--- title: "Essential English Vocabulary Dropped Flour and Added Mortgage" description: "A 1953 list of essential English words and its 2023 replacement differ by 1,700 entries. The change tracks a shift from manual work to institutions." dek: "Two lists of the words a person needs to function in English, published 70 years apart, disagree about roughly 1,700 entries. The disagreement is not really about language." published_at: "2026-08-05T04:57:44.845Z" updated_at: "2026-08-05T04:57:44.845Z" tags: - "Labour" - "Science" source_url: "https://pudding.cool/2026/07/essential-words/" source_domain: "pudding.cool" canonical: "https://hex37.com/essential-english-vocabulary-dropped-flour-added-mortgage" --- *Bread* appears on both lists. *Flour*, *wheat*, *harvest* and *bake* appear on only one. The lists in question are the two most widely used inventories of essential English vocabulary — the words a person learning the language needs before anything else. The first, the General Service List, was published in 1953 and contained about 2,284 words. Its replacement, the New General Service List, arrived in 2013 and was revised in 2023, and contains 2,809. Between them, roughly 600 words were dropped, more than 1,100 added, and 1,656 kept. An [interactive analysis of the two lists](https://pudding.cool/2026/07/essential-words/) published this year lays the differences side by side. Some of the changes are exactly what you would guess. *Telegraph* left; *computer*, *website* and *blog* arrived. *Tobacco* was replaced by *cigarette*. *Motherhood* became *mom*, and *dad* was added, having never been there before. Nobody needs an explanation for those. The harder ones are *apple*, *fork*, *soap*, *umbrella* and *leaf*, none of which survived. *Dog* stayed; *goat* and *donkey* did not. *Cook* is on the new list, but *boil*, *bake* and *fry* are not. Meanwhile the arrivals include *mortgage*, *corporation*, *analysis*, *appropriate*, *fairly* and *despite*. The obvious reading is that this is a curriculum update — one committee's taste replacing another's, seventy years on. That reading is worth testing, because of how these lists are actually built. ## A word has to earn its way on Neither list is a set of opinions about which words are worthwhile. Both are frequency counts, assembled from a corpus — a large collection of real text and speech, gathered to be representative of how a language is actually used, and then counted. The field that does this is called [corpus linguistics](https://en.wikipedia.org/wiki/Corpus_linguistics), and its whole discipline is refusing to substitute intuition for measurement. A word gets on the list by turning up often enough, in enough different contexts, that an ordinary person could not plausibly avoid it. The design goal is compression: the fewest words that cover the most language. The 1953 list is estimated to account for about 84% of general English usage; the 2023 list, over 90%. To hit those numbers, a list has to shadow a large share of whatever people are genuinely reading and saying. That constraint has a second edge. The lists are deliberately small, because they exist to be taught, and classroom time is finite. Every word admitted is a claim on hours that some other word cannot have. *Wheat* and *mortgage* are not neighbours on a shelf; they are competitors for the same slot. When one displaces the other, something has been measured, not merely preferred. ## The categories closest to the body shrank Running every word through a standard linguistic tool that sorts vocabulary by meaning — assigning each to one of 21 subject areas such as Food and Farming, the Body and the Self, or Government and Public — produces a pattern that is hard to miss when the categories are arranged by how close they sit to the person speaking. The domains that describe your body, your immediate surroundings, your food and your tools lost ground. The domains that describe institutions, systems and pure abstraction gained it. Categories with no physical location at all — reasoning, processes, general concepts — grew the most. So the words did not scatter randomly. They moved in one direction, away from arm's reach. ## By 1957, most American workers were no longer making things with their hands Four years after the 1953 list was published, white-collar workers in the United States outnumbered blue-collar workers for the first time. By 2000, fewer than one in four workers did manual labour at all. This is the step that connects the economy to the dictionary, and it is a short one. What a person needs to talk about is set by what that person runs into. A working life spent with tools, animals, materials and food produces daily conversation full of tools, animals, materials and food. A working life spent inside institutions produces conversation about contracts, categories, procedures and claims. Nobody chooses this; you discuss what your day contains. An economy that reorganises what most people do all day therefore reorganises what most people say all day. Because the lists are frequency counts of what people say, they had no choice but to follow. The 600 departures and 1,100 arrivals are not a change of educational philosophy. They are the readout. ## Words you cannot hold cost more to learn The shift also shows up on a scale that has nothing to do with subject matter. Researchers have rated tens of thousands of English words for "concreteness" — how directly you can experience the thing through your senses, from 5 (fully) down to 1 (not at all). The most concrete words fell from 21% of the 1953 list to 14% of the 2023 one. That matters for anyone doing the learning. Psychologists describe concrete words as dual-coded: read *axe* and your mind retrieves both the word and a weight, an image, a gesture. Two routes to the same meaning. Abstract words have one route, through language alone. *Assumption*, *evaluation* and *involvement* arrive with no sensory anchor, which makes them harder to hold and slower to recall. Abstract words also need more scaffolding, and the new list supplies it. More than a hundred adverbs were added — *approximately*, *relatively*, *somewhat*, *precisely*, *definitely*, *virtually*, *presumably*. Almost all of them do the same job: they tell you how much of a statement is true, how often, and how sure the speaker is. *Axe* needs no such help. *Adequate* and *acceptable* are useless without it. ## What the list is a record of Read this way, an essential vocabulary list is not a neutral census of important words. It is a footprint. It records what a population had to deal with often enough that the language could not route around it — and the two footprints, seventy years apart, describe a life that has moved away from its own making. Less bound to tools, animals, food and the body. More bound to economies, regulations and institutions that operate at a distance and mediate almost everything, without ever occupying physical space. Each list also becomes a starting condition for the learners who get it. The 2023 list equips someone to sign a lease, contest a bill and read legislation. It also assumes they will rarely need to say *fork*. Whether that trade is a loss or simply the correct adaptation to the world as it is, the lists cannot tell you. What they can tell you is that it happened, and why. Language, as the record shows, keeps count whether or not anyone asks it to.