Re: {soi}

[ucleaar <ucleaar@UCL.AC.UK>]

la djan cusku die [incidentally, if John is {djan}, and Jen
is {djen}, what is Jan? Are Jan & Jen homophonous in Gen.
American? Ah, no, Jan would be {dji,yn} wouldn't it, because
Ann is homophonous with Ian.]

> > Thanks for the explanation. Is the {dy}/{ri} difference that {dy}
> > refers to whatever its antecedent refers to, whereas {ri} repeats
> > or reactivates its antecedent (so the reference remains constant).
> > Is this degree of subtlety necessary?
> Yes, that was my intent.  No, that degree of subtlety is probably not
> necessary, but somone may find a counterexample (porbably involving
> references to references) so I wanted to be prepared.

Perhaps this relates to the issue of how anaphoric references to
pluralities work: does {re prenu cu prami py} mean they each love
themself, or does it mean they each love each of them? I think
Jorge & others were debating this for each of the systems of
anaphora, but I forget (or never knew) what the upshot was.

> > You say the grammar is
> >   "soi <sumti-reference-1> <sumti-reference-2> [se'u]"
> > Is that 'sumti' in the syntactic or the semantic sense (i.e. is it
> > necessarily lexical)?
> I'm confused.  Syntactically, any sumti can appear; if a sumti that
> doesn't refer to another sumti appears, the meaning is indeterminate.

Does it refer (in some sense) to a previous word/phrase? Or does it
refer to the referent of a specific previous word/phrase? Or does it
refer to an individual (typically not a word) that has already been
referred to? If the last of these, then it ought to be possible to
use a {le} or {la} expression within {soi ... seu}. I gather from
what you say that that should indeed be possible.

I discern an ambiguity in the (English) use of "sumti" here. On the
one hand it usually means "syntactic argument" or "semantic argument",
while on the other hand it here means something more like "nominal
expression", "NP" rather than "S", an expression with type <e> rather
than type <t>.

> Not all sumti that refer to other sumti are lexical items;
> "le se go'i"; "le go'e", etc.

But they're still lexical: they're {lo valsi}, albeit not {pa valsi}.

> > And is that 'reference' in the sense of 'referent'
> > or in the sense of 'cross-reference/pointer'?
> I'm not sure I can make this distinction.

In a book most words refer to something (in the sense of building up
a communicable picture of the world), but only expressions like
"see page 30" are cross-references/pointers.

And:
> As we are on this point, could you perhaps say whether x1, x2, x3 of {sumti}
> and x1, x2, x3 of {bridi} refer to logicosemantic or to syntactic objects?
> The definitions make it sound like they are logicosemantic, but in actual
> usage they are almost always syntactic.
> We should distinguish either between
>    sumti               v.  vlasui/sumvla
>    duu, bridi          v.  vlabri/brivla
>                            (but this last standardly means selbrivla)
> or
>    sibsui/sumsio       v.  sumti
>    duu, sibbri/brisio  v.  bridi
> The giuste supports the former. Actual usage supports the latter.

Jorge:
> That's my impression too. When I asked Goran whether he thought
> {mi sumti} or {zo mi sumti} was correct in some context like {mi klama},
> he said that the second one was right. That's how we use "sumti", even
> though the definition suggests that {mi sumti} should be right.

John:
> I don't know what the actual usage in Lojban is these days.  :-)
> I don't think that English usage is necessarily determinative:
> English is a notorious magpie borrower that perverts words from their
> original senses with abandon, e.g. Sp. >sombrero< 'hat' > Eng.
> >sombrero< 'Mexican-type hat'.

I would prefer to take John's point, and see our mainly in-English
use of {sumti, bridi} as loose and ambiguous, as English words often
are, and say that the Lojban meanings of {sumti, bridi} really are
as the giuste has them. In that case, if {mi klama}, then {mi sumti},
and not if {lu mi klama liu se bacru} then {zo mi sumti}.

I prefer this, since I think the logical meanings are more basic and
more firmly grounded than the syntactic ones.

In this case, we should start being pedantic and use

   sumvla    = valsi be lo sumti
   brivla    = valsi be lo bridi & not = "brivla" as currently used
   selbrivla = valsi be lo selbri

I forget the in-English difference between "selbri" and "brivla".
"Selbri" denotes a syntactic funtion, the grammatical predicate,
and "brivla" denotes words that can function as "selbri".
But in that I see no basis for a terminological decision.

---
And