Re: sumti categories

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

A very belated reply to Jorge, continuing the discussion of terbri klesi
(terbri categories)

> la kolin cusku di'e
>
> > The way it works is that if the terbri is +feat for some feature, and the
> > sumti is -feat, then the combination fails category consistency (I agree
> > that this does not per se make them ungrammatical); similarly
> > if the terbri is -feat, and the sumti +feat. But if either happens to be
> > unspecified for that feature, the combination is OK as far as that feature
> > goes.
>
> Ok, now I will give each feature a value of "usefulness", from 0 to 1,
> such that 0 is useless and 1 most useful.

Fair enough - we will see whether I agree with your values.

>
> If there is no terbri with either +feat or -feat, then the feature is
> useless, because the category doesn't restrict anything. If every terbri is
> either +feat or -feat then the feature has usefulness=1. In general, the
> more terbri there are with the feature unspecified, the less useful the
> feature.

Yes, provided you realise that this is not strictly limited to terbri. I
have indicated that gadri, sumtcita and other cmavo may have these features
as well.
One could imagine for example a classification whereby only certain terbri
could match a particular BAI - not that I think there really are any such
examples. A better example (though of course it does occur with real terbri
as well) is the feature I call +/-location - items which can be governed
by spatial locators such as 'vi'.

>
> > > > Most terbri are -set but unspecified for mass
> > >
> > > Is any terbri unspecified for set? Probably not, in which case +/-set
> > > is a useful category for this purpose.
> > >
> > Yes, there are several. For example 'se badri' and 'lidne',
> > 'se casnu', 'banro', 'se galfi'.
>
> I really don't see how a set (as opposed to its elements) can be
> a topic of discussion 'se casnu', or an object of sadness 'se badri'.
> 'banro' and 'se galfi' are dubious, but I guess we could define
> time dependent sets. 'lidne' seems ok, but I think this belongs
> to a special class of selbri where x1 and x2 have to be the same type
> of object, and there isn't much restriction on what they are as long
> as they are of the same type.

I disagree with your first doubt (obviously).
I take your point about 'lidne', but would enlarge it - there are certainly
a number of selbri which have co-occurrence restrictions on the features
of their terbri, but I suspect there may be cases where the restriction is
not one of identity, ie if the x2 is +feat1 then the x3 must be +feat2.
I can't think of any examples, though, so I might be wrong.

>
> I would say the feature 'set' has a high usefulness. Most terbri
> are -set, some are +set, and very few (but I guess some) are
> undefined %set.
>
> Another very useful feature would be 'number'. For instance namcu,
> se mitre, se cacra, etc, are all +number. Most terbri are -number.
> I can't think of any that would be %number, in which case this
> has usefulness = 1. (Maybe those unspecified for set are also
> unspecified for number, though.)

I think you're right.
na'ipei mi casnu li re
"? We discuss the number two"

(What about na'ipei for the linguist's "?" (doubtful grammaticality)?
I'm not happy about it - I suspect we need a question word on the
jo'a/na'i dimension)

>
> A third useful feature (borrowing from Colin's list) is 'proposition'
> (or du'u). Again most terbri are -proposition, some are +proposition
> (and most of these are clearly so marked in the gismu list) and very
> few if any are %proposition.
>
> Yet another is 'text' (sedu'u).
>
> A nice thing about those four features, on top of them having high
> usefulness, is that they are all mutually exclusive: if a terbri is
> + for one of them, it cannot be + for any of the others.
>
> Properties like 'mass' or 'plant' have very low usefulness (in my
> scale) and so in my opinion it is not so interesting to know the
> value of these features for every terbri, which will be mostly
> unspecified anyway.
>
I agree with you about 'plant' - but I suspect that such categories will
be useful for machine checking and parsing - but not about Mass. I
believe that mass/set/individual is one of the fundamental grammatical
distinctions of Lojban, and is important even if it is comparatively
rarely specified for a terbri.

I also want to make explicit a point that I think has been implicit in
my postings on this subject: I believe that specifying the categories of
terbri (specifically of tergi'u) makes their meanings clearer, and will
also show us where there is vagueness.
In particular, it will help us to see where a meaning has been generalised
by simply relaxing the specification of a feature, or where the gismu
now applies to two different kind of sumti, but with a slightly different
meaning.
For example, I am quite unsure as to the features of 'banxa'. Consider the
feature +/- concrete.

If 'banxa' is +concrete, then it refers to a physical establishment (which
need not of course be a building - it could be the bit of street where the
moneychanger has set up his stall today), and we can say
        mi klama le banxa       "I go to the bank"
but not
        *le banxa cu ponse le gigrupnu be li 15 "The bank has 15 billion pounds"
(We would probably need to talk about the se banxa for that)

If it is -concrete, these judgements are presumably reversed, and I have
to say
        mi klama le baxydi'u    "I go to the bank-building"

A third possibility is that it is unspecified for 'concrete' - I am not sure
what this would mean, but I guess that for some terbri this can be useful.

The fourth possibility, which I would regard as very unfortunate, is that
it may mean both of the above - in other words, unspecified for the
feature, but with a different meaning in the two senses.

Colin