Sample Analysis Guide

How to systematically analyze writing samples to extract voice DNA patterns.

---

The Analysis Philosophy

Sample analysis is the foundation, but not the whole building:

1. Samples reveal patterns — They show what the writer does, not always why.
2. Interview confirms patterns — Samples might be atypical; interview validates.
3. Patterns need examples — Always capture quotes that demonstrate patterns.
4. Gaps must be noted — What can't we learn from this sample?

Never finalize a pattern based on samples alone without interview confirmation.

---

Before You Begin

Gather Context

Before analyzing, know:

• What register is this? (blog, fiction, technical, essay, email)
• Is this typical? (or was it written for an unusual context)
• Was it edited? (by the writer alone, or by others)
• How old is it? (recent, or potentially outdated)
• How long is it? (longer samples = more reliable patterns)

Ideal Sample Characteristics

More reliable:

• Recent (last 1-2 years)
• Self-edited (not heavily touched by others)
• Typical of what they normally write
• At least 1000+ words
• Multiple samples (2-3 minimum)

Less reliable:

• Old (style may have evolved)
• Heavily edited by others
• Written for unusual circumstances
• Very short (under 500 words)
• Single sample only

---

The Three-Pass Approach

First Pass: Gut Check (2-3 minutes)

Read the sample quickly. Note immediate impressions.

Ask yourself:

• What 3 things jump out about this writing?
• What's the overall temperature? (warm, cool, neutral)
• What register/mode is this?
• Does anything feel distinctive or unusual?

Capture gut impressions before they fade. First reactions often identify the most distinctive patterns.

Example gut notes:

"Short sentences. Very direct. Almost aggressive. Lots of 'But' at sentence starts. No hedging at all."

---

Second Pass: Quantitative Scan (5-10 minutes)

Measure objective patterns.

Sentence Metrics

Punctuation Count

Word Patterns

Quick Vocabulary Check

• Any recurring words?
• Latinate or Anglo-Saxon preference?
• Technical terms present?
• Jargon or colloquialisms?

---

Third Pass: Qualitative Analysis (10-15 minutes)

This is where you find the distinctive patterns.

Sentence Architecture

• Internal structure: Simple, compound, or complex sentences?
• Clause patterns: Where do they nest clauses?
• Emphasis placement: Front-loaded or end-weighted?
• Fragments: Present? Intentional?

Look for:

• Unusual constructions
• Signature rhythms
• Consistent patterns across the sample

Paragraph Patterns

• Construction: How do they build paragraphs?
• Topic sentences: First, last, implied?
• Length variation: All similar, or strategic variety?
• Endings: How do paragraphs conclude?

Transitions

• Explicit markers: "However," "Furthermore," etc.
• Implicit flow: Ideas connect without markers
• Bridge words: Repeated concepts that link ideas
• White space: Strategic breaks

Opening & Closing

• How does the piece begin? (hook, scene, question, statement)
• How does it end? (callback, question, definitive, trailing)
• Are these moves typical? (compare across samples if possible)

Tone Markers

• Warmth indicators: Personal anecdotes, "I/you" language, emotional words
• Coolness indicators: Impersonal constructions, analytical distance
• Confidence level: Hedging vs. assertion
• Authority stance: Expert, guide, fellow traveler

Signature Elements

• What's unusual? Things you don't see in most writing
• What's consistent? Patterns that repeat
• What's recognizable? Things that would identify this writer

---

Comparative Analysis

When you have multiple samples:

Consistency Check

Consistent patterns = High confidence, can include in profile without caveat

Inconsistent patterns = Need investigation:

• Different registers?
• Evolution over time?
• Context-dependent variation?
• One sample is atypical?

Variation Analysis

When patterns differ across samples:

1. Identify the variation: What's different?
2. Seek explanation: Is there a reason?
3. Determine whether to:
   • Capture dominant pattern (most frequent)
   • Capture context-dependent pattern (when X, do Y)
   • Treat as different registers (separate DNA docs)
   • Investigate further in interview

---

What to Look For (Quick Reference)

Highly Distinctive Patterns

These are often the most identifying:

• Unusual punctuation habits
• Sentence opening patterns
• Paragraph ending moves
• Distinctive transitions
• Pet phrases or constructions
• Unusual word choices
• Signature rhythms

Common But Important Patterns

These establish the baseline:

• Sentence length tendencies
• Formality level
• First person presence
• Contraction usage
• Paragraph length
• Tone/temperature

Often Overlooked Patterns

Don't forget to check:

• What's absent (things they never do)
• Opening moves across multiple pieces
• How they handle uncertainty
• Question usage patterns
• Humor placement and style

---

Gap Identification

After analysis, note what you CAN'T determine from samples:

Common Gaps

• Intent: Why do they make certain choices?
• Aspirational vs. actual: Is this how they want to write or how they do?
• Context-dependence: Does this change in other situations?
• Evolution: Is this current or outdated?
• Editing impact: How much was changed by others?
• Anti-patterns: What do they avoid? (Absence is hard to prove)
• Humor: If not present in samples, does that mean none exists?

Interview to Fill Gaps

For each gap, identify the interview question that would fill it:

---

Red Flags

Signs the sample may not be representative:

Content Red Flags

• Heavy jargon in supposedly casual writing
• Sudden formality shifts mid-piece
• Inconsistent voice within the same piece
• Generic AI-like patterns (might not be their writing)

Context Red Flags

• Sample is very old (5+ years)
• Sample was for unusual context (school, job application)
• Sample was heavily edited by others
• Sample is much shorter than typical work
• Only one sample available

Pattern Red Flags

• Patterns that contradict user's self-description
• Patterns wildly different from other samples
• Patterns that match AI writing tells

When red flags appear: Note them. Ask clarifying questions in interview. Don't assume sample represents true voice.

---

Sample Analysis Worksheet

Use this for systematic analysis:

SAMPLE INFO
- Title/description:
- Date written:
- Word count:
- Register/mode:
- Edited by others?:

FIRST PASS (gut impressions)
- 3 things that jump out:
  1.
  2.
  3.
- Overall temperature:
- Initial distinctiveness notes:

SECOND PASS (metrics)
- Avg sentence length:
- Sentence range:
- Punctuation notable:
- Paragraph lengths:
- "I" frequency:
- Contraction use:

THIRD PASS (qualitative)
- Sentence architecture:
- Paragraph patterns:
- Transitions:
- Opening move:
- Closing move:
- Tone markers:
- Signature elements:

PATTERNS IDENTIFIED
- Strong patterns (confident):
  1.
  2.
  3.
- Tentative patterns (need confirmation):
  1.
  2.
- Potential anti-patterns:
  1.

GAPS TO FILL
- What can't I determine?
- Questions for interview:

RED FLAGS
- Any concerns about this sample?

---

From Analysis to DNA Document

Translating Findings

Confidence Levels

Rate each pattern:

• High confidence: Consistent across samples, confirmed in interview
• Medium confidence: Consistent in samples, not yet confirmed
• Low confidence: Based on single sample or inconsistent across samples

---

Common Analysis Mistakes

1. Overgeneralizing from one sample — Wait for confirmation before finalizing patterns.

2. Missing the forest for trees — Don't get lost in metrics. What makes this writer recognizable?

3. Ignoring gut reactions — First impressions often catch the most distinctive patterns.

4. Treating absence as avoidance — Just because something's not in the sample doesn't mean they never do it.

5. Assuming samples are representative — They might be unusual. Always check.

6. Skipping gap identification — Know what you don't know.