word or morpheme gloss and an English translation. If the source’s own Latin transcription is the form being cited, retain its notation on the form line and identify any conversion on the reading line. The pitch interpretation is recoverable only if it is marked or supplied by the linguistic analysis. An accent or diacritic must be interpreted under the transcription key before it is carried over into a length or pitch notation. (Chamberlain 1895, p. 1) (Miyara 2010, pp. 55–84)
Unicode forms and visually similar characters
Digital text can display the same kana through different character sequences. A voiced kana can be stored as one precomposed character or as a base kana followed by combining dakuten. The difference affects exact string comparison and character counts. Kana size, script and the choice of a length mark are other differences, but these can be meaningful parts of the spelling.
Voicing marks and character composition
Dakuten ゙ and handakuten ゚ occur as combining marks. The spacing versions ゛ and ゜ are separate characters, as are the halfwidth marks used with halfwidth katakana. The following table gives all three pairs of encoding forms for these two marks. A combining mark in isolation may be displayed with a dotted circle by the renderer.
| Encoding form | Dakuten | Handakuten | Function in the character sequence |
|---|---|---|---|
| Combining | U+3099 | U+309A | Attaches to the preceding base character |
| Spacing | U+309B ゛ | U+309C ゜ | Occupies its own character position |
| Halfwidth | U+FF9E ゙ | U+FF9F ゚ | Follows a halfwidth kana in a halfwidth spelling |
For example, が can be stored as U+304C or as か U+304B followed by U+3099. Likewise, ぱ can be stored as U+3071 or as は U+306F followed by U+309A. The two encodings of each kana are canonically equivalent. NFC composes these base-and-mark sequences. Where Unicode has no corresponding precomposed kana, a combining sequence can remain after NFC.
Spacing dakuten and a combining dakuten attached to a kana are different encodings. A freestanding mark copied from an edition may need interpretation before it can be treated as the voicing mark of an adjacent kana. A conversion from halfwidth katakana is another operation. For example, NFKC converts ティー to ティー and グ to グ. It leaves ティー in katakana. Rendering it as てぃー requires an additional script conversion.
Size, length and script
Small kana and full-size kana are distinct Unicode characters. Changing the size can change whether the kana joins the preceding sign or is read independently. The distinctions remain in NFC. The same applies to hiragana and katakana counterparts. Converting between them can be useful for building a shared reading, but the original script is part of the written form.
| Signs | Code points | Distinction |
|---|---|---|
ぃ and い | U+3043, U+3044 | Small and full-size hiragana i |
ィ and イ | U+30A3, U+30A4 | Small and full-size katakana i |
っ and つ | U+3063, U+3064 | Small and full-size hiragana tsu |
ッ and ツ | U+30C3, U+30C4 | Small and full-size katakana tsu |
ゎ and わ | U+308E, U+308F | Small and full-size hiragana wa |
ヮ and ワ | U+30EE, U+30EF | Small and full-size katakana wa |
ん and ン | U+3093, U+30F3 | Hiragana and katakana moraic nasal |
ー and ー | U+30FC, U+FF70 | Fullwidth and halfwidth prolonged sound mark |
The prolonged sound mark ー can resemble a dash, a minus sign or the kanji numeral 一 in horizontal type. They are distinct characters. The prolonged sound mark can appear vertically when the layout is vertical, without a change in the encoded character. A dash or a numeral found next to kana requires its own reading from context.
There are also similarities in Latin transcription. The glottal stop letter ʔ is U+0294. A straight apostrophe U+0027, a right single quotation mark U+2019 and a modifier letter apostrophe U+02BC are distinct characters. Use the documented sign of a transcription when preserving it. Mapping several such signs to a common glottal stop is an interpretation of that transcription and requires evidence that they serve that function in the relevant forms. A macron above a Latin vowel may likewise be precomposed, as in ī U+012B, or combining, as in i followed by U+0304. Those two macron encodings are canonically equivalent. A double vowel spelling such as ii requires a separate conversion rule.
For comparison and citation, てぃー, ティー and 手ぃー can all be given the reading tii, ‘hand’ (constructed). The first two differ in script and the third uses kanji for the lexical portion. Unicode normalization alone does not derive the common reading. That requires the kana correspondences and, for the third form, the lexical reading.
How Mova handles scripts and spelling variants
tools/mova-parse assigns sentence and span languages at level 1 through tools/langseg, then uses the assigned language’s model to parse each span at level 2. In Latin script, tools/langseg distinguishes English and French using a Latin table, language lexicons and a character four-gram model. It also handles Hebrew-script varieties and scholarly transliteration of Hebrew or Aramaic. Kana and kanji fall under other, and Central Okinawan is outside the Latin-script coverage described here.
There is no level-2 model for ryu. When the model for an assigned language is missing, tools/mova-parse reports that in the # parse comment. The token rows have _ in LEMMA, UPOS, XPOS, FEATS, HEAD and DEPREL, with Parse=None in MISC. Mova does not tokenize, tag, lemmatize or parse Central Okinawan. It holds no annotated Central Okinawan treebank, and no treebank baseline has been measured. The English trees used by prag and world provide no Central Okinawan analysis.
Squid reads the spelling and script statements in this chapter as claims. Formalizable rules and lists enter its queue. The chapter’s rules, including the treatment of kana combinations, length marks, boundaries and spelling variants, are unimplemented. The first points needed for language support are those required by a tokenizer, a tagger and a Universal Dependencies treebank baseline.
Sources for this chapter
- Chamberlain, Basil Hall. 1895. Essay in Aid of a Grammar and Dictionary of the Luchuan Language. Transactions of the Asiatic Society of Japan 23, supplement. Yokohama. — public domain; not held in the library, to verify
- Shimoji, Michinori, and Thomas Pellard, eds. 2010. An Introduction to Ryukyuan Languages. Tokyo: Research Institute for Languages and Cultures of Asia and Africa, Tokyo University of Foreign Studies. — CC BY-NC-SA; not held in the library, to verify
This page in Markdown → https://md.mova.elamur.ai/free/en/docs/languages/ja/ryu/06-scripts-and-spelling/