Dash frequency analyser

By Jack Stovell X · Instagram · LinkedIn · published 2026-09-24

Dash frequency is how often dashes appear in your writing, expressed as a count per 1,000 words. It covers em dashes, en dashes, double hyphens and spaced hyphens. Since November 2026, code counts it from your samples when a profile is built, so for a profile built since then the number is a count, not an estimate.

Analyse your dash frequency free

Your free voice profile measures dash frequency among all 45 attributes, from a sample of your own writing. No card needed.

Try the six-signal check first

At a glance

What dash frequency measures

Dash frequency measures how often a writer reaches for a dash, per 1,000 words of their samples. The app's own definition is plain: how often dashes (em, en, a double hyphen or a spaced hyphen) appear, counted from your samples. Four marks count: the em dash, the en dash, the double hyphen (--) and the spaced hyphen ( - ).

Three things never count. Hyphens inside words, like "well-known", are ignored. So are markdown bullet markers and horizontal rules. Otherwise a formatted document would look dash-heavy when it isn't.

So why bother? Because the dash is a habit, and habits show. Some writers drop an aside in the middle of a sentence. Others use a dash as a pause before the payoff. Plenty never touch one. The attribute sits in the Syntax group, next to average sentence length and parenthetical rate, because a dash and a bracket often do the same job.

Most values in a ScriptGrain voice profile are the extraction model's reading of your samples. This attribute is one of two exceptions (the other is contraction rate). Since November 2026, code counts dash frequency from your samples whenever a profile is built or re-analysed, and stores that number. Profiles built before then keep the extraction model's earlier reading. The free analyser and the free style guide use the same code, so all three give the same number for the same text.

How dash frequency is scored

Dash frequency does not move a voice match score at all. The voice-match engine has no comparator for this attribute, so it never changes the score and polish cannot target it. The full method sits on the voice measurement framework page, and the score itself is defined in the voice match glossary entry.

That doesn't make the number pointless. It's a description of your style, and the em dash is on ScriptGrain's published AI-tell list. The AI-tell scanner also counts em and en dashes as its own "Em dashes" meter, kept outside the tell density. So the attribute is a measurement, not a grade. Your count describes you; nothing in the match score marks you down for it.

How to analyse dash frequency in your own writing

You can work out your dash frequency by hand. You need a sample of your own writing, ideally a few thousand words.

  1. Pick samples that sound like you: published posts, long emails, a newsletter. Skip anything heavily edited by someone else.
  2. Count the words. Your word processor will give you a total.
  3. Search the text for each mark in turn: the em dash and the en dash, then the double hyphen, and a hyphen with a space either side.
  4. Ignore hyphens inside words, bullet markers and horizontal rules. They don't count.
  5. Add up the dashes you found.
  6. Divide by the word count, then multiply by 1,000. That's your dashes per 1,000 words.

ScriptGrain's free voice profile reads dash frequency from a writing sample, among all 45 attributes. For a wider view of the same ground, see the writing style analysis tool.

Examples of dash frequency in real writing

Dash frequency shows up in how a sentence breathes. These three examples are illustrative, written for this page.

Example, heavy use: "The launch went fine - mostly - until the pricing page broke, and then we all stopped talking."

Example, light use: "The launch went fine, mostly. Then the pricing page broke and everyone went quiet."

Example, none at all: "The launch went fine (mostly), until the pricing page broke."

Same event, three habits. The first leans on spaced hyphens for asides and a pause. The second swaps them for commas and a full stop. The third tucks the aside into brackets. None is wrong. Which one feels like you is the question, and the count tells you.

How to get AI to use dashes the way you do

AI uses dashes your way only if you tell it which dash you mean and show it the habit. Left alone, a model picks its own.

  1. Say which dash you use: spaced hyphen, en dash or em dash. Then say what for: asides, a pause before the payoff, or ranges.
  2. Show the habit with examples from your own writing.
  3. Compare the dash count in the output with your samples. The habit is easy to overdo, so check it.

One caution. ScriptGrain generation never writes em or en dashes, so these steps apply to your own writing or to other tools. When a ScriptGrain profile writes, every generation prompt says "ABSOLUTELY NEVER use em dashes. Not a single one." and stripEmDashes replaces any em or en dash left in the output with a comma. Spaced hyphens are not stripped. The profile reaches ChatGPT, Claude and other tools through the API and MCP server.

How to make AI stop using em dashes

To make AI stop using em dashes, ban the mark outright and name what should replace it. If you want the background, why models overuse the mark is covered in the markers of AI writing reference.

  1. Ban the em dash in your prompt, and name the replacements: comma, colon, full stop or brackets.
  2. Search the draft for the em dash and en dash characters before publishing.
  3. Use a generator that enforces the ban. ScriptGrain strips any em dash that slips through.

The evidence is stark. In SGR-001, the bare model used ten em dashes on the banking brief; the profiled arm used none. In SGR-002, bare GPT-5 averaged 17.5 em dashes per draft, while ScriptGrain's drafts had none. The prompt arm also removed every em dash, so a firm instruction does work. It just needs checking. For other tells, the AI cliché checker covers the phrases.

Questions

What is a dash frequency analyser?

A dash frequency analyser counts the dashes in a piece of text and reports them per 1,000 words. ScriptGrain's version counts em dashes, en dashes, double hyphens and spaced hyphens, and ignores hyphens inside words, bullet markers and horizontal rules. It's the same code the free style guide uses, so the numbers match.

Is an em dash checker the same as an em dash counter?

Mostly, yes. Both find em dashes and tell you how many there are. The dash frequency analyser goes a step further: it counts en dashes, double hyphens and spaced hyphens too, then scales the total per 1,000 words. That makes two texts of different lengths comparable.

How is dash frequency scored?

It isn't scored. The voice-match engine has no comparator for dash frequency, so it never moves a voice match score. The value is a count per 1,000 words, taken from your samples when a profile is built or re-analysed. It describes your habit without grading it.

Why does AI use so many em dashes?

ScriptGrain's studies measure the habit, not its cause: in SGR-002, bare GPT-5 averaged 17.5 em dashes per draft. A habit that strong is easy to spot, which is why the em dash is now read as a tell.

Are em dashes a sign of AI writing?

They're a signal, not proof. Plenty of human writers love the em dash, and plenty of AI drafts carry none once told to avoid them. The AI-tell scanner reports em and en dashes as their own meter, separate from the tell density, because one mark shouldn't decide the verdict.

Related

Sources