Detect nearby word repetition in your writing

Nearby word repetition can make otherwise sound writing feel accidental, monotonous, or under-edited. This check reads your text, ignores common English stopwords, and identifies meaningful words that appear again within the number of word positions you choose. Every finding includes both positions and the distance between them, giving you enough context to review the phrase without pretending that every repetition is a mistake. It is deterministic, fast, and useful for drafts ranging from short emails to longer articles.

Choose a window that matches the kind of draft

The window size defines how close two occurrences must be before the later one is flagged. Distance is counted across every word in the text, including common words that are not themselves checked as candidates. A smaller window is useful when you want to catch obvious echoes inside a sentence, such as using “report” twice within a short clause.

A larger window can reveal repeated vocabulary across neighboring sentences or a compact paragraph. There is no universally correct setting because repetition has different effects in a slogan, an email, a technical explanation, and a chapter. Start with a modest window, read each result in context, and increase it only when you want a broader stylistic sweep.

The minimum is two words; a value below two is rejected instead of being silently corrected. That explicit error protects automated editing workflows from producing misleading results when a configuration is invalid. The checker reports word positions using one-based numbering, making the findings straightforward to compare with tokenized text or another editing stage.

Understand what is counted and what is ignored

Words are matched without regard to capitalization, so “Draft” and “draft” count as the same word. Unicode letters and numbers are recognized, and apostrophes inside words are retained, which keeps common contractions together instead of splitting them into misleading fragments. A concise set of frequent English stopwords is excluded from candidacy.

Articles, conjunctions, pronouns, and similar structural words often repeat naturally, so flagging every nearby “the” or “and” would bury useful findings in noise. Those stopwords still occupy positions and therefore still affect the measured distance between meaningful words. For each qualifying recurrence, the result names the normalized word, its earlier position, its current position, and the distance.

If a word occurs several times, each later occurrence is compared with the nearest earlier occurrence. This focuses the report on the closest echo, which is usually the most relevant one during self-editing. Punctuation does not count as a word, and differences in surrounding punctuation do not prevent otherwise identical words from matching.

Use findings as editorial prompts, not automatic corrections

A repeated word is not automatically bad writing. Deliberate repetition can establish rhythm, preserve a technical term, reinforce a contrast, or avoid an ambiguous synonym. Treat every flag as a prompt to inspect the passage rather than an instruction to replace a word mechanically.

When a repetition sounds accidental, you might remove one occurrence, combine two sentences, use a precise alternative, or restructure the passage so the repeated term carries intentional emphasis. When it is necessary, leave it in place. The total word count and selected window are returned as useful audit context.

Because the algorithm uses no network, random values, or current time, identical input always produces identical output. You can run it during drafting, before publication, or as one check among several quality rules.