More on Meanings of grammatical gismu

[Colin Fine <C.J.Fine@bradford.ac.uk>]

And comments:
++++++>
 > However, if this is correct, then the 'language' place of 'gadri' is
> inappropriate.

I don't understand why.
>++++++++

OK, I have been refining the argument a little.

I claim that there are two parallel series of concepts which have not been
 clearly
distinguished up to now.

One is of the general form
'x1 is a grammatical word/structure of class/type x2 with function x3 in
 language
x4'

The other is of the general form
'x1 is a word/structure of type/class x2 with function/meaning x3 in
string/phrase/sentence x4'

I would call them 'type' and 'instance' respectively - the best I have come up
 with in
Lojban is 'sucta' and 'selsucta', but these aren't good. I don't think 'mupli'
 works.

The comment And picks up above amounts to saying that the second (instance) case
does not need a 'language' place ('inappropriate' was a bit strong), as the item
 is by
definition embedded in a phrase which presumably has a language.

Examples of the first type ('type') are:
-----------------------------------
cmavo x1 is a structure word, grammar exemplified by word x2, with
meaning/function x3 in language x4
gismu x1 is a (Lojban) root word expressing relation x2 among argument roles x3,
with affix(es) x4 (note the 'argument roles', a dead giveaway. All its
 conversions
are also 'type'. It would presumably need a 'language' place if it were not
 restricted
to Lojban)
cmene  x1 (quoted word(s)) is a/the name/title/tag of x2 to/used-by namer/name-
user x3 (person) (No language place, as names are theoretically language-
independent. In fact of course they're not, and maybe it should have the place
 in).
tanru   x1 is a binary metaphor formed with words/concepts x2 and x3, giving
meaning x4 in language x5
lujvo   x1 is a compound (Lojban) predicate word with meaning x2 and arguments
 x3
built from metaphor x4

Examples of the second type ('instance') are:
----------------------------------------
gadri x1 is an article/descriptor labelling description x2 in sentence x3,
 language
x4, semantics x5 (As I suggested, I don't think this needs the 'language' place,
 which
is in any case ordered inconsistently with 'cmavo'. I have also previously
 pointed out
the ambiguity in the definition: I don't know whether the se gadri is the whole
description including the gadri, or the phrase which it introduces.
sumti   x1 is a/the argument of predicate/function x2 filling place x3
 (kindanumber);
(Actually not necessarily linguistic, but as defined it is clearly an instance.
 Notice
that a 'te sumti' is also an instance, and so is the place of an actual
 predicate or
function, as opposed to a 'te gismu' which is an argument role).
valsi   x1 is a word meaning/causing x2 in language x3

Not fitting in either are:
----------------------
bridi   x1 (du'u) is a predicate relationship with relation x2 among arguments
(sequence/set) x3 (not necessarily a linguistic object at all, but I think it's
 an
'instance' in my sense.)
jufra x1 is a sentence about x2 (topic/subject/predicate/focus) in language x3
 (is an
instance, but not part of a sentence, so  does not satisfy my definition for
 type 2)
gerna   x1 is the grammar/rules/defining form of language x2 for structure/text
 x3
(not in either type. But 'te gerna' is by definition either of them, which I
 think is an
unfortunate ambiguity).


For words whose definition is as 'types' there is little harm in using them for
 the
instances as well - to be precise, we can use a tanru with 'mupli' or something.
 But
for those defined as 'instance', I don't think we can satisfactorily discuss
 their
grammar. In particular, note that we are in the habit of using the term 'sumti'
 for
anything which has the grammatical structure which would be legal as the sumti
 of a
sentence, but as the word is defined, unless it is actually made a (specific)
 argument
of a predicate, it is not a sumti!

        Colin