Lesson 21
repamoi seltadni

Pro-sumti
sumti cmavo


We know by now that there are several kinds of pro-sumti.  In this
sublesson, we will explore and summarize ways of referring to sumti.
We will indicate how to subscript pro-sumti, a technique that we will
later learn can be applied to many other grammatical constructs.  We
will also discuss in some depth how the various forms of first person
and second person pro-sumti are defined and used.  Finally, we will
explore the implications of one technique for defining pro-sumti, the
relativized space scalars, as they apply to the system of tense and
location.


21.1
First and Second Person
cusku je tercusku cmavo

Let us start with some simple pro-sumti: "mi" and "do", the first two
that we learned.  While simple in concept and expression, these really
aren't that simple in usage.  One of the few things that can be said
is that "mi" and "do" are mutually exclusive; once you have defined
one of them, the other is restricted from including any overlapping
individuals (unless you in some way clearly indicate that you are
talking to yourself).

Of the two, "do" is the easier to define: "do" is 'lo te cusku', the
person or persons to whom the speaker/writer/expressor is directing
the expression.  This need not necessarily be limited to those who
directly hear or read the expression.  As an example we will return to
below, someone who gives an interview to the press may express
something intended for the readers of the publication (who are not
present).

"mi" always includes the expressor.  Of course, there is still the
need to determine who else might be included; "mi" is not necessarily
limited to just the expressor.  "mi" can be translated either to
"I/me" or to "We/us".  Those English pronouns, however, are quite
ambiguous.  Who do they refer to?  In Lojban, we have defined
conventions that can aid in interpretation.  We also have ways that an
expressor can explicitly define his/her intent.

The obvious way to define "mi" is with a restrictive relative phrase:
"mi po'u la [cmene]" or "mi po'u lai [cmene]".  In later lessons, we
will see how to enumerate sets or lists of sumti values to fill in the
relative phrase when several specific people are to be included in
"mi".

We can also use descriptions with "po'u" just as well, if the referent
of the description is sufficiently well-described to be known to the
listener:

mi po'u ro lo prenu poi xabju la .iunaitydsteits.

which is similar to the translation used in Lesson 5 (which used
"... poi prenu ..."), but is more specific - it indicates that "mi"
includes ALL of the people of the United States.

"do" can be similarly defined using a relative clause.  "do" may also
be defined, even without using the word "do", when any vocative
(selma'o COI or DOI) is used, except "mi'e"; "mi'e" is used, of
course, to define "mi".

"mi" and "do", once defined, retain the same definition until
redefined.  That redefinition requires a restrictive relative clause.
There is a way, covered in a later lesson, of 'freeing' all
definitions of pro-sumti with a single word, but there is no way to
limit such effects only to "mi" or any other individual pro-sumti.

As we said above, "mi" doesn't generally include "do" ("you").  This
must be strongly qualified, however.  "mi" doesn't include "do"
provided that any of the words "do", "mi'o", "ko", or "do'o" or "ma'a"
(which we will define in a moment) have been defined or used in the
context of the discussion.

If "do" has been defined or used, you must use "mi'o" ("me/we and
you") to include "do" in a translation of "we".  "mi'o" implicitly
includes a reference to "do", so that after a reference to "mi'o", a
reference to "mi" is clearly seen to not include "do".

"ma'a", which means "me/we and you and others unspecified", similarly
includes a reference to "do"; its use makes a later reference to "mi"
a contrasting exclusion of "do".  "do'o" denotes "you and others".

In addition to these, "mi'a" is the remaining pro-sumti that is part
of the first/second person set.  It is symmetric to "do'o", and
denotes "me and others, explicitly excluding anyone within 'do'."
"mi'a" and "do" are mutually exclusive, like "mi" and "do".
Similarly, "do'o" and "mi" are mutually exclusive.

There will be examples of each of these following this section.

We might want to use "mi" to include "do" (or at least to not
necessarily exclude "do") whenever the expressor is expressing to an
unknown or undefined audience.  The English version of the preamble to
the U.S. Constitution is such an example.  When the writers used "We
the people", they did not know whether the readers would be among 'the
people' or not.  So when they used "we", they certainly weren't
automatically excluding the reader, nor were they automatically
including a reader who wasn't "of the United States".

Returning to "do", this pro-sumti is defined symmetrically with "mi".
"do" doesn't include the speaker/expressor, nor does it include anyone
previously defined to be included in "mi" (but see the note next
paragraph).  "do" always includes the referent of a vocative; a new
vocative redefines "do".  "do" can also be defined with a restrictive
clause, and the definition can be changed with a new restrictive
clause.

(It is possible to 'talk to yourself': "doi mi", or to include "mi" in
a restrictive definition of "do" in which case you are equating or
incorporating the definition of "mi" into "do".  You can also add "do"
into "mi" using a restrictive clause definition.  These are specific
cases of using one convention to override another.  When dealing with
semantics - as opposed to grammar - Lojban users always try to find a
plausible interpretation of what might otherwise be nonsense.)

"mi'o" and "ma'a" are always dependent on the definition of "mi" and
"do".  They amount to combining or mixing the two together (and adding
extras, in the case of "ma'a").  If you change the definition of
either "mi" or "do", then a later reference to "mi'o" or "ma'a" may
reference a different set than such a reference before the
redefinition.  Similarly, "do'o" and "mi'a" are dependent on the
definitions of "do" and "mi", respectively.  "ko" is, of course,
defined identically to whatever "do" is at the moment.

Clearly, first and second person pro-sumti are vague (not 'ambiguous')
when their referent is not restricted in a relative clause or a
vocative.  This is in keeping with the normal practice for Lojban: you
can be intentionally vague without being ambiguous, and you can be
arbitrarily specific when this is necessary for communication.
Remember, however, that communications is a partnership between
speaker and listener; while the speaker can be intentionally (or
unintentionally) vague, the listener is entitled to use her/his
standards of plausibility to interpret the reference.  Normally, this
will mean that a vague "mi" refers to the actual speaker(s) only, and
a vague "do" refers to those who actually listen to or read the
expression.


21.2
Example
mupli

Try to read this on your own.  Use the translation only if you must.

lai KEpiulet. ba penmi lai mantygius. le sanmi pe la djuliet. KEpiulet. mantygius.

djuliet:	lemi lanzu ba klama ca le lamji djedi
romios:	lemi lanzu ba klama ba lenu ledo lanzu ba klama
djuliet:	mi'o ba jukpa ma pe ba le sanmi
romios:	mi'a nelci loi grute  .i xu do'o nelci loi grute
djuliet: mi'a milxe nelci loi grute .i mi'a mutce nelci loi titla
                nanba
romios:	mi'a nelci loi titla nanba  .i.ai mi'o jukpa lo grute ke titla nanba
djuliet:	.ie.ui ma'a ba gleki citka lo grute ke titla nanba ba le sanmi


21.3
Translation Of The Example
mupli xe fanva

lai KEpiulet. ba penmi lai mantygius. le sanmi pe la
djuliet. KEpiulet. mantygius.  Capulets will meet Montagues at the
meal pertaining to Juliet Capulet Montague.

djuliet:	lemi lanzu ba klama ca le lamji djedi
	My family will come on the adjacent day (tomorrow).
romios:	lemi lanzu ba klama ba lenu ledo lanzu ba klama
	My family will come after the event of your family coming.
djuliet:	mi'o ba jukpa ma pe ba le sanmi
	We will cook what? pertaining to after the meal?
romios:	mi'a nelci loi grute  .i xu do'o nelci loi grute
	Me and others unspecified excluding you are fond of fruit.  
Is it true that you-and-others-unspecified (that aren't part of
"mi'a") are fond of fruit?.  djuliet: mi'a milxe nelci loi grute .i
mi'a mutce nelci loi titla nanba
	Me and others unspecified excluding "do'o" are mildly fond of
fruit.  We are very fond of cake.  romios: mi'a nelci loi titla nanba
.i.ai mi'o jukpa lo grute ke titla nanba
	We (also) are fond of cake.  (Intention!) We cook some fruit-cake.
djuliet:	.ie.ui ma'a ba gleki citka lo grute ke titla nanba ba le sanmi
	(Agree! Happy!) Me-and-you-and-others unspecified will happily
eat some fruit- cake after the meal.


21.4
Quantification of Pro-sumti
sumti cmavo nunkancu

The discussion that follows applies to all pro-sumti, although it has
specific applicability to the questions of defining and using first
and second person pro-sumti.

As stated in Lesson 5, when a description or a pro-sumti refers to an
individual, that individual is taken to be a 'mass'.  If "mi jgari le
cukta" is referring to an individual "mi", then we are not claiming
that every part of the individual is actively grasping the book.  As
with other mass individuals, if part of the mass performs an action,
then the mass as a whole can be said to be performing the action.

What happens when we talk of pro-sumti that are not singular, or that
are non-specific as to singularity or plurality?  The answer is that
such pro-sumti are treated as a mass for default quantification.  The
default quantification value is "piro" ("all").  This means that "do
klama", with a plural "do" means that all of the individuals within
"do" come, not that a portion of the mass of "do" suffices for the
whole.  Note that this differs from the default quantification for
"loi", "lei", and "lai" descriptions, for which "pisu'o" is the
default quantifier before the description.

It is generally assumed that any fractional quantifier on a pro-sumti
similarly is treating the pro- sumti as a mass individual.  If I say
"pimu do bevri le botpi", I do not mean that half of one individual
among you is doing the carrying, but that some half of the mass of
'you' is actively involved in the carrying (with the other half
uninvolved).

Unlike masses that are marked with descriptors, however, we can also
treat pro-sumti as sets of individuals instead of masses.  If we say
"re do", we are treating "do" as a set of individuals, and selecting
two of them, hence "two of you".  Similarly, "pa do" translates as
"one of you" and "ro do" translates as "each of you".  The convention
that results is that, if the quantifier is expressed as a fraction or
portion, then it is a fraction of the mass individual; if it is a
quantity greater than or equal to one, then it is treating the
pro-sumti as a set, and selecting from that set.  Note that if you
select more than one from a set represented by a pro-sumti, you are
implicitly stating that the pro- sumti is plural.

(Warning - quantificational logic is a complex topic; it is not
difficult to devise questions which are outside the scope of this
course.  You will seldom need to express ideas that require resolution
of the 'gray' areas we cannot cover.)

One way to help define a pro-sumti is to quantify it.  This settles
the question of singularity vs.  plurality, among other things.  We
have two ways to specify the number of individuals in a pro- sumti
set/mass without providing a list of the specific members.  One way is
with a restrictive clause:

  "do po'u re da [cu] klama"
"You-who-are-identified-as-two-somethings come."

Another way is to quantify the pro-sumti as a description:

  "le re do ku [cu] klama"
"The two-of-you come."

The English translation is misleading; as with all descriptions where
a quantifier is provided after the descriptor, the "re" in the above
example quantifies the number of individuals in the description - it
does not select two out of the set of 'you'.


21.5
Back-Counting Pro-sumti
ti'ernunkancu ke sumti cmavo

There is one other kind of pro-sumti that we are going to talk about
in this lesson, the 'back- counting' cmavo "ri", "ra", and "ru".
These always refer to some previously defined sumti.  By comparison,
we know that "ko'a" pro-sumti can be used with the definition
occurring after the usage of the pro-sumti.

"ri" is the simplest and easiest to understand.  It back references
the just previous sumti.  This can have several vital uses in 'normal'
speech, which is heavily dominated by afterthoughts:

"le cilce danlu cu melbi .i ri goi ko'a mabru"
"The wild animal is beautiful.  It (which I'll call ko'a) is a mammal."

In this example, the speaker was talking, perhaps, about the animal
he/she saw.  Then, in afterthought, she/he (goi ko'e) realized that
ko'e wanted to say something more about it.  ko'e had not assigned it
to a pro-sumti.  Without "ri", ko'e would have had to repeat the sumti
in full: "ko'a goi le cilce danlu".  If the sumti had been even
longer, this could be a distracting nuisance.

A second use is in 'possessive' expressions such as:

"la .alis. sipna vi leri kumfa"
"Alice sleeps in her room."

In this case, unless we have assigned a "ko'a"-series pro-sumti to "la
.alis." previously, "ri" is the only way to avoid repeating "la
.alis." twice in one sentence:

"la .alis. sipna le la .alis. kumfa"
"la .alis. sipna le kumfa po la .alis."

both of which are quite clumsy and long-winded.

How do we know what "ri" refers to?  We simply look back to the next
previous sumti which is 'significant'.  A sumti is considered
significant if there is some meaningful reason to use "ri" as a
replacement for it.  Simply speaking, nearly all of the pro-sumti are
not significant.  There is minimal advantage to saying:

"ko'a sipna leri kumfa"

instead of:

"ko'a sipna leko'a kumfa"

Similar argument eliminates using "ri" to stand for "mi" and "do" and
their relatives.  "ri" can replace "ti" and its relatives in some
circumstances - those in which for some reason the speaker might
necessarily be pointing to something else:

"ti zmadu ti poi se jbena ri vau loni slabu"
"These are more than those-which-are-born-to-them in age."

The last example demonstrates the other major property of "ri": when
counting back, you cannot refer to a sumti of which the "ri" will be a
part.  In the case given, the "ri" is part of the relative clause
defining the second "ti", the x2 place of "zmadu"; it therefore cannot
refer to the x2 place of "zmadu" as a whole. (You can use "ke'a", for
that purpose, of course.)  Similarly, a reference within a quotation
cannot refer to the sumti that includes the quotation:

le prenu cu cusku lu ri klama li'u

In this example, "ri" cannot refer to the x2 place of "cusku", the
quotation.  In fact, in the case of quotations such as this, you
cannot include anything outside of the 'quotation universe' in back-
counting (such as "le prenu"), since you must back-count from the
point of view of the person quoted, who obviously cannot have been
referring to a sumti not yet expressed.  (You can, on the other hand,
refer from outside a quotation to a sumti on the inside.)

The most important point about "ri" is that it is very ephemeral in
definition.  EVERY TIME IT IS USED, IT MUST BE RE-COUNTED.  "ri" thus
acts more like "ti" than like "ko'a", or "mi" which retain their
definitions indefinitely.  You can chain "ri" references together;
"ri" itself is significant:

"la djos. viska le tricu .i ri se jadni le ri jimca poi jorne ri vi le
sirji" "Joe sees the tree.  It is adorned by its branches which are
joined to it at the straight (the trunk)."

For each of the "ri"s after the first in this mildly stilted example,
counting back points to the previous "ri".

"ra" and "ru" are much simpler than "ri" to use: you needn't count
back.  "ra" is some previous sumti which is nearby.  It perhaps is
even the last one, which could be referred to by "ri".  If "ri" is
used in a sentence, however, "ra" always refers to something before
"ri".  "ru", like "ra" refers to an earlier sumti, but never the last
one, and always before either "ri" or "ra".

The first question that comes to everyone's mind now is "But Lojban is
supposed to be unambiguous.  How would a computer figure out what "ra"
and "ru" refer to?"  The answer is that, without exceptional
plausibility programming, a computer could not.  It would have to ask
"ra ki'a" or "ru goi ma".  Lojban is designed for humans, and only
incidentally is it valuable for computers.  This type of usage is
something humans would use, because we are lazy and don't want to
bother defining pro-sumti every time we want to use them in
afterthought (which is when we would use "ra" and "ru".)  We are
dealing with a case of vagueness, semantic ambiguity, and not
grammatical ambiguity.  Just as tanru are vague and semantically
ambiguous, but can be clarified on request, so are back-counted
pro-sumti.

As with all such 'convenience features', the speaker must be fair to
the listener in the use of back- counting.  In extremely convoluted
sentence structures, it can be nearly impossible to figure out exactly
which is the last significant sumti - remember that the listener isn't
aware that the speaker is about to spring one of the back-counting
pro-sumti on ru until ra actually does so.  As the last sentence
shows, use of "ra" and "ru" are somewhat easier to interpret than "ri"
would be.  Try having someone read the sentence aloud to you, and you
can verify this.

"ra" and "ru", by not being exactly back-counted, allow the speaker to
retain their definitions for short periods of time.  "ra" at the
beginning of a sentence could reasonably refer to the same thing as
"ra" at the end of the sentence, and "ru" assignments can reasonably
last even longer.  "ri", on the other hand, must be re-figured every
time it is used, and thus is primarily used only to chain together
immediate afterthoughts long enough to finish them of or to assigned
them to a "ko'a"- series variable.  The example at the end of the next
section demonstrates some of these considerations.


21.6
On xi
me zo xi

You can, it turns out, back-count exactly to some sumti prior to the
first one.  You reference such additional back-counted sumti by
appending a subscript onto "ri".  A subscript is a quantifier which is
appended using the cmavo "xi".  There are several possible numbering
conventions, as we'll see in a moment; the simplest is to assume that
the first back-counted sumti is "ri" and the second is "ri xi re"
("ri2"), the third is "rixici", etc.  The only easily translatable
examples of this usage are when you want to talk of "the former" and
"the latter" or similar references, and for some reason you want to
avoid using "ra" (either because it would be too vague, or because it
is already in use for another purpose).  Creating an example:

le gusta prenu cu bevri lo palta .i lo dakfu .i lo smuci .i lo forca
.i la rik. pilno rixire .i la .alis. pilno riximu "... Rick uses it (a
fork).  Alice uses it (a spoon)."

Do you get the same references for the two back-counted sumti as we
did?  As you can see, the back-counting gets large in a hurry.  If
someone reads this sentence to you, you are not easily going to be
able to count the sumti in your head, especially if the speaker is
continuing while you are trying to count.  On the other hand, the
speaker cannot use "ra" or "ru" meaningfully - there are too many
possible referents in the example given.

In natural speech, we suspect that people will simply repeat the sumti
rather than try to use back- counting.  In writing, where the reader
can look back to count (as can the writer), subscripted back- counting
pro-sumti may be useful (if only rarely).  One thing is sure - if
there hadn't been a way to do this designed into the language, someone
would have quickly come up with an example that needed it.

Subscripting is relatively new to the language, and has never been
extensively used.  As such, the numbering conventions are only
tentative, and you can feel free to try others, provided that your
listener is aware that you are doing so.  Try playing with the
following moderately complex example:

"la djan. ckire lenu la djordj. troci lenu ri klama ti ta kei lenu
cadzu .i rixixo ..."  "John is grateful for the event of George trying
to come to here from there by walking.  It? ..."

where "xo" is used to ask what subscript should be used to reference a
given sumti.  The following is the 'standard' method of back-counting
the sumti from the indicated point:

ri	lenu cadzu
rixire	ta
rixici	ti
rixivo	ri(la djorj.)
riximu	lenu ri (la djorj.) klama ti ta
rixixa	la djorj.
rixize	lenu la djorj. troci lenu ri (la djorj.) klama ti ta lenu cadzu
rixibi la djan.

Can you come up with another counting method that is more useful?  (If
so, let la lojbangirz.  know.)  As one help, studies have shown that
back references are almost always either to the very last sumti or to
the first sumti of various previous sentences.  This could suggest
some other counting approach.

The subscript that is attached can be any quantifier.  Quantifiers
need not be ordinary numbers: "ru" might be considered equivalent to
"rixiso'i" ("rimany"), and "ra" is usually equivalent to "rixisu'ore"
("ri>=2"), though it can be used to access "rixipa".

Subscripts have other uses besides in back-counting.  If you run out
of free variables in the "ko'a" or "fo'a" series, you can start with
"ko'axire".  Similarly, you can use "daxire" and other subscripted
bound variables when you need more than the three of those provided in
the Lojban cmavo.  In order to find out a complete list of words
and/or structures that can be subscripted within Lojban, see a
reference list of cmavo (on "xi" subscripting), or examine the formal
grammar definition.

21.7
Back-Counting Example
ti'ernunkancu mupli

la djeryldin. xolmz. meisn. cu facki lo melbi ke cizra spati .i ri
xunre .i ri plana .i ri xabju le caxno vi le lalxu .i ra citka loi
sfani .i ra citka la meris. .elen. mai,rs.  .i ri pu penmi ra goi
rixici ca lenu cadzu va le lalxu .i ru xenru lenu ri pu facki le spati

Geraldine Holmes Mason discovers a beautiful type of strange plant.
It (the plant) is red.  It (still the plant) is fat.  It (still the
plant) inhabits the shallows in the lake.  It (the plant - almost
certainly not Geraldine, the shallows, or the lake) eats flies.  It
(probably still the plant) eats Mary Ellen Meyers.  She (Mary Ellen)
met it (which is back-counted to the last "ra" that we are assuming is
the plant.) during the walking near the lake.  She (something earlier
than "ra" and thus necessarily Geraldine) regrets the event of she
(again Geraldine) discovering the plant (At this point, "ra" is fairly
vague - there is more than one plausible thing that Geraldine could
regret discovering, such as Mary Ellen - if anything is left of her -
yecch!), so we explicitly repeat the description of the plant.

The usage of back-counting pro-sumti here is reasonable for a written
story in a fairly informal mood.  It probably is understandable even
told orally, due to the small number of different sumti.  Following
are the ways it would have to be told if (1) we used "ko'a"-series
variables, but only in afterthought since the mood is informal; (2) we
use no back-counting cmavo.

(1) la djeryldin. xolmz. meisn. cu facki lo melbi ke cizra spati .i
ko'a goi lo melbi ke cizra spati cu xunre .i ko'a plana .i ko'a xabju
le caxno vi le lalxu .i ko'a citka loi sfani .i ko'a citka la
meris. .elen. mai,rs.  .i la meris. .elen. mai,rs. pu penmi ko'a ca
lenu cadzu va le lalxu .i ko'e goi la djeryldin. xolmz. meisn. xenru
lenu ko'e pu facki le spati

(2) la djeryldin. xolmz. meisn. cu facki lo melbi ke cizra spati .i le
spati cu xunre .i le spati cu plana .i le spati cu xabju le caxno vi
le lalxu .i le spati cu citka loi sfani .i le spati cu citka la
meris. .elen. mai,rs.  .i la meris. .elen. mai,rs. pu penmi le spati
ca lenu cadzu va le lalxu .i la djeryldin. xolmz. meisn. xenru lenu la
djeryldin. xolmz. meisn. pu facki le spati

I'm sure that you can see that these two varieties are quite stilted,
each in its own way.  Neither sounds quite as relaxed (as human?) as
the original version.