Modern English spelling exists for two main reasons: to torment children and to create jobs for English teachers and editors. Even the spellchecker, the device installed for putting a veneer of civilization on universal illiteracy, is not of much use, because a computer cannot know whether we mean
Vowels consisting of two parts are called diphthongs, as opposed to monophthongs. In English, the most audible of them is the one at the end of words like no. (Before going on, I cannot refrain from a remark on the word diphthong. The pronunciation dipthong is not uncommon; the same is true of diphtheria, which competes with diptheria. Although dictionaries do not object to the variants with p, those are hardly to be recommended.) People without special training have a hard time recognizing other diphthongs; even o may cause trouble. Some English-speakers admit that a and o are complex sounds only when a tape with their recording is played back to them; then the presence of something like i-e and u-o can no longer be denied. Another aggravating factor is the lack of uniformity among regional norms. In the speech of most Midwesterners, poor and pour are homonyms (a poor man went to pour himself a drink), whereas my poor (with a diphthong), rhyming, as it does, with demure, allure, manure, is distinct from pour (with a monophthong or with a different diphthong) and rhymes with shore, score, and galore. Likewise, merry, Mary and marry have merged in the Midwest (historically, Mary had a diphthong, as in mare), and Old Harry is supposed to be hairy. In other parts of the world, merry Mary can marry and preserve some linguistic dignity.
Be that as it may, diphthongs existed in Middle English. A vowel close to o in modern no was present in the numeral four. It developed from long o (that is, the o as in Shaw in the speech of those who distinguish Shaw from shah; I am sorry that my exposition is so confusing—clunky, as they say now–but it is not my fault: I wish everybody spoke English like Professor Higgins and me) and was later again monophthongized (to use a technical term), or smoothed (a much more human term), but our spelling reflects the stage at which four sounded approximately foh-ur. (It had such a diphthong as late as the beginning of the 17th century.) Naturally, forty and four contain the same root. But in Middle English, forty was a trisyllabic word; rather early, its first vowel lost its length (a common occurrence in words consisting of three syllables) and never became a diphthong. Hence the modern spelling forty. It would, of course, be rash to expect consistency in an institution as capricious as English spelling. Fourteen was also pronounced in three syllables but is spelled like four, while fortnight “two weeks” (from “fourteen nights,” a counterpart of the archaic sennight “week,” from “seven nights”) does without the letter o in the middle. The adverb forth never had a diphthong and therefore needs no digraph (a digraph is a combination of two letters rendering one sound in writing).
The rest of my story will strike some readers as anticlimactic, for there is almost no intrigue left to unravel. Taught and brought were also pronounced with diphthongs (and indeed, with different ones: approximately tah-u-kht- and broh-u-kht-). Then the diphthongs merged and were “smoothed,” but their medieval spelling has come down to us intact. The same changes happened in the history of bought, daughter, naught (from which we have naughty: it first meant “poor, needy; bad, of inferior quality; depraved”), and many other words with -ought-/-aught- in modern spelling.
All and awl are homonyms; several centuries ago, they already were such, but were both pronounced with the diphthong au. The adverb all is spelled as though it had never had that diphthong, whereas awl reveals its past to a modern speaker of English. (The digraph aw is a variant of au, especially in word final position: cf. paw, draw, straw, thaw. However, sometimesw turns up in the middle of a word, as in awl; haul and maul have au, while drawl and shawl have aw—a depressing inconsistsency.) The Romance words balm, calm, and palm, along with the native words calf and half, at one time must also have had au. Unexpectedly, au did not yield the same sound in them that developed from au in all and aught. Equally puzzling is the preservation of l in realm, helm, and overwhelm, as opposed to its absence in half and calf.
I know from long experience that most people who think that they want to know something about etymology are eager for entertaining word histories, and I see nothing wrong in their expectations. But etymology is a chapter in language history, and its conclusions depend on a good understanding of phonetics. A study of Middle English diphthongs may not be the most exhilarating subject in the world. Yet linguists cannot hope to dine on the roof a posh hotel all their lives. Someone must lay the hotel’s foundation and enjoy a series of hasty brownbag lunches. (“Please stop speaking in metaphors,” said an irritated faculty member to the dean at one of the meetings I attended years ago.) Being a card carrying etymologist, I often hear the question how one becomes a serious student of word origins. This is how. Start at the foundation, dig enthusiastically, build one floor after another, get to the roof, and, if you happen to live long, order a six-course dinner there in the winter of your days.