Count repeated words

Decide what you want repeated words to tell you

A frequency report ranks word forms by how often they occur. It can reveal a filler word repeated across a draft, a term that appears more often than you expected, or a name spelled in two different ways. It does not judge whether a word is useful, infer the subject of a paragraph or tell you what percentage you should aim for. Start with a concrete editorial question, such as whether a product description uses two different names for the same feature.

Paste the part you actually want to examine. Navigation labels, quoted source material and boilerplate can dominate the counts if you include them with the draft. If you are reviewing several sections, measure each separately first. A term that occurs often in one technical section can disappear in the frequency ranking when combined with a much longer unrelated passage.

Keep a copy of the original draft. A ranking is a view of that text, not a replacement for it. When you spot an unusual term, go back to the passages where it occurs and decide whether there is an issue. Repetition can be a problem, a necessary technical term, or a deliberate convention that helps readers stay oriented.

Try a short passage with mixed case and tied counts

Paste the following sample with Match case off. The count groups Blue and blue together, and repeats of ink form one row. Punctuation and whitespace do not become words in the report.

Blue ink, black ink. Blue paper. Ink again.

The output lists ink three times and blue twice. Black, paper and again each appear once. Tied counts follow first appearance. Turn on Match case to compare what happens to Ink and ink. Their counts become separate, while two identically written instances of Blue still count together. Choose the setting based on whether capitalization is the question you are checking.

This is also a practical way to verify the effect of a title or repeated section label. Add a sentence containing a term you know already appears, then see if its count moves. If the result does not look as expected, inspect the exact input, including curly apostrophes, hyphens and compound forms. A word form is not always the same unit as the concept you meant to track.

Distinguish word forms from concepts and phrases

Edit, edited and editing are separate forms. Two words with the same spelling can refer to different things in different sentences. The counter does not group inflections, identify synonyms or decide whether a word was used as a noun or a verb. If you need a count of a specific phrase, use Find and replace as a literal match check without applying changes. Review its matching options and the text around each occurrence.

A hyphenated term or a term written with an apostrophe can be segmented differently depending on the browser's language-aware support. Text from writing systems that do not put spaces between every word also depends on segmentation. Use a short known example if counts from two applications differ. Record the convention rather than assuming one application is wrong in every context.

The tool normalizes counted forms to a consistent canonical Unicode form, so visually equivalent representations of an accent can be grouped. That does not strip accents or merge every variant of a word. If a repeated term still appears in more than one form, inspect its spelling and punctuation. Use Normalize Unicode on a copy only if a broader normalization is appropriate for the intended comparison.

Choose how many rows to show without confusing that with the total

The Show setting limits the number of rows included in the ranked output. A top 20 report is useful for a quick scan; a larger selection can expose rare forms in a longer draft. The absence of a term from a limited report does not prove it never appeared. It may simply rank below the displayed cutoff. Use a literal search when you need to check one specific term regardless of rank.

Read the total and unique counts separately from the number of displayed rows. Repeating a word raises the total count without adding a new unique form. Changing the case rule can change the number of unique forms even when the same text was pasted. For a simple total of words, sentences or paragraphs, use the word and character counter rather than treating the size of a frequency table as the length of the text.

When comparing reports from two versions, use the same case setting and the same part of the draft. If one report was created from an entire page and the other from only its body, a difference in rankings may reflect the sample rather than the editing work. Label the two exports clearly.

Use the TSV report in a spreadsheet or another converter

Copy or download the report as a tab-separated table. The header identifies the word and count columns. If the receiving application does not split them automatically, select Tab as the delimiter in its import settings. Avoid replacing tabs with spaces by hand: the column separator is intentional. Check one known count and the last exported row before relying on the imported version.

If you need a structured representation for a script, open the report in the CSV and JSON converter and choose Tab for the source. Remember that the table-to-JSON direction produces string values, so a consumer that expects a numeric count may need a separate type conversion. Keep the raw report and a note of the original case and row limit settings.

For a project review, a compact table of a few interesting words can be clearer than a full report. Include the part of the document that was checked and the reason each listed term matters. The report shows counts; the surrounding passages tell another reviewer whether the repetition is useful, unnecessary or a possible spelling inconsistency.

Use the report to guide a concrete editorial change

Pick one term to investigate and locate several occurrences. If the word is an essential name or technical concept, consistency can help readers and a high count can be appropriate. If it is filler or an accidental duplicated sentence, rewrite the sentence in context. Do not replace every repetition with a synonym simply to make the count lower. That can introduce ambiguity where a consistent term was helpful.

After editing, compare the original and revised text with the text diff checker. Look for unintended changes to names or links if you made a bulk replacement. Run the same frequency settings again to see whether the intended forms changed, and read the edited passage independently of the numbers. The guide to reviewing text changes shows how to combine these checks.

When handing off a review, describe an observation and what it means, rather than presenting an unexplained count. For example, point out that the same setting has two spellings in adjacent sections and say which one should stay. That is a useful editorial result even if the ranking itself hardly changes.

Compare samples of comparable length

A longer draft can have more occurrences of a word even if the word is less prominent in the writing. Check the size and scope of each selection before drawing a conclusion from counts alone. If the two samples have different purposes, compare the relevant passages instead of combining them into one number. Keep the source and a short note of the query so the result can be reproduced later.

Check the surrounding text before interpreting an unusual count

A term may appear often because it is part of a repeated heading, label or quoted policy. Search the exact form in the source and read a few occurrences in context. If all the matches appear in boilerplate, measure the body separately before deciding that the draft overuses the term. If several spellings appear, decide which represents the same concept and which is a genuinely different word. The report is useful for finding candidates, not for making that editorial decision for you.

When comparing revisions, keep the exact scope and case rule with each exported table. Otherwise a difference in the selected text can masquerade as a change in vocabulary.