[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Lojbab on subject/object
- To: John Cowan <cowan@SNARK.THYRSUS.COM>, Eric Raymond <eric@SNARK.THYRSUS.COM>, Eric Tiedemann <est@SNARK.THYRSUS.COM>
- Subject: Re: Lojbab on subject/object
- From: John Cowan <cbmvax!UUNET.UU.NET!snark.thyrsus.com!cowan>
- In-Reply-To: <email@example.com>; from "And Rosta" at Oct 8, 91 3:30 pm
- Reply-To: John Cowan <cbmvax!UUNET.UU.NET!snark.thyrsus.com!cowan>
- Sender: Lojban list <cbmvax!uunet!CUVMA.BITNET!pucc.PRINCETON.EDU!LOJBAN>
[This message is entirely about Lojban, so I've moved it from conlang to
Lojban List. Apologies to non-conlang subscribers.]
la .and. rostas. cusku di'e:
> Two very general principles are involved in determining clause structure:
> (1) Semantic roles must be matched with grammatical relations, which are
> signalled by case or by word order.
> If I understand Lojbab right, Lojban uses both techniques.
> For signalling by word order, unmarked X1 precedes unmarked X2, and so on.
> For signalling by case, Lojban uses _fe_, _fi_, _fo_, etc.
This is correct, except for a minor point of Lojban jargon. To avoid the
slippery term "case" with its multiple meanings (semantic vs. morphological)
we call the argument slots of a predicate its "places".
> (2) Word order must signal informational prominence. If you mark semantic
> roles by case markers, then the word order can be free within the clause,
> so it is easy to signal informational prominence. But if marking of
> grammatical relations (& hence roles) is shown by word order, you need
> processes like passive and the sort of rules Relational Grammar loves
> to swap around grammatical relations and roles.
> Lojban uses (i) case markers, or (ii) a number of particles, such as _se_,
> which signal passive-like processes.
This is also correct.
> Lojban, it may be seen, liking to eat its cake and have it, kills both birds
> with several stones! Is it the 'Lojban way', that if there are several
> solutions to a problem then all are adopted?
In general, yes. Sometimes one is more highly marked (needs more particles
than the other). For example, the normal order of compounds is AN (modifier
before modificand) but the particle "co" produces the reverse order.
Likewise (returning to where this discussion began) SVO and SOV are unmarked
orders, but VSO requires minimal marking and the other orders require
heavier marking. On the other hand, subordinating and coordinating causals
require about the same amount: "A because B" can be "A ri'a le nu B" or
"A .i ri'a bo B" respectively.
> Lojbab also discusses how to omit arguments in Lojban by putting some sort
> of dummy in its position. Is it also possible to use the case markers and
> then not signal omitted arguments? From the examples given this seems
Certainly. Putting in the dummy (equivalent to the non-quantified use of
English "something", as in "something bit me!") or using the place markers
are yet another structural option; which is more heavyweight depends on the
complexity of the construction.
> Finally, I have been rather ... um ... shocked to find predicates with more
> than three arguments. Does this not impose an unnecessarily large memory load
> in learning a verb - it's easier to learn two argument places than six.
> Extra arguments can be expressed by using additional predicates (like
> prepositions in English).
Lojban does have the analogue of prepositions for adding extra places,
but the underlying theory (not necessarily carried out perfectly for every
predicate!) is that the numbered places represent those which are >necessary<
to the claim.
For example, the predicate "klama" (which can be glossed "come" or "go")
requires five places: a doer, a destination, an origin, a route, and a means.
Any state of affairs which does not involve all five of those cannot properly
be called a "klama". This is, of course, not the same as saying that all
five need be >expressed< in a given utterance, simply that they must >exist<.
The movement of the Earth around the Sun, for example, has neither origin
nor destination, and so it is not true that "la terdi cu klama"; but the
word "litru" (roughly "travel"), which involves only a traveler, a route, and a
means, would be suitable.
> I don't (yet) know Lojban, and am therefore
A most encouraging parenthesis!
> likely to be mistaken, but this seems like an unwarranted flaw in the design
> of the language: you buy brevity at the cost of a significant multiplication
> of the amount of information listed in lexical entries.
So you do. The desire for brevity tends to increase the number of places,
whereas the desire for ease of recall tends to decrease them. The issue
of "metaphysical necessity" serves as a tertium quid. There are many sets
of words in the language which are contrasted precisely by the number of
places metaphysically necessary for each:
fasnu x1 happens (no actor)
gasnu x1 does x2, x1 is the actor in event x2
zukte x1 does x2 for x3, x1 employs means x2 to end x3
fatci x1 is a fact (absolute)
jetnu x1 is true under conditions x2 by standard x3
binxo x1 changes into x2 under conditions x3 (no actor)
galfi event x1 changes x2 into x3
cenba x1 changes in property x2 by amount x3
under conditions x4
litru x1 travels route x2 by means x3
(no origin or destination)
cliva x1 leaves x2 via route x3 by means x4 (no destination)
klama x1 comes/goes to x2 from x3 via route x4 by means x5
e'osai ko sarji la lojban