Exercises — Chapter 6: Sentences That Work
These exercises are mostly find the error and revise this tasks, because that's how sentence repair becomes automatic. For each fix, name the error (dangling modifier, comma splice, faulty parallelism, etc.) — naming is what makes the skill transfer to your own drafts. Stars mark difficulty: ⭐ recognize, ⭐⭐ revise, ⭐⭐⭐ produce/judge, ⭐⭐⭐⭐ extend.
Selected solutions and rubrics live in
appendices/answers-to-selected.md. Where a task is open-ended, use the self-assessment rubric given here.
Part A — Analyze This ⭐ (find and name the error)
For each sentence, (a) identify whether it's broken and (b) name the specific error from the chapter. Some may have more than one. A few are correct — say so.
A1. "Running the benchmark suite for the first time, several memory leaks were detected."
A2. "The orchestrator schedules the jobs, then it reports the results to the dashboard." (Two services exist: an orchestrator and a reporter. Is this clear?)
A3. "The set of integration tests that exercise the payment flow are currently failing."
A4. "The script reads the config file, it validates each field, and the connection is then opened."
A5. "To improve reliability, we added retries, a circuit breaker, and monitoring the queue depth."
A6. "The deployment succeeded; however, the cache was never cleared."
A7. "Each of the three regions have a dedicated failover cluster."
A8. "The migration locked the table for an hour, which caused the support tickets to spike."
A9. "The function reads the input file, parses each line, and the results were written to the database."
A10. "Having reviewed the proposal carefully, the budget seemed insufficient to the committee."
A11. "The accuracy of the new model is significantly better than the previous version, this is a result we did not expect."
Answers A1–A11
- **A1.** Dangling modifier. *Running the benchmark suite* implies a human; the subject is *several memory leaks*, which didn't run anything. Fix: *"Running the benchmark suite for the first time, **we detected** several memory leaks."* - **A2.** Ambiguous pronoun. *It* could be the orchestrator or the reporter. Fix by repeating the noun: *"…then **the orchestrator** reports the results."* - **A3.** Subject–verb disagreement across distance. Subject is *set* (singular), not *tests*. Fix: *"The set of integration tests … **is** currently failing."* - **A4.** Two problems: a **comma splice** (*reads the config file, it validates*) and **faulty parallelism / inconsistent voice** (two active clauses, then a passive *the connection is then opened*). Fix: *"The script reads the config file, validates each field, and opens the connection."* (Now three parallel active verbs, one subject, no splice.) - **A5.** Faulty parallelism in a series. *retries* (noun), *a circuit breaker* (noun), *monitoring the queue depth* (gerund phrase). Fix: *"…we added retries, a circuit breaker, and **queue-depth monitoring**."* (three nouns) — or recast all three as verbs. - **A6.** **Correct.** Semicolon before *however*, comma after — exactly right. (This is the right form; A6 is the trap that tempts you to "fix" something that isn't broken.) - **A7.** Subject–verb disagreement. *Each* is singular. Fix: *"Each of the three regions **has** a dedicated failover cluster."* - **A8.** Vague *which*. Grammatically *which* points at *hour* or *table*; you mean the *lock*. Acceptable as-is to many readers, but tighter: *"The migration locked the table for an hour, and **the lock** caused the support tickets to spike."* - **A9.** Faulty parallelism / inconsistent voice and tense. Two present-active verbs (*reads, parses*), then a past-passive clause (*the results were written*). Fix: *"The function reads the input file, parses each line, and **writes** the results to the database."* (three parallel present-active verbs, one subject) - **A10.** Dangling modifier. *Having reviewed the proposal* implies the committee did the reviewing, but the subject is *the budget*, which reviewed nothing. Fix: *"Having reviewed the proposal carefully, **the committee found** the budget insufficient."* - **A11.** Two problems. **Illogical comparison**: *the accuracy … is better than the previous version* compares accuracy (a number) to a version (a thing) — fix with *that of* (*"…better than **that of** the previous version"*). **Orphan-ish *this***: *this is a result* actually has a noun (*result*), so it's acceptable, though *"a result we did not expect"* could be folded in more tightly. Fixed: *"The accuracy of the new model is significantly better than that of the previous version — a result we did not expect."*Part B — Revise This ⭐⭐ (rewrite for correctness)
Rewrite each passage to fix every error, then note which errors you fixed.
B1. "After upgrading the firmware, the diagnostic LED was still blinking, this indicated that the boot sequence had not completed, the technician escalated the issue."
B2. "The report summarizes the findings, it includes recommendations, and three appendices are attached."
B3. A bulleted onboarding checklist: - Clone the repository - The dependencies should be installed - Running the local test suite - You will need to request VPN access
B4. "There are several factors that, due to the fact that they were not properly accounted for during the planning phase, had the ability to contribute to the delay of the project timeline."
B5. "The new caching layer, which was designed to reduce database load and that the team had benchmarked extensively before the rollout to confirm that it met the latency targets that had been agreed upon with the product team, was deployed on Tuesday." (This one is correct but unreadable — fix the rhythm, not just grammar.)
B6. "Neither the primary database nor the read replicas was available during the incident, the on-call engineer, who had been paged at 2 a.m., were unable to fail over."
Sample revisions B1–B6
- **B1.** *"After upgrading the firmware, the technician found the diagnostic LED still blinking. **This signal** indicated that the boot sequence had not completed, so the technician escalated the issue."* — Fixed: dangling modifier (supplied actor *the technician*), orphan *this* (→ *this signal*), and a comma splice / run-on chain (broke into sentences + subordinated with *so*). - **B2.** *"The report summarizes the findings, includes recommendations, and attaches three appendices."* — Fixed: comma splice and faulty parallelism (three parallel active verbs under one subject, *the report*). - **B3.** All imperative verbs: *Clone the repository · **Install** the dependencies · **Run** the local test suite · **Request** VPN access.* — Fixed: non-parallel bullets (was imperative, passive, participle, second-person). - **B4.** *"Several factors, unaccounted for during planning, contributed to the delay."* — Fixed: expletive (*there are…that* → promoted *factors*), *due to the fact that* (folded into *unaccounted for*), *had the ability to contribute* → *contributed*. Cut ~70%. - **B5.** Break the 50-word sentence for rhythm: *"The new caching layer was deployed on Tuesday. It had been designed to reduce database load, and the team benchmarked it extensively before rollout — confirming it met the latency targets agreed with the product team."* — No grammar error in the original; the fix is sentence *variety* (§6.7): one suspended monster split into a short sentence plus a medium one. - **B6.** *"Neither the primary database nor the read replicas **were** available during the incident**.** The on-call engineer, who had been paged at 2 a.m., **was** unable to fail over."* — Fixed: with *neither…nor*, the verb agrees with the nearer subject (*replicas*, plural → *were*); the second clause's subject is *engineer* (singular → *was*); and the comma splice between the two sentences became a period.Part C — Write This ⭐⭐–⭐⭐⭐ (produce text)
C1. Write a four-item deployment checklist for any process you know (deploying code, running a lab procedure, closing the monthly books). Make every item a parallel imperative verb. Then deliberately write a non-parallel version of the same list and label what's wrong with each item — proving you can see the difference.
C2. Take this single fact and write it as: (a) three short choppy sentences, then (b) one well-subordinated sentence that shows the causal chain. Fact: the disk filled up; logs couldn't be written; the service health check failed; the load balancer removed the node. Which version better shows what caused what — and why?
C3. Write a 60–80 word paragraph explaining any technical process to a colleague. Then audit it against the §6.9 list: mark any of the 20 errors you find in your own writing. Report your error tally. (Almost everyone finds at least one — usually an orphan this or an expletive.)
C4. Write one sentence that correctly uses a semicolon to join two independent clauses, one that correctly uses a colon to introduce a list, and one that correctly uses an em-dash for emphasis. Then write the same three ideas using only periods and commas, and decide which version reads better.
Self-assessment rubric (C1–C4): Full marks if (1) every produced list/series is grammatically parallel, (2) your subordinated version in C2 makes the cause–effect chain explicit, (3) your C3 audit names errors by number/family, not by vague feel, and (4) your C4 punctuation is correct (semicolon between independent clauses, colon after a complete clause, em-dash sparingly).
Part D — Synthesis & Critical Thinking ⭐⭐⭐
D1. Find the flaw. A colleague argues: "Grammar rules don't matter as long as the reader gets the point." Using two examples from this chapter (the only-placement deletion script and a dangling modifier in a methods section), explain a case where a grammar error doesn't just look sloppy but changes the meaning or causes real harm. Then state the actual stakes of sentence correctness in technical writing.
D2. Translate for audience. Take this sentence and rewrite it (a) for a peer engineer in a Slack message, (b) for a non-technical executive in a status email, and (c) for a formal incident report. Keep it correct in all three; notice how sentence structure and length shift with register: "The root cause was a race condition between the cache-warming job and the first incoming request, which returned stale data for roughly 90 seconds after each deploy." (This previews Chapter 7 — register — but tests this chapter's variety and correctness.)
D3. Defend a choice. §6.4 gave five ways to fix a comma splice. For the splice "The model overfit, we added regularization," pick the fix you'd use in (a) a research paper and (b) a quick team chat, and defend each choice by the relationship and register it implies.
D4. Cross-chapter integration. Chapter 3 said "cut 20–40%." This chapter said "vary sentence length; don't make everything short." Take a paragraph of your own writing and apply both: first cut inert words (Ch 3), then adjust the resulting sentence lengths for rhythm (Ch 6). Show before, after-cut, and after-rhythm. Reflect: did the two passes ever pull in opposite directions? Where?
Self-assessment rubric (D1–D4): Strong answers (1) tie a grammar error to a concrete consequence (data loss, misread instruction), not just "looks unprofessional"; (2) in D2/D3, justify choices by audience and relationship, not personal taste; and (3) in D4, treat concision and variety as complementary stages, identifying that Ch 3 removes inert words while Ch 6 shapes the rhythm of what remains.
Part M — Mixed Practice (Interleaved) ⭐⭐–⭐⭐⭐
These mix this chapter with earlier ones, so you must choose the right tool — sometimes it's a Chapter 6 grammar fix, sometimes a Chapter 3 clarity cut, sometimes both, sometimes neither.
M1. "It is important to note that the implementation of the new caching strategy, which the team had been working on, resulted in a reduction of the latency." — Is the main problem here a Chapter 6 grammar error or a Chapter 3 clarity problem? Diagnose, then fix. (Name which chapter's tools you used.)
M2. "The methodology section describes the apparatus, the procedure is explained in detail, and we discuss the limitations." — This has a parallelism problem (Ch 6) and a voice-consistency question (Ch 3, active/passive). Fix both and explain your voice choice.
M3. You're reviewing a draft. Sort these three problems into "structure" (Ch 4), "clarity" (Ch 3), or "sentence-level" (Ch 6): (i) the conclusion is buried in the last paragraph; (ii) the sentence "There are three options that we considered" pads its subject; (iii) the sentence "After analyzing the data, the trend was clear" dangles. For each, name the chapter and the fix.
M4. "The system should be fast and handle a lot of users." — This is vague (Ch 3) and will need a parallelism/precision fix when made testable (previews Ch 33). Rewrite it as one specific, parallel, measurable requirement.
M5. Take a real paragraph from something you wrote this week. Run the Chapter 5 question first ("which stage am I in?"), then the Chapter 3 clarity cut, then the Chapter 6 five-pass error hunt. Which pass found the most problems in your writing? That's your weak spot — name it.
Hints M1–M4
- **M1.** *Both.* Primary problem is [Chapter 3](../../part-01-writing-is-thinking/chapter-03-clarity/index.md) clarity (empty opener *it is important to note that*, nominalizations *implementation/reduction*). No grammar error per se. Fixed ([Ch 3](../../part-01-writing-is-thinking/chapter-03-clarity/index.md) tools): *"The new caching strategy cut latency."* - **M2.** Parallelism fix: make all three clauses share a voice. Active: *"The methodology section describes the apparatus, explains the procedure, and discusses the limitations."* (All active, parallel verbs.) Voice choice: active is cleaner here; if the field convention demands passive throughout the methods section ([Ch 35](../../part-07-writing-for-specific-fields/chapter-35-writing-for-science/index.md)), make all three passive instead — the point is *consistency*. - **M3.** (i) **Structure / [Ch 4](../../part-01-writing-is-thinking/chapter-04-structure/index.md)** — buried conclusion, fix with inverted pyramid. (ii) **Clarity / [Ch 3](../../part-01-writing-is-thinking/chapter-03-clarity/index.md) (and Ch 6 §6.8)** — expletive padding, → *"We considered three options."* (iii) **Sentence-level / Ch 6** — dangling modifier, → *"After analyzing the data, **we saw** the trend clearly."* - **M4.** e.g. *"The system **shall** return search results within 200 ms and **support** at least 10,000 concurrent users."* — Two parallel, measurable *shall* clauses (the testable-requirement pattern of [Ch 33](../../part-07-writing-for-specific-fields/chapter-33-writing-for-engineering/index.md)).Part E — Extension ⭐⭐⭐⭐ (optional, for the Deep Dive track)
E1. Style under constraint. Rewrite a paragraph from a paper or report in your field so that no sentence exceeds 15 words and no two consecutive sentences share the same opening structure. Then write a one-paragraph reflection: where did the constraint improve the prose, and where did it force you to drop a necessary subordinate relationship (forcing an all-short failure)? Use this to argue for or against hard sentence-length limits.
E2. Build your error signature. Collect five documents you've written over the past month. Run the §6.9 checklist on all five and tally errors by number. Identify your top three personal errors. Write a 150-word "self-editing protocol" tailored to your signature — the three things you'll check for first, every time. (This is more valuable than memorizing all twenty: you make the same handful repeatedly.)
E3. The grammar-superstition audit. Find three "rules" you were taught that this chapter calls superstitions (never split an infinitive, never start with and/but, never end with a preposition). For each, find one real sentence where breaking the "rule" produces clearer prose than obeying it, and explain why the superstition exists historically. Conclude with a principle for telling a real rule (prevents ambiguity) from a superstition (enforces etiquette).