Counting rules and selection
Count English Words with Explicit Punctuation Rules
Count English prose consistently by deciding how contractions, compounds, numbers and punctuation form tokens, then check a small passage before a longer one.
To count English words consistently, select the text to be measured and apply one rule to contractions, compounds, numbers and punctuation. Ordinary sentences are usually straightforward, but dates, decimals, technical identifiers and copied punctuation can expose differences between counters. State the method when the exact total matters.
Start with ordinary prose
The sentence The green folder is ready. contains five tokens under the local Word Counter rule: The, green, folder, is and ready. The final period ends the sentence but does not add a word. Replacing spaces with tabs or line breaks does not create extra tokens.
A heading or signature contributes words only if it is included in the selected text. For Project Update followed by that sentence and the signature Mira, the complete selection contains eight words. The sentence-only selection remains five. Decide which selection the task requires before comparing totals.
Treat contractions and compounds consistently
The local tokenizer starts a token with a Unicode letter or number and allows further letters, numbers and combining marks. It also keeps certain internal apostrophes and hyphens when another letter or number follows them. Thus don't re-enter 00123 produces three tokens.
This is an implementation rule, not a claim that every hyphen shape is equivalent. A regular hyphen inside red-blue joins the parts for this count. An em dash in red—blue separates them, and a slash in red/blue also separates them. The complete fixture red—blue red-blue red/blue therefore gives five words.
Inspect numeric expressions separately
Numbers count as tokens, but punctuation can divide them. In the tested fixture 1.5 3/4 2026-09-29, the decimal produces two tokens, the slash expression produces two and the hyphenated date produces one. The total is five.
If a required manual convention treats a decimal as one written number, record that rule and follow it instead. Do not alter the source by removing decimal points merely to force a counter to match the desired convention. The count should describe the text, while the text should retain its correct meaning.
Distinguish words from other characters
Emoji, standalone mathematical symbols and punctuation marks are visible characters but are not necessarily word tokens. The fixture 🙂 — + gives zero words in this implementation. Character count and word count therefore answer different questions even for the same input.
URLs, email addresses and programming identifiers can contain separators that divide them into several tokens. If your document contains many such items, test representative examples before relying on a total. An English prose counter is not a parser for every technical notation.
Do not confuse frequency grouping with counting
Alpha and ALPHA each contribute one occurrence. The local frequency analysis groups them after lowercasing, so alpha ALPHA beta gives three total words and two unique values. That grouping changes the unique-word result without reducing the total word count.
Similarly, write writes writing contains three occurrences and three distinct values in this tool. It does not perform stemming or decide that all three represent a single dictionary headword. A vocabulary study may need a different analysis from a submission word count.
Check copied text before measuring it
Text copied from a PDF can contain repeated page headers, artificial line breaks or hyphens inserted by layout. Compare the paste against the source. A line-ending hyphen may be part of a real compound or only a printed break within one word; do not remove every hyphen automatically.
Keep a source copy and record any cleanup used to reconstruct the intended prose. Then count the cleaned version and preserve it with the result. Otherwise, a later recount from the original paste may differ for reasons unrelated to the counting tool.
Keep the language limit visible
English-oriented token rules are not a universal definition of words. Unicode’s segmentation guidance discusses language-specific boundary requirements. For multilingual passages or a submission with a prescribed counting application, use a suitable method and record it rather than assuming this English-focused workflow covers every case.
Related guides
Count a passage by hand with an explicit word rule, marked groups and a second pass, then investigate differences without guessing from page length.
Use this guide for how to count words in writing; it explains the workflow and how to verify meaning, Unicode, whitespace and destination formatting.
Measure notes with an explicit scope for headings, bullets, tags and quotations, then distinguish a stored-text count from a summary of new writing.