Write for the reader, not for the sentence

Scientific writing does not become more rigorous when every sentence is built from several layers of subordinate clauses. In practice, dense syntax often turns a paper into a reading-comprehension exercise. The first priority should be to make the meaning immediately available to the reader. Sentence style comes afterward.

This is especially important for papers written in English. Readers and reviewers generally prefer direct statements, clear logic, and confident claims. If a sentence is difficult to understand, a reviewer is unlikely to assume that the complexity is intentional; they may instead conclude that the meaning is unclear or that the authors are trying to conceal a weak point.

These principles were developed in a course designed largely for students in the United States, so they do not address every feature of academic writing in China. The underlying problems, however, are shared across languages. The same emphasis on clarity is useful when writing Chinese scientific papers as well.

Cut words that do not carry meaning

A useful first step is to write a complete draft without worrying too much about elegance, then remove anything that does not improve the argument. Common candidates include:

• empty introductory phrases such as “As it is well known”;
• vague expressions such as “basic tenets of”;
• unnecessary abbreviations and specialist labels, including abbreviations that save very little space;
• pairs of words that repeat the same function, such as “illustrate” and “demonstrate” when only one is needed;
• weak or unnecessary adverbs such as “very,” “really,” “quite,” “basically,” and “generally.”

The following example shows how a long formulation can be made more direct:

Dysregulation of physiologic microRNA (miR) activity has been shown to play an important role in tumor initiation and progression, including gliomagenesis. Therefore, molecular species that can regulate miR activity on their target RNAs without affecting the expression of relevant mature miRs may play equally relevant roles in cancer. –〉 Changes in microRNA expression play a role in cancer, including glioma. Therefore, events that disrupt microRNAsfrom binding to their target RNAsmay also promote cancer.

Several habits help achieve this kind of compression. Use fewer negative constructions when a positive statement will do. Avoid beginning sentences with “there is” or “there are” when the real subject can appear first. Also remove prepositions or connective words that add no difference in meaning. Readers can often infer a relationship without being guided through every possible grammatical link.

The goal is not to make every sentence short. It is to ensure that each sentence has a clear subject, a clear action, and a clear point.

Prefer active, simple sentences

The passive voice is often treated as more objective, but objectivity comes from the design of the experiment and the quality of the reasoning, not from grammatical voice. Passive constructions frequently omit the person or group responsible for an action, which can make a statement sound evasive or incomplete.

Active sentences are usually more vivid, more precise, and easier to follow. They also make responsibility clear. High-level journals are not necessarily opposed to first-person writing, and some landmark scientific papers use it openly. There is no reason to treat the active voice as unprofessional merely because it is less common in some local academic traditions.

The active voice can be used in almost every part of a paper. It can be combined with passive constructions when variation improves the flow. The methods section is the main exception: there, the procedure often matters more than the person who performed it, and readers usually look for specific methodological details rather than a narrative account of who did what. Passive voice can therefore remain useful in describing experimental procedures.

To convert a passive sentence, identify the person, group, or object performing the action and place it before the verb. The basic pattern is:

who does what to whom

Many passive sentences use a gerund or an abstract noun as the grammatical subject. Replacing that subject with the real actor often reveals a shorter and more direct sentence.

Choose verbs that show action

A sentence becomes difficult to read when its main verb is hidden inside a noun phrase or replaced by a weak form of “to be.” Strong verbs tell the reader what happened and reduce the amount of grammatical material surrounding the central idea.

Several transformations are especially useful:

• Replace a neutral verb plus an adjective with a single verb whenever possible.
• Look for opportunities to replace “to be” with a verb that expresses the actual action.
• Do not turn an active verb into a gerund merely to make the sentence sound formal.
• Keep the subject close to its verb so the reader does not have to search for the sentence’s action.

A few usage points are worth checking carefully. “Data” is conventionally treated as plural in formal scientific writing, while “datum” is singular. “Affect” usually means “to influence,” whereas “effect” is generally a noun meaning a result or influence. “Compared with” is often the safer choice when discussing differences between similar objects; “compared to” is more commonly used when emphasizing a similarity between different objects.

The distinction between “that” and “which” also affects precision. A restrictive “that” clause changes the meaning by identifying which object is meant and normally appears without a comma. A nonrestrictive “which” clause adds information that can be removed without changing the basic meaning and is set off by commas.

Finally, make pronouns agree with their subjects. Plural forms can sometimes avoid awkward gendered pronouns, but consistency matters more than any single stylistic preference.

Clarity is the common purpose

Concise sentences, active voice, and strong verbs are not three unrelated rules. They usually emerge from the same decision: write from the reader’s point of view. A paper overloaded with technical terms and elaborate sentence structures may look sophisticated to its author, but if the reader cannot follow the argument, reviewers may interpret the problem as poor English or poor reasoning.

Complex vocabulary and elegant syntax can improve an already clear paper. They cannot compensate for an unclear one. In scientific writing, accessibility is not a loss of authority; it is part of making the evidence persuasive.

Punctuation and paragraph structure

Let punctuation create rhythm and hierarchy

Scientific writing does not have to rely only on commas and full stops. Different punctuation marks divide information with different degrees of force. A useful rough hierarchy, from the lightest separation to the strongest, is:

comma < colon < em dash < parentheses < semicolon < full stop

In terms of formality, the em dash is generally less formal than parentheses, while the other marks tend to be more neutral in formal prose.

Each mark has a distinct role:

Semicolons

A semicolon can connect two closely related independent clauses. It can also separate items in a list when the items themselves contain commas, including a list introduced by a colon.

Parentheses

Parenthetical information provides an explanation or qualification that can be removed without damaging the main sentence. It is useful for secondary details, but excessive use can make the argument feel fragmented.

Colons

A colon can introduce a list of at least three items, a conclusion, a detailed explanation, a quotation, or an elaboration of the statement before it.

Em dashes

An em dash is effective for emphasis, definition, or an inserted description. It is more conspicuous than a comma and should be used when the interruption or emphasis is intentional.

The broader principle is that punctuation should reflect the logic of the sentence. If a sentence has become too complicated, replacing a punctuation mark with a full stop may be the clearest solution.

Give each paragraph one job

Paragraphing is not cosmetic. A paragraph should normally develop one idea, and keeping it to roughly five sentences or fewer often helps maintain focus. The first sentence should usually state the main point because some readers, including busy reviewers, may read the opening sentence of each paragraph more carefully than the rest.

The following sentences can explain, qualify, or support that point. A final sentence is useful when the paragraph is long or when one result deserves emphasis, but not every paragraph needs a formal conclusion.

A paragraph also needs internal order. Its sentences may proceed in parallel, from general to specific, or from evidence to implication, but the relationship should be apparent. This can resemble a rigid formula, yet research papers are not literary works in which ambiguity is an artistic virtue. Their purpose is to let readers understand the argument efficiently.

Transitional words are helpful when a real relationship needs to be signaled. They should not be inserted mechanically. In many cases, “but” is enough to express a contrast; a more elaborate transition is not automatically more precise.

Punctuation divides sentences, and paragraphing divides the paper. Both divisions should follow the logic of the argument rather than interrupt it arbitrarily.

Spend most of the writing time before drafting

The time required to write a paper cannot be separated completely from the earlier stages of research. Forming the hypothesis, designing and conducting experiments, and analyzing the data are not usually counted as “writing,” but a paper is difficult to produce if those stages have not generated an organized body of evidence and relevant literature.

Assuming the research itself is in reasonable shape, the actual writing process can be divided into four broad stages.

Prewriting: about 70 percent

The largest share of the work should go into organizing the material before drafting sentences. Searching for supporting papers while writing every paragraph at the same time often produces a disordered manuscript. It is more efficient to handle one task at a time: first determine what information belongs in the paper and where it belongs.

A classification system is essential. When reading a paper, you should be able to decide whether it belongs in the introduction, methods, results, or discussion, and what specific point it will support. A mind map can help. Put the central idea in the middle and branch out into background, methods, related studies, mechanisms, and other themes. This creates a tree-like structure for the literature.

Once the material is organized, close the computer and sketch the manuscript on paper. Decide the purpose and structure of each paragraph before writing it. This planning can begin well before the data analysis is complete because many structural questions can be considered while the research is still under way.

When the outline is finished, drafting becomes a matter of turning an existing structure into sentences rather than discovering the structure in real time.

Drafting: about 10 percent

A well-prepared outline makes the first draft much easier. The initial version should not be polished sentence by sentence. Follow the order of the ideas, make the logic visible, and keep moving. Word choice, grammar, and stylistic refinement can wait.

Stopping repeatedly to perfect a sentence interferes with thinking about the argument as a whole. The first draft’s job is to tell the story of the research. Later drafts can make that story more precise and economical.

Revision: about 20 percent

Revision is where most attention to expression belongs. Reading the paper aloud is particularly useful. Reading silently allows the mind to supply missing words and smooth over awkward transitions; speaking the sentences makes those problems easier to hear.

Check whether the verbs express real actions and whether active or passive voice is being used intentionally. These are not merely grammatical details: they affect how confidently the evidence is presented.

It is also valuable to ask a colleague from the same broad field but a different specialty to read the paper. Such readers are often closer to the audience that must understand the work than the author’s immediate collaborators are. If a scientifically trained colleague cannot follow the argument, the problem is probably in the writing rather than in the reader.

A professional editor can help when one is available, especially with recurring language problems. However, editing cannot repair a study whose design or evidence is fundamentally weak. The scientific level of a paper is largely determined while the experiment is being conceived and conducted.

Journal fit also matters. Manuscripts may be rejected because the subject does not match the journal’s scope, or because the authors have ignored the journal’s submission instructions. Careful language cannot compensate for an unsuitable target or failure to follow basic requirements.

Final checks

The last stage is for omissions and consistency. Check numbering, figure and table references, abbreviations, cross-references, and citations. Verify that every cited paper actually supports the statement attached to it.

References should be checked against the original articles whenever possible. Citation chains are not automatically reliable: published papers can misrepresent or oversimplify the work they cite, and those errors can be repeated by later authors.

The point of dividing the process into stages is not to impose an exact schedule. It is to prevent several different problems from competing for attention at once. First solve the problem of what the paper says; then solve how it says it; finally check whether all the details are consistent.

A practical order for writing the manuscript

A useful sequence is:

figures and tables → results → methods → introduction → discussion → conclusion → abstract

This order follows the way the research becomes concrete. Once the data have been analyzed, the first products are usually the figures and tables. The results section then explains what those displays show. The methods can be written next because they describe work that has already been completed. The introduction is a focused review of the relevant background. The discussion is usually the most demanding part because it must place the findings in a wider context. Once that argument is settled, the conclusion and abstract are much easier to write.

Figures and tables come first

When reading a large number of papers, researchers often move from the abstract and introduction directly to the figures. Visual displays can reveal the central idea faster than several pages of prose. The same principle applies when preparing a manuscript.

A paper’s contribution should be concentrated in a small number of figures or tables whenever possible. Each display should answer a recognizable question and, ideally, stand on its own. More figures do not necessarily make a paper stronger. The best ones reveal a pattern or relationship rather than merely placing raw measurements on the page.

The growing use of graphical abstracts reflects the same need: readers should be able to grasp the main idea quickly. A figure should therefore be both visually clear and scientifically informative.

Results describe what the data show

When a journal separates results from discussion, the results section should report the observed findings without turning into a second discussion. It should present the figures and tables, identify important relationships and trends, and explain information that cannot be shown conveniently in the visual display.

Do not repeat every number already visible in a table or graph unless the value is especially important. At the same time, describe the control experiments clearly so that the logic of the evidence remains complete.

A claim of significant difference should refer to statistical evidence, not simply to an apparent visual separation. Methodological explanations and comparisons with other researchers generally belong elsewhere.

Use the past tense for the experiments and observations themselves. When explaining what a figure contains, the present tense is appropriate: “Figure 1 shows …”

Methods should be precise and reproducible

The methods section is not the place for literary variation. Describe the experimental design accurately and in enough detail for readers to understand how the work was performed. If a published protocol was followed, cite the source. If the procedure was designed by the authors, a flowchart may make the sequence easier to follow.

Past tense and passive voice are both acceptable here, particularly when the procedure is more important than the identity of the person performing it. Statistical methods deserve the same level of precision as the experimental procedures; vague descriptions of analysis weaken confidence in the results.

The introduction moves from known to unknown

The introduction can be treated as a short, selective review rather than a complete survey of the field. A useful structure is:

1. What is already known about the research subject?
2. What remains unknown or unresolved?
3. What hypothesis or question does this study address?

This known–unknown–hypothesis progression mirrors the reader’s need to understand why the study was necessary. Begin with the larger context and gradually narrow the focus toward the specific work. Avoid burying the introduction in details that belong in a full review article.

The discussion returns to the larger field

The discussion is often the hardest section because it requires both evidence and judgment. It should first answer the question raised by the hypothesis. Then use the data to support that answer, bring in relevant literature, and examine why the findings agree with or differ from previous studies.

In other words, the discussion expands the unknown: it explains what the new results add to an unresolved issue. It then places those results back into the known background and considers what they may clarify.

Do not extend the interpretation beyond what the experiment can support. A careful discussion should address limitations and possible future directions rather than presenting every implication as established fact. At the same time, avoid hiding the main result behind excessive caution. Strong, direct statements are more useful than a chain of vague qualifications, provided the evidence supports them.

Keep the conclusion short

The conclusion does not need to repeat the entire paper. State the central finding and its most important implication clearly. This is the place to show what the study adds, not to introduce a new line of argument.

Write the abstract last

A compact abstract can often be built from a few essential sentences: one for the background, one for the research question or hypothesis, one for the methods, one for the central result, and one for the conclusion or a reasonable implication.

Other structures may suit particular journals or types of study, but the same principle applies. Many readers will decide whether to continue based on the abstract alone. Writing it after the rest of the manuscript forces the author to identify the paper’s actual story rather than the story originally imagined at the beginning of the project.

The writing order therefore moves from relatively concrete material to increasingly interpretive material. Beginning with the discussion can encourage speculation that has not yet been anchored in the actual data.

Keep the ideal model flexible

Not every field writes papers in exactly this way. In some areas, especially those with highly standardized experimental workflows, the manuscript may follow a nearly automatic path from characterization to application. A researcher can sometimes learn the structure simply by studying well-written papers in the field.

That variation does not make writing principles irrelevant. Different disciplines may use different conventions, but every paper still benefits from clear claims, visible logic, disciplined paragraphing, and careful attention to the reader.

The most useful habit is to examine the work from the perspective of a demanding reviewer from the beginning of the project. Ask difficult questions about the hypothesis, controls, analysis, interpretation, and limitations before someone else asks them during peer review. The more problems the authors identify and resolve themselves, the fewer problems the manuscript will leave for reviewers to discover.