TECH: pe'a/po'a proposal (long)

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

John Cowan writes:
>Subject:      TECH: Figurative speech: a minor change proposal

I talked with Nora about this tonite, and her reaction was somewhat
negative.  I, too, am unconvinced, but think that Frank's implied
suggestion that led to John's proposal here may be a good idea worth
making an experimental part of the language (i.e., if unused, the status
quo would prevail).  I thus will at the end suggest a counterproposal
which I think supports the status quo as well as a variation on this
proposal which is fully optional.

Since both versions of the proposal incur rafsi assignment changes, one
decision must be made quickly.  It has been my intent to baseline the
modified rafsi list as of 1 June, which is Tuesday.  I CAN make a change
in the assignment for "pe'a" (which was assigned "peb" in the status quo
baseline proposal) such that it allows consideration of one or both
versions of change - we can then add a new rafsi for "po'a" which is
mnemonically consistent.  This would not necessarily commit us to either
version of change, and allow a stable baseline.  Or given my
counterproposal, we can make the indicated change involving only 1
rafsi, which allows later consideration and addition of a 2nd rafsi
should it be considered necessary.

****************************************************************************
*                                                                          *
* I especially want to hear NICK's opinion on this issue, since he is our  *
* resident lujvo-making policy-proposer, and IVAN's opinion as well, since *
* he has studied comparative tanru-making across a spectrum of the world's *
* languages and may have insights into types of metaphors that we might    *
* tolerate in Lojban according to our current policies but which have not  *
* actually appeared in current usage.  Failing that, or in addition, I     *
* would like John to look at Ivan's unpublished paper on the subject,      *
* since I know John has specifically looked at the Lojban aspects of the   *
* tanru Ivan discovered.                                                   *
*                                                                          *
****************************************************************************

>Frank Schultz has been talking here and on conlang about different kinds
>of compounds:  semantic restriction ("blue house"), semantic extension
>("wolf man"), poetic figurative forms ("broken heart").  Most
>tanru/lujvo are of the first kind, but the second kind is also possible;
>we have special marking particles for "figurative speech".  This posting
>discusses how the current markers work and some of the motivation behind
>them, and proposes a slight change to make them both simpler and more
>flexible.

The norm for Lojban tanru is of the first (restriction) type, as John
said.  Nora (and John, I think) are skeptical about the second type, or
at least about Frank's example - is a (werewolf) "wolf man" NOT a man of
some type?  After all, the gismu are designed with broad meanings.
Furthermore, if one were to assume that a werewolf were NOT a type of
man, then whatever it really is (a semantically extended base "man") is
still being restricted by the modifier "wolf".  Thus, Nora sees this
example as being either purely or partially a semantic restriction.
Probably most examples that people would propose as being 'semantic
extensions' would indeed be a combination of an extension of one term
and a restriction on the other term.  If a tanru were composed of
expansions of BOTH terms, it would probably be considered by most people
to be of the third type.

Other examples we came up with also fit this analysis:  "space ship" is
in some sense an expansion of the concept "ship", but the result is
still being reduced by the modifier "space".  We probably would consider
a tanru based on "space ship" to be valid Lojban since you could come up
with values for all places of "ship", as well as all places of "space"
to fit the intended meaning.  Another example, "cat house" in the sense
meaning "brothel", is a kind of house, but the modifier "cat" has been
expanded probably beyond acceptablility for Lojban to apply to human
females.  The rejected Loglan usage of "dog woman" (i.e.  "bitch") as a
derogative term for a human female also fits this latter category.

Attempting to counter Nora's analysis, I suggested that if the right
term is modified such that the resulting concept would use a consistent
place structure with the final term, but the semantics is such that the
final term alone would not be accurate, then you have what Frank is
calling a 'semantic expansion' of meaning while NOT having a 'poetic
expansion'.  There can exist tanru in Lojban that meet this criteria,
since the ultimate rule of Lojban tanru is that they always have the
place structure of the final term (unless explicitly marked with "pe'a"
or "po'a" as John defined in his posting, in which case the place
structure is explicitly undefined).  Unfortunately, I can't think of
any examples from actual usage.

Nora suggested back that "NAhE"-based tanru fit this version of
second-category definition.  In a sense, they do (a "na'e nanmu"
probably must be something that has a place structure consistent with
nanmu based on the tanru place structure rules, but which is not in
itself a "nanmu").  But we have NOT traditionally considered NAhE and SE
tanru to be normal tanru in the modifier/ modificand sense (and SE
tanru, by definition, result in a place structure DIFFERENT from the
original).

It might be that an unmarked usage of the type for which the cmavo
"xo'e" is being proposed (in that thread on this list) would be a
semantic extension, by the analogy that ADDING a place, as with BAI,
inherently imposes a restriction.  Thus a "zbasu" relationship without a
maker or without a material would be a semantic expansion of "zbasu".
But of course, under current Lojban, tanru based on such expansions are
not permitted (although LUJVO which reduce places through modification
are permitted and probably will be common - I would consider
"plutyclaxyklama" (long-form lujvo for "route-without-go") as an
acceptable lujvo for teleportation as I've defined in the other thread
with example "klama fo xo'e", and having exactly 4 places, excluding the
route place.)

However, Nora feels that these examples "cat house", "dog woman", and
"xo'e"-type semantic modifications are all 'poetic' - they rely on an
assumed understanding of the implied meanings of one or more gismu which
extends beyond what she would consider acceptable in a culturally
neutral language.  The sense in which a human female is in ANY way a
"dog" or a "cat" is to her very cultural, indeed unintelligible without
recourse to a specific cultural point of view.  And she dislikes the
idea of "klama fo xo'e" (which is in any case not a tanru) being in
any way considered a kind of "klama" relation(ship).

But in these, at least the place structure has some relation to the
original and official tanru place structure, unlike Frank's third
category of metaphor ("broken heart" does not in any way refer to the
organ for pumping blood; nor is it in any real sense 'broken').  The
less extreme example "desert ship" for "camel" (i.e. ship of the desert)
is a borderline case somewhere between types 2 and 3:  it is after all
restricted to the desert in some way, but it is a poetic and possibly
culturally restricted metaphor to see a camel as a kind of ship.  An
example that has seen Lojbanic usage is Michael Helsem's "purple
Lojban", which is used to apply to some of his writings that he admits
is probably invalid Lojban.  Such text is in no way "zirpu" = "purple",
and it is indeed (especially in some of Michael's more extreme writings)
arguable as to whether the text is in any sense 'Lojban'.

Whether Michael's lujvo (as opposed to tanru) based on "purple Lojban"
is valid Lojban without "pe'a" or "po'a" is arguable, as noted in the
current thread being led by And on the segmentation of lujvo-meanings.
While technically permitted by the lack of an official policy forbidding
it, I consider even lujvo that are that 'poetic' to be "mabla"
(derogative) unles marked, and proposed the rafsi for "pe'a" as part of
the current change proposal as an answer.

>Currently, we have two cmavo for figurative speech:  "pe'a" (selma'o
>PEhA) and "po'a" (selma'o POhA).  Both are grammarless particles -- they
>can occur anywhere.  When both are used, they are the left and right
>markers of figurative speech:
...
>The use of the markers ... signals a culturally dependent metaphor --
>people from non-English-speaking cultures are not expected to understand
>it, but are at least cautioned not to take it literally.

>When "pe'a" is used without "po'a", the figurative intention persists
>indefinitely:  this could be used in skaldic poetry or the like, where
>all tanru are probably figurative.  However, it is also valid to use
>"po'a" without "pe'a".  In this case, "po'a" is treated as if it
>belonged to selma'o UI:  it applies to the previous word; if the
>previous word begins or ends a grammatical construct, it applies to the
>whole construct.  The regular mechanism for extending the scope of
>attitudinals ("fu'e" and "fu'o") does not apply to "po'a".

>The proposal is to abandon this grammatical separation, and make both
>"pe'a" and "po'a" regular members of UI.  The intention is to then
>separate them semantically.  "pe'a" would be assigned to poetic
>metaphor, whereas "po'a" would be used for semantic extension.  (Memory
>hook:  pe'a/pemci.)  This change (techfix 33) would simplify the grammar
>a bit, as three selma'o (PEhA, POhA, UI) would be merged; this would
>allow removing some hard-coded C support as well, although not enough to
>be a substantial consideration.  For occasions when figurative speech
>must be prolonged, the regular UI mechanism with "fu'e" and "fu'o" would
>be used.

>All UI cmavo can be negated with "-nai", giving a polar negation.
>"pe'anai" would mean "literal no matter how absurd", and "po'anai" would
>mean "a semantic restriction even though it looks like an extension".

Further thought leads me to query whether "literal" ever means anything
other than "restriction".

>In addition, both "pe'a" and "po'a" would be given rafsi, -pev- and
>-pov- respectively, to allow the creation of figurative lujvo.  A lujvo
>beginning with pev- might have a totally erratic place structure.

The first thing noticably missing from this proposal is that there is no
defined way to carry the polar negations into lujvo.  This can be
remedied using the rafsi "na'e" (natfe=deny), or "dukti" (polar
opposite).  But such should be explicitly mentioned in any teaching of the
principle if adopted.

However, further thought, and the discussion with Nora, suggests that
the status quo would suffice, given John's description, with only a
small broadening of the use of "po'a".  If Nora's basic contention is
true, and Frank's second category is really not truly independent, then
the whole of metaphor can be viewed as a continuum from restriction to
(poetic) expansion.  Examples that Frank or others might place in the
2nd category might be considered as being on the midpoint of that scale.

Thus I counterpropose that "po'a" and "pe'a" retain their current scopes
and basic meaning of "poetic figurative".  However, it would be
explicitly stated that these two words could take the scalar modifiers
in selma'o NAI and CAI (I'm not sure whether this would actually take a
YACC grammar change). tanru and lujvo of the norm (i.e., type 1
restrictions) would probably be unmarked, except when an exceptionally
figurative lujvo has somehow crept into the system, in which case the
modifier "po'anai" could follow the lujvo, or the rafsi-suffix
"povna'e/"povdukti" could be incorporated into the lujvo, to signify
that the "literal", restrictionist sense would apply.

Type 2 "expansions" which are more-or-less borderline cases could be
handled with "po'ano'e" or in lujvo with "povno'e" (since "no'e" is
being assigned "-no'e" as a rafsi in the current change proposal),
signifying a middle position on the restriction/poetic-expansion scale.

Given the UI-like nature of "po'a" which would typically follow AFTER
the tanru or lujvo and apply to the limited structure, I suggest that it
be "po'a", and NOT "pe'a" that gets assigned a rafsi.  "pe'a" as John
explains it will seldom be used on a tanru/lujvo, but rather on a larger
scope of text, since it will normally require a "po'a" to close it.
Given John's explanation and a UI-like "po'a", ".ipo'a", and not
".ipe'a", would be the way, for example, to make an entire
sentence/bridi figurative, "ni'opo'a" would give a non-literal
interpretation to an entire paragraph, and "tu'epo'a" or "tu'upo'a"
would handle a text of longer scope, leaving "pe'a" to be used only for
irregular figurative blocking that is independent of the normal
syntactic structures.  "po'a" will thus be the far more-used of the two
cmavo.

The obvious question, John, is whether you see "po'a" as being UI-like
in this extreme a sense.  I hadn't thought about it this way until
writing this.

The use of "pe'ano'e", and "pe'anai" becomes a bit strange under this
analysis, possibly meaningless.  What is a non-poetic expansion of a
large irregular block of text?  Why would one mark a explicitly
restrictionist or literal variety of such a text?  Allowing NAI and CAI
on "po'a" only may simplify any implementation.  It would have to be
decided, though whether a "po'ano'e" or a "po'anai" would close a
"pe'a".  I would rather see the modifiers as forcing a UI-like
interpretation independent of the "pe'a"-closing function (and lujvo
varieties definitely do not close "pe'a"), since this would allow a
local "po'anai" inside a "pe'a/po'a" bracketing to serve as a local
'non-figurative' exception to a generally (and probably vaguely)
figurative larger body of text.  Sort of "I know you aren't taking any
of this in its literal sense as a whole, but this one
tanru/lujvo/sumti/sentence really can be taken at face value."

The one problem with this is that bare "po'a" will usually appear AFTER
the tanru/lujvo, and one would prefer that the lujvo form similarly
appear as a suffix.  This would not be possible with "pov", although the
modified forms "povnai" and "povno'e" work fine.  The only available
rafsi for final position even remotely close to either "po'a" or "pe'a"
is "pe'e".  This could be assigned in addition or instead of "pov",
giving suffix "pe'e" or (optionally, if both are used) "pe'enai" and
"pe'eno'e" (I think I would be inclined to assign both, since the latter
adds an unseemly 4 syllables and would probably as a result never be
used).  A somewhat more extreme change would be to simply move cmavo
"po'a" to "pe'e" (and selma'o POhA to PEhE), still a recognizable
pairing to close "pe'a".  Alternatively, within lujvo, type 3 figurative
lujvo would be left-marked with "pev", consistent with "pe'a", while
"restrictions" and "expansions" would be marked with "povnai" and
"povno'e".  Not symmetrical, but consistent with the grammar of "pe'a"
and "po'a".

For the short term, while we try to baseline the rafsi, I therefore
propose dropping "pef" for "pe'a" in favor of "pov" for "po'a", and
possibly adding "pev" for "pe'a").  We can also consider adding "pe'e"
as well for "po'a", and I will leave it to the other commenters whether
this addition, and/or changing "po'a" to "pe'e" (along with the
corresponding selma'o) would be worthwhile.  This set or subset of
changes allows the rafsi list to be baselined, (and makes a minimal
change to the grammar if the selma'o name is changed), while allowing my
counter proposal, or possibly some form of John's version, to be tried
experimentally.  If either or both proposals turn out to be invalid, or
simply too 'baroque' for real usage, the default turns out to be the
status quo (which I'll admit is already baroque enough that it has only
been used by our baroque poet Helsem).

But after all, if it ain't baroque, don't fix it zo'o.iunai

lojbab