Script

← Guide index · 中文 · third item in the navigation bar, between Phonology and Sound changes

The Script page manages a language’s custom writing systems: alphabets, syllabaries or abugidas of your own design, or ready-made Unicode blocks you borrow (runes, Tengwar in the Private Use Area, Devanagari…). Once a script is defined, the lexicon gets a Script column, corpus sentences and phrases get a script line, and dictionary exports include it — all generated automatically from rules.

A language can have several scripts (say, a formal hand and a cursive). Next to the page title is the switch between the three sub-pages Glyphs / Mapping rules / Preview; further right, each script of the language has its own bookmark tab. Click a tab to switch; press and drag it onto another tab to change their order.

1. Creating a script and its basic properties

Use New script at the far right of the title bar, then set in the inspector:

Field	Description
Name	e.g. “Aelith runes”
Type	Alphabet / Abjad / Abugida / Syllabary / Logographic / Featural / Mixed / Other. The type doesn’t affect the engine; it only decides which hints the Mapping rules page gives you
Direction	Left to right / right to left / top to bottom. RTL makes displays use dir="rtl"; top to bottom is vertical by itself
Vertical layout	Off by default. When on, everywhere this script is displayed (corpus, phrasebook, lexicon, entry cards, character panel, the preview on this page) it is laid out vertically (writing-mode) with upright glyphs; columns run right-to-left, or left-to-right if the direction is set to left to right
Parentheses	How parentheses in the source text (half- or full-width) are written in the script: Keep them, transliterate inside and outside separately (default, e.g. “kala（mira kala）”) / Drop them, merge the content into the word (optional letters such as “tal(a)n” are merged into the word; a parenthesised phrase becomes a space) / Drop them together with their content
Font	A system font name, or an imported font file embedded in the project
Notes	Free text

2. Fonts

• System font name: for fonts installed on the computer, just type the name.
• Import font file: TTF / OTF / TTC / WOFF / WOFF2. The file is stored in the project file as a data URL, so it displays on any computer. TTF / OTF / TTC can also provide the glyph list (see below).
• With a font, the glyph cards, the inspector title and the Character input use the script’s font; Private Use Area glyphs show as boxes without one.
• Changing the font in Skin: below Skin → Font slots, every script of the current project is listed and can be given its own font (installed from the font library or imported locally), which takes priority over the setting here; it only changes the display on this computer and doesn’t touch the project file.
• Without a font, script text falls back to the Skin’s “custom script” font slot, and then to the language data font.

3. Glyphs

Three sources, which can be mixed:

1. Read glyphs from font: parses the font’s cmap (character-to-glyph map) and post table (glyph names). With fonts exported from tools like FontCreator, the names you gave the glyphs appear under Name. For basic Latin letters and digits, the transliteration is filled in with the character itself (many home-made fonts map their glyphs onto a–z, so they work immediately); other characters are left empty for you to fill in. After choosing a font you confirm above the glyph table: the inspector shows an import preview (glyph cards in that font; characters already present are faded and will be skipped), and Import glyphs embeds the font into the script as well.
2. Paste glyph list: one line per glyph — “character transliteration name”, separated by tabs or spaces, e.g. ᚠ f fehu. The import preview in the inspector updates as you paste, and the card for the line you changed flashes; characters already present are faded and skipped.
3. Add glyph: add them one by one by hand.

Each glyph has:

Field	Description
Character	The glyph itself (combining marks allowed; Private Use Area code points too)
Transliteration	The corresponding spelling in the primary orthography; the mapping rules are generated from it. Leave it empty to exclude the glyph from automatic mapping
Name	Glyph name
Category	Letter / Vowel sign / Consonant / Syllable / Mark / Number / Punctuation / PUA glyph / Space / Other, or a custom one. Auto-categorize looks at the transliteration first: it is read as sounds using the primary orthography (if Transcribe from is set to a pronunciation, the value is taken as sounds directly) — all vowels make a vowel sign, all consonants a consonant, both a syllable; transliterations that are punctuation or digits count as such. Only glyphs without a transliteration are classified by the character itself (letter, mark, PUA glyph…). Categories you named yourself are left alone, and a toast reports how many went into each category. New glyphs read from a font or pasted in are categorised the same way
Notes

Glyph cards can be filtered by category; click a card to edit it in the inspector.

4. Mapping rules

The rule language is the same as on the Sound changes page (see Rule language), with a list view and a source view.

The special part is the line @glyphs: it expands into the mapping generated from the glyph table — one transliteration > character rule per glyph with a transliteration, sorted by transliteration length, longest first (th before t, so the short one doesn’t eat the long one). You can inspect the expansion under Automatic mapping at the bottom of the page. Transliterations containing symbols that mean something in rules (? . # *), or capital letters that share a name with a class, get a backslash in front when expanded (\? > ⸮) so they are matched literally — giving a glyph to a question mark no longer turns entire words into that glyph.

Rules written before @glyphs run first; rules after it run last. That is how the different script types are handled:

Type	Typical approach
Alphabet	Usually just @glyphs; for digraphs, give the dedicated glyph a longer transliteration
Abjad	Delete unwritten vowels before @glyphs: [aiu] > / C_
Abugida	First delete the inherent vowel after consonants, a > / C_, then map the other vowels to vowel signs; add a vowel killer to consonants without a vowel at the end of a word, e.g. C > C◌् (◌ is only a placeholder — write the actual character)
Syllabary	One glyph per syllable with its transliteration (ka, ki…); longest match is handled automatically. For complex packing patterns use Syllable packing below
Logographic	Fill in each word by hand in the lexicon’s Script form; rules only handle the regular part
Featural	Map classes to components first, then assemble them with rules
Mixed	Write each part as a separate block of rules and use -* stage snapshots to inspect intermediate results

Rules can refer to the classes and digraphs from the Phonology page. In the inspector’s Try it box, type a transliteration (words separated by spaces) and the script appears immediately.

The inspector also has Transcribe from: by default a script transcribes the entry’s headword, but you can switch it to a stem, the pronunciation in one orthography, or an inspector module (for instance a field that holds another transcription); when that field is empty it falls back to the headword. The lexicon’s script column, entry cards and this script’s line in the corpus all follow it.

Syllable packing and Automatic mapping are two titled sections below the rule list; the triangle to the right of each title collapses it. When Add rule makes the rule list taller, they simply move down.

Syllable packing

Syllable packing, below the rules tab, is an alternative route for CV / VC syllabaries: when enabled, the rule chain is skipped and the transliteration is packed into syllables using the glyph readings.

The packing grid is derived from the glyph table itself: every glyph whose reading splits into exactly “one consonant + one vowel” (or the reverse) is a grid cell, and readings longer than that (whole words, root blocks) are preferred as whole chunks. So once the glyph readings are right, packing works.

The remaining settings are things the software can’t infer and you need to tell it:

Setting	Description
Killer (vowel-cancelling) reading	The reading of the glyph that removes a vowel or marks a voiceless sound, e.g. ∅
Borrowed vowel	Which vowel to borrow when a consonant has no vowel to pack with; the killer then cancels it
Letters needing a mark	Space-separated. These consonants read as voiced when they appear alone, so they are either repeated across two cells or paired with the killer
Vowels	Space-separated; tells the software which units are vowels
Transcription units	Space-separated, digraphs included, e.g. th dh ch á â a e …
Spelling → glyph letter	One per line, e.g. dh=th, b=p, ch=h, folding voiced sounds and digraphs into the letters used in the glyph table
How many copies a vowel gets	e.g. á é = 2, â ê = 3: half-long and long vowels are written once more in the middle
Base vowel of a long vowel	e.g. á=a, telling the software which plain vowel a vowel with a diacritic is based on

The packing logic: a short vowel between two grid cells is written twice; a voiceless sound appears across two cells while a voiced one appears only once; word-initial voiceless sounds and consonants with no vowel to pack with get the killer; in the headword part before a middle dot, the killer and long-vowel marks may be omitted. Without a killer reading, a word-final consonant with no vowel to pack with is simply written as the consonant itself, without borrowing a vowel — so the preview doesn’t grow an extra vowel out of nowhere.

The Kessar script of Theusrin (瑟乌丝林语) is configured this way following the “Writing rules” chapter of its grammar. Checked against the spellings of 414 words in that grammar, about 60% match exactly; most of the rest are cases where the grammar writes a whole root with its inherent-sound glyph — that is lexical information, and can be filled in by hand in the entry’s Script form to override.

5. Preview

The Preview sub-page shows the first 40 lexicon entries and the first 10 corpus sentences converted to the script, to help check the rules.

6. Connections

Where	What you see
Lexicon list	Tick “Script: name” under Columns to add a column shown in the script’s font
Lexicon Edit mode	The Script form field overrides the automatic result by hand (for logographic scripts and irregular spellings)
Entry card / hover card	A script line below the headword
Corpus	A script line at the top of the editor and of list cards; all exports (Leipzig, Markdown, HTML, LaTeX) include the script line; custom templates use ``
Phrasebook	A script line at the top of each card
Dictionary export	The script follows the headword (can be turned off)
Character panel	The project characters tab lists every glyph of the current language; click to insert
Skin	Each script can be given its own font (taking priority over the script’s own font); the “custom script” font slot is the default when there is no font

7. Tips

• Want the script to apply only to some words? Set the Script form of the words you don’t want converted to the same text as the headword.
• No system font can display Private Use Area (PUA) glyphs, so embed the font. A font assigned to a script in Skin only applies on your computer; projects you send to others rely on the embedded font.
• A large font file (several MB) makes the project file larger too. That is a deliberate trade-off: the project carries everything it needs.