Sentence Length Variance Analyzer

Why sentence-length variance matters

In 1985 the writing teacher Gary Provost published a paragraph that has become required reading for editors and copywriters. It begins:

"This sentence has five words. Here are five more words. Five-word sentences are fine. But several together become monotonous. Listen to what is happening. The writing is getting boring."

Then Provost varies his sentence lengths and the prose comes alive. The paragraph is famous because it teaches the lesson by demonstrating it. Length variation is to prose what dynamics are to music — short and long sentences carry different weight, and alternating them keeps the reader's ear engaged. Even technically clean writing falls flat if every sentence is the same length.

The length tiers the analyzer uses

The Rhythm Score formula

The Rhythm Score is a 0 to 100 number computed from three components:

\( \text{Rhythm Score} = \text{CV-score} + \text{Mix-score} \) \( {} + 20 - \text{Monotony-penalty} \)

• CV-score (0 to 50): rewards a coefficient of variation close to 55%. The formula is \( 50 - |CV - 55| \) clamped at zero.
• Mix-score (0 to 30): rewards using more length tiers. Each tier (very short, short, medium, long, very long) is worth 6 points.
• Base bonus (+20): a flat reward for having more than one sentence — without it, very short texts score near zero.
• Monotony penalty (0 to 20): subtracts based on the fraction of consecutive sentences within ±2 words of each other.

Scores above 80 are rhythmic — the prose sings. 60 to 80 is balanced. 40 to 60 is uneven. Below 40 is usually one of choppy, monotonous, or run-on heavy, depending on the mean.

The statistics the analyzer reports

How to read the sentence skyline

The skyline is the most important visual on the page. Each bar is one sentence, ordered left to right. Bar height is the word count. Color shifts from cool (short) to warm (long). Patterns to look for:

• Flat skyline: every bar at roughly the same height. The CV will be low and the verdict will say "monotonous" — even if the average length is reasonable, the lack of variation tires the reader.
• All short: only blue and teal bars, all stubby. Choppy and staccato. Combine some pairs.
• All tall: only orange and red bars, towering. Run-on heavy. Split the tallest sentences.
• Cathedral skyline: short bars punctuated by occasional very tall ones. This is what rhythmic prose looks like. The Provost paragraph and most Hemingway short fiction have this shape.
• Yellow caps on bars: mark sentences within ±2 words of the previous one. Three or four caps in a row signal a streak that may bore the reader, even if the overall CV is healthy.

How to Use the Sentence Length Variance Analyzer

1. Paste your writing into the editor. The live coach on the right counts sentences and previews mean and CV as you type.
2. Click Analyze rhythm. The page renders a sentence skyline where each bar is one sentence and its height is the word count.
3. Read the Rhythm Score and verdict. Above 80 is rhythmic; below 40 usually means choppy, monotone, or run-on heavy.
4. Look for yellow streak caps on the skyline. Three or more in a row signal a stretch where sentence lengths are too similar. Break one sentence in half or merge two with a semicolon.
5. Compare your skyline to Gary Provost's in the benchmark section. Provost mixes 3 to 23 word sentences for a CV near 62%. Match the shape, not the exact lengths.
6. Click any bar in the skyline to jump to that sentence in the table. This is the fastest way to find an outlier or the start of a streak.
7. Apply the diagnosis tips, paste the revised draft back in, and watch the score climb.

Six edits that move the Rhythm Score most

1. Split your single longest sentence in half. If it is 35+ words, this alone can push CV down toward 55% and reduce the monotony penalty by introducing two new lengths.
2. Insert a 3 to 5 word sentence into a streak. Almost always cheap to add ("It mattered.", "She knew.", "Then it stopped.") and it breaks a flat run.
3. Merge two adjacent short sentences with a conjunction. Cuts choppiness without losing meaning. "The door opened. He stepped in." → "The door opened and he stepped in."
4. Add one very short sentence per paragraph as a punch. The reader's eye lands on it like a percussion hit.
5. Trim subordinate clauses from your longest sentences. A 28-word sentence with two subordinate clauses can usually shed one to drop into the long tier.
6. Re-read aloud. If you have to take a breath in the middle of a sentence, it is probably too long. If three sentences in a row feel identical, the rhythm is dead.

How this analyzer differs from the Hemingway editor

The two complement each other. Use the Hemingway editor to clean individual sentences, then use this analyzer to mix them into music. A clean draft can still be monotonous; a rhythmic draft can still have lazy verbs. Run both.

What the Rhythm Score does not capture

• Genre conventions. Legal contracts and academic papers genuinely need longer sentences. A low Rhythm Score on a brief is not bad writing — it is house style.
• Intentional staccato. Crime novels and ad copy often use deliberate choppy rhythm. The score will be low and that may be exactly what the writer wants.
• Sentence quality. A 12-word sentence and an 18-word sentence look great in the skyline, but if both say nothing, length variation is decoration. Use this tool alongside a content-edit pass.
• Internal rhythm. Comma placement, em-dashes, and clause structure also drive rhythm and are not part of word-count analysis.

Privacy and limits

The analyzer processes your text on our server only long enough to render this page and never stores it. The live coach previews while you type run entirely in your browser. There is no third-party tracking of the content you paste. Maximum input is 60,000 characters per analysis (roughly a 20 page document).

FAQ

What is sentence length variance and why does it matter?

Sentence length variance measures how much your sentence word counts differ from the average. Prose with high variance feels musical: short sentences set the pace, long ones carry the argument. Prose with low variance feels monotonous, choppy, or run-on. Standard deviation and coefficient of variation quantify this. The famous Gary Provost paragraph deliberately mixes 3 to 23 word sentences for a coefficient of variation around 60 percent, which is the gold standard.

What is the Rhythm Score?

A 0 to 100 score that blends three signals: how close the coefficient of variation is to the Provost ideal of 55 percent, how many of five length tiers (very short, short, medium, long, very long) your text uses, and a penalty for consecutive sentences within two words of each other. Above 80 is rhythmic. Below 40 usually means choppy, monotone, or run-on heavy.

What is coefficient of variation?

Coefficient of variation (CV) is standard deviation divided by the mean, expressed as a percent. It lets you compare variation across drafts with different average sentence lengths. A CV of 20 percent means your sentences are all roughly the same length. A CV of 60 percent means your sentences vary widely. The Gary Provost paragraph sits at about 62 percent.

Why does the tool flag streaks of similar-length sentences?

Even with a healthy overall coefficient of variation, a run of four or more sentences all within two words of each other lulls the reader. The skyline view shows these streaks as a flat plateau. Breaking one sentence in half, or inserting a short punchy sentence, restores cadence.

Does this tool replace the Hemingway editor?

No. The Hemingway editor flags long sentences, adverbs, passive voice, and complex phrases — it is about clarity word by word. This analyzer is about rhythm sentence by sentence. The two are complementary. Use Hemingway to clean your sentences, then use this analyzer to mix them into music.

How are sentences counted?

A sentence ends at a period, question mark, or exclamation mark followed by whitespace and a capital letter, digit, or opening quote. Abbreviations such as Mr., Dr., Inc., U.S. are not treated as sentence ends because the next character is usually a lowercase letter or a continuation of the same sentence.

Will it work on languages other than English?

The sentence splitter expects Latin-script punctuation (. ? ! …) and an uppercase next-word convention, so it works well for most European languages, less well for Chinese, Japanese, and Thai which use different punctuation conventions. Word counting itself is language-agnostic for any space-separated script.