Science Communication & Plain-Language Translation

Novice · Communication & Dissemination track · ~35 min hands-on + readings and quiz

← Back to the Communication & Dissemination track

Turn dense findings into clear public-facing materials, at a reading level your audience can use.

What you’ll be able to do

  • Translate technical findings into plain language for a specific audience
  • Control reading level, tone, and format for fact sheets, briefs, and social posts
  • Check that simplification didn’t distort the science

Overview

AI is fast at turning a result into a brief, a fact sheet, or a social post — and at adjusting reading level on request. You decide the audience and confirm the simplified version is still accurate.

Plain language is an equity tool: clearer materials reach more people. But oversimplification can mislead, so you check that nothing important was lost or overstated.

Practice activities

Activity 1 · Novice — Translate one finding for three audiences

Time ~18 min · Tools ChatGPT Edu

Goal. Turn a technical finding into clear materials for three different readers without distorting it.

Setup. Pick one public finding from your work, or use this example: “In Philadelphia, pedestrian-involved crashes peak between 4–7 PM and are more common on wide arterial roads.” Use public, non-sensitive content only.

Steps.

  1. Paste the finding and ask for three versions:

    Rewrite this public-health finding three ways: (a) for community members at an 8th-grade reading level, (b) for city policymakers, (c) as a social post under 280 characters. Keep every number accurate. [finding]

  2. Check each version against the original — did any overstate the result or drop a caveat?

  3. Push one further:

    The community version is still a little technical. Simplify to a 6th-grade level, under 60 words, without losing the “4–7 PM” detail.

Expected result. Three audience-tailored versions, all factually faithful.

Check your work. Read the social post: could a reader screenshot it as a stronger claim than the data support? If so, tighten it.

Common pitfalls. Simplifying often turns “associated with” into “causes” — guard against it. Public data only.

Stretch (optional). Ask for a flyer-ready headline plus one supporting sentence.

Activity 2 · Novice — Set and verify reading level

Time ~12 min · Tools ChatGPT Edu + a readability check

Goal. Deliberately set reading level and tone, then verify it independently.

Setup. Use this dense paragraph (or substitute one of your own non-sensitive writing):

Multilevel logistic regression models with random intercepts at the census-tract level were employed to estimate the adjusted odds of fatal or serious pedestrian injury (KSI) as a function of road-segment characteristics, controlling for tract-level sociodemographic covariates. After adjustment for posted speed limit, number of travel lanes, presence of marked crosswalks, and ambient lighting conditions, intersections lacking pedestrian countdown signals exhibited 1.9 times the odds of KSI (95% CI: 1.4–2.5) relative to signalized intersections. The intraclass correlation coefficient (ICC) of 0.12 indicated that approximately 12% of the variance in KSI outcomes was attributable to between-tract differences, suggesting that neighborhood-level factors beyond road design contribute to pedestrian injury severity.

Steps.

  1. Ask for a leveled rewrite:

    Rewrite this at a 6th–8th grade reading level, warm and direct. Then tell me the approximate reading level of your rewrite and what you changed.

  2. Verify with a real tool — paste into the Hemingway editor or check Word’s readability stats — and compare to the AI’s claim.

  3. Get a contrasting version:

    Now a more formal version for a stakeholder briefing, same facts.

Expected result. Two versions at different levels, plus a readability check that roughly matches.

Check your work. Don’t trust the AI’s self-reported grade level — confirm it with a real readability metric.

Common pitfalls. Plain language is not “dumbed down” — keep the accuracy. AI reading-level estimates are approximate.

Stretch (optional). Add a one-line “bottom line” a busy reader could act on.

Check your readiness

Answer these, then check — your score suggests whether to dive in or skim the readings first.

Useful resources