A page is a count of words, and a list of words either has each one or does not. On a passage from a novel, the 2,800 words of the list cover 91.0% of what is printed, and 94.3% once names are set apart. A reader needs 98%. The gap is not a feeling about difficulty: it is a list of words you can read off the page, and this lesson shows you how to find it for any text.
Core ideas
Three ideas. The third is the one most people get wrong.
The passage has 936 words and 361 different ones. If the list has the, then all 26 times that the appears count as known. This is why a short list of common words does so much: the first 1,000 words cover 85.0% of the passage, and the next 1,000 add only 4.2 points, to 89.2%. The rest of the list adds 1.8 more, to 91.0%.
A word in capitals in the middle of a sentence, such as Darcy or Meryton, is a name, and nobody looks a name up. The lab sets names apart. With them the passage is covered 94.3%, the play 94.6% and the modern documents 92.7%. This is the fairer figure to hold against a reader's 98%.
Nation (2006) puts the share a reader needs for an easy read at 98%, which is one unknown word in every 50. The passage has 936 words, so 98% would leave about 19 words unknown. After names and families, the lab finds 41. The words that make up the gap are printed, and a list of 41 can be learned.
The material
The table under the numbers lists every word the lab could not cover. Read it once through before you read the next section.
How much of a page do you know? Most people answer with a feeling. This lesson answers with a count. The lab takes a printed text and asks one question of every word: is it on the list? The list is the 2,800 headwords of the New General Service List. A headword is the word the list files the others under: walk for walks, walked and walking. The answer is a share, and the share is the coverage of the list on that text.
The list comes in three bands. The first holds the commonest 1,000 headwords and, beside them, the numbers, the days and the months, 1,050 in all; the second holds the next 1,000, and the third the last 809. That is 2,859 headwords, which this Subject calls 2,800 everywhere. On the passage, 936 words from the end of a chapter of Pride and Prejudice, the first 1,000 cover 85.0%, the first 2,000 cover 89.2%, and all 2,800 cover 91.0%. On the part of the play printed here, 1,972 words, the same steps give 79.5%, 83.9% and 85.9%. On the two modern documents, 1,685 words in all, they give 76.0%, 84.7% and 87.3%.
The list holds headwords, and a page holds forms. The lab closes the gap with the rules of the earlier courses, run backwards. It takes -s, -ed and -ing off a verb, with the spelling changes put back; it takes a plural off a noun, the irregular ones too; it takes -er, -est and -ly off a word; it turns n't into not; it files him and his under he, as the list does; and it knows the irregular verbs. If the word that comes out is on the list, the form is covered. The menu calls the passage 936 words, the count this lab reads, with don't and aunt's one word each.
Does this work? It was checked away from this page, against the full list of 10,296 forms the list records. The rules take 9,925 of them back to a headword, and 9,920 of those to their own headword. The 371 they miss are of six kinds: 152 British spellings (colour, centre, words in -ise); 74 spoken forms and words cut short (gonna, comin); 64 number words (fourth, twentieth); 17 old or irregular forms (cometh, bade, borne, farther); 11 plurals from Latin and Greek (crises, phenomena); and 53 wrong spellings and the rest (arguement, e-mail). The rules also reach words the list does not record, so the lab gives a little more cover than looking each word up in the list of forms would: 91.0% against 90.8% on the passage and 87.3% against 87.1% on the modern documents. On the play the gap is wider, 85.9% against 83.6%, and most of it is don't and the other short forms with an apostrophe: the forms list does not record them, and the rules take them apart. One extra hit is wrong: Worthing, a family name, is read as a form of worth six times.
The lab asks about the list first, so a listed word that happens to carry a capital is covered, not a name. A word the list lacks is a name if its capital letter does not start a sentence. So is a word in capitals all through, which is how the play prints the name of each speaker, and a word that stands with a capital in the middle of a sentence somewhere else in the text, even where it opens a sentence. Mr and Mrs fall in this set too. The capital is the whole test, and the lab cannot tell a name from a word that happens to have one. Read the passage's 31 names and you find Lord, from Oh, Lord!. In the play, Ahem, a sound written as a word, is a name three times.
Now compare the share with names with the 98% a reader needs. For the passage the gap is 5.7 points at 94.3%. The play is at 94.6% and the modern documents at 92.7%. All three fall short, and the table says which words make the difference. The words are not the same in each. Read the three tables: the passage lacks words of manner, such as affection and gallantry; the play lacks words of feeling, such as darling and passionately, and its biggest single miss is the word of surprise it says fourteen times; and the modern documents lack the words of two trades, law and population counting.
Choose “a text of your own” and paste a page. Below fifty words the lab says so and counts nothing. Above it, you get the same tiles and the same table. The words under not covered are the ones to learn first for that page, and the ones that appear most often come first.
Interactive
Every word of the text checked against the list, by the rules of the nine courses before
Method
Five steps, in this order.
Look at the first number. A share of a short text moves a lot when one word changes, which is why the lab asks for fifty words or more.
The first 1,000 words, the first 2,000 and all 2,800. Each step adds less than the one before, and that is how the list was made: the commonest words first.
This is the figure to hold against 98%. If it is above, you can read the text without looking words up. If it is below, the next step says how far.
Take the share from 100 and divide 100 by what is left. At 94.3% the gap is 5.7, and 100 divided by 5.7 is about 18, so about one word in 18 is a word you may not know. At 98% it is one in 50.
Sort them into kinds before you learn them. A word that appears twice is worth more than a word that appears once.
Here are all 41 words from the table under the passage, sorted into seven kinds by what they have in common. The table prints 40 rows, because oh appears twice.
Ten, five, eleven, two, four, six and three make 41, the number the lab prints. The first kind is a matter of the year: words of manners and feeling in a novel from 1813. The second is a matter of the setting: a country house and a church. The third is the one to read with care. The table lists forms, not headwords, so one missing verb can fill two rows: provoke and provoked, and the same for sermon and for interrupt. For provoked and sermons the rules would reach the second row if the first were on the list; for interruption they would not even then, because the lab cuts -tion, not -ion. None of the three is on the list, so every row stands. Learn the eleven as nine words. Two words are only a British spelling of a listed word (honor, humor). Four are made from listed words by a cut the lab does not make: after with an ending, condition with two, separate with one, and twelve and month joined. Six are formal words whose roots are not listed either. Oh is a word said in surprise, and de is the small word in the middle of one name, with no capital to mark it.
Now read the other two tables the same way. The play has 91 words not covered in 1,972 words, and the modern documents have 106 in 1,685. A reader of the first is short of words about love and manners; a reader of the second is short of the words of law and population, mortality and discrimination among them.
You should be able to read the share for the first 1,000, 2,000 and 2,800 words off the lab, say what a name does to the share, compare it with the 98% a reader needs, work out how many words of a page that gap stands for, and test a text of your own.
Something to change about this lesson? Write it down, tick it off when it is handled, and export the list when you want to act on it. Kept in this browser only: nothing is sent anywhere, and nobody is notified.
No recommendations yet for this lesson.