Writing systems

A taxonomy of writing; the invention of writing; phonetic writing systems; optimality

What was it about the letter-shapes that struck his soul with the force of a tribal mystery? – Don DeLillo, The Names

We read and write with such ease that we seldom recognize that the invention of writing was a remarkable intellectual achievement. Writing allows us to communicate over great distances of space and time, record our own thoughts and memories, and even decipher the languages of the ancients. But the insight that underlies writing – that visual signs can represent spoken language – is far from obvious, and it took several millennia of refinements for the earliest writing to evolve into our modern alphabet. The great variety of scripts in use today – the Latin alphabet, Arabic, Chinese characters, the scripts of South Asia, among many others – attests both to the diversity of human language and the long and tortuous history of writing itself.

A taxonomy of writing

Since the invention of writing five and a half millennia ago, a great number of writing systems have come into and fallen out of use, but almost all of them can be classified as either logographic, syllabic, or alphabetic, according to what kind of linguistic unit the individual symbols represent.

The symbols of a logography represent semantic units – words or parts of words – rather than sounds. Typically each symbol corresponds to a single syllable, so that words with multiple syllables are composed of multiple symbols. Chinese characters are the best-known logographic system in use today; aside from Chinese, they are also used in Japan, and historically in Korea and Vietnam as well. They are not purely logographic – such a system would demand an impractically large number of distinct symbols. Rather, most Chinese characters have one visual part that suggests the semantic meaning and another that suggests the pronunciation. This hybrid system is often described as logosyllabic.

In a syllabary, the symbols represent syllables, but unlike in a logosyllabary, the symbols are strictly phonetic, so the same symbol is used for a syllable no matter its meaning in context. Two languages that use syllabaries are Japanese (called kana), and Cherokee, an indigenous language of North America.

The alphabet is readily familiar to English speakers. Aside from true alphabets, in which consonants and vowels have equal status, there are also abugidas, where the consonants are the primary signs and the vowels are secondary markings, and abjads, where the vowels are optional and often omitted entirely. Most of the scripts of South Asia are abugidas, as is the Geʽez script of Ethiopia. Arabic and Hebrew are abjads.

The invention of writing

Writing was invented independently at least three times in human history: in ancient Mesopotamia (Sumerian cuneiform, c. 3000 BCE), China (the oracle bone script, c. 1200 BCE), and Mesoamerica (the Maya script, c. 500 BCE). Some scholars consider the hieroglyphs of ancient Egypt to be a fourth instance; the ancient Egyptians likely knew of the concept of writing from Mesopotamia, so hieroglyphs were not an entirely novel invention, but the concrete influence of cuneiform seems to be minimal.¹

Why did writing arise in these places, but not in otherwise sophisticated civilizations like the Incas of South America? After all, true writing arose in each case out of 'pictographic' writing where simple drawings (an eye, an ox, a river, etc.) represent a limited set of concepts, and many societies that did not independently develop writing had similar pictographic systems. The crucial insight – called the rebus principle – to transforming pictograms into true writing is that a pictographic sign can be reused to represent another word with the same pronunciation; for instance, in English a drawing of an eye can also represent the abstract pronoun I, which might be otherwise difficult to unambiguously render as a drawing. Applied enough times, a true writing system can be constructed out of pictograms.

One possibility is that this intellectual leap could only happen for certain types of languages. Each of the three languages for which writing was invented – Sumerian, Old Chinese, and Classic Maya – had a large number of monosyllabic words and thus a large number of homophones or near-homophones, which makes the application of the rebus principle much more practical.² If true, then, given the importance of writing to the development of civilization, it would be a striking instance of a language's internal grammatical structure influencing the course of human history.

While writing arose in multiple places, the alphabet was invented only once, in the ancient Near East some time between 2000 and 1500 BCE for a Semitic language. This alphabet, by way of its better-known descendant the Phoenician alphabet, is the common ancestor of all the alphabets of modern Europe (Latin, Cyrillic, Greek), the Middle East (Arabic, Hebrew), and South Asia (the Brahmic scripts).

The creation of the alphabet rested on another insight: that syllables can be further divided into the smaller units of consonants and vowels. As natural as this ‘alphabetic principle’ seems to us today, it is unintuitive for people without linguistic training, and, just as the original invention of writing was dependent on certain features of the languages involved, the invention of the alphabet may have been made possible by certain properties of Semitic languages.³

A candidate for a second independent invention of the alphabet is Hangul, the Korean writing system invented apparently out of whole cloth by King Sejong the Great in the 15th century. Hangul is without doubt a linguistically sophisticated system: not only does it distinguish consonants and vowels, unlike the existing logographic scripts of East Asia at the time, but the shapes of the letters even suggest the shape of the mouth as they are pronounced. Whether Hangul was an entirely novel invention is not known, as knowledge of other alphabets may have diffused to Korea by then. Regardless, its eminent advantages over hanja, Chinese characters used to write Korean, led it to become the dominant script in Korea by the 20th century.

Aside from entirely independent inventions, there have been a dozen or more documented instances where an illiterate person, exposed to the idea of writing by contact with literate societies, invented their own form of writing. The best-known case is that of Sequoyah, a 19th-century Cherokee man who created the Cherokee syllabary which is still in use today. Others include the Bamum script of Cameroon, an Inuit script of Alaska, the Ndjuka script of Suriname, and two scripts of the Caroline Islands in the Pacific.

Interestingly, these cases follow a consistent pattern. First the inventor tries to create a symbol for each word – a pure logography. Realizing the futility of that approach, they switch to a syllabary (or logosyllabary) where the basic units are consonant–vowel pairs. Typically there is little or no relationship between phonetic qualities and visual form; that is, syllables that sound alike do not look alike. Nor does the order of the symbols reflect any phonetic relationships. The fact that these re-inventions of writing always result in a syllabary, never an alphabet, shows how counterintuitive the alphabetic principle is.⁴

Phonetic writing systems

Alphabetic systems differ in the regularity of their spelling. This fact is well-known to anyone who has wrestled with the vagaries of English's famously inconsistent spelling system. English is often compared unfavorably to languages with a more consistent correspondence between letters and sounds.

The most literal correspondence between letter and sound is phonetic, where each symbol represents a distinct speech sound – 'distinct' meaning that it can be distinguished by its acoustic properties and how it is physically produced in the vocal tract. The International Phonetic Alphabet (IPA) is an example of phonetic alphabet, and moreover a universal one: it aspires to represent every sound in every spoken language. The IPA has a larger inventory of symbols than English, but not unmanageably so – the entire alphabet fits on a single one-page chart.

It might seem desirable that all languages use a phonetic alphabet. In practice, such an alphabet would be encumbered by unnecessary detail, and require a level of phonetic awareness that most speakers do not possess. Consider the pairs grate/grater and grade/grader in English. One might suppose that the difference between grate and grater is the r sound at the end, and the difference between the two pairs is in the t and d sounds. This analysis is mistaken; close examination reveals that the addition of the r at the end also transforms the pronunciation of the t/d sound into the same, 'weakened' sound. The words grater and grader are actually distinguished not by the t/d sound but by the a vowel, which is longer in grader than in grater.⁵ A phonetic alphabet would require a writer to be aware of these fine distinctions, which are anyway totally superfluous because the sound changes involved are completely regular: the a, t, and d sounds always behave in this way under the same circumstances.

Aside from technical alphabets like the IPA, all real alphabets are phonemic rather than phonetic, meaning that they represent the abstract sounds rather than the concrete way they are pronounced. Phonemic alphabets take advantage of the regularity in a language to cut down on redundant information.

Even in a phonemic system, it is not always possible to unambiguously spell a word, knowing its pronunciation. In Russian, vowels change their sounds depending on whether they are stressed or unstressed. The vowel 'o' for instance is pronounced as a full 'oh' when stressed, a partially reduced 'ah' when preceding a stressed syllable, and a fully reduced 'uh' otherwise. Similarly, the vowel 'a' is also said 'uh' when unstressed. A Russian speaker hearing the sound 'uh' cannot be sure whether it corresponds to the letter 'o' or 'a'.

English writing, needless to say, is neither phonetic nor phonemic. But the apparent irregularity of English spelling belies a degree of underlying consistency. The spellings of sane and sanity, south and southern, and know and knowledge, for instance, sacrifice strict phonemic correspondence for the sake of preserving the form of the common root.⁶

A looser correspondence between sound and symbol can be advantageous in some respects. For a language with many dialects a strictly phonemic system may be inadequate. In one recorded instance of a newly-created writing system for an Austronesian language in Papua New Guinea, it was decided for the system to be phonemic for one of two dialects; speakers of the second dialect had to simply memorize the spellings of certain words which were pronounced differently in the first dialect.⁷ Morphophonemic systems are also more durable to language change – although it is often language change itself which turns a phonemic system into a morphophonemic one.

Optimality

There is no fundamental relationship between a spoken language and the script conventionally used to represent it. Indeed, many languages remain entirely unwritten, and some languages (Japanese, Serbo-Croatian, Hindi-Urdu) are commonly written in multiple scripts.

Nonetheless, for a particular language, some writing systems are better-suited than others. Syllabaries are ideal for languages with simple consonant–vowel syllables, which allows for a manageable number of symbols (the Japanese syllabaries and Cherokee each have less than a hundred); a syllabary for English would require thousands of symbols.

While the omission of vowels in an abjad might seem to pose tremendous difficulties to the reader, it is actually reasonably well-suited for Semitic languages like Arabic and Hebrew, as words are typically made of a fixed set of consonants, with vowels changed to mark grammatical function or to derive related words. The omitted vowels can usually be recovered by a trained reader because it is the consonants that determine the fundamental meaning of the word.

An oft-cited advantage of a logographic writing system is the ability to distinguish homophones, of which Chinese has a large number – though Chinese speakers are still able to distinguish homophones in speech without the aid of different characters, and Old Chinese, for which the logographic system was originally devised, had fewer homophones than modern Chinese.

Whether some writing systems are more optimal in general is less clear. Readers and writers, and learners and experienced users, have different needs that are sometimes at odds. Writing systems that are easier to learn tend to be less efficient for experienced users, and vice versa. An extreme example is the IPA: if all the world's languages used it, then a reader would only need to memorize a small set of symbols to be able to correctly pronounce a written text in any language. But writing a text in this level of detail would be oppressive.

A writing system that merges or omits symbols, such as an abjad, may be more efficient to write but more difficult to read. Conversely, a writing system with a large array of distinct symbols – such as Chinese – may be faster to read but more laborious to write. Syllabaries and logographies are typically more space-efficient, but alphabetic systems with fewer symbols are easier to input and reproduce digitally.

Perhaps surprisingly, logographies may be easier to learn, at least initially, than alphabetic writing systems. A 1971 study found that American children with learning disabilities could be taught to read simple sentences in English written using Chinese characters (i.e., using the Chinese character for person to represent the English word 'person'), despite failing to read the English alphabet after years of instruction.⁸ This is yet another example of the counterintuitive-ness of the alphabetic principle: it is much easier for children to understand that a symbol stands for a whole word than for a single speech sound. The difficulty is even greater when the alphabet is not even regular: progress in reading is slower for English-speaking children (and speakers of other languages with irregular or complicated spelling, like French and Danish) compared to Finnish, which has a highly regular alphabet.

Further reading

Writing systems are an endlessly fascinating topic. A good reference is The World's Writing Systems, edited by Peter T. Daniels and William Bright and available at the Internet Archive. It covers dozens of historical and current writing systems from a modern linguistic standpoint, and its bibliography is a good starting point for further research. You can also look at the papers that cite it on Google Scholar to find more recent sources, as the book was published in 1996. For concrete examples of scripts, consult ScriptSource.org and Omniglot.

Bibliography

Aro, Mikko (2004). "Learning to read: the effect of orthography." Dissertation, University of Jyväskylä.

Berry, Jack (1977). "'The making of alphabets' revisited." In Advances in the Creation and Revision of Writing Systems (ed. Joshua A. Fishman), pp. 3–16.

Cheung, Him et al (2001). "The development of phonological awareness: effects of spoken language experience and orthography." Cognition, vol. 81, no. 3 (October 2001), pp. 227–241

Daniels, Peter T. (1992) "The syllabic origin of writing and the segmental origin of the alphabet." In The Linguistics of Literacy (ed. Pamela Downing, Susan D. Lima, & Michael Noonan), pp. 83–110.

— (1996). "The invention of writing". In The World's Writing Systems, pp. 579–586.

DeFrancis, John (1996). "How efficient is the Chinese writing system?" Visible Language, vol. 30, no. 1, pp. 6–44.

Gleitman, Lila R. & Paul Rozin (1977). "The structure and acquisition of reading I: Relations between orthographies and the structure of language." In Toward a Psychology of Reading (OCLC 1082251166), pp. 1–54.

Krepel, Alexander, Elise H. de Bree & Peter F. de Jong (2021). "Does the availability of orthography support L2 word learning?" Reading and Writing, vol. 34, pp. 467–496.

Regulski, Ilona (2016). "The origins and early development of writing in Egypt." In The Oxford Handbook of Topics in Archaeology (online edition).

Rogers, Henry (1995). "Optimal orthographies." In Scripts and Literacy (ed. Insup Taylor & David R. Olson), pp. 31–43.

Rozin, Paul, Susan Poritsky, & Raina Sotsky (1971). "American children with reading problems can easily learn to read English represented by Chinese characters." Science, vol. 171, no. 3977, pp. 1264–1267.

Seifart, Frank (2006). "Orthography development." In Essentials of Language Documentation (ed. Jost Gippert, Nikolaus P. Himmelmann, & Ulrike Mosel), pub. De Gruyter Mouton.

Venezky, Richard L. (1970). "Principles for the design of practical writing systems." Anthropological Linguistics, vol. 12, no. 7 (October 1970), pp. 256–270.

Wiese, Richard (2004). "How to optimize orthography." Written Language & Literacy, vol. 7, no. 2, pp. 305–331.

Ziegel, Johannes C. & Ludovic Ferrand (1998). "Orthography shapes the perception of speech: The consistency effect in auditory word recognition." Psychonomic Bulletin & Review, vol. 5, no. 4, pp. 683–689.

1. Regulski (2016)

2. Daniels (1992)

3. Daniels (1992)

4. Daniels (1996)

5. Example courtesy of Gleitman & Rozin (1977)

6. Examples courtesy of Rogers (1995)

7. Seifart (2006)

8. Rozin, Poritsky, & Sotsky (1971)