Abugidas

Chapter 4 said a syllabary gives every syllable its own unrelated symbol. Chapter 6 said an alphabet gives vowels equal status with consonants. The abugida sits between them: each symbol is a consonant with a vowel already attached, and the other vowels are marked by changing it. This chapter covers the largest family, the Brahmic scripts of South and Southeast Asia, and Ethiopia's Ge'ez script, which reached the same idea separately.

The Word and the Idea

• The term. From Ge'ez, where it names the Ethiopic letters; it derives from the letters "ä, bu, gi, and da". Peter T. Daniels proposed it as a linguistic label in 1990, the same year as "abjad" (Chapter 5).
• Against a syllabary. In a syllabary, a single symbol denotes the consonant-vowel pair, and letters sharing a consonant or vowel usually look unrelated. In an abugida the related syllables share a visible element.
• Against an alphabet. In an alphabet vowels have status equal to consonants; in an abugida vowels are secondary marks on the consonant.
• Silencing the vowel. Many abugidas have a diacritic that removes the inherent vowel to leave a bare consonant; in Sanskrit it is called the virāma or halantam.
• Clusters. Consonant clusters may be written by merging letters (conjuncts); some scripts stack the consonants vertically and others use special half-forms.

The Brahmic Family

The Abugida article says Indic scripts originated in India and spread to Southeast Asia, and reached Bangladesh, Sri Lanka, Nepal, Tibet, Mongolia and Russia. It says all surviving Indic scripts descend from the Brahmi alphabet. The article's examples include Devanagari (Hindi, Sanskrit, Marathi, Nepali), Bengali, Tamil, Telugu, Kannada, Malayalam, Thai, Lao, Burmese, Khmer, Tibetan, Javanese, Balinese and Baybayin.

Devanagari

From Wikipedia's Devanagari article (first 100,000 of 151,714 characters read).

• Use. Hindi, Sanskrit, Marathi and Nepali among over 120 languages; the article calls it the fourth most widely adopted writing system in the world.
• Origin. From the 3rd-century BCE Brāhmī script, through Nāgarī.
• Inventory. 48 primary characters: 14 vowels and 34 consonants. Written left to right, with a horizontal line (shirorekha) running along the top of full letters.
• Vowels. The consonant क carries the inherent /a/. Other vowels are dependent signs (matras) attached to it: का for ā, कि for i. A virama, as in क्, removes the vowel.

Ge'ez (Ethiopic)

From Wikipedia's Ge'ez script article (first 100,000 of 163,469 characters read) and the Abugida article.

• Not Brahmic. The Abugida article says it derives from a different abjad, the Sabean script of Yemen; the Ge'ez article says from Ancient South Arabian, though its origins are debated.
• History. First an abjad (the Ge'ez article dates this to about the 1st century CE; the Abugida article says until about AD 350), then vowel marking was added in the 4th century.
• How vowels are shown. By modifying the shape of the consonant itself, in seven basic vowel orders (ä, u, i, a, e, ə, o), not by separate diacritics. The article says the vowel marks were fused with the consonants, "so the system is laid out as a syllabary".
• Use and size. Ge'ez, Amharic, Tigrinya and other languages of Ethiopia and Eritrea; 26 consonantal letters. Written left to right today, right to left as an abjad.
• Naming. Unicode and ISO 15924 call it "Ethiopic". Amharic and Tigrinya speakers often call it fidäl (ፊደል).

Thai

From Wikipedia's Thai script article (first 100,000 of 219,119 characters read).

• Origin. From Old Khmer script, rooted in the South Indian Pallava alphabet and a southern branch of the Brahmi tradition. The infobox gives 1283 to present, with Ramkhamhaeng as creator; the article dates the Ram Khamhaeng Inscription to 1292 and says many scholars question its authenticity.
• Letters. 44 consonant symbols representing 21 distinct consonant sounds.
• Vowels. Vowel signs can go above, below, left of or right of the consonant, or in combinations of these places.
• Tones. Four tone diacritics; tone depends on the consonant's class, vowel length, the closing consonant and the mark, if present.
• No spaces. Words are not separated by spaces, except in certain linguistically motivated cases.

Seeing the Marks in Python

My own example; I ran it and copied the output. It builds the Devanagari consonant क (ka) and adds the vowel signs from the article.

Three observations, all mine from the output. First, in Devanagari the vowel sign and the virama are separate code points stored after the consonant, and the conjunct क्ष is stored as ka + virama + ssa, so the merged shape is something the font draws. (That also explains why the Hindi word in Chapter 1 counted as four code points, not as the number of shapes a reader sees.) Second, Unicode stores Ge'ez as whole syllables, one code point per consonant-vowel pair, treating it like a syllabary even though its shapes are related; seven orders of the same consonant (HA, HU, HI, HAA, HEE, HE, HO) appear in a row. Third, in Thai the vowel sign SARA E is stored before the consonant, matching its position on the left in writing.

Try It Yourself