dikyjvo misunderstood

[cbmvax!uunet!MULLIAN.EE.MU.OZ.AU!pucc.PRINCETON.EDU!nsn]

Lojbab says:
>Jimc writes:
>>In wannabe-soldier, the tanru is sonci djica, the lujvo is soidji, and
>>in both forms you are supposed to interpret the phrase as "x1 djica
>>lodu'u x1 sonci".
>Jim chose a good example as to why I dislike dikyjvo.  WHY must "sonci
>djica" 'mean' "djica ledu'u sonci"?  Why can't it 'mean' "djica lo
>sonci"?  (a military camp follower, perhaps), or lo djica poi sonci (the
>soldier in the foxhole, yearning for home, beloved, and a steak dinner)?

No reason. In fact, Bob, my congrats: you have just outlined the three
possible interpretations of a dikyjvo. The whole point is that a) there
is nothing out there that constrains us to select one or the other alternatives;

(b)
>The tanru COULD mean any of those, and a bunch of other things besides.

The only distinction between dikyjvo and looser lujvo is that a dikyjvo can
only mean any of those, and may (*may*) have a well-defined, predictable
place structure; (c) this prediction, and the selection of the dikyjvo
flavour , is a matter of pragmatics - and will probably boil down to a
combination of Zipf and separate rules for each end-gismu, as Jim predicts;
(d) what people are ignoring is the distinction between trivial lujvo
formation, like factitives (ri'a or gau), and real nasties - give me the
place structure for a lujvo for feminism. And when you have nasties like
that to consider (and not only will they be nasty, they may be the most
serious impediment to communication in the language), why spend *any* effort
on deciding the stucture of a facticitve? That's crazy.

In the case of soidji, the reason I'd go for Jimc's interpretation (belenu
flavour) is that it compresses a longer deep structure phrasing than the
other two flavours do - the compressing being both  in length and in
structural complexity. Jimc's interpretation makes things easy for any would-be
expressor of the not infrequent expression "would-be soldier"; the others give
you such a small saving in space and conceptualisation, that there's no reason
for them to be in lujvo space anyway.

And that, Virginia, is all there is to this horrid dikyjvo monster.

>It is only pragmatics (real world knowledge) that might let us determine
>which of the possible meanings is most plausible for a given context.

See above.

>Thus, to exercise Lojban as a 'logical' language, you would probably
>avoid tanru.

Yeah right, I'll believe that in a hurry...

Lojbab complains that he won't have soidji as would-be soldier, because it
doesn't come to him intuitively. I believe that {belenu} dikyjvo should be
written up in JL (I'll do the job), because they *don't* readily intuit to
an English-speaker - but once you know them, they are incredibly productive.
I ask Lojbab how he *would* express would-be-soldier in lujvo. Lojbab might
not like {soidji}; I look at {-dji} as a "suffix", and cannot think of any
better interpretation as a general rule. It's a more productive pattern.
But there is an essential conflict between the {be} and the {belenu} inter-
pretation, and I won't be too loud on it.


If you people still don't know what I'm talking about, ask Cowan or somebody.
This is the fifth or so time I've had to explain dikyjvo-as-a-reality. There
is *no* means of a priori predicting which flavour dikyjvo is to be used, and
I wish Lojbab would stop reading such an intent in (or Jimc leaving such an
impression in Lojban dikyjvo).

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Nick Nicholas, Melbourne Uni, Australia.  nsn@{munagin.ee|mullauna.cs}.mu.oz.au
"Despite millions of dollars of research, death continues to be this nation's
number one killer"      - Henry Gibson, Kentucky Fried Movie
_______________________________________________________________________________