The curious linguistic histories of ump, imp, amp, omp, and empt
As every linguist learns in college, the relationship between a word's form and its meaning is purely arbitrary. Except when it isn't.
Of course there are words that imitate sounds, like bang and oink, and sometimes those are used metaphorically, like zoom. The rest, it is generally agreed, are like glue, clue, and blue: the connection between form and sense is effectively random, and any acceptable combination of sounds is theoretically about as likely as any other. If words that sound alike have similar meanings, or if some bunchings of sounds are over-represented, it's because they're historically related or it's coincidental.
Just don't mention the mp clan.
The mp clan don't care about this "arbitrary" and "basically random" stuff. It cramps their style. Some of them are champs, some are chumps, some are wimps, but they all romp together, and they won't be pre-empted.
Who are the mp clan? They're all the words that end in mp. The mp clan has five houses, each with its own characteristics.
The house of ump is the house of bump, frump, hump, lump, rump, thump… Its members are generally dull, blunt, or heavy things or actions: three-quarters of English words ending in ump have that in common, and some others are arguable, like jump and pump. It is the biggest house in the clan of mp, with about 20 well-known members.
The house of amp is the house of champ, stamp, amp, cramp, revamp: aggressive or energetic things and actions. Nearly two-thirds of its members have that in common.
The house of imp is the house of chimp, imp, limp, scrimp, shrimp, wimp… nearly all of its 13 members have something to do with small, weak, or bad things.
The house of omp has 7 well-known members. It's like the house of amp but its members are perhaps just a bit heavier, but more positive than ump: chomp, clomp, romp, stomp, and whomp. The only two common omp words that don't express that kind of thing are comp and pomp.
The house of emp is… almost empty. It has only two commonly used members: hemp and temp. It just can't seem to get it together.
But there is an impostor house: the house of empt. It is not a huge house, but it has attempt, contempt, exempt, pre-empt, tempt, and unkempt.
You would think the house of empt would hang out with its own mpt clan. But what mpt clan? If you set out to look for other members of this clan, your results will be prompt. Literally: Prompt is the only other commonly used English word ending in mpt. If you look for uncommon words, you can add dompt and consumpt, and just maybe crimpt.
So… why is this? Why this uneven set of houses, and why do the members of each house — which for the most part don't have common historical sources — have these meanings in common?
The second question is easier to answer than the first. To start with, some of the mp words imitate noises made by what they refer to: bump, clump, thump; clomp, stomp, whomp. This helps set a pattern of similar-sounding words that are imitative or expressive. In general, we are used to smaller things making sounds more like "imp" and larger things making sounds more like "omp" and "ump" (and don't forget that in Old English — as in some modern English dialects — those u's were like the oo in book, even more deep and hollow-sounding). Birds say "tweet" and "cheep"; cows say "moo"; big dogs say "woof" and small dogs say "yap." This can help shape our attitudes towards other words that have those sounds.
When people encounter a new word, they tend to guess its meaning under the influence of familiar words it sounds like, and this can cause the sense to shift over time. And when people want to choose a word to express something, they can be more likely to use a word that "sounds right" — like other words with similar senses. It's not just the mp clan that has this kind of peer pressure. But the mp clan does it better than most.
But why are there all those ump and amp and imp, but fewer omp, and almost no emp? Did some of them shrink?
Sure, they shrink, shrank, shrunk because they drink, drank, drunk and sing, sang, sung… You know irregular "strong" verbs in English, verbs that change the vowel to make the past tense instead of adding –ed? We have a few dozen of them, and there used to be more. They have several vowel patterns they can follow, but when there is an n or m after the vowel, the normal pattern is i-a-u: I begin, I began, I have begun. If crimp were a strong verb, it would be I crimp it, I cramp it, I have crump it. Several words in the mp clan came from such old strong verbs that eventually became regular (crimp, crimped). In some cases, such as ramp and dump, the amp or ump version lasted instead of the imp. This helped start the peer pressure.
What about the house of omp? Chomp, romp, and stomp are actually modified forms of champ, ramp, and stamp to express a heavier sound. Clomp is related to clump. Whomp is imitative (and was invented less than a century ago). That leaves just pomp, which comes from a Greek root, and comp, which is a shortened form of complimentary (or, in some cases, comprehensive). So the house of omp is sort of like one of those organizations where most of the members are infiltrators and spies.
And the house of emp? Temp is another shortening, and hemp comes originally from henep — though it had become hemp by the 14th century. Why not any other words? Is emp just not imitative enough? An at least equally good reason is that imp, emp, and amp are harder to tell apart than, say, ick, eck, and ack. The added nasalization of the vowel actually makes both imp and amp sound closer to emp, leaving less room in the middle, so emp tends to be avoided. We can see a similar effect with nk: we have plenty of ink, ank, onk, and unk words, but try to think of an enk word (tip: give up).
Okay, so fine. Where do these empt words come from, then? Did they just add a t to stand out more?
Nope, they really are impostors. With one exception, they come from Latin. Also, in nearly all of them, the p is actually there just because the m hardens to a "p" sound just before the t (the nose closes and the voice stops before the lips open). Attempt and tempt come from Latin temtare meaning "handle, touch, try"; contempt comes from con plus the root temn– (meaning "scorn, despise") plus the past tense ending tus (trimmed down in English); exempt and pre-empt — and prompt too — start with the root em– ("take") plus a prefix and the past tense ending tus. The only non-Latin one is unkempt (and kempt, of course). It comes from an Old English verb root kemb meaning "comb," plus a past-tense ending. Why not kembed? For the same reason we got burnt to go along with burned: we used to use –t rather than –ed sometimes.
So the house of empt is really just a random collection of exceptions coming together by coincidence. It doesn't belong in the clan of mp at all. House of empt… ya burnt!