editor - VIBE score

V - Viewpoint score

What is it? Why did we make it? How to get the most out of it.

From all the research we did in building the VIBE score, the most commonly researched characteristic on what makes for great writing is your viewpoint.

It turns out that it's a very human thing to have. An opinion.

Now, how you express that opinion, and all it's subtleties are taken into account with the V score.

Stance

What it is

How balanced and intentional your stance sounds — the mix of hedges (might, could) and boosters (clearly, definitely), not just their presence but their ratio across the text.

What it looks for

Presence of hedges and boosters per sentence
Balance around ~1:1 hedge/booster (not all swagger, not all hedging)
Coverage across windows (not just one spiky paragraph)

Why it matters

Balanced stance correlates with credibility and reader trust: too many boosters → bluster; too many hedges → timidity. Reader perception of author confidence is highly sensitive to these markers. scholar.google.com

Research anchors (quick takes)

Hyland – hedges/boosters as interactional metadiscourse and authorial identity; balance supports credibility and reader alignment. (Multiple papers on stance/self-mention/engagement). scholar.google.com

Scoring gist (you’re using):
Per sentence: score = GaussianPenalty(|hedges/boosters − 1|) + 1.2*(hedges + boosters)
Per window: combine with mean × coverage (geometric-ish softening).

How to improve (tactics)

Add one hedge where you currently over-boost (“This clearly works” → “This clearly works, but it might fail at scale”).
Add one booster where you’re over-hedged (“It might help” → “It definitely helps if you do X”).
Spread these across sections (coverage).

Perspective

What it is

Cohesive viewpoint management with discourse connectives that acknowledge alternatives and contrast (however, yet, at the same time, whereas, even though…).

What it looks for

Variety and density of contrastive/connective markers
Distribution across windows (coverage)
Diminishing returns for repeats

Why it matters

Discourse connectives make relations explicit (contrast, concession, elaboration), improving coherence and perceived sophistication. WAC Clearinghouse

Research anchors

Pitler & Nenkova (2009) – connectives signal discourse relations; diversity relates to coherence modeling. WAC Clearinghouse

Scoring gist:
Per sentence: log-discounted gains per unique connective; window uses mean × coverage.

How to improve

Add contrastive pairs once per section (“however”, “yet”, “even though”).
Mix types (contrast + concession + elaboration) — avoid hammering just “however”.

Novelty

What it is

Lexical sophistication without jargon dump: proportion of uncommon / morphologically rich words at a sane density.

What it looks for

Words outside common 5k list
Morphological signals (-tion, -ment, hyphen compounds, length > 7)
Cap the density so it never becomes sludge

Why it matters

Lexical sophistication correlates with perceived expertise only up to a point; over-use hurts readability. Measures like lexical frequency profiles and sophistication are standard in applied linguistics. Oaji

Research anchors

Laufer & Nation (1995) – lexical richness/sophistication measures. Oaji

Scoring gist:
Per sentence: count sophisticated tokens; score only if density ≤ ~25%. Window: mean × coverage.

How to improve

Replace 1–2 garden-variety words with precise terms (“bad” → “pathological”, “help” → “ameliorate”) — but don’t stack them back-to-back.
Use coined or hyphenated task terms sparingly to lift signal.

Clarity

What it is

Evaluative clarity: explicit, compact appraisal language that makes judgments and criteria legible.

What it looks for

Evaluative adjectives/adverbs (clear, wasteful, brittle, elegant) at a sweet-spot density
Gaussian penalty when density is too low or too high

Why it matters

Appraisal theory shows explicit evaluation helps readers track stance, criteria, and judgments; over-marking reads as hype. PMC

Research anchors

Martin & White (2005) – Appraisal/The Language of Evaluation (framework for attitude, engagement, graduation). PMC

Scoring gist:
Per sentence: score = cap * exp(−(pct − 3.5%)² / 2σ²); window with coverage.

How to improve

Add one precise judgment per key claim (“This migration plan is risky because…”).
Trim vague intensifiers (very, really) and swap for crisp appraisals.

Empathy

What it is

Cognitive empathy and concession: markers that acknowledge audience feelings, constraints, or counter-positions (“I get it…”, “you’re right that…”, “granted…”).

What it looks for

Empathy markers and concession frames
Spread across sections (coverage), not clumped

Why it matters

Perspective-taking and concession increase persuasion and reduce reactance; readers feel “seen,” lowering resistance. SAGE Journals

Research anchors

Perspective-taking & persuasion (behavioral work showing empathy/perspective-taking improves receptivity). SAGE Journals

Scoring gist:
Per sentence: weighted counts of empathy+concession; window: mean × coverage.

How to improve

Add a one-line concession before advice (“You’re swamped; granted, a rewrite sounds expensive. Here’s the 30-minute fix.”).
Mirror the reader’s felt risk (“If pricing hikes make you look bad to your boss, this rollout avoids surprises.”).

Polarity

What it is

Healthy emotional range (positive/negative/neutral) at the token level, measured as entropy across the three classes.

What it looks for

Tokens from positive and negative lexicons (not just neutral)
Window scores based on entropy thresholds (≥ 0.8 = full credit; 0.5–0.79 = half)

Why it matters

Flat affect reduces engagement; mixed valence increases attention and recall. We operationalize it via Shannon entropy over sentiment classes; the method is standard in sentiment analysis surveys. shura.shu.ac.uk+1

Research anchors

Pang & Lee (2008) – survey of opinion mining methods. shura.shu.ac.uk
Shannon (1948) – entropy as uncertainty/information. journals.muni.cz

Scoring gist:
Per sentence: count pos/neg/neutral; compute entropy H. Map H to tiered score; window with coverage.

How to improve

Pair praise + pain (“It’s a beautiful API, but the auth flow punishes beginners.”).
Don’t fear negative affect; it earns trust when it’s precise.

Explicit

What it is

Direct, first-person viewpoint markers and commitment phrases that make ownership and judgments unambiguous (“I think…”, “we recommend…”, “let me be clear…”).

What it looks for

First-person stance openers and explicit commitment phrases
Diminishing returns; coverage beats clumps

Why it matters

Self-mention and engagement markers signal accountable authorship and guide the reader’s alignment. ijsshr.in

Research anchors

Hyland (self-mention/engagement) – authorial presence and reader engagement devices. ijsshr.in

Scoring gist:
Per sentence: count explicit markers (with a light first-token heuristic); window: mean × coverage.

How to improve

Start summary or guidance paragraphs with I/We + action (“We recommend shipping the pricing note 7 days prior.”).
Use one commitment intensifier per section (“Make no mistake: this is a trust breaker.”).

Rhetorical

What it is

Device mix: measured presence of engaging moves — rhetorical questions, concise imperatives, and occasional punchy (≤6-word) sentences.

What it looks for

Q-forms (“So what changes?”), imperatives (“Ship it.”), short punches scattered
No spam: variety over repetition

Why it matters

Rhetorical questions increase processing depth and engagement; directives create momentum; punchiness adds rhythm when used sparingly. libguides.macewan.ca+1

Research anchors

Rhetorical questions & processing – evidence that questions can boost persuasion/engagement under central processing. libguides.macewan.ca
Directive/imperative as speech act – pragmatics of directives in guiding action. academypublication.com

Scoring gist:
Per sentence: devices = rq + imperatives + punchFlag; window: mean × coverage with caps.