ARGIRA v1.4.0: Accessibility Verification and Spanish Localization of the Three-Layer Uncertainty Communication Prototype

ARGIRA v1.4.0 is an incremental release of the ARGIRA prototype for automatic alt-text generation and three-layer uncertainty communication.

This release extends the v1.3.0-alpha prototype in two main areas: accessibility verification and Spanish localization.

Accessibility verification

The release includes manual accessibility verification performed by the developer using Android TalkBack on the forced experimental variants used in Phase 3C. The verification followed the project’s predefined screen-reader checklist. Eight of nine checklist points were confirmed; one interaction remains explicitly unresolved: the “More information” section should open only through activation, rather than merely when it receives navigation focus.

Apple VoiceOver was not evaluated because an Apple/VoiceOver testing environment was not available. This is therefore an untested scope rather than a failed validation.

Spanish localization

The release adds a Spanish-language variant of the prototype:

• argira_demo_v2.html — original English prototype.
• argira_demo_es.html — Spanish-localized variant.

The two language variants are provided as separate, independently executable HTML artifacts. Language selection in this release is performed by choosing the corresponding artifact. No in-application language selector, language toggle, automatic browser-language detection, or cross-link between the English and Spanish variants is provided.

Both variants derive from the same frozen v1.3.0-alpha prototype. The localization preserves the same experimental session mechanism, identifiers, classification logic, calibration logic, thresholds, and data flow.

The localization covers the previously defined interface and generated-content perimeter, including the accessibility description function and the audited generated-text templates. Where the English implementation uses language-specific article logic, the Spanish variant uses native Spanish grammatical construction while preserving the same semantic output and underlying classification and calibration behavior.

Structural verification confirmed parity between the English and Spanish artifacts, including the JavaScript function structure, element identifiers, SESSION_ID, and SESSION_CONDITION. Differences between the variants are restricted to the authorized localization content and the language-specific grammatical construction required for Spanish.

No new experimental condition, classification rule, calibration mechanism, or language-selection interface is introduced.

Continuity with v1.3.0-alpha

ARGIRA v1.4.0 does not introduce a new image-classification algorithm, calibration model, or experimental condition. The release preserves the underlying prototype and extends its accessibility documentation and verification, as well as its language availability.

The prototype remains an experimental research artifact and is not presented as a finished product or as validated accessibility technology.

This release is intended to support continued research, reproducibility, accessibility-oriented inspection, bilingual use through separate artifacts, and future evaluation with external users and additional assistive-technology environments.