Representational Consistency Analysis (Language of Thought)

Description

Audits whether interface language and symbols compose systematically, so that users can combine known terms and structures to understand new commands and states.

Fodor's language of thought hypothesis proposes that thinking occurs in a compositional mental representational system, where meanings of complex thoughts are built from the meanings of their parts and structure, giving thought systematicity: anyone who can think 'A acts on B' can think 'B acts on A'. Applied to interface design, representational consistency analysis asks whether the product's language, icons and syntax are similarly compositional, so users can generalise from known constructions to novel ones. The analyst inventories terms, symbols and constructions, checks that each maps consistently to one concept, and tests whether combinations behave predictably. Rigour comes from a systematic inventory, explicit grammar rules, and verification with users attempting novel combinations.

Attributes

Loss type
Extinct
Stage
Design
Effort
Days

Steps

  1. Compositionality audit for interface language: for each multi-word label or complex instruction, test:
  2. Meaning of the whole = function of meaning of parts?
  3. (2) Ask users to predict what 'X Y' means, given they know what 'X' means and what 'Y' means separately.
  4. (3) Compute prediction accuracy. < 70% accuracy on a label's composed meaning = the composition is non-transparent. Redesign: either break the composition or make components more predictable.

Sources

  • Language of Thought — Jerry Fodor

Also known as

  • Mentalese consistency analysis
  • Compositional interface language audit

Categories

  • Content
  • Linguistics, Semiotics & Discourse