Exercises — Chapter 7: Word Choice, Tone, and Voice

Word choice and tone are learned by producing and re-toning text, not by memorizing rules. Most of these ask you to rewrite real-looking passages for connotation, register, certainty, inclusivity, or voice. For open-ended tasks, a self-assessment rubric appears instead of a single "right" answer. Selected solutions: appendices/answers-to-selected.md.

How to work these: for every change you make, name the lever you turned — connotation (§7.1), register (§7.3), voice (§7.4), certainty/hedging (§7.5), weasel/euphemism (§7.6), inclusivity (§7.7), plain-language swap (§7.8), or terminology consistency (§7.9). Naming the lever is what transfers; changing by feel does not.

Part A — Analyze This ⭐

Diagnose, don't yet rewrite. Name the lever and the effect.

A1. For each pair, the denotation is identical. Name the connotation of each word and say which you'd choose to describe your own work to a manager: - a) "a thrifty plan" vs. "a cheap plan" - b) "we refined the estimate" vs. "we fudged the estimate" - c) "the rollout was aggressive" vs. "the rollout was bold" vs. "the rollout was reckless"

A2. Read these three versions of one finding and label each register (formal / neutral / informal). Then name two specific word-or-format choices that mark each register: - a) "A statistically significant reduction in error rate was observed following the intervention (p < .01)." - b) "After the change, errors dropped noticeably — and the difference is real, not noise." - c) "errors went way down after the fix 🎉 p<.01, want it in the report?"

A3. This sentence is grammatically fine and perfectly clear, but tonally wrong for a blameless postmortem. Name what's wrong and which lever fixes it: "The on-call engineer carelessly neglected to check the dashboard, allowing the outage to spiral."

A4. Count the hedges in this sentence and decide whether it's honest calibration or empty stacking: "It could perhaps be tentatively argued that the data might possibly be somewhat consistent with the hypothesis, at least under certain conditions."

A5. Each of these is a weasel word or euphemism. Name the dodge — what is it letting the writer avoid committing to? - a) "Our tool can help to reduce load times by up to 60%." - b) "The role is being eliminated as part of an organizational realignment." - c) "Studies have shown this method to be highly effective."

A6. Spot the exclusionary or imprecise word in each, and say what specific meaning it blurs: - a) "Each developer should configure his environment." - b) "The bug crippled the checkout flow." - c) "We'll do a quick sanity check before merging."

A7. This README sentence is "corporate no-voice." Identify three no-voice markers and explain what the sentence actually fails to tell a developer: "This solution delivers a robust, scalable framework that empowers users to leverage best-in-class data persistence capabilities."

A8. Find the elegant variation problem: how many distinct things might a reader think are being described, and why is that a failure here? "The handler receives the request, the listener validates the call, and the endpoint logs the transaction."

Part B — Revise This ⭐⭐

Rewrite each, naming the lever you turned. Keep the facts constant unless the task says otherwise.

B1. (Connotation — neutralize the blame) Rewrite for a blameless tone, removing the moral loading but keeping the fact: "The junior analyst botched the query and corrupted half the dataset."

B2. (Register — fix the mismatch) This is a Slack message to a teammate, written like a formal memo. Rewrite it for the channel: "I am writing to advise you that the build is currently failing and to request that you refrain from merging until such time as the issue has been resolved."

B3. (Register — the other mismatch) This is a sentence from a journal abstract, written like a tweet. Rewrite it for the venue: "Basically our model totally smokes the baseline — like, way more accurate and a lot faster. Super promising!"

B4. (Certainty — calibrate down) You have one preliminary, unreplicated result. This sentence overclaims. Rewrite it honestly: "This experiment proves that the compound cures the infection."

B5. (Certainty — calibrate up) You have strong, replicated evidence, but the sentence under-claims so hard it wastes it. Rewrite it to commit: "It may perhaps be the case that there could possibly be some association between the treatment and recovery."

B6. (Weasel words — pin it down) Replace every weasel with a committed, checkable claim (invent plausible specifics): "Experts agree our approach can help to significantly improve performance in many scenarios."

B7. (Euphemism — say it straight) An incident notice, fogged by euphemism. Rewrite it to respect the reader who needs the truth: "Due to an unforeseen service interruption, a subset of users may have encountered intermittent challenges accessing certain functionality during the affected period."

B8. (Inclusive language) Rewrite to remove the gendered default and the ableist metaphor, and note how the fix also improves precision: "Every user must trust that his data is safe, even if a server goes crazy and the system gets crippled."

B9. (Plain-language swap — keep the precise one) Swap the bloated words for plain ones, but watch for one word that may be a genuine term of art and should stay: "We will utilize the optimal configuration in order to facilitate the commencement of the migration prior to deployment."

B10. (Voice — kill the no-voice) Rewrite this "About us" line so a real person could only have written it about a real product (make up a plausible product): "We are committed to leveraging innovative solutions to deliver exceptional value to our stakeholders across the ecosystem."

B11. (Terminology consistency) Rewrite so one concept gets one term throughout, removing the elegant variation: "The user submits the form. The system stores the submission in the data store, and the application later reads the record from the repository."

B12. (Everything at once) A composite "before" — diagnose and rewrite, naming every lever:

"Our team neglected to fully leverage the available bandwidth, and as a result it could potentially be suggested that some degree of suboptimal performance may have been experienced by a number of users during the affected window; we will endeavor to ascertain the root cause and provide a status update in due course."

Part C — Write This ⭐⭐–⭐⭐⭐

Produce the document, then check your own draft against the levers.

C1. (The three-register drill ⭐) Take one factual claim from your field — a finding, a result, a recommendation — and write it three ways, keeping the facts identical: - a) Formal, as it would appear in a journal article, formal report, or specification. - b) Neutral, as it would appear in a blog post for an interested general reader. - c) Informal, as you'd say it to a trusted colleague in a Slack message.

Then write two sentences: what stayed constant across the three (the facts), and what moved (register markers — contractions, word formality, the writer's visibility, sentence length). This is the chapter in miniature.

C2. (Connotation, on purpose) You're writing about a competitor's product that genuinely underperforms yours. Write one sentence using a neutral connotation (factual, no spin) and one using a deliberately loaded connotation that's still honest. Then say which you'd use in (a) a formal comparison report and (b) an internal sales-strategy memo, and why.

C3. (Calibrated certainty) Write three one-sentence claims about the same topic at three evidence levels: (a) a hunch with no data, (b) one preliminary study, (c) strong replicated evidence. Use calibrated language for each (§7.5 table), and underline the marker that sets the certainty level in each.

C4. (Voice charter) Write a 3–5 sentence "voice charter" for your own professional writing (this feeds your Project Checkpoint). Name: warm or reserved? direct or diplomatic? plain or precise-technical? List three "no-voice" words you ban yourself from using.

C5. (De-euphemize a real message) Find a real corporate or institutional message you've received — a layoff notice, an outage notice, a policy change, a rejection — that used euphemism to soften bad news. Rewrite it straight: what would honest, plain, humane language have said? (Note: straight is not cruel — see Case Study 2.)

C6. (Introduce a voice into documentation) Take any feature or tool from your field and write a two-sentence description twice: once in corporate no-voice, once with a deliberate, human voice. Then name the specific choices (a committed claim, a pointed word, a touch of opinion) that gave the second version a voice.

Part D — Synthesis & Critical Thinking ⭐⭐⭐

D1. (Translate a hedge across the hedging gap) Here is a careful, properly-hedged research sentence:

"In this sample, daily mindfulness practice was associated with a modest reduction in self-reported stress; the effect size was small and the design was correlational." Now imagine a press office turns it into a headline. Write the dishonest over-claimed version a bad press release would produce, then write an honest general-audience version (Chapter 28) that preserves the hedge. In two sentences, explain why preserving the hedge is an ethical act (Chapter 38).

D2. (Find the flaw in the advice) A style guide says: "Always be confident. Never hedge. Always use the simplest word. Avoid all jargon." For each of those four "always/never," give one concrete counterexample from this chapter where the rule produces worse or dishonest writing, and write the better formulation.

D3. (Defend register against "authenticity") A colleague refuses to adjust register, saying, "Writing differently for different audiences is fake — I just write like myself." Write a short, respectful response (3–5 sentences) that distinguishes voice (which can stay constant) from register (which should flex), using the suit-vs-jeans analogy or your own.

D4. (Word choice vs. clarity — draw the line) You're handed a draft and told "the words feel off, but I can't say why." List the seven word-choice/tone levers from this chapter as a diagnostic checklist, and for each, write the specific question you'd ask of the draft. Then explain how this checklist differs from Chapter 3's clarity checklist — what does each catch that the other misses?

D5. (Person-first vs. identity-first) You're writing a healthcare app's onboarding copy for users with diabetes and users on the autism spectrum. Research (briefly) or reason about the typical community preferences for each, and explain why a single blanket rule ("always use person-first") would be the wrong approach. What would you actually do when you don't know a specific user's preference?

Part M — Mixed Practice (Interleaved) ⭐⭐–⭐⭐⭐

These mix this chapter with Chapters 2, 3, and 6, so you must choose the right tool — not just apply the last thing you read.

M1. (Ch 3 + Ch 7) You've cut a sentence to the bone (Chapter 3) — no bloat, every word carries a fact — and your reader still reacts badly. The sentence is "The team neglected the backups, so we lost three days of data." The facts are right and it's concise. What's wrong, and which chapter's tool do you reach for?

M2. (Ch 6 + Ch 7) A reviewer flags this sentence. Decide whether the problem is a mechanical one (Chapter 6 — fix it, there's a right answer) or a tonal one (Chapter 7 — choose among valid options), or both, and act: "The vendor, who we had trusted, neglected to deliver, and it could be argued this may have impacted us somewhat."

M3. (Ch 2 + Ch 7) Same outage, two messages: one to your engineering peers in a postmortem, one to a non-technical VP. Decide the register and the connotation for each, then write the one-sentence summary for each audience. Explain, using K-R-A-C from Chapter 2, why the VP's version isn't "dumbed down."

M4. (Ch 2 + Ch 3 + Ch 7) Order these four questions the way you'd actually ask them when drafting, and justify the order: (i) "Is every word carrying meaning?" (ii) "What does this word make the reader feel?" (iii) "Who is my reader and what do they know?" (iv) "What register does this genre call for?"

M5. (Ch 7 internal) You have time for only one tone pass before sending a hard email delivering bad news to a client. Which lever gives you the most protection against a bad outcome — connotation, register, certainty, euphemism-check, or voice — and why? (Defend your choice; there's a strong case for more than one.)

M6. (Ch 3 + Ch 7) This sentence has both a clarity problem and a tone problem. Fix both, and label which fix is which: "It is important to note that the team utilized a methodology that, it could be argued, may have potentially resulted in some degree of suboptimal outcomes for certain individuals."

Part E — Extension ⭐⭐⭐⭐ (optional; Deep Dive track)

E1. (Hedging corpus study) Find two documents about the same scientific finding: the original paper (or its abstract) and a popular-press article about it. Identify three places where the press version stripped or weakened a hedge that the paper carried. For each, decide whether the change is acceptable simplification or dishonest over-claiming, and justify. (This is the §7.5 "Going Deeper" sidebar applied to real text.)

E2. (Build your own no-voice blacklist) Over one week, collect ten "no-voice" words or phrases from your own field's writing (your inbox, your company's marketing, papers you read) — the leverage / solutions / robust family, plus field-specific ones. For each, write the committed, human alternative. This becomes your personal voice-charter ban list.

E3. (Register range under constraint) Take a single 40-word factual paragraph and rewrite it at five points on the register spectrum: a legal/contract register, a formal journal register, a neutral business register, a casual blog register, and a text-message register. Keep the facts identical. Then write a paragraph on which choices changed at each step and where the register "broke" (became inappropriate for any real audience).

E4. (The connotation of verbs in incident reports) Collect five verbs commonly used to describe failures (failed, neglected, missed, dropped, broke, degraded, crashed, lost). Rank them from most blame-loaded to most neutral, and write one sentence on which you'd choose for (a) a blameless internal postmortem and (b) a vendor-accountability letter, and why the connotation matters differently in each.

Self-assessment rubric (for the "Write This" and open tasks): - Connotation: Did you choose words whose feeling matches your intent — and can you name the connotation you picked? - Register: Does the formality fit the genre and the audience? In the three-register drill, did the facts stay constant while only register moved? - Certainty: Is each claim calibrated — strong claims committed, weak claims honestly hedged, no empty stacking and no wasteful under-claiming? - Honesty: Did you replace weasel words and euphemisms with committed, checkable claims? - Inclusivity: Does every word serve the actual reader without excluding, insulting, or blurring? Did the inclusive fix also improve precision? - Voice: Does the prose sound like a human who has taken a position — not a committee-approved press release? - Consistency: One term per concept, repeated — no elegant variation?