Say the dog bit the man in Spanish, and the words can move: a small word, a, marks the man as the one bitten, and the verb agrees with the dog. Russian puts the job in an ending on the noun itself. Japanese and Korean use a small word after the noun. English, for the most part, does none of this. This lesson is about what it does instead.
What English kept
Old English had endings on nouns. Almost all of them are gone. What is left is a small set of pairs of pronouns. I and me. He and him. She and her. We and us. They and them. The first of each pair is the subject. The second is who or what it happens to. Her also does a second job, belonging, as in her book, which is why the lab leaves it out when it counts the object forms.
Because the pronoun carries the role, there is a rule that needs no reading of meaning. A subject pronoun is followed by its verb. A machine can look for he and read the next word. Nobody has to mark up the page first.
The score
The lab runs the rule over a 936-word passage from Pride and Prejudice, printed on this page. It finds 80 of I, he, she, we and they, and in 72 of them, 90.0%, the next word is a verb. The passage also holds 46 of you and it, and the scan leaves every one out, because those two words do not change with the job.
The same rule was once run over the whole novel, novel text only, 122,396 words. It found 5,990 of the five pronouns, and in 88.7% the next word was a verb. No page can carry the novel, so that figure is quoted here, not counted in your browser; the passage is the part you can check.
A number like that is easy to read wrongly in two ways. The first is to call the other tenth a failure rate. The second is to call it noise. It is neither. It is a few things, and each can be named.
The four groups
- A modifier in between, 63% of the misses in the whole-novel run, 5 of the 8 on this page. They both knew. We all agreed. I almost envy you. A small word sits between the pronoun and the verb, and the verb is still the next main word.
- A word outside the list, 16%, and 2 of the 8 here. The tool only knows the words in its list. Austen wrote beg and inquired where the list has neither, and on this page I interrupt is scored a miss because the scan does not know interrupt, and so is I don’t, because the list holds no word with n’t on it. This is a limit of the tool, not of the sentence.
- A speech tag, 15%, and 1 of the 8 here. “But it is,” returned she. “My dear sister,” said he. This is a fixed device of telling a story, and it only happens beside quoted speech. A check in the whole-novel run, quoted, found 88% of these cases sitting next to quoted speech. The lab counts it in the box called word order broken, because the verb does come before its pronoun; the lesson's claim is that it is a habit, not an error.
- Other, 6%. We have not sorted these. The printed passage has none.
What was not found
The result worth knowing is the one that is missing. Among the cases we read, there was no sentence where a subject pronoun stood in a place English does not allow. Every apparent exception was one of the groups above. English order is strict, and the one device that bends it is a named habit of telling a story.
We say among the cases we read on purpose. A sample of each group was read and the speech tags were checked in full by a second method. The 16% word list gap and the 6% other were not read case by case. So the claim is that we found no mistake, which is a weaker claim than that there is none. On this page the claim is stronger, because the eight misses are all in the table and you can read each one.
Why a novel from the 1810s
The lab also counts the same pronouns in two modern public documents, printed in full under the passage: the opinion of the US Supreme Court in Stanley v. City of Sanford, 606 U.S. 46 (2025), and the Census Bureau story “U.S. Population Aging as Nation Turns 250” of 9 April 2026. Both are free to use. The passage has 85.5 subject pronouns for every thousand words, 80 in 936; the two documents have 13.9, 24 in 1,723. That is about 6.1 times fewer. For the object forms the gap is wider still: 18.2 against 1.7 per thousand, about 10.4 times fewer. Formal modern writing is mostly nouns, and the structures a speaker needs are thin on the page.
So the measurements use a novel, and that has a cost. It is 1810s English, and some of what looks like a rule may be the habit of that time. The speech tag is the clearest case. This lesson tells you where the age of the text limits a figure, and the figure that depends on it most is the 15%.
There is a lesson in the gap itself. If you learn English from formal reading only, you meet far fewer of the forms you need when you talk. Read stories and speech too.