English ↔ Spanish language learning

Second-language practice bands, explained

MindCraft has one progression for Spanish speakers practicing English and a different progression for English speakers practicing Spanish. The target-language sentence still has to pass its English-native or Spanish-native text envelope; the second-language layer adds communicative purpose, vocabulary, grammar, and reversible support. It does not translate an English curriculum into Spanish or the reverse.

Validation status: these are versioned product hypotheses. L2-specialist review, representative learner calibration, reliability evidence, and outcome studies are pending. MindCraft does not issue CEFR, ACTFL, DELE, school-grade, or other official ratings. [L2-BANDS-01]

Two directions, two owned progressions

Spanish → English

English text envelope controls text difficulty. A separate item bank and progression control L2 functions and structures. Spanish support is optional and never determines the target sentence's level.

1. Orientation words — one word or one short phrase
2. Familiar actions — one short complete sentence
3. Connected meaning — one longer simple sentence
4. Practical reading — two connected clauses
5. Explained choices — one controlled complex sentence
6. Extended meaning — one extended multi-clause sentence
7. Irregular spellings — one complex sentence with an irregular spelling
8. Irregular families — one multi-clause sentence with two irregular spellings
9. Dense meaning — one longer dense sentence
10. Extended discourse — one extended multi-clause sentence with irregular spellings

English → Spanish

Spanish text envelope controls text difficulty. A separate item bank and progression control L2 functions and structures. English support is optional and never determines the target sentence's level.

1. Orientation words — una palabra o frase breve
2. Familiar actions — una oración completa breve
3. Connected meaning — una oración simple más larga
4. Practical reading — dos cláusulas conectadas
5. Explained choices — una oración compleja controlada
6. Extended meaning — una oración extensa con varias cláusulas
7. Irregular spellings — una oración compleja con ortografía poco transparente
8. Irregular families — una oración de varias cláusulas con dos formas opacas
9. Dense meaning — una oración densa más larga
10. Extended discourse — una oración extensa con formas irregulares

What is measured—and what is not inferred

Target-text difficulty

The language-owned text engine checks length, vocabulary, orthography, morphology, and sentence structure.

Oral reading

Observed word-level reading is recorded separately. It is not an overall language-proficiency rating.

Pronunciation

A separate, uncertainty-aware signal. A speech transcript is not treated as ground truth.

Vocabulary and comprehension

Each needs its own task and evidence. Reading words aloud does not prove either construct.

ACTFL's official public overview likewise describes proficiency separately across listening, speaking, reading, and writing and organizes evidence by functions/tasks, accuracy, context/content, and text type. We cite that distinction; we do not reproduce its descriptors or infer an ACTFL rating from a MindCraft reading. [L2-CONSTRUCT-01]

Communication first, with target-language-specific content

The Council of Europe describes the CEFR as a framework for curricula, teaching materials, and assessment that must be adapted to a particular context rather than a ready-made solution. That informs our decision to progress from familiar identification and actions toward connected explanations. MindCraft's ten bands, thresholds, names, and Minecraft items are our own product design; we do not assign CEFR levels. [L2-CONTEXT-01]

Spanish is not a translation of the English inventory. Instituto Cervantes documents Spanish-specific grammar, pronunciation, orthography, functions, discourse, notions, and intercultural content. We use that as a reason to keep a Spanish-owned progression, not as copied curriculum or certification. [L2-SPANISH-01]

Scaffolding and progression rules

AI candidates must pass both the target language's native text envelope and a narrow, direction-and-band-specific gate for observable vocabulary, grammar, and communicative-purpose signals. Candidates that miss any gate are discarded and replaced from the separate frozen L2 bank.
Whole-sentence native-language support is available in the first four bands only when the learner has translations enabled. It is never the text-difficulty engine.
Early bands can use modeled help; middle bands use a brief cue; later bands are target-language-only. Removing support is band-based, explicit, and reversible through a manual difficulty adjustment—not an age assumption.
Soft credit and model disagreement keep gameplay moving but are neutral for automatic promotion.
Promotion requires high-confidence oral-reading, vocabulary, and comprehension evidence across at least two sessions, occurs only at a session boundary, and moves at most one band.
Evidence is tagged to one direction. English-native, Spanish-native, English→Spanish L2, and Spanish→English L2 evidence cannot be mixed by the progression contract.

These support-removal and promotion rules are MindCraft product rules. They are not CEFR, ACTFL, or DELE descriptors, and no external framework currently validates their sequence or counts. [L2-SCAFFOLD-01]

Claim and evidence registry

Version methodology-evidence-1.0.0. Each record limits the claim to the source's scope and labels product decisions separately.

L2-CONSTRUCT-01: Listening, speaking, reading, writing, vocabulary, pronunciation, and comprehension must not be collapsed into one MindCraft result.

official-framework

Supported scope: ACTFL's official overview separates language modes and describes multiple evidence dimensions.

Does not establish: MindCraft does not reproduce ACTFL descriptors, claim alignment, or issue an ACTFL rating.

Status: source-checked

ACTFL Proficiency Guidelines overview — Official overview: modes and proficiency criteria. Official framework overview for language proficiency across modes.

L2-CONTEXT-01: Language frameworks require context-specific interpretation; they are not ready-made MindCraft scoring systems.

official-framework

Supported scope: The Council of Europe describes CEFR uses for curriculum, materials, and assessment while emphasizing adaptation to context.

Does not establish: The CEFR does not validate MindCraft's ten bands, thresholds, Minecraft items, or progression decisions.

Status: source-checked

Council of Europe, Purposes of the CEFR — Official purposes and uses. Official framework guidance for language curricula, materials, and assessment.

L2-SPANISH-01: Spanish L2 content needs a Spanish-owned analysis rather than an English inventory translated at the end.

official-framework

Supported scope: Instituto Cervantes documents Spanish-specific treatment of grammar, pronunciation, orthography, functions, discourse, notions, and intercultural content.

Does not establish: MindCraft does not copy the inventories, claim PCIC alignment, or issue CEFR/DELE certification.

Status: source-checked

Instituto Cervantes, Plan Curricular introduction — Introduction and inventory structure. Official Spanish-as-a-foreign-language reference inventory.

L2-BANDS-01: MindCraft's ten L2 bands, names, target signals, and thresholds are product hypotheses, not CEFR/ACTFL/DELE mappings.

product-design

Supported scope: The rules are versioned, direction-owned, and regression-tested.

Does not establish: Specialist review, representative calibration, reliability, fairness, and outcome evidence remain pending.

Status: product-calibration-pending

Product decision; not attributed to an external source.

L2-SCAFFOLD-01: Whole-sentence support bands, cue removal, and multi-session promotion counts are explicit MindCraft product rules.

product-design

Supported scope: The implementation keeps target-language evidence and optional native-language support separate and reversible.

Does not establish: No cited framework validates the exact scaffold sequence or promotion counts; L2 specialist review and learner calibration are pending.

Status: product-calibration-pending

Product decision; not attributed to an external source.

Framework attribution and licensing boundary

MindCraft links to official sources and paraphrases only high-level organizing concepts. Its band labels, item banks, thresholds, and decisions are original product content.
The Council of Europe says a citation alone needs no permission, while commercial reuse of protected material can require prior authorization (copyright and permissions). No CEFR tables, scales, logos, or descriptor text are reproduced here.
ACTFL states that commercial use of its Guidelines is not authorized without permission and that official ratings come only from official tests (ACTFL use and official-rating policy). MindCraft does not claim ACTFL alignment or ratings.
Instituto Cervantes reserves intellectual-property rights in its content (Instituto Cervantes legal notice). MindCraft links to the Plan Curricular but does not reproduce its inventories. Any future descriptor-level mapping requires a separate legal and specialist review.

Known limitations

The frozen item banks are regression and fallback content, not standardized assessment items.
L2-specialist review and representative calibration for children and adults are pending in both directions.
Progression is stored independently for all four English/Spanish pathways. A session freezes its band and hard envelope for at most ten completed readings; inactivity also creates a boundary. Current single-transcript oral-reading results cannot automatically promote an L2 band, and L2 promotion remains unavailable until qualifying oral-reading, vocabulary, and comprehension evidence exists across sessions.
Automatic checks cannot establish naturalness, dialect fit, cultural fit, comprehension, or learning outcomes.
The semantic gate is intentionally conservative and lexical: passing it proves only that required observable signals are present, not that a specialist has judged the sentence instructionally ideal.
False-friend and cognate notes are teaching safeguards, not a complete bilingual lexicon.

Version and change log

English L2 engine: english-l2-1.1.0
Spanish L2 engine: spanish-l2-1.1.0
Published: August 14, 2026
Review status: MindCraft product review complete; English/Spanish L2 specialist review and representative calibration pending
Changes: Added four higher bands so advanced readers have more room. Bands 7-10 require irregular spellings such as through and enough. Direction-owned item banks, language-owned text envelopes, translation scaffolds, and conservative multi-session progression remain.