Academic Writing Style Guide

Table of Contents

• Core Principles
  • 1. Clarity Over Complexity
  • 2. Precision Over Vagueness
  • 3. Active Voice (When Appropriate)
• Sentence Length Guidelines
• Paragraph Structure
  • Topic Sentence
  • Supporting Sentences
  • Transition
• Academic Vocabulary
  • Reporting Verbs (Neutral)
  • Reporting Verbs (Strong Agreement)
  • Reporting Verbs (Tentative)
  • Reporting Verbs (Critical)
• Transition Words
  • Addition
  • Contrast
  • Cause/Effect
  • Example
  • Sequence
• Citation Integration
  • Integral (Author as Subject)
  • Non-Integral (Content Focus)
  • Paraphrase (Preferred)
  • Direct Quote (Sparingly)
• Common Section Patterns
  • Introduction
  • Related Work
  • Methodology
  • Results
  • Conclusion

---

Core Principles

1. Clarity Over Complexity

• One idea per sentence
• Avoid nested clauses when possible
• Define terms on first use

2. Precision Over Vagueness

• Use specific numbers instead of "several" or "many"
• Quantify claims whenever possible
• Avoid hedging without evidence

3. Active Voice (When Appropriate)

• ✅ "We propose a novel method..."
• ✅ "This paper presents..."
• ❌ "A novel method is proposed by us..."

Sentence Length Guidelines

Type	Word Count	Use Case
Short	10-15	Key findings, transitions
Medium	15-25	Most content
Long	25-40	Complex relationships
Very Long	>40	⚠️ Consider splitting

Paragraph Structure

Topic Sentence

First sentence states the main point.

Supporting Sentences

• Evidence, examples, or elaboration
• 3-5 sentences typical
• Clear logical flow

Transition

Connect to next paragraph if needed.

Academic Vocabulary

Reporting Verbs (Neutral)

• states, notes, observes, reports, describes

Reporting Verbs (Strong Agreement)

• demonstrates, shows, proves, confirms, establishes

Reporting Verbs (Tentative)

• suggests, implies, indicates, proposes, hypothesizes

Reporting Verbs (Critical)

• claims, argues, asserts, alleges, maintains

Transition Words

Addition

• Furthermore, Moreover, Additionally, In addition

Contrast

• However, Nevertheless, Conversely, On the other hand

Cause/Effect

• Therefore, Consequently, As a result, Thus

Example

• For instance, For example, Specifically, In particular

Sequence

• First, Second, Subsequently, Finally

Citation Integration

Integral (Author as Subject)

"Smith et al. [1] demonstrated that..."

Non-Integral (Content Focus)

"Deep learning has shown remarkable success [1-3]."

Paraphrase (Preferred)

Restate the idea in your own words with citation.

Direct Quote (Sparingly)

Only for definitions or exceptionally well-phrased ideas.

Anti-Citation-Stacking Rules (Introduction & Related Work)

Stacking 3+ references without individual discussion is a common AI writing pattern and is unacceptable in top-tier venues. These rules apply to both Introduction and Related Work sections.

Rules:

1. Max 2 clustered citations without discussion per sentence

   • ❌ "Many methods have been proposed [1], [2], [3], [4], [5]."
   • ✅ "Smith et al. [1] proposed X for scenario A. Building on this, Jones [2] extended the approach to B, while Wang et al. [3] addressed limitation C."

2. Every cited work must earn its citation with at least one of:

   • A summary of its core contribution (1 clause minimum)
   • A comparison with another cited work
   • A specific limitation that motivates your work

3. Narrative over parenthetical in Introduction and Related Work body:

   • Use integral citations (author as subject) for key works: "Smith et al. [1] demonstrated..."
   • Reserve non-integral citations (content focus) for well-established facts only: "Gradient descent is widely used [1], [2]."

4. Funnel-appropriate density (Introduction):

   • Background paragraph (broad context): up to 2 clustered citations for established facts
   • Problem statement paragraph: each citation must be individually discussed
   • Gap/motivation paragraph: every cited limitation must reference a specific paper

5. Categorical discussion (Related Work):

   • Group works by methodology/approach, not chronologically
   • Within each group, discuss each work's specific contribution and limitation
   • Use comparative language between works: "Unlike [1], method [2] addresses..."

Positive patterns (Introduction):

Smith et al. [1] proposed method X, achieving Y% accuracy on dataset Z.
However, their approach assumes A, which limits applicability to B.
Jones [2] relaxed this assumption by introducing C, but at the cost of D.
In contrast, our method addresses both limitations by...

Positive patterns (Related Work):

\textbf{Transformer-based methods.} Vaswani et al. [1] introduced the
self-attention mechanism for sequence modeling. Li et al. [2] adapted this
architecture for time series, but their method requires O(n^2) memory.
Zhou et al. [3] proposed ProbSparse attention to reduce complexity to O(n log n),
though at the cost of approximation error.

Negative patterns (FORBIDDEN):

Many researchers have studied this problem [1], [2], [3], [4], [5].
Several methods have been proposed [6], [7], [8], [9], [10], [11], [12].
Recent advances include [13], [14], [15], [16], [17], [18], [19], [20].

Common Section Patterns

Introduction

1. General context → Specific problem
2. Gap in existing work
3. Contributions of this work
4. Paper organization

Related Work

1. Categorize existing approaches
2. Compare and contrast
3. Position your work

Methodology

1. Overview of approach
2. Detailed steps
3. Justification for choices

Results

1. Experimental setup
2. Quantitative results
3. Qualitative analysis
4. Comparison with baselines

Conclusion

1. Summary of contributions
2. Limitations
3. Future work