VISUAL TOWNSCAPE · MALAYSIA

Discover the language of the visible townscape.

A research control surface for moving from naturally occurring language to candidate lexical structure, then into human validation and psychometrics.

Working construct

Townscape is the relational organisation and appearance of the visible external environment as perceived within a visual field and, where relevant, through movement.

CURRENT STATE

Study at a glance

Data export required

—Malaysia pages selected

—sentence units

—pages represented in sentences

500planned human QA sample

Pipeline

Construct definitionComplete

Corpus ingestionComplete

Malaysia preprocessingComplete

Corpus QAHuman review required

Lexical-universe annotationnext

Lexical candidate extractionnext

Lexical clusteringnext

Human psychometricslater

Research logic

CorpusNatural language before theory seeding

Lexical candidatesWords and short expressions with source context

Lexical familiesSemantic clustering, not factors

PsychometricsOptional downstream validation

Study principle

“Townscape is not the sum of visible objects.”

The computational stage is designed to recover the vocabulary people use to describe the visible scene, preserving object attributes, relations, appearance, evaluation, metaphor, and local expressions before any latent structure is proposed.

Upstream ingestion snapshot: 68,444 source pages, collection run WV-2026-08. The dashboard treats the local DuckDB export as derived data and Parquet as authoritative.