# Talk Board credits

## Word list (app/lexicon.js) — CC BY-SA 4.0

The prediction word list is an adaptation of **wordfreq** by Robyn Speer
(https://github.com/rspeer/wordfreq, doi:10.5281/zenodo.7199437). wordfreq's data is licensed
**Creative Commons Attribution-ShareAlike 4.0** (https://creativecommons.org/licenses/by-sa/4.0/);
`app/lexicon.js` is shared under the same license. Changes: the 30,000 most common English words made of letters
(and an apostrophe), words not suitable for children removed, frequencies rounded to two decimals
(tools/build-lexicon.py).

wordfreq is built from, and credits:
- **SUBTLEX** word lists (SUBTLEX-US: Brysbaert, M. & New, B. (2009), Behavior Research Methods 41(4), 977-990; and
  SUBTLEX-UK/-CH/-DE/-NL). SUBTLEX is freely available data; it is used here only as it is contained in wordfreq,
  whose author has the SUBTLEX authors' permission for any use with credit.
- **OpenSubtitles 2018** via OPUS (Lison & Tiedemann 2016), with attribution to OpenSubtitles (https://www.opensubtitles.org).
- **Google Books Ngrams** (http://books.google.com/ngrams) and Google Books Syntactic Ngrams.
- Wikipedia; the Leeds Internet Corpus; ParaCrawl; OSCAR; NewsCrawl 2014 and GlobalVoices; Twitter and Reddit word counts.

The raw SUBTLEX-US file that sits in the words-with-sam repository (wordlist-1000/src) is **not** shipped in this app:
its own terms ("free for research purposes") do not clearly cover a family app.

## Phrases, greetings, next-word table

Written for this app (app/phrases.js, app/words.js). Word pairs and triples for "Next word" are counted from these
phrases on the device; nothing is downloaded.

## Voice

Josh's own voice, made with ElevenLabs (voice of the app's owner, used with his permission).