Why English Has So Many Irregular Verbs

English has roughly two hundred irregular verbs still in everyday use, and the honest answer for why they exist isn't "English is illogical" — it's that modern English is running two separate historical systems for forming the past tense at once, and they never merged into one. Regular verbs (walk/walked, text/texted) use what linguists call the "weak" conjugation: add a dental suffix, -ed, and you're done. Irregular verbs like sing (sang, sung), swim (swam, swum), and begin (began, begun) use the older "strong" conjugation, which marked the past tense by changing the vowel inside the verb instead of adding an ending at all.

Where the strong-verb system actually comes from

This vowel-changing pattern, called ablaut, is a genuine inheritance from Proto-Germanic, the common ancestor of English, German, Dutch, and the Scandinavian languages. Old English organized its strong verbs into seven distinct classes, each with its own predictable vowel pattern across present, past-singular, past-plural, and past-participle forms — sing/sang/sungon/sungen is a tidier, more systematic version of the modern sing/sang/sung than most people realize once existed. German still shows the family resemblance clearly: singen/sang/gesungen runs on the exact same ablaut mechanism as English's sing/sang/sung, because both languages inherited it from the same Germanic ancestor rather than one borrowing it from the other.

Centuries of sound change eroded the clean boundaries between those seven classes. Vowels that used to be distinct merged through general shifts in English pronunciation (the Great Vowel Shift among them), verbs drifted between classes through analogy with more common neighbors, and some verbs picked up features from more than one class at once. That's why modern irregular verbs look like a scattered, memorize-each-one-separately list rather than the seven neat patterns they descend from — the patterns are still there underneath, just obscured by a thousand years of erosion.

Norman French and the regular -ed pattern's takeover

The other half of the story is why the -ed pattern became the default for new verbs at all. After the Norman Conquest of 1066, Old English stopped being the language of law, government, and high culture in England for roughly three centuries, replaced by Anglo-Norman French and Latin in those domains. When English re-emerged as a full literary and administrative language in the 14th century, it had absorbed thousands of French and Latin loanwords — and every single one of them needed a past-tense form assigned to it from scratch. The strong-verb ablaut system was a closed, native-only pattern; it was never going to be productively extended to a borrowed verb like "judge" or "continue." The regular weak -ed pattern, being a simple additive rule rather than a memorized vowel-class assignment, became the automatic default for every new verb entering the language from that point on — a pattern that has held for every subsequently coined or borrowed English verb since, from "telephone" to "google."

Why the list keeps shrinking, not growing

Harvard researcher Erez Lieberman led a 2007 study, published in the journal Nature, that put a number on something linguists had long suspected: an irregular verb's odds of survival track its frequency of use almost mathematically. Digging through roughly a millennium of English text, Lieberman's team calculated that a verb used one hundred times less often than a comparison verb regularizes about ten times faster — meaning the irregular forms clinging on today aren't randomly surviving relics, they're overwhelmingly the verbs speakers say and hear constantly enough that the brain never bothers substituting the "easier" regular pattern. It's the reason a rarely used irregular like "chide" has quietly slid into the regular "chided" within living memory, while a workhorse verb like "have" hasn't budged in any traceable way and shows no realistic prospect of doing so.

Look closely at today's irregular verbs and you can catch several of them mid-transition, not yet fully settled one way or the other. "Learnt" survives as the everyday past tense in Britain while Americans overwhelmingly write "learned"; the same split plays out with "spelt"/"spelled" and "dreamt"/"dreamed," none of which represents an error so much as two dialects landing on opposite sides of an ongoing shift. "Dive" tells the opposite story, and it's worth pausing on precisely because it runs against the general regularizing trend: originally a perfectly ordinary weak verb ("dived"), it picked up a competing irregular past tense, "dove," sometime in the 1800s by analogy with existing strong verbs like "drive"/"drove" — and "dove" has since become the more frequent choice in American speech, even though it's moving toward irregularity rather than away from it. Regularization is the dominant direction verbs travel over time, in other words, but "dive" is a working reminder that it isn't the only direction available.

The net effect is a language carrying two systems from two different historical eras, unevenly worn down by a thousand years of sound change and frequency-driven survival, which is exactly why irregular verbs feel arbitrary at first glance and turn out to have real, traceable structure once you know where to look for it.

Prefixed verbs inherit their root's irregularity automatically

A large share of the roughly two hundred irregular verbs in modern use aren't base verbs at all, but compounds built by prefixing an already-irregular root: understand and withstand both conjugate like "stand" (stood), overthrow and forgo both conjugate like go (went, gone), and mistake and undertake both conjugate like take (took, taken). This isn't coincidence — English productively builds new verbs by attaching a prefix (over-, under-, mis-, out-, with-, be-, re-) to an existing verb, and the resulting compound always conjugates identically to its root, without exception. Recognizing the root buried inside an unfamiliar compound verb is often the fastest way to correctly guess its past tense without needing to look it up at all — if you already know "stand" is irregular, you already know "understand," "withstand," and every other stand-based compound is irregular the exact same way.

Suppletion: when the past tense comes from an entirely different verb

A smaller, stranger group of irregular verbs doesn't fit the ablaut-vowel-change pattern described above at all, because their past tense isn't built from the present-tense stem in any way — it's borrowed wholesale from a different, historically unrelated verb, a process linguists call suppletion. The clearest case in English is go/went/gone: "went" began its life as the past tense of an entirely separate verb, "wend" (meaning to travel or proceed, which survives today only in the fossilized phrase "wend your way"), and gradually took over as the past tense of "go" once "go" itself lost its own native past-tense form. "Be" pushes suppletion about as far as it can go: its full paradigm — am, is, are, was, were, been, being, every single one of it — is actually stitched together from three separate ancient verb roots that have nothing to do with each other, one supplying am/is, a second supplying are, a third supplying was/were, all now operating as if they'd always belonged to a single verb. Nothing else in the language's conjugation system draws from that many genuinely unrelated historical sources at once.