I don’t know whether this will get lost amidst the insanity or provide a bit of dry, stodgy relief, but it’s finally ready to go, so I’ll take my chances.
Far too long ago I wrote a diary, The Prehistory of English: The Case of the Hidden Suffix, about the origin of the ‑th in such nouns as filth, length, and strength that clearly seem to be related to corresponding adjectives, in this case foul, long, and strong. It derives ultimately from a late Proto-Indo-European (PIE) suffix that was added to adjectives to produce abstract nouns in much the same way that we can add ‑ness and ‑ity to adjectives to produce such abstract nouns as boldness and rotundity. This suffix survived into Proto-Germanic (PGmc) as a suffix *‑iþō, (for more details see the previous diary), which was used in much the same way as its ancestor to form nouns from adjectives within PGmc.
There are also many nouns ending in ‑t or ‑th that look — at least if we keep the previous suffix in mind — as if they might have been formed by adding ‑th or ‑t to a related verb, perhaps mangling the verb a bit in the process; some of the more readily apparent are gift (to give), weight (to weigh), cleft (to cleave), birth (to bear), and flight (to fly). And indeed this is the case. I should warn you, however, that this story is a bit more complicated, and I can’t avoid going into a little more technical detail. I’ve tried to find a reasonable compromise between gross oversimplification and incomprehensible honesty; I’m not at all sure, however, that I’ve succeeded.
Our starting point is a pair of PIE suffixes, *‑ti- and *‑tu-, that were used to form abstract nouns from verbs. There were originally both formal and semantic differences in the way they were used, but these distinctions are quite technical and were starting to blur already in PIE, so I’ll ignore them.
These suffixes survived into PGmc, and *‑ti- at least remained productive in PGmc: its PGmc reflex was used to form abstract nouns from PGmc verbs that have no known non-Germanic cognates. (In fact most of the known examples are of this type.) Of course they were subject to the various sound changes that are part of the history of PGmc; for the benefit of those who are interested in the details I will mention some of these sound changes by name and link to the corresponding Wikipedia articles, but the basic story should be reasonably clear without those details. Thanks to Grimm’s law and the Germanic spirant law, their main PGmc reflexes were *‑þi- and *‑ti-, and *‑þu- and *‑tu-, respectively, which for the most part have come down to us as ‑th and ‑t.
None of the examples that I can find has a completely straightforward history in English, but the pair to cleave and cleft comes closer than most. Underlying both is the PIE root *glewbh- ‘to split’, whose PGmc reflex is *kleub-. This is clearly visible in the PGmc verb *kleubaną ‘to split’, which by regular sound changes gave rise to Old English (OE) clēofan and then Present-Day English (PDE) cleave. (Note that PGmc *b was actually a voiced bilabial fricative, like the b in Spanish haber.)
The noun cleft has a more complicated history, one that is easier to present in reverse chronological order. Its immediate ancestor is Middle English (ME) clift ‘a crack, fissure, cleft’. The change to PDE cleft is irregular: ME clift would be expected to produce PDE clift. Apparently the noun was assimilated to the newly formed past participle cleft of the verb, which by the late 14th century is already sometimes found instead of the regular past participle cloven.
The most straightforward explanation of ME clift is that it is the regular reflex of an unrecorded OE *clyft regularly derived from PGmc *kluftiz. The extant OE corpus contains only about 35,000 distinct lexical items, so gaps in it are very much to be expected, and reflexes in Old High German and Old Norse (ON) confirm the existence of *kluftiz. And *kluftiz, despite appearances, really is the result of appending the suffix in question to the root of the verb. (The *‑z is a second suffix, an inflectional marker for the nominative case of the noun.)
So why does it appear that we are adding *‑ti- to *kluf- instead of to *kleub-? This is quite technical, and I’ll not go into very great detail. That we have *‑ft- instead of *‑bt- or *‑bþ- is a consequence of the Germanic spirant law. The change of vowel is an instance of a much more fundamental phenomenon known as Indo-European ablaut. There was a standard set of changes that could be rung on the vocalic nucleus of a PIE root to produce different inflectional forms and even altogether different words with related meanings. For example, the verb sing goes back to the PIE root *sengwh- ‘to chant’, while the noun song goes back to *songwhos ‘a chant’, formed from *songwh-, the so-called o-grade of the same root, which is obtained by changing the vowel e to o.
The basic form of the ‘split’ root is *glewbh-, the so-called e-grade. This is the form whose PGmc reflex is *kleub-. One of its standard variants, called the zero grade, is *glubh-, obtained by dropping the basic e vowel altogether and vocalizing the semivowel *w to u. Appending the suffix *‑ti- yields *glubhti-, whose regular PGmc reflex is *klufti-.
To drive and drift form a similar pair. The underlying PIE root is *dhreybh- ‘to push, to drive’; its regular PGmc reflex is *drīb-, which we do indeed see in PGmc *drībaną ‘to drive’. Its zero grade is *dhribh-, obtained by dropping the *e and vocalizing the semivowel *y to *i, and the regular PGmc reflex of *dhribhti- is *drifti-, whence e.g. ON drift ‘snowdrift’, Middle High German trift ‘a floating of timber; something driven; pasturage (place where cattle are driven); a drove’, and Middle Dutch drift ‘pasturage; a drove’. PDE drift can be traced back to ME drift ‘act of driving; snowdrift; floating’, which is either from an unattested OE cognate *drift or a borrowing from Scandinavian or Middle Dutch but in any case goes back to PGmc *driftiz.
For an example in which the PIE root does not change grades we may use to weigh and weight. The verb is from PGmc *weganą ‘to move’, and the noun is from PGmc *wihtiz ‘weight’. (Here the h is a conventional notation for a sound like the German ch in Bach and the Scottish ch in loch.) The underlying PIE root is *weǵh- ‘to go, move, transport in a vehicle’; its regular PGmc reflex is *weg-, as in *weganą. Owing to the Germanic spirant law, the PGmc reflex of PIE *‑ǵht- is *‑ht-. The change of vowel from *e to *i is the result of i-umlaut, which I discussed in the previous diary, triggered by the *i of the suffix, and the PGmc output of PIE *weǵhti- is therefore *wihti-. The OE intermediates are wegan ‘to carry, transport; to bear; to support; to weigh’ and wiht ‘weight, a weighing’, both regular reflexes of the PGmc words.
I’ll give three more examples in comparable detail; the first is to give and gift. The verb is from OE (specifically Old West Saxon) giefan‘to give’, a regular reflex of PGmc *gebaną with the same meaning. Had this verb developed regularly, it would most likely have become yive: OE initial g was generally pronounced like the y in PDE yet when followed by i or e (cf. PDE yield from OE gieldan). Our give is the result of Scandinavian influence in the Danelaw (cf. Old Swedish giva).
The noun may be directly from OE gift with similar influence from ON gift‘a gift’ (since the OE word would regularly have yielded modern yift). However, the OE word is recorded only in the sense ‘payment for a wife’, a sense that does not seem to have survived into Middle English (ME), so it is quite likely that our gift is actually a borrowing from Scandinavian. At any rate, both the OE and the ON words are from PGmc *giftiz‘that which is given, a gift’, formed by appending the suffix *‑ti- to the root of the verb. (Here again the i in the root is the result of i‑umlaut.)
The second is one of the relatively few examples deriving from the PIE suffix *‑tu-: to flow and flood. The underlying PIE root is *plew-‘to flow, fly, run’ in its lengthened o-grade form *plōw-. The PGmc verb was *flōwaną, whence by regular changes OE flōwan and PDE flow. Apparently the PIE noun had the accent on the suffix, so that Verner’s law applied to yield the PGmc reflex *flōðuz, with *ð (the voiced sound of th in this and other) instead of *þ (the unvoiced sound of th in thin and ether). PGmc *ð regularly hardened to d in OE, and the normal loss of final *‑uz left OE flōd, of which PDE flood is the expected outcome.
Finally, for an example in which the suffix survives as *‑th we may take to die and death. The verb goes back to a PIE root *dhew-‘to pass away, die’ via PGm *dawjaną. This developed regularly into ON deyja ‘to die’. The Scandinavian word was borrowed into early ME as dēȝen, of which PDE die is the regular reflex. PIE *dhowtus, with the o-grade of the PIE root and the suffix *‑tu-, (and inflectional marker *‑s) developed into PGmc *dauþuz ‘death’, with regular OE reflex dēaþ, whence PDE death.
Now I need to correct a lie by omission that simplified the presentation a bit: it’s possible that some of the PGmc nouns that I’ve mentioned were not descended from PIE nouns but rather were new creations in PGmc. We know that speakers of PGmc did create new nouns on this pattern by analogy with inherited ones. In fact, this construction remained productive in at least one daughter language of PGmc. PGmc eventually split into Proto-East Germanic and Proto-Northwest Germanic (PNWGmc), the ancestor of English, and we have an example from PNWGmc.
After borrowing the Latin verb scrībere ‘to write’ as *skrīban, speakers of PNWGmc formed from it the noun *skrifti ‘writing’ on the pattern of PWGmc *drīban and *drifti (from the PGmc verb and noun*drībaną and *driftiz, discussed above). These developed regularly into OE scrīfan ‘to prescribe, ordain, allot, assign; to shrive (hear confession and exact penance)’ and OE scrift ‘confession’, the etymons of PDE shrive and shrift.
And it continued into the history of English proper. The noun heft ‘weight, heaviness, quality of weight’, for instance, only appears in the mid-15th century, created from the verb to heave by analogy with such pairs as weave ~ weft and thieve ~ theft. A century later we find growth, formed from the verb to grow.
It is noteworthy that in this case the suffix was added as ‑th rather than as ‑t, because this is exactly what we would expect had the noun developed normally from a hypothetical PGmc **grōþiz. (The PGmc verb is *grōaną, with root *grō-, and since neither Verner’s law nor the Germanic spirant law would have applied, the PGmc suffix corresponding to PIE *‑ti- would have been *‑þi-.) Even at this very late date the analogy was being applied correctly.
I can’t resist concluding with the bizarre example of to choose and gusto, two words that have no obvious semantic connection at all.
The verb is from PGmc *keusaną ‘to test; to choose’ by way of OE cēosan ‘to choose’; the underlying PIE root is *ǵews- ‘to taste, try, test’, and all of these developments are regular. Then things went a bit astray. The regular ME reflex of OE cēosan is chēsen, which does occur; had it developed normally, the PDE verb would be to cheese. At some point, however, it was displaced by later ME chōse, the actual source of PDE choose. Now the OE diphthong ēo was a falling diphthong, meaning that its first element was more prominent acoustically than its second; if it had a rising variant in some dialect(s), this variant would have yielded ME chōse. The native lexicon of PDE is descended from a variety of ME and OE dialects, so some such history is entirely plausible.
Such diphthong variants did exist. For example, the OE name of what is now Yafforth in North Yorkshire was Ēaford (‘river ford’), with the OE falling diphthong ēa. By 1283 the name appears in record as Yafford, a spelling that clearly shows that the initial diphthong had been replaced by the corresponding rising diphthong.
The noun is not native to English, but it does have a straightforward history back to a PIE form containing the noun-forming suffix *‑tu-. In English it’s a borrowing of Italian gusto ‘taste; enjoyment, pleasure, relish’ that appears in the early 17th century. The Italian word, unsurprisingly, is from Latin gustus ‘flavor, taste’, which in turn is from PIE *ǵustus with the zero grade variant of the root.
It’s not hard to see that the notions of trying, testing, and tasting overlap, and choosing typically entails testing the alternatives, at least mentally; it’s a bit harder to see how the sense ‘keen enjoyment, relish’ fits in, but the related Latin verb gustāre offers a clue. It’s primary sense is ‘to taste’, but it also came to mean ‘to take a light meal’, ‘to partake of’ and ‘to enjoy’, and those senses form a pretty natural progression towards the English meaning.
And that’s probably more about penguins these suffixes than you cared to know.