misc responses to Jorge from last month, esp. SEI

[Logical Language Group <lojbab@ACCESS.DIGEX.NET>]

>la lojbab cusku di'e
>> Jorge:
>> >In any case, what do you think of {le voksa cu cusku sei krefu sei krefu
>> >lu li'o li'u} for "the voice said over and over again and again ..."
>> >Would that be an appropriate use for an "indicator"?
>>
>> That is not how I would interpret the text.  More like
>> The voice SAID said said (kind of like an echoing)
>
>Echoing the word "said" rather than what the voice said?  I really don't
>understand how you can get that from {krefu}, can you explain more
>fully?
>
>> or perhaps
>> The voice said (I say again) (And AGAIN).
>
>This is similar to what John said, but I still don't understand how
>{krefu}, which means "event x1 is a repetition of event x2" can be
>closer to "I say again" than to "happens repeatedly".  Where is the "I"
>coming from, and why is {krefu} suddenly being used as "repeat" in the
>sense of "say again"?  Isn't that very malglico?

[BTW, Jorge, after I get your signoff from the list, I will save
backlogged responses to things that you post until you are again on-line
so you won't miss them %^).  Oh wait - you are on the list manually to
prevent auto-deletes, so don't forget to send me email telling me to
shut you off.]

Since "sei" is a marker of metalinguistic comment, I am inclined to use
a default "dei", or "di'u" depending on context as a default for an
unspecified x1 or other significant sumti.  Alternatively within a
sentence, it could be a reference to the marked grammatical construct
within the sentence.  I would probably NOT interpret sei to be a comment
about la'edei or la'edi'u unless it were made explicit.  That seems more
likely to be the type of thing one would mark with a normal
parenthetical to/toi.

If the x1 is generally agentive, I tend to agree with Chris that the
presumption is speaker point-of-view, unless there is contravening
context (in stories, for example, I think we developed a convention
whereby the speaker is empathically identifying with some character in
the story - in reported conversation, it might on the other hand be
attempting to report the attitudes of the speakers in the conversations.

So the metalinguistic occurance (the utterance?) is what is happening
repeatedly, and not the thing referred to by the utterance.

Later:
>Chris:
>> The "I" comes from the presumption that attitudinals deal with the speaker's
>> attitude towards the subject matter.
>
>But {sei} is an indicator, not an attitudinal.  (Don't ask me what's the
>difference because I don't know yet.  I'm defending this use of {sei}
>because I find it extremely useful.)

No - "sei" is a 'discursive' marker.  Indicator and attitudinal may be
synonymous or close to it - I may include the evidentials in indicators
but exclude them from the attitudinals, but the distinction is subtle.

The distinction between the attitudinals and discursives (leaving out
"indicators") is that the former "express" whereas the latter "claim".
We have defined the attitudinals such that they are implicitly "true"
(one is presumed not to lie about emotions, but in any event we presume
that it is not possible to assign truth values to something so
implicitly beyond logic).  Thus an attitudinal makes no claim,
metalinguistic or otherwise.  The discursive members of UI, on the other
hand DO make a metalinguistic claim.  They are thus abbreviations for
some metalinguistic bridi that could be expressed explicitly with "sei +
selbri".  There exist metalinguistic bridi that parallel the
attitudinals as well, but these are not presumed to be synonymous with
the attitudinals.  This distinction is important and dates back to and
caused a major dispute between JCB/pc and Jim Carter, which dispute
later was instrumental in the collapse of the Institute organization
meaningfully distinct from JCB.

>> If you want it to refer to something other than the speaker, you should fill
>> in the x1 with something else to override the {mi} default.  e.g. {le nanmu
>> cu klama le zarci sei la noras. cusku}
>
>I refuse to accept that {sei} has a {mi} default for its selbri.  An
>empty slot there should be filled by context as is the usual case.  If
>the selbri needs a person as the x1, then I might agree that "I" may be
>a probable choice, although one of the sumti of the main bridi might
>also be possible, but in the case of {krefu} {mi} doesn't make any sense
>at all.

The main time when you would want to use a default "mi" is when you are
emulating an attitudinal, and talking about emotions.  Lojban includes
the presumption common to many languages that one simply does not
express someone else's emotions.  You CAN override this assumption using
"sei" (the result then is a metalinguistic claim about someone's
emotions and not an attitudinal).

On the other hand, the original bona fide Lojban usage for "sei" ("sei"
was in the language earlier - dating as I said from my inclusive
solution to the Carter/pc disagreement, but it had not seen any usage)
was for the reported conversations of Saki's "The Open Window", when we
realized that English and most other languages, when citing
conversations, tended to insert the "he said"/"she said" in the middle
of the quote (Lojban seisa'a).  In such inserts, there often is an
adverb or even longer claim about the speaker's manner or behavior while
speaking.  In such cases, the default x1 is thus clearly the speaker,
and not the narrator.

ni'o
>> and you have effectively
>> said that there exist cases where you would have an ambiguity with
>> multiple ke'a in a single sentence that would HAVE to be resolved using
>> subscripts,
>
>If I said that, I retract it.  With And's smart use of the prenex all
>need for subscripts becomes obsolete.  Besides, that is nothing new.
>The problem of embedded {ke'a}s already exists even if you stay with
>relative clauses only.

Thus far we have managed to avoid requiring prenexes in order to say
most things in the language.  Prenexes are elegant, but the fact that
most Lojbanists avoid them confirms the hypothesis that they are not
especially natural even though they are clear.

ni'o
>I propose that the place structure of {skari} be changed to "x1 is of
>the colour of x2", which is much more useful than the current one.

1. Because the truth condition is indeterminate in that case.  Colors
truly are observer and conditions dependent (at least to some people -
other consider color an innate property implicit in the atomic or
molecular structure).

It is not necessarily true that two things are the same color to all
observers and under all conditions.

2. What do you put in x2 for x1 = ultraviolet light, or for that matter,
he sun, which has a mixture of several frequencies of light.

The problem is that this has been debated over and over for several
years of Loglan history, inconclusively, and no one can agree as to what
the place structure of color words should be.  Between skari and the
individual colors, I tried to cover all bases.

BTW, try for x2 putting in hue/brightness/saturation as another
alternative.  The current place structure was generalized from JCB's
which required specification of those values (which doesn't work too
well for color mixtures or for people who do not use scientific
instruments to decide what color something is - i.e. most of us).

>> and maybe Nick needs to include
>> a note about these kinds of words. rango, danlu a nd a few others have
>> this problem too.
>
>{danlu} seems to be perfectly normal, and {rango} is pretty standard
>too.  What is the problem with them?

It seems likely that lujvo based on those words might not be intended to
use more than the x1 place.

ni'o
>PS1:  Instead of {ca} I would have preferred to use the proposed new
>ZAhO for "already".

bapu'o or ba'opu'o ???

>> 2. How do we express confidence intervals?
>
>There is not much flexibility to do that, at least nobody has shown how
>to do it.  For approximate numbers using {ji'i} there are two schools:
>the official take is to use for example 3ji'i45 to mean 345 +/- 50.  My
>approach is to use the two ends of the interval, e.g. 300ji'i400.

You can't do it with MAI, but you can with XI, which can take a MEX
construct.  BIhI has both endpoint and center-range formulations.

lojbab