A carmen (garden house) in Granada

Corpus

Corpus Granada

Spanish from a semester in Granada: speech samples set in the city, graded by level and annotated following the Instituto Cervantes curriculum (PCIC).

What is it?

A corpus is a collection of real or realistic texts, organised and annotated so it can be queried: which words appear, in which contexts, at which level. The large Spanish learner corpora (CAES, CEDEL2) serve research; this one is smaller and has a different purpose: to accompany whoever spends a semester in Granada.

Each sample is a situation the student will actually live — ordering at a bar, asking at the pharmacy, flat-hunting, talking to teachers — written in the Spanish you hear here and graded by CEFR level. Annotations mark communicative functions, grammar, vocabulary, pragmatics, culture and pronunciation features.

Alongside the reference layer grows a learner corpus: pieces written by Clinic students, pseudonymised and only with their explicit consent. On this page, only aggregate statistics of it are shown.

How it is made

  1. 1

    PCIC first

    Before a sample is written, the Instituto Cervantes curriculum inventories are consulted: functions, grammar, notions and sociocultural knowledge for the level.

  2. 2

    Ten agents, one veto

    La Banda, a system of ten agents with fixed roles, documents the situation, drafts, annotates and reviews. One of them can only veto: if a fact about Granada is undocumented or the level is off, the piece does not pass.

  3. 3

    Closed tagset

    Annotation uses a fixed, versioned set of tags only. An invented code invalidates the annotation. That keeps queries comparable across pieces and levels.

  4. 4

    Declared provenance

    Every piece states whether it is authentic (with source and licence), adapted, or generated for the corpus. Granada is never made up: prices, opening hours and places are documented or flagged as unverified.

  5. 5

    Human review

    No sample is published until a teacher validates and publishes it from the Clinic panel. The machine proposes; the person decides.

Published samples

Clear

The corpus in numbers

Published samples by level

  • A10
  • A24
  • B11
  • B21
  • C12
  • C22

Annotated learner texts by level

Not enough data yet.

Most frequent phenomena by level

Not enough data yet.

Learner figures are aggregates: no student texts or data are ever shown.

Licence and how to cite

Published samples are released under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 (CC BY-NC-SA 4.0): you may use them in class and in research, citing the source, for non-commercial purposes and sharing under the same licence. Authentic pieces keep the licence of their source. CC BY-NC-SA 4.0.

How to cite: Clínica Cultural y Lingüística (Granada). Corpus Granada: Spanish from a semester in Granada. https://www.clinicacultural.com/corpus

If you are spending a semester in Granada, the Clinic accompanies you with this corpus and everything else.