| name | reader-expectations-prose |
| description | Improve prose clarity using the reader-expectations methodology of Gopen & Swan ("The Science of Scientific Writing", American Scientist, Nov-Dec 1990). Use when the user asks for writing improvement, prose review, revision suggestions, help making writing clearer or more readable, feedback on a draft, or mentions clarity, flow, emphasis, sentence structure, or "why this passage is hard to read". Works on any expository prose (papers, reports, memos, docs, essays), not only scientific writing. |
Reader-Expectations Prose Review
Source attribution
This skill applies the rhetorical principles articulated by George Gopen and Judith Swan in "The Science of Scientific Writing", originally published in American Scientist, NovemberโDecember 1990 (the methodology itself draws on prior work by Joseph M. Williams and Gregory G. Colomb). When you use this skill, briefly tell the user that the recommendations come from that article, so they can read the original if they want.
The article's core claim: readers interpret prose using structural expectations about where information should appear. Writing is clearer when its structure satisfies those expectations. The point is not to simplify the substance โ it is to remove the structural obstacles that make substance hard to extract.
When to invoke this skill
Invoke when the user asks for help improving a piece of writing โ for example:
- "Can you improve this paragraph?"
- "Why is this hard to read?"
- "How can I make this clearer?"
- "Review my draft."
- "This feels clunky โ can you tighten it?"
Apply the skill to the prose the user provides (or points to). Do not silently rewrite โ show the user what is wrong, why, and what to change.
The seven principles
These are principles, not rules. Slavish adherence will not produce good prose. Any one can be violated to good effect; skilled stylists violate them deliberately at exceptional moments. Use them as diagnostic lenses, not as a checklist to enforce.
-
Subjectโverb proximity. Follow a grammatical subject as soon as possible with its verb. Anything of length between them is read as interruption, and therefore as lesser importance โ even when the interrupting material is in fact important.
-
Stress position. Readers naturally emphasize material that arrives at the moment of syntactic closure โ the end of a sentence, or just before a properly used colon or semicolon. Put the information you want emphasized there. A sentence is "too long" not at a fixed word count but when it contains more stress-worthy items than it has stress positions.
-
Topic position. The beginning of a sentence frames it: readers assume the sentence is "a story about" whoever or whatever appears first. Put the person, thing, or concept whose story it is in the topic position.
-
Old information links backward; new information goes to the stress position. The topic position is for material that connects to what came before (linkage and context). The stress position is for the new information you want the reader to take away. The common failure โ the No. 1 problem in American professional writing, per the authors โ is reversing this: writers rush new information to the front and bury linkage at the end.
-
Action lives in the verb. Express the real action of each clause as its verb, not as a nominalization buried inside a noun phrase ("inhibits" rather than "produces an inhibition of"). When actions hide in nouns, readers cannot tell what is happening.
-
Context before novelty. Within a sentence, paragraph, or section, give the reader orientation before asking them to absorb something new.
-
Structural emphasis must match substantive emphasis. The places the structure tells the reader to emphasize must be the places the writer actually wants emphasized. Mismatch is the most reliable source of misreading.
A corollary: when old information is absent (not just misplaced), the reader has to invent the logical link. This is how revision exposes conceptual gaps โ once you arrange old/new properly, you often discover that a connecting sentence was never written.
Review procedure
Work in this order. Each step typically uncovers issues that make later steps easier.
1. Read the passage and identify what it is trying to do
Before diagnosing anything, state in one sentence what story the passage seems to be telling and who/what is its main subject. If you can't, that itself is the first finding โ and tells you the topic positions are not doing their job.
2. Scan topic positions
List the grammatical subject (or topic phrase) of each sentence in order. Then ask:
- Is there a consistent thread? Does the same "character" recur where it should?
- Do the topic positions contain old information that links to the previous sentence โ or are they introducing new material the reader has not yet been prepared for?
- Where the topic shifts, is the shift signalled and justified?
3. Scan stress positions
Identify what sits at the end of each sentence (and immediately before any colon or semicolon). Then ask:
- Is that the information the writer most wants emphasized?
- If two or three stress-worthy items are crammed into one sentence with only one stress position, the sentence needs to be split or restructured.
- If the stress position holds something trivial (a date, a hedging clause, a parenthetical), the real emphasis-worthy material is being deprioritized.
4. Check subjectโverb distance
For long sentences, locate subject and verb. If they are widely separated, decide whether the intervening material is (a) important enough to warrant its own clause/sentence, or (b) an aside that should be deleted or relocated. Either fix is acceptable; leaving the interruption in place is not.
5. Check that verbs carry the action
Underline the verbs. If they are mostly forms of is/are/has/was, or vague linkers (involves, concerns, relates to, occurs), look for the real action โ usually it is trapped inside a noun. Reanimate it as a verb.
6. Look for missing connectives (logical gaps)
Where you have rearranged old/new information and the sentences still don't flow, suspect a missing sentence โ a connection the writer left implicit because it was obvious to them. Name what is missing; offer to supply it only if the technical content is within reach, otherwise flag it for the writer.
7. Compose specific revisions
For each diagnosis, give the user:
- The original sentence(s).
- The specific diagnosis (which principle is violated and how).
- A concrete revised version.
- A brief note on what the revision changed and why.
Do not rewrite silently. The user should be able to see the principle in action and apply it themselves next time.
Output format
When reviewing a passage, structure the response like this:
- What this passage seems to be about โ one sentence stating the apparent subject/story.
- Diagnoses โ a short list of the structural issues found, each tagged with the relevant principle (e.g., "Topic position โ sentence 3"). Be specific: quote the offending text.
- Proposed revision โ the rewritten passage in full, so the user can see how the pieces fit together. If you had to fill a logical gap with content the author did not provide, mark that addition clearly and recommend the author confirm it.
- What changed and why โ a short explanation pairing each substantive edit with the principle behind it. Avoid line-by-line pedantry; group related edits.
- Caveats โ note where you guessed at the author's intended emphasis, and invite correction. Per Gopen & Swan, "meaning requires the combined participation of text and reader" โ the reviewer cannot know the author's intent with certainty.
Briefly cite the source on first invocation in a conversation: "Applying the reader-expectations methodology from Gopen & Swan, 'The Science of Scientific Writing' (American Scientist, 1990)."
What to avoid
- Don't pursue "plain English" or simplification as a goal. The article is explicit that the aim is clarification, not simplification; jargon and technical vocabulary are not the target.
- Don't enforce word-count limits on sentences. Length is a symptom, not the disease.
- Don't blame the passive voice reflexively. "Pollen is dispersed by bees" is the correct sentence when the paragraph's story is about pollen.
- Don't apply the principles as rigid rules. Note when a violation is deliberate and effective, and leave it alone.
- Don't pile on small grammatical or stylistic nits unrelated to reader expectations โ that is a different review. If the user wants line edits too, ask before mixing them in.
A worked-example shape (for reference, not output)
From the article: when the topic positions of a paragraph are scanned and reveal new information in nearly every slot โ "Large earthquakes / The rates / Therefore one / subsequent mainshocks / great plate boundary ruptures / the southern segment of the San Andreas fault / the smaller the standard deviation" โ the diagnosis is that the recurring concept ("recurrence intervals") is being denied the topic position it deserves. The fix is to rebuild each sentence so that recurrence-interval information opens it and the new, emphasis-worthy claim closes it. Use this scanning move whenever a paragraph feels disorienting on first read.