Comparison

Hudum

Comparison with Quejingzhabu (2013). Quejingzhabu (2013) introduces MONGOLIAN LETTER NA MEDIAL SEPARATE FORM for the letter n. This presentation form is used as the initial N in Todo n, which is never used in Hudum. Therefore, this written form is not included in the Hudum part of this document.

More importantly, the naming principle for written forms used by Quejingzhabu (2013) differs significantly from this document, for example, naming the initial A of the letter a as the “third medial” form of the letter a. This is because it names the written form based on its position in the Mongolian word, where the front of chachlag and particles are not considered word boundaries. So the reason it says a written form is “medial” is because it is not preceded or followed by a word boundary. However, this document defines “medial” as joined both forward and backward.

Comparison with GB/T 25914—2010. This document contains all the written forms of all the letters in GB/T 25914—2010 and can be regarded as an implementation of it.

Comparison with GB/T 25914—2023. This document contains all the written forms of all the letters in GB/T 25914—2023 and can be regarded as an implementation of it.

Comparison with Liang (2019). This document uses the shaping process framework of Liang (2019) and presents the written units, phonetic letters, and logic of the shaping process in the same format. The main difference is that this document introduces positional default forms, fully uses dictionary-based particle shaping logic, and adds shaping rules as required by GB/T 25914—2023; whereas Liang (2019) uses fallback forms instead of default forms in the shaping process, minimizes the use of dictionary-based particle shaping logic, and does not include these additional shaping rules.

Todo

Comparison with Quejingzhabu (2013). Quejingzhabu (2013) introduces MONGOLIAN LETTER NA THIRD MEDIAL FORM for the letter n. This presentation form is used as the final N in Hudum n, which is never used in Todo. Therefore, this written form is not included in the Todo part of this document.

Quejingzhabu (2013) introduces MONGOLIAN LETTER TODO LONG VOWEL SIGN FINAL FORM for the letter lvs. This presentation form consists of a chachlag and a long vowel sign, and is part of the ligatures like b a lvs, p a lvs and so on. Since the behavior of the long vowel sign when added to a ligature is considered similar to a combining mark, the chachlag is not considered to be part of the long vowel sign, and thus this written form is not included.

Comparison with GB/T 36649—2018. This document contains all the written forms of all the letters in GB/T 36649—2018 and can be regarded as an implementation of it.

Sibe

Comparison with Quejingzhabu (2013). Quejingzhabu (2013) introduces MONGOLIAN LETTER YA SECOND INITIAL FORM for the letter y. This presentation form is used as the medial I in Hudum y, which is never used in Sibe. Therefore, this written form is not included in the Sibe part of this document.

Quejingzhabu (2013) introduces MONGOLIAN LETTER SIBE I THIRD FINAL FORM for the letter i. This presentation form is used as the final Iy in Manchu i, which is never used in Sibe. Therefore, this written form is not included in the Sibe part of this document.

Quejingzhabu (2013) introduces MONGOLIAN LETTER SIBE ZA SECOND INITIAL FORM and MONGOLIAN LETTER SIBE ZA SECOND MEDIAL FORM for the letter z. These presentation forms are used in the upper part of Sibe ligature z i, which is never used alone in Sibe. Therefore, this written form is not included in the Sibe part of this document.

The presentation form MONGOLIAN LETTER SIBE ZHA MEDIAL FORM for letter zr in Quejingzhabu (2013) is a J2 form with a ring on the right. However, examining today’s Sibe newspapers, books and dictionaries, this form has changed, so it is not included in the Sibe part of this document.

Comparison with GB/T 36641—2018. GB/T 36641—2018 introduces MONGOLIAN LETTER SIBE ZA SECOND INITIAL FORM and MONGOLIAN LETTER SIBE ZA SECOND MEDIAL FORM for the letter z. These presentation forms are used in the upper part of Sibe ligature z i, which is never used alone in Sibe. Therefore, this written form is not included in the Sibe part of this document.

Manchu

Comparison with Quejingzhabu (2013). Quejingzhabu (2013) introduces MONGOLIAN LETTER SA THIRD FINAL FORM for the letter s. This presentation form is used as the final S3 in Manchu Ali Gali s, which is never used in Manchu. Therefore, this written form is not included in the Manchu part of this document.

Quejingzhabu (2013) introduces MONGOLIAN LETTER SIBE ANG FINAL FORM for the letter ng. This presentation form is used as the final A G3 in Sibe ng, which is never used in Manchu. Therefore, this written form is not included in the Manchu part of this document.

Quejingzhabu (2013) introduces MONGOLIAN LETTER SIBE ZA SECOND INITIAL FORM and MONGOLIAN LETTER SIBE ZA SECOND MEDIAL FORM for the letter z. For the same reasons as in the previous section, these written forms are not included in the Manchu part of this document.

Quejingzhabu (2013) introduces MONGOLIAN LETTER SIBE ZHA MEDIAL FORM for the letter zr. This presentation form is different from the final Jc in Manchu zr, so this written form is not included in the Manchu part of this document.

Quejingzhabu (2013) introduces MONGOLIAN LETTER MANCHU I SECOND MEDIAL FORM for the letter i. This presentation form is used as the medial A I in Sibe i, which does occur in Manchu literature, but is regarded as a non-normative writing variant, so this written form is not included in the Manchu part of this document.

Comparison with GB/T 36645—2018. GB/T 36645—2018 introduces MONGOLIAN LETTER SIBE ZA SECOND INITIAL FORM and MONGOLIAN LETTER SIBE ZA SECOND MEDIAL FORM for the letter z. For the same reasons as in the previous section, these written forms are not included in the Manchu part of this document.

Hudum Ali Gali

Comparison with Quejingzhabu (2013). Quejingzhabu (2013) introduces MONGOLIAN LETTER ALI-GALI A SECOND ISOLATE FORM for the letter a. This presentation form is a stylistic toothed variant of the isolate A, so this written form is not included in the Hudum Ali Gali part of this document.

Quejingzhabu (2013) introduces MONGOLIAN LETTER ALI-GALI A FINAL FORWARD FORM for the corresponding character. This presentation form is used in the Hudum Ali Gali ligature, and is equivalent to the final A of letter a, so this written form is not included in the Hudum Ali Gali part of this document. For the same reason, both MONGOLIAN LETTER ALI-GALI NGA INITIAL SHORT FORM and MONGOLIAN LETTER ALI-GALI NGA MEDIAL SHORT FORM are not included either.

Comparison with GB/T 25914—2010. For the same reasons as in comparison with Quejingzhabu (2013) in this section, the MONGOLIAN LETTER ALI-GALI A SECOND ISOLATE FORM, MONGOLIAN LETTER ALI-GALI A FINAL FORWARD FORM, MONGOLIAN LETTER ALI-GALI NGA INITIAL SHORT FORM, and MONGOLIAN LETTER ALI-GALI NGA MEDIAL SHORT FORM are not included in the Hudum Ali Gali part of this document.

Comparison with GB/T 25914—2023. For the same reasons as in comparison with Quejingzhabu (2013) in this section, the MONGOLIAN LETTER ALI-GALI A SECOND ISOLATE FORM and MONGOLIAN LETTER ALI-GALI A THIRD FINAL FORM are not included in the Hudum Ali Gali part of this document.

Since the shaping step for MONGOLIAN LETTER ALI-GALI A SECOND FINAL FORM is quite clear, this document adds the shaping step for this form.

Todo Ali Gali

Comparison with Quejingzhabu (2013). Quejingzhabu (2013) introduces MONGOLIAN LETTER TODO ALI-GALI I TA. This letter is not used in the Todo Ali Gali, thus not included in the Todo Ali Gali part of this document.

Comparison with GB/T 36649—2018. GB/T 36649—2018 introduces MONGOLIAN LETTER TODO GA SECOND FINAL FORM. This presentation form is used as the initial N of letter n, but since it is analyzed as the letter g, this document analyzes it as the initial Hp of letter g.

GB/T 36649—2018 introduces MONGOLIAN LETTER TODO GA THIRD FINAL FORM. This presentation form is the stylistic variant of the MONGOLIAN LETTER TODO GA FIRST FINAL FORM, and thus is not included in the Todo Ali Gali part of this document.

GB/T 36649—2018 introduces MONGOLIAN LETTER ALI-GALI PA THIRD INITIAL FORM and MONGOLIAN LETTER ALI-GALI PA THIRD MEDIAL FORM. These presentation forms record different phonemes with MONGOLIAN LETTER ALI-GALI PA SECOND INITIAL FORM and MONGOLIAN LETTER ALI-GALI PA SECOND MEDIAL FORM, and thus are analyzed as the variant of p in this document.

GB/T 36649—2018 introduces MONGOLIAN LETTER ALI-GALI HALF U SECOND MEDIAL FORM. This presentation form is not used in the Todo Ali Gali, thus not included in the Todo Ali Gali part of this document.

Manchu Ali Gali

Comparison with Quejingzhabu (2013). Quejingzhabu (2013) does not add the feminine forms for the letters tX and dhX, this document adds the feminine forms for the letters tX and dhX.

Quejingzhabu (2013) introduces MONGOLIAN LETTER ALI-GALI I SECOND FINAL FORM. This presentation form is equivalent to the final Iy of letter i, so this written form is not included in the Manchu Ali Gali part of this document.

Comparison with GB/T 36645—2018. For the same reasons as in comparison with Quejingzhabu (2013) in this section, the MONGOLIAN LETTER ALI-GALI I SECOND FINAL FORM is not included in the Manchu Ali Gali part of this document.

Implementations

This document is the model; a font is one thing that renders it, and a normalizer is another. Both have to answer the same question—which written unit does this text produce—but they are asked it in different circumstances, and reading how each of them answers it is a way to test whether the model is stated clearly enough to be followed.

Where they meet is the shaping process. A font carries the process into the glyph image layer: it draws the written units and writes the substitution and positioning rules that the process prescribes. An engine that has no font to ask still has the same question to answer whenever it compares two texts, and the model answers it without a font, because the written unit a text produces is decided by the characters and the rules alone.

mongfontbuilder. The reference implementation that this repository maintains. It reads the rule data of this document, composes the glyph image layer of a font from a minimal glyph set, and validates the result by shaping the test cases of the national standards with HarfBuzz.

mongol-norm. A shape-aware normalizer, independent of this repository, by Satsrag. It runs the shaping rules of this document without a font, so that two texts which look alike are found to be alike: it maps every encoding of a word—the same word typed with a or e, with o or u, with an FVS or a joiner—to one key, for search, deduplication, and indexing. Its engine is written in Rust and reaches Python as a thin binding, with a command line beside it.

The normalizer is not a font, and the difference is the point of comparing them. A font is asked for the written units of the text in front of it and stops there; the normalizer is asked whether two texts are the same, so it must also decide what to do with the freedoms that the model leaves to the text. Two things in it are worth a reader’s attention, because they are readings of this document rather than restatements of it.

Reduction without a glyph set. The lower layers of the process name glyphs—the basic mapping of phase Ia is a cmap, and the cursive joining of phase IIa assigns a joining position to the glyph of a letter. The normalizer has no glyph set, so it holds the written units a font would draw as names of its own, and the phases act on those names. This is what a font-independent implementation has to do, and the data of this document are what makes it possible.

Normalization is not identity. The output of its normalize is a key, not a recommended spelling. It encodes by glyph shape, so a word can come back written with different characters than it went in with—the normalizer’s own documentation gives m o ng g o l becoming m o a g ² n ¹ n ¹ o l—and it publishes a canonical_version for the key so that a caller can rebuild its index when the encoding changes. A reader who wants the standard spelling of a word should read the shape the normalizer reports, not the key it stores.

Adding an implementation. The list above is not closed. An implementation that follows this document is a reading of it, and a disagreement between the model and an implementation is usually a place where the model is not yet stated clearly enough to be followed—which is a report this document wants. If you have one, open an issue or send a revision to the UTN.