Skip to content
Cantonese Fonts

Basic Concepts

This section outlines the elements involved in the Cantonese Font. While the terms recur in these documentation, I suggest just skimming over them for now. You can always re-visit them when you need, and you can search using the cmd/ctrl-K keyboard shortcut while reading other pages.

Font

A font, in the most basic form, contains instructions (paths) about how each character should be drawn. These instructions describes the shape and knows not about color; I’ll refer to these (standard) fonts as monochrome fonts. While they are by far the most common kind of font you’ll encounter, most of the Cantonese Fonts are color fonts (which also contains monochrome instructions as a fall-back fail-safe).

Normal fonts contain about 2,000 glyphs (drawings), and contains Latin, Cyrillic, and Greek characters as well as numbers and punctuations. A Chinese-Japanese-Korean (CJK) font contains many more glyphs to represent the characters; the Cantonese Font, with 65,355 glyphs, maxes out what could be packaged in a font.

CJK characters are referred to by a “standard ID number,” assigned by the Unicode Consortium. This ID number is called a codepoint.

Modern, sophisticated fonts contain not only shapes, but addition information about how glyphs may interact with one another. These rules, or features, describes the interactions. An example is a ligature, where particular pairs of letters gets swapped with a different glyph. Many other kinds of rules exists, and the Cantonese Font uses about 150,000 features to specify the intricate behaviours.

Fonts are interpreted by software, often on the operating system level. The mechanism is complex and takes place over multiple stages, including layout, shaping, and rendering. I often refer to this generically as “the font renderer”.

The Cantonese Fonts are not true fonts; instead of relying on paths, each glyph actually contains a (mathematical) vector image in the SVG format. This kind of “false font” is called an Opentype SVG (OT-SVG) font.

The Cantonese Fonts is planned as full families so that the Bold, No Jyutping, Only Jyutping font variants shares the same features, and can be used in a mix-and-match fashion.

All the fonts are / will be available under the SIL Open Font 1.1 License, which means that they can be used for commercial projects. The only restriction is you cannot resell the fonts. Special font variants will be embargoed for a few years but can be accessed now under the Lab 🧪 sponsorship program.

Cantonese and Jyutping

Cantonese uses the Chinese script, which consists of ideographic characters 字. A word 詞 is a higher unit organization that is formed from multiple words. One character is spoken with one syllable (but there are exceptions).

Romanization is using alphanumeric characters to denote the sound. Jyutping is currently the most popular Cantonese romanization system, though you will also see a similar system called Yale (which does use some tone marks).

Jyutping conceptualizes Cantonese sounds into four components, and in the extended Jyutping (Jyutping+) used by the Canto Font, these four components are represented with different colors:

  1. Onset (brown)
  2. Nucleus (blue)
  3. Coda (blue or gray)
  4. Tone (red)

Cantonese contains six tones, which vary by how high they sound (pitch) and whether the utterance stays at the same pitch (inflection). The sequence of numbering has a historical basis.

CJK ideographs have three components:

  1. How they are drawn,
  2. How they sound (pronunciation), and
  3. What it means (semantic)

The Cantonese Font is deeply engaged in relating the shape with the pronunciation.

Most characters corresponds to a unique sound but some characters can be pronounced in many ways (e.g., 行 can be haang4, hang4, hang6, hong2, hong4 amongst others). How a character is pronounced is context-sensitive, that is, it depends on the meaning. For example, in “銀行 bank”, the 行 must be hong4 and not the alternatives. Sometimes this dependence is relaxed: 妹妹 can be pronounced with a variety of tones but carrying the same meaning.

In certain cases, the sound also compels the meaning.

Some characters are written in multiple ways, for a number of historical reasons; some have their own codepoints while others do not. Sometimes one written shape corresponds to multiple characters (e.g., 体 is used as simplified for 體 but it was a Traditional character in its own right).

Some sounds have a meaning, but no standardized / accepted glyphs to represent it. An example is he3, which means roughly “carefree”.

Most characters are pronounced with one Jyutping but a handful of characters are read with two. These are called bi-syllabic characters. An example is 卅 (saa1 aa6).

In the Cantonese Font, each character codepoint is represented by one or more glyphs: each unique combination of Chinese character and Jyutping is its own glyph. 行 haang4 is a different glyph than 行 hong2. The default reading is the glyph/Jyutping that will be shown when a character is typed without any neighbors.

Chinese language

Chinese ideographs (script) can be used to express a number of languages. Standard written Chinese 書面語 is a form of grammatical / vocabulary formulation that aligns closely with Mandarin usage. This can be uttered in Cantonese but represents a formal, stilted register. Written Cantonese directly corresponds to how Cantonese is spoken. The presence of two parallel forms of writing is called di-glossic.

Chinese script, unlike Latin/Cyrillic, is written without spaces to denote different words. The reader is expected to “cut up” the sentence into fragments, and this is not a certain process. Segmentation refers to this action of cutting up sentence into words.