Home > Misc. > Archive

In Defense of English Spelling

Posted by Jew from Jersey
22 March 2007

A friend writes:

Do you have any idea how many countries have spelling bees? Isn’t a spelling bee illogical in most languages, given that in most you generally say what you see? So, how many other languages are as twisted as English that spelling becomes such a challenge and such a hefty requirement leveled on the educational system?
Just as all-rock gardens are unique to Japan and bull fighting and bull runs are confined mainly to the Spanish-speaking world, the spelling bee is a unique cultural institution of the Anglosphere. At first glance, it may appear that this is because no other language except English has a spelling system so opaque as to make the spelling of words at all challenging. A famous example, popularly attributed to George Bernard Shaw, suggests that the English word fish might plausibly have been spelled ghoti: gh as in cough, o as in women and ti as in nation.

Closer analysis will show that English spelling is not really so chaotic as the ghoti example would suggest, nor are the other spelling systems of the languages of the world quite so predictable either. The spelling bee is in fact specifically American in origin, and while it did later become known to various degrees in other English-speaking countries, it is largely viewed as an American import. Much like American football, it is a homegrown cultural tradition. Many European countries, both Eastern and Western, use spelling dictations as classrooms exercises. These are carried out along much the same lines as spelling bees, except that they are a compulsory part of the curriculum, and do not involve cash prizes. Their enduring use as a pedagogical device suggests that English is hardly the only language in which the spelling of words is less than obvious based on their sound.

All writing systems include some degree of symbol-to-sound correspondence. This is true even of Chinese characters, which have been called logographic or even ideographic. Yet no writing system completely reflects its spoken language. The two chief culprits here are natural sound change and the fact that writing systems throughout the world have always taken on a cultural life of their own that encompasses more than just providing a transcript for the spoken word.

Yet, just as all writing systems are condemned to eternally deviate from pronunciation, the desire to reform spelling and bring it closer to speech seems to spring eternal as well. English spelling reform efforts have been ongoing since the 12th century. Other languages that have had robust spelling reform efforts include Armenian, Chinese, Dutch, French, German, Greek, Indonesian, Japanese, Korean, Latvian, Norwegian, Portuguese, Russian, Spanish, Turkish, and Vietnamese. Thus, while English is unusual in having spelling bees, this is not because it is the only language that has a spelling issue. It is simply the only language that has bothered to make it into a national pastime.

The Politics of Spelling Reform

Frank Zappa once said that no change in musical style will survive unless it is accompanied by a change in clothing style. For literate populations, no spelling reform will survive unless it is accompanied by a very popular political movement or an incredibly powerful central government. In either case, it is almost always part of a wider series of social changes.

Korean was written using Chinese characters for most of its history. The more phonetic hangul writing system was introduced by king Sejong in the 15th century. However, it was not widely adopted until the 19th century, coinciding with the rise of Korean nationalism. The result is that Korean has a relatively high correspondence of sound and spelling. However, in the 20th century, spoken Korean has undergone further sound changes that are not represented in its spelling, chiefly involving the pronunciation of plosive consonants at the ends of words.

Following the Russian revolution, the Bolsheviks eliminated the letters “I” and “yat” from the Russian alphabet. They also eliminated the convention of writing the “hard sign” at the ends of words. That’s several million eliminated people per each eliminated letter. They also changed the calendar, which is why the anniversary of the October revolution is celebrated on November 7. The writing system the Bolsheviks took the ax to had itself been introduced two centuries earlier by Tsar Peter I to replace an even older version. That was concurrent with Peter I’s other modernizing efforts, including an edict ordering Russian men to shave their beards.

Shah Reza Khan of Iran also ordered his subjects to shave their beards. This was after he deposed Ahmad Shah Qajar in 1925. Not surprisingly, he also attempted to introduce new “Persified” words into Persian to replace Perisan words of Arabic origin.

Indeed, most changes to writing systems are probably for the most part politically motivated. What is certain is that regardless of motives, the consequences of such changes are always more political than anything else.

The People’s Republic of China introduced simplified Chinese characters in 1956 to replace the traditional characters. It is possible this has been a factor in the increased levels of literacy in China (90.9% according to the United Nations Human Development Index of 2006). However, Taiwan stubbornly sticks to using the traditional characters, and has an even higher literacy rate (97.2% according to its own statistics, which are not reported by the UN since the UN doesn’t recognize Taiwan). What is for certain is that the political tension between Taiwan and mainland China now extends to their writing systems as well as their political and economic systems.

Mustafa Kemal Attaturk’s junking of the Arabic-based Ottoman writing system may or may not have helped Turkish children learn to read, but it sure did make all Ottoman and Islamic documents and culture seem alien and incomprehensible. Serb and Croat intellectuals and politicians have spent the last 15 years feverishly devising new ways to make their writing systems different, after 50 years of complete mutual intelligibility. In central Asia, the Communists changed the Arabic-based writing systems to Cyrillic-based systems in the 1920s. Now, central Asians are junking those systems for wholly new Latin-based systems. Russians are now complaining that this is unnecessary and will lead to inefficiency. They are right, and that is exactly why the central Asians are insisting on it.

It is probably not a coincidence that the lone successful example of widespread spelling reform in English coincided with the largest single political rift in the otherwise politically stable English-speaking world. I refer of course to the introduction of the typical American spellings of words like center and color (vs. British centre and colour). These spellings were introduced in Noah Webster’s first Compendious Dictionary of the English Language, published in 1806 when anti-British sentiment ran high in the newly independent United States. In the intervening 200 years, is there any evidence that American children learn to spell sooner or more easily than British children? Do Americans save time by writing with a more “logical” spelling system compared to the British? In 200 years, the one indisputable accomplishment of this spelling reform has been to accentuate the cultural and political rift in perceptions Americans and Britons have of each other.

Why English Spelling Is So Atrocious

The degree of divergence between spelling and pronunciation is largely a function of a language’s cultural and political history. While spoken language changes constantly in a spontaneous and non-uniform manner, writing systems tend to change only in periods of unusual political volatility. Thus, the degree of sound to spelling correspondence in a language is largely a function of how much time has passed since the language’s literary tradition began, and also of the degree of political turmoil experienced by its speakers. When spelling systems are first codified, they usually have a high degree of sound to spelling correspondence. As time goes by, the spoken language will diverge from the spelling. At the same time, the older the literary tradition, the more social resistance there is to changing the spelling system, precisely because of the weight of the literary tradition that the society risks being alienated from once the spelling system is changed. Political upheaval facilitates spelling reform and often even instigates it, since the existing literary tradition may be seen as part of the political order that is being overthrown.

The degree of sound to spelling correspondence found in a writing system may also be slightly correlated with the level of grammatical inflection of the language. Of course, it is possible to use any writing system to write any spoken language, but as writing systems have developed naturally, there does seem to be some correlation between the degree of inflection in a language and the phonetic detail of the writing system. Chinese has hardly any inflection to speak of (no case marking, no gender or number agreement), which would make it a good candidate for logographic characters. Finnish has a high degree of inflection (16 cases and a full agreement system in all parts of speech), making it a good candidate for a more phonetic system. English is somewhere in between (2 cases, but only marked on pronouns; number and person agreement, but only in verbs).

In sum, considering its middling degree of inflection, and considering its long literary history and relative political stability, the lack of one-to-one sound correspondence in English spelling strikes me as, well, about what you’d expect.

What Do Spelling Reformers Really Want?

While it’s hard to guess the motives behind the many proponents of spelling reform, reform proposals usually follow a single principle that is seen to be the ultimate goal of the writing system. The English spelling reforms of the late Middle Ages focused on making English spelling as similar as possible to Latin spelling. While the utility of this seems questionable today, modern-day spelling reformers have taken up the cause of one-to-one spelling correspondence with a similar zeal.

And just as spelling reform is usually only successful in periods of political upheaval, spelling reformers (whether successful or not) are often also proponents of other wide-sweeping social changes. George Bernard Shaw, who was one of the most dedicated advocates for spelling reform in English, was also a radical socialist. Richard Feynman, while hardly a long-term crusader for spelling reform, seems to have viewed it as part of scientific development, along the lines of electricity, running water, penicillin, etc. He seems to have viewed scholars of the humanities as derelict in this respect, considering the great strides wrought by natural scientists. Isaac Asimov took this train of thought further and suggested English grammar should be made more scientific as well. For instance, he suggested the plural of man should be mans and not men.

The reasoning here is somewhat reminiscent of crusades that were popular in the United States in the 1950s like the campaign against breast-feeding and the quest for “the all electric kitchen”. Such schemes are religious in nature, even if the higher principle all else is sacrificed to is said to be science, progress, logic, modernism, or the like. Science is an epistemological system. It is not a decision-making system or a value system. No matter how rational the method of the scientist whose work makes the electric stove possible, there is nothing rational per se about throwing out a gas or wood-burning stove just to be able to replace it with an electric one. Rational decision-making involves a cost and benefits analysis and even then, may often devolve onto a subjective valuation. Is the benefit of not having to pay a separate gas bill more important to you than having to wait several minutes for your burner to heat up and cool down?

The yearning for spelling reform usually avoids cost and benefit analysis altogether and posits one-to-one spelling correspondence as the trump card that outweighs all other considerations. It’s like someone suggesting leveling a mountain or draining a swamp for no reason at all besides that the landscape would “look more scientific that way”. Shaw was so obsessed with replacing English spelling with his own proposed 42-letter alphabet that he argued for its implementation by any means necessary, even civil war:

“The waste of war is negligible in comparison to the daily waste of trying to communicate with one another in English through an alphabet with sixteen letters missing. That must be remedied, come what may.” (Preface to R.A. Wilson, The Miraculous Birth of Language, 1948)
Of course, electric stoves and artificial milk are a matter of personal choice, no matter how much or little they are in vogue at any given time. Even institutional reforms like the British changes from their old monetary system of shillings and pence are fairly simple and relatively harmless, since pounds sterling are only issued by the central bank anyway. Changing spelling conventions, on the other hand, requires a far more pervasive use of power. While this may be simple oversight on the part of some spelling reformers, for others it may be precisely the point.

The spelling changes instituted in Russian after the revolution, rational as they may have been, had the effect of making all pre-revolutionary texts difficult to read for all future generations. Of course, old texts could be transliterated into the new Russian spelling format, but this would require special effort by specialists. As it turned out, a society that was centralized enough to control spelling conventions was also centralized enough to control the specialists and to control the printing presses. Thus, even old texts that were not destroyed were effectively rendered inaccessible to all but the most dedicated modern readers. This has the effect of resetting history, as the French revolutionaries tried to do by creating a new calendar year zero. Robspierre didn’t change French spelling, but he did redraw the boundaries of the French provinces and even tried to replace the 7-day week with a 10-day week. There was a movement in Israel in the 1990s to use the Latin alphabet to write Hebrew. Its proponents were the same kinds of people who wanted to outlaw circumcision and kosher slaughter.

One of the objections to the latinizing of Hebrew spelling was that it would sever future generations of Hebrew speakers from the age-old religious texts. The response of the would-be latinizers was that this was a small price to pay for “progress”. It then becomes clear that what they mean by “progress” is precisely the elimination all religious and traditional influence from Hebrew culture, in the same way that the motivation for Robspierre’s 10-day week was the elimination of Sunday mass.

Advantages of Spelling Reform

Shaw lived out his life believing that English spelling reform would have great practical benefits that outweighed any costs they might incur (he had similarly unwavering lifelong faith in the leadership and wisdom of Joseph Stalin). But what are the benefits of one-to-one spelling correspondence?

There is some argument to be made that lack of sound-to-spelling correspondence makes literacy more difficult. The case of Chinese is often brought up. However, what really matters is whether China’s illiteracy is really higher once other factors (poverty, available schools, etc.) have been adjusted for. I would argue that Japan is an informative test case, especially when compared with China. In order to be literate, Japanese speakers must master two separate alphabets (or syllabaries, as it were), hiregana for old Japanese words, and katakana for recent loanwords. Furthermore, the large numbers of Chinese loanwords in Japanese are written in their original Chinese characters. So Japanese speakers also memorize in school the 2000 or so most common Chinese characters, the same ones that are supposed to be the cause of illiteracy in China. And yet, as per the UN HDI 2006, Japan’s literacy rate (99.0%) is higher than China’s (90.9%).

It is perhaps worth looking into the UN HDI data a little more closely. The Human Development Index is computed from a number of statistics, including literacy, per capita GDP, life expectancy, and educational enrollment. Literacy data is often unavailable, and even when it is available, it is computed differently by different countries. For the purpose of computing HDI, the UN uses the rate of 99.0% literacy for countries that are thought to have high literacy levels. A total of 42 out of 177 countries were treated as having the 99.0% rate. Not surprisingly, most of these countries also scored very well on the other HDI statistics. It should probably also be noted that these 42 countries included all countries in Europe, most countries in the former Soviet Union, and no countries in South America or Africa. Thus, it would seem that whatever effect language and writing systems have on literacy, they are far outweighed by developmental and regional factors.

The writing system of Spanish is often compared favorably with that of English as being more logical, rational, easier to learn, etc. Presumably, all Spanish-speaking countries use a similar writing system. Yet only two Spanish-speaking countries score among the top 42 for literacy: Spain and Cuba. Yet this same list of 42 countries includes all four major English-speaking countries: Australia, Canada, New Zealand, and the United States. This is not to say that the Spanish writing system is not superior to the English writing system, just that this superiority is of limited benefit as far as literacy is concerned.

The entire question of how to measure literacy is of course controversial and the UN data should be taken with a grain of salt. Many have suggested alternative definitions of “functional literacy” that are more stringent than those used by most countries and included in the UN HDI statistics. But however literacy is defined, and however much the actual numbers change, there is little reason to assume the general inter-country rankings will change much relative to each other.

The ranking of writing systems according to their complexity or ease of use is also less than clear. I’ve assumed here that the Japanese writing system is more complex than the Chinese, that the Chinese is more complex than the English, and that the English is more complex than the Spanish. I hope these are relatively uncontroversial claims, but they would be impossible to quantify. There is no objective standard for how complex a writing system is. Even if spelling reform would make English easier to learn, it would be almost impossible to measure what would really qualify as a simpler system.

For instance, I have been assuming that the Chinese writing system is complex because it has so many unique characters. There are probably at least several thousand commonly used characters, although some dictionaries list over 10,000. I think this is what most people mean when they blame the complexity of the writing system for China’s historically low literacy. However, you could argue that Chinese spelling is easy since nearly each word corresponds to a unique symbol or pair of symbols. In that sense, it is actually closer to being one-to-one than English.

No one even knows how to quantify the complexity of spoken languages. The Universal Grammar hypothesis currently in vogue among American linguists says that all natural languages (including sign languages) are generated by the same module in the brain. So you wouldn’t expect big differences in complexity among languages. This has absolutely nothing to say about writing systems. However, even in the case of spoken languages, it doesn’t guarantee that all languages have exactly the same complexity, not that anyone would know how to quantify that anyway. The Russian humorists Ilf and Petrov claimed there is a cannibal language called “Mimba Lyumba” that has only 20 words. Any linguist today would tell you this is nonsense. However, I don’t know any who would commit to saying that all languages are exactly equally complex. I know only one, the esteemed Roger Higgins, emeritus of the University of Massachusetts, who has said that he thinks some languages are more complex than others. I asked him if he didn’t just mean subsets of a language, like the verb system or vowel system. He said no, he meant there were significant differences in the complexity of entire languages. I don’t think he meant to say he could quantify this, just that as someone who’s studied these things more than most, that was his distinct impression.

It is probably impossible to deny that some writing systems are better than English in some way, but it is almost equally impossible to measure to what extent they are better. This problem is complicated further by the equally difficult to measure differences between the languages themselves, regardless of their writing systems. Finally, whatever advantages are posed by some writing systems over others, there is little evidence to suggest that these advantages provide much practical benefit.

Disadvantages of Spelling Reform

Henry Rogers of the University of Toronto, who is generally sympathetic to spelling reform in English, estimates that after a reformed spelling system is introduced in English: “Most people, certainly university students, would have to learn to read both systems for at least 50-75 years; for scholars, knowledge of both systems would be required for much longer.” If it is true that the English spelling system is a burden, then surely spending a century learning two systems is even more of a burden. How many additional centuries will it take to compensate for the increased burden of the first century of reform?

This problem is complicated by the fact that spoken language is constantly changing. Bill Labov at the University of Pennsylvania has documented a number of noticeable shifts in pronunciation in American English dialects over the last 30 years. It should be obvious that there are serious disadvantages to changing spelling conventions every 30 years. Yet, if spelling reform is not repeated, we will end up with sound-to-spelling divergence all over again. If professor Rogers is right and it will take the better part of a century to complete English spelling reform, then by the time the reforms are complete, new reforms will be necessary.

Dialectal differences are another serious problem with one-to-one sound correspondence. No matter how it is pronounced, an English word is written the same from Manitoba to Mississippi to Manhattan and, notwithstanding Webster’s innovations like center/centre, to the slums of Glasgow and the Australian outback as well. And most differences between dialects of English do not involve words like center or colour. To paraphrase the old song: I say toh-may-toh, you say toh-mah-toh, but we all spell tomato. This flexibility would not be possible if English had one-to-one sound correspondence. If Mr. Shaw had his way, whose dialect would we be using as the basis of our one-to-one writing system? His? This would mean we would all have to speak in his accent when reading aloud. Either that, or America and Britain would have to use two different spelling systems. This same problem would be replicated within Britain and the US on regional and social lines as well. What would be the advantage in that?

The dialects of Chinese spoken in different parts of China are not mutually intelligible as they are spoken. But all literate Chinese can read the same newspaper. Chinese movies shown in China may have subtitles if the dialect spoken in the film is different than the dialect of the locale where the film is being shown. The dialects of Arabic are similarly not mutually intelligible. Yet, Arabic has far too complex an inflection system to have developed a writing system as logographic as Chinese. Also, speakers of all dialects of Arabic have a fierce attachment to their historical writing system, even though it no longer represents any spoken dialect. This situation has led to the creation of an entirely new language: Modern Standard Arabic. No one actually speaks this language as a native language, but everyone must learn it in school. It is the benchmark for literacy in the Arab world. It might be argued that this is why Arabs have a relatively high level of illiteracy. Yet illiteracy rates have been falling significantly in the Arab world in recent decades despite Arab refusal to compromise on this issue. Mohamed Maamouri of the University of Pennsylvania reports illiteracy in the Arab Region dropping from 73.5% in 1970 to 43.5% in 2000. According to UN HDI 2006, the adult literacy rate for Arab states in 2006 is 69.9%, which is an illiteracy rate of only 30.1%.

I often think that the EU might consider adopting a common writing system along similar lines. Most European languages are not really that different from each other. Romance languages certainly are more similar to each other than modern Chinese dialects. The fact that the former are considered a language family, while the latter are considered a single language with many dialects is more of a political distinction than a linguistic one.

Just as some languages or dialects are united under one writing system for the sake of political unity, others can become split into different writing systems for the sake of separatism. The case of Serbs and Croats has already been mentioned. Hindi and Urdu are in fact mutually intelligible as they are spoken, but are considered separate languages because the Hindus write in Devanāgarī characters, while the Muslims use characters based on the Arabic script. But whether linguistic groups decide to unite under their spelling systems (like the Arabs, the Chinese, and the Anglos) or go it alone (like the Romance languages, the Hindi/Urdu speakers, and the former Yugoslavs), this is a political decision more than a philological one. Of course, there is a measurable advantage to having a single system over a large region of dialects, but if you do not like the speakers of these other dialects, this advantage is outweighed. Similarly, if you speak a language with a complicated inflection system, there may be an advantage to using a writing system with a high sound to symbol correspondence. However, if your ancestral writing system is important to you, and you regard speakers of unintelligible dialects as your brothers, then that advantage is overruled.

Of course, using a writing system based on a single dialect has the effect of privileging speakers of that dialect at the expense of speakers of other dialects, who must in effect learn a second dialect in order to become literate. This is the case in Chinese, whose writing system is based on a Mandarin dialect. This issue is also at the heart of the debate on "Ebonics" in the United States. Unfortunately, there are really only two alternatives. Either use a writing system based on a dialect that no-one speaks, like Modern Standard Arabic, or else let each dialect use its own writing system. Yet, while this latter option does bestow some additional prestige on speakers of minority dialects, it may not alleviate the burden imposed on them in the long run. Residents of the largely Catholic province of Latgale in Latvia have developed their own writing system, which better represents their regional dialect of Latvian. This modified writing system is used in local newspapers and by the Catholic Church in Latvia. There are also some writers and poets who publish in it. However, Latgalians still must learn standard Latvian orthography (and pronunciation) to have a career anywhere outside of Latgale. The only other option would be to go the way of Yugoslavia and declare Latgale an independent country.

The situation in English spelling is somewhere in between these scenarios. English spelling doesn’t really reflect the way anybody speaks (as spelling reformers love to point out), but it isn’t quite as divorced from most spoken dialects as is the case with Modern Standard Arabic. Ultimately, it’s up to the speakers of every language (or dialect) group to work these issues out for themselves, depending on what they feel are the goals of their language, their past, and what kinds of sound-symbol correspondence their language is easily amenable to.

Spelling reformers seem to assume that the goal of writing is simply to reflect speech. Thus, for them, any deviation from perfect one-to-one sound-symbol correspondence is a defect that must be corrected. There in fact already exists a writing system that can accurately reflect the speech of any spoken language or dialect. It is called the International Phonetic Alphabet (IPA) and it is used every day by linguists in their work. Yet no natural language has ever developed a writing system that works like the IPA. Actual written languages always include cultural and political aspects that exist alongside and to some degree at the expense of pure sound correspondence. The goal may be to preserve historic spelling, as in Hebrew, to preserve the spelling unity across derived forms of the same root, as in English, to unite speakers of widely divergent dialects, as in Chinese or Arabic, to separate themselves from speakers of similar dialects, as in Serb and Croat, or any other number of articulated or unarticulated goals. Furthermore, there is something about learning to read in any language that implies a process of acculturation, even for speakers whose spoken dialect doesn’t differ much from the spelling system. Many languages incorporate slang and dialect into the writing system. Chinese has special “dialect characters” for pronunciations that are not represented by any standard character. Yet their use is restricted. In English, we write "ain’t" in novels, but not in PhD dissertations, regardless of how we talk. We all say cuz and wanna, yet we are reluctant to spell these words as we say them even though the English spelling system gives us the option. Even when languages do undergo spelling reform, they never completely correspond to spoken language, because that doesn’t appear to be what most people think writing is for, at least not exclusively.

Advantages of English Spelling

Richard Venezky, widely considered to have been an authority on English orthography, has written a number of books which actually argue for certain advantages in the existing spelling system of English. In The Structure of English Orthography he tabulates exactly when and where English letters are pronounced and when and how English sounds are written: “[English] o corresponds to seventeen different sounds, a to ten, e to nine, and the combined group [of vowel spellings] to forty-eight...” He goes into far more detail than Shaw ever did, and not surprisingly, finds far more regularity in the madness.

It is this level of analysis which clearly shows why ghoti is not in fact a plausible English spelling of fish, even given the current spelling system. First, it is true that gh is sometimes pronounced as f, but this is never true at the beginning of a word. Word initially, gh is always pronounced like a hard g. Check it in your dictionary. Second, while ti is sometimes pronounced as sh, this is never true at the end of a word. Word finally, it is always pronounced like the word tea. You’ll need a reverse dictionary to check that one. Finally, women is the only word in the entire English language where o is pronounced as a lax i. Checking this requires a searchable database. See if you can think of another such word. It is highly unlikely that any new word in English containing this sound would be spelled with an o, or that any new spelling containing an o would be interpreted by English readers as making a lax i sound. Following Venezky, considering that the o in ghoti does not precede a doubled consonant or an r and that it neither precedes nor follows another vowel, I’d say that in ghoti, the o would be pronounced either tense like in note or lax like in not. I’d predict any other interpretation would be very rare. I’ll go further and predict that tense o will predominate over lax o. There are two reasons for this. First, this is the kind of spelling environment that Venezky refers to as “free” as opposed to “checked”. If Venezky’s distinction is correct, the consonant-vowel sequence following o in ghoti is predicted to have the same effect as silent e. Second, both the gh beginning and the ti end conspire to make the word look like it is of foreign origin. English speakers will assume (more or less correctly) that lax sounds are less likely in a loanword.

The language, its spelling, and its speakers have a lot more reason to them than people like Shaw give them credit for.

Here’s a proposal for a simple experiment to put these predictions to the test. Get 100 randomly selected literate native speakers of English. Show them each the word ghoti on a piece of paper. Ask each speaker to read the word into a microphone. Use a wave editor to cut the pronunciations out of the recordings and have them phonetically transcribed by two trained phoneticians who are blind to the purpose of the experiment. Make sure that the transcriptions of the two transcribers correlate. Then, tabulate the token types. If I am right, all of these will begin with a hard g. Slightly more than half should rhyme with throaty. Slightly less than half should rhyme with spotty. All other responses will total less than 5% of the results. On the other hand, if Mr. Shaw is right, there should be a far larger number of token types, all of about equal proportion, one of which sounds just like the English word fish.

Venezky has argued that while English spelling may initially take longer to learn than one-to-one sound correspondence spelling, it makes reading easier once that learning is complete. This is due to the ability of English spelling to map one spelling to more than one pronunciation, as well as to map one pronunciation to more than one spelling. Such one-to-many and many-to-one mappings are precisely what critics of English spelling see as irrational chaos. Yet in practice, these mappings are used rather judiciously in ways that provide many benefits to experienced readers as well as to writers. None of these proposed advantages of English spelling are particularly huge. Yet, as I hope I have shown earlier, the supposed advantages of English spelling reform aren’t very great either. Let us examine a few examples.

English spelling uses one-to-many mappings to pronunciations to preserve the spellings of a single meaningful unit of speech. For example, the plural s in English is sometimes pronounced s as in cats and sometimes z as in dogs. A true one-to-one spelling system would have to write dogz. This obscures the fact that both suffixes are instances of the plural marker s, no matter how it is pronounced. The point is again that a blind insistence on one-to-one sound correspondence ignores a wealth of internal patterns that may be of value to speakers in subtle ways. In their book The Sound Pattern of English, Noam Chomsky and Morris Halle of the Massachusetts Institute of Technology have documented a wealth of other patterns of this type. For instance, English modifies the pronunciation of vowels depending on their position in a word. So, for example, the a in sane is pronounced differently that the a in sanity. In a one-to-one writing system, we would have to write the two a’s differently, since they are pronounced differently. Yet we know that the root san- is the same root in both words. It is not clear why preserving the consistency of the sound to symbol correspondence is more important than preserving the consistency of the spelling of the root from one derived from to another.

English spelling uses many-to-one mappings to pronunciations to disambiguate homonyms such as bye, buy, by and there, their, they’re. This makes the use of such words within a text unambiguous to the reader. In a one-to-one system, they would be spelled the same and be distinguishable only by context.

The possibility of many-to-one mappings also allows English to preserve foreign spelling in loanwords from other Latin-based systems. For instance, the English word football has been borrowed into Spanish, but in Spanish it must be spelled futbol due to the inflexibility of Spanish spelling. In English, however, we can reborrow it back as futbol alongside the original English spelling football. These two spelling will be pronounced more or less identically in English, but the difference in spellings still lets us indicate that in one instance we mean association football and not American football.

Because of this ability of English to swallow foreign spellings whole, new spelling patterns have been introduced, especially from French, which gave us "lingerie" and "tableaux". English speakers have often drawn their own generalizations, which can then be extended to new words. For instance, before there were large numbers of Italians in the English speaking world, I’m not sure what English speakers would have made of a word ending in ucci. But now we pronounce such words as easily as cat, even though we were never taught in school what sound ucci is supposed to make. Not only that, but the spelling ucci (as opposed to the homonymous oochie) actually encodes an additional piece of information, namely that the word or name in question is Italian in origin. An experienced reader of English will eventually learn to encode a myriad of such generalizations about foreign loanwords without ever being explicitly taught any of them.

Another advantage of such a flexible (but highly patterned) system is that people can use alternative spelling to express their individuality, without sacrificing readability. According to the Social Security Administration, three of the most popular names given to girls born in the last 20 years are Ashley, Ashleigh, and Ashlee. And then of course there’s Condoleezza. Businessmen also take advantage of this flexibility to brand their products. Think Kleenex or Cingular. None of this would be possible with one-to-one spelling correspondence. And where is the downside? How many people have mispronounced Ashleigh or Cingular recently?

The English spelling system incorporates many embedded patterns that are intuitively understood and used on a daily basis by English speakers. It simply does not do this by means of one-to-one sound correspondence. Some people act as if it is self-evident or carved in stone somewhere that all languages have to use one-to-one sound spelling. If you think this is all-important, then English spelling is a travesty. If you don’t — it’s not. In fact, English spelling is more like an ecosystem, or a market economy, or a neural network. It contains all the necessary information to respond very effectively to most challenges, it just doesn’t say anywhere in longhand how it does it. This kind of implied, networked decision-making disturbs some people, and it’s usually the same people. But such systems often work as well as or even better than finite control systems like Turing machines, rule-based grammars, or command economies.

If you’re really in favor of piecemeal spelling reform, your best bet is precisely with such a flexible system. Precisely because there are so many different ways to spell the same sound sequence in English, people can decide which variants they find more sensible, and gradually replace them by spontaneously phasing the old ones out. This is impossible in a one-to-one correspondence system, where the only way spelling can be changed is for a central authority to decide that from now on, certain sounds will be represented by different symbols.

Spelling conventions in English do change eventually, as you surely have noticed when reading old texts and seeing spellings like majestick. The change from ick to ic at the ends of words was one of Noah Webster’s innovations, although, unlike his others, it seems to have caught on in Britain as well. Some people today spell through as thru. This is due to a number of proposed changes introduced by the Chicago Tribune in the 1930s. Other innovations of theirs such as hocky for hockey didn’t fare as well, but thru has caught on at least in the word thruway.

Left to its own devices, English spelling will probably continue to reform very gradually, without the use of government power, without the dictates of intellectuals, without massive short-term learning costs and without jettisoning any of the hard-earned but hidden generalizations it contains. Just don’t expect it to ever spontaneously develop perfect one-to-one spelling correspondence.


One of the reasons there may be disproportionately much griping about English spelling is that there are more non-native speakers learning English than non-native speakers learning any other language.

Persian spelling isn’t particularly reasonable by any objective means, but there are few non-Iranians studying Persian, and most of us who do are people who are obsessed with how beautiful and exotic we think it is, so we’re not likely to complain about spelling inconsistency. This is very different than the situation of millions of people around the world who feel compelled to learn English for mundane purposes for their jobs.

This leads me to restate that English spelling can’t really be so much of a cognitive burden. While I don’t believe it is anything about the language itself or its writing system that makes English so influential the world over, there doesn’t seem to be anything linguistic or orthographic about it that’s hindering it, either. Finnish supposedly has one of the closest sound-to-spelling correspondences of any written language. Yet how many non-Finns are studying it? Similarly, how hard can English spelling be for children to learn, when the English speaking countries all share the same high literacy rate as Finland?

Home > Misc. > Archive