← Encyclopedia Skill

Reading a Draft Before Editing: Diagnosis by Levels

Introduction A draft does not need the same kind of repair everywhere. One sentence may be unclear because of its construction; another may expose information missing several pages earlier. Diagnosis by levels locates the actual problem so editing does not become a sequence of local corrections leaving the cause intact. What to inspect first Draft diagnosis begins with large-scale problems such as purpose, structure, and causality before word-level polishing, because perfecting sentences in a scene that will be cut wastes effort. Good feedback separates symptom from cause; a chapter that feels slow may suffer from repeated objectives rather than merely long sentences. A scene map shows who wants what and what changes in each unit without getting lost in local prose style. Continuity checking includes time, place, knowledge, objects, and emotional state rather than only physical errors. Reading aloud can reveal rhythm, repetition, and syntax that remain less visible during silent reading. Imagine a draft in which someone enters a meeting already knowing an outcome no character has told them and whose cause they did not witness. The entrance’s wording and the entire dialogue’s rhythm can improve while the knowledge problem remains. Supplying information in an earlier scene may alter the meaning of other passages, so the repair extends beyond the sentence where the symptom appeared. A chapter may likewise feel slow because it repeats a request already settled, not because its sentences are long. Summarizing each unit’s function reveals repetition concealed by different wording. Conversely, if a chapter’s goal and progression are clear and the reader stumbles over one pronoun reference, reconstructing the whole work is unnecessary. Good diagnosis distinguishes a wide problem with multiple symptoms from a local defect that can be repaired within its boundaries. It resists applying a reviewer’s preferred editing method to every kind of difficulty. Reading procedure Step 1: Summarize each chapter or scene in one line naming the action and change. Step 2: Classify each issue at the level of whole work, chapter, paragraph, or sentence. Step 3: Fix the highest level first and reread before moving downward. Step 4: Keep a dated version before each major revision so effects can be compared and changes reversed if necessary. Map each scene through its action and change rather than retelling all its details. Record new information and who possesses it at the end, then compare that condition with the next scene’s opening. When an object moves, a relationship changes, or knowledge is transferred, locate the passage explaining the transition. One dedicated reading can track an object’s possession and another knowledge or time, keeping observations distinct. Group problems that appear to share a cause and test them together before making scattered corrections. Preserve a separate file before each major revision and record its target and the dependent passages rechecked afterward. Once principal structural issues are settled, read sentences aloud to locate burdensome repetition or unclear connections. This final reading does not establish the correctness of events. It reveals a different class of problem that the scene map was not designed to detect. Connecting observation to context Professional editing commonly distinguishes developmental or structural editing, copyediting, and proofreading, although terminology varies among organizations. Digital publishing has made versioning and collaborative comments easier while increasing the need to control which version is authoritative. Literary revision may be individual or collaborative with an editor or readers, and each method has different strengths and biases. Variation in editorial-stage terminology does not erase the practical differences among organizing a story, refining expression, and correcting final errors. A project should define each role and who approves the authoritative version instead of assuming “review” means the same task to everyone. Collaborative comments also belong to particular versions. An observation may accurately describe an older text but no longer apply after a scene moves. Suggestions from different files should therefore not be combined without checking sequence. Writers’ self-reading benefits from knowledge of purpose but may skip over information held mentally and never placed on the page. A new reader can expose that absence while also bringing expectations unsuitable to the form. Adding more readers does not eliminate these differences; recording what each read and understood makes them usable. Limits of inference Applying every reader comment literally can create contradictions because readers differ in experience and expectation. Polishing language before structure is stable can hide a deeper problem under attractive prose. Automated editing tools should not replace human checking of meaning, tone, and context. The map must not replace attention to tone and detail. Two summaries may assign scenes the same function while their effects on character voice or setting differ. Examine what the summary omits before deleting either scene. Nor should every recommendation be implemented literally. One proposal may resolve ambiguity by adding explanation, while another removes the confusing construction that caused it; combining both can overburden the text. Choose the intervention based on the demonstrated cause, then retest the effect. When using automated editing, examine changes in attribution, time, and certainty as well as grammar. A smoother sentence may turn a possibility into a definite claim or attribute speech to the wrong person. Following meaning from the previous version to the new one remains part of diagnosis, not a separate task postponed until after approval. Conclusion Diagnostic reading is complete when every observation has a location, level, and plausible cause, and repairs are tested across related passages. More changes do not necessarily mean improvement, and fluent sentences do not guarantee a sound story. A clearer version resolves the defect at its actual level and returns to other levels to check that a successful local correction has not created a new problem elsewhere.