11.4 Mongolian Mongolian: U+1800 U+18AF (additional material) Shaping rules

Transcription

1 11.4 Mongolian Mongolian: U+1800 U+18AF (additional material) Shaping rules Mongolian letters frequently take one of several forms depending on the surrounding context within the word. The following rules define the normal shaping behaviour. Some letters (mostly Ali Gali) have just one appearance. Certain words have traditional spellings whose appearance differs from the normal one. Foreign words often have an appearance that cannot be predicted from the rules. Alternative shapes are obtained by using a Mongolian Free Variation Selector after the affected letter. Tables specifying the precise alternative shaping behaviour for running text are in the midst of the standardisation process. The shaping rules are described in distinct stages to aid understanding. Implementations are free to combine or split stages as they wish provided an identical end result is obtained. The rules must be applied in turn to each character from the start of the text to the end. Implementations can either apply the first three stages in turn to each individual character or can apply the stages one stage at a time to all the characters. The last two stages must be done after the first three have been applied to all characters. The stages are: Cursive joining. Letters that change shape have one of four basic forms depending on their position within the word: isolated, initial, medial or final. The basic positional form is determined in a similar manner to the Arabic script. Subsequent steps may select a variant form for that position but the basic position is not changed. Gender specific shaping. Mongolian words are masculine or feminine. (Exceptionally some compound and foreign words have mixed gender.) Some letters change their appearance depending on the gender of the word. The Unicode encoding assigns a single code point to these letters, unlike some other encodings. This step determines the appropriate variant for these letters. Contextual shaping. Some letters change their appearance depending on the surrounding letters. Variants are selected depending on the surrounding one or two letters. Mapping to glyphs. Each individual variant maps to a particular glyph. Some glyphs are shared between several letters and variants and this step makes the following stage simpler. Ligature formation. There are many compulsory ligatures, mostly between a rounded consonant and a following vowel. Cursive joining Joining Classes. Most Mongolian letters may be depicted by one of a number of possible contextual glyph forms. The appropriate form is initially determined on the basis of a letter s joining class and the joining class of adjacent characters. Mongolian letters are willing to join with both the preceding and following letters. Each Mongolian character falls into one of the classes shown in Table 11-3 (see file???????.txt for a complete list). Mongolian Free Variation Selectors are not taken into account at this stage and are treated as transparent characters. Table 11-4 defines derived superclasses of the primary Mongolian joining classes; those superclasses are used in the cursive joining rules. In the following tables and text preceding and following refer to the visual order of the characters. In traditional vertical lines of text they are above and below respectively. When the text is embedded within horizontal lines they are left and right respectively.

2 Table Primary Mongolian Joining Classes Joining Class Symbols Members Dual-joining D Most Mongolian letters, TODO LONG VOWEL SIGN Join-causing C ZERO WIDTH JOINER, NIRUGU, SIBE SYLLABLE BOUNDARY MARKER, HYPHEN-MINUS, HYPHEN, some ALI ALI letters... These characters are distinguished from the dual-joining characters in that they do not change shape (or are invisible). Mid-word space M NARROW NO-BREAK SPACE, MONGOLIAN VOWEL SEPARATOR. These characters add white space in the middle of a word. They cause the preceding character to take a final form and "join" with the following letter. Non-joining U ZERO WIDTH NON-JOINER. All digits, all punctuation except HYPHENS and SIBE SYLLABLE BOUNDARY MARKER, all non- Mongolian letters. Transparent T All combining and format marks, including ALI GALI DAGALGA, MONGOLIAN FREE VARIATION SELECTOR ONE, TWO and THREE, WORD JOINER, ZERO WIDTH NO-BREAK SPACE. TODO SOFT HYPHEN. These characters are ignored when determining joining. Table Derived Mongolian Joining Classes Joining Class Preceding-join-causing Following-join-causing Members Superset of Dual-joining, Join-Causing. Superset of Dual-joining, Join-Causing, Mid-word space. Positional forms. The dual-joining letters can take four different positional forms depending on the joining classes of the preceding and following characters. The forms are isolated, initial, medial and final. The join causing characters have a single nominal form as shown in the code charts or are invisible. Table 11-5 describes the notation for the various forms. Notation Table 11-5 Positional Form Notation Description X N Nominal form X L0 Default Isolated form X L1 Isolated form variant 1 X I0 X I1, X I2,... X M0 X M1, X M2,... X F0 X F1, X F2,... Default Initial form. This joins with the following character. Initial form variants 1, 2,... These variants may be selected by certain rules. Default Medial form. This joins with the preceding and following characters. Medial form variants 1, 2,... These variants may be selected by certain rules. Default Final form. This joins with the preceding character. Final form variants 1, 2,... These variants may be selected by certain rules. Missing positional forms. Some letters only occur in certain locations within complete words and do not have certain positional forms, e.g. initial or final forms. The consonants generally do not have isolated forms. When a piece of text is being entered or edited partial words are common and letters

3 may occur in situations for which there is no traditional positional form. Table 11-6 lists recommended fallback forms for the missing positional forms. If the form listed as the first fallback is not available either, then the one listed as second fallback should be used instead. Table Fallbacks for Missing Positional Forms Missing Form 1st Fallback Form 2nd Fallback Form Isolated Initial Final Initial Medial Final Medial Final Initial Final Medial Initial The cursive joining rules are: J1 Transparent characters do not affect the joining behaviour of other characters. J2 A join-causing character X will adopt the form X N. J3 J4 J5 J6 A dual-joining character X that has no following-join-causing character preceding it and a preceding-join-causing character following it will adopt the form X I0. A dual-joining character X that has a following-join-causing character preceding it and a preceding-join-causing character following it will adopt the form X M0. A dual-joining character X that has a following-join-causing character preceding it and no preceding-join-causing character following it will adopt the form X F0. A dual-joining character X that has no following-join-causing character preceding it and no preceding-join-causing character following it will adopt the form X L0. Gender Determination and Shaping The gender of a word is important in determining the shape of some letters. Mongolian words are masculine or feminine. The gender is determined by the vowels in a word, and vowels of opposite gender are normally not mixed in the same word. Some vowels are neuter in gender and can be freely mixed with masculine and feminine vowels. Table 11-7 lists the genders of the vowels. When looking at preceding or following letters certain characters are ignored and skipped over; these are also listed. Gender Class Symbols Members Table Vowel Genders Masculine M A, O, U, TODO O, TODO U, SIBE U, ALI GALI A, ALI GALI HALF U. Feminine F E, OE, UE, EE, TODO E, TODO OE, TODO UE, SIBE E. Neuter N I, TODO LONG VOWEL SIGN, TODO I, SIBE I, SIBE UE, MANCHU I, ALI GALI I. Ignore I ZERO WIDTH JOINER, ZERO WIDTH NON-JOINER, MONGOLIAN FREE VARIATION SELECTOR ONE, TWO and THREE, TODO SOFT HYPHEN, NIRUGU, all combining and format marks, including ALI GALI DAGALGA. Some compound words or foreign words have mixed gender and Manchu and Sibe are not a strict as Mongolian. The following rules are only applied to gender sensitive consonants. The first rule whose conditions are met determines the form of the consonant and subsequent rules are ignored. The letter assumes a variant depending on the gender determined by the rules and the letter s positional form. Table 11-8

4 lists the variants for all the gender sensitive consonants. Table Forms of Gender Sensitive Consonants Consonant Position Masculine Form Feminine Form QA Initial QA IO QA I2 ã QA Medial QA M0 Ì QA M4 ã QA Final QA F0 ì QA F0 ì GA Initial GA IO - GA I2 ã GA Medial GA M0 Ì GA M3 ã / Î GA Final GA F0 ì GA F1 ì TODO QA Initial TODO QA IO TODO QA I1 TODO QA Medial TODO QA M0 TODO QA M1 TODO QA Final TODO QA F0 TODO QA F0 TODO GA Initial TODO GA IO TODO GA I1 TODO GA Medial TODO GA M0 TODO GA M2 TODO GA Final TODO GA F0 TODO GA F1 SIBE KA Initial SIBE KA I0 SIBE KA I? SIBE KA Medial SIBE KA M0 SIBE KA M? SIBE KA Final SIBE KA F0 SIBE KA F? SIBE GA Initial SIBE GA IO SIBE GA I? SIBE GA Medial SIBE GA M0 SIBE GA M? SIBE GA Final SIBE GA F0 SIBE GA F? SIBE HA Initial SIBE HA I0 SIBE HA I? SIBE HA Medial SIBE HA M0 SIBE HA M? SIBE HA Final SIBE HA F0 SIBE HA F? MANCHU GA Initial MANCHU GA IO MANCHU GA I? MANCHU GA Medial MANCHU GA M0 MANCHU GA M? MANCHU GA Final MANCHU GA F0 MANCHU GA F? [The above table is to give the feel of the table - it needs checking. It also needs the glyphs adding.] The gender shaping rules are: G1. If the current letter is in its isolated form then it does not change. G2. If the following character (skipping over ignorables) is a masculine vowel then the current letter takes its masculine form. G3. If the following character (skipping over ignorables) is a feminine or neuter vowel then the current letter takes its feminine form. G4. Search backwards within the current word looking for a masculine or feminine vowel. If one is found the current letter takes the corresponding form. G5. Search forwards within the current word looking for any vowel. If a feminine or neuter vowel is found the current letter takes its feminine form. Otherwise it takes its masculine form. The feminine vowels OE and UE in medial position normally look like the masculine O and E but take a special medial form variant to distinguish themselves if the reader would mistake them because a feminine vowel has not already been seen in the word.

5 G6. If OE or UE is in its medial form, search backwards in the word looking for a feminine or masculine vowel, but noting the presence of any neuter vowel encountered. If a masculine vowel is found, or no vowel of any kind is found then change the form of the current letter to OE M1 or UE M1 respectively. [The rule above spots changes of gender mid-word. A simpler rule that only detects the first syllable situation is...] G6. If OE or UE is in its medial form, search backwards in the word looking for a vowel. If no vowel is found then change the form of the current letter to OE M1 or UE M1 respectively. Contextual Shaping Many Mongolian letters assume a variant form when they are next to certain other letters or classes of letter. In addition the Mongolian Free Variation Selectors may be used in certain circumstances to choose alternative variants the normal rules do not produce in a particular context. The classes used for these rules are listed in Table Table 11-9 Contextual Shaping Classes Context Class Symbols Members Vowel V This is a superset of the gender classes masculine vowel, feminine vowel and neuter vowel. Consonant C Mongolian letters that are not vowels, NARROW NO- BREAK SPACE, MONGOLIAN VOWEL SEPARATOR Ignore I The same as the gender class of the same name Other O All other characters. These delimit a Mongolian word. For the purposes of these rules when considering a particular letter, if a Mongolian Free Variation Selector immediately follows it then the pair is treated as a combination when checking for an appropriate rule. If there is no matching rule for the pair then the Mongolian Free Variation Selector must be ignored and a search made for a rule that matches just the letter. Rules apply to letter in a particular positional form. Some rules only apply to a particular variant of a positional form. (These variants will have been selected by the gender shaping rules above.) The context required for the rules is limited to a maximum of two characters (apart from ignorables) on either side of the character being considered. Some rules consider the specific positional form and sometimes the exact variant of the character preceding the current one. The positional form is not considered for any of the other contextual characters. At most one rule is applied to any letter, the first one that matches the context. [Either an extensive list of rules, or a few typical examples and a statement saying the full rules are available on a file.] Example rule: When medial NA is followed by a vowel it gains a dot. NA M + V NA M1 + V Á + V Æ + V [Does showing the glyphs help or hinder? The glyph shown is also that for medial A, but the rule only applies to NA]

6 Mapping to glyphs Each letter positional form variant has a single glyph associated with it. In many cases several letters variants map to the same glyph. E.g. A M0 is the same as NA M0. This stage reduces the variants to the smaller set of glyphs which simplifies the following ligature formation. BIG TABLE Ligature Formation There are a large number of compulsory ligatures between curved consonants and following vowels. Some fonts may wish to add extra ligatures where certain combinations of letters have shapes that clash. ANOTHER TABLE Line breaking Line breaking within a word is preferred at NARROW NO-BREAK SPACE and MONGOLIAN VOWEL SEPARATOR when present. If a break occurs at one of these points the visible gap is omitted from rendering in the same manner as an inter-word space. [Perhaps the text for NNBS should mention that the no-break aspect is about word break not line break as with other uses of the phrase no-break] Implementation notes Some implementations may prefer to work with glyphs at an earlier stage in the process than the mapping to glyphs step. If this is the case care must be taken to assign additional glyph numbers so that the various letters, forms and variants can be distinguished from each other when obeying the various rules. Failure to do this will result in ambiguities that will give incorrect results. The mapping to glyphs step is still useful to reduce the number of glyphs. This reduces the size of the ligature tables.