# Frequent words dataset — sources and licence

`frequent-words.json` is generated by `scripts/build-vocabulary.mjs` and committed. It
is a derivative of two Share-Alike sources and is therefore offered under the same
terms: **CC BY-SA 3.0**.

## Ranking and word forms

The lemma ranking, the part of speech, and the word forms grouped under each lemma come
from [doozan/spanish_data](https://github.com/doozan/spanish_data) (`frequency.csv`),
which credits [hermitdave/FrequencyWords](https://github.com/hermitdave/FrequencyWords).
That corpus is built from film and television subtitles, which is why it leans towards
spoken register. **CC BY-SA 3.0.**

The ranking is a practical study order for Spanish generally. It is not a measurement of
Argentine speech, and the page says so.

## Meanings, region and register labels

English glosses and the region/register qualifiers attached to each sense come from
[Wiktionary](https://en.wiktionary.org), via the same repository (`es-en.data`).
**CC BY-SA.**

## Authored supplement

Dropping upstream rows the tagger could not resolve removes English-subtitle noise
(`john`, `the`, `um`) but takes a little real Spanish with it. Eight entries — the
preposition `a` and the courtesy abbreviations — carry hand-written glosses instead, and
are marked `"authored": true` in the data.

## Derived in this repository

Two fields are computed here, not taken from upstream:

- `vos` marks a form as distinctly Rioplatense. Candidates are derived from the
  infinitive and are kept only when the form actually occurs in the corpus. Forms shared
  with `tú` are not marked, because they carry no regional information.
- `rioplatense` marks an entry whose Wiktionary sense is labelled Rioplatense,
  Argentina, Uruguay or Cono Sur.

## Not included

The RAE's [CORPES country lemma lists](https://www.rae.es/corpes/contenidos/lemas-pais)
are the Argentine-specific supplement, but they are built from press texts rather than
conversation and are not offered for redistribution. The page links to them instead.
