AI Language Learning Experiment

This project generates a seeded synthetic language from the 1000 most common English words and creates parallel training samples.

Included Files

• data/english_1000.txt: 1000-word English base vocabulary.
• language_experiment.py: generator + bidirectional translator CLI.
• output/lexicon.csv: English to synthetic-language lexicon.
• output/sentence_pairs.csv: 100 parallel sentence pairs.
• output/manifest.json: seed + generation metadata.

Generate Language + 100 Sentence Pairs

python language_experiment.py generate --seed ai-language-seed-42 --sentence-count 100 --output-dir output

This produces:

• output/lexicon.csv
• output/sentence_pairs.csv
• output/manifest.json

Translate Text

English to synthetic:

python language_experiment.py translate --seed ai-language-seed-42 --direction en2new --text "We will read the story tonight."

Synthetic to English:

python language_experiment.py translate --seed ai-language-seed-42 --direction new2en --text "Griert broismiob tozgriask quacjoul hotqoo goovual."

You can also translate files:

python language_experiment.py translate --seed ai-language-seed-42 --direction en2new --input-file input.txt --output-file output.txt