2.3 Academic Style, Tense, and Logical Flow¶
What you will learn on this page
- A practical workflow for writing English papers with AI (division of roles between human and AI)
- Key style conventions in academic English (subjects and verbs, hedging, active vs. passive voice, nominalization, wordiness)
- Section-specific tense conventions and tense choices in citations (reporting verbs)
- How to make logical connections explicit (connectors, given-new, paragraph structure)
- “Pre-editing” tips for writing Japanese (L1) that is intended for translation
About prompt design and practical templates
Design principles for prompts based on the rules in this page, a set of templates by purpose, and a step-by-step refinement workflow are covered in detail in: 3.1 Prompt Design and Practical Refinement
A practical workflow for writing English papers with AI¶
Yanase (2023) proposes a basic four-stage workflow for writing English papers with AI, specifically targeted at native Japanese speakers (L1 Japanese speakers). Although these tips are intended for Japanese writers, I would like readers to consider them as a useful reference for effectively leveraging their L1.
- Planning and writing a Japanese draft: Design the story strategically (what to include and in what order)
- AI-based translation into English: Translate with machine translation (for example, DeepL)
- AI-based style improvement: Improve style with ChatGPT or similar tools
- Final proofreading by the author: Critically read the output and finalize the manuscript
Reference: Yanase, Y. (2023). “Writing English academic papers with AI: challenges for Japanese speakers and practical countermeasures.” Information Science and Technology, 73(6), 219–224. https://doi.org/10.18919/jkg.73.6_219
A key message in Yanase (2023) is this principle: the quality of the final product improves when humans do their best in areas where AI is weak. Current AI is relatively reliable for linguistic aspects such as spelling, usage, and style, but it has limits in understanding the research content in specialized fields. AI should be used as a general language assistant, not as a domain-specific research assistant (Yanase, 2023).
Why letting AI ''generate'' the text is problematic
Tools like ChatGPT can generate text, but what they produce tends to be generic and is not suitable for research writing where originality matters (Yanase, 2023). Such an approach also raises ethical concerns (see: 1.2 Academic Integrity and Plagiarism Prevention). This page focuses only on a workflow where you translate and revise a draft that the author has written.
Core style conventions in academic English¶
Academic English has conventions that differ from everyday English. The points below are especially important for Japanese writers.
Keep the subject short and place the verb early¶
Japanese writing often produces long subjects, but academic English prefers compact subjects and earlier verbs.
| Less effective style | Preferred style |
|---|---|
| The results of the analysis of the data collected from the participants in the experiment showed... | The analysis results showed... |
| What we found in our study was... | Our findings indicated... |
Hedges and boosters¶
Academic writing requires careful control of claim strength through hedges and boosters. Use hedges to present claims cautiously. Use boosters only for claims that are strongly supported by evidence.
| Strength | Hedges (weaker claims) | Boosters (stronger claims) |
|---|---|---|
| Modals | may, might, could | must, will |
| Verbs | suggest, indicate, appear to | demonstrate, confirm, establish |
| Adverbs | possibly, perhaps, arguably | clearly, certainly, undoubtedly |
| Adjectives | possible, probable, likely | clear, obvious, definitive |
Avoid overusing boosters
If you use expressions like clearly, obviously, or undoubtedly without strong evidence, reviewers may respond with “Is it really that certain?” Use boosters only when evidence strongly supports the claim.
More on hedging
For the role of hedging and common patterns, see: 2.2 Checking Conventional Expressions with Corpora
Active vs. passive voice and first-person pronouns¶
Academic writing uses active and passive voice strategically, depending on the section and purpose. Use of first person (we / I) also varies by field and journal.
Typical tendencies by section:
| Section | Typical tendency | Examples |
|---|---|---|
| Methods | More passive voice | Participants were recruited... |
| Results | Both active and passive | The analysis revealed... / A significant difference was found... |
| Discussion | More active voice | We argue that... / These findings suggest... |
Practical principles:
- When you want to make decisions visible → active voice (We selected... / We adopted...)
- When describing procedures generally → passive voice (The data were collected...)
- When presenting the authors’ claims → active voice (We argue that... / We propose...)
Tendencies by field:
| Field | Typical tendency |
|---|---|
| Natural sciences and engineering | Passive voice remains common; we is also increasingly used |
| Social sciences and applied linguistics | we is common in co-authored work |
| Humanities | I is often accepted |
Confirm conventions in your target venue
Read 2–3 recent papers from your target journal and check how often first-person pronouns are used. It can be efficient to ask AI to extract and list first-person instances from those papers.
Using nominalization¶
Academic English often uses nominalization (turning verbs or adjectives into nouns). This can compress information and help keep sentences concise.
| Verb/adjective | Nominalization | Example |
|---|---|---|
| analyze | analysis | The analysis of the data revealed... |
| differ | difference | A significant difference was found... |
| apply | application | The application of this method... |
| complex | complexity | The complexity of the task... |
| improve | improvement | A notable improvement was observed... |
What nominalization helps you do:
- Link sentences by referring back to the previous sentence with a noun phrase
The participants completed the task. → This completion took approximately 30 minutes. - Increase information density and shorten phrasing
Avoid excessive nominalization
Overusing nominalization can make writing too abstract. Avoid noun chains such as “The implementation of the improvement of the system...”.
Common wordy expressions¶
Literal translation from Japanese often leads to wordiness. Below are common patterns and more concise alternatives. Conventions vary across fields, so treat these as general guidelines.
| Wordy expression | Concise alternative | Note |
|---|---|---|
| in order to | to | to is usually enough |
| due to the fact that | because | expressing cause |
| it is important to note that | (delete and state directly) | weak preface |
| a large number of | many / numerous | quantity |
| at the present time | currently / now | time |
| in the case of | for / in | condition |
| has the ability to | can | ability |
| it should be noted that | (delete and state directly) | weak preface |
| the reason is because | the reason is that / because | redundancy |
| as a matter of fact | in fact / (delete) | overemphasis |
Tense conventions by section¶
Academic papers follow section-specific tense conventions. Incorrect tense choices can confuse readers.
| Section | Typical tense | Why | Example |
|---|---|---|---|
| Introduction (general background) | Present | general facts, established knowledge | Research shows that... |
| Introduction (specific prior studies) | Past | what a specific study did | Smith (2020) found that... |
| Methods | Past | completed procedures | Participants were recruited... |
| Results | Past | reporting obtained results | The analysis revealed... |
| Discussion (interpretation) | Present | interpreting meaning | These findings suggest... |
| Discussion (limitations) | Past or present | constraints of the study | The sample size was limited... |
A common mistake
Many Japanese writers use present tense in Results (The results show...), but past tense (The results showed...) is standard when reporting your own results. Present tense is often acceptable when referring to figures and tables (Table 1 shows...).
Reporting verbs and tense choice¶
When citing prior work, the tense of the reporting verb changes the meaning. For reporting verb types (neutral reporting, strong claims, suggesting, and so on), see: 2.2 Checking Conventional Expressions with Corpora
| Tense | Typical meaning | Example |
|---|---|---|
| Past | a specific action in a specific study | Smith (2020) found that... |
| Present perfect | a continuing stream of research | Several studies have shown that... |
| Present | widely accepted knowledge in the field | Research shows that... |
Decision rules:
- Citing a specific author and year → past (Tanaka (2023) reported...)
- Summarizing multiple studies → present perfect (Studies have indicated...)
- Referring to well-established knowledge → present (It is well established that...)
Typical points where tense shifts¶
Tense shifts are frequent, but they tend to follow patterns.
Typical pattern in the Introduction:
[Present] Research on X is essential for understanding Y.
[Present perfect] Several studies have examined the relationship between A and B.
[Past] Smith (2020) found that A was correlated with B.
[Present] However, the role of C remains unclear.
[Past/Present] The present study aimed to / aims to address this gap.
Typical pattern in the Discussion:
[Past] The results showed that A significantly affected B.
[Present] This finding suggests that...
[Past] Smith (2020) reported a similar pattern.
[Present] Taken together, these results indicate that...
Shift tense intentionally
Each tense shift should have a reason. If you cannot explain why the shift happens, it may be an error.
AI prompt example: tense and reporting verb check
Please check tense usage in the following manuscript excerpt.
Checklist:
(1) In Methods and Results, is past tense consistently used for what the authors did and found?
(2) In the Introduction/Literature Review, is past tense used for specific studies (Author, Year)?
(3) Are reporting verbs (found, suggested, demonstrated, etc.) appropriately strong for the claim?
If you find issues, list the sentence and explain what is wrong. Do not rewrite the text.
[Paste text here]
Logical flow and readability¶
Connectors¶
Academic English often uses connectors to clarify relationships between sentences and paragraphs. If every sentence begins with a connector, the text becomes heavy, so focus on major transitions such as contrast and cause.
Addition: Furthermore / In addition / Moreover Contrast/concession: However / Nevertheless / Although ..., ... Cause/effect: Therefore / Consequently / As a result / This is because ... / ... due to ... / ... , which led to ... Examples: For example / Specifically / In particular Summary: In summary / Overall / Taken together
Formality levels of connectors¶
Expressions can have similar meanings but different formality. Academic writing generally prefers the expressions on the right.
| Casual | More formal (academic) |
|---|---|
| But | However, / Nevertheless, |
| So | Therefore, / Consequently, |
| Also | Furthermore, / In addition, / Moreover, |
| Anyway | Nonetheless, / Regardless, |
| Plus | Additionally, |
| On top of that | Moreover, |
Avoid starting sentences with And / But / So
Starting a sentence with And, But, or So is common in casual English, but it is safer to avoid in academic writing. Starting with So is especially likely to be flagged by reviewers.
Information structure: the given-new principle¶
English sentences usually place information the reader already knows (given) early, and new information later. Following this principle makes sentence-to-sentence flow smoother.
Good example (given → new):
The participants completed a vocabulary test. The test consisted of 50 multiple-choice items.
Less effective example (new → given):
The participants completed a vocabulary test. 50 multiple-choice items constituted the test.
Three ways to maintain given-new flow:
- Use pronouns: The method was effective. It produced significant results.
- Use paraphrase: A survey was conducted. The questionnaire included...
- Use “This + noun”: Performance improved dramatically. This improvement can be attributed to...
Avoid using ''This'' alone
In phrases such as “This suggests...”, this can be unclear. Prefer “This finding suggests...” or “This result indicates...”.
Paragraph structure (TSSC)¶
A typical paragraph in academic English often follows this internal structure.
| Element | Role | Example |
|---|---|---|
| Topic sentence | State the main point of the paragraph | The results indicated a significant effect of... |
| Supporting sentences | Provide reasons, details, or examples | Specifically, the experimental group... |
| Summary/Synthesis | Summarize and return to the main point | These results suggest that... |
| Connection | Bridge to the next paragraph | However, the effect was not uniform across... |
One paragraph, one topic
Do not pack multiple topics into one paragraph. Start a new paragraph when you move to a new topic. Ideally, readers should be able to follow your argument by skimming topic sentences.
AI prompt example: paragraph-level logic check
Please analyze the following paragraphs.
Tasks:
(1) Identify the topic sentence of each paragraph.
(2) Check whether each paragraph has one clear topic.
(3) If a paragraph mixes multiple topics, point out where the topic shifts.
Do not rewrite the paragraphs. Provide feedback only.
[Paste text here]
Pre-editing Japanese drafts for translation¶
When you write a Japanese draft and translate it with AI, the quality of the Japanese input strongly affects the English output. Yanase (2023) frames this preparation as “pre-editing” and proposes principles like the following.
Principle 1: Essential point first¶
Make it a rule to state your main point early in the paragraph. Japanese writers often place the most important point at the end, but English readers expect the key point at the beginning. This is the same idea as placing the topic sentence first in the paragraph structure described above.
Principle 2: Keep perspective consistent¶
Keep a consistent perspective within a paragraph. For example, in a paragraph comparing X and Y to argue that X is advantageous, it is clearer when each sentence maintains X as the viewpoint.
| Mixed perspective | Consistent perspective |
|---|---|
| X has advantage A. Meanwhile, Y has drawback B. In addition, Y has disadvantage C. However, X has benefit D. | X has advantage A and benefit D. In addition, X does not have drawback B or disadvantage C, both of which Y has. |
Principle 3: Old information before new information¶
Place information the reader already knows early in the sentence, and new information later. See the given-new section above for examples.
Principle 4: One important idea per sentence¶
Keep sentences short and include one main idea in each sentence. Japanese writers often include many ideas in a single long sentence, which is a major cause of mistranslation in machine translation.
Also avoid sentences that are too short
If each sentence contains too little information, readers must integrate many short sentences and the text can become difficult to read (Yanase, 2023). Choose an appropriate granularity by imagining your reader.
Principle 5: Agent + action¶
English often prefers sentences where the agent appears as the subject and the action appears early as the verb. If you write Japanese with clear “X does Y” structure, translation quality tends to improve.
This also reduces overuse of structures that frequently appear in Japanese writers’ English, such as passive voice, It is … to/that …, There is …, and If/When …, ….
Do not be afraid of inanimate subjects
Try writing Japanese so that the topic behaves like an agent with an appropriate predicate. This often leads to inanimate-subject sentences that are unusual in Japanese but natural in English, improving translation quality.
Principle 6: Avoid excessive ellipsis¶
Japanese often omits subjects and objects, but when writing Japanese intended for translation, make sentence elements explicit as much as possible. This increases the chance that AI parses the structure correctly and reduces the burden of fixing wrong subjects or objects.
Input quality determines output quality
Many complaints such as “AI translation is unusable” originate from the Japanese input. Japanese written without translation in mind often turns into hard-to-read English (Yanase, 2023). Improving input Japanese meaningfully improves translated English.
Practical prompt templates for this page¶
Below are practical prompts that match the themes of this page. They are designed to reduce “generic rewrites” and keep you in control of meaning.
Prompt: academic style revision without changing meaning
Please revise the following text to sound natural in academic English.
Requirements:
- Do not add new claims, interpretations, or citations
- Keep technical terms unchanged
- Preserve hedging strength (do not make claims stronger)
- Keep the author’s intended meaning, but improve clarity and flow
Output format:
- Provide the revised text only
[Paste text here]
Prompt: reduce wordiness while preserving stance
Please identify wordy or redundant expressions in the following text and suggest concise alternatives.
Requirements:
- Keep the stance unchanged (do not increase certainty)
- Do not remove necessary methodological detail
- Provide a table with: Original phrase | Suggested concise phrase | Reason
[Paste text here]
Prompt: check logic signals and paragraph transitions
Please check whether the logical relationships between sentences and paragraphs are explicit.
Checklist:
- Are contrast and cause relationships signaled clearly?
- Are there abrupt jumps between paragraphs?
- Is “This” used without a clear noun reference (e.g., “This suggests...”)?
If you find issues, quote the exact sentence and explain the problem.
Do not rewrite the manuscript.
[Paste text here]
Next step: refine prompts in 3.1¶
To apply the style rules, tense conventions, and logic techniques explained on this page in a repeatable workflow, see: 3.1 Prompt Design and Practical Refinement