Two Jays

That’s 12 out of 18 or so vowels covered. We need to look at r. It modifies the sound of a preceding consonant like this:

spelling	IPA	as in …	fallback
a	ɑː	cart, hard	aa
e	ɜː	her, learn	eu
i	ɜː	fir, squirt	eu
o	ɔː	bore, four	au
u	ɜː	cur, hurd	eu
oo	uː	boor, hooray	ui

Most of these sounds are new, but one is not: the ‘r’ has no detectable effect on the sound of ‘oo,’ so we’ve already dealt with it. Of the other 5 vowels, 3 are identical: her, fir and cur all rhyme.

So the words above will be written ‘cart,’ ‘hard,’ ‘her,’ ‘lern,’ ‘fer,’ ‘sqwert,’ ‘bor,’ ‘for,’ ‘ker,’ ‘herd,’ ‘boor’ and ‘hooray.’

If you are non-rhotic speaker of English (like me), you are probably feeling annoyed at this point, because you can’t hear the letter ‘r’ as in the table (except in ‘hooray,’ where there’s a following vowel). You want to spell them ‘caat,’ ‘haad,’ ‘heu,’ ‘leun,’ ‘feu,’ ‘sqweut,’ ‘bau,’ ‘fau,’ ‘keu,’ ‘heu,’ ‘booꝛ’ and ‘hooray’ (using the fallback spelling). This would make lots of sense if there were no rhotic speakers, but we need to compromise here. Do we want to spell ‘cart’ as ‘cart’ or ‘caat’? The former is much more helpful to non-rhotic speakers than the latter is to rhotic speakers, so it’s the obvious choice. The latter would result in ‘r’ being annihilated from so many words where it provides such useful legibility that the other choice practically withdraws before the election even happens.

These sounds can all also occur without an ‘r’ being present at all. In that case, we have to use the fallback sequence instead of the single-letter form.

spelling	IPA	as in …
aa	ɑː	glass, past, father
eu	ɜː	colonel, uh
au	ɔː	taut, thought

So the words in the table become ‘glaas,’ ‘paast,’ ‘faadher,’ ‘keunel,’ ‘eu,’ ‘taut’ and ‘thaut.’ For a non-rhotic speaker, strictly speaking these are ambiguous: they can’t tell from the pronunciation ‘glaas’ if they should write ‘glaas’ or ‘glars,’ and similarly ‘paast’/‘parst,’ ‘faadher’/‘fardher,’ ‘keunel’/‘kernel,’ ‘eu’/‘er,’ ‘taut’/‘tort’ and ‘thaut’/‘thort.’ To this extent, this whole project could be considered as failure. But let’s just call it a blemish instead, and continue.

The sound ‘aa’ has an unusual history: it is another sound (like ‘dh’ or ‘zh’) that has no written cue at all. It has only existed since the 17th century, it started in Southern England, and it soon became a shibboleth for middle- and upper-class speakers. This is called the trap–bath split (or “trap–baath split”). It is not universal: many speakers would say ‘trap’ and ‘bath’ have the same vowel. Nothing in the above stops anyone from pronouncing ‘a’ and ‘aa’ the same, of course, in which case they just have to learn which words are spelled with ‘a,’ which with ‘aa.’ Another blemish. Ah well.

Also, some speakers have pronounce the lax ‘o’ and the r-tense ‘o’ (or ‘au’) the same. This is called the cot–caught merger. At least in this case the difference is already in the spelling which we already learn, so we aren’t making anyone’s life any harder.

There’s also a thing called the lot–cloth split which, like the trap–bath split, once enjoyed vogue among the upper classes of English society. You can hear it the Pirates Of Penzance, where some amusement is derived from the fact that ‘orphan’ sounds exactly the same as ‘often.’ This need not detain us, as we’ll just ignore it.

One last thought: there is a sound that non-rhotic speakers make that has no alphabetic representation at all. It is a schwa (‘ə’) sound that marks an important distinction in words where the vowel is modified by an ‘r,’ but the ‘r’ itself is not pronounced, at least not as any recognisable consonant. If we were going to use the r-free spellings ‘caat,’ ‘haad,’ ‘heu,’ ‘leun,’ ‘feu,’ ‘sqweut,’ ‘bau,’ ‘fau,’ ‘keu,’ ‘heu,’ ‘booꝛ’ and ‘hooray’ from above, we would need some completely new letter, already hinted at in ‘booꝛ,’ to mark that sound. This has to happen even if not inventing new letters is very high on the priority list, as it is. ‘ꝛ’ (r-rotunda) is suitable for this because it’s just a fragment of a letter ‘r,’ which fits what it’s doing here. The list of possible places it would go is this:

spelling	IPA	as in …
aꝛ	eɪə	mayor
eꝛ	iːə	near
iꝛ	æɪə	fire
oꝛ	ʌʊə	lower
uꝛ	ɪʊə	cure
ooꝛ	uːə	boor
oiꝛ	ɒɪə	destroyer
ouꝛ	æʊə	flour

(Some of these are arguably 2-syllable words, but we’ll ignore that.) If we had adopted r-free spelling, we would have to write maꝛ, neꝛ, fiꝛ, loꝛ, cuꝛ, booꝛ, destroiꝛ, flouꝛ, which would be quite a remarkable invention. But we didn’t, so we actually will write mair, nere, fire, loar, cure, boor, destroir, flour, as we will soon see.

This where it ges interesting. The 6 vowels also have another form, the “tense” form, which is signalled in spelling by a systematic and regular system of lengthening. Here they are:

Spelling	IPA	as in …	fallback
a	eɪ	bale, wake, fail	ai
e	iː	beep, whee, peel	ee
i	æɪ	lied, vile, mike	ia
o	ʌʊ	bowl, whole, toll	oa
u	ɪʊ	cute, lure, butte	ue
oo	uː	fool, boom, tomb	ui

The vowel (one of ‘a,’ ‘e,’ ‘i,’ ‘o,’ ‘u,’ ‘oo’) is followed by a single consonant letter and then another vowel, which is arbitrary.

But obviously in the above table, not all the words have the sequence vowel-consonant-vowel. We are going to say they are irregular, and respell them so that they do, like this: ‘bale,’ ‘wake,’ ‘fale,’ ‘bepe,’ ‘whe,’ ‘pele,’ ‘lide,’ ‘vile,’ ‘mike,’ ‘bole,’ ‘whole,’ ‘tole,’ ‘cute,’ ‘lure,’ ‘bute,’ ‘foole,’ ‘boome,’ ‘toome.’ The final ‘e’ is a “magic” ‘e’ that isn’t pronounced, but it still counts as the final vowel in the vowel-consonant-vowel sequence. (When we mention polysyllabic words, which we should probably do soon, other vowels than ‘e’ can take this rôle.)

We also need to think about what to do if these sounds occur when they are not before a consonant and a vowel. This can happen in words like ‘baste,’ ‘wheeled,’ ‘heist,’ ‘toast,’ ‘lutes’ and ‘doomed.’ So we need an unambiguous way to spell each of these, as a fallback when the vowel letter isn’t in its natural position of vowel-consonant-vowel. Exhaustive research indicates that the sounds ‘eɪ’ ‘iː’ ‘æɪ’ ‘ʌʊ’ ‘ɪʊ’ ‘uː’ are often spelled by the pairs ‘ai,’ ‘ee,’ ‘ia,’ ‘oa,’ ‘ue’ and ‘ui’ (as in ‘plaice,’ ‘teeth,’ ‘briar,’ ‘goats,’ ‘cued’ and ‘sluice’), which gives us the spellings ‘baist,’ ‘wheeld,’ ‘hiast,’ ‘toast,’ ‘luets’ and ‘duimd.’ —How does it look? ⊢Not too alien. Great.

English has 18 vowels. Or maybe 17. Or maybe some other number. Let’s pick a number that seems nice, such as 18, and do the best we can with that number.

Some of these vowels are short and unproblematic. They are called the “lax” vowels. Let’s say these:

Spelling	IPA	examples …
a	æ	cat, lamb, has
e	e	get, debt, well
i	ɪ	ill, jim, kick
o	ɒ	hot, what, moll
u	ʌ	hull, dumb, pup
oo	ʊ	book, put, could

There are 5 vowels, ‘a,’ ‘e,’ ‘i,’ ‘o,’ ‘u,’ but there’s also a 6th, ‘oo,’ which turns out to be useful. I’ve put the IPA in the table, but it doesn’t really help, as there are 6 different symbols for 6 different sounds, so if you don’t know them, you learn nothing. Maybe later they will come in handy or something. The problem now is that we’ve used up all the vowels letters (and we cheated by using ‘oo’ as well), and we still have 12 sounds left to spell. This means we need to use context in some way.

So, that’s it for the consonants. If you like IPA, the consonants of English are:

spelling	IPA	as in …
p	p	pen, spin, tip
b	b	but, web
t	t	two, sting, bet
d	d	do, odd
ch	tʃ	chair, nature, teach
j (g before [ei])	dʒ	gin, joy, edge
k (c before [^ei], q before w)	k	cat, kill, skin, queen, thick
gh (g before [^ei])	ɡ	go, get, beg
f	f	fool, enough, leaf
v	v	voice, have
th	θ	thing, teeth
dh	ð	this, breathe, father
s (c before [ei])	s	see, city, pass
z	z	zoo, rose
sh	ʃ	she, sure, emotion, leash
zh	ʒ	pleasure, beige
h	h	ham
m	m	man, ham
n	n	no, tin
ng	ŋ	ringer, sing, drink
l	l	left, bell
r	ɹ	run, very
w	w	we
y	j	yes
wh	ʍ	what
x	ks	hexagon

If we are going outside English into adjacent languages, we might also like these. Let’s take them, just in case.

spelling	IPA	as in …
kh	x	loch, ankh (in Scottish)
lh	ɬ	Llanelli (in Welsh)
jh	ɣ	voiced version of kh (in Old English)

That’s nearly it!

The last thing we need is some way of spelling words where these various digraphs appear, but are pronounced separately, as in ‘jachammer’ (the rule for ‘c’ make this the spelling of ‘jackhammer’), ‘longhorn,’ ‘carthorse,’ ‘redhanded,’ ‘mishap,’ ‘hogzhed’ (how we will spell ‘hogshead’), ‘ungainly’ or ‘engine,’ ‘foolhardy’ and ‘hejhog’ (the new ‘hedgehog’). These are mostly at syllable boundaries, which is why the ambiguity in spelling doesn’t matter so much. But we have to do better—a word is a word—so we need some sort of separator to keep them apart. We won’t be using the regular apostrophe any more, as that is a feature of spelling that has no pronunciation, so we can reuse it. But just for idiosyncratic fun, let’s point it other way typographically, and use the left quote instead the right quote. So these words become ‘jac‘hammer,’ ‘long‘horn,’ ‘cart‘horse,’ ‘red‘handed,’ ‘mis‘hap,’ ‘hogz‘hed,’ ‘un‘gainly,’ ‘en‘gine,’ ‘fool‘hardy’ and ‘hej‘hog.’

Wow. Next, the vowels.

One more consonant remains, ‘x.’

Every rational spelling of English acknowledges that ‘x’ is redundant, and should be spelled ‘ks.’ Let’s ignore the fact that according to the rules we have so far, it would be ‘cs,’ and instead embrace it. Here’s the rule: wherever the sounds ‘k,’ ‘s’ occur in order, they are written ‘x.’ Always!

You might think that this would lead to many funny spellings, but it doesn’t kick in that often. We get, for example, ‘sox,’ ‘hax’ and ‘sux,’ and more interesting examples like ‘suxeed,’ ‘blaxmith’ and ‘oxidental,’ which all look amusing rather than ugly. So this is obviously the right decision.

That’s a Hebrew letter yod. It’s a sort of joke.

So, yod coalescence is a thing that even has its own page on Wikipædia. No-one says ‘action’ as ‘ac-tee-on’, though some may say ‘question’ as ‘kwes-tee-on.’ True, dat. Many people say ‘duke’ as ‘jook,’ though also many say it as ‘dook.’ These are all instances of yod coalescence, which is a pronunciation feature of some people that converts consonants into other consonants when they are written before ‘e’ or ‘i’ (and the tense sound of ‘u’ counts as ‘i-oo’ for this purpose), like this: ‘d’ → ‘j,’ ‘s’ → ‘sh,’ ‘t’ → ‘ch’ or ‘sh’ and ‘z’ → ‘zh.’

Examples are ‘duke,’ ‘fission,’ ‘christian,’ ‘action,’ ‘fusion’ which become, if you like, ‘jook,’ ‘fishon,’ ‘chrischan,’ ‘acshon,’ ‘fuzhon.’ The problem is knowing where to stop. I’m sure some people do say ‘chrischan,’ though a lot more probably say ‘chris-tee-an’. Even more may alternate between the 2, without being aware of it much.

The one certainty is that no-one says ‘ac-tee-on’. Let’s commit! The rule will be yod coalescence everywhere. It’s weird! It has to be done! After ‘dh,’ ‘z,’ ‘gh,’ and ‘zh’ as described above, probably no-one will even notice.

How could we forget q? It’s just the way you write the hard ‘k’ sound when it’s before a ‘w’ sound. So here we go for ‘qween,’ ‘qwick’ and ‘conqwest.’ (We haven’t mentioned vowels yet, but it seems clear that after a ‘q,’ a ‘u’ is going to be written ‘w,’ as it’s obviously a consonant there.) The only weirdness here is that to retain the 1–1 correspondence between written and spoken forms, every ‘k’ before a ‘w’ must be written as a ‘q’—else how would you know?

So that gives us ‘awqward,’ ‘baqwards’ and ‘booqworm.’ These are strange combinations that occur at syllable boundaries, but of course, we can’t just arbitrarily say that syllable boundaries have special status—at least not without some hint in the spelling. If it’s a word, it’s a word. End of story. So ruthlessness prevails.

We have one more sound that has no unambiguous representation in English, and we are going to have to resort to invention again. Boo! (Or is it hooray!?)

It’s the consonant sound in ‘fusion,’ ‘Asian’ and ‘beige.’ Since it differs from the sound in ‘ship,’ ‘shop’ and ‘shall’ in voice, it makes sense to use the voiced version of ‘sh,’ which would be ‘zh.’ So we get ‘fuzhon,’ ‘Azhan’ and ‘beizhe.’ Nice, or nasty? Whichever it is, I see no alternative. So that’s where we are.

For some people, ‘w’ and ‘wh’ are 2 different sounds. They differ in the same way that ‘t’ and ‘d’ differ: by voice. Unadorned ‘w’ is voiced, but ‘wh’ is more like a whisper. Which makes that word an onomatopœia. How about that?

It‘s true that for most people, the difference is lost. But there must be 1 person—maybe more—who still pronounce those words differently. So let’s commit to them!

Another reason to do that is that the spellings are so evocative. When you see a word starting ‘wh,’ you know it’s a question. Or a relative. Or something like that. It’s cute. It would be sad to lose it. So the decision is that the English we are respelling is one from before the wine-whine merger. If you can’t hear it, use your imagination. Let’s do it!

These two letters have the same relationship as ‘s’ and ‘c’. Specifically, ‘g’ before ‘e’ or ‘i’ is pronounced as a ‘j,’ (a “soft” ‘g’) and elsewhere as a “hard” ‘g.’

Except there is no equivalant of ‘k,’ no letter which is always pronounced hard. This is especially bad because there are so many common words which are exceptions to the rule, with a hard ‘g’ before an ‘e’ or an ‘i,’ like ‘give’ and ‘get.’ (This is often because they come to us through German, rather than French, which is where the soft/hard rule comes from.) So, let’s invent a letter! Since we already have so many digraphs ending in ‘h,’ and since ‘gh’ is already pronounced ‘g’ in words like ‘ghost’ and ‘ghetto,’ we can use that. Then we can adopt exactly the same rules for j/g/gh as for s/c/k, like this: ‘g’ before ‘e’ or ‘i’ is pronounced ‘j’, otherwise ‘gh,’ and the other side of the coin: the sound ‘j’ before ‘e’ or ‘i’ is always spelled ‘g;’ the sound ‘gh’ not before ‘e’ or ‘i’ is also always spelled ‘g.’ More funny looking words! ‘Gelly,’ ‘gester’ and ‘gigsaw’ for example, and also ‘ghive’ and ‘ghet’.

This is going well. Are we done with consonants yet?