Skills Reference

The Computational Linguistics suite contains 18 skills organized across a 5-phase pipeline (Scope → Acquire → Analyze → Evaluate → Release). The magic-linguistic-orchestrator coordinates routing; 14 specialist skills own phase-specific content; 3 optional Mindset stubs cover cross-cutting scenarios that can fire at any stage.

Pipeline Overview

Phase 0: Scope    → scope, scripts, tokenize, ethics
Phase 1: Acquire  → corpus, bitext, transfer
Phase 2: Analyze  → morph, syntax, annotate, semantics, discourse, speech
Phase 3: Evaluate → eval
Phase 4: Release  → ethics (release gate)
         + orchestrator (coordinates all phases)
         Cross-cutting / optional → codeswitch, historical, lexicon

Skills by Phase

Phase 0 — Scope

Identify the target language precisely and set strategic direction before touching any data.

Skill	Purpose
`magic-linguistic-scope`	ISO 639-3 + Glottolog resolution, Joshi classification, URIEL typological profiling, transfer-source selection
`magic-linguistic-scripts`	Unicode normalization policy (NFC/NFKC), confusable folding, diacritic preservation for tone languages
`magic-linguistic-tokenize`	Fertility audit, SentencePiece config, vocab-extension method (FOCUS/OFA/HyperOfa)
`magic-linguistic-ethics`	CARE/FPIC, license audit, sacred-text gating, attribution registry — runs at Scope AND Release

Phase 1 — Acquire

Gather monolingual and parallel data ethically and reproducibly.

Skill	Purpose
`magic-linguistic-corpus`	Catalog (OLDI/CulturaX/MADLAD-400/Glot500), paragraph LID, MinHash dedup, contamination audit
`magic-linguistic-bitext`	LASER3/SONAR mining, Vecalign alignment, margin threshold tuning, synthetic bitext
`magic-linguistic-transfer`	LoRA rank by URIEL distance, MAD-X adapters, forgetting mitigation, tool selection

Phase 2 — Analyze

Run linguistic analysis layers needed for evaluation, augmentation, or training.

Skill	Purpose
`magic-linguistic-morph`	UniMorph, SIGMORPHON segmenters, FST/HFST, paradigm-completion augmentation
`magic-linguistic-syntax`	UD treebanks, cross-lingual parser transfer, agreement-probe construction
`magic-linguistic-annotate`	IAA metric selection (κ/α/γ), guideline authoring, adjudication, active learning
`magic-linguistic-semantics`	WordNet/OMW coverage, FrameNet/PropBank SRL, MWE/PARSEME, semantic-equivalence eval
`magic-linguistic-discourse`	RST/PDTB/GUM frameworks, coreference (zero-anaphora for pro-drop), coherence eval
`magic-linguistic-speech`	ELAN/Praat/FLEx → Lhotse CutSet, G2P/IPA, MMS/Whisper ASR, VITS TTS

Phase 3 — Evaluate

Honestly measure performance with metrics fit for the target language.

Skill	Purpose
`magic-linguistic-eval`	chrF++/COMET/GEMBA-MQM, BLiMP-style probes, contamination-aware reporting, per-dialect breakdowns

Orchestrator

Skill	Purpose
`magic-linguistic-orchestrator`	Entry point; phase routing; workspace state management

Phase 4 — Release

Ship the model responsibly. magic-linguistic-ethics returns here as the release gate — final license-compatibility check, attribution-registry completeness, community sign-off, and model-card authoring (release modes: Open / Community-gated / Restricted). See How It Works for details.

Cross-Cutting / Optional (Mindset Stubs)

Not a pipeline phase — these supplements activate when the specific scenario applies, at whatever stage it arises.

Skill	Purpose
`magic-linguistic-codeswitch`	Code-switching awareness for Hinglish/Spanglish/Singlish/MSA+dialect communities
`magic-linguistic-historical`	Cognate sets, Swadesh lists, sound correspondences for Class 0–1 bootstrap
`magic-linguistic-lexicon`	Dictionary-building, sense splitting/lumping, MWE inventories for RAG/MT post-edit

Shared Utilities

The _linguistic_shared/ directory contains interaction_utils.py and findings_presenter.py — shared utilities used across all skills. See Shared Utilities for details.

Was this page helpful?

On this page