[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Lojban verbosity, the theory of intervals, etc.
- To: cbmvax!uunet!PRC.Unisys.COM!dave
- Subject: Lojban verbosity, the theory of intervals, etc.
- From: cowan (John Cowan)
- Date: Wed, 9 Jan 91 13:29:55 EST
- Cc: lojban-list
- In-reply-to: <9101091715.AA00266@gem.PRC.Unisys.COM>; from "PRC.Unisys.COM!dave" at Jan 9, 91 12:15 pm
> If you have to express [the interval]
> from winter 91 until spring 91
> ca le temci be le dunra be li 91 bei le vensa be li 91
> then this is ridiculously long (at least, for native English
> speakers), and is going to get contracted. You can be sufficiently
> precise in any language if you just use enough words. Since the whole
> point of Lojban is to be precise and unambiguous, something is wrong
> if description of a simple time interval requires this many words.
Well, even though more words are required, the expansion in terms of syllables
(which measures >spoken< language verbosity) isn't that bad. The English
has 12 syllables (each "91" requires three), the Lojban has 20 syllables
(each "91" requires only two: "sopa").
> I have no doubt that anyone "really" speaking Lojban would find a
> shorter, if less precise, way of expressing this idea.
Lojbab pointed out to me that "bi'o", the connective meaning "from...to..."
(but infix) can be used in this context to achieve a 17-syllable version:
ca le dunra be li 91 bi'o le vensa be li 91
> BTW, I have done some work in the use of time intervals in artifical
> intelligence, and might be able to help some if you want precise ways
> of describing relationships between intervals (e.g. the interval that
> Brad is not here and the interval of winter to spring). Someone else
> would have to supply the Lojban words, though.
Please, please, send this stuff in! We have only four words for handling
intervals in Lojban now (one for ordered intervals, "from...to", and one
for unordered intervals, "between...and", plus modifiers that distinguish
between open and closed endpoints), and better theory would be welcome.
> The most widely used scheme for describing the relationship of two
> intervals is James Allen's. He defines the 13 possible exact
> relationships of two intervals.
If you have either hard-copy or (better) electronic versions of this,
please forward to the list!
> I've invented some schemes that are simpler when the relationship of
> the intervals is incompletely known, e.g. if you know one interval
> overlaps the other, but you don't know which interval started first or
> which ended first. I'd use the above sentence fragment as an example,
> but I don't know what it means; the translation back into English was
> given as
> during the time-interval from the winter of year 91 to the
> spring of year 91
> which sounds to me like a zero-length interval, since spring begins when
> winter ends. Which of these is the intent? How much is known?
The context was that Brad wouldn't be home during the winter inter-semester
break, so "winter" and "spring" are being used in their university senses
rather than in the general sense. A zero-length interval was surely not
intended by the original author of the English sentence (not me).
In fact, a more thoughtful paraphrase might have been:
ca la dunra be li 91
during the-thing-called "Winter of the year 91"
where the use of "la" rather than "le" signals that "Winter" is a name in
this context and not to be understood fully literally. The winter
inter-semester break is named "winter" even though it does not coincide
with (astronomical) winter.
[pictures of four possible intervals deleted]
> Or something else? Perhaps the Lojban is precise on this, and it is
> simply the translation back into English that is ambiguous; I don't
> know enough Lojban to tell. All the English tells me is that Brad
> will not be here at some point during the winter. If the Lojban is
> equally ambiguous, then it seems to me that it fails in its primary
> goal, in addition to being too wordy.
The Lojban is ambiguous as to endpoints, which seems to be the thrust
of your pictures. We don't know, in the Lojban as in the English,
whether or not endpoints are included. (Here, of course, the "endpoint"
is a "fat" object.) It is possible to make the Lojban precise by using
the endpoint flags, at the expense of even more wordiness.
> BTW, a distinction should be made between a statement that is
> ambiguous because it's intent cannot be discerned (as above), and one
> that is merely incomplete, because it deliberately conveys partial
> knowledge (e.g. "sometime in winter to sometime in spring").
The Lojban project usually uses the word "vague" rather than "incomplete"
here. The appropriate aphorism (due to Jeff Prothero) is "The price of
infinite precision is infinite verbosity."