Fixing the Latin Alphabet

Many of you may agree with me that our spelling is crazy, but ask yourselves why we don't just fix our current Latin-based alphabet, instead of changing to a new one. This page will explore that option in more detail, so you can see for yourself whether you like the results. We'll aim for a solution where the letters stand for the same sounds (or close enough) in every language.

Unicode, the standard computer encoding, now includes 1350 different Latin letters: almost 60 times more than the classical Latin alphabet. About half of the world's languages now write with this extended Latin alphabet, including most languages in the Americas, Africa and southeast Asia. Those languages use a bewildering variety of sounds, and the International Phonetic Alphabet (IPA) needs 107 letters and 52 diacritic marks to record those sounds.

Is that really the best we can do?


Let's start our experiment by keeping all the Latin consonants that have pretty standard values. In addition,

Consonants Labial Coronal Sibilant Palatal Velar
Nasal m n
Voiced Stop b d g
Unvoiced Stop p t c k
Voiced Fricative v z
Unvoiced Fricative f s x
Rhotic r
Lateral l
Semivowel w j h

The letter r stands for any kind of r:

It's just not worthwhile trying to capture the whole range of pronunciations in writing.

Likewise, the b sound in Spanish haba is pronounced with both lips, but we'll still write it as a v. And the "dark" l at the end of English syllables, as in all, is written using the normal l letter.

Let's use "new letters" for a few consonants:

The entire palatal column is empty, except for j; Latin didn't have palatal sounds, but most languages do. I'm going to represent them all by adding j to the corresponding coronal:

Consonants Labial Coronal Palatal Velar
Nasal m n Sibilant nj ŋ
Voiced Stop b d ƶ dj gj g
Unvoiced Stop p t c cj kj k
Voiced Fricative v ð z zj ɣ
Unvoiced Fricative f þ s sj xj x
Rhotic r rj
Lateral l lj
Semivowel w j & wj h

That's a total of 35 consonants, written with 24 letters.

The Slavic languages and Irish also make a distinction between hard and soft consonants (called broad and slender in Irish): the soft/slender consonants are palatalized (which is not the same as palatal). I'll write the palatalization with a y. So Polish miara "measure" is written myara, while mara "ghost" doesn't change.

In Polish, there are three series of sibilants and their corresponding affricates:

Hissing Rustling Humming
z → z ż → zj ź → zy
s → s sz → sj ś → sy
dz → ƶ → ƶj → ƶy
c → c cz → cj ć → cy


Now let's look at the vowels. Unlike consonants, vowels don't have definite positions: they vary between languages, within a language, and even for each individual speaker. The height of your tongue - and thus the size of the opening - varies continuously, as does the position - front to back - of the opening and the roundedness of your lips. You can stress it or not, hold it longer, or move it to another position as you speak. The IPA uses about 30 letters for vowels, and another 10 or so diacritics to modify them.

But we don't need such detail. For a practical orthography, we can get by with only 15 stressed vowels:

Vowels Front Back
Spread Rounded Spread Rounded
Close (high) í ü ï ú
Near-Close ì ù
Close-Mid é ö ë ó
Open-Mid è ò
Open (low) ä á à

They all have accents: an acúte accent for the higher version, a gràve accent for the lower version, and a diëresis for the central version. These accents also indicate stress; when the vowel is unstressed, we don't need to be so precise about the pronunciation. For example, ï is the sound of Polish y, Russian ы and Turkish undotted ı. We also have this sound in English, but only in unstressed syllables, like in little buttons merit roses. But there, we don't need the accent: we can use a normal (dotted) i.

Vowels are the nuclei of their syllables, so there's only one vowel per syllable. But often, other vowels are used as on-glides before the nucleus or off-glides after the nucleus. In these cases, they're actually being used as consonants, and we call them semivowels. The high vowels i u are very often used this way, and we will write them as j w. Note that the diphthong ju makes some preceding consonants palatal, just as the u in sugar makes the s into an sh.

When the off-glide is low, we write -h or -r. Long vowels are short vowels that are repeated as an offglide, so long a as in English bra or pa would be written ah.

Here are the English vowels, shown using the excellent color vowel system:

In island English - British, Irish, Australian, New Zealand, South African - all the vowels ending in -r in mainland English are written with -h; those dialects are called non-rhotic.

We need a few more letters: Portuguese, Polish and French have nasal vowels, which we'll write with a tilde above the vowel. In French, they're not just nasalized; they're also lowered. Here are all the French vowels:

French
Vowels
Front Back
Spread Rounded Spread Rounded
Close (high) í
si
ü
su
ú
sous
Close-Mid é
chez
ö
ceux
soeur
e
ce
(muet)
ó
sceau
Open-Mid è
sait
ò
sol
Nasal (low) ĩ
sain
ũ
un
ã
cent
õ
son
Open (low) á
sa
à
sable


25 consonants and 5 vowels with one (or none) of 5 accents: that's our European alphabet: 35 letters. Two final "letters": the ' apostrophe, representing a glottal stop, and the - hyphen, representing a separation within a word. We need the apostrophe to separate sounds that would otherwise run together, like in German Be'amter or Czech po'užívat. The hyphen, on the other hand, is silent: we use it to separate false digraphs like mìd-jír (mid-year), and to link two words more closely than a space, like the name Jean-Claude.

But we'd need many more letters for the rest of the world's languages:

So you see we are a long way from a universal alphabet! But maybe we shouldn't be writing non-European languages in a European alphabet, anyway.


OK, now that we have our new Latin alphabet, let's see how good it looks:

English: Fòr skòr and sèvin yírz egów, áwr fáðirz bròt fòrþ an ðìs kántinent e núw néjsjin, kensíjvd in lìbirtij and dèdikejtid te ðe prapezìsjin ðat òl mèn ár krijéjtid íjkwel.

Spanish: Nó ávles a ménos ke pwédhes mexorár el silénþijo.

German: Probléme kán mán níjmalz mit derzèlbin dèŋkvajze lözin, dùrx dij zij entsjtándin zìnt.

French: Kã zje zjú evèk mõ sjá, kí sè síl ne semüz pá plüs de mwá ke zje le fè de lwjí?

Russian: Volkòv bojátsa - v lyés nye xodyíty

What do you think? Daunting, isn't it?

But maybe if we forget about other languages and just try to fix English spelling, we can do much better. That idea is explored on the page Fixing the English Alphabet.


Most people are slightly surprised to discover that this "fixed" Latin alphabet is so hard to read - after all, we already know the letters! But it turns out that knowing the letters of the alphabet is only a very small part of knowing how to read and write. In fact, as is well known, fluent readers don't read individuals letters at all, and they certainly don't sound them out mentally in order to decipher every word. Instead, people recognize entire words at once, which is why Chinese readers can read just as fast as readers of alphabetic scripts.

But this doesn't mean that all writing systems are equal! There is a huge difference when it comes to the difficulty of learning to read and write in various systems, and systems in which there is a regular correspondence between the written and spoken languages are much easier. This is especially true in situations in which people learn the spoken language from the written one, as is usually the case when literate people learn a foreign language in a classroom. It's even how people learn most of their native vocabulary: from reading!

Naive reformers often assume that retaining the Roman alphabet will make learning a "fixed" orthography easier - after all, we already know the letters. But it's pretty easy to learn new letters; look how quickly we all learned @. What's hard is learning to spell, and if a new alphabet makes that easier, then the investment in learning new letters is paid back very quickly.


© 2002-2021 The Musa Academy musa@musa.bet 04apr19