response to jimc on "only"

[Logical Language Group <lojbab@GREBYN.COM>]

jimc writes:
>The word "only" was one of the first that I broke my
>teeth on.  Here is a short list of meanings, supplementing what Colin
>wrote:
>
>(1)     The only way is love            There is exactly one X which is
>                                        a tadji (way), AND X is love
>(2)     I ate only two cookies          I ate two cookies, AND two is less
>                                        than the expected number for this
>                                        situation
>(3)     I only ate two cookies          I ate two cookies, AND that event
>                                        is less than what would normally
>                                        justify the criticism or punishment
>                                        that you are putting me through
>(4)     She is only a servant           She is a servant, AND this condition
>                                        implies a social status that is less
>                                        than what is normally expected for
>                                        the present situation
>(5)     Smoke if you want, only not     Smoke if you want [discursive of
>        where I can smell it.           contrast] don't smoke where...

I will agree that there seems to be a secondary implication in some of
these of a denial of expected value.  But the essential point in all of these
is "...and no other"

The way (and no other) is love

I ate two (and no other) cookies) (for both 2 and 3, though the English
  positioning does change the attitudinal implication; Lojban would use a
  different attitudinal, .i'anai in addition.  Note, of course, that the
  "only" is unnecessary in quantificational situations since, per our
  earlier discussion, "two" implicitly means "only two".)

She is only a servant (and nothing else).  (This for me actually causes
  the most challenge to the concept of "only" as an attitudinal of the
  cases.  It suggests that "only" might be a value on the scalar negation
  scale rather than an attitudinal.  I could live with interpreting my
  problems words as additions to the je'a/na'e scale instead of
  attitudinals (or both if appropriate).)

Smoke where you want, (but/such that/intersection) not where I can smell it
  (and no other place).  Here the only is an intensifier of the not.  It
  changes the type of connective to be chosen, but there is still an
  implied "and no other" in the result; it just happens to be identical to
  the result of "but not".

Note that I am not saying that I want a word to stand for all uses of
"only", but rather for this underlying central one.  I see this as
do-able using one or all of:  an attitudinal, a brivla (key word
"exclusively" is fine), or as an addition to the scalar modifiers
related to na'e) My concern about non-attitudinal solutions is whether
the result is flexible enough for all "only" situations.

>Meanings 2,3,4 are closely related.  In -gua!spi, to solve the "only"
>problem, I came up with the following list of gismu (showing only the
>definitions):

Recognizing Cowan's commentary on these, which I agree with, I give my
own independent first reactions, since they lead in a new direction.

>(a)  Object x1 is sufficient in dimension x3 for event/condition x2 to be
>     true about it (x2 and x3 contain a cleft place replicating / ke'a'ing
>     x1).  Default x2 is "to qualify as x3".  Scalar neutral negation gives
>     "insufficient".

I think this is "banzu".  Cowan's answer may be better than mine.

>(b)  Object x1 is not quite sufficient ("almost" sufficient) in dimension
>     x3 for event/condition x2 to be true about it.  Polar
>     opposite negation gives "barely".

I could agree to a brivla for this, or a na'e series entry.  There seems
to be a close relation to "so'a" "almost-all-of", which makes me wonder
if so'amei would do the trick.  It also makes me wonder if the grammar
does (or could) allow scalar negation of quantifiers, as used in the
non-MEX grammar (I'm sure they can be used on the MEX side, but it would
be nice to avoid bracketing).  That would suggest that "so'a" should be
in the "na'e" series and not in the indefinite quantifier series.
Probably the same for "du'e" then as well, which would be jimc's (d)
below).  But as a NAhE it would have a wider use than as a quantifier, I
think.  Since you can combine NAhEs, na'eso'a would be "barely", which
is what I think I meant by "just" ("just" may mean equal to or just
slightly greater than, whereas this would mean just slightly greater
than.  I can live with such differences from the English, since I'm not
looking for encoded English.

Cowan seems to agree that this isn't presently covered except through so'a.


>(c) Antecedent event x1 is a sufficient condition for consequence x2 to
>occur

I'm not clear how this is different from some form of (a), or maybe vice versa.

>(d)  x1 is more than usual or expected on dimension x2 for members of set x3.
>     Scalar neutral negation gives "fairly", i.e. at the high end of the
>     usual range.

see (b) above re "du'e", and "na'edu'e" might then mean jimc's "fairly".
Per Cowan, the selbri version of this is dukse.

>(e)  x1 is less than usual or expected...  Scalar neutral negation gives
>     "somewhat", i.e. at the low end of the usual range.  Polar opposite
>     negation could conceivably be interpreted as converting (d) into (e),
>     but I wanted a clear route to obtain "somewhat", so I gave a separate
>     gismu.

This sounds like so'o, though Cowan is right that to'e dukse covers the
selbri version.  I start feeling that I am risk for suggesting that the
whole so'a series should be moved to NAhE, which I don't think is
appropriate.  But perhaps some of them may be misplaced, at least the
two high end members.

Thus the question comes to be whether so'Vmei (brivla) does the job,
some of these need to be in "na'e", or somethings else.  Discussion
welcome.  Seems like a selma'o change would be a significant change in
some long-stable cmavo, but I'm not sure it would effect much historical
usage, since the words being questioned are the less-used ones.  "du'e"
was a late addition to the series, of course.  But using the "mei"
versions of these if they could be applicable would seem the most
conservative solution.

>Negated versions of (a) or (c), with an implied pronoun in x2 connecting
>to the implied enclosing circumstance presumably expressed in preceeding
>discourse, give the "only" meanings of 2, 3 and 4. (e) can also be used
>to translate example 2.

I don't think I see how this covers the "and no other" aspect of those
meanings, unless your hypothetical predicates have the exactness attribute
of the numbers.  This is evidence for the "so'Vmei" approach.  But I still
think that the "and no others" is a significant need in the language.

So:

option 1 status quo

for dealing with just/barely/somewhat
option 2a interpretation of so'Vmei to meet these semi-quantified scalar usages
option 2b conversion of some of the so'V series to NAhE
option 2c removing the so'V series to a separate selma'o that can be used
          as a quantifier or a scalar, if grammatically feasible (probably not)

for dealing with my sense of only/exclusively
option 3a addition of attitudinal for "and no other"
option 3b addition of a NAhE for "and no other"
option 3c addition/identification of a brivla for "and no other"

(I used brivla in the above, since I would not rule out a good lujvo instead
of a gismu for these concepts.)

lojbab