First of all, I'd like to say welcome to this forum, Bob.
I saw your "200k-word dictionary" link with rapt anticipation. After all, I have a list online of conlangs with over 10,000 words (https://www.frathwiki.com/Conlangs_with ... ,000_words
), and the largest I've encountered is Mark P. Line's Classical Yiklamu, with 91,591. My own Kankonian is in fifth place, having over 66,000 words in its lexicon.
When I checked out your word lists, though, I was disappointed. What you have is basically a cipher of English (a computer-generated cipher, as a matter of fact). For example, you have "figure , gureg20" and "figure , tazit52". (You don't even specify which one is the noun and which one is the verb.) For a word like "figure", my Kankonian dictionary lists nine translations. Meila
is build, like "She has a nice figure". Inehwil
is to be a factor, as in "That all figures in this equation". Praisi
is statistics, as in "The 98% figure on the Trump question shocked me". Barukh
is a numeral, as in "a figure 8". Nalosh
is also used for the meaning of 'to assume', while kinir
means to use reasoning, or 'to reckon', and is used in sentences like "I figure that if I go to bed at 7:30 p.m, I'll be able to get up at 4:30 a.m. and be at the station by 6 to catch that train . . ." Instar
is a scientific illustration/diagram, as in "See fig. 4". In fact, the word isnaloshas
(lit. "I figure" or "I assume") is placed at the end of Kankonian sentences to form tag questions. Dren
is used for things like an authority figure or a father figure. You'd use *omos
for the "hitokage" sense of "figure", as in, "I saw a shadowy figure". Then there's azraum
for "to figure out", and eabospa
for "to figure big", and stoiza
for "to figure prominently". A "figure of speech" is doinktrak
What makes it even more of a cipher is all the suppletive forms for the plural nouns and different verb forms. "Dog", for instance, is either "zemubin29" or "dahol72", while "dogs" is "kekil55". In any human language, these would be visibly related. A little suppletion is OK (English has "be, is, am, are, was, were" and "go, went"), but you have things like "ber8" for "smile", "yeypag52" for "smiled", "mezih64" for "smiling", and "tanon16" for "smiles".
Then you have computer-generated equivalents for proper names. How is "delen41" supposed to have evolved from Stephen, or "kropat32" from Michael, or "zewuk18" from Kate? You even have surname puzzlers like "holos22" for Johnson.
Instead of just computer-generating a suppletion-filled relex for 200,000 English wordforms, why not use the Landau Core Vocabulary? It's a word/concept list I've been developing since 1997. Different meanings and the distinctions in semantic concepts that real natlangs make are covered, and I have the most important words in English (a naturalistic conlang based on the LCV, with no other words, would have about 6,500 entries in its dictionary, but even if you computer-generate a relex of the LCV, you'll have more than 10,000 words; you'll also have an extreme splitlang, but that's better than a near-cipher of English).
The LCV also has an appendix of 250 girls' names and 250 boys' names with their major forms across Earth's languages (Anne, Ana, Anna, Anita, Annika, Oneika, Anja, Anya, Annette, etc.) You can hand-make Atlantean names that are related to each of these (maybe Anne can be Onah, and Greg can be Kreekhor, or whatever) and you'll have names that would plausibly be translations of these. The LCV also has 500 surnames (the 10 most common names from each of 50 ethno-linguistics groups, with no duplicates across the 50 categories), which you'll almost certainly just transliterate into Atlantean (So Davis would be something like Deybis instead of "lapor9".)
If you PM me your email address, I can send you the LCV, as I have it polished up right now (it's still
not ready to go up online). You can work on translating it (you might consider doing it by hand instead of computer-generating the words; that way you can decide which concepts will share the same word and which won't).