Even the best books prefer not to discuss this situation. Skeat has the entry Thrash, Thresh and says “Thresh is older,” a correct but somewhat cryptic remark. The Century Dictionary and H. C. Wyld’s The Universal Dictionary of the English Language, both usually so helpful in matters of etymology (regardless of whether their solutions carry conviction), are equally uninformative. Even the great Karl Luick, the author of the unsurpassed history of English sounds, sits on the fence, and that is where we too will stay. As noted, I will assume (with many others) that the modern doublets arose in Shakespeare’s lifetime, rather than continuing þerscan ~ þærscan. Nowadays the verbs are no longer synonyms, for thresh refers to beating corn (grain), and thrash to beating enemies and opponents, though we can also thrash out a problem. In the vocabulary of all languages, such division of labor between close neighbors is typical, unless one of them manages to oust its rival altogether and take over its senses.
The distant etymology of thresh and the passage of thresh through the centuries need not delay us here. We should only observe that the old verbs began with þ– (= th-) and a vowel followed by r, whereas the present day forms are thresh and thrash, in which the vowel follows thr-. This is not a complication because r is the most frequent partner in the game of leapfrog called metathesis (compare Engl. burn and burst versus German brennen and bresten). The real problem is the variation e ~ a. We don’t know why Old English had two similar verbs, but, since we pretended that this fact is of no consequence to us, we have to explore only the causes of what happened in the sixteenth century.
The variation reminiscent of thresh ~ thrash occurs in all kinds of words. Sometimes it marks social dialects. Demned for damned, a form well-known from direct observation and often ridiculed in fiction and film, is a case in point. Unlike damned, thresh is a word of the neutral style, mainly used by peasants, but thrash is different, and Wilhelm Horn (see more about him below) ascribed the change to emphasis. Perhaps when one beats up an offender, the homey verb thresh is insufficient and in describing the process one wants to open the mouth wide. However tempting such conjectures (like references to sound imitation) may be, we have no way of proving them, which does not mean that they are necessarily wrong. In etymology, once linguists step outside trivial phonetic correspondences, very little can be “proved.”
In the middle of the fifteenth century, the spelling Wanysday “Wednesday” turned up several times. Later, nafew “nephew” appeared in private letters, along with “reverse spellings” (that is, with e for a): bechelor and cheryte for bachelor and charity. Henry Cecil Wyld (the same H. C. Wyld, whose name graces the title page of a superb one-volume dictionary) dug out many such examples, and some of them are discussed in Laut und Leben (“Sound and Life”), a book by Horn-Lehnert (Lehnert was the editor of his teacher’s posthumous opus magnum.) Horn also thought that frantic, which superseded frentic in the sixteen-hundreds, is an emphatic form. Indeed, the word’s expressive meaning does not contradict such a hypothesis. The adjective goes back to French frénétique. The change of e to a ruined for English speakers the ties between frantic and frenzy. In dialects, the formfranzy is not uncommon.
Short i was also often broadened in the seventeenth century, as evidenced by such spellings as cheldren, denner, desh, shep, and tell for children, dinner, dish, ship, and till. Those changes need not strike us as absolutely chaotic and unpredictable. In the Early Modern period, the long vowels of English underwent a major restructuring known as the Great Vowel Shift. This is the shift that, among other things, drove a wedge between the names of the letters in English and in the other European languages. (For example, outside English, a is pronounced as Engl. a in spa.) In the languages of the Germanic group, when long vowels begin to change, short vowels usually follow suit and “try” to look more like their long partners. Consequently, a good deal of vacillation that looks odd can be expected.
The sound r frequently affects its neighbors (without establishing causal connections, we can at least say that in its vicinity all kinds of events take place). It would be tempting to ascribe the appearance of thrash from thresh to its influence, but the outwardly erratic behavior of vowels, so surprising to the scholar (a linguistic analog of Brownian motion!), makes all conclusions risky. It will be enough to say that the short vowels received an impulse rather than a command to modify their values and either obeyed it or remained stable. Fresh, fret, press, dress, crest, and so forth did not yield forms with a.
A few cases are special. I’ll cite only one. The word errand was pronounced in the eighteenth century and some time later as arrand, though, for some reason, spelled with an e. The pronunciation made sense, for the Middle English form was arunde. Language teachers suggested that perhaps people pronounce this word according to its visual image, and that is what they did. An amazing example of successful and evidently painless language planning! It follows that, when one encounters examples like thresh ~ thrash, all the facts have to be examined before even preliminary conclusions can be offered. (Politicians, with their penchant for buzzwords and clichés, would have probably announced here that all options are on the table.) So, as they teach in grade school: “Don’t generalize.”
When all is said and done, it seems that thrash owes nothing to its Old English lookalike, whose origin remains unclear, and that it developed from thresh for some vaguely stated phonetic reasons. To keep its distance from thrash, the new verb acquired a light-hearted, almost slangy tone. It also seems that the writing thresh ~ thrash with a tilde between them does not distort the history of either word.