Re: arguments of verbs of motion

[And Rosta <cbmvax!uunet!pucc.princeton.edu!ucleaar>]

rAtZAnkauEnNo_ZU (John Cowan writes:)
>
>la .and. rostas. cusku di'e
>
>> This is undeniably long-winded, but I wish to make the point that these
>> are the only necessary arguments. I disagree with John Cowan (I think it
>> was) and lojban in that I don't believe there is such a thing as motion
>> without a destination, as distinct from motion without *specified*
>> destination.
>
>The point is that Lojban calls that kind of motion "litru", and of course
>such a path may be closed, without beginning or end (example: the Earth's
>orbit around the Sun).  "klama", on the other hand, describes the kind of
>motion that does have a beginning and an end, whether or not specified
>by the speaker.  So "klama" vs. "litru" makes a semantic contrast:  "motion
>over a path with beginning and end (and a means)" vs. "motion over a path
>possibly without a beginning or an end (with a means)".

I understand the reasoning, but I question the need for the additional
predicate. Before explaining precisely why, let me suggest for consideration
a general principle constructed languages might adhere to:
    To express a compositional meaning, it is preferable for the
    lexicalization of the meaning to be compositional. For example,
    _kill_ and _die_ according to Lojban principles might well belong
    to different lexemes (as in English), because they have distinct
    argument structures, but the sense of _kill_ is composed of "cause
    to die", so my principle advocates not having a word _kill_, but
    instead using _cause to die_.
This principle is why I don't just say "so what" when I see distinct
lexemes for _klama_ and _litru_.

In my previous message I suggested that going really only has two arguments:
theme and location/path. How does this square with John's examples? I think
that in the orbiting example the path is not delimited (in the way that
the referent of a mass noun, or of a verb in imperfective aspect is not
delimited). So in Sta, _klama_ would be:

     go -LOC-> thing <-LOC- delimited_amount

while _litru_ would be :

     go -LOC-> thing <-LOC- nondelimited_amount

I previously defined 'source' and 'goal' as first and last (equivalent
to start and end) parts of the location of go. A nondelimited entity can't
have a first and last part.

I'd like to know what you think of the 'principle of lexicalization of
compositional meaning', and of the analysis presented here. I'd also like
to know whether I can calque the Sta examples in Lojban (using relative
clauses or such like, and predicates for delimitation & nondelimitation).

>
>> Extending the discussion, it is the case that almost every action may or
>> may not involve the use of an instrument. Surely it is mad to duplicate
>> every action verb in the lexicon in order to show whether an instrument
>> is or isn't involved.
>
>True, and that's why we have the "extra arguments" mechanism, usually called
>BAI places after the prototype word of the semantic category of extra-argument
>flags.  ("bai" = roughly "compelled by...").  "With instrument" would be
>marked by "sepi'o":
>
>        mi kakpa zo'e loi dertu sepi'o lemi canpa
>        I dig something-unspecified from-the-mass-of dirt with-instrument
>                my shovel.
>
But why have both _klama_ and _litru_ but only _kakpa_? Why not abandon
_klama_ and use _litru_ with BAI places for _klama_'s extra arguments?

----------------
And

ps: John: could you resend me a copy of your message I'm replying to here:
I accidentally destroyed my copy.

pps: Lojbab: I received today a ginormous wodge of Lojban info, for which:
many thanks.