Radial Category Analysis

Description

Model a category, or all the senses of one term, as a central case plus extensions linked to it by named, independently justified principles, exposing why members belong.

A radial category has a central subcategory and conventional variants motivated by the centre but not predictable by rule, so they must be learned. The analysis collects every member of a category or every use of a term, identifies the central member or sense, and for each non-central item finds the link that motivates it: a shared domain of experience, a belief, an image-schema transformation, metonymy or metaphor. Each link type must be justified by other cases, not invented for one item. Chains (A to B to C) and unexplained exceptions are recorded. Lakoff's examples include the Dyirbal classifier balan, the Japanese classifier hon and the many senses of 'over'. For UX, the same structure explains why an existing navigation category or overloaded label holds members users cannot predict, and which links break first.

Attributes

Loss type
Unmigrated
Effort
Weeks

Steps

  1. Collect all members of the category, or all uses of the term in content, logs or interviews, with example contexts.
  2. Identify the central member or sense, usually concrete and basic-level, using centrality ratings where feasible.
  3. For each non-central item, name the link that motivates it: experiential domain, belief, image-schema transformation, metonymy or metaphor.
  4. Require each link type to be supported by other cases rather than invented for a single item.
  5. Record chains of links and list exceptions that no link explains.
  6. Use contrastive sentences or tasks to separate senses of an ambiguous term.
  7. Check the network against users' own explanations and against where users fail to predict membership, then decide which chains to keep, split or relabel.

Sources

  • Women, Fire and Dangerous Things (Lakoff, 1987)

Also known as

  • Radial category analysis (of a category's members or a term's senses)
  • Radial network analysis
  • Polysemy network analysis
  • Sense network analysis
  • Chaining analysis
  • Dixon's principles of classification

Categories

  • Content
  • Information Architecture & Content