Evaluating meritthe evolution of writing reconsidered

Florian Coulmas

Abstract
This article examines the question whether the merits of a writing system have a bearing on its spreading. While it is well known that the diffusion of writing systems and scripts follows that of religions, empires and other non-linguistic factors, the adoption and reform of a writing system or orthography is invariably promoted on the grounds of systematic merits. Since writing is an artefact it stands to reason that some systems are superior to others. Yet, it is difficult to define an absolute standard of goodness against which different writing systems could be measured and compared. Instead this article raises the question why, if it is true that in the long run quality prevails, after 5000 years of writing a universal visual code for language has not won world-wide recognition. Discussing a variety of examples it offers six partial answers.

Correspondence: Florian Coulmas, Jochi Kioizaka Bldg., 7-1 Kioicho, Chiyoda-ku, Tokyo 102-0094, Japan. E-mail: [email protected]

1 Introduction
Writing has been called the technology of the intellect by Jack Goody (1968). Writing systems are specific instances of the technology of writing. If this is more than a felicitous metaphor, this technology should be expected to behave, in some ways at least, like other technologies. In particular, the spread and improvement of writing systems should conform to technological innovation in general and hence allow for the analytic tools to be applied that have been developed for the study of innovation. These tools are found for the most part in economics. Two basic assumptions are as follows: (1) collective behaviour can be explained in terms of choices by individuals, and (2) Competition in markets will lead to the replacement of inferior by superior technologies, where superior is understood as having greater utility. This article discusses the questions of how writing systems can be evaluated and whether and to what extent their utility has a bearing on their diffusion. To this end, another notation system, that of music, is referred for comparison. Chapter VI of Ignace J. Gelbs famous book A Study of Writing is

entitled The Evolution of Writing. In it, Gelb presents his view of the history of writing, which he saw as proceeding in successive stages of development. His ambitious project was to develop a coherent theory of evolution valid for all writing systems. He speaks of the necessary steps of a unidirectional development, declaring that writing must pass through the stages of logography, syllabography, and alphabetography in this, and no other, order (Gelb, 1963, p. 201). This sequence of stages reflects what Gelb in the diction of his time called the stages of primitive psychology [], three great steps by which writing evolved from the primitive stages to a full alphabet (p. 203). These remarks, among others, are indicative of Gelbs strong evolutionist beliefs, which informed his entire approach to the history of writing. The idea of progress guided by quasi-natural laws was the unquestioned premise on which he built his theory. Accordingly, the differences between writing systems could be mapped onto a scale of developmental advance. The alphabet, in this view, was the conclusion of a goal-directed, logically necessary progression and improvement.
5

Writing Systems Research, Vol. 1, No. 1, 2009. The Author 2009. Published by Oxford University Press. For Permissions, please email: [email protected] doi:10.1093/wsr/wsp001

F. Coulmas

Gelb tied the evolution of writing systems to social progress, specifically the spread of literacy. In tune with Diringers (1968) notion of theocratic versus democratic writing systems, he saw writing as a force of social change. This is clearly in evidence in his comments about Chinese writing, which he characterized as a type of writing which perfectly suited the needs of a small bureaucratic clique, but was totally inaccessible to 90 per cent of the population (Gelb, 1963, p. 203). He thus established a conceptual link between writing system and mass literacy: the further removed a writing system from perfection, the less accessible it is to the population. There are a number of implicit assumptions to this argument, which are of interest for the question at issue here; i.e. the question of how the relative merits of the writing systems can be assessed. These assumptions are: (1) The evolution of writing is driven by utility maximization. (2) Utility is measured in terms of accessibility to the population. (3) The spread of literacy is desirable. (4) The spread of literacy depends on structural features of writing systems. I want to discuss these assumptions with regards to the bearing they have on the question of the relative merits of writing systems. Gelbs evolutionist perspective presupposes improvement over time, and hence an evaluation to the effect that alphabetic writing is superior to syllabic writing, which in turn is superior to logographic writing. More generally put, the universe of writing can be described not just as an unordered set of diverse systems, but as a hierarchy of merit. Some writing systems are good, others are poor; some are better than others. It is very common, especially among non-specialists, to regard the Greek alphabet as an evolutionary breakthrough and a major improvement over extant writing systems. However, while the differential quality of writing systems is very intuitive and seems to be borne out by much empirical evidence, it is an entirely different question whether this qualitative variety is a reflection of evolutionary progress, as some of the most eminent scholars in the field of writing systems research have argued (Cohen, 1958; Gelb, 1963). Goody (1968, p. 19) speaks of the technological
6 Writing Systems Research, Vol. 1, No. 1, 2009

restrictions imposed by non-phonetic systems of writing, where the sheer difficulties of learning the skill mean that it can be available only to a limited number of people. This is the most intuitive connection between writing system and social organization, which especially in the Western world is considered as common knowledge.

2 Notation Improvement
Writing is a technology, which means that it is an artefact. It stands to reason, therefore, that it can be evaluated for its utility. In this regard, writing resembles numerals and musical notation systems. There can be no denying that Indian numerals, which are often called Arabic, are highly efficient and thus superior to both Roman and Chinese numerals. They are more efficient for writing and for reading, as they take up less space, and are most importantly better for making calculations. This is due to two structural innovations: (1) the invention of the zero and (2) the place value system according to which each number sign changes its meaning (value) depending on its place in a sequence of number signs. The first 1 in 11 means ten, while the second one means one. In contrast, Roman and Chinese numerals have the same meaning in every context (Fig. 1). These two innovations not just facilitate making calculations, but make higher order calculations possible (Ifrah, 1999). That is why Arabic numerals became the general standard used throughout the world. Musical concepts, that is notes, pitches and durations, too, have been made subject to visible notations such as the standard five-line notation on a diatonic staff (Fig. 2) spanning an octave. In the

Fig. 1 The superiority of Arabic numerals

Evaluating writing

Fig. 2 Five-line diatonic staff notation, C major

diatonic pattern of notes, only seven notes have their own line or space on the staff, which is based on the key of C major, leaving five notes that must be modified with sharp or flat signs. This pattern has been found inconvenient by many and, accordingly, alternative music notations have been proposed. The not-for-profit Music Notation Project organization, which explores systems that seek to improve upon traditional music notation, lists a number of criteria for superior music notations, some of which are quoted below. Desirable criteria for alternative music notation systems according to the Music Notation Project (MNP, Internet, 2009) are as follows: (1) The notation is convenient for a human writer (as contrasted with a machine) to express musical ideas. The notation is convenient for a human performer to recreate those musical ideas. (2) The notation can be written conveniently and quickly with nothing more than a writing tool (such as a pencil) without the absolute necessity of a ruler or other drawing aids or specially prepared paper. In other words, a plain piece of paper and a pencil or a chalkboard and chalk should be sufcient for quickly notating music in the notation if desired. (3) The notation is independent of all musical instruments for intelligibility so that the notation is readily adaptable to all instruments including the human voice. (4) The notation can express music of all reasonable degrees of complexitynot only simple music. (5) The notation is exible enough so as to be appropriate for the music of the past, present, and foreseeable future, as well as to music of various cultures, and to both solo and ensemble performance.

(6) The notation is writable using only a single colour on a contrasting background (e.g. black on white) without shading or tinting. Such a monochrome system offers the maximum in simplicity and convenience, and is considered essential, especially since many people have some degrees of colour blindness.

3 Writing System, Script, and Orthography

Can comparable criteria be devised for writing systems? Since writing is a technology and a notation, it ought to be possible to define criteria of merit for writing, too. Writing shares with musical notation that it is a symbolic system based on the coordination of hand, eye, and brain. However, language differs in characteristic ways from music, which implies different conditions for visualization and the evaluation of the two kinds of notation. With regard to writing, it is useful to make a threefold distinction between writing system, script, and orthography. The conceptual distinction was present already in antiquity. In the ancient Near East, for example, Cuneiform and Egyptian were different, both in terms of script and the underlying writing system. But as Cuneiform was interpreted first in terms of Sumerian and then Akkadian and other languages, different writing systems evolved using the same script. The monumental rock inscription of King Darius (522486) in Behistum, Iran, dating from the early fifth-century BCE, is in three languages, Old Persian, Babylonian, and Elamite, all using the cuneiform script according to different principles resulting in different writing systems. By the same token, Egyptian scribes developed three different
Writing Systems Research, Vol. 1, No. 1, 2009 7

F. Coulmas

scripts to give expression to the same writing system: hieroglyphs, a cursive variety derived thereof called hieratic, and an even more cursive form called demotic. Hieroglyphs and demotic are displayed on what is probably the most famous inscription of the world, the Rosetta Stone. (The third inscription on the stone is in Greek.) The Chinese also developed several different scripts for the same underlying writing system, characterized by differing degrees of cursivity. Most commonly, five major styles of writing are distinguished: seal script, clerical script, standard script, cursive script, and swift script. Differences between these styles refer to the script. However, there are potential implications on the level of the writing system. As a general pattern, Chinese characters consist of two elements, semantic and phonetic. In character reform schemes, standard forms have sometimes been replaced by cursive forms of these elements, some of which are less distinct than the standard forms. Thus, the delicate balance between the semantic and phonetic information load of characters is potentially affected. In the second round character simplification promulgated in China in 1977, too little attention was paid to this point, which is one of the reasons why it was eventually nullified by the authorities (Bkset, 2006). As this example shows, while writing system and script are conceptually clearly distinct at any point in time, they may interact over time. The notion of a codified norman orthographyis different from both script and writing system, but sometimes confusion arises when the same script is used for different languages that necessarily have different orthographies, orthographies that may or may not correspond to different writing systems. Do the written forms of English, French, Spanish, Danish, Polish, German, etc. represent different writing systems, different orthographies, or different scripts? Arguably, English and Spanish represent different writing systems, becausequite apart from the fact that the latter has some diacritics which the former lacksthe underlying rules for interpreting the graphemes are very different in the two languages. On occasion, rival orthographies exist for the same language, as for example German in Germany and Switzerland and German
8 Writing Systems Research, Vol. 1, No. 1, 2009

before and after the 1996 spelling reform. American versus British orthography is another example. Divergent normative standards also exist for Chinese, simplified characters being used in China, Singapore and Malaysia, while traditional complex characters continue to be employed in Taiwan and Hong Kong. From a European point of view, where the Roman alphabet is dominant, the multiplicity of coexisting writing systems and scripts is often overlooked or, informed by Gelbs evolutionism, regarded as an antiquated relic. Europeans and, more so, their American offspring are no strangers to the idea of Western supremacy that has been linked more than once to alphabetic writing. This idea has a long tradition reaching back to enlightenment thinkers such as Samuel Johnson who called East Asians, including the Chinese, barbarians because they had no alphabet (Boswell, 1969, p. 984), and Hegel (1970, p. 274) who opined that the alphabet of and in itself is the most intelligent writing system. It was further promoted more recently by linguists and literacy scholars (Saussure, 1959; Havelock, 2003). Anthropologist Goody (1968), too, supported the idea of a non-arbitrary link between the complexity of writing systems, literacy levels, and the social systems their users established. Although the idea that, because of its relative simplicity, the Greek alphabet is superior to previous writing systems has been called into question occasionally (Harris, 1986; Grosswiler, 2004), it continues to enjoy currency. Moreover, there is a widespread conviction that, in combination, market principles and human rationality determine the dissemination of good as opposed to poor technologies. Proponents of Rational Choice Theory are convinced that in free markets quality prevails (Olson, 1965). This is debatable, but the inverse conclusion that what prevails is of best quality is simplistic and sometimes clearly wrong. The fact that the Roman alphabet is the most widely used script on earth cannot be taken as evidence that it is the best system. The spread of a technology is driven by many factors in addition to utility, such as power (market dominance) and ignorance (limited information), which have been taken into account in recent decades in further developing Rational Choice Theory (Kahneman et al., 1991). This clearly holds

Evaluating writing

for the dissemination of writing systems which, it is well known, has followed the conquest of lands and minds through military expansion and proselytism (Coulmas, 2003, p. 201). Moreover, the Roman alphabet is in reality two distinct things, and one should not be confused with the other. As the alphabet of Latin, it is the writing system of Latin; whereas, when interpreted for other languages, it only furnishes the notation for different systems that may or may not resemble that of Latin. As suggested by Harris (2000), the letters of the Roman alphabet should therefore be regarded as elements not of a writing system but of a notation. This notation can be put to use to various ends including purposes unrelated to writing in the proper sense of the word. For example, the nomenclature of chemistry makes use of the Roman alphabet, and it is a matter of course that the human genome is depicted as a sequence of Roman letters, ATGC, to be precise, for Adenine, Thymine, Guanine, and Cytosine, respectively. For the code of life to be written in anything but the Roman alphabetArabic letters, Chinese charactersis unthinkable, but, upon reflection, this is an inadvertent expression of the belief in the superior nature of the Roman alphabet which is part of Western suprematism. Another problem is that there are good reasons to doubt that the relative merits of different writing systems are critical factors in their dissemination. Recent studies of script choice (e.g. Eira, 1998; Grivelet, 2001; Unseth, 2005, 2008) suggest that socio-linguistic factors weigh heavier than systemic factors in determining how communities choose their writing systems, scripts, and orthographies.

4.1 Convenience
The notation is convenient for a human writer (as contrasted with a machine) to express musical ideas. The notation is convenient for a human performer to recreate those musical ideas. This criterion can easily be applied to writing in the sense that a good writing system/script/orthography must be convenient for both the reader and the writer. However, in reality, these are two separate criteria, which must be balanced. Maximizing graphic discrimination is good for the reader but cumbersome for the writer, with implications on all three levels of script, system, and spelling conventions. Two wellknown cases are the use of capitalization to mark nouns in German and the use of Chinese characters for content morphemes in Japanese. Where the best definition of a noun is that it is a word written with a capital letter, capital letters are clearly an obstacle for the writer, especially for the learner-writer. However, the reader appreciates capitals as an aid for parsing. This innovation, i.e. the systematic differentiation of upper case and lower case letters, was introduced through the Carolingian reform late in the eighth century CE. It meant the doubling of the inventory of symbols. However, improved legibility evidently outweighed the additional burden for the writer. Similarly, writing Chinese characters is more difficult than syllabic kana, but a text of mixed Chinese characters and kana orthography is easier to read than one written entirely in kana. Parsimony an apparently obvious criterion for the merit of a notationmeans different things to reader and writer, and one must be weighed against the other.

4 Criteria of Merit
In discussions about the selection of a suitable visible form for a language, the question of quality is often touched upon. Writing reforms concerning any of the three levels discussed would have no rationale if it were not for quality improvement. Let us consider then whether the relative merits of writing can be sensibly assessed. To start, we will use the criteria developed for the assessment of musical notation quoted aboven.

4.2 Tools
The notation can be written conveniently and quickly with nothing more than a writing tool (such as a pencil) without the absolute necessity of a ruler or other drawing aids or specially prepared paper. In other words, a plain piece of paper and a pencil or a chalkboard and chalk should be sufficient for quickly notating music in the notation if desired.
Writing Systems Research, Vol. 1, No. 1, 2009 9

F. Coulmas

Talking about pen and paper may seem outdated, but the tools are still essential for writing. As Cardona (1986, p. 30) emphasized, writing is essentially the coordination of hand and brain. Unless writing is done by voice input, which is not really writing, this is still the case, and therefore the criterion applies to writing. In this regard, the technology of writing clearly exhibits improvements over time. Lead pencils are more convenient than quills, and paper is easier both for mass production and to handle than good-quality vellum or papyrus, not to mention clay. At the same time, utility is often counterbalanced by inertia or path dependence that hinders the diffusion of superior tools. The development and adoption of the QWERTY keyboard is an example frequently cited by critics of rational choice theory, because this configuration has survived in spite of the availability of ergonomically and economically superior keyboards (David, 1997). Defenders of the qualities of the QWERTY keyboard have also made themselves heard, arguing that it is unsuitable as a counterexample that falsifies rational choice theory (Liebowitz and Margolis, 1990). What the discussion of the QWERTY keyboard has revealed is that functional utility (convenience) is just one criterion among others that affects the development and diffusion of new technologies. Walker (2000) has stressed the interrelation of technological regimes with institutional, political, and economic commitments which add to the cost of switching from one technology to another and may lead to institutional entrapment. Writing systems are a case in point, being institutionally embedded in the school system and public administration, relating to economic commitments, for instance of the publishing industry, as well as to political commitments, being associated as they are with religious and other ideological traditions. All of these factors interfere with purely instrumental improvements.

4.3 General applicability and linguistic t

The notation is independent of all musical instruments for intelligibility so that the notation is readily adaptable to all instruments including the human voice.
10 Writing Systems Research, Vol. 1, No. 1, 2009

While it is perhaps farfetched to liken a language to a musical instrument, with some adjustments this criterion can be interpreted as follows: a good writing system is independent of all languages for intelligibility so that it is readily adaptable to all languages. This is an interesting criterion, relating as it does to the spread of writing systems as a measure of relative merit often cited as a key advantage of the Graeco-Roman alphabet. Easy transfer from one language to another increases the utility of a writing system, especially in multilingual and foreign language-learning contexts. Egyptian hieroglyphs, although they were in continuous use for 3,000 years, did not spread far beyond their land of origin and were not adapted to other languages. The Semitic consonant alphabet, on the other hand, has been variously adapted to many languages. Accordingly, writing systems can be compared with regard to the diversity of languages adapted to their interpretation. The linguistic interpretation of graphic symbols always began in the context of and for a particular language. Although proto-writing records information in various ways that do not necessarily relate to a language in a systematic way, Damerow (2006) and Gnanadesikan (2009), among others, have stressed that no writing system ever came into existence independent of a particular language. Although it is theoretically conceivable to develop a graphic code whose expressive power matches that of a natural language, this has never happened. Once graphic recording systems acquired a certain degree of complexity, they became associated with language (Coulmas, 2003). Because of this, writing systems have a linguistic fit and are variably suitable for other languages. Japanese kana is simple and elegant for Japanese, but less so for languages with more complex syllable structures such as Russian or English, for that matter (Fig. 3). Kana are interpreted as vowels or as consonant plus vowel syllables. For writing English in kana, one could, of course, get used to interpreting them as consonant plus vowel syllables in some contexts and consonants in others, but the resulting system would not be as elegant and simple as Japanese. As a matter of fact, in the past, English has been taught in Japanese schools through kana. The fact that the pronunciation of many English words

Evaluating writing

Fig. 4 English without vowels

Fig. 3 English in katakana

by Japanese students is noticeably off-target has been attributed to this practice, although it is hard to say whether interference is caused by the phonological system, which does not allow for consonant clusters, or by the syllabic writing system. Similarly, thanks to the lexical root structure of Semitic languages, Semitic alphabets do not need vowel letters. Thus, Modern Hebrew and Arabic do not normally write short vowels, although Hebrew has a system of indicating vowels by diacritical pointing marks. Since this is an obvious difference from the Greek and Roman alphabets that are conventionally considered as the standard by which others are measured, these scripts have been characterized as defective. An early example is from nineteenth-century Hebrew grammar: No system of writing is ever so perfect as to be able to reproduce the sounds of a language in all their various shades, and the writing of the Semites has one striking fundamental defect, viz. that only consonants are written as real letters. (W. Gesenius, quoted from Ravid and Haimowitz, 2006, p. 67) The interpretation of the intended meaning in Hebrew and Arabic writing is quite possible without vocalization. However, because of these systems lack of a feature that is present in the supposedly ideal systems, they are described as defective. Semitists are, of course, quite aware that this is a defect only from the point of view of the Graeco-Roman alphabet and less so if typological considerations are taken into account. The choice of the term defective is indicative of this point of view. For Indo-European languages, Semitic alphabets are not as fitting as for Semitic languages (Fig. 4),

and even less so for very vowel-rich languages such as Hawaiian and other Polynesian languages. For Italian too, a language that features word patterns such as cvvcv (e.g. baule trunk) and even vvvcv (e.g. aiuola flowerbed), an alphabet without vowel letters would be rather awkward. The Chinese writing system works well for Chinese and can be made to work for English, which like Chinese has a very simple morphology. But it would be unwieldy for a language such as Latin, which expresses grammatical relationships by means of inflexions. Chinese has been used to write typologically very different languages, notably Korean and Japanese, but in the course of time characters were modified, reinterpreted, and supplemented to the effect that very different systems emerged. In modern times, thanks to the commercial military religious complex of European imperialism in the nineteenth century and American predominance in the twentieth, Roman is most frequently used for providing hitherto unwritten languages with a written form. Did the systemic characteristics of this system also play a role in its diffusion? To put it differently, is Roman less language-bound than other systems, and is the fact that it has been applied to many languages due to its systemic superiority? If we consider the various augmentations and modifications of Roman, its multiple adaptations are the result rather than the precondition of being put to use for many languages. This is apparent once languages outside the IndoEuropean family are taken into account. For instance, Roman is rather ill-suited for tone languages (Voorhoeve, 1963). The example of Vietnamese is instructive. It not only has many vowels, but also is a tone language incorporating an entire dimension of phonetic differentiation that is absent from the Semitic languages for which the alphabet was first developed, as well as from Greek for which some Phoenician letters were reinterpreted as vowels. The point at issue here is not just that in Roman Vietnamese every vowel letter appears in multiple forms, but the representation of tone as such (Fig. 5).
Writing Systems Research, Vol. 1, No. 1, 2009 11

F. Coulmas

Fig. 5 Vietnamese

Bassetti (2006, p. 96) has argued and presented convincing empirical evidence to the effect that different types of writing systems (alphabetic, consonantal, etc.) provide different models for the analysis of language, segmenting language in different units and making these units apparent to their readers. This is a most important observation. Linell (1982) and Coulmas (2003) have argued that this is also true for linguists, who base their analyses on writinginduced concepts. Faber (1992) has made a strong case that segmental phonology is an epiphenomenon of alphabetic writing, an argument which runs counter to conventional knowledge, according to which alphabetic letters represent entities that actually exist. This raises the question of whether the visual form of a language makes linguistic units apparent or imposes them on the speech continuum. The representation of tone calls for reflection on this issue. In alphabetic transcription tone appears to be detachable from the sequence of letters that supposedly represent the flow of speech, like some sort of additional filigree put on a different line. This is like saying a consonant is a syllable without a vowel a possible analysis to be sure, but clearly an artefact. In speech, there is no toneless syllable and no awareness thereof. The many diacritics that are needed for writing Vietnamese with Roman letters testify to the Semitic origin of the Roman alphabet and to its continued language dependence. Diacritical tone marks in an alphabet illustrate the role of historical accident in technological innovation and diffusion. Had the alphabet come into existence in the context of tone languages, one would not expect tone to be treated as a detachable part of the vowel. The upshot of this discussion is that there is no language independence in the strict sense. When, early in the history of writing, graphical marks were given a linguistic interpretation, all writing systems
12 Writing Systems Research, Vol. 1, No. 1, 2009

developed a linguistic fit. What exactly this means is a question that should be reconsidered in light of the findings quoted above. It is certainly true, to borrow Bassettis words, that different types of writing systems provide different models for the analysis of language; but what is the nature of the relationship between model and that to which it corresponds? Does the model reproduce or impose structure? My understanding is that the medial difference between audible speech and visible writing implies necessarily that both are the case. A world map is a model that maps a sphere onto a plane. Imposing the properties of two-dimensionality on the model is inevitable. Similarly, mapping a structured temporal event onto a spatial expanse cannot result in isomorphism, because the structural properties of shapes arranged on a plane are not the same as those of an event occurring in time. This is evidenced, for example, by the phenomenon of anticipatory coarticulation where the articulatory gestures of a subsequent sound have an effect on those of a preceding sound. What in writing by virtue of the two dimensions of the writing surface appears as a sequence of discrete elements is an overlapping occurrence of gestures in speech. The great variety of writing systems teaches us that different structures can be highlighted. Many languages for which more than one written form has been developedChinese, Vietnamese, Mongolian, Turkish, Rumanian, among others (Grivelet, 2001)and the many writing systems that have been applied to several languagesChinese, Latin, Arabic, Greek, among othersprovide ample material for comparative study to show that writing systems vary with regard to their linguistic fit, which is more or less fine-grained and more or less accurate, but none meets the requirement of being independent of all languages.

4.4 Expressive power

The notation can express music of all reasonable degrees of complexity not only simple music. How can this criterion be applied to writing? We know that language is the most highly complex symbolic system the human mind has developed. Thus, the criterion can be understood to mean that

Evaluating writing

the notation must express language. That amounts to a definition or conceptual division between graphic systems whose expressive power is on a par with natural languages and others that can communicate simple messages but not content of all reasonable degrees of complexity. Property marks and traffic signs, for example, are visible signs that convey messages, but fall short of the expressive power of natural language and are, therefore, not recognized as writing. Writing must have the expressive power of language. Can this be accomplished only by rendering a particular language visible? This is certainly the way writing systems developed over the course of history. Philosophers, such as the great enlightenment thinker Gottfried Wilhelm Leibniz and the mathematician and logician Gottlob Frege, have probed the possibility of a universal writing system, a characteristica universalis, which has the expressive power of language without being tied to a particular language. They have not been successful. No poem or novel has ever been composed in Freges (1879) Begriffsschrift. It would seem, therefore, that a writing system can satisfy this criterion only if it systematically maps onto a language. However, this does not mean that writing is subordinate to speech. The relationship between language, speech, and writing is more complex, for the expressive power of a given language is not only reproduced by a writing system that has a good fit to it, but it is also expanded. With writing, a language gains additional functions. Thus, on the one hand, a writing system maps onto a language to be equivalent with regard to expressive power; on the other hand, by so doing, amplifies it. Analogously, the development of complex instrumental music is intimately linked to that of symbolic music notation. The contrapuntal composition, for example, first fully developed by Bach, would not be possible without chromatic staff notation.

4.5 Simplicity
The notation is relatively simple so as to be practical for both children and adults. The criterion of simplicity has been invoked both in dealing with the history of writing (Gelb, 1963) and writing or spelling reform (Fodor and

Hagge, 198389). It echoes Occams time-tested razor: Entia non sunt multiplicanda praeter necessitatem; the principle that for the explanation of a phenomenon no more entities than necessary should be assumed. In the sense that a writing system is a model of a language, it constitutes a kind of explanation of it. An explanation that makes do with fewer elementary units can be said to be better than one with more elementary units. The small number of basic signs of the Greek alphabet is often cited as its main advantage, while Chinese writing is said to be cumbersome because of the great number of characters. However, when it comes to mastering the system, the number of basic signs is only one of the factors that have a bearing on simplicity. There are at least two aspects to the simplicity of a writing system, ease of learning and ease of use. The former depends partly, but not entirely, on the parsimony of the sign inventory. Simply memorizing the ABC is a long way from becoming literate in English or French, for example. Conversely, Chinese characters are structured. Learning 1,000 or 2,000 characters does not mean committing to memory that many unique signs. The building blocks from which Chinese characters are composed are much fewer in number. There are just eleven basic strokes, from which a certain historically contingent number of radicals are composed. In modern times, 214 radicals are conventionally used for arranging characters in dictionaries. These are used for building all characters and ordering them lexicographically. On the other hand, fluent readers of English need to know several hundred phonemegrapheme combinations. According to one analysis, the forty phonemes of English correspond to in excess of 1,700 graphemes and grapheme combinations (Nyikos, 1988). While statistics of this sort may not do justice to the English spelling system, which maps more onto the morphological level than on the phonological one, it is clear that English includes a large number of word-specific spellings (Van Berkel, 2005). The ratio of transparent rule-governed spellings and word-specific spellings depends on the corpus used for the analysis. Whatever the exact outcome, English is a prime example of a writing system that makes use of the small set of Roman letters to produce a set of phonemegrapheme correspondences
Writing Systems Research, Vol. 1, No. 1, 2009 13

F. Coulmas

so numerous as to be comparable, for example, to the set of 2,000 commonly used Chinese characters in Japanese (j y kanji). Chinese writing has stood the test of time, for while simpler systems such as the Indian scripts, Semitic-derived Sogdian, as well as the Latin alphabet have been known in China for many centuries, Chinese characters have persisted. Cultural tradition comes to bear here, but utility for the user cannot be excluded as an explanatory factor. Chinese texts are short and according to proficient readers, easier to read than texts in Pinyin, romanized Chinese. In addition to the size of the sign inventory, the rules governing their use must be taken into account when simplicity is considered. There is always a trade-off between memory and combination, which is hard to measure. For this reason alone, if not for others, psychological simplicity cannot be measured by anything as straightforward as counting entities.

4.7 Monochrome coding

The notation is writable using only a single colour on a contrasting background (e.g. black on white) without shading or tinting. Such a monochrome system offers the maximum in simplicity and convenience, and is considered essential, especially since many people have some degree of colour blindness. Today colour coding is not a systematic feature of any writing system. It seems that in early Arabic writing, vowels were written with symbols of different colours; and in the Peruvian quipu recording system, the colours of strings are significant. For the rest, however, to the extent that colour has been used in writing, it served ornamental and symbolic rather than referential functions. Yet, this criterion reminds us of the fact that writing is a visible code that crucially depends on the physical capacities of the human eye. Therefore, this last criterion has been taken into consideration in some definitions of writing (e.g. Gelb, 1980, p. 21f.).

4.6 Stability through time

The notation is flexible enough so as to be appropriate for the music of the past, present, and foreseeable future, as well as to music of various cultures, and to both solo and ensemble performance. One of the most consequential differences between language and writing has to do with their localization in time and space. Speech has a temporal extension which writing maps onto a space. Writing freezes and divides into discrete parts what in speech is continuous, overlapping, ephemeral, and in flux. It arrests change. The inevitable consequence is that shifts occur in the relationship between one and the other, a relationship which typically was complex to begin with. The smaller the analytic unit of the writing system, the more the system will be affected by historical change. A 1000-year old Chinese text, therefore, presents less of a problem to the proficient reader than a 1000-year old French text, for example. The historicity of language affects the sustainability of writing systems of different types in different ways. This must be taken into consideration if stability though time is accepted as a criterion of the relative merits of writing systems.
14 Writing Systems Research, Vol. 1, No. 1, 2009

5 Merit
Where does this list of criteria for a good writing system leave us? Two lessons can be drawn from this discussion. First, writing systems are historically organic structures that, typically, have not been designed at the drawing board. Second, writing serves a variety of functions, both instrumental and symbolic. Every community makes a selection of the functions for which writing is employed, and they do not necessarily all make the same selection. Writing is a tool, but not only a tool. As is the case with language, instrumental, communicative, and emblematic functions coexist. Regarding the history of writing, this means that evolution cannot be explained exhaustively in terms of pure instrumental utility and that merit must be conceived as merit for a particular purpose. We have reviewed convenience for writer and reader, production by means of a simple implement, linguistic fit, expressive power, simplicity, stability through time, and monochrome coding. No writing system excels in all categories. The reason is that

Evaluating writing

Fig. 6 Three scripts used for Tuareg: Roman, Tinagh, and Arabic (source: Savage, 2008, p. 18)

every writing system is a historical accomplishment, which means that it is subject to change on one hand, and to cultural attachment on the other. Savage (2008) reports the instructive example of the three scripts used for Tuareg: the authentic Tifinagh, invented and used exclusively by the Tuareg, Arabic, and Roman (Fig. 6). At a conference in Morocco deliberating which of these scripts should be used for writing the related Berber language, Tifinagh, paradoxically, emerged as the neutral solution (Savage, 2008, p. 11), because the defenders of the other two could not reach agreement. This is one of many cases of script choice where utility considerations were subordinate to symbolic ones (Unseth, 2005, 2008; Sebba, 2007). No matter how good a writing system is and how well conceived the norm that it embodies, by virtue of the fact that it is permanently visible, it suggests fixedness and inevitability, a strong precondition for path dependence, that is, the notion that technological change depends on its own past. Hence the resistance writing reforms regularly encounter. There is something extremely conservative about writing systems, not least because of their twin relationships with language, whose structure they both map and constitute in the literate persons mind. This twofold nature of writing is at the bottom of the most potent criticism of an evolutionary theory of writing along the lines laid down by Gelb, as well as the standard of merit derived from it. Such a theory ignores what the technology of writing does to those who use it, ignoring the complexity of the relationship among language, speech, and writing in a social

context. Evolutionism forces us to believe that writing developed in the direction of something it helped to produce; that is, linguistic analysis. Instrumental utilitarianism forces us to believe that maximizing utility is the major force in the development of writing. However, as is true of many technologies, its applications became apparent only after it had been invented, and thus cannot sensibly be conceived as the end to which the invention or development had occurred. Moreover, once a writing system, an orthography and a script have received official blessing and are subject to institutional support, functional criteria are often overruled by inertia and an interest in stability which may block the road towards future improvement for a long time, if not permanently. Writing systems are not just technologies, but also emblems of identity, and as such interfere with the rationality of utility. If the diffusion of writing systems can be analysed in terms of technological innovation, as has been suggested in this article, these systems should therefore be conceived not as objects of individual choice, but rather as public goods that are subject to cultural attachments and political decisions. However, this must be the subject of another paper.

References
Bassetti, B. (2006). Orthographic input and phonological representation in learners of Chinese as a foreign language. Written Language and Literacy, 9(1): 95114. Boswell, J. (1969 [1904]). Life of Johnson. London: Oxford University Press.
Writing Systems Research, Vol. 1, No. 1, 2009 15

F. Coulmas

Bkset, R. (2006). Long Story of Short Forms: The Evolution of Simplied Chinese Characters. Stockholm University: Stockholm East Asian Monographs, No. 11. Cardona, G. R. (1986). Storia universale della scrittura. Milano: Arnoldo Mondadori Editore. Cohen, M. (1958). La grande invention de lcriture et son valution. Paris: Imprimerie Nationale. Coulmas, F. (2003). Writing Systems. An Introduction to their Linguistic Analysis. Cambridge: Cambridge University Press. Damerow, P. (2006). The origins of writing as a problem of historical epistemology. Cuneiform Digital Library Journal, 1. http://cdli.ucla.edu/pubs/cdlj/2006/cdlj2006_ 001.html (accessed 22 January 2009). David, P. A. (1997). Path dependence and the quest for historical economics: one more chorus of the ballad of QWERTY. University of Oxford Discussion Papers in Economic and Social History, No. 20. http://www .nuff.ox.ac.uk/economics/history/paper20/david3.pdf (accessed 22 January 2009). de Saussure, F. (1959). Course in General Linguistics. (Baskin, W., Trans.). New York: The Philosophical Society. Diringer, D. (1968). The Alphabet: A Key to the History of Mankind. New York: Philosophical Library. Eira, C. (1998). Authority and discourse: towards a model for orthography selection. Written Language and Literacy, 1: 171224. Faber, A. (1992). Phonemic segmentation as epiphenomenon: evidence from the history of alphabetic writing. In Downing, P., Lima, S. D. and Noonan, M. (eds), The Linguistics of Literacy. Amsterdam: John Benjamins, pp. 11134. Fodor, I. and Hagge, C. (eds). (198389). La Rforme des langues. Histoire et avenir. Language reform. History and future. Sprachreform. Geschichte und Zukunft. Hamburg: Buske. Frege, G. (1879). Begriffsschrift, eine der Arithmetischen Nachgebildete Formelsprache des Reinen Denkens. Halle a.S.: Verlag von Louis Nebert. Gelb, I. J. (1963 [1952]). A Study of Writing. Chicago and London: University of Chicago Press. Gelb, I. J. (1980). Principles of writing systems within the frame of visual communication. In Kolers, P. A. et al. (eds), Processing of Visible Language 2. New York: Plenum Press, pp. 724.

Gnanadesikan, A. E. (2009). The Writing Revolution. Cuneiform to the Internet. Oxford: Wiley-Blackwell. Goody, J. (ed.) (1968). Introduction. Literacy in Traditional Societies. Grivelet, S. (2001). Introduction. International Journal of the Sociology of Language, 150(special issue): 110. Grosswiler, P. (2004). Dispelling the alphabet effect. Canadian Journal of Communication, 29(2). http:// cjc-online.ca/index.php/journal/article/viewArticle/ 1432/1540. Harris, R. (1986). The Origin of Writing. London: Duckworth. Harris, R. (2000). Rethinking Writing. Bloomington: Indiana University Press. Havelock, E. (2003). The Greek legacy. In Crowley, D. and Heyer, P. (eds), Communication in History: Technology, Culture, Society. Boston: Allyn & Bacon, pp. 538. Hegel, G. W. F. (1970). Enzyklopdie der philosophischen Wissenschaften. Werke in zwanzig Bnden, vol. 4. Frankfurt/M.: Suhrkamp. Ifrah, G. (1999). The Universal History of Numbers: From Prehistory to the Invention of the Computer. New York: Wiley. Kahneman, D., Slovic, P. and Tversky, A. (eds). (1991). Judgment under Uncertainty. Heuristics and Biases. Cambridge: Cambridge University Press. Liebowitz, S. J. and Margolis, S. E. (1990). The fable of the keys. Journal of Law and Economics, 33: 125. Linell, P. (1982). The Written Language Bias in Linguistics. Linkping, Sweden: Linkping University, Department of Communication Studies. MNP. (2009). Desirable criteria for alternative music notation systems. http://musicnotation.org/musicnotations/ criteria.html (accessed 22 January 2009). Nyikos, J. (1988). A linguistic perspective of illiteracy. In Empleton, S. (ed.), The Fourteenth LACUS Forum 1987. Lake Bluff, IL: Linguistic Association of Canada and the United States, pp. 14663. Olson, M. (1965). The Logic of Collective Action. Cambridge, MA: Harvard University Press. Ravid, D. and Haimowitz, S. (2006). The vowel path. Learning about vowel representation. Written Language and Literacy, 9(1): 6793.

16

Writing Systems Research, Vol. 1, No. 1, 2009

Evaluating writing

Savage, A. (2008). Writing Tuaregthe three script options. International Journal of the Sociology of Language, 192: 513. Sebba, M. (2007). Spelling and Society. Cambridge: Cambridge University Press. Unseth, P. (2005). Sociolinguistic parallels between choosing scripts and languages. Written Language and Literacy, 8: 1942. Unseth, P. (2008). The sociolinguistics of script choice: an introduction. International Journal of the Sociology of Language, 192(special issue): 14.

Van Berkel, A. (2005). Learning to spell English as a second language. In Cook, V. and Bassetti, B. (eds), Second Language Writing Systems. Clevedon: Multilingual Matters, pp. 97121. Voorhoeve, J. (1963). Some problems in writing tone. In William A. Smalley (ed.), Orthography Studies. Articles on New Writing Systems. London: United Bible Societies, pp. 127131. Walker, W. B. (2000). Entrapment in large technical systems: institutional commitment and power relations. Research Policy, 29: 83346.

Writing Systems Research, Vol. 1, No. 1, 2009

17