Place structure paper, version 1.2

[cowan@snark.thyrsus.com (John Cowan)]

"To Boston Via The Road Go I": Manipulating Selbri Place Structures
$Revision: 1.2 $


1.  Introductory

This paper describes the various ways in which the place structure of
Lojban bridi is expressed and by which it can be manipulated.  The place
structure of a selbri is a sequence of empty slots into which the sumti
associated with that selbri are placed.  Sumti are said to occupy the
places of a selbri.

For our present purposes, every selbri is assumed to have a natural place
structure.  If the selbri is a brivla, the place structure can be looked
up in a dictionary (or, if the brivla is a lujvo, inferred from the
principles of lujvo construction); if the selbri is a tanru, the place
structure is always the same as that of the final component in the tanru.

As an example of a place structure, we will use the place structure of
the gismu "klama":

	x1 comes/goes to destination x2 from origin x3 via route x4
		employing means of transport x5.

The "x1...x5" indicates that "klama" is a five-place predicate, and show
the natural order (as assigned by the language engineers) of those
places:  agent, destination, origin, route, means.

The tables of cmavo at the beginning of various sections of this paper
have a common form.  There are three columns: a cmavo, the selma'o to
which the cmavo belongs, and a rough English gloss (not necessarily a
full translation.)


2. Standard bridi Form

The most usual way of constructing a bridi from a selbri such as "klama"
and an appropriate number of sumti is to place the sumti intended for the
x1 place before the selbri, and all the other sumti in order after the
selbri, thus:

2.1)	mi cu klama la bastn. la .atlantas. le dargu le karce
	I go to-Boston from-Atlanta via-the road using-the car.

Here the sumti are assigned to the places as follows:

		x1	agent		mi
		x2	destination	la bastn.
		x3	origin		la .atlantas.
		x4	route		le dargu
		x5	means		le karce

(Unless otherwise mentioned, the examples in the rest of the paper will
have the same meaning as Example 2.1; this fact will not be reiterated.)

This order is called "standard bridi form", and is found in the bulk of
Lojban bridi, whether used in main sentences or in subordinate clauses.
However, many other forms are possible, such as:

2.2)	mi la bastn. la .atlantas. le dargu le karce cu klama
	I to-Boston from-Atlanta via-the road using-the car, go.

Here the selbri is at the end; all the sumti are placed before it.  However,
the same order is maintained.

Similarly, we may split up the sumti, putting some before the selbri and
others after it:

2.3)	mi la bastn. cu klama la .atlantas. le dargu le karce
	I to-Boston go from-Atlanta via-the road using-the car.

All of the variant forms in this section and following sections can be
used to place emphasis on the part or parts which have been moved out
of their standard places.  Thus, Example 2.2 places emphasis on the
selbri (because it is at the end); Example 2.3 emphasizes "la bastn.",
because it has been moved before the selbri.

The general rule, then, is that the selbri may occur anywhere in the
bridi as long as the sumti maintain their order.  The only exception
(and it is an important one) is that if the selbri appears first, the
x1 sumti is taken to have been omitted:

2.4)	klama la bastn. la .atlantas. le dargu le karce
	A-goer to-Boston from-Atlanta via-the road using-the car.

Here the x1 place is empty: the listener must guess from context who is
going to Boston.  In Example 2.4, "klama" is glossed "a goer" rather than
"go" because "Go" at the beginning of an English sentence would suggest a
command: "Go to Boston!".  Example 2.4 is not a command, simply a normal
statement with the x1 place unspecified, causing the emphasis to fall on the
selbri "klama".  Such a bridi, with empty x1, is called an "observative",
because it calls on the listener to observe something in the environment
which would belong in the x1 place.

There is a way to both provide a sumti for the x1 place and place the
selbri first in the bridi: see Example 3.7.

Suppose the speaker desires to omit a place other than the x1 place?
(Presumably it is obvious or, for one reason or another, not worth
saying.)  Places at the end may simply be dropped:

2.5)	mi klama la bastn. la .atlantas.
	I go to-Boston from-Atlanta.

Example 2.5 has empty x4 and x5 places: the speaker does not specify the
route or the means of transport.  However, simple omission will not work
for places whose successors are to be specified: in

2.6)	mi klama la bastn. la .atlantas. le karce
	I go to-Boston from-Atlanta via-the car.

"le karce" occupies the x4 place, and therefore Example 2.6 means:

	I go to Boston from Atlanta, using the car as a route.

This is nonsense, since a car cannot be a route.  What the speaker
presumably meant is expressed by:

2.7)	mi klama la bastn. la .atlantas. zo'e le karce.
	I go to-Boston from-Atlanta via-something-unspecified using-the car.

Here the sumti cmavo "zo'e" is used to explicitly fill the x4 place;
"zo'e" means "the unspecified thing" and has the same meaning as leaving
the place empty: the listener must infer the correct meaning from context.


3. Tagging Places: FA

		fa	FA	tags x1 place
		fe	FA	tags x2 place
		fi	FA	tags x3 place
		fo	FA	tags x4 place
		fu	FA	tags x5 place
		fi'a	FA	place structure question
                                                                                                                                      
In sentences like Example 2.1, it is easy to get lost and forget which
sumti falls in which place, especially if the sumti are more complicated
than simple names or descriptions.  The place structure tags of selma'o
FA may be used to help clarify place structures.  The five cmavo "fa", "fe",
"fi", "fo", and "fu" may be inserted just before the sumti in the x1 to
x5 places respectively:

3.1)	fa mi cu klama fe la bastn. fi la .atlantas. fo le dargu fu le karce
	x1= I go x2= Boston x3= Atlanta x4= the road x5= the car.
	I go to Boston from Atlanta via the road using the car.

In Example 3.1, the tag "fu" before "le karce" clarifies that "le karce"
occupies the x5 place of "klama".  The use of "fu" tells us nothing about
the purpose or meaning of the x5 place; it simply says that "le karce"
occupies it and no other.

In Example 3.1, the tags are overkill; they serve only to make Example 2.1
even longer than it is.  Here is a better illustration of the use of FA tags
for clarification:

3.2)	fa mi klama fe le zdani be mi poi nurma vau fi la nuIORK.
	x1= I go x2= the house of me which is-rural x3= New York.

In Example 3.2, the place structure of "klama" is as follows:

		x1	agent		mi
		x2	destination	le zdani be mi poi nurma vau
		x3	origin		la nuIORK.
		x4	route		(empty)
		x5	means		(empty)

The "fi" tag serves to remind the hearer that what follows is in the x3
place of "klama" and not part of the complex sumti occupying the x2 place.

Of course, once the sumti have been tagged, the order in which they are
specified no longer carries the burden of distinguishing the places.
Therefore, it is perfectly all right to scramble them into any order
desired, and to move the selbri to anywhere in the bridi, even the
beginning:

3.3)	klama fa mi fi la .atlantas. fu le karce fe la bastn. fo le dargu
	go x1= I x3= Atlanta x5= the car x2= Boston x4= the road.
	Go I from Atlanta using the car to Boston via the road.

3.4)	fu le karce fo le dargu fi la .atlantas. fe la bastn. cu klama fa mi
	x5= the car x4= the road x3= Atlanta x2= Boston go x1=I
	Using the car, via the road, from Atlanta to Boston go I.

Example 3.4 exhibits the reverse of the standard bridi form seen in
Examples 2.1 and 3.1, but still means exactly the same thing.  If the
FA tags were left out, however, producing:

3.5)	le karce le dargu la .atlantas. la bastn. cu klama mi
	The car to-the road from-Atlanta via-Boston goes using-me.
	The car goes to the road from Atlanta, with Boston as the route,
		using me as a means of transport.

the meaning would be wholly changed, and in fact nonsensical.

Tagging places with FA cmavo makes it easy not only to reorder the places
but also to omit undesirable ones, without any need for "zo'e" or special
rules about the x1 place:

3.6)	klama fi la .atlantas. fe la bastn. fu le karce
	A-goer x3 = Atlanta x2= Boston x5 = the car.
	A goer from Atlanta to Boston using the car.

Here the x1 and x4 places are empty, and so no sumti are tagged with "fa"
or "fo"; in addition, the x2 and x3 places appear in reverse order.

What if some sumti have FA tags and others do not?  The rule is that after
a tagged sumti, any following untagged sumti occupy the succeeding places,
subject to the proviso that an already-filled place is skipped:

3.7)	klama fa mi la bastn. la .atlantas. le dargu le karce
	Go x1= I x2= Boston x3= Atlanta x4= the road x5= the car.
	Go I to Boston from Atlanta via the road using the car.

In Example 3.7, the "fa" causes "mi" to occupy the x1 place, and then
the following untagged sumti occupy in order the x2 through x5 places.
This is the mechanism by which Lojban allows placing the selbri first
while specifying a sumti for the x1 place.

Here is a more complex (and more confusing) example:

3.8)	mi klama fi la .atlantas. le dargu fe la bastn. le karce
	I go x3= Atlanta the road x2= Boston the car.
	I go from Atlanta via the road to Boston using the car.

In Example 3.8, "mi" occupies the x1 place because it is the first sumti
in the sentence (and is before the selbri).  The second sumti, "la .atlantas.",
occupies the x3 place by virtue of the tag "fi", and "le dargu" occupies
the x4 place as a result of following "la .atlantas.".  Finally, "la bastn."
occupies the x2 place because of its tag "fe", and "le karce" skips over
the already-occupied x3 and x4 places to land in the x5 place.  Such a
convoluted use of tags should probably be avoided except when trying for
a literal translation of some English (or other natural-language) sentence;
the rules stated here are merely to make some interpretation possible.

It is grammatically permitted to tag more than one sumti with the same
FA cmavo.  The effect is that of making more than one claim:

3.9)	[fa] la rik. fa la djein. klama
		[fe] le skina fe le zdani fe le zarci
	[x1=] Rick x1= Jane goes-to
		x2= the movie x2= the house x2= the office

may be taken to say that both Rick and Jane go to the movie, the house, and
the office, merging six claims into one.  More likely, however, it will
simply confuse the listener.  There are better ways, involving logical
connectives (explained elsewhere), to say such things in Lojban.

The cmavo "fi'a" also belongs to selma'o FA, and allows Lojban users to
ask questions about place structures.  A bridi containing "fi'a" is a
question, asking the listener to supply the appropriate other member of
FA which will make the bridi a true statement:

3.10)	fi'a do dunda fe le vi rozgu
	[what place]? you give x2= the nearby rose
	Are you the giver or the receiver of this rose?

In Example 3.10, the speaker uses the selbri "dunda", whose place structure
is:

	x1 gives x2 to x3

The tagged sumti "fi'a do" indicates that the speaker wishes to know whether
the sumti "do" falls in the x1 or the x3 place (the x2 place is already
occupied by "le rozgu").  The listener can reply with a sentence consisting
solely of a FA cmavo:  "fa" if the listener is the giver, "fi" if he/she
is the receiver.

There is also another member of FA, namely "fai", which is discussed in
Section 12.


4. Conversion: SE

		se	SE	2nd place conversion
		te	SE	3rd place conversion
		ve	SE	4th place conversion
		xe	SE	5th place conversion

So far we have seen ways to move sumti around within a bridi, but
the actual place structure of the selbri has always remained untouched.
The conversion cmavo of selma'o SE are incorporated within the selbri
itself, and produce a new selbri (called a converted selbri) with a
different place structure.  In particular, after the application of any SE
cmavo, the number and purposes of the places remain the same, but two of
them have been exchanged, the x1 place and another.  Which place has been
exchanged with x1 depends on the cmavo chosen.

Note that the cmavo of SE begin with consecutive consonants in alphabetical
order.  There is no "1st place conversion" cmavo, because exchanging the x1
place with itself is a pointless maneuver.

Here are the place structures of "se klama":

	x1 is the destination of x2's going from x3 via x4 using x5

and "te klama":

	x1 is the origin of x2's going to x3 via x4 using x5

and "ve klama":

	x1 is the route of x2's going to x3 from x4 via x5

and "xe klama":

	x1 is the means used by x2 in going to x3 from x4 via x5

Note that the place structure numbers in each case continue to be listed
in the usual order, x1 to x5.

Consider the following pair of examples:

4.1)	la bastn. cu se klama mi
	Boston is-the-destination of-me.
	Boston is my destination.

4.2)	fe la bastn. cu klama fa mi
	x2 = Boston go x1=I.
	To Boston go I.

Examples 4.1 and 4.2 mean the same thing, in the sense that there is a
relationship of going with the speaker as the agent and Boston as the
destination (and with unspecified origin, route, and means).  Structurally,
however, they are quite different.  Example 4.1 has "la bastn." in the
x1 place and "mi" in the x2 place of the selbri "se klama", and uses
standard bridi order; Example 4.2 has "mi" in the x1 place and "la bastn."
in the x2 place of the selbri "klama", and uses a non-standard order.

The most important use of conversion is in the construction of descriptions.
A description is a sumti which begins with a cmavo of selma'o LA or LE,
called the descriptor, and contains (in the simplest case) a selbri.  We
have already seen the descriptions "le dargu" and "le karce".  To this
we could add:

4.3)	le klama
	the go-er, the one who goes

In every case, the description is about something which fits into the x1
place of the selbri.  In order to get a description of a destination (that
is, something fitting the x2 place of "klama"), we must convert the selbri
to "se klama", whose x1 place is a destination.  The result is

4.4)	le se klama
	the destination gone to by someone

Likewise, we can create three more converted descriptions:

4.5)	le te klama
	the origin of someone's going

4.6)	le ve klama
	the route of someone's going

4.7)	le xe klama
	the means by which someone goes

Example 4.6 does not mean "the route" plain and simple: that is "le pluta",
using a different selbri.  It means a route that is used by someone for
an act of "klama"; that is, a journey with origin and destination.

When converting selbri that are more complex than a single brivla, it is
important to realize that the scope of a SE cmavo is only the following
brivla (or equivalent unit).  In order to convert a tanru, it is necessary
to enclose the tanru in "ke...ke'e" brackets:

4.8)	mi se ke blanu zdani [ke'e] ti
	I [2nd conversion] blue house this-thing

The place structure of "blanu zdani" (blue house) is the same as that
of "zdani", by the rule given in Section 1; the place structure of "zdani"
is:

	x1 is a house/nest/lair/den for inhabitant x2

The place structure of "se zdani" is therefore:

	x1 is the inhabitant of house (etc.) x2

Consequently, Example 4.7 means:

	I am the inhabitant of the blue house which is this thing.

Conversion applied to only part of a tanru has subtler effects which are
explained elsewhere.

It is grammatical to convert a selbri more than once with SE; later (inner)
conversions are applied before earlier (outer) ones.  For example, the
place structure of "se te klama" is achieved by exchanging the x1 and
x2 place of "te klama", producing:

	x1 goes from x2 to x3 via x4 using x5

On the other hand, "te se klama" has a place structure derived from swapping
the x1 and x3 places of "se klama":

	x1 is the origin of x2's going to x3 via x4 using x5

which is quite different.  However, multiple conversions like this are
never necessary.  Arbitrary scrambling of places can be achieved more
easily and far more intelligibly with FA tags, and only a single 
conversion is ever needed in a description.

(Although no one has made any real use of it, it is perhaps worth noting
that compound conversions of the form "setese", where the first and third
cmavo are the same, effectively swap the two places mentioned while leaving
the others, including x1, alone:  "setese" (or equivalently "tesete") swap
the x2 and x3 places, whereas "texete" (or "xetexe") swap the x3 and x5
places.)


5. Modal Places: FIhO, FEhU

		fi'o	FIhO	modal place prefix
		fe'u	FEhU	modal terminator

Sometimes the place structures engineered into Lojban are inadequate to
meet the needs of actual speech.  Consider the gismu "viska", whose
place structure is:

	x1 sees x2 under conditions x3

Seeing is a threefold relationship, involving an agent (le viska), an
object of sight (le se viska), and an environment that makes seeing
possible (le te viska).  Seeing is done with one or more eyes, of course;
in general, the eyes belong to the entity in the x1 place.

Suppose, however, that you are blind in one eye and are talking to someone
who doesn't know that.  You might want to say, "I see you with the left
eye."  There is no place in the place structure of "viska" such as "with
eye x4" or the like.  Lojban allows you to solve the problem by adding
a new place, changing the relationship:

5.1)	mi viska do fi'o kanla [fe'u] le zunle
	I see you [modal] eye: the left-thing
	I see you with the left eye.

The three-place relation "viska" has now acquired a fourth place specifying
the eye used for seeing.  The cmavo "fi'o" (of selma'o FIhO) followed by
a selbri, in this case the gismu "kanla", is prefixed to the sumti which
fills the new place, "le zunle".  The semantics of "fi'o kanla le zunle"
is that "le zunle" fills the x1 place of "kanla", whose place structure is

	x1 is an/the eye of body x2

The x2 place of "kanla" is unspecified and must be inferred from the
context.  It is important to remember that even though "le zunle" is
placed following "fi'o kanla", semantically it belongs in the x1 place.
The selbri may be terminated with "fe'u" (of selma'o FEhU), an elidable
terminator which is rarely required.

The term for such an added place is a "modal place", as distinguished
from the regular numbered places.  (This use of "modal" is particular to
the Loglan Project.)  The "fi'o" construction marking a modal place is called
a "modal tag", and the sumti which follows it a "modal sumti".  Modal sumti
may be placed anywhere within the bridi, in any order; they have no effect
whatever on the rules for assigning unmarked bridi to numbered places, and
they may not be marked with FA cmavo.

Consider Example 5.1 again.  Another way to view the situation is to
consider the speaker's left eye as a tool, a tool for seeing.  The
relevant selbri then becomes "pilno", whose place structure is

	x1 uses x2 as a tool

and we can rewrite Example 5.1 as

5.2)	mi viska do fi'o se pilno le zunle kanla
	I see you [modal] [conversion] use: the left eye
	I see you using my left eye.

Here the selbri belonging to the modal is "se pilno".  The conversion
of "pilno" is necessary in order to get the "tool" place into x1, since
only x1 can be the modal sumti.  The "tool user" place is the x2 of
"se pilno" (because it is the x1 of "pilno") and remains unspecified.


6. Modal Tags: BAI

There are certain selbri which seem particularly useful in constructing
modal tags.  In particular, "pilno" is one of them.  The place structure
of "pilno" is:

	x1 uses tool x2

and almost any selbri which represents an action may need to specify a tool.
Having to say "fi'o se pilno" frequently would make many Lojban sentences
unnecessarily verbose and clunky, so an abbreviation is provided in the
language design: the compound cmavo "sepi'o".

Here "se" is used before a cmavo, namely "pi'o", rather than before a brivla.
The meaning of this cmavo, which belongs to selma'o BAI, is exactly the same
as that of "fi'o pilno fe'u".  Since what we want is a tag based on
"se pilno" rather than "pilno" -- the tool, not the tool user -- the grammar
allows a BAI cmavo to be converted using a SE cmavo.  Example 5.2 may
therefore be rewritten as:

6.1)	mi viska do sepi'o le zunle kanla
	I see you with-tool: the left eye
	I see you using my left eye.

The compound cmavo "sepi'o" is much shorter than "fi'o se pilno [fe'u]"
and can be thought of as a single word meaning "with-tool".  The modal
tag "pi'o", with no "se", similarly means "with-tool-user", probably a
less useful concept.  Nevertheless, the parallelism with the place
structure of "pilno" makes the additional syllable worthwhile.  Some
BAI cmavo make sense with as well as without a SE cmavo; for example,
"ka'a", the BAI corresponding to the gismu "klama", has five useful
forms corresponding to the five places of "klama" respectively:

		ka'a		with-goer
		seka'a		with-destination
		teka'a		with-origin
		veka'a		with-route
		xeka'a		with-means-of-transport

Any of these may be used to provide modal places for bridi, as in the
following examples:

6.2)	la .eivn. cu vecnu loi flira cinta ka'a mi
	Avon sells a-mass-of face paint with-goer me.
	I am a traveling cosmetics salesperson for Avon.

(Example 6.2 may seem a bit strained, but it illustrates the way in which an
existing selbri, "vecnu" in this case, may have a place added to it which
might otherwise seem utterly unrelated.)

6.3)	mi cadzu seka'a la bratfyd.
	I walk with-destination Bradford.
	I am walking to Bradford.

6.4)	bloti teka'a la nuIORK.
	[Observative:] is-a-boat with-origin New York
	A boat from New York!

6.5)	do bajra veka'a lo djine
	You run with-route a circle.
	You are running in circles.

6.6)	[to be supplied]

There are almost one hundred cmavo of selma'o BAI, based on selected gismu
that seemed useful in a variety of settings.  The list is somewhat biased
toward English, because many of the cmavo were selected on the basis of
corresponding English prepositions and preposition compounds such as
"with", "without", and "by means of".  The BAI cmavo, however, are far
more precise than English prepositions, because their meanings are fixed
by the place structures of the corresponding gismu.

All BAI cmavo have the form CV'V or CVV.  Most of them are CV'V, where
the C is the first consonant of the corresponding gismu and the two Vs are
the two vowels of the gismu.  The table in Section 17 show the exceptions.


7. Modal Sentence Connection:  The Causals

	ri'a	BAI	rinka modal: physical cause
	ki'u	BAI	krinu modal: justification
	mu'i	BAI	mukti modal: motivation
	ni'i	BAI	nibli modal: logical entailment

This section has two purposes.  On the one hand, it explains the grammatical
construct called "modal sentence connection".  On the other, it exemplifies
some of the more useful BAI cmavo: the causals.

There are four causal gismu in Lojban, or maybe six, distinguishing
different versions of the relationships lumped in English as "causal":

	rinka: event x1 physically causes event x2
	krinu: event x1 is the justification for event x2
	mukti: event x1 is the (human) motive for event x2
	nibli: event x1 logically entails event x2

Each of these gismu has a related modal:  "ri'a", "ki'u", "mu'i", and "ni'i"
respectively.  Using these gismu and these modals, we can create
various causal sentences with different implications:

7.1)	le spati cu banro
		ri'a le nu do djacu dunda fi le spati
	the plant grows
		with-physical-cause the event-of you water give to the plant.
	The plant grows because you water it.

7.2)	la djan. cpacu le pamoi seljinga ki'u le nu la djan. jinga
	John gets the first prize with-justification the event-of John wins.
	John got the first prize because he won.

7.3)	mi lebna le cukta mu'i le nu mi djica le cukta
	I took the book with-motivation the event-of I wanted the book.
	I took the book because I wanted it.

7.4)	la sokrates. morsi binxo ni'i le nu la sokrates. remna
	Socrates dead-became with-logical-justification Socrates is-human.
	Socrates died because Socrates is human.

In Examples 7.1-7.4, the same English word "because" is used to translate
all four modals, but the types of cause being expressed are quite different.
Let us now focus on Example 7.1, and explore some variations on it.

As written, Example 7.1 claims that the plant grows, but only refers to
the event of watering it without actually making a claim.  This is semantically
asymmetrical.  Suppose we wanted to claim that the plant was being watered,
and only mention its growth as ancillary information?  Then we could say:

7.5)	do djacu dunda fi le spati seri'a le nu ri banro
	You water-give to the plant with-physical-effect it grows.
	You water the plant; therefore, it grows.

In addition, there are also symmetrical forms:

7.6)	le nu do djacu dunda fi le spati cu rinka
		le nu le spati cu banro
	The event-of (you water-give to the plant) causes
		the event-of (the plant grows).
	Your watering the plant causes its growth.
	If you water the plant, then it grows.

does not claim either event, but asserts only the causal relationship
between them.  The second colloquial translation shows a form of "if-then"
in English quite distinct from the logical connective "if-then" explained
elsewhere.

Suppose we wish to claim both events as well as their causal relationship?
We can use one of two methods:

7.7)	le spati cu banro .iri'abo do djacu dunda fi le spati
	The plant grows.  Because you water-give to the plant.
	The plant grows because you water it.

7.8)	do djacu dunda fi le spati .iseri'abo le spati cu banro
	You water-give to the plant.  Therefore it grows.
	You water the plant; therefore, it grows.

The compound cmavo ".iri'abo" and ".iseri'abo" serve to connect two bridi,
as the initial ".i" indicates.  The final "bo" is necessary to prevent
the modal from "taking over" the following sumti.  If the "bo" were omitted
from Example 7.7 we would have:

7.9)	le spati cu banro .i ri'a do djacu dunda fi le spati
	The plant grows.  Because of you, [observative] water-give to the plant.
	The plant grows.  Because of you, water is given to the plant.

Because "ri'a do" is now a tagged sumti, there is no longer an explicit
sumti in the x1 place of "djacu dunda", and the translation must be changed.

The effect of sentences like Example 7.7-7.9 is that the modal, "ri'a" in
this example, no longer modifies an explicit sumti.  Instead, the sumti
is implicit, the event given by a full bridi.  Furthermore, there is a
second implication:  that the first bridi fills the x2 place of the
gismu "rinka"; it specifies an event which is the effect.

In principle, any modal tag can appear in a sentence connective of the
type exemplified by Examples 7.7-7.9.  However, it makes little sense to
use any modals which do not expect events or other abstractions to fill
the places of the corresponding gismu.  The sentence connective ".ibaubo"
is perfectly grammatical, but it is hard to imagine any two sentences
which could be connected by an "in-language" modal.


8. Other Modal Connections

Like many Lojban grammatical constructions, sentence modal connection has
both forethought and afterthought forms.  Section 7 exemplifies only
afterthought modal connection, illustrated here by:

8.1)	mi jgari lei djacu .iri'abo mi jgari le kabri
	I grasp the-mass-of water with-physical-cause I grasp the cup.
	I grasp the water because I grasp the cup.

An afterthought connection is one that is signaled only by a cmavo (or
compound cmavo, in this case) between the two constructs being connected.
Forethought connection uses a signal both before the first construct and
between the two:  the use of "both" and "and" in the first half of this
sentence represents a forethought connection (though not a modal one).

To make forethought modal sentence connections in Lojban, place the modal
plus "gi" before the first bridi, and "gi" between the two.  No ".i" is
used.  The forethought equivalent of Example 8.1 is:

8.2)	ri'agi mi jgari le kabri gi mi jgari lei djacu
	With-physical-cause I grasp the cup , I grasp the-mass-of water.
	Because I grasp the cup, I grasp the water.

Note that the cause, the x1 of "rinka" is now placed first.  To keep the
two bridi in the original order of Example 8.1, we could say:

8.3)	seri'agi mi jgari lei djacu gi mi jgari le kabri
	With-physical-effect I grasp the-mass-of water , I grasp the cup.

In English, the sentence "*Therefore I grasp the water, I grasp the cup"
is ungrammatical, because "therefore" is not a conjunction like "because".
In Lojban, "seri'agi" can be used just like "ri'agi".

When the two bridi joined by a modal connection have one or more elements
(selbri or sumti or both) in common, there are various condensed forms
that can be used in place of full modal sentence connection with both
bridi completely stated.

When the bridi are the same except for a single sumti, as in Examples 8.1-8.3,
then a sumti modal connection may be employed:

8.4)	mi jgari ri'agi le kabri gi lei djacu
	I grasp because the cup , the-mass-of water.

Example 8.4 defies a simple translation, but it means exactly the same as
Examples 8.1-8.3.

If the two connected bridi are different in more than one sumti, then a
termset may be employed.  Termsets are explained more fully elsewhere,
but are essentially a mechanism for creating connections between multiple
sumti simultaneously.

8.5)	mi dunda le cukta la djan. .imu'ibo la djan. dunda lei jdini mi
	I gave the book to John.  Motivated-by John gave the-mass-of money to-me.
	I gave the book to John, because John gave money to me.

means the same as:

8.6)	nu'i mu'igi mi le cukta la djan. gi la djan. lei jdini mi nu'u dunda
	[start] because I, the book, John; John, the-mass-of money, me [end] gives.

Here there are three sumti in each half of the termset, because the two
bridi share only their selbri.

There is no modal connection between selbri as such: bridi which differ
only in the selbri can be modally connected using bridi-tail modal
connection.  Bridi-tails are more fully explained elsewhere, but essentially
consist of a selbri with optional sumti following it.  Example 7.3 is
suitable for bridi-tail connection, and could be shortened to:

8.7)	mi mu'igi djica le cukta gi lebna le cukta
	I, because desired the book, took the book.

Again, no straightforward English translation exists.  It is even possible
to shorten Example 8.7 further to:

8.8)	mi mu'igi djica gi lebna vau le cukta
	I because desired, therefore took, the book.

where "le cukta" is set off by the non-elidable "vau" and is made to belong
to both bridi-tails.  See elsewhere for more explanations.

Finally, mathematical operands can be modally connected.

8.9)	li ny. du li vo
		.ini'ibo li ny. du li re su'i re
	the number n = the-number 4.
		Motivated-by the-number n = the-number 2 + 2.
	n = 4 because n = 2 + 2.

can be reduced to:

8.10)	li ny. du li ni'igi vei re su'i re [ve'o] gi vo
	the-number n = the-number because ( 2 + 2 ) therefore 4.

The cmavo "vei" and "ve'o" represent mathematical parentheses, and are
required so that "ni'igi" affects more than just the immediately following
operand, namely the first "re".  (The right parenthesis, "ve'o", is an
elidable terminator.)  As usual, no English translation does
Example 8.10 justice.

Note:  Due to mechanical restrictions on the Lojban parsing algorithm,
it is not possible to form modal connectives using the "fi'o"-plus-selbri
form of modal.  Only the predefined modals of selma'o BAI can be compounded
as shown in Sections 7 and 8.


9. Modal Selbri

Consider the example:

9.1)	mi tavla bau la lojban. bai la lojbab.
	I speak in-language Lojban with-compeller Lojbab.
	I speak in Lojban, under compulsion by Lojbab.

Example 9.1 has two modal sumti, using the modals "bau" and "bai".
Suppose we wanted to specify the language explicitly but be vague about
who's doing the compelling.  We can simplify Example 9.1 to:

9.2)	mi tavla bau la lojban. bai [ku].
	I speak in-language Lojban under-compulsion.

In Example 9.2, the elidable terminator "ku" has taken the place of the
sumti which would normally follow "bai".  Alternatively, we could specify
the compeller but keep the language vague:

9.3)	mi tavla bau [ku] bai la lojbab.
	I speak in-some-language under-compulsion-by Lojbab.

We are also free to move the modal-plus-"ku" around the bridi:

9.4)	bau [ku] bai ku mi tavla
	In-some-language under-compulsion I speak.

An alternative to using "ku" is to place the modal cmavo right before the
selbri, following the "cu" which often appears there.  When a modal is
present, the "cu" is almost never necessary.

9.5)	mi bai tavla bau la lojban.
	I compelledly speak in-language Lojban.

In this use, the modal is like a tanru modifier semantically, although
grammatically it is quite distinct.  Example 9.5 is very similar in
meaning to:

9.6)	mi se bapli tavla bau la lojban.
	I compelledly-speak in-language Lojban.

The "se" conversion is needed because "bapli tavla" would be a compeller
type of speaker rather than a compelled (by someone) type of speaker, which
is what a "bai tavla" is.

There are two other related uses of modals.  A modal can be attached to
a pair of bridi-tails that have already been connected by a logical,
non-logical, or modal connection (see elsewhere for more on logical
and non-logical connections):

9.7)	mi bai ke ge klama le zarci gi cadzu le bisli [ke'e]
	I under-compulsion ( both go to-the market and walk on-the ice )
	Under compulsion, I both go to the market and walk on the ice.

Here the "bai" is spread over both "klama le zarci" and "cadzu le bisli",
and the "ge...gi" represents the logical connection "both-and" between the
two.

Similarly, a modal can be attached to multiple sentences that have been
combined with "tu'e" and "tu'u", which are explained in more detail 
elsewhere:

9.8)	bai tu'e mi klama le zarci .i mi cadzu le bisli [tu'u]
	Under-compulsion [start] I go to-the market. I walk on-the ice [end]

means the same thing as Example 9.7.

Note:  The uses of modals discussed in this section are applicable both to
BAI modals and to "fi'o"-plus-selbri modals.


10. Modal Relative Phrases: Comparison

	pe	GOI	restrictive relative phrase
	ne	GOI	incidental relative phrase
	mau	BAI	zmadu modal
	me'a	BAI	mleca modal

Relative phrases and clauses are explained in much more detail elsewhere.
However, there is a construction which combines a modal with a relative
phrase which is relevant to this paper.  Consider the following examples
of relative clauses:

10.1)	la .apasionatas. poi se cusku la .artr. rubnstain. se nelci mi
	The Appassionata which is-expressed-by Artur Rubenstein is-liked-by me.

10.2)	la .apasionatas. noi se finti la betovn. se nelci mi
	The Appassionata, which is-created-by Beethoven, is-liked-by me.

In Example 10.1, "la .apasionatas." refers to a particular performance
of the sonata, namely the one performed by Rubenstein.  Therefore, the
relative clause "poi se cusku" uses the cmavo "poi" (of selma'o NOI) to
restrict the meaning of "la .apasionatas" to the performance in question.

In Example 10.2, however, "la .apasionatas." refers to the sonata as a whole,
and the information that it was composed by Beethoven is merely incidental.
The cmavo "noi" (also of selma'o NOI) expresses the incidental nature of
this relationship.

The cmavo "pe" and "ne" (of selma'o GOI) are roughly equivalent to "poi"
and "noi" respectively, but are followed by sumti rather than full bridi.
We can abbreviate Examples 10.1 and 10.2 to:

10.3)	la .apasionatas pe la .artr. rubnstain. se nelci mi
	The Appassionata of Artur Rubenstein is-liked-by me.

10.4)	la .apasionatas ne la betovn. se nelci mi
	The Appassionata, which is of Beethoven, is-liked-by me.

Here the precise selbri of the relative clauses is lost: all we can tell
is that the Appassionata is connected in some way with Rubenstein (in
Example 10.3) and Beethoven (in Example 10.4), and that the relationships
are respectively restrictive and incidental.

It happens that both "cusku" and "finti" have BAI cmavo, namely "cu'u" and
"fi'e".  We can recast Examples 10.3 and 10.4 as:

10.5)	la .apasionatas pe cu'u la .artr. rubnstain. se nelci mi
	The Appassionata expressed-by Artur Rubenstein is-liked-by me.

10.6)	la .apasionatas ne fi'e la betovn. se nelci mi
	The Appassionata, invented-by Beethoven, is-liked-by me.

Examples 10.5 and 10.6 have the full semantic content of Examples 10.1 and
10.2 respectively.

Modal relative phrases are often used with the BAI cmavo "mau" and "me'a",
which are based on the comparative gismu "zmadu" (more than) and "mleca"
(less than) respectively.  The place structures are:

	zmadu	x1 is more than x2 in property/quantity x3 by amount x4
	mleca	x1 is less than x2 in property/quantity x3 by amount x4

Here are some examples:

10.7)	la frank. nelci la betis. ne semau la meiris.
	Frank likes Betty, which-is more-than Mary.
	Frank likes Betty more than (he likes) Mary.

Example 10.7 requires that Frank likes Betty, but adds the information that
his liking for Betty exceeds his liking for Mary.  The modal appears in
the form "semau" because the x2 place of "zmadu" is the basis for comparison:
in this case, Frank's liking for Mary.

10.8)	la frank. nelci la meiris. ne seme'a la betis.
	Frank likes Mary, which-is less-than Betty.
	Frank likes Mary less than (he likes) Betty.

Here we are told that Frank likes Mary rather than Betty; the information
about the comparison is the same.  It would be possible to rephrase Example
10.7 using "me'a" rather than "semau", and Example 10.8 using "mau" rather
than "seme'a", but such usage would be unnecessarily confusing.  Like many
BAI cmavo, "mau" and "me'a" are more useful when converted with "se".

Pure comparison, which states only the comparative information but says
nothing about whether Frank actually likes either Mary or Betty (he may
like neither, but dislike Betty less), would be expressed differently, as:

10.9)	le ni la frank. nelci la betis.
		cu zmadu le ni la frank. nelci la meiris.
	The quantity-of Frank's liking Betty
		is-more-than the quantity-of Frank's liking Mary.

Note:  The uses of modals discussed in this section are applicable both to
BAI modals and to "fi'o"-plus-selbri modals.



11. Mixed Modal Connection

It is possible to mix logical connection (explained elsewhere) with modal
connection, in a way that simultaneously asserts the logical connection
and the modal relationship.  Consider the sentences:

11.1)	mi nelci do .ije mi nelci la djein.
	I like you.  And I like Jane

which is a logical connection, and

11.2)	mi nelci do .iki'ubo mi nelci la djein.
	I like you.  Justified-by I like Jane.

Examples 11.1 and 11.2 can be simultaneously expressed by combining the
two compound cmavo, thus:

11.3)	mi nelci do .ijeki'ubo mi nelci la djein.
	I like you.  And justified-by I like Jane.

Here the two sentences "mi nelci do" and "mi nelci la djein." are
simultaneously asserted, their logical connection is asserted, and their
causal relationship is asserted.  The logical connective "je" comes before
the modal "ki'u" in all such mixed connections.

Since "mi nelci do" and "mi nelci la djein." differ only in the final
sumti, we can transform Example 11.3 into a mixed sumti connection:

11.4)	mi nelci do .eki'ubo la djein.
	I like you and/because Jane.

Note that this connection is an afterthought one.  Mixed connectives are
always afterthought; forethought connectives must be one or the other.

There are numerous other afterthought logical and non-logical connectives
that can have modal information planted within them.  For example, a
bridi-tail connected version of Example 11.4 would be:

11.5)	mi nelci do gi'eki'ubo nelci la djein.
	I like you and/because like Jane.

The following three complex examples all mean the same thing.

11.6)	mi bevri le dakli
		.ije seri'a tu'e mi bevri le gerku .ijadu'ibo mi bevri le mlatu tu'u
	I carry the sack.
		And [effect] (I carry the dog.  And/or [equal] I carry the cat.)
	I carry the sack.
		As a result I carry the dog or I carry the cat, equally.

11.7)	mi bevri le dakli
		gi'eseri'ake bevri le gerku gi'acabo bevri le mlatu
	I carry the sack
		and [effect] (carry the dog and/or [equal] carry the cat).
	I carry the sack
		and as a result carry the dog or carry the cat equally.

11.8)	mi bevri le dakli .eseri'ake bevri le gerku .adu'ibo le mlatu
	I carry the sack and [effect] (the cat and/or [equal] the dog).
	I carry the sack, and as a result the cat or the dog equally.

Note:  Mixed modal conversions cannot involve modals made with
"fi'o"-plus-selbri; only the predefined modals of selma'o BAI can be used.


12. Modal Conversion: JAI

	jai	JAI	modal conversion
	fai	FA	modal place structure tag

So far, conversion of numbered bridi places with SE and the addition of
modal places with BAI have been two entirely separate operations.  However,
it is possible to convert a selbri in such a way that, rather than exchanging
two numbered places, a modal place is made into a numbered place.  For
example,

12.1)	mi tavla bau la lojban.
	I speak in-language Lojban.

has an explicit x1 place occupied by "mi" and an explicit "bau" place
occupied by "la lojban."  To exchange these two, we use a modal conversion
operator consisting of "jai" (of selma'o JAI) followed by the modal cmavo.
Thus, the modal conversion of Example 12.1 is:

12.2)	la lojban. jai bau tavla fai mi
	Lojban is-the-language-of-speech used-by me.

In Example 12.2, the modal place "la lojban." has become the x1 place
of the new selbri "jai bau tavla".  What has happened to the old x1 place?
There is no numbered place for it to move to, so it moves to a special
"unnumbered place" marked by the tag "fai" of selma'o FA.

Like SE conversions, JAI conversions are especially convenient in descriptions.
We may refer to "the language used in speech" as "le jai bau tavla", for
example.

In addition, it is grammatical to use "jai" without a following modal.
This usage is not related to modals, but is explained here for completeness.
The effect of "jai" by itself is to send the x1 place, which should be
an abstraction, into the "fai" position, and to raise one of the sumti
from the subordinate abstract bridi into the x1 place of the main bridi.
This feature is discussed in more detail elsewhere.  The following two
examples mean the same thing:

12.3)	le nu mi lebna le cukta
		cu se krinu le nu mi djica le cukta
	The event-of I take the book
		is-justified-by the event-of I want the book.
	My taking the book is justified by my wanting it.

12.4)	mi jai se krinu le nu mi djica
		[fai le nu mi lebna le cukta]
	I am-justified by the event-of I want the book
		[namely, the event-of I take the book]
	I am justified by wanting the book.

Example 12.4, with the bracketed part omitted, allows us to say that "I am
justified" whereas in fact it is my action that is justified.  This
construction is vague, but useful in representing natural-language methods
of expression.

Note:  The uses of modals discussed in this section are applicable both to
BAI modals and to "fi'o"-plus-selbri modals.


13. Modals Versus Tenses

Grammatically, every use of modals seen so far is exactly paralleled by
some use of tenses.  Tenses can be followed by sumti, can appear before the
selbri, can be used in pure and mixed connections, can participate in
JAI conversions.  The parallelism is perfect.  However, there is a deep
difference in the semantics of tense constructs and modal constructs,
grounded in historical differences between the two forms.  Originally,
modals and tenses were utterly different things in earlier versions of
Loglan; only in Lojban have they become grammatically interchangeable.
And even now, differences in semantics continue to be maintained.

The core distinction is that whereas

13.1)	mi nelci do mu'i le nu do nelci mi
	I like you with-motivation the event-of you like me.
	I like you because you like me.

places the "le nu" sumti in the x1 place of the underlying gismu "mukti",
namely the motivating event,

13.2)	mi nelci do ba le nu do nelci mi
	I like you after the event-of you like me.
	I like you after you like me.

places the "le nu" sumti in the x2 place of the underlying gismu "balvi",
namely the point of reference for the future tense.  As a result, the
afterthought sentence-connective forms of Examples 13.1 and 13.2 are,
respectively:

13.3)	mi nelci do .imu'ibo do nelci mi
	I like you.  [That is] Because you like me.

13.4)	do nelci mi .ibabo mi nelci do
	You like me.  Afterward, I like you.

In Example 13.3, the order of the two bridi "mi nelci do" and "do nelci mi"
is the same as in Example 13.1.  In Example 13.4, however, the order is
reversed: the origin point "do nelci mi" physically appears before the
future-time event "mi nelci do".  In both cases, the bridi characterizing
the event in the x2 place appears before the bridi characterizing the event
in the x1 place of "mukti" or "balvi".

Schematically, "X .i+BAI+bo Y" may be understood as "X BAI le nu Y", whereas
"X .i+TENSE+bo Y" really means "Y TENSE le nu X".  These equivalences 
neglect the fact that the sentence-connective forms claim both X and Y,
whereas the alternative "le nu" forms do not.

In forethought connections, however, the asymmetry between modals and
tenses is not found.  The forethought equivalents of Examples 13.3 and 13.4
are simply

13.5)	mu'igi do nelci mi gi mi nelci do
	Because you like me, I like you.

and

13.6)	bagi do nelci mi gi mi nelci do
	After you like me, I like you.


14. Modal Negation

Negation is explained in detail elsewhere.  There are two forms of
negation in Lojban: contradictory and scalar negation.  Contradictory
negation expresses what is false, whereas scalar negation says that
some alternative to what has been stated is true.  A simple example
is the difference between "John didn't go to Paris" (contradictory
negation) and "John went to (somewhere) other than Paris" (scalar
negation).

Contradictory negation involving BAI cmavo is performed by appending
"-nai" (of selma'o NAI) to the BAI.  A common use of modals with "-nai"
is to deny a causal relationship:

14.1)	mi nelci do mu'inai le nu do nelci mi
	I like you, but not because you like me.

Example 14.1 denies that the relationship between my liking you (which is
asserted) and your liking me (which is not asserted) is one of motivation.
Nothing is said about whether you like me or not, merely that that
hypothetical liking is not the motivation for my liking you.

Scalar negation is achieved by prefixing "na'e" (of selma'o NAhE), or any
of the other cmavo of NAhE, to the BAI cmavo.

14.2)	le spati cu banro na'emu'i le nu
		do djacu dunda fi le spati
	The plant grows other-than-motivated-by the event-of
		you water-give to the plant.

Example 14.2 says that the relationship between the plant's growth and
your watering it is not one of motivation: the plant is not motivated
to grow, as plants are not something which can have motivation as a rule.
Implicitly, some other relationship between watering and growth exists,
but Example 14.2 doesn't say what it is (presumably "ri'a").

Note:  Modals made with "fi'o" plus a selbri cannot be negated directly.
The selbri can itself be negated either with contradictory or with scalar
negation, however.


15. Sticky Modals

	ki	KI	stickiness flag

Like tenses, modals can be made persistent from the bridi in which they
appear to all following bridi.  The effect of this "stickiness" is to
make the modal, along with its following sumti, act as if it appeared in
every successive bridi.  Stickiness is put into effect by following the modal
(but not any following sumti) with the cmavo "ki" of selma'o KI.  For
example,

15.1)	mi tavla bau la lojban. bai ki la lojbab.
		.ibabo mi tavla bau la gliban.
	I speak in-language Lojban compelled-by Lojbab.
		Afterward, I speak in-language English.

means the same as:

15.2)	mi tavla bau la lojban. bai la lojbab.
		.ibabo mi tavla bau la gliban. bai la lojbab.
	I speak in-language Lojban compelled-by Lojbab.
		Afterward, I speak in-language English compelled-by Lojbab.

In Example 15.1, "bai" is made sticky, and so Lojbab's compelling is
made applicable to every following bridi.  "bau" is not sticky, and so
the language may vary from bridi to bridi, and if not specified in a particular
bridi, no assumption can safely be made about its value.

To cancel stickiness, use the form "BAI ki ku", which stops any modal
value for the specified BAI from being passed to the next bridi.  There is
no simple way to cancel stickiness for all modals simultaneously.

Note:  Modals made with "fi'o"-plus-selbri cannot be made sticky.  This is
an unfortunate, but unavoidable, restriction.


16. Logical And Non-Logical Connection Of Modals

Logical and non-logical connectives are explained in detail elsewhere.
For the purposes of this paper, it suffices to point out that a logical
(or non-logical) connection between two bridi which differ only in a
modal can be reduced to a single bridi with a connective between the
modals.  As a result, Examples 16.1 and 16.2 mean the same thing:

16.1)	la frank. nelci la betis. semau la meiris.
		.ija la frank. nelci la betis. du'i la meiris.
	Frank likes Betty more-than Mary.
		Or Frank likes Betty equally-with Mary.

16.2)	la frank. nelci la betis. semau ja du'i la meiris.
	Frank likes Betty more-than or equally-with Mary.


17. CV'V cmavo of selma'o BAI with irregular forms

The following cmavo of selma'o BAI do not follow the regular rule for
BAI cmavo, namely that they are of the form CV'V, where C is the first
consonant of the corresponding gismu, and the Vs are the two vowels of the
gismu.  The cmavo are divided into sub-tables according to the nature of the
exception; some cmavo appear in more than one sub-table, and are so noted.

		cmavo	gismu		comments

Monosyllables of the form CVV:

		bai	bapli
		bau	bangu
		cau	claxu
		fau	fasnu
		gau	gasnu
		kai	ckaji		also uses 2nd consonant of gismu
		mau	zmadu		also uses 2nd consonant of gismu
		koi	korbi
		rai	traji		also uses 2nd consonant of gismu
		sau	srasu
		tai	tamsmi		also based on lujvo, not gismu
		zau	zanru

Second consonant of the gismu (which is always of the form CCVCV) as the C:

		ga'a	zgana
		kai	ckaji		also has CVV form (monosyllable)
		ki'i	ckini
		la'u	klani		also has irregular 2nd V
		le'a	klesi		also has irregular 2nd V
		mau	zmadu		also has CVV form (monosyllable)
		me'e	cmene
		ra'a	srana
		ra'i	krasi
		rai	traji		also has CVV form (monosyllable)
		ti'i	stidi
		tu'i	stuzi

Irregular 2nd V:

		fi'e	finti
		la'u	klani		also uses 2nd consonant of gismu
		le'a	klesi		also uses 2nd consonant of gismu
		ma'e	marji
		mu'u	mupli
		ti'u	tcika
		va'o	vanbi

Special cases:

		ri'i	lifri		uses 3rd consonant of gismu
		tai	tamsmi		based on lujvo, not gismu
		va'u	xamgu		CV'V cmavo can't begin with "x"

The lujvo "tamsmi" on which "tai" is based is derived from the tanru
"tarmi simsa" and has the place structure:

	x1 has form x2, similar in form to x3 in property/quality x4

This lujvo is employed because "tarmi" does not have a useful place
structure.

-- 
John Cowan	cowan@snark.thyrsus.com		...!uunet!lock60!snark!cowan
			e'osai ko sarji la lojban.