IELTS vocabulary: how many words you need, according to research

Why there is no "words for band 7" number

The study most often cited, by Milton, Wade and Hopkins, is summarised in Milton (2013): among 30 students on a pre-sessional course in the UK, a written vocabulary test correlated with the Reading score at r = 0.70. That is a real relationship on a very small group, measured with a test that tops out at 5,000 words. It says that people with larger vocabularies tend to score higher. It does not give a threshold.

A larger study points the same way with more caution. Drummond (2018) tested 205 foundation-year students and found a correlation of .54 between vocabulary size and the Reading band. Students with the same band differed enormously from one another, and the author's conclusion is that an IELTS score is a poor predictor of vocabulary size. If the score does not predict the vocabulary, a vocabulary number cannot promise the score.

What is known: how many words cover IELTS texts

Ha, Le, Phung and Nguyen (2022) ran 186 Reading texts from the Cambridge practice-test books through word-frequency lists. For Academic Reading, the most frequent 5,000 word families covered 95% of the running words, and 9,000 families covered 98%. For most General Training texts the figures were 3,000 and 6,000; the final General Training text was close to Academic in vocabulary. A word family is a base word with its common forms, so analyse, analysis and analytical count once.

Coverage describes texts, not candidates. It is not a score threshold.

What 95% and 98% mean for a reader

At 95% coverage one word in twenty is unknown; at 98% it is one in fifty. Hu and Nation (2000) gave readers a story with different shares of unknown words: at 80% known words nobody understood it adequately, at 90% and 95% a few did, and their regression pointed to about 98% for comfortable unassisted reading. Laufer and Ravenhorst-Kalovski (2010), working with 745 students, proposed two thresholds: a minimal one at 95% coverage and 4,000 to 5,000 word families, and an optimal one at 98% and 8,000 families.

Two cautions. Schmitt, Jiang and Grabe (2011) tested 661 readers in eight countries and found that comprehension rises steadily with the share of known words, with no sudden cliff, and a 2023 replication of the 98% figure reproduced it only in part. And Nation (2006), who calculated that 98% coverage of written text takes 8,000 to 9,000 word families, is explicit that the 98% goal assumes reading without a dictionary or other help. With help at hand, a somewhat lower share is workable. In the exam you have no help, which is the argument for practising on texts you can almost fully read.

Academic words are the efficient target

Averil Coxhead's Academic Word List (2000) contains 570 word families that are frequent across academic subjects and fall outside the 2,000 most common English words. In her 3.5-million-word academic corpus they accounted for about 10% of all words, against 1.4% in fiction. The list is free on the Victoria University of Wellington site, split into sublists by frequency. The newer Academic Vocabulary List (Gardner and Davies, 2014), built from 120 million words of academic text, covers about 14%.

For IELTS Academic this is the best return on study time once everyday vocabulary is in place: a few hundred families that show up in every passage about research, evidence and argument, whatever the topic.

How reading builds vocabulary, and how much it takes

Meta-analyses of extensive reading find positive effects on vocabulary of moderate size: 0.47 in Jeon and Day (2016), alongside 0.83 for reading rate and 0.54 for comprehension. One book is not enough. Waring and Takaki (2003) tested learners after a single graded reader and found that most of the new words were not retained. Words stick when you meet them again and again.

Nation (2014) estimated how much reading it takes to meet the words of each frequency band about twelve times. His figures assume 150 words a minute, five days a week, forty weeks a year, and he calls the estimate optimistic.

Source: Nation (2014), Reading in a Foreign Language 26(2).

Set these against the coverage figures. The fifth thousand, where Academic texts reach 95% coverage, is by Nation's estimate a year of reading for about half an hour a day. The ninth thousand alone would take 100 minutes a day. That gap is the case for combining reading with targeted study of academic words.

A routine that follows from this

This routine is our suggestion, drawn from the research above. It is not an official method.

1. Read every day, mostly texts where you know nearly every word. Free sources of the right kind are listed on the page about what to read for IELTS Academic.
2. Look up a word when it blocks the meaning or when you have now seen it three times. Skip the rare ones you will not meet again.
3. Save the words you look up and review them, so that a word you met once gets the repeat meetings it needs.
4. Work through the Academic Word List sublist by sublist and tick off the families you already know from reading.
5. Test the result on official material. The four-week plan links to the free official tests.

Where Readlang Mobile fits

Readlang Mobile is an iPhone and iPad reading app. Tap a word in a book or an article and the translation appears in the line; every word you translate is saved to a vocabulary list without any extra step, and Premium adds practice with the saved words. That covers steps two and three of the routine. The free plan allows 30 translations a day, Premium is $8.99 a month or $59.99 a year.

What it does not have: the Academic Word List is not built in, vocabulary does not export to Anki or CSV, and there are no practice tests. It is a reading tool, and the exam practice belongs on the official sites.

Questions people ask