Being spatial, writing is visible. But being visible is not crucial to its definition. Braille, for example, is a writing system for the blind designed to be felt with the fingers. It represents letters as a series of raised bumps that can be read by touch. In both reading by touch and reading by sight, time has been translated into space. There are also forms of language which are inherently visible and spatial, such as American Sign Language (ASL). But such languages are akin to spoken languages in their essential properties: they too unfold in time. Like spoken words, signed words are gone the moment they are produced. By contrast, writing is a transformation of language, a technology applied to language, not language itself.
Writing takes words and turns them into objects, visible or tangible. Written down, words remain on the page like butterflies stuck onto boards with pins. They can be examined, analyzed, and dissected. They can be pointed to and discussed. Spoken words, by contrast, are inherently ephemeral. So written language seems more real to us than spoken language. Nevertheless, writing is only a means of expressing language; it is not language itself. In a highly literate culture it is easy to confuse the two, since much communication is mediated by writing, and the standards of written language influence our sense of “proper” language. But writing is not language, nor is it necessary to language.
Humans everywhere use language. It is a natural and normal human behavior. Although babies are not born speaking a language, all children who are raised around other people, who can perceive the language spoken around them (they are not, say, deaf in an environment where no sign language is used), and who are within normal range in certain mental and physical facilities will inevitably learn at least one language. They pick up their mother tongue naturally over the first few years of life. Indeed they cannot really be taught it, and will resist instruction if parents try too hard to correct their baby talk. Reading and writing do not come so naturally and must be taught. By the time children learn to read and write the vast majority of their language learning (other than further vocabulary growth) has already taken place.
As far as we can tell, language has been with us since the human race began. By contrast, writing is not a fundamental aspect of human life despite the profound impact it has had on human history. All human societies have had language, but many have had no writing. The organization SIL (originally the Summer Institute of Linguistics) has counted 6,912 languages spoken in the world today. Thousands more were once spoken but are now dead. The exact tally of languages is open to dispute, as it is often difficult to determine what forms of speech are dialects of a single language and which are different languages; also, languages change constantly, and two dialects may grow into distinct languages (especially in the absence of a common written form); languages may also die out, and are now doing so at increasing rates. Thousands of the world’s languages use no writing system; no more than a hundred languages have produced a significant literary tradition.
Although writing is secondary to language, it often enjoys higher prestige. Writing is generally done more deliberately than speaking, so finished written pieces are much more carefully crafted than a typical spoken sentence. Written texts can thus convey their message more precisely, adding to the sense that writing is worth more than speech. Until the development of modern recording and broadcasting techniques, writing could reach a larger audience than the spoken word, and continue to communicate to people over a long period of time. Writing is associated with education, and education with wealth and power. The small percentage of languages that have a well-established written tradition include all the languages of national and international influence. Most of the unwritten languages are spoken by small minority groups, and many of these languages are not expected to survive the twenty-first century. Language conservation efforts must therefore include the development of writing systems and literacy programs.
Nowadays individuals faced with the task of designing a writing system for a language can draw on a wealth of literacy experience and linguistic theory. The original inventors had no such luxury. Later pioneers had the benefit of knowing that writing was possible, but still had to make most of it up as they went along.
Take King Njoya, for instance. King Ibrahim Njoya ruled the Bamum people of Cameroon from 1880 to 1931, the seventeenth king to rule from the ancient capital of Foumban. Njoya lived in a changing world, as strange people with strange new technologies encroached on traditional lands. To the north were invading Arabs, and they gave credit for their victories to a small book. Impressed, Njoya became a Muslim. Then Europeans came along with superior fire power. When asked where their strength came from, they also pointed to a book. Their book was larger, and their power the greater. Njoya therefore considered adopting Christianity, but could not accept its requirement of monogamy.
One thing was clear, however: writing was a powerful technology, and his people needed it. So in 1896 Njoya set out to invent a writing system for his language, Shü-mom, gathering together his best thinkers and best artists to help him.
The job he faced was not an easy one. His advisors were bright, but none of them had any prior experience with writing, and so none knew how the technology worked. What should Njoya write? What aspects of the Shü-mom language should be recorded?
Could he perhaps bypass the words of language and just record the thoughts he wanted to convey? When European scholars first encountered Egyptian hieroglyphs they thought the elaborate drawings represented pure thought. They believed that the hieroglyphic signs were ideograms – symbols that stood for ideas, not specific words. This misunderstanding set the decipherment of Egyptian hieroglyphs back considerably. The ideogram hypothesis was more than just a bad guess for Egyptian, however. As it turns out, a full writing system that bypasses the encryption process of language is not possible. In other words, information separate from language is not the place to begin writing.
Rudimentary systems of such a type do exist. A road sign that shows a car skidding will convey its meaning whether you say to yourself, “Slippery when wet,” or, “Watch out, you might skid,” as you “read” it. Similarly, mathematical symbols and equations convey a meaning that can be expressed in any one of many languages, or even several ways within a language. What is essential in an expression such as ∫ dx/(a + bx2)2 is not what it sounds like in English words, but what mathematical operation it refers to.
The graphical systems of road signs and mathematics work because they apply to a very limited part of human communication. By contrast, one of the essential properties of human language is the infinite range of what can be communicated using only a finite number of basic words. If we could distill human thoughts into a finite number of concepts that could be written down, could we resist giving them names – words? No. We would “read” the symbols by pronouncing them as words. Written symbols cannot systematically bypass language.
So King Njoya’s writing system had to encode language. But this did not make the problem much easier. The system of encoding and communicating information that we call language has many layers. Which layer or layers should Njoya make symbols for?
The most obvious layer of language is its words. However, to make a truly different symbol for each word of a language would result in far too many symbols. To take an example from English, the 160,000 entries of the second edition of Webster’s New World College Dictionary would require 160,000 different symbols. But the number of entries in a dictionary actually underestimates the number of words in a language. For example, the entry for girlish also mentions girlishly and girlishness– both words of English, but not given their own entries. It would be silly, though, to try to create a writing system that had one symbol for girl, an entirely different one for girlish, and another completely different one for girlishness. The words girl, girlish, and girlishness have pieces in common. They all contain the piece girl, while girlish and girlishness share -ish as well. The -ness of girlishness is also a piece that recurs over and over in English. These
pieces of words are called morphemes. There are far fewer morphemes in a language than words, and the morphemes can be combined and recombined in so many ways that it is hard to say how many words a language actually has. It is not surprising, therefore, that no one has ever managed to create a usable writing system that uses full words rather than morphemes as its level of encoding.
A morpheme has two aspects, its meaning and its pronunciation. Writing systems that concentrate on representing morphemes – as complete meaning–pronunciation complexes – are called logographic (the name, meaning “word-writing,” is traditional, though it ignores the difference between morphemes and words), and the individual symbols are called logograms, as shown in figure 1.1 Although those of us who have been trained to use an alphabet find it natural to divide words up into individual vowels and consonants (in other words, separating meaning from its pronunciation and representing only pronunciation), the first inventors of scripts did not. For them it was more natural to consider the morphemes as a whole. Core morphemes at least (those like girl, rather than -ish or -ness) can be uttered on their own in many languages and thus are natural units in which to think of language.
The first version of King Njoya’s writing system was therefore logographic. He compiled a list of little schematized pictures that could stand for individual morphemes. After a while he had 465 of them. A symbol for every morpheme in the language was clearly going to take a lot more than that. And so he was forced to a decision that all complete writing systems have had to make in some form or another: he was obliged to begin using symbols to represent pronunciation.
The pronunciation (or phonology) of language also has several layers. Words are made of one or more morphemes, but they are also made of one or more syllables in the way they are pronounced. A word like cat has a single morpheme and a single syllable, but a word like undesirable contains three morphemes and five syllables. Thus a logo-graphic writing system would give cat one symbol and undesirable three, while a syllabary would give cat one and undesirable five. That lengthens the spelling of undesirable, but lessens the number of symbols needed in all, as there are fewer distinct syllables in a language than there are distinct morphemes.
Figure 1.1 How different writing systems represent language. Logograms represent morphemes, both their meaning and pronunciation, while syllabaries and alphabets represent only pronunciation. In the column of examples, the word undesirable is used to illustrate how the various writing systems would divide up such a word. A morphemic (logographic) system would use three symbols, a syllabary five, and so forth. In an akvara system, the vowels are written as appendages to the consonants.
So King Njoya converted a number of his symbols into syllabograms, standing for syllables – just a pronunciation, unconnected to any meaning. The meaning would come only when the syllabograms were put together to make up words. He worked on his script over a period of many years, ending with a syllabary of 73 signs, plus 10 numerals. He put the writing system to good use, compiling a law code, designing a calendar, and founding schools.
Other ways of writing were theoretically open to him. Syllabaries come in different kinds. Most represent only core syllables (a single consonant + short vowel sequence) and find a variety of workarounds to represent other kinds of syllables. A few include symbols for closed syllables (those that contain a final consonant), and a few writing systems of the world split the syllable in two, representing the consonant(s) at the beginning (the onset) with one symbol and the rest (the rhyme) with another.
More familiar to Westerners is the kind of writing system that ignores syllables entirely and looks at the individual sounds out of which syllables are made. This requires knowing what counts as an “individual sound.” Consider for a moment the words feel and leaf. They appear to contain the same sounds, in reverse order. However, if you say the two words slowly, and pay close attention to your tongue as you say the l s, you may notice that the l in feel has the back of the tongue pulled back and upward compared to the l in leaf. Chances are, however, that you have never noticed it before. Similarly, the p in spoof is pronounced quite differently from the p in poof – you can blow out a candle by pronouncing the latter but not the former.
There are many such variations in sound that native speakers of a language disregard and typically have lost the ability to hear unless they have had training in phonetics. Native speakers of a given language will consider an entire range of sounds to be the “same.” That “same sound” that native speakers perceive is called by linguists a phoneme of that language. The actual sounds of language are infinitely varied, as they are uttered by different people in different circumstances. It would be pointless to try to capture this variation in writing. But most languages have between 20 and 37 phonemes, and phonemes can be written down. An alphabet that is strictly phonemic would have the same number of letters as phonemes (though English does not).
Technicalities aside, an important point here about these abstract phonemes and syllables is that although writing represents information about how words are pronounced, it does not record the identifying details of any individual utterance of those words. It records language, but not actual speech. Even in cases of dictation or courtroom stenography, much information about the actual speech is lost, such as intonation and emotional content. As a result, reading is not at all the same as listening to a recording (and can therefore, fortunately, proceed much faster).
Writing systems that represent individual phonemes are called alphabets. It is therefore inaccurate to refer to the “Chinese alphabet” or the “Japanese alphabet,” as these writing systems do not work at the phonemic level. A further level of distinction separates alphabets into those that represent only or primarily consonants (consonantal alphabets, also known as abjads), those that represent vowels as somehow dependent on the preceding consonant (akara systems or alphasyllabaries), and those that give vowels and consonants equal status (“true” or voweled alphabets).
All writing systems find themselves somewhere in the range from morphemic to phonemic (see figure 1.1). The more morphemic writing systems may also do a little to directly represent the semantic aspect of a morpheme in the form of clues to meaning known as determinatives (thus the symbol for “cat” might include a symbol showing that it is an animal). But no writing system is so completely morphemic that it pays no attention to the phonology (syllables and/or phonemes) of the language. Some scripts are fully phonological, representing either the phonemes or the syllables of the language. On the other hand, no written language is simply a record of uttered sounds: that is left up to a less significant invention, the phonograph, and its modern descendants.
The earliest writing systems were, like King Njoya’s first efforts, all highly logographic. Later writing systems are typically more phonologically based and use far fewer logograms. This is not to say that logographically based scripts are primitive. Logograms have the advantage of using space very efficiently, needing only one sign per morpheme, where alphabets need several. They are also more convenient in contexts where pronunciation varies significantly, making phonologically based writing hard to standardize. Yet an alphabet, with its limited number of signs, is the more easily memorized and can therefore spread faster in a context of limited schooling. What kind of writing system a language uses is largely determined by the accidents of history and by the properties of the language itself.
King Njoya’s labors had a sad ending. The French colonial forces burned his books and exiled him from Foumban in 1931. Today, despite Cameroon independence, his writing system is nearly forgotten. His grandson, the present king, sponsors classes in it at the royal palace in Foumban, but it sees very little actual use. Other scripts have been luckier. Born into more propitious time they have enjoyed a more extensive history. The following chapters tell their stories.
First, in chapters 2 through 5 are the stories of ancient logographic systems – Mesopotamian cuneiform, Egyptian hieroglyphs, Chinese characters, and Maya g
lyphs – along with their syllabic or consonantal compromises. Next, chapters 6 through 8 tell of syllabaries, from the Bronze-Age Linear B used for Greek, through the two Japanese syllabaries, to the modern invention of the Cherokee script. Phonemic scripts follow, with consonantal alphabets, akvara systems, and voweled alphabets in turn. In the final chapter the effects of secondary writing technologies – printing, typing, word processing, and the Internet – are considered, along with the globalization of the Roman alphabet.
A book about writing systems faces one significant obstacle: transliteration. The phonemes, syllables, and morphemes recorded by the world’s writing systems cannot all be recast into the Roman alphabet in a single, unambiguous way. The languages of the world contain some 600 distinct consonants and 200 different vowels. Not all of these have yet been converted into writing, but clearly there needs to be a way to translate the scripts we do not know into one we recognize, so that we know what they say. Many languages already have established ways of being transliterated into the Roman alphabet or use the Roman alphabet themselves. For many transliterated languages the general operative principle is “consonants as in English, vowels as in Italian.” Such a system glosses over a lot, as there are only so many consonants in English and so many vowels in Italian. Furthermore, languages that already use the Roman alphabet do so in many different, mutually incompatible ways.
Therefore I will use standard spellings and transliteration systems where their meaning is clear, but will supplement them where necessary with the International Phonetic Alphabet (IPA). The IPA is designed to represent all the phonemes of human languages. By transcribing an alphabet into the IPA we can tell what phonemes that alphabet encodes: it is a sort of decoder ring for alphabets.
The IPA is reprinted in the appendix (figure A.1). Examples of English phonemes transcribed into the IPA are given in figure A.2. When using IPA symbols to describe a pronunciation, I will write them between square brackets. This is to emphasize that what is being referred to is a sound, not a letter of the Roman alphabet. Thus b is a letter, but [b] is a sound. In many cases the IPA symbol represents the same sound that the Roman letter does in English, but this is not always so, especially in the case of vowels, where the symbols taken from the Roman alphabet generally have the sound values that they do in most continental European languages (such as Italian).
The Writing Revolution Page 2