VAD cube
Valence, arousal and dominance on −1..1. Where the state sits. Mehrabian & Russell, used not owned.
Lexicon · Version 1.0.0 · 131 terms
Not thousands of rated tokens. One hundred and thirty-one authored feeling words you can show someone back to themselves — with VAD, wheel tier, and family intensity, and without cactus, trashbag, or a clinical insult sitting next to a coordinate.
Nearest-neighbour dumps from a general word list can label a person cactus, trashbag, fart, or dialysis because those tokens happen to sit near a VAD point. That is clinically damaging and absurd. This lexicon contains only plain-English feeling words suitable to show a client. It is authored, not scraped. It is not NRC-VAD.
Valence, arousal and dominance on −1..1. Where the state sits. Mehrabian & Russell, used not owned.
Broad, mid, fine — generalisation versus specificity, like an emotion wheel. Far zoom stays at the hub.
Authored rank inside one family. Not arousal. Tired can be more intense than depleted-looking activation.
On a classic emotion wheel, general/weak sit at the hub and specific/strong sit on the petals — one radius. They often travel together. They are not the same question. Angry is general and fairly strong. Irritated is weaker — everyday, not a hub. Furious is specific and strong. Tier is what a zoom may say. Intensity is whether it escalated inside the family.
Where in the cube. Sensors may emit raw VAD; words still come from this list.
Hub is general. Petals are specific. A theme must not be labelled with a fine petal.
Strength inside one family, weakest first. Cross-family comparison is unavailable.
Rows are intensity families. Columns are wheel tiers. Empty cells are allowed: not every family has a hub word, and some states have no intensity at all.
General, and fairly strong. A hub word for a whole stretch of session.
Weaker. Everyday mid-tier — not a hub word, and not an escalation.
Specific, and strong. The Plutchik outer petal — here the two axes line up.
The case that looks like one wheel radius: more specific and more intense together. Still two moves.
A Plutchik wheel fuses those into one radius. We split them because zoom must not confuse “how precise may I be?” with “did this escalate?”
A lexical match is evidence, not a diagnosis. Phrase-sense analysis (fine, I’m done) is not a second lexicon. Categorical centroids (Calvo & Kim) remain a separate hub layer. Synthetic oversampling may rebalance observations; it must not invent new feeling words.