Small synthetic language for testing seq2seq model

espancanese.lua

-- > te grillo wa silencieusement matte iru
-- = your cricket is sleeping silently

genders = {
    m={
        the='el',
        a='un',
        my='me',
        your='te',
    },
    f={
        the='la',
        a='une',
        my='ma',
        your='ta',
    },
}

nouns = {
    {'dog', 'perro', 'm'},
    {'cat', 'gato', 'm'},
    {'bird', 'pajaro', 'm'},
    {'goose', 'oca', 'f'},
    {'giraffe', 'jirafa', 'f'},
    {'jellyfish', 'medusa', 'f'},
    {'hedgehog', 'erizo', 'm'},
    {'cricket', 'grillo', 'm'},
    {'computer', 'computadora', 'f'},
    {'door', 'puerta', 'f'},
    {'window', 'ventana', 'f'},
    {'shoe', 'zapato', 'm'},
    {'knife', 'cuchillo', 'm'},
    {'book', 'libro', 'm'},
    {'garden', 'jardin', 'm'},
    {'boat', 'barco', 'm'},
    {'desk', 'escritorio', 'm'},
    {'box', 'caja', 'f'},
    {'steak', 'bistec', 'm'},
    {'salad', 'ensalada', 'f'},
}

adjs = {
    {'blue', 'bleu'},
    {'green', 'verde'},
    {'red', 'rouge'},
    {'black', 'noir'},
    {'white', 'blanc'},
    {'small', 'petit'},
    {'large', 'grand'},
    {'soft', 'doux'},
    {'healthy', 'sain'},
    {'dirty', 'sal'},
    {'lazy', 'paresseux'},
    {'nervous', 'nerveux'},
    {'sad', 'triste'},
    {'cold', 'froid'},
    {'evil', 'mechant'},
    {'new', 'noveau'},
    {'heavy', 'lourd'},
    {'slow', 'lent'},
    {'silent', 'silencieux'},
}

advs = {
    {'clearly', 'clairement'},
    {'slowly', 'lentement'},
    {'silently', 'silencieusement'},
    {'politely', 'poliment'},
    {'quickly', 'vite'},
    {'well', 'bien'},
    {'badly', 'mal'},
    {'elegantly', 'elegamment'},
    {'a lot', 'beaucoup'},
    {'enough', 'assez'},
}

vintr_presents = {
    {'standing', 'tatte'},
    {'drinking', 'nonde'},
    {'running', 'hashitte'},
    {'jogging', 'jogingu shite'},
    {'talking', 'hanashite'},
    {'reading', 'yonde'},
    {'swimming', 'oyoide'},
    {'waiting', 'matte'},
    {'sleeping', 'matte'},
}

vintr_pasts = {
    {'sang', 'utatta'},
    {'drank', 'nonda'},
    {'flew', 'tonda'},
    {'escaped', 'nigeta'},
    {'bathed', 'abita'},
    {'laughed', 'waratta'},
    {'swam', 'oyoida'},
    {'called', 'yonda'},
    {'waited', 'matta'},
    {'ate', 'tabeta'},
    {'slept', 'neta'},
}

vintr_futures = {
    {'eat', 'taberu'},
    {'drink', 'nomu'},
    {'fly', 'tobu'},
    {'run', 'hashiru'},
    {'talk', 'hanasu'},
    {'bathe', 'abiru'},
    {'laugh', 'warau'},
    {'ponder', 'omoiiru'},
    {'swim', 'oyogu'},
    {'call', 'yobu'},
    {'sleep', 'neru'},
}

vtr_presents = {
    {'kicking', 'kette'},
    {'eating', 'tabete'},
    {'flying', 'tonde'},
    {'drawing', 'kaite'},
    {'grilling', 'yaite'},
    {'reading', 'yonde'},
    {'seeing', 'mite'},
    {'protecting', 'mamotte'},
}

vtr_pasts = {
    {'kicked', 'ketta'},
    {'ate', 'tabeta'},
    {'grilled', 'yaita'},
    {'called', 'yaita'},
    {'saw', 'mita'},
    {'bought', 'katta'},
    {'crushed', 'kudaita'},
    {'protected', 'mamotta'},
    {'drank', 'nonda'},
}

vtr_futures = {
    {'kick', 'keru'},
    {'cut', 'kiru'},
    {'eat', 'taberu'},
    {'grill', 'yaku'},
    {'avoid', 'sakeru'},
    {'drink', 'nomu'},
    {'see', 'miru'},
    {'comfort', 'nagusameru'},
    {'hug', 'daku'},
    {'inhale', 'suu'},
    {'mock', 'yajiru'},
    {'protect', 'mamoru'},
}

sentences = {
    {'i am $vintr_present', 'watashi wa $vintr_present iru'},
    {'i am definitely $vintr_present', '! watashi wa $vintr_present iru !'},
    {'i am $vintr_present $adv', 'watashi wa $adv $vintr_present iru'},
    {'i am definitely $vintr_present $adv', '! watashi wa $adv $vintr_present iru !'},
    {'am i $vintr_present ?', '? watashi wa $vintr_present iru ka ?'},

    {'i am $vtr_present $noun', '$noun wo $vtr_present iru'},
    {'i $vtr_past $noun', '$noun wo $vtr_past'},
    {'i $vtr_past $noun and $noun2', '$noun et $noun2 wo $vtr_past'},
    {'i will $vtr_future $noun', '$noun wo $vtr_future'},
    {'i will $vtr_future $noun and $noun2', '$noun et $noun2 wo $vtr_future'},
    {'i will definitely $vtr_future $noun', '! $noun wo $vtr_future !'},
    {'i will not $vtr_future $noun', '$noun wo $vtr_future sen'},
    {'i definitely will not $vtr_future $noun', '! $noun wo $vtr_future sen !'},

    {'$noun is $vintr_present', '$noun wa $vintr_present iru'},
    {'$noun is $vintr_present $adv', '$noun wa $adv $vintr_present iru'},
    {'$noun will $vintr_future', '$noun wa $vintr_future'},
    {'$noun $vintr_past', '$noun wa $vintr_past'},
    {'$noun $vtr_past $noun2', '$noun wa $noun2 o $vtr_past'},
    {'$noun $vtr_past $noun2 $adv', '$noun wa $noun2 o $adv $vtr_past'},

    {'this is $noun', 'kore wa $noun desu'},
    {'this is definitely $noun', '! kore wa $noun desu !'},

    {'where is $noun ?', '? $noun wa doko desu ka ?'},
    {'is this $noun ?', '? kore wa $noun desu ka ?'},
    {'who $vtr_past $noun ?', '? dare ga $noun o $vtr_past ka ?'},
    {'who $vtr_past $noun and $noun2 ?', '? dare ga $noun et $noun2 o $vtr_past ka ?'},
    {'who is $vtr_present $noun ?', '? dare ga $noun o $vtr_present iru nodesu ka ?'},
}