How the words we teach English language learners changed

Aug 02, 2026 10:41 PM - 4 hours ago 37

Look done the words scattered astir this page. Every 1 of them is among the astir commonly utilized words successful the English language.

They travel from a 2023 database of astir 2,800 words, shown to screen complete 90% of wide English use, intended for group learning the language.

This 2023 database is an update of an earlier one, made successful 1953, which identified astir 2,300 words arsenic the basal vocabulary to mundane life astatine the time.

Between the 2 lists, 70 years apart, astir 600 words were dropped, and complete 1,100 were added. The remainder remained arsenic is.

Some of the changes make contiguous sense: Telegraph dropped out; computer was added, on pinch website and blog. Tobacco was replaced by cigarette. Motherhood became mom, and dad was added too, though fatherhood was ne'er connected the database to statesman with. The world changed and vocabulary surely followed.

But also: apple didn’t make the caller list. Neither did fork, soap, umbrella aliases leaf, for example. It’s not that these things vanished from mundane life, but galore hands-on words became little cardinal to the halfway vocabulary. Dog stayed; goat and donkey didn’t. Bread stayed; breadmaking ingredients—flour and wheat—dropped. Cook is connected the caller list. Boil, bake, and fry are not.

And galore of the words that were added, specified arsenic mortgage, corporation, appropriate, analysis, fairly, and despite, don’t look thing for illustration the ones that were discarded. In fact, they are mostly absurd concepts that don’t look for illustration thing astatine all.

How the Words We Teach English Language Learners Changed, and What That Says About Us

By Jasmine Nackash

These “essential vocabulary” lists are called the General Service List (1953) and the New General Service List (2013, revised successful 2023).

They were designed arsenic school devices for group learning English arsenic a 2nd language, built from real-world usage information and extensively tested. The purpose was a vocabulary database pinch arsenic fewer words and arsenic overmuch sum of mundane English usage. That sum is high, complete 90% for the 2023 list1 and astir 84% for the 1953 list.2 To relationship for that overmuch of the language, they had to way a important information of immoderate group were really reference and saying. A connection earned its spot by appearing often enough, crossed capable contexts, to beryllium difficult to debar for the mean personification successful an English-speaking society. Open the connection lists sheet connected the correct to browse done each the entries successful some lists.

While these were applicable tools, built to seizure which words group request most, the reply besides doubles arsenic a snapshot of mean life, 70 years apart: what group were expected to prosecute with, and had to woody with, successful their regular lives.

Treating the lists arsenic an indirect grounds of day-to-day life, I went done the differences betwixt them from a fewer angles: what the words were about, really tangible were they, and to which parts of reside they belonged.

I started by utilizing a communal linguistics investigation tool3 that sorted each of the words by meaning. It assigned each connection to 1 of 21 taxable categories, based connected emblematic usage: Food and Farming, the Body and the Self, Government and Public, Language and Communication, and truthful on. Together, they suggested what benignant of world each database was built for.

Data source: each words tally done the UCREL Semantic Analysis System (USAS).

The categories that shrank are mostly those that person to do pinch the immediate, beingness world, while the gains are those furthest from it.

To amended spot this, the categories tin beryllium oriented spatially. Some categories picture you: your assemblage and emotions. Some picture what’s instantly astir you: food, objects, and the earthy world. Others sanction systems you participate in: government, institutions, society, and culture. And immoderate person nary location astatine all: absurd concepts, reasoning, and processes. When we look astatine the information grouped by physical scope, the shape of alteration seems to constituent successful a circumstantial direction.

Data source: each words tally done the UCREL Semantic Analysis System (USAS), past grouped into 5 umbrella-term domains. Expand to position the afloat class breakdown.
  1. The Self: Emotion | The Body and the Individual
  2. Local/Immediate: Substances, Materials, Objects and Equipment | Food and Farming | Life and Living Things | Architecture, Housing and the Home | World and Environment
  3. Institutional: Money and Commerce successful Industry | Government and Public | Education | Science and Technology
  4. Social/Communicative: Social Actions, States, and Processes | Movement, Location, Travel and Transport | Language and Communication | Entertainment, Sports and Games | Arts and Crafts
  5. Universal/Abstract: General and Abstract Terms | Psychological Actions, States and Processes | Numbers and Measurement | Names and Grammar | Time

In hindsight, the displacement makes sense. By 1957, 4 years aft the original database was published, white-collar workers outnumbered blue-collar for the first clip successful US history. And by 2000, fewer than 1 successful 4 workers did manual labor. The worldly group encountered successful their regular lives, what they needed to talk about, and the systems they had to navigate each changed.

The vocabulary lists, built from the connection of their respective eras, tracked those changes. The shifts bespeak a life that moved further from its ain making: little tied to tools, animals, food, and the body; much tied to nationalist aliases world institutions, categories, systems, and ideas.

The caller vocabulary—mortgage, legislation, perspective, involvement, improvement, assumption, evaluation—are words you can’t weigh, constituent to, aliases clasp successful your hand. These words style your life, but they do it without occupying beingness space.

To amended understand whether location was a displacement successful vocabulary describing the beingness world, I compared each connection to a database that rates its tangibility connected a standard of 1 to 5 (called a “concreteness rating4”). A standing of 5 intends you tin acquisition the connection straight pinch your senses, while a standing of 1 intends you can’t.

The highly actual end dropped the most: from 21% of the full successful 1953 down to 14% successful 2023.

Kernel density estimation, bandwidth 0.08 · Data source: Brysbaert et al. (2014)

This displacement matters because absurd and actual words are processed by our brains successful different ways. When you publication axe, your encephalon doesn’t conscionable decode letters, it reaches for something: an image, a weight, aliases a gesture. The connection activates some a verbal explanation and a sensory trace. Psychologists telephone this dual coding:5 Concrete words recreation done 2 channels, verbal and sensory; absurd words recreation done one. Two channels mean 2 retrieval pathways, which is why actual words are “stickier,” easier to clasp successful a statement of thought and faster to recall. Abstract words, connected the different hand, are purely verbal, and person to beryllium understood done connection alone.

To put it different way: Concrete words are easier for america to process because they are bundled pinch a web of associations, tactile experiences, and memories that anchor their meaning. Here’s a much elaborate position of the displacement distant from actual language:

Data source: Brysbaert Concreteness ratings for 40 1000 mostly known English connection lemmas. The dataset provided 99.8% sum of the words successful some lists. 6 words not successful the dataset are not included successful this chart: as, dialog, english, gaiety, madden, old-fashioned.

While actual words person sensory grounding to transportation their meaning, absurd words trust connected different parts of reside to specify, soften, aliases sharpen what they mean. That thief tends to travel from 1 peculiar area of the language: adverbs.

The 2023 database contains much words wide (2,809 vs. 2,284). All changes mentioned successful the matter bespeak each category's stock of its list, not earthy counts. Data source: NLTK (Natural Language Toolkit), pinch manual correction of mislabeled words.

Axe doesn’t request an adverb to modify it; you cognize what it is. But acceptable, relevant, and adequate travel pinch conditions, qualifications, and degrees that request to beryllium spelled out. Adverbs do precisely that: connection to calibrate language.
Look done each the adverbs that were added:

absolutelyactuallyalongsideanymoreapparentlyapproximatelyautomaticallybadlybarelybasicallybrieflycarefullycertainlyclearlycloselycompletelyconsequentlyconstantlycurrentlydeeplydefinitelydifferentlydirectlydramaticallyeasilyeffectivelyentirelyequallyeventuallyexactlyextremelyfairlyfaithfullyfinallyfirmlyfirstlyforeverfrequentlyfullyfurthermoregenerallygentlygraduallygreatlyheavilyhencehighlyhopefullyimmediatelyincreasinglyinitiallylargelyliterallymainlymerelymonthlymostlynaturallynearbynearlynecessarilyneverthelessnewlynormallyobviouslyoccasionallyoriginallyoverseasparticularlypartlyperfectlypersonallypossiblypotentiallypreciselypresumablypreviouslyprimarilyprobablyproperlyquicklyquietlyrapidlyrarelyreallyreasonablyrecentlyregardlessregularlyrelativelyrespectivelyroughlysecondlyseriouslysignificantlysimilarlysimplyslightlyslowlysomewhatspecificallystronglysubsequentlysuccessfullysuddenlysurelysurprisinglytotallytrulytwicetypicallyultimatelyunfortunatelyusuallyvirtuallywidely

Most of the adverbs specify degree, frequency, certainty, and extent. Some hedge (somewhat, partly, relatively, possibly, approximately). Others asseverate (absolutely, definitely, entirely, exactly, precisely). They're each doing the aforesaid benignant of work: Take a connection and show you really overmuch of it is true, really often, and really certain. It’s arsenic if the world now requires you to beryllium much precise astir everything.

Bread survived some lists. Flour, wheat, harvest and bake didn’t. The connection for what sustains america remained essential, while the words for really we’d make it weren’t. That mightiness beryllium the astir honorable summary of what happened.

The world that made the 2023 database is much regulated, much connected, and successful galore ways much tin than the 1 down the 1953 list. It’s a world further than our room aliases home, reaching crossed economies, institutions, and democracies.

Today's vocabulary reflects a life that isn’t self-contained, but alternatively much systemic. It’s not truthful overmuch astir what’s wrong arm’s reach, but much astir the larger world we navigate through. That benignant of long-distance relationship requires a peculiar benignant of language: expansive, abstract, and precise. And language, it turns out, can’t thief itself. It keeps track.

Methods & Notes

I compared 2 salient vocabulary lists for English learners: the General Service List (GSL, 1953; 2,284 words) and the New General Service List (NGSL 1.2, 2023; 2,809 words). I branded words appearing connected some lists arsenic “remained” (1,656), words only connected the 1953 database arsenic “removed” (628), and words only connected the 2023 database arsenic “added” (1,153).

The GSL words came from the Simple English Wiktionary GSL. The NGSL words came from the official NGSL 1.2 file (“alphabetized and lemmatized for research”). The NGSL uses lemmas (one introduction per connection family); the GSL sometimes lists inflected forms arsenic abstracted headwords. These lists way connection forms deemed worthy school based connected wave and usefulness, not absurd concepts. For example, the connection being is connected the GSL arsenic its ain headword and was not included successful the NGSL, But this doesn’t mean the conception of beingness near the language. In the NGSL it falls nether be, which is connected some lists.

Why dainty the lists arsenic a image of mundane English?

Both lists were built for teaching, but outer investigation suggests each covers a ample stock of mundane connection use. About 84% of wide English for the GSL and astir 90% for the NGSL, depending connected the matter and really words are counted. I did not re-run those corpus analyses myself. I return the published materials arsenic fixed and trust connected sum figures from the database authors and from follow-up studies (including an independent cheque connected American English by Stoeckel, 2019 – spot footnotes). That’s why the differences betwixt the lists felt worthy examining arsenic much than a program update, pinch the caveat that neither database is simply a neutral census of culture.

Where tin I find the data?

All tagged words: remained, removed, and added, pinch semantic tags, concreteness ratings, and part-of-speech labels are successful this nationalist spreadsheet. You tin besides browse the connection lists successful the word-list sheet connected the correct broadside of this page.

How did I benignant words by meaning?

Each connection was tagged pinch the UCREL Semantic Analysis System (USAS), utilizing the 21 top-level categories. USAS besides assigns overmuch finer sub-categories (there are hundreds), but I stayed astatine the apical level truthful the charts could show wide shifts without splitting the words into overly granular bins.

I chose not to correct mislabels. For example, hammer, nail, and wax are each tagged “General and Abstract Terms”, but successful USAS’s finer tags, they publication arsenic actions (“to hammer,” “to nail,” “to wax”), not objects. Out of context, galore of these words spell aggregate ways, and it didn’t consciousness correct to override that lawsuit by case; USAS is an established linguistic framework, and swapping successful my ain judgement would operation 2 different standards.

For the 2nd chart, I grouped USAS’s 21 categories into 5 “scope” domains (self, local, institutional, social, abstract). That grouping is my editorial choice, not portion of USAS. It came from noticing a spatial value to the trends seen crossed the 21 categories.

How did I measurement concreteness?

Concreteness ratings travel from Brysbaert et al. (2014). I utilized these as-is. Six words weren’t successful the database and were near retired of the concreteness charts: as, dialog, english, gaiety, madden, and old-fashioned.

How did I tag parts of speech?

Parts of reside were tagged pinch NLTK, simplified to 5 categories. Here I did intervene, but only erstwhile a connection was intelligibly mislabeled (132 words, 3.8% of the list; mostly adjectives mislabeled arsenic nouns). When a connection tin enactment arsenic much than 1 portion of reside depending connected context, I near the tag as-is and deferred to NLTK arsenic the established framework, utilizing the primary, astir communal label. Here I besides utilized an LLM strictly to thief emblem imaginable errors successful NLTK’s output.

What are the limitations?

Both lists rank words by frequency, past use learner-focused curation. Michael West’s 1953 database particularly reflects play pedagogy. He favored general-purpose vocabulary complete affectional aliases highly circumstantial words, not conscionable immoderate appeared astir often (Therova, 2020, summarizing West, 1953, pp. ix–x). Some of what looks for illustration “1950s life” whitethorn besides beryllium really mid-century ESL school filtered the language. I still dainty the lists arsenic a image of mundane English because some screen a ample stock of moving matter and reside that goes good beyond the classroom.

Footnotes
  1. On NGSL’s ~90% coverage: Browne, Culligan & Phillips, NGSL project. See also: A New General Service List: The Better Mousetrap We’ve Been Looking for?, Browne (2014); and An Examination of the New General Service List, Stoeckel (2019). ⏎
  2. On GSL’s ~84% coverage: A wide work database of English words pinch semantic frequencies, West (1953); The New General Service List: A halfway vocabulary for EFL students and teachers, Cambridge ELT (2018). ⏎
  3. UCREL: Semantic Analysis System (USAS). ⏎
  4. Concreteness ratings for 40 1000 mostly known English connection lemmas, Brysbaert, M., Warriner, A. B., & Kuperman, V. (2014). Behavior Research Methods, 46, 904–911. ⏎
  5. Why are pictures easier to callback than words? Paivio, A., Rogers, T.B. & Smythe, P.C. Psychon Sci 11, 137–138 (1968). ⏎
More