Re: Universal Translation Language

On Fri, 28 May 1999 01:44:25 -0700, Charles <catty@...>
skribis:

>obviously a self-stultifyingly invalid argument. Anyway,
>as an admirer of Ido and some of its intentions,
>I've tinkered quite a bit with it and can show a few
>of my own experiments, perhaps best to be avoided ...
Anyway, I'll read them with interest.

>> I'm developing a logical language (that is, an unambiguous,
>> full-expressive one) which relies on western world vocabulary and
>> syntactics. To make it, I'm taking advantage of Esperanto/Ido
>> structure and lexicon.
>
>The Ido vocabulary is really a good starting point;
I think Eo's vocabulary would be even better, as (I think) it's more
developed. However, as UTL is going to use Ido's alfabet, I think Ido
forms will be the standard to follow.

>using under 4000 roots it still has the feel of a fully
>developed and usable language. The syntactics are still
>(I think) a bit rough, being too inflexible and irregular.
>The central issue is ambiguous-direct versus precise-indirect
>derivation. I'd keep the "ambiguous" -o noun ending but allow
Meaning? I think that's dangerous cause people will feel inclined to
use the short, underived -o form, which for it's ambiguity will be
unsuitable for MT purposes. =20

>also the more specific -ilo, etc. and go for consistency.
>I like the proposed -ao / -oa type of endings and I've
>been trying something similar.
>
>> Part-of-speech endings are -o (noun) -e (plural noun) -a (adj) and -am
>> (adv).
>
>Can all adverbs be derived from adjectives? Hmm.
No, that's not what I meant. -am is for derived adverbs. Other adverbs
will have a series of endings (Eo's -aw, -el, -am, etc) different of
-am. Btw, it's still to decide whether -am is the best selection. I
think it works well, but if I take -n for accusative and it's applied
not just on nouns but on adjectives as well, we would have an ending
-an, too similar to adverbial -am. In that case, -im could be a good
alternative.

>Mandatory number on nouns may be a bad idea, and something
>that Anglo-Euro speakers could easily learn to do without;
>"five sheep" sounds more logical than "no cows", IMO.
(just one comment before responding: "no cows" doesn't seem very
logical, but it wouldn't be said that way on our AL; rather, it would
be said something like "nula bovino").
Perhaps you're right on non-mandatory number in an AL, but I prefer
not taking too brave steps like that. I normally prefer using tools
that have been already tested (in Eo and Ido, for example). Btw, it's
to keep in mind that one of the objectives of this language will be
MT, so that almost every distinction made in NL should be made in the
AL too, unless we want the machine to make hard guesses.
Non-numberality may be a source of ambiguity. Btw, I think some way
could be devised to express non-numberality in certain specific cases.

>It is tempting to use -s for a perhaps-optional plural.
>Also, it is not really necessary to tense every verb.
I say the same for this.

>> Verbs work as in Esperanto: -i -as -is -os -us -u
>>
>> -ant- -int- -ont-, -at- -it- -ot- forms are not used. Active
>> participle is derived directly from verb (e.g. reganta homo is said
>> rega homo). Passive verb is made by the suffix -at-.
>
>Ido has a passive but I'd rather make that a true inverse voice.
>Currently, "me havas lu" is legal but "lu havesas me" is not.
In Ido and Eo this inversion is made by subject-object order exchange
(obviously, with marked object). I think that's right, cause that way
they can have both inversive and passive constructions.

>So I suggest adding -at, -it, -ot as verb endings. Also, keep
>-r infinitive; -z imperative, "-u" reserved for something better.
Well, that's Ido's scheme. I however am not fully Ido biased, as you
see. There are certain features I prefer on Eo over Ido, but they are
less than the other way around. Anyway, even if I didn't, there are
certain strategic reasons to base UTL on Eo&Ido and not just on Ido.

>Where did that -s for active voice come from, anyway? Prefer -n.
There have been some Eo reform projects which used -n for that (mainly
to free -s and use it for plurals). They have had not much success
though. I personally prefer -s, though there's no ethymologic reason
for it, I think. Anyway:

>Needless to say (?) there should be no accusative ending at all.
I think accusative ending (as used in Ido at least) will be
indispensable if we want some flexibility on sentence structure
without loosing unambiguity.

>> I recommend you to apply real words to this system to better
>> understand how it works; e.g. take "bona" (good) or "broso" (brush)
>> and take from the table the meanings of their derivates.
>
>As stated above, these are very interesting ideas.
>
>> I think this system gets at once rid of Eo system's impreciseness and
>> inconsistency, and of Ido system's rigidity.
>
>I'd like to eliminate the inconsistencies remaining in Ido,
>such as adverbs with non-conforming endings, and pronouns
Agreed.

>(personal, deictics, etc.) which are not well-designed.
Yes, though I would like to keep some mnemonic referents in the final
desing of it.

>How about something more like Rick Morneau's comprehensive set?
>
>        1:      s-                        Pers:   -p  Gen:    -x
>        2:      p-    Singular:     -a-
>        3:      g-                        Dem:    -d
>        1+2:    v-    Plural:       -i-
>        1+3:    j-                        Loc:    -z
>        2+3:    l-    Unspecified:  -u-
>        1+2+3:  m-                        Tem:    -k
The set is interesting, but "you" is said "pap" in this scheme? %-/
Btw, "Gen" stands for "genitive"? Is it the same for the rest of the
vocab?=20
Other thing is that there's something lacking in the table:
person/thing distiction.
And masc/fem distinction too (at least on 3rd person).

>Next, I wish to inflict upon you my conlang's (custom) BNF syntax.
>It is only about 2 hours old, so ... The notation below uses
>(0 or 1), [1 or more], and {0 or more}, for repetitions.
>
>>> VOWEL ::=3D a | e | i | o | u
>>> CONS  ::=3D b | c | d | f | g | x | j | k | l | m | n | p | q | r | s=
 | t | v | z
>>> CV    ::=3D CONS - VOWEL -
>>> MORPH ::=3D [CV] - CONS -
>
>(Makes "closed" roots without consonant clusters or diphthongs,
>suitable for unambiguous compounding, unlike Ido's.)
Yes, though that would be for an a-priori vocab, I guess...

>>> VOICE  ::=3D i | u
>>> VERB   ::=3D [MORPH] -  VOICE
>>> ADJECT ::=3D [MORPH] - (VOICE) - a
>>> ADVERB ::=3D [MORPH] - (VOICE) - e
>>> NOUN   ::=3D [MORPH] - (VOICE) - o
>
>(The "i" is for active voice, "u" for inverse.
>It can be infixed to each part-of-speech.
Interesting. I guess those -i- | -u- are like my -i- | -ati- (except
that -ati- is for passive and not inverse, of course).

>A diphthong at the end of a word is easy to recognize.
>I also like the French way of stressing the final syllable,
>otherwise English speakers mangle the final vowels.)
Don't know what you meant with this commentary... You would put stress
on last syllable of your conlang words?

>>> OBLIQUE    ::=3D ADVERB (NOUNPHRASE)
>>> PARTICIP   ::=3D ADJECT (NOUNPHRASE)
>>> PREDICATE  ::=3D VERB   (NOUNPHRASE)
>>> NOUNPHRASE ::=3D {ADJECT} NOUN {PARTICIP}
>>> SENTENCE   ::=3D (NOUNPHRASE) PREDICATE {OBLIQUE}
>
>So the result is something like this, generated randomly:
>"Zopa xesizedqoloca zezapuo padteza digpuzabospupu fasuo."
>Gorgeous, eh?
Gorgeous, gorgeous :) Would that be adj adj vn adj v vn?


Saludos,
Marcos