There are two hundred English verbs that refuse to follow the rules. It is the only thing they have in common. When a verb wants to talk about the past, English hands it one of exactly two options: add -ed and move on, or take a detour through history and emerge as a different word entirely. Two hundred verbs chose the detour. Nobody asked them to. Nobody benefits.
This article is about those verbs, and about a single question that English has avoided for a thousand years: what are you for?
The Tax Bill
Let us be precise about the cost. Across the full landscape of modern English, there are roughly 200 common irregular verbs. Most English learners—the ones trying to pass a Chinese high school exam, for instance—encounter 142 of them as mandatory. That set breaks down as follows:
| Pattern | Example | Count | New Words to Memorize |
|---|---|---|---|
| AAA (all three forms identical) | cut–cut–cut | 14 | 0 |
| AAB (base = past ≠ participle) | beat–beat–beaten | 1 | 1 |
| ABA (base = participle ≠ past) | come–came–come | 5 | 5 |
| ABB (past = participle ≠ base) | bring–brought–brought | 88 | 88 |
| ABC (all three different) | begin–began–begun | 30 | 60 |
| Special (be, go, etc.) | be–was/were–been | 4 | — |
Total memorization burden for the core set: 154 distinct past-form and participle-form words. Each one must be learned individually, because no rule predicts them. You cannot look at bring and deduce brought. You cannot look at sing and derive sang and sung. You either know them or you do not.
If every one of these verbs followed the -ed rule, the number of additional words a learner would need to memorize would be exactly zero. Past and participle forms would be mechanically predictable from the base. You learn the base, you learn the rule, and you own every past tense in the language.
So What Do We Get in Return?
Here is the defining test of any feature of a language: does it carry meaning? If a feature encodes a distinction that would otherwise be lost, it has a purpose. If it encodes nothing, it is noise. Noise is not harmless. Noise is a tax paid in childhood hours and learner frustration, collected every year, worldwide, by approximately one and a half billion people who are trying to learn English.
Does brought carry any meaning that bringed would not?
To answer this, imagine a parallel-universe English where all 200 verbs regularized five hundred years ago. In that universe, a speaker says: Yesterday I bringed my lunch to work, but I forgetted my keys. I thinked about calling home. I runned into a friend. We talked. I goed home.
Every thought in that sentence is perfectly understandable. No ambiguity. No loss of precision. A speaker of our universe might wince at goed, but they would not misunderstand it. The meaning is carried entirely by the base form. The past tense is signaled by the suffix. The irregular stem contributes exactly nothing.
This is the uncomfortable fact that English teachers do not want to say out loud: irregular past forms are semantically empty. They are fossils—hardened remains of earlier sound changes in Old English and Proto-Germanic that, at the time they occurred, were regular. The vowel shifts that produced sing–sang–sung were once predictable phonological rules. Then the rules died. The words stayed. Every English learner since has been paying the inheritance tax on an estate nobody lives in.
The Common Defense, Examined
Defenders of irregular verbs typically make one of three arguments. None of them survive contact with reality.
1. “They make the language richer and more expressive.”
Richer how? Does sang give you access to an emotional register that singed would deny you? If we had all grown up saying singed, it would sound perfectly natural, and sang would sound like a mistake. There is no inherent beauty in irregularity. There is only familiarity. The person who calls irregular verbs beautiful is the person who no longer has to learn them.
2. “They are the most common verbs. Changing them would be impossible.”
This is actually true—and it is an argument not for their value, but for their inertia. High-frequency items resist regularization precisely because we hear them so often that the irregular form gets reinforced before the rule can take over. This is the linguistic equivalent of saying “this law is too entrenched to repeal.” It may be factually correct. It does not make the law just.
3. “Children acquire them naturally. It is not a real burden.”
If that were true, five-year-olds would not say goed and runned and breaked. Over-regularization is one of the most robustly documented phenomena in child language acquisition. Children try to apply the rule first. They are corrected into the irregular. The irregular is an interference with the natural learning process, not a natural product of it. Every goed a five-year-old says is the language’s own immune system trying to clean up the mess. The adults stop it.
The Practical Arithmetic
For the Chinese secondary-school student preparing for the gaokao, the cost is 154 individual word-forms that must be memorized, none of which follow any productive rule. For the international learner, the cost is 216 past and participle forms across 200 verbs. For the native-speaking child, the cost is a year or two of correction and confusion—roughly ages four through six, during which almost every child produces regularized forms that adults then overwrite.
Here is the part that should make any reasonable person angry: the AAA verbs prove that irregularity is optional even among the irregulars. Put, cut, hurt, read—these verbs have irregular past forms that happen to be identical to their base forms. Nobody objects to them. Nobody is confused by them. If put-put works, why would bring-bringed be a disaster? The system already tolerates non-change. It is only the direction of the change that it polices.
What Would Regularization Actually Look Like?
Nobody is proposing that we issue a decree tomorrow declaring went illegal. Language does not work by decree. But we can do something simpler and more powerful: we can stop pretending the irregular forms are legitimate.
A grading rubric that deducts points for bringed is a grading rubric that treats historical accident as moral failing. A textbook that makes ten-year-olds chant sing-sang-sung, ring-rang-rung, drink-drank-drunk is a textbook that has confused memorization with education. An English test that hinges on whether the student remembers that strive is strove and not strived—while thrive does take thrived—is a test that has stopped measuring language ability and started measuring tolerance for pointless complexity.
The first step is not to abolish irregular verbs. The first step is to stop grading them. Accept thinked alongside thought. Accept teached alongside taught. Let the language decide which one it wants to keep, without the artificial life support of the red pen.
Chinese does not mark past tense on the verb at all. A single particle handles the entire domain of temporal reference. If the most widely spoken language on Earth can manage time without touching the verb, English can certainly manage it with one regular suffix instead of two hundred random ones.
The irregular verb is not a feature. It is a scar. And scars are not heritage—they are just where the wound was.