Re: basic morphemes of a loglang

On Fri, 28 Nov 2003, Ray Brown skribis :

>> Almost everybody seems to agree that about 1000 basic morphemes are
>> needed (and enough).
>
>Do they?  I'm been pondering this for many years now. One gets quite a few
>different guestimates IME.  1000 might, I guess, be OK to get the language going -
>so, yes, usable in that sense. But whether you'd eventually find that too limiting
>is another matter. My guess is that if the language is a loglan you will need a
>larger vocablary.
(snip)
>> The examples of Basic English (ca. 800 words),
>
>A bit of a cheat IMHO.  To keep the word count down to 800, Basic English
>has to resort to the _idiomatic_ use of so-called phrasal verbs and other
>circumlocutions. I do not think idioms help paricularly; one might be better having separate
>morphemes.
>
>> Esperanto and Volapk (no idea of word count, but probably slightly over
>> 1000) confirm this.
>But Esperanto has been increasing its vocabulary over the past century - I
>don't know what the current morpheme count is.
>
>> The 881 "essential characters" of Japanese also go that way, while not
>> being European-biased.
>
>Reginald Dutton endowed his Speedwords with only 511 morphemes - 491
>'radicals' and 20 'particles'. He claimed:
"It is readily deducible that the total obtainable by combining every
>radical and its derivates with every other would be 491 x 20 x 490 x 20; that is
>millions of words, and all without additional memory effort once the few
>fundamental have been learnt."
>
>In practice this leads to so many idiomatic compounds that additional
>memory is most certainly needed.  Better a new morpheme than an arbitrary
>idiomatic compounding.
>
>Jeffrey Henning's Dublex claims to need only some 400 (IIRC) basic
>morphemes - but from what I've seen of it, it has to resort to some kludgey and
>potentially ambiguous compounding.Indeed, I cannot see how this can be avoided if
>>vocabulary is so restricted.
My approach is to use a limited set of "elementary morphemes", in the sense of
"elementary particles". Then the "real" morphemes are built by composition and
derivation, and the initial morphemes are then buried into the incounscious
reptilian memory of the proto-conlang. Who cares that the word "origin" includes
the proto-IE root "or" (to arise, to spring up)? That's the kind of thing I'm
trying to re-create.

For this purpose, 1000+ roots may be enough. But surely imply a lot of
idioms, which IMHO is not a problem while you're designing a conlang rather than
an aux- or lo-lang. And additional memory is a bargain nowadays :-)

From this point of view, Basic English is not a cheat. But it is one to claim it is
usable in real life (IMHO). Consciously building idioms in real-time is not likely,
and learning arbitrary ones amounts to learning new words - plus you get the extra
confusion for free.

>>What about, "neither (neither my sister nor herself) nor
(neither me nor myself)", meaning "my sister and I"? :)))
>
>If I recall my Boolean algebra classes of many years ago, I think you can
>also get by with just NAND   :)
Sure you can. But I can't see any English conjunction for NAND.

>> It is not obvious to me that a tree is "something that grows" (kreskajho
>> in Esperanto, IIRC),
>
>?? Esperanto for 'tree' is _arbo_
>
>> although it is so in many conlangs;
>
>Such as?  All the conlangs I've met actually have a word for tree.  Even
>Dutton included one among his 491 root words: _bo_
>
>> why not focus on "bearing fruits", probably more useful in real life?
>
>Other plants besides trees bear fruit.  Why not have a morpheme that means
>"tree"? There seems to be some misunderstanding here.
There is indeed. I was just meaning to exemplify, but having no precise reference
at hand I typed what I only vaguely recorded. Having checked an Esperanto lexicon,
"tree" is indeed "arbo", and "kreskajho" is any vegetal (BTW, I discovered that
"farto" is "health", which must be obvious to English-native Esperantists :)

But I agree that something as a tree deserves a root by itself (no pun). And IMO
vegetals deserve a whole noun class. Bearing fruit is another concept, that could
also be used for animals bearing youngs.

>> (1) How to get enough roots, while keeping a simple syllabic structure?
(snip)
>
>This is something I've spent about 40 years or more trying to resolve  :-)
>
>My BrScA has the written structure just CVC (none of this CC business!) -
>but I did allow it 7 vowels (the 5 'canonical' vowels + two diphthongs) - three
>seems very restrictive. I've concluded that I cannot get enough roots; that's
>why I've been developing BrScB recently.
>
>> (2) Diacritics are not welcome on mail software, as I could see browsing
> the list.
>
>Very true :)
>
>> I tried restrincting the initial first 8 (yes, plus three special letters,
>> if you care to know), but the use of di- and tri-graphs eventually seems
>> unavoidable. Thus _hs_ for [C], _sh_ for [S]; and hence _hhs_ for [C:]
>> and _ssh_ for [S:]. And what about _h_ [X] + _s_ [s]??? Urgent help neede!
>> !!
>
>Ach - avoid digraphs and eschew trigraphs!  If you have only 11 consonants,
> there should be a few spare ones.  How are you using letters like |c| and |x|,
>for example?
>
>Maybe you could mail your orthography to the list.  I'm sure some here
>will be only too ready
>to give suggestions  :)
>
I have then to state my problem more precisely. Three vowels and eleven consonants
provide 4389 syllables of the forms VC, CVC, CVCC or CCVC. I really _hate_ self-
seggregation (it disables puns) but, of course, not all CC pairs are allowed. The
only ones permitted are those I can pronounce :) The grand total varies with the
hour of the day and the number of beers I've been drinking.

My three vowels are /a/, /e/, /o/ (slashes denote conventional symbols, and what is
included is _not_ some IPA equivalent; for these I use [brackets]). But of course,
each one comes in a series of flavors, e.g. for /o/: closed [o], open [O], front
[u], "mute" [w], nasal [O~], long [O:], diphtong [ow]. See? 24 actual vowels in
all. And I abandoned tones :)

Similar for consonants. Stops are /k/, /t/, /p/, and e.g. /t/ has variants /t/ [t],
/d/ [d], /dh/ [D] and /tt/ [tt]. Sibilants are /h/, /hs/, /sh/ and /s/ with short,
long and "emphatic" (pharyngealized) variants; /h/ is [x] or [X], /hs/ or /c with
cedilla/ is [C], /sh/ or /s with caron/ is [S], /s/ is fortunately [s]. "Vocalics"
are /r/, /l/, /n/, /m/ with variants according to my mood of the moment.

So I need many diacritics, mostly for vowels. For consonants, I can still use the
"x" for the glottal stop (the apostrophe is busy elsewhere), and "nq" for the eng I
was using at a time - "q" is not used otherwise, but distinguishing plain "n" from
"eng" is important.

Variants _are_ phonemic. They help distinguish noun cases, verb tenses, and the
like. E.g. nominative /lok/, ergative /log/, locative /loghra/, and so on.
To get things even more simple, flexion and derivation use all kinds of affixes and
infixes: /lurgan/ is a flexion of the above with "-r-an" postinfix and vowel
mutation /o/ > /u/.

It is this morphology that reduces my syllabic possibilities. Not all syllables
with a CC cluster will allow an affix with an infixed part, even if the infix is
always a "vocalic". And introducing additional vowels to make things pronounceable
amounts to have the lovely monosyllabic roots dissolve into an ocean of unrelated
phonemes. I _love_ puns, but don't want to make one (or several ones) each time I
say "pass me the salt".

Thank you for your advice,

Francois
    ^ this one should wear a cedilla!!!

Reply