🇬🇧 English · 🇪🇸 Español

Lexicon Reviewer Log

This log records who reviewed each language lexicon, against which anchor documents, and on what date. Required for any _<DOMAIN>_<lang> lexicon shipped under puremacro.narrative.indices._lexicons.

`_TECH_DOMAIN`

Language	Reviewer	Anchor docs (≥ 3)	Date
en	Jorge Alonso Ortiz	BBD AI-EPU 2023 paper §2; Felten-Raj-Seamans 2021 SMJ Tables 1–2; 3 FOMC speeches Powell (2023-05, 2024-01, 2024-06) referencing AI	2026-05-11
es	Jorge Alonso Ortiz	Banxico Informe Trimestral 2024-Q1 §III; BCRA Informe de Política Monetaria 2024-03; CEPAL "AI en el mercado laboral latinoamericano" 2023	2026-05-11
pt	Jorge Alonso Ortiz	BCB COPOM Ata 2024-09 (minutes 264); BCB Relatório de Inflação 2024-Q2 §3; Discurso Roberto Campos Neto 2024-04 (digitalização)	2026-05-11
de	Jorge Alonso Ortiz	Bundesbank Monatsbericht 2024-04 (KI-Wirtschaft); ECB Lagarde speech 2023-10-25 (AI / labour); Bundesbank Speech Mauderer 2024-03	2026-05-11
fr	Jorge Alonso Ortiz	Banque de France Bulletin 2024-Q1 §2 (numérique); ECB Lagarde 2024 Sintra opening remarks; Trésor Trésor-Éco n°337 (mai 2024)	2026-05-11
it	Jorge Alonso Ortiz	Banca d'Italia Relazione Annuale 2023 §III (digitalizzazione); Banca d'Italia speech Visco 2024-02; ISTAT Rapporto Annuale 2024 §3	2026-05-11
ja	Jorge Alonso Ortiz	BoJ Outlook Report 2024-10 (生成AI section); BoJ speech Ueda 2024-05; Cabinet Office White Paper on the Economy 2024	2026-05-11
zh	Jorge Alonso Ortiz	PBoC 2024Q1 Monetary Policy Report §专栏; PBoC Pan Gongsheng 2024 speech (人工智能); China Statistical Yearbook 2024 ch.20	2026-05-11

2026-05-12 — α.2 stability variants

Reviewer: Jorge Alonso Ortiz Scope: _*_EXPANDED lexicon variants for the 5 in-scope indices (lui, ltui, ltui_up, ltui_down, lwui) × 8 languages. Six families touched — _LABOR_DOMAIN, _UNCERTAINTY_TONE, _TECH_DOMAIN, _TECH_LABOR_UPSIDE, _TECH_LABOR_DOWNSIDE, _WAR_DOMAIN — for a total of 48 distinct _EXPANDED frozensets (plus 8 _LUI_PHRASES_* passthroughs reused verbatim).

Each expansion adds ≤ floor(0.20 × |base|) terms (cap enforced by puremacro/tests/test_narrative_lexicons_expanded.py), sourced from a per-family English seed list and translated into each target language:

_UNCERTAINTY_TONE_*: subdued / muted / cloudy / wary / hesitant.
_LABOR_DOMAIN_*: workforce / staffing / payrolls / labour-supply trend / manpower.
_TECH_DOMAIN_*: large language model / generative model / AI assistant / digitisation / automation wave.
_TECH_LABOR_UPSIDE_*: skill upgrade / productivity boost / new occupations / human–AI complementarity.
_TECH_LABOR_DOWNSIDE_*: labour-saving / obsolete skills / redundant roles / AI replacement.
_WAR_DOMAIN_*: hostilities / armed conflict / refugee flows / war economy / sanctions regime / military mobilisation.

Files: puremacro/narrative/indices/_lexicons_expanded.py, puremacro/tests/test_narrative_lexicons_expanded.py.

Purpose: stability-report perturbation only. Not used by any production index helper. Consumed exclusively by NarrativeStabilityReport (Sprint α-14) to measure how sensitive the headline cross-country correlations are to lexicon-curation choices.

Caveat: non-English additions are best-effort translations of the English seed, not curated against per-language anchor docs. This is intentional — the goal is a robustness perturbation, not a production lexicon. The base _lexicons.py (with full reviewer-doc backing) is not modified.