1.Statement of findings
Principal finding
—
| Communities examined | — | |
|---|---|---|
| Posts examined | — | Top posts of the year, sampled per community |
| Posts containing an em dash | — | |
| Movement over the period | — |
2.Frequency over the period
The heavy rule is every community combined. The grey bands show how far individual communities sit either side of it.
View:Compiling figure
Figure 1. Share of sampled posts containing at least one em dash, by month. Rates are posts weighted across the communities in the sample.
3.Distribution by community
The ten communities with the highest em dash rate in the month under review.
View:Compiling figure
Figure 2. Highest em dash rates in the month under review, with the number of posts sampled beside each rate. Select a bar to trace that community in Figure 1.
4.Notes on method and limitations
- The record was collected on 4 May 2025 through Reddit's API, taking the top 1,000 posts of the preceding year in each community. Recent posts therefore appear only if they gained score quickly, which biases the sample against the closing months.
- Rates are posts weighted: posts containing an em dash divided by all posts sampled in the selection. A community's own rate is computed the same way within that community.
- Small communities can return extreme percentages off a handful of posts. Every bar in Figure 2 carries its post count for that reason, and the accompanying tables give the counts in full.
- January 2025 is excluded. The sample for that month falls to 325 posts across 107 communities, which is too thin to report beside the others.
- A rise in em dash usage is consistent with machine written text entering the corpus. It does not establish it. Editorial habit, subject matter and the sampling bias above are all live alternatives.