Why avoid
the em dash?
By Adam Frick · Illustrated draftWhy leave the reader to supply the connection?
A dash is quick to type. Explaining why the editor postponed publication takes a little more work. In the first example below, the dash leaves that explanation for the reader to infer. If the missed deadline caused the postponement, why make a colleague supply the word “because”? I want the reason on the page.
See the supporting references.Compare the wording, then judge the effect.
Name the cause.
Constructed teaching exampleThe editor postponed publication—the team missed the deadline.
The editor postponed publication because the team missed the deadline.
Name a different relationship.
Constructed teaching exampleThe draft is shorter—the argument remains hard to follow.
The draft is shorter. In spite of the reduced length, the argument remains hard to follow.
Move the interruption.
Constructed teaching exampleThe committee—which met twice—approved the proposal.
The committee approved the proposal after meeting twice.
Even a small range gives you a choice.
A related preference appears in June Casagrande’s “Dashes are poor substitutes for words”[6]. Her examples concern marks used in ranges, rather than the em-dash interruptions above. She prefers words in running prose while making room for compact ranges in information boxes. That distinction interests me because the same information can call for different treatment in a sentence and a timetable.
Match the form to the reading task.
Constructed example, not a quotation from CasagrandeWorkshop: 10:00–12:00
The workshop runs from 10:00 to 12:00.
The Australian Government Style Manual[3] recommends phrases for most spans in general prose. Within that guidance, the manual permits compact ranges in technical content and where space is limited. My preference follows the same practical question: which form helps this reader use this information?
When would I keep the dash?
I would keep a dash when a deliberate interruption serves the passage better than a fuller explanation. A compact range can also belong in a timetable, as the comparison above shows. My preference concerns the work required in the particular sentence, not a belief that the mark has no legitimate uses. Research on embedded clauses gives me a reason to inspect interruptions; that research does not prove a punctuation ban.
References and evidence
- LemonGrad figures attributed to Ethnologue’s 2022 edition ↩
- MLA Style Center ↩
- Australian Government Style Manual ↩
- discussion of subject–verb separation by Gopen and Swan ↩
- The 2022 legal-language experiments ↩
- Casagrande, June (6 October 2026). Dashes are poor substitutes for words. Grammar Underground. A personal editing preference about ranges, not an experiment on reading. ↩
Read the evidence and its limits.
The LemonGrad figures attributed to Ethnologue’s 2022 edition[1] imply nearly three non-native speakers per native speaker. My preference for dependable signals follows from that shared audience. I regard the em dash as a “wild west” punctuation mark. The problem is the breadth of the permitted functions and the scope for individual judgement. A flexible mark can become a substitute for deciding how the sentence should work.
Style guides do give instructions for dashes. The MLA Style Center[2] describes em dashes used to set off a portion of a sentence in a way similar to parentheses. The Australian Government Style Manual[3] prefers spaced en dashes for additional material and notes that other guides use em dashes.
The first version is grammatical, but the added information separates the subject from the main verb. The reader meets the committee and must then work through the interruption before reaching the action. The discussion of subject–verb separation by Gopen and Swan[4] explains why the placement of added information matters.
The structure of the interruption has experimental support. The 2022 legal-language experiments[5] found that clauses embedded inside other clauses impaired recall more than the other difficult features examined. The experiments tested linguistic features rather than a punctuation ban. My practical conclusion concerns the placement of added material, so replacing a dash with commas cannot by itself resolve the problem.