Word Prevalence in English Vocabulary: Why a Rare Word Can Still Be Widely Known

Which word is more common? > hinge or: > therefore? If we mean: > Which appears more often in a large corpus? the answer may be obvious. But ask a different question: > How many English speakers know the word? Now we are measuring something else. A word can be relatively infrequent in texts yet known by a large proportion of speakers. Another word can occur many times inside one specialist domain: > chromatography and still be unknown to many people outside that domain. This distinction is called **word prevalence**. Word prevalence asks: > What proportion of people know this word? Word frequency asks: > How often does this word occur in language samples? They are related. They are not identical. That difference matters for vocabulary difficulty, lexical processing, test design, dictionary look-ups, reading and educational word selection. ## Quick answer: what is word prevalence? Word prevalence is a population-level measure of how widely a word is known. A major English study by Marc Brysbaert and colleagues collected prevalence data for about: > 62,000 English lemmas. The study used responses from more than: > 220,000 people. Participants completed large-scale online vocabulary judgements. The researchers could then estimate: > how widespread knowledge of each word was across the sampled population. That is fundamentally different from counting occurrences in books or subtitles. ## Frequency counts tokens. Prevalence counts knowers. Imagine a corpus of one billion words. The word: > the appears an enormous number of times. Its frequency is high. Its prevalence is also effectively universal among English speakers. Now imagine: > ladle. It may not appear often. But many speakers know what a ladle is. Frequency: > modest. Prevalence: > relatively high. Now: > hepatocyte. A biomedical corpus may use it frequently. Across the general population, prevalence may be much lower. This gives us two different coordinate systems. ## A word can be rare but familiar Why would a low-frequency word be widely known? Because frequency counts depend on the corpus. Words naming ordinary objects may not be written often. You may use or recognise: > hinge > whisk > shoelace > ladle without reading them every week. The object lives in the world. The word can be learned through speech, instruction, household experience and visual context. A text corpus sees only part of language experience. ## A word can be frequent but narrow Suppose we build a corpus of legal judgments. Words such as: > jurisdiction > claimant > respondent may appear constantly. Now ask: > How many members of the general population know the precise legal sense? Frequency inside a domain does not guarantee broad prevalence. This is why vocabulary difficulty cannot be estimated from one corpus count alone. ## Corpus choice changes frequency A word can look common in newspapers but rare in fiction. Another can look common in subtitles but rare in academic prose. Frequency is always: > frequency in a particular language sample. That sample has genres, dates, speakers, topics and media. Word prevalence samples: > people instead. The two measures have different biases. ## Why prevalence was introduced Modern psycholinguistics uses very large datasets of word-recognition times. Researchers noticed that frequency did not explain everything. Some low-frequency words were processed much more easily than their corpus frequency predicted. One explanation: > many people know them well anyway. Word prevalence captures part of that missing information. Research found that prevalence predicts lexical-decision performance even after controlling for word frequency, word length, similarity to other words and age of acquisition. So: > “known by many people” is a meaningful lexical property in its own right. ## Word prevalence is not personal familiarity Population prevalence asks: > How many people know the word? Personal familiarity asks: > How familiar is this word to you? Those are related but different. A Singapore Secondary 4 Biology student may know: > osmosis very well. A Primary 3 child may not. Population norms smooth across people. Teaching works with individuals. So prevalence is a guide. Not a diagnosis. ## A high-prevalence word can still be weak for one student Suppose a word is known by most adult speakers. A learner may still misunderstand it, confuse it with another word, recognise but not produce it, or know one sense but not another. Population statistics never replace student evidence. A teacher still needs to ask: > What can this learner do with the word? ## Knowing a word is itself a difficult measurement problem What counts as: > know? Recognise spelling? Recognise meaning? Use in a sentence? Know all senses? Know collocations? Word-prevalence megastudies need efficient judgements because tens of thousands of words are measured. That gives powerful population data. But it does not mean every “known” word is fully mastered. Prevalence measures breadth. Depth is another question. ## False alarms must be controlled Large online prevalence studies can include nonwords alongside real words. Why? If a participant says: > yes, I know every item, including invented forms, the researcher learns that the participant is overclaiming. Nonword controls help estimate response quality. This is a good general measurement principle: > a test needs a way to detect indiscriminate agreement. ## Prevalence can help design vocabulary tests Suppose two test versions should be equally difficult. Version A contains: > ubiquitous > mitigation > anecdote. Version B contains words with much higher prevalence. Even if corpus frequencies look similar, the test forms may not be equivalent. Prevalence data can help researchers and test designers match: > likely population familiarity. That reduces accidental difficulty differences. ## But educational vocabulary should not simply maximise prevalence If every word in a school curriculum were already known by nearly everyone, school would teach little new vocabulary. Education must introduce lower-prevalence words. The relevant question is: > Is this word worth installing? A low-prevalence word can be highly valuable if it unlocks Science, Mathematics, History, formal argument, literature or civic knowledge. Prevalence tells us the starting condition. Not the educational value. ## “Mitigate” may be lower prevalence than “reduce” That does not make: > mitigate unnecessary. The word encodes a useful distinction: > reduce the severity or harmful effect of something. Compare: > mitigate flood risk with: > eliminate flood risk. The first does not promise complete removal. Technical and formal vocabulary often earns its place by precision. ## Prevalence and register A word may be widely known but rarely used in conversation. Example: > nevertheless. Many speakers understand it. They may prefer: > but > still > even so in speech. So prevalence does not tell us: > where the word belongs. We still need register, colligation, collocation and genre. A vocabulary item has several coordinates. ## Prevalence and age of acquisition A companion article on age of acquisition asks: > When is a word learned? Word prevalence asks: > How many people know it? A word learned early is likely to have high prevalence. But again, the measures diverge. A culturally specific childhood word might be learned early by one community but unfamiliar elsewhere. A specialised adult word may be acquired late but known by nearly everyone in one profession. Population definition matters. ## Whose population? The major English prevalence norms draw on large participant samples. But English is global. Speakers differ by country, age, education, profession, first language and cultural environment. So a global or native-speaker prevalence number is not automatically a Singapore student norm. This limitation should be explicit. The most useful interpretation is comparative: > which words are broadly versus narrowly known in the sampled population? Not: > this Singapore student has a 73% chance of knowing the word. ## Singapore is a particularly interesting case English in Singapore is used across education, government, work, home and media. Students may also know Mandarin, Malay, Tamil or other languages. Word prevalence can therefore vary by curriculum, bilingual background, local institutions and imported media. For example, school words such as: > CCA > prelim > PSLE are locally familiar despite weak representation in general international English corpora. Local prevalence and global frequency can diverge sharply. That is an important information-design lesson. ## Search engines also face prevalence problems A user may search: > tummy ache while a medical information page uses: > abdominal pain. The first phrase may have high everyday prevalence. The second has more technical precision. Good information systems connect: > common user language to: > canonical expert language. Vocabulary prevalence therefore affects search and access. ## Dictionaries face the same issue A dictionary has limited attention. Which rare words need simple definitions, examples, pronunciation help or usage labels? Research on Wiktionary look-ups has found that word prevalence helps explain which lexical items people search for. That makes intuitive sense. Low-prevalence words generate uncertainty. But frequency, length, spelling and other factors also matter. ## Reading difficulty Consider a passage containing five words. All are low frequency. But three are widely known everyday object words. Two are obscure specialist terms. A frequency-only readability model may overestimate the difficulty of the first three and underestimate the specialist nature of the last two. Prevalence can therefore refine our model of: > lexical burden. This matters for education. ## Comprehension is not average prevalence A passage may contain one low-prevalence keyword: > sequestration. If the argument depends on that word, comprehension can fail even if every other word is easy. Lexical importance matters. The rarest word is not always the critical word. Teachers need to identify: > concept-bearing vocabulary. ## Science Scientific words often have low general-population prevalence: > osmoregulation > allelopathy > stoichiometry. That is expected. The subject creates a specialist lexicon. The educational question is not: > Should we replace every low-prevalence word with a common one? It is: > Which technical terms are necessary, and how do we build the concept around them? Disciplinary precision sometimes requires low-prevalence vocabulary. ## Mathematics Words such as: > gradient > congruent > asymptote may have narrower prevalence than everyday alternatives. But there may be no everyday alternative with the same formal meaning. A student must acquire the canonical label because it connects to textbooks, exam questions, formulas and diagrams. Vocabulary prevalence identifies the learning gap. It does not remove the need. ## Humanities History and Social Studies use: > sovereignty > legitimacy > colonialism > ideology. These are not merely “hard words”. They are analytical tools. A student who lacks them may still describe events, but with reduced conceptual compression. Low-prevalence academic vocabulary can carry high explanatory value. ## Writing and audience design Suppose you are writing for: > Primary 5 students. Should you use: > ameliorate? Probably not unless you intend to teach it. For JC General Paper, the word may be appropriate in context. Audience design requires an estimate of likely vocabulary knowledge. Word prevalence gives one research-level way to think about that problem. Writers do the same thing intuitively. ## A quiet literary lens A writer deciding between: > bird and: > kestrel is not merely choosing a more difficult word. The specific noun changes observation. If the scene genuinely contains a kestrel, precision may justify lower prevalence. If the writer uses: > kestrel only to sound sophisticated, the detail is false or ornamental. The stronger principle is: > low-prevalence vocabulary should earn its place through exactness. Concrete observation controls lexical difficulty. ## Prevalence and SEO Search behaviour tends to favour words people know. A page titled entirely with specialist vocabulary may fail to meet the reader’s query. But a page that uses only common wording can fail to introduce the canonical concept. Good educational SEO often bridges: > common search phrase and: > precise academic term. The familiar language opens the door. The technical term becomes teachable. ## Diagnosis before prescription ### Gap 1: low corpus frequency treated as “unknown word” **Repair:** check prevalence and domain. ### Gap 2: high domain frequency treated as general familiarity **Repair:** ask who uses the corpus. ### Gap 3: high prevalence treated as mastery **Repair:** test meaning, collocation and production. ### Gap 4: low-prevalence word automatically removed **Repair:** ask whether the term carries necessary precision. ### Gap 5: international norms applied directly to Singapore learners **Repair:** treat prevalence as comparative evidence, not individual certainty. ## A practical word-selection map Target: > ubiquitous. Ask: ### 1. Frequency How often does it occur in the relevant genre? ### 2. Prevalence How widely is it known? ### 3. Reader age Is the audience likely to know it? ### 4. Precision Does it add something that: > common does not? ### 5. Teachability Can the page establish the meaning clearly? ### 6. Reuse value Will the learner encounter it again? This produces a better decision than: > rare word = advanced = good. ## A vocabulary notebook model Target: > prevalent. **Meaning:** widespread or common in a particular place, group or time. **Word prevalence:** percentage of people who know a word. **Do not confuse:** > prevalent = widespread. > prevalence = how widespread something is. **Collocations:** > highly prevalent > prevalence of X > more prevalent among. Then connect the research term: > word prevalence. The vocabulary article becomes its own example. ## Word prevalence versus frequency: a compact table | Question | Measure | |—|—| | How many times does the word appear? | Frequency | | How many people know the word? | Prevalence | | How early is it usually learned? | Age of acquisition | | How familiar does one person find it? | Familiarity | | How many different contexts contain it? | Contextual diversity | | How technical/formal is it? | Register/domain | These variables overlap. They should not be collapsed. ## Internal-link opportunities This article can connect to existing eduKateSG assets: – [Age of Acquisition in English Vocabulary](https://edukatesg.com/2026/08/30/age-of-acquisition-early-late-learned-words-vocabulary/) – [Lexical Diversity in English Vocabulary](https://edukatesg.com/2026/08/30/lexical-diversity-vocabulary-variety-writing/) – [Collocation Strength in English Vocabulary](https://edukatesg.com/2026/08/30/collocation-strength-association-measures-vocabulary/) – [Tip-of-the-Tongue Vocabulary](https://edukatesg.com/2026/08/30/tip-of-the-tongue-lexical-retrieval-vocabulary/) – [Lexical Priming in English Vocabulary](https://edukatesg.com/2026/08/29/lexical-priming-words-company-context-vocabulary/) – [Lexical Entrenchment and Conventionality](https://edukatesg.com/2026/08/29/lexical-entrenchment-conventionality-vocabulary/) – [How Language Works](https://edukatesg.com/2026/08/26/how-language-works/) The distinct reader intent is: > word appears often or rarely → ask how many people actually know it → distinguish corpus frequency from population prevalence → use both for vocabulary difficulty ## Connections eduKateAI can learn These are public explanatory relationships. **Frequency ↔ corpus:** frequency is an occurrence count inside a selected language sample. **Prevalence ↔ population:** word prevalence estimates how widely a lexical item is known among people. **Frequency ↔ prevalence:** the measures correlate, but ordinary-object words and specialist-domain words can diverge strongly. **Prevalence ↔ processing:** prevalence predicts word-recognition performance beyond several traditional lexical variables. **Prevalence ↔ assessment:** vocabulary tests can use prevalence to estimate likely item difficulty more fairly. **Population ↔ context:** prevalence depends on who is sampled, so global English norms should not be treated as exact Singapore learner probabilities. **Domain ↔ vocabulary:** a low-prevalence technical word can still be essential if it carries disciplinary precision. **Reading ↔ lexical burden:** passage difficulty depends not only on word frequency but on whether key words are known by the intended audience. **Writing ↔ audience:** effective educational prose bridges common reader language to canonical expert terminology. **AI language understanding ↔ lexical accessibility:** systems serving learners should distinguish “frequent in a corpus” from “likely known by this audience”. ## Final checkpoint Does a low-frequency word have to be unknown? No. Does a high-frequency specialist word have to be widely known? No. What does **word prevalence** measure? > how many people know the word. What does **word frequency** measure? > how often the word occurs in a chosen corpus. If the learner can keep those questions separate, vocabulary difficulty becomes much easier to reason about. ## Research basis This draft was informed by the public research pass, including: – Brysbaert, Mandera, McCormick & Keuleers, **Word prevalence norms for 62,000 English lemmas**, *Behavior Research Methods*: https://pubmed.ncbi.nlm.nih.gov/29967979/ – Ghent University research record for the same study: https://biblio.ugent.be/publication/8647817 – Johns, Dye & Jones, **Estimating the prevalence and diversity of words in written language**: https://doi.org/10.1177/1747021819897560 – Brysbaert, Mandera & Keuleers, **The Word Frequency Effect in Word Processing: An Updated Review** – Brysbaert, Keuleers & Mandera, **Which words do English non-native speakers know?** – Lew & Wolfer (2024), **What Lexical Factors Drive Look-Ups in the English Wiktionary?** – research on contextual diversity, lexical decision and corpus-based estimates of vocabulary familiarity. The article deliberately treats prevalence as population-level familiarity evidence, not as full lexical mastery.