Character and word counter

Three metrics from the technical brief cover water, classical and academic nausea, and three cover readability. They are calculated using the open formulas described below. They do not have to match another service character for character because each one uses its own stop-word dictionary.

• Water — the share of function words: prepositions, conjunctions, particles, pronouns, and introductory words. It is calculated using a dictionary of 280 words, which are also highlighted as stop words in the field. Up to 40% is normal, natural text; 40–60% is somewhat high; above that, trim the fluff.
• Classical nausea — the square root of the number of repetitions of the most frequent significant word. Function words are deliberately excluded: in Advego, the most frequent word is almost always a preposition, which makes the result meaningless. That is why our value may be lower; the word that produced it is shown beside it.
• Academic nausea — the share of repetitions of significant words in the entire text: all occurrences of words that appear more than once are added together. The normal range is 5–15%: below that, the text lacks a specific point; above that, it is over-optimized.
• Flesch index — 206.835 minus 1.52 times the average sentence length in words minus 65.14 times the average number of syllables per word. These coefficients are for Russian, from Irina Oborneva. English uses different coefficients, so applying the English formula to Russian text is meaningless.
• Syllables are counted by vowels: in a Russian word, the number of vowels equals the number of syllables. For Latin text, consecutive vowel groups are counted — "house" is one syllable, not four.

Words are grouped by stem: "текста", "тексту", and "текстом" count as one word, not three. The Porter stemmer extracts the stem, so it may not resemble the dictionary form; the table shows the most frequent form found. Function words are excluded from the top list; otherwise, the first thirty rows would be prepositions. They are collected in a separate list below the table.

Which paragraph has grown too long and where a forty-word sentence is hiding is visible at a glance, without rereading the entire text. Sorting changes the row order, but the paragraph number stays the same.

How to use it

1. Paste text into the field or drop a .txt, .md, .docx, or .odt file onto the page. All figures are recalculated as you type; there is nothing to click.

2. Choose a platform if you are writing to a limit: a page title, an X post, a Telegram caption, a Direct ad, or an SMS. The scale shows how many characters remain; for SMS, it also shows how many segments the message will use.

3. Set a length target if your work must be submitted by character or word count. The bar shows how much you have entered and how much remains; the draft and target are saved in this browser and remain after you close the tab.

4. Check water and nausea if you are submitting text to a client: these three figures are usually specified in the technical brief. Each one includes a benchmark and a plain-language assessment.

5. Open the frequency analysis to see what the text actually revolves around. Words are grouped by stem, and pairs and triples of words are shown separately, making repeated phrases easy to spot.

6. Turn on highlighting when the numbers have already identified a problem: long sentences, repetitions, and stop words are colored directly in the field, so you can see where to edit.

7. Export the result: a summary to the clipboard, a full report as a text file, and the frequency analysis and paragraph breakdown as CSV for a spreadsheet.

The formula and what it includes

• Characters including spaces — the full line length: letters, digits, punctuation, spaces, and line breaks.
• Characters without spaces — the same count minus all spaces, tabs, and line breaks. This is the figure usually used to measure copywriting volume.
• Word — a continuous chain of letters and digits; a hyphen or apostrophe inside the chain does not split it, so “из-за” is one word, not two.
• Sentence — a section of text ending with a period, question mark, or exclamation mark followed by a space or the end of the text. Two exceptions apply to abbreviations: a period after a single capital letter is an initial, so “А. С. Пушкин написал это.” counts as one sentence; a period followed by a lowercase letter or digit is part of an abbreviation, so “и т. д.” and “рис. 5” do not end a sentence.
• Paragraph — a section of text between blank lines. A simple line break does not end a paragraph—it creates a new line.
• Word stem — what remains after Porter's stemmer removes an ending and suffix. It puts “текста” and “тексту” under the same entry in the frequency analysis. It is not a dictionary lemma: “стали” from “стать” and “сталь” produce the same stem, and the stemmer always makes conflations like this.
• Function word — a preposition, conjunction, particle, pronoun, or introductory word from the built-in dictionary of 280 entries. These words make up the water figure and are highlighted in the field as stop words.
• A4 page — 1800 characters including spaces—14-point type, 1.5 line spacing, and two-centimeter margins. A typewritten page has 1860 characters: thirty lines of sixty-two characters. An author's sheet is 40,000 characters including spaces.
• SMS segment — one message in the carrier's network. Latin text uses the GSM-7 7-bit alphabet: 160 characters in a single-part message and 153 per part in a multipart message. Cyrillic does not fit in GSM-7, so the message uses UCS-2: 70 characters in a single-part message and 67 per part in a multipart message. One Russian letter in the text means the entire text is counted using the 70-character limit.

Example analysis: two short phrases

The input data is the same as the default field values: you can calculate it mentally and confirm that the tool is telling the truth.

• Text: Brevity is the soul of wit. But a character limit is no reason to write badly.
• Reading speed: 200 words per minute; speaking speed: 125 words per minute

1. The text contains 13 sections separated by spaces, so it has 12 spaces. There are 71 characters including spaces, and 71 − 12 = 59 without spaces.

2. There are 11 words: the two em dashes do not count as words because they are neither letters nor digits. All eleven words are different, so there are also 11 unique words.

3. There are two periods at the ends of phrases, so there are two sentences. There are no blank lines, so there is one paragraph and one line.

4. There are three function words: “но”, “по”, and “не”. Water = 3 ÷ 11 × 100 = 27.3%—below forty, which is normal.

5. No significant word is repeated: the most frequent one occurs once, and the square root of one is one, so classic nausea is 1.00. There are no repetitions at all, so academic nausea is 0%.

6. The text contains 20 syllables, 11 words, and 2 sentences. Flesch = 206,835 − 1,52 × (11 ÷ 2) − 65,14 × (20 ÷ 11) = 206,835 − 8,36 − 118,44 ≈ 80—very easy, at an elementary-school level.

7. Reading time: 11 ÷ 200 = 0,055 minutes, which is 0,055 × 60 ≈ 3 seconds. Spoken aloud, the same 11 words at 125 words\/min take 5 seconds.

71 characters including spaces, 59 without spaces, 11 words, and 2 sentences. Water: 27.3%; classic nausea: 1.00; academic nausea: 0%; Flesch ≈ 80. As a page description, the length fits within 160 characters with room to spare; as a headline, it does not.

Frequently asked questions

• Why do different counters show different word counts?

  Because everyone defines “word” differently. Here, a word is a chain of letters and digits; a hyphen or apostrophe inside it does not split it: “из-за” counts as one word, and so does “5-7”. If a counter splits text at spaces, “из-за” is still one word, but a dash surrounded by spaces becomes a word. The difference is usually a few percentage points.

• Why don't the water and nausea figures match Advego and PR-CY?

  The formulas are public, but the dictionaries are not. Water depends entirely on which words are treated as function words: our dictionary contains 280 entries and is built right into the highlighting, so you can see exactly what was counted as water. We calculate classic nausea from the most frequent significant word, not the most frequent word overall. In Advego, that is almost always a preposition, so its figure reflects the preposition rather than the text. A difference of a couple of percentage points is normal; a difference of several times means the services are measuring different things.

• Are the text and file sent anywhere?

  No. The file is read in the browser, and .docx and .odt files are unpacked right here using built-in decompression—this page has no server of its own at all. The draft and settings are stored in the browser on this device; the “Clear” button deletes them along with the text field. Text is included in the analysis link only when it is short, and the button says so explicitly.

• How accurate is the reading time?

  It is only as accurate as the speed you entered. The formula is simple: words divided by words per minute. It does not account for text complexity, tables, or images, so it is an order-of-magnitude estimate, not a measurement. The same applies to speaking time: 125 words per minute is a calm presentation pace, while speaking rapidly is twice as fast.

• What is the word stem in the frequency analysis?

  A truncated ending, not a dictionary form. Porter's stemmer for Russian removes endings and some suffixes according to rules, without a dictionary: “текста”, “тексту”, and “текстом” produce the same stem. The trade-off is conflation: “стали” from the verb and “сталь” converge on one stem. That is enough for estimating repetitions, but not for linguistics.

• How are SMS segments counted?

  The alphabet is checked first. If the text contains only Latin letters, digits, and characters from the GSM-7 table, the message uses the 7-bit alphabet: 160 characters in a single-part message, or 153 per part in a multipart message. A single Cyrillic letter means the entire message uses UCS-2, with two bytes per character: 70 characters in a single-part message, or 67 per part in a multipart message. Parentheses, the tilde, and the euro sign take two positions in GSM-7; this is accounted for.

• Are emoji and ideographs counted?

  As characters, yes, but with a caveat: some emoji consist of several encoding units, and the counter will see more than one character there. Most forms where you paste this text count them the same way, so the limit usually matches. Emoji do not count as words: a word consists of letters and digits.

• What counts as a paragraph?

  Here, a paragraph is a section between blank lines. A single line break creates a new line, but not a new paragraph. That is why a text can contain more lines than paragraphs, and that is normal. In .docx and .odt files, each document paragraph comes in as a separate section, so the paragraph analysis matches what you see in the editor.