What Is Writing?

Writing Systems of the World

Course 1 · Chapter 1 · What Is Writing?

You can read this page because a set of symbols has been tied to a language by convention. Every script on Earth does the same job, but the ties differ: some symbols stand for meanings, some for syllables, some for single sounds. This chapter sets out the vocabulary for the whole course and asks where writing came from.

The idea in a box

Wikipedia defines a writing system as "any conventional system for representing a particular language using a set of symbols". The useful question to ask of any script is: what does one symbol stand for? The answer sorts almost every script in the world into a type.

Writing Is Not Language

Speech comes first. The Writing system article says that while many spoken or signed languages are not written, there are no written languages without a spoken counterpart.
One language, several scripts. A script is a set of symbols, and the same language can be written in more than one. (Serbian written in both Latin and Cyrillic letters is my own example, not the article's.)
One script, many languages. The article calls the Latin alphabet by far the most common script used by writing systems.
Drawing is not writing. On the narrow definition the article uses, drawings and maps are excluded. Proto-writing (ideographic and mnemonic symbols such as the Inca quipu) lacks the capability to fully encode language, and most proto-writing systems did not become writing systems.

The Vocabulary

The article's typology divides systems first by what the symbols represent: morphographic (meaning-based) or phonographic (sound-based). Phonographic systems split into syllabic and alphabetic ones, and alphabets into abjads, pure alphabets and abugidas. These are the six terms the rest of the course relies on.

Type	One symbol stands for	Example (chapter)
Logographic	a unit of meaning, such as a word or morpheme	Chinese characters (3)
Syllabary	a syllable or mora	Japanese kana (4)
Abjad	a consonant only	Hebrew, Arabic (5)
Alphabet	a phoneme, roughly (consonants and vowels)	Greek, Latin (6)
Abugida	a consonant plus vowel, with the vowel marked by a predictable change to the symbol	Devanagari (7)
Featural	a sub-phonetic element (a feature of how a sound is made)	Hangul (8)

Watch out: "alphabet" is not a synonym for "script"

People say "the Chinese alphabet" or "the Japanese alphabet", but neither is an alphabet in this typology. Also, the article says most systems are mixed, combining sound-based and meaning-based elements, and that the featural class, proposed by Geoffrey Sampson, is disputed by several scholars, including John DeFrancis. The table is a map of tendencies, not sealed boxes.

How many symbols?

The type largely predicts the size of the symbol set. According to the article, alphabets typically use fewer than 100 symbols, syllabaries may use hundreds, and logographies thousands. It gives Japanese kana as covering roughly 100 moras and the Yi script as having 756 symbols. The World's Writing Systems project, which the article links to, lists 294 known writing systems.

Where Writing Began

The article describes writing as invented independently in several places. Dates are as the article gives them.

System	Place	Date (per the article)
Cuneiform	Southern Mesopotamia (Sumerian)	Akkadian writing appears in significant quantities c. 2350 BC
Egyptian hieroglyphs	Egypt	Evolved from proto-writing between 3400 and 3100 BC; earliest coherent texts c. 2600 BC; generally regarded as independent of cuneiform
Chinese characters	Yellow River valley	c. 1200 BC, with no evidence of contact with the Near East
Mesoamerican scripts	Including Olmec and Maya	Also considered independent inventions
Indus script	Indus valley	c. 2600 to c. 2000 BC; undeciphered, and whether it was true writing is disputed
First alphabetic writing	Sinai Peninsula (a Semitic language)	c. 1800 BC; most alphabets descend from or were inspired by Proto-Sinaitic

The last row matters for the rest of the course: the great majority of alphabetic scripts, and several abjads and abugidas, trace back to one source. Chapters 5 to 7 follow those lines.

Seeing Script Types in Python

Unicode (met in Computational Linguistics 1) gives every character a name that includes its script. The example is my own; I ran it and copied the output.

import unicodedata
for ch in ["A", "Ж", "א", "ع", "क", "あ", "한", "語"]:
    print(ch, unicodedata.name(ch))

A LATIN CAPITAL LETTER A Ж CYRILLIC CAPITAL LETTER ZHE א HEBREW LETTER ALEF ع ARABIC LETTER AIN क DEVANAGARI LETTER KA あ HIRAGANA LETTER A 한 HANGUL SYLLABLE HAN 語 CJK UNIFIED IDEOGRAPH-8A9E

Eight characters, eight names, and every one of the six types above is represented by at least one. The same word "water" in different scripts shows how the symbol count varies (again my own example, run and copied):

words = {"English": "water", "Greek": "νερό", "Hebrew": "מים",
         "Hindi": "पानी", "Japanese": "水", "Korean": "물"}
for lang, w in words.items():
    print(lang, w, len(w))

English water 5 Greek νερό 4 Hebrew מים 3 Hindi पानी 4 Japanese 水 1 Korean 물 1

Watch out: this counts code points, not symbols a reader sees

Python's len counts Unicode code points. The Hindi word shows 4 because vowel signs are separate code points; how a reader groups them is a different question, taken up in Chapter 7. The Korean syllable block 물 counts as 1 here because the string holds the precomposed form. The numbers illustrate that "length" depends on what you count, not that one script is shorter than another.

Note on sources

This chapter uses Wikipedia's Writing system article, of which I read the first 100,000 of 154,316 characters; the rest was not read. Definitions, symbol counts and invention dates are the article's, and I have not checked them against the primary scholarship. The Serbian example, the chapter-to-type mapping and the Python examples are mine.

Try It Yourself

Exercise 1

Classify each as logographic, syllabary, abjad, alphabet, abugida or featural, using the table above: Greek, Hebrew, Japanese kana, Devanagari, Chinese characters, Hangul. Which classification does the article say is disputed?

📄 View solution

Exercise 2

Run the first Python example with three characters of your own from other scripts (for example Ω, ა and ス). Which scripts do the names report, and does any character surprise you?

📄 View solution

Exercise 3

In two or three sentences, explain why a drawing of a deer is not writing but a symbol that is always read as the word "deer" might be. Use the article's idea of proto-writing in your answer.

📄 View solution

Chapter 1 Quick Reference

Writing system: a conventional set of symbols for representing a particular language
Types by what one symbol stands for: logographic (meaning), syllabary, abjad, alphabet, abugida, featural
Most real systems are mixed; the featural class is disputed
Independent inventions per the article: Mesopotamia, Egypt, China, Mesoamerica (Indus disputed); most alphabets descend from Proto-Sinaitic

Coming next

Writing Systems of the World 2 looks at the first scripts, cuneiform and hieroglyphs, and the trick that turned pictures into sounds.