On 5/26/10 12:17 AM, David Peterson wrote:
> On May 25, 2010, at 8◊57 PM, Jesse Bangs wrote (quoting Jim Henry):
>
>>> That's a potential problem; we've discussed it before on the list (see
>>> links below).   Avoiding (unintentional) coining of new words with the
>>> same form as existing words should be pretty simple if you have your
>>> conlang's dictionary in any halfway decent database, or even
>>> spreadsheet software.
>>
>> Honestly, I don't consider this a problem. Every language synonyms and
>> homophones; creating them by accident is a fine way to add this
>> realism.
>
> Indeed. And even if one considered it an actual problem, it's a
> relatively simple matter to fix. If I don't do a find/search when
> coining a new word and I happen to coin a new form for an
> already-created word, I usually figure it out when I enter the
> form into the English-to-Conlang side of the dictionary. Then
> it's just a few minutes' work to decide if I want to give the new
> form a completely different definition, a synonymous definition
> (perhaps with a slightly different shade of meaning), or to simply
> abandon the new form for the time being.

This really depends on the type of language being developed.  With an 
auxlang realism isn't a goal but utility is, and homophones are 
something to be avoided if possible.  In SASXSEK, I not only work to 
avoid homophones but also near-homophones.



>
> As for being influenced by other languages (a valid concern) in
> coining either words or coming up with phonetic forms, I don't
> think a random word generator is the solution. It can be a useful
> tool (and there are a number out there), but it can't replicate
> sound symbolism. Sound symbolism is something that exists in
> every language, but which differs from language to language.
> If one is creating a naturalistic language, one will likely want to
> include sound symbolism, without reproducing the sound symbolism
> of one's native language unconsciously. Sound symbolism seems
> like something a program will never be able to reproduce, as it
> entails coming up with a phonological pattern (not difficult), and
> then applying it semi-systematically to a series of semantically-
> related words (very difficult).
>
> Also, I see problems here:
>
> On May 25, 2010, at 4◊16 PM, Dave Babb wrote:
>
>> b)    A 2000 word list that the community accepts as canonical. (at least without everyone flaming each other in unending debate over minutia)
>
>
> Specifically, this will introduce a problem that most conlangers
> are good about avoiding: Reproducing the vocabulary of an
> already-existing language. Essentially, this program would be
> generating a series of relexifications of the same language, and
> a simple difference like that between Enlgish's "hand" and "arm"
> and Russian's "ruka" would be impossible to reproduce. This
> might be ideal for a language one intends to use live (in which
> case, one would probably want a word for every word one
> wants to use in a conversation [I'd probably come up with a
> shorter word for "cell phone"]), but wouldn't be ideal for a
> language that's attempting to look like a natural language. I
> guess it depends what kind of language you're trying to create...
>
> -David
> *******************************************************************
> "A male love inevivi i'ala'i oku i ue pokulu'ume o heki a."
> "No eternal reward will forgive us now for wasting the dawn."
>
> -Jim Morrison
>
> http://dedalvs.com/
>
> LCS Member Since 2007
> http://conlang.org/